Research at the Christian de Duve Institute of Cellular ... Report DDU… · Structural studies on...

Research at

the Christian de Duve

Institute of Cellular Pathology (ICP)

and at

the Brussels Branch of

the Ludwig Institute for Cancer Research (LICR)

August 2004

TABLE OF CONTENTS

CHRISTIAN DE DUVE INSTITUTE OF CELLULAR PATHOLOGY (ICP)............... 1ICP: an International Biomedical Research Institute................................................................................. 3ADMINISTRATION AND GENERAL SERVICES ............................................................................ 5DEVELOPMENT AND EXPANSION COUNCIL (DEC) ................................................................ 6ACKNOWLEDGMENTS........................................................................................................................... 7

CARBOHYDRATE METABOLISM.................................................................................................................. 8Emile VAN SCHAFTINGEN, MemberMaria VEIGA-da-CUNHA, Associate Member

Protein deglycation......................................................................................................................................... 8Fructosamine metabolism in bacteria ........................................................................................................ 10Regulation of Vitamin C synthesis ............................................................................................................. 11Metabolism of L- and D-2-hydroxyglutarate............................................................................................ 12Inborn errors of metabolism....................................................................................................................... 12Selected publications .................................................................................................................................... 13

PURINE METABOLISM.................................................................................................................................... 14Françoise BONTEMPS, Associate MemberMarie-Françoise VINCENT, Associate Member

Adenylosuccinate lyase deficiency.............................................................................................................. 14A new defect in the purine biosynthesis.................................................................................................... 15Antileukaemic properties of 2-chloro-2'-deoxyadenosine....................................................................... 16Selected publications .................................................................................................................................... 17

TISSUE-SPECIFIC TRANSCRIPTION FACTORS IN DEVELOPMENT............................................. 19Guy G. ROUSSEAU, MemberFrédéric P. LEMAIGRE, Member

Control and molecular mode of action of HNF-6................................................................................... 20Control of pancreas development by HNF-6........................................................................................... 20Control of liver development by HNF-6 .................................................................................................. 21Role of OC-2 and OC-3 in the development of endoderm derivatives................................................ 22Selected publications .................................................................................................................................... 22

SIGNAL TRANSDUCTION AND PROTEIN PHOSPHORYLATION................................................. 24Louis HUE, MemberMark H. RIDER, Member

Insulin signalling ........................................................................................................................................... 25AMP-activated protein kinase..................................................................................................................... 25Control of smooth muscle contraction ..................................................................................................... 27Selected publications .................................................................................................................................... 27

ENDOCYTOSIS ................................................................................................................................................... 29Pierre J. COURTOY, MemberMarie-France VANDENBROUCKE, Member

Oncogene-induced macropinocytosis in fibroblasts................................................................................ 29Regulation of endocytosis by v-Src in polarized cells.............................................................................. 29Relation between endocytosis and cell motility........................................................................................ 30Regulation of transcytosis............................................................................................................................ 31Apical endocytosis regulates thyroid hormone production in human disease; molecular dissectionin vitro.............................................................................................................................................................. 31Alterations in the endocytic apparatus of mice lacking renal chloride channel, ClC-5, account forproteinuria in Dent’s disease....................................................................................................................... 31Selected publications .................................................................................................................................... 32

EXTRACELLULAR MATRIX BREAKDOWN............................................................................................. 33Pierre J. COURTOY, MemberYves EECKHOUT, MemberEtienne MARBAIX, Member

Regulation of the expression of endometrial MMPs and cytokines ...................................................... 33Selective binding of active MMP-7 to human epithelial cell membrane............................................... 34Receptor-mediated endocytic clearance of proMMP-2:TIMP-2 complexes ........................................ 34Role of matrix metalloproteinases in abnormal endometrial bleeding.................................................. 35Selected publications .................................................................................................................................... 35

CONNECTIVE TISSUE AND ARTHRITIS.................................................................................................. 37Daniel MANICOURT, Member

Effects of nonsteroidal anti-inflammatory drugs on the overall metabolism of articular cartilage ... 37Markers of connective tissue metabolism in health and disease ............................................................ 38Role of the subchondral bone in the initiation and progression of the osteoarthritic diseaseprocess ........................................................................................................................................................... 38Towards a better understanding of the metabolism of hyaluronan in connective tissues.................. 38Selected publications .................................................................................................................................... 39

METABOLIC COMPARTMENTATION IN TRYPANOSOMES............................................................ 40Fred R. OPPERDOES, MemberPaul A.M. MICHELS, Member

Enzymes of carbohydrate metabolism ...................................................................................................... 41Enzymes of the hexose monophosphate pathway (HPM) ..................................................................... 43The presence of plant traits in the Trypanosomatidae ................................................................................. 43Analysis of the control of the glycolytic flux ............................................................................................ 44Biogenesis of glycosomes ............................................................................................................................ 44Analysis of glycosomal membrane solute.................................................................................................. 45Selected publications .................................................................................................................................... 45

GENETICS OF HUMAN CARDIOVASCULAR ANOMALIES, SKELETAL DISORDERS ANDCEREBRAL TUMORS ........................................................................................................................................ 46

Miikka VIKKULA, MemberVenous malformations, glomuvenous malformations (“glomangiomas”) and Maffucci syndrome. 46Lymphedema................................................................................................................................................. 47Vascular anomalies affecting capillaries..................................................................................................... 47Cardiopathies................................................................................................................................................. 48Cleft lip and palate........................................................................................................................................ 48Other cutaneous disorders .......................................................................................................................... 49Cerebral tumors ............................................................................................................................................ 49Conclusions ................................................................................................................................................... 49Selected publications .................................................................................................................................... 50

THEILER'S ENCEPHALOMYELITIS VIRUS.............................................................................................. 51Thomas MICHIELS, Associate Member

Analysis of viral proteins involved in Theiler's virus escape of the host immune response .............. 52Inhibition of type-I interferon production by the leader protein........................................................... 52Influence of the L* protein on macrophage infection and viral persistence........................................ 52Characterization of the type-I IFN response............................................................................................ 53Selected publications .................................................................................................................................... 53

ANTIVIRAL IMMUNITY AND PATHOGENESIS .................................................................................... 54Jean-Paul COUTELIER, Associate Member

Viral infections result in a dramatic increase in the proportion of IgG2a ............................................ 54Activation of natural killer cells .................................................................................................................. 54Activation of macrophages ......................................................................................................................... 55Autoimmune thrombocytopenia ................................................................................................................ 55Selected publications .................................................................................................................................... 55

SIGNAL TRANSDUCTION AND GENE REGULATION BY PLATELET-DERIVEDGROWTH FACTORS.......................................................................................................................................... 57

Jean-Baptiste DEMOULIN, Assistant MemberGene regulation by PDGF.......................................................................................................................... 58Regulation of the PDGF receptor kinase activity by its C-terminal tail and NHERF........................ 58Role of Gab1 in PDGF signal transduction ............................................................................................. 58Selected publications: ................................................................................................................................... 59

HUMAN TUMOR IMMUNOLOGY*.............................................................................................................. 60Pierre G. COULIE, Member

Correlation between tumor regression and T cell responses in melanoma patients vaccinated witha MAGE antigen........................................................................................................................................... 60Tumor regressions observed after vaccination: a possible role for tumor-specific cytolytic Tlymphocytes that do not recognize the vaccine antigens ........................................................................ 62Selected publications .................................................................................................................................... 63

LUDWIG INSTITUTE FOR CANCER RESEARCH (LICR) ........................................ 65

ADMINISTRATION AND GENERAL SERVICES .................................................................................... 68Administration and secretariat .................................................................................................................... 68Pre-clinical Research Resources Group:.................................................................................................... 68Technical support ......................................................................................................................................... 68

TUMOR IMMUNOLOGY AND ANTIGEN PROCESSING.................................................................... 69Benoît VAN DEN EYNDE, Member

Differential processing of tumor antigens by the standard proteasome and theimmunoproteasome ..................................................................................................................................... 70An antigenic peptide produced by peptide splicing in the proteasome................................................. 70Identification of new antigens recognized by autologous CTL on human melanoma ....................... 71A novel tumor immune escape mechanism based on tryptophan degradation by indoleamine 2,3dioxygenase.................................................................................................................................................... 71Development of an inducible mouse melanoma model for immunotherapy ...................................... 72Antigen presentation by dendritic cells in Systemic Lupus Erythematosus ......................................... 72Selected publications .................................................................................................................................... 73

GENES EXPRESSED IN CANCER AND GERMLINE CELLS.............................................................. 74Etienne De PLAEN, Assistant member

Isolation and characterization of new cancer-germline genes ................................................................ 74MAGE-A1 acts as a transcriptional repressor.......................................................................................... 74Genome hypomethylation in the activation of cancer-germline genes in tumors............................... 75Selected publications .................................................................................................................................... 75

IDENTIFICATION OF HUMAN TUMOR ANTIGENS........................................................................... 77Pierre van der BRUGGEN, Member

New MAGE antigens recognized by CD8+ and CD4+ T cells .............................................................. 77A novel approach to identify antigens recognized by CD4 T cells using complement-opsonizedbacteria expressing a cDNA library............................................................................................................ 80The normal anti-MAGE-3.A1 repertoire.................................................................................................. 80A reversible functional defect of CD8+ T lymphocytes involving loss of tetramer labeling.............. 80Detection of anti-vaccine CD4 T cell response in vaccinated patients................................................. 81Selected publications .................................................................................................................................... 81

THERAPEUTIC VACCINATION OF CANCER PATIENTS WITH TUMOR-SPECIFICANTIGENS............................................................................................................................................................ 83

Marie MARCHAND, Senior Clinical InvestigatorTherapeutic vaccination with MAGE tumor antigens ............................................................................ 83Selected publications .................................................................................................................................... 85

ANALYSIS OF T CELL RESPONSES OF VACCINATED CANCER PATIENTS *........................... 86Aline VAN PEL, Associate Member

Methods for evaluation of T-cell responses in vaccinated cancer patients........................................... 86A melanoma patient vaccinated with a MAGE-3 recombinant canarypox virus ................................ 87A melanoma patient vaccinated with dendritic cells pulsed with a MAGE-3 peptide presented byHLA-A1......................................................................................................................................................... 88Selected publications .................................................................................................................................... 89

CYTOKINES IN IMMUNITY AND INFLAMMATION ........................................................................... 90Jean-Christophe RENAULD, Member

Interleukin 9 .................................................................................................................................................. 90IL-9 receptor and signal transduction........................................................................................................ 91Anti-apoptotic activity of I-309receptor and signal transduction.......................................................... 92IL-9-induced genes....................................................................................................................................... 92LICR2: a new cytokine receptor mediating antiviral activities ............................................................... 93Selected publications .................................................................................................................................... 93

SIGNAL TRANSDUCTION GROUP: STRUCTURE AND FUNCTION OF CYTOKINERECEPTORS ......................................................................................................................................................... 95

Stefan CONSTANTINESCU, Associate MemberDetermination of the interface and orientation of the activated erythropoietin receptor dimer....... 95Structural studies on the transmembrane and juxtamembrane cytosolic sequences of the EpoR .... 97Signaling by the thrombopoietin receptor................................................................................................. 97Signaling by the IL9R via the common g chain (gc) and traffic of cytokine receptors to the cell-surface ............................................................................................................................................................ 97Sequence–specific interactions between transmembrane domains ...................................................... 98Constitutive activation of JAK-STAT signaling pathways and genes targeted by STAT5 intransformed hematopoietic and patient-derived leukemia cells ............................................................. 98Selected publications .................................................................................................................................... 99

1

Christian de Duve Institute of Cellular Pathology (ICP)

3

ICP: an International Biomedical Research Institute

When Christian de Duve founded the

Institute of Cellular Pathology (ICP) in 1974, he

was acutely aware of the constrast between the

enormous progress in biological sciences that

had occurred in the 20 preceding years and the

modesty of the medical advances that had

followed. He therefore created a research

institution based on the principle that basic

research in biology would be pursued by the

investigators with complete freedom, but that

special attention would be paid to the

exploitation of basic advances for medical

progress. It was therefore highly appropriate for

the Institute to be located on the campus of the

Faculty of Medicine of the University of Louvain

(UCL). This campus is located in Brussels. The

University hospital (Clinique St Luc) is located

within walking distance of ICP.

Emile Van Schaftingen

Benoît Van den Eynde The main commitment of the members of

ICP is research. Discovery is the endpoint of

their efforts and the only element taken into

account for their evaluation. But the Institute

functions in symbiosis with the Faculty of

Medicine and many of its senior members hold a

Faculty posit ion and have teaching

appointments. The influx of doctoral students

and postdoctoral fellows from the University is

also a key element in the success of the Institute.

.

4

In 1978 the Ludwig Institute for Cancer Research decided to base its Belgian branch within the walls

of ICP. A happy collaboration between the two Institutions has been pursued since that time. Even

though the two Institutes are completely independent, the collaboration between the scientists of ICP and

the Ludwig Institute is extremely close and the sharing of resources is considerable.

ICP is managed by a directorate of three scientists, presently composed of Emile Van Schaftingen,

Benoît Van den Eynde, and Miikka Vikkula. The directorate is appointed by the Board of directors, which

comprises the Rector of the University of Louvain, one of the Pro-rectors, the General Administrator of

the University and the Dean of the Faculty of Medicine. Also present in the Board of directors are

eminent members of the business community.

About 170 researchers work in ICP and in

the Ludwig Institute, assisted by a technical and

administrative staff of about 80 members.

Despite this relatively small size, ICP has the

ambition of pursuing research projects of high

quality under conditions that allow original,

long-term projects to be pursued. The Institute

has a limited endowment, which is a source of

key financing for priority issues, such as the

creation of new laboratories for promising

young researchers. We expect that the quality of

our researchers, supported by sound

organisational approaches, will enable ICP to

stand at the forefront of European Research.

Miikka Vikkula

5

ADMINISTRATION AND GENERAL SERVICES

Rolande BOUCKAERT, External Relations Manager

Lydia HAUBOLD, Accountant

Philippe PENNINCKX, Finance and Administration Manager

Monique VAN de MAELE, Executive Secretary

Contact!:

ICP 75.50, Avenue Hippocrate 75, 1200 Bruxelles, Belgium

Tel!: +32 (2) 764 7550Fax!: +32 (2) 764 7573

6

DEVELOPMENT AND EXPANSION COUNCIL (DEC)

Membership

President : Baron PETERBROECK

Honorary Presidents : Me P. HOURDEAUBaron van der REST (+)Baron de TILLESSE, Honorary President

of the Board of Directors

Honorary Vice-Presidents :Mr A. LAGASSE de LOCHT (+)Baronne M. VELGE

Fellowship and/or Laboratory Sponsors :

Mrs C. DELORIMrs G. DELORIMrs H. DELORIMr P. DELORIMrs L. DRESSE de LEBIOLESMrs Ph. FABRIComte B. de GRUNNEBaron van GYSEL de MEISEMr P. HOLLENFELTZ du TREUXMr & Mrs M. HUYNENMr & Mrs A. LACROIXMrs M. LHOISTMr & Mrs D. MATHIEUBaron & Baronne PETERBROECKMr M. RUIZ de ARCAUTEVicomte P. de SPOELBERCHDr J. TROQUETBaron M. VELGEMrs M. de VISSCHERFONDS JACQUES GOORUMICORE

MembersMr E. van BARENMr J. BERTEAU

Mr & Mrs R. BOON-FALLEURMr V. BRUNEAUMrs G. van CUTSEM-STAES POLETVicomtesse de DUVEMr S. EMSENSBaronne de GERLACHE de GOMERYMr P. GUILMOTMr V. JACOBS van MERLENMr F. JAMARMr L. LHOISTComte J. de LIEDEKERKEDr J. MAILLETMr & Mrs O. MAIRLOTMr B. MATTLETMr E. MOREL de BOUCLE ST DENISMrs A. PIERONMr J. SAMAINMr & Mrs SOLVAY de LA HULPEBaronne H. van der STRATEN WAILLETDr E. VAN ASSCHEMr E. VAN CAMPENHOUTMrs J. VAN CAUWENBERGHEDr P. VAN HONSEBROUCKMr Ph. de VISSCHER

AXAINGBANQUE DEWAAYBEKAERTCABLERIE D'EUPENCARLSBERG IMPORTERSCARMEUSECHEMITEXFONDATION MAGGY et ROBERT de

HOVREGBLGROUPE LHOISTPFIZERSIBEKA

7

ACKNOWLEDGMENTS

In 2004, the ICP has attracted major gifts from several foundations, companies and individuals who havebeen very generous. These sponsors are providing the resources which enable our scientists to betterunderstand and treat diseases that afflict people around the world. Gifts are the lifeblood of new researchinitiatives and private resources are crucial in underwriting the costs of new laboratories. On an annualbasis, fund-raising from private sources has nearly tripled during the past decade over levels achievedpreviously and now support 6 % of the ICP’s budget.

The appeal for sponsoring postdoctoral fellowships was also widely followed. In 2004 the ICP has beenable to allocate the following fellowships, entirely supported by our donors :

the "Haas-Teichen" fellowship to Tomoko SO

the "Michel de Visscher" fellowship to Bruno GUIGAS

the "Philippe Delori" fellowship to Katharina KUBATZKY

the "Pierre Lacroix" fellowship to Anders KALLIN

the "Umicore" fellowship to Delphine GERBOD

We express our gratitude to all who contributed to the financing of post-doctoral fellows and state-of-the-art research laboratories at the ICP, ensuring that this institute will remain at the top of the field inbiomedical research.

Jean PETERBROECK,President of the Development and Expansion Council

8

CARBOHYDRATE METABOLISMEmile VAN SCHAFTINGEN, MemberMaria VEIGA-da-CUNHA, Associate Member

Thierry de BARSY, MemberYounes ACHOURI, Postdoctoral fellowJean-François COLLET, Postdoctoral FellowGhislain DELPIERRE, Postdoctoral FellowJérôme DELPLANQUE, Postdoctoral FellowFrançois COLLARD, Graduate StudentJuliette FORTPIED, Graduate StudentRita GEMAYEL, Graduate StudentCarole LINSTER, Graduate StudentPushpa MALIEKAL, Graduate StudentRim RZEM, Graduate StudentElsa WIAME, Graduate StudentGeneviève CONNEROTTE, TechnicianGaëtane NOËL, TechnicianCatherine PEEL, TechnicianIlse SCHMIDT, SecretaryKarim ACHERKI, Technical Staff

For many years, the main interest of our group has been the regulation of carbohydrate metabolism in mammals.Two important contributions have been the discovery of fructose 2,6-bisphosphate (in 1980, in collaboration withL Hue, HORM unit, and HG Hers, former head of this group) and that of the regulatory protein of glucokinase.Our laboratory has also identified several “new” enzymatic deficiencies in patients with inborn errors of metabolism(serine biosynthesis and degradation defects; phosphomannomutase deficiency, in collaboration with prof. J. Jaeken,KULeuven) and the identification of the gene mutated in glycogen storage disease type Ib. As a result of this, partof our work has been devoted to the biochemical characterization of some of the enzymes involved in thesedeficiencies. More recently, the study of the mechanism of formation of the intriguing phosphate ester, fructose 3-phosphate, has led us to identify fructosamine 3-kinase. This has brought us into a very different field, that ofprotein repair.

Protein deglycation

Gh. Delpierre, F. Collard, J. Fortpied, R. Gemayel, E.Wiame, , G. Connerotte, K. Peel, M. Veiga-da-Cunha,E. Van Schaftingen in collaboration with M.H. Riderand D. Vertommen, Horm Unit

Fructosamine 3-kinase

Chronic elevation of the blood glucoseconcentration in diabetes appears to beresponsible for the long-term complications ofthis disease. The link between the elevatedconcentration of glucose and the developmentof these complications is not yet clear. One ofthe theories on this link emphasizes the role offructosamines. These are formed through aspontaneous reaction (known as ‘glycation’) of

glucose with primary amines, followed by anAmadori rearrangement. Fructosamine 3-kinase(FN3K) is a recently identified enzyme thatphosphorylates both low-molecular-weight andprotein-bound fructosamines [1]. Fructosamine3-phosphates are unstable, breaking downspontaneously to 3-deoxyglucosone, inorganicphosphate and the amino compound thatoriginally reacted with glucose (Fig. 1). FN3K istherefore a ‘deglycating’ enzyme. Evidence hasbeen provided that this enzyme indeed removessome of the fructosamine residues present onhemoglobin in erythrocytes, a cell type in whichFN3K is particularly active and from which itwas first isolated. Taken together all thesefindings indicate that FN3K initiates a newprotein repair mechanism.

9

Figure 1. Role of fructosamine 3-kinase in the deglycation of proteins

It is often stated that proteins, being easilyrenewed, do not need to be repaired, unlikeDNA where mutations can threaten the life ofthe cell. However, two types of protein repairmechanisms have been quite well established.These are the reduction of methionine-S-oxideresidues to methionyl residues and theconversion of L-isoaspartate residues to L-aspartate residues by a methyl transferase.Inactivation of these repair mechanisms in miceresults in a diseased state and in a decreasedlifespan, underlining their physiologicalimportance. Phosphorylation of fructosamineresidues by FN3K represents a third proteinrepair mechanism, the physiological significanceof which is presently evaluated.

The results of our previous studies indicatedthat only part of the fructosamines that form onhemoglobin can be removed as a result of theaction of FN3K in intact erythrocytes. We haverecently determined the identity of thesefructosamine residues [4]. This was done bycomparing the glycation sites of hemoglobinderived from erythrocytes incubated with anelevated glucose concentration in the absence orin the presence of deoxymorpholinofructose, a(substrate and) competitive inhibitor of FN3K.The identification of the residues was facilitatedby their prior tagging by in vitro incubation ofhemoglobin with FN3K and [g-32P]ATP. Afterreduction of fructosamine 3-phosphates withborohydride, the labeled hemoglobin wasdigested with trypsin. Peptides were separated byreverse-phase HPLC and the radioactive peaks

were analyzed by mass spectrometry. This typeof analysis indicated that fructosamines boundto Lys-a-16, Lys-a-139 and Lys-b-17 werereadily deglycated as a result of the action ofFN3K in intact cells. By contrast, the amount offructosamines bound to Lys-a-61, Val-b-1, Lys-b -59 and Lys-b -66 was unaffected by thepresence of the FN3K inhibitor, indicating apoor clearing of these residues in intacterythrocytes. Accessibility to FN3K seems to bethe major factor driving the potential deglycationof lysine residues in hemoglobin.

Fructosamine-3-kinase related protein

We have also cloned human and mouse cDNAsencoding proteins sharing 65 % sequenceidentity with FN3K [3]. The gene encodingfructosamine-3-kinase related protein (FN3K-RP) is present next to the FN3K gene on humanchromosome 17q25, and they both have asimilar 6-exon structure. Northern blots indicatehigh level of expression of both genes in bonemarrow, brain, kidney and spleen. HumanFN3K-RP was transfected in HEK cells and theexpressed protein was partially purified bychromatography on Blue-Sepharose. UnlikeFN3K, FN3K-RP did not phosphorylatefructosamines (fructoselysine, deoxymorpholino-fructose, lysozyme glycated with glucose). BothFN3K-RP and FN3K were found, however, tophosphorylate ketoamines with a Dconfigurat ion in C3 (ps icoselys ine ,deoxymorpholinopsicose, deoxymorpholino-ribulose, lysozyme glycated with allose or with

10

Figure 2. Role of FN3K and FN3K-RP in protein deglycation. FN3K-RP phosphorylates psicosamines andribulosamines whereas FN3K also phosphorylates fructosamines. Psicosamines may in principle arise from thecondensation of amines with allose (not shown), but, since this sugar is absent from mammalian cells, a more likelysource is the epimerisation of fructosamines. Ribulosamines may arise from a condensation of amines with ribose orderivatives thereof such as ribose 5-phosphate.

ribose). Tandem mass spectrometry and NMRanalysis of the product of phosphorylationof.deoxymorpholinopsicose by FN3K-RPindicated that FN3K-RP also phosphorylates thethird carbon of the sugar moiety. These resultsindicate that FN3K-RP is a ketoamine-3-kinase,which presumably plays a role in the removal ofribulosamines and psicosamines from proteins.This role is shared by FN3K, which has, inaddition, the unique capacity to phosphorylatefructosamines.

To investigate the possibility that FN3K-RPalso participates in protein deglycation [4], weused human erythrocytes. First, we verified thatthese cells also contain FN3K-RP, and foundindeed that this enzyme is about three times asactive as FN3K in these cells when thephosphorylation of psicosamines is assayed. Weverified that deoxymorpholinopsicose (asubstrate and competitive inhibitor of FN3K-RP) penetrated intact erythrocytes. We alsodetermined that protein-bound psicosamine 3-phosphates and ribulosamine 3-phosphates wereunstable, decomposing at pH 7.1 and 37°C withhalf-lives of 8.8 h and 25 min, respectively, ascompared to 7.0 h for fructosamine 3-phosphates. The proof that FN3K-RP could act

as a deglycating enzyme could therefore beestablished by incubating erythrocytes with 50mM allose or 10 mM ribose for 24 h in theabsence or in the presence ofdeoxymorpholinopsicose. The presence of thelatter caused an about 1.9-2.6-fold higheraccumulation of glycated hemoglobin, an effectthat was not mimicked by deoxymorpholino-fructose, a specific inhibitor of FN3K.Furthermore, incubation with 50 mM allose alsocaused the accumulation of ketoamine 3-phosphates, which was inhibited bydeoxymorpholinopsicose but not bydeoxymorpholinofructose. These data indicatedthat FN3K-RP can phosphorylate intracellular,protein-bound psicosamines and ribulosamines,thus leading to deglycation. Our presenthypothesis is that the physiological substrates ofFN3K-RP are ribulosamines and that these arisefrom the reaction of proteins with the potentglycating agent ribose 5-phosphate.

Fructosamine metabolism inbacteria

E. Wiame, E. Van Schaftingen

Until recently, the only enzymes known to beinvolved in the metabolism of low-molecular-

11

weight fructosamines were oxidases that severone of the bonds between the aglycone andsugar portions of these compounds. Wereported recently an entirely differentmetabolism in E. coli [5]. This bacterium, whichwas found to grow on fructose-e-lysine,phosphorylates this fructosamine on the sixthcarbon of its fructose moiety thanks to a‘fructoselysine 6-kinase’. Fructoselysine 6-phosphate is then hydrolysed to lysine and theglycolytic intermediate glucose 6-phosphate by a‘deglycase’ (Fig. 3). Both enzymes are encodedby the same operon, now termed Frl operon,which comprises four open-reading-frames:FrlA, encoding a transporter related to cationicamino acid transporters; F r l B, encodingfructoselysine 6-phosphate deglycase; FrlC, anepimerase (see below); and FrlD fructoselysine6-kinase. Free fructoselysine is most probablyproduced through proteolysis of glycatedproteins. It is apparently not absorbed by thegut, but degraded in the intestine by themicrobial flora. Our data indicate that E. coli isone of the bacteria that participates in thismetabolism. In addition to this, fructoselysine 6-kinase and fructoselysine 6-phosphate deglycaseare useful tools to assay fructoselysine.

Figure 3. Metabolism of fructoselysine andpsicoselysine in Escherichia coli.

BLAST searches indicated that FrlC (see above)shares ≈ 20 % identity with tagatose 3-epimerase, an enzyme that isomerizes D-tagatoseto D-sorbose, as well as D-fructose to D-psicose. This finding suggested that FrlC couldcatalyse the interconversion of fructoselysinewith its C3 epimer, psicoselysine. FrlC wasoverexpressed and found indeed to catalyse thisreaction. The enzyme can be easily assayedthrough the formation of tritiated water from [3-3H]fructoselysine, which indicates that thereaction proceeds through an enediolintermediate. Psicoselysine was also found tosupport the growth of E. coli, causing the

induction of the three enzymes of the Frl operon[6]. The existence of a specific enzyme for themetabolism of psicoselysine suggests theoccurrence of this unusual Amadori compoundin nature. It most likely originates from thecondensation of the rare sugar allose with lysineor from the epimerisation of fructoselysineresidues during food processing.

Regulation of Vitamin C synthesis

C. Linster, E. Van Schaftingen

Vitamin C synthesis in animals proceeds fromglucuronate, which is first reduced to L-gulonate. The latter is lactonized and oxidized byL-gulonolactone oxidase, an enzyme that isdeficient in man. Vitamin C formation inanimals (and, in man, the formation of thepentose L-xylulose, also derived fromglucuronate) is known to be enhanced by a seriesof xenobiotics, including aminopyrine andchloretone. The mechanism and thephysiological significance of this effect areunknown.

Figure 4. Formation of L-ascorbate and L-xylulose from UDP-glucuronate in liver. Theformation of glucuronate from UDP-glucuronate isstimulated by aminopyrine and other agents. Theprecise mechanism of this formation is still unknown.

Using a simple enzymatic assay ofglucuronate that we have developed [7], werecently investigated the mechanism by whichaminopyrine and other agents stimulate theformation of vitamin C in isolated rathepatocytes [8]. We found that aminopyrine andseveral other agents (antipyrine, chloretone,clotrimazole, metyrapone, proadifen, barbital)induced in a few minutes an up to 8-foldincrease in the formation of glucuronate, whichwas best observed in the presence of sorbinil, aninhibitor of glucuronate reductase. They alsocaused an about 2-fold decrease in the

12

concentration of UDP-glucuronate, but little ifany change in the concentration of UDP-glucose. Depletion of UDP-glucuronate withgalactosamine or resorcinol markedly decreasedthe formation of glucuronate both in thepresence and in the absence of aminopyrine,confirming the precursor-product relationshipbetween UDP-glucuronate and free glucuronate.Most of the investigated agents did not inducethe formation of detectable amounts ofglucuronides, indicating that the formation ofglucuronate is not due to a glucuronidation-deglucuronidation cycle. To the exception ofbarbital (which inhibits glucuronate reductase),all of the above-mentioned agents caused also anincrease in the concentration of ascorbic acid inthe absence of sorbinil. They had little effect onglutathione concentration and their effect onglucuronate and vitamin C formation was notmimicked by glutathione-depleting agents suchas diamide and buthionine sulfoximide. It isconcluded that the stimulation of vitamin Csynthesis exerted by some xenobiotics ismediated through a rapid increase in theconversion of UDP-glucuronate to glucuronate,which does not apparently involve aglucuronidation-deglucuronidation cycle. Ourpresent aim is to identify the enzyme responsiblefor the formation of D-glucuronate from UDP-glucuronate.

Metabolism of L- and D-2-hydroxyglutarate

Y. Achouri, R. Rzem, G.Noël, M. Veiga-da-Cunha,E. Van Schaftingen in collaboration with D.Vertommen and M. Rider, Horm unit

D- and L-2-hydroxyglutaric acidurias are distinctneurometabolic diseases characterized by theaccumulation of abnormal amounts of either D-or L-2-hydroxyglutarate in cerebrospinal fluid,blood and urine. The biochemical lesionsresponsible for these disorders are not identified,largely due to the fact that the enzymesresponsible for the utilization of these two a-hydroxyacids are not well characterized and thattheir molecular identity is unknown.

In order to characterize enzymes involved inthe metabolism of these compounds, we haveprepared DL-2-hydroxy[2-3H]glutarate, whichwas likely to be a convenient substrate for thedevelopment of sensitive enzymatic assays.Extracts of frozen rat liver catalysed the

formation of tritiated water from DL-2-hydroxy[2-3H]glutarate. Chromatography of aliver extract on DEAE-Sepharose separated twomajor peaks of enzyme activities. The firstcorresponded to an enzyme acting on L-2-hydroxyglutarate and the second to an enzymeacting on D-2-hydroxyglutarate, as indicated bycompetitive inhibition of the detritiation of theracemic radioactive compound by the unlabelledL- and D-isomers, respectively. The enzymeacting on the D-form was further characterized.It was independent of NAD or NADP andconverted D-2-hydroxyglutarate to a -ketoglutarate, transferring electrons to artificialelectron acceptors. It also oxidized D-lactate, D-malate, and mesotartrate, and was stimulated byZn2+, Co2+ and Mn2+, but not by Mg2+ or Ca2+.Subcellular fractionation indicated that it waspresent in the mitochondrial fraction. Theenzyme was further purified by chromatographyon Blue Trisacryl and Phenyl-Sepharose, up to astage where only a few bands were still visible bySDS-PAGE. Four candidate polypeptides wereanalysed by trypsin digestion and massspectrometry. One of them corresponded to apredicted mitochondrial protein homologous toFAD-dependent D-lactate dehydrogenase. Thecorresponding human protein was expressed inHEK 293 cells and shown to catalyse thedetritiation of DL-2-hydroxy[2-3H]glutarate withsimilar properties as the purified rat enzyme. Inconclusion, we have identified the enzyme that isresponsible for the metabolism of D-2-hydroxyglutarate in mammalian tissues. Thisenzyme is most likely deficient in D-2-hydroxyglutaric aciduria [9].

Inborn errors of metabolism

T. de Barsy, J.F. Collet, G. Noël, M Veiga da Cunha,E. Van Schaftingen

In 2003, samples from about 100 patients wereanalysed, allowing the diagnosis of carnitinepalmitoyl transferase deficiency (3 cases), ofaldolase B deficiency (3 cases), of various formsof glycogen storage disease (9 cases) and ofphosphomannomutase deficiency (7 cases). Wehave also identified the mutations in thephosphoserine phosphatase gene resulting indeficiency of this enzyme in the unique case ofphosphoserine phosphatase deficiency identifieduntil now [10].

13

Selected publications

1. Delpierre G, Rider MH, Collard F, StroobantV, Vanstapel F, Santos H, Van Schaftingen E.Identification, cloning, and heterologous expression of amammalian fructosamine-3-kinase. Diabetes 2000;49:1627-34.

2. Delpierre G, Vertommen D, Communi D,Rider MH, Van Schaftingen E. Identification offructosamine residues deglycated by fructosamine-3-kinasein human hemoglobin. J Biol Chem 2 0 0 4 ;79:27613-20.

3. Collard F, Delpierre G, Stroobant V, MatthijsG, Van Schaftingen E. A mammalian proteinhomologous to fructosamine-3-kinase is a ketosamine-3-kinase acting on psicosamines and ribulosamines but noton fructosamines. Diabetes 2003;52:2888-95.

4. Collard F, Wiame E, Bergans N, Fortpied J,Vertommen D, Vanstapel F, Delpierre G, VanSchaftingen E. Fructosamine 3-kinase-related proteinand deglycation in human erythrocytes. Biochem J2004;382:1-7.

5. Wiame E, Delpierre G, Collard F, VanSchaftingen E. (2002) Identification of a pathway forthe utilization of the Amadori product fructoselysine inEscherichia coli. J Biol Chem 2002;277:42523-9.

6. Wiame E, Van Schaftingen E. Fructoselysine 3-epimerase, an enzyme involved in the metabolism of theunusual Amadori compound psicoselysine inEscherichia coli. Biochem J 2004;378:1047-52.

7. Linster C, Van Schaftingen E. Rapid stimulationof free glucuronate formation by non-glucuronidablexenobiotics in isolated rat hepatocytes. J Biol Chem2003;278:36328-33.

8. Linster C, Van Schaftingen E. Aspectrophotometric assay of D-glucuronate based onEscherichia coli uronate isomerase and mannonatedehydrogenase. Prot Exp Purif 2004, in press.

9. Achouri Y, Noël G, Vertommen D, RiderMH, Veiga-da-Cunha M, Van Schaftingen E.Identification of a dehydrogenase acting on D-2-hydroxyglutarate. Biochem J 2004; 381:35-42.

10. Veiga-da-Cunha M, Collet JF, Prieur B,Jaeken J, Peeraer Y, Rabijns A, Van SchaftingenE. Mutations responsible for 3-phosphoserinephosphatase deficiency. Eur J Hum Genet 2004;12:163-6.

14

PURINE METABOLISM

Françoise BONTEMPS, Associate MemberMarie-Françoise VINCENT, Associate MemberGeorges VAN DEN BERGHE, Emeritus

Sandrine MARIE, Postdoctoral FellowEric VAN DEN NESTE, Postdoctoral FellowSabine CARDOEN, Graduate StudentValérie RACE, Graduate StudentCaroline SMAL, Graduate StudentAnne DELACAUW, TechnicianThérèse TIMMERMAN, TechnicianIlse SCHMIDT, SecretaryKarim ACHERKI, Technical Staff

Purine metabolism is essential to the body: it provides components of the nucleic acids, DNA and RNA, and theenergy currency of the cell, ATP. Purine catabolism leads to the formation of a poorly soluble compound, uric acid,which can precipitate when elevated, and thereby causes gout. Our major present interests are the genetic defects ofpurine metabolism, and the mechanisms of action of select synthetic purine nucleoside analogues which possesstherapeutic, mainly anticancer and antiviral properties.

Adenylosuccinate lyase deficiency

M.-F. Vincent, S. Marie, V. Race, T. Timmerman,

Collaboration with the Department ofPaediatrics of the University HospitalGasthuisberg in Leuven has led us to thediscovery, in 1984, of adenylosuccinate lyase(adenylosuccinase, ADSL) deficiency, the firstenzyme deficiency described on the 'de novo'pathway of purine synthesis in man. Thisdisorder causes accumulation in cerebrospinalfluid and urine of two normally undetectablecompounds,succinylaminoimidazolecarboxamide riboside(SAICA-riboside) and succinyladenosine (S-Ado). These are the dephosphorylatedderivatives of the two substrates of ADSL,SAICA-ribotide (SAICAR) and adenylosuccinate(S-AMP), respectively (see Figure 1). Affectedchildren display variable, but mostly profoundpsychomotor delay, often epilepsy and/orautistic features, occasionally growth retardationand muscular wasting (1). We study themutations that lead to ADSL deficiency (2-4),and the pathophysiologic mechanisms of thedisorder.

Mutation analysis

ADSL deficiency has been diagnosed in morethan 60 patients worldwide. In accordance withthe variability of the clinical picture, 40 differentmutations in the ADSL gene have beenidentified to date in 38 unrelated families (seeAdenylosuccinate Lyase Mutations Database athttp://www.icp.ucl.ac.be/adsldb/). The majorityare missense mutations. In about half of thefamilies, the patients are compoundheterozygotes. Most frequently encountered,accounting for about one third of the allelesinvestigated, is a R426H mutation, which hasbeen found in 13 families.

Pathogenetic studies

The symptoms of ADSL deficiency could apriori be due to a distal deficiency of purine,particularly adenine nucleotides, and/or to atoxic effect of proximally accumulating SAICARand S-AMP, and/or of their dephosphorylatedderivatives, SAICA-riboside and S-Ado,respectively.

15

Figure 1. Pathways of purine metabolism. The ten-step synthetic route, often termed 'de novo' pathway,leads from phosphoribosyl pyrophosphate (PRPP) toIMP. From IMP, the nucleoside monophosphates,AMP and GMP, and the corresponding di- andtriphosphates (not shown) are formed. The catabolicpathway starts from the nucleoside monophosphatesand, in man, produces uric acid, a poorly solublecompound. In lower mammals, uricase (not shown)converts uric acid into allantoin, which is much moresoluble. The salvage pathway, composed of twoe n z y m e s , h y p o x a n t h i n e - g u a n i n ephosphor ibosy l t r ans fe rase and aden inephosphoribosyltransferase, converts the purine bases,guanine, hypoxanthine and adenine, into thecorresponding nucleoside monophosphates.Adenosine kinase can also be considered a salvageenzyme. AICAR, aminoimidazolecarboxamideribotide; Fum, fumarate, S-Ado, succinyladenosine;SAICAR, succinylaminoimidazolecarboxamider ibot ide ; S-AMP, adenylosuccinate . (1)adenylosuccinate lyase; (2) cytosolic 5'-nucleotidase;(3) adenylosuccinate synthetase; (4) AMP deaminase;(5) adenosine deaminase; (6) adenosine kinase; (7)hypoxanthine-guanine phosphoribosyltransferase; (8)adenine phosphoribosyltransferase. Bars indicate thedefect in adenylosuccinate lyase deficiency.

The observation that the levels of SAICA-riboside are comparable in severely and mildlyretarded patients, whereas those of S-Ado aremarkedly higher in the latter, has led to thehypothesis that SAICA-riboside is theneurotoxic compound, and that S-Ado couldcounteract its noxious effects. However, wecould not demonstrate a cytoxic effect of thesuccinylpurines on cultured rat neurones.

These findings have led us to consider thepossibility that the intracellular accumulation ofthe substrates of the enzyme might be toxic. Totest this hypothesis, construction of ADSL-deficient models was initiated. Severalapproaches are under way to obtain ADSL-deficient cells: creation of dominant negativemutants by overexpression of mutated ADSL inmammalian cells, expression of antisense RNAsto repress endogenous ADSL, and inhibition ofADSL by compounds such a sadenylophosphonopropionate. We are also inthe process of creating an ADSL-deficientneuronal cell line by knock-in through insertionof the most frequent mutation, R426H, in themurine gene.

Our studies were also devoted to themechanisms involved in the dephosphorylationof the two substrates of ADSL, giving newinsight in the pathophysiology of the neurologicsymptoms of the deficiency.

A new defect in the purinebiosynthesis

S. Marie, T. Timmerman and MF. Vincent

In a female infant with dysmorphic features,severe neurological defects, and congenitalblindness, a positive urinary Bratton-Marshalltest led to identification of a massive excretionof 5-amino-4-imidazolecarboxamide (AICA)-riboside, the dephosphorylated counterpart ofAICAR (also termed "ZMP"), an intermediate ofde novo purine biosynthesis. ZMP and its di-and triphosphate accumulated in the patient'serythrocytes. Incubation of her fibroblasts withAICA-riboside led to accumulation of AICAR,not observed in control cells, suggestingimpairment of the final steps of purinebiosynthesis, catalyzed by the bifunctionalenzyme AICAR transformylase/IMPcyclohydrolase (ATIC). AICAR transformylase

16

was profoundly deficient, whereas the IMPcyclohydrolase level was 40% of normal.Sequencing of ATIC showed a K426R change inthe transformylase region in one allele and aframeshift in the other. Recombinant proteincarrying mutation K426R completely lacksAICAR transformylase activity (5).

Antileukaemic properties of2-chloro-2'-deoxyadenosine

F. Bontemps, S. Cardoen, A. Delacauw, C. Smal, E.Van!Den Neste, in collaboration with A. Ferrant,Department of Haematology, University Hospital Saint-Luc

In 1997, a collaborative study of theantileukaemic nucleoside, 2-chloro-2'-deoxy-adenosine (CdA), was started with theDepartment of Haematology of the UniversityHospital Saint-Luc. This adenosine deaminase-resistant deoxyadenosine analogue displaysremarkable therapeutic properties in indolentlymphoid malignancies including hairy cellleukaemia and B-cell chronic lymphocyticleukaemia (B-CLL). Nevertheless, resistance isalso observed, and CdA does not confer asurvival advantage when compared to moreconventional therapies such as alkylating agents.The aims of the project are to understand themechanisms that lead to resistance to CdA, andto improve its therapeutic efficacy by searchingfor synergisms with other compounds.

To exert is antileukaemic effect, CdA has tobe phosphorylated by deoxycytidine kinase(dCK) into CdAMP, followed by conversioninto CdADP and CdATP. The latter, the activemetabolite of CdA, has been shown to inhibit avariety of enzymes involved in DNA synthesis,including ribonucleotide reductase and DNApolymerase a . Moreover, CdATP can beincorporated into newly synthesised DNA,causing chain termination. Together, theseactions result in arrest of DNA synthesis and inthe progressive accumulation of DNA strandbreaks, leading to apoptosis by mechanismswhich are not yet entirely clear. 2-Chloroadenine, the major catabolite of CdA, wasfound to be actively phosphorylated, but poorlycytotoxic (6).

Effects of CdA in EHEB cells

To improve our understanding of themechanisms by which CdA induces apoptosis in

B-CLL cells, we investigate EHEB cells, acontinous cell line derived from a patient with B-CLL. The EHEB cell line was found to be lesssensitive (10- to 1000-fold) to the nucleosideanalogue CdA than other human lymphoblasticcell lines. This can be explained by a lowerintracellular accumulation of CdATP, thecytotoxic metabolite of CdA, due to a reduceddCK activity. Unexpectedly, DNA synthesis,measured by thymidine incorporation intoDNA, was increased in EHEB cells, up to 2-fold, after a 24 h-incubation with CdA atconcentrations close to the IC50 (5 – 10 µM) (7).Analysis by flow cytometry, using doublelabelling with propidium iodide andbromodeoxyuridine, has shown that CdA, inEHEB cells, provokes an increase in theproportion of cells in S phase, synthesisingactively DNA. These results contrast with thosereported in other leukaemic cell lines, likeCCRF-CEM cells, in which CdA inhibits DNAsynthesis and provokes an accumulation of mostcells in either early S phase or at the G1-Sborder. Kinetics and synchronisationexperiments have shown that 10 µM CdAstimulates the progression of EHEB cells fromG1 to S phase, rather than blocking them in Sphase. This led us to study the effect of CdA onproteins regulating the G1/S checkpoint of thecell cycle, and firstly on the phosphorylation ofthe retinoblastoma (Rb) protein, which isincreased during the G1/S transition. We haveobserved that CdA enhances thephosphorylation of Rb in EHEB cells.Additional preliminary experiments have shownthat CdA also increases the activity of the cyclin-dependent kinase 2 (cdk2), a kinase thatphosphorylates Rb and plays a crucial role in theprogression of cells from G1 to S phase. Theexpression of p21, a key inhibitor of cdk2, tendsto decrease after CdA, despite an increase of thelevel of p53. The p53 status of this cell line wasdetermined and found unmutated. Inconclusion, we show a new mode of cellularresponse to CdA, implying modification ofcomponents involved in cell cycle regulation. Weare currently studying if the activation of cellcycle by CdA in EHEB cells is correlated withthe onset of apoptosis.

Regulation of dCK activity

Since dCK activates numerous nucleosideanalogues used in anticancer and antiviraltherapy, knowledge of its regulation can beexpected to allow optimization of the activation

17

of these analogues. Recently, it has been shownby others and by us that dCK activity can beincreased by various genotoxic agents, includingCdA, aphidicolin, etoposide, and UV-Cirradiation (8). This activation is not explained byan allosteric effect or by an increase of thequantity of enzyme. A post-translationalactivation of dCK by intracellular signallingpathways was suggested. To unravel themechanism of the activation of dCK by CdA, wefirst investigated the effect of a variety ofactivators and inhibitors of protein kinases onthe basal activity of dCK and on its activation byCdA. We discovered that dCK can be activatedby several inhibitors of protein kinases, includinggenistein, an unspecific inhibitor of proteintyrosine kinases, AG-490, an inhibitor of theprotein tyrosine kinase JAK-2 and JAK-3, andPD-98059 and U0126, two specific inhibitors ofthe MAPK/ERK pathway. We also observedthat these inhibitors potentiated the activatingeffect of CdA. On the other hand, we haveshown that dCK activity can be markedlyincreased in intact EHEB cells by incubationwith okadaic acid, an inhibitor of proteinphosphatase PP2A. Taken together, these resultsdo not allow to identify the pathway by whichdCK is activated, but clearly indicate that itsactivity can be regulated by protein kinase(s) andphosphatase(s). This was confirmed by ourobservation that dCK, activated or not by CdA,can be inactivated in a crude cell extract bypurified PP2A. This last result also demonstratesthat activation of dCK results from itsphosphorylation (9).

To further unravel its regulation, dCK wasoverexpressed in HEK-293 cells as a His-tagfusion protein. Western blot analysis showedthat purified overexpressed dCK appear asdoublet protein bands. The slower banddisappeared after treatment with proteinphosphatase lambda in parallel with a decreaseof dCK activity, providing additional argumentsin favor of both phosphorylated andunphosphorylated forms of dCK. We plan nowto verify if dCK can be effectively labelled with[32P]orthophosphate and if this labelling can bemodulated by effectors of dCK.

Search for potentiation of antileukaemiceffect of CdA

In recent years, we have shown that combinationof CdA with DNA-damaging agents, such ascyclophosphamide (CP) derivatives (10) or UV-

light (11), resulted in synergistic cytotoxicity inB-CLL lymphocytes. The in vitro synergybetween CdA and CP derivatives have providedthe rationale for a clinical trial of thiscombination, which gives encouraging results(12).

We are currently analysing if efficacy of CdAcould be strengthened by combination withMAP kinase inhibitors. MAP kinases indeedregulate important cellular processes, includingcell proliferation and survival, and programmedcell death, which explains that inhibitors of theseenzymes are being explored as anticancer agents.

Drug sensitivity profiles

With standard treatments of B-CLL, clinicalresponse is variable, depending on drug andpatient. To tailor therapy on a more individualbasis, we develop an in vitro assay whichevaluates the sensitivity of the patients’lymphocytes to various drugs (nucleosideanalogues, alkylating agents, steroids,anthracyclin, etc.). The lymphocytes will also becharacterised by conventional cytogenetics,hybridisation in situ (FISH), and molecularbiology techniques. By revealing correlationsbetween chemoresistance profiles and geneticanomalies, these studies might also allowidentification of mechanisms of resistance.


1. Van den Berghe G, Jaeken J. Adenylosuccinatelyase deficiency, in The Metabolic and MolecularBases of Inherited Disease (Scriver CL,Beaudet AL, Sly WS, Valle D, eds) 8 th ed,McGraw-Hill, New York. 2001; pp. 2653-62.

2. Marie S, Cuppens H, Heuterspreute M,Jaspers M, Zambrano Tola E, Gu XX, Legius E,Vincent MF, Jaeken J, Cassiman JJ, Van denBerghe G. Mutation analysis in adenylosuccinate lyasedeficiency!: eight novel mutations in the re-evaluated fullADSL coding sequence. Hum Mut 1999;13:197-202.

3. Race V, Marie S, Vincent MF, Van denBerghe G. Clinical, biochemical and molecular geneticcorrelations in adenylosuccinate lyase deficiency. HumMolec Genet 2000;9:2159-65.

4. Marie S, Race V, Nassogne MC, Vincent MF,Van den Berghe G. Mutation of a nuclear respiratory

18

factor 2 binding site in the 5’untranslated region of theADSL gene in three patients with adenylosuccinate lyasedeficiency. Am J Hum Genet 2002;71:14-21.

5. Marie S, Heron B, Bitoun P, Timmerman T,Van Den Berghe G, Vincent MF. A I C A -ribosiduria: a novel, neurologically devastating inbornerror of purine biosynthesis caused by mutation of ATIC.Am J Hum Genet 2004;74:1276-81.

6. Bontemps F, Delacauw A, Cardoen S, VanDen Neste E, Van den Berghe G. Metabolism andcytotoxic effects of 2-chloroadenine, the major catabolite of2-chloro-2’-deoxyadenosine. Biochem Pharmacol2000;59:1237-43.

7. Cardoen S, Van Den Neste E, Smal C, RosierJF, Delacauw A, Ferrant A, Van den Berghe G,Bontemps F. Resistance to 2-chloro-2’-deoxyadenosineof the human B-cell leukemia cell line EHEB. ClinCancer Res 2001;7:3559-66.

8. Van Den Neste E, Smal C, Cardoen S,Delacauw A, Frankard J, Ferrant A, Van denBerghe G, Bontemps F. Activation of deoxycytidinekinase by UV-C-irradiation in chronic lymphocyticleukemia B-lymphocytes. Biochem Pharmacol2003;65: 573-80.

9. Smal C, Cardoen S, Bertrand L, Delacauw A,Ferrant A, Van den Berghe G, Van Den NesteE, Bontemps F. Activation of deoxycytidine kinase byprotein kinase inhibitors and okadaic acid in leukemiccells. Biochem Pharmacol 2004;68:95-103.

10. Van Den Neste E, Bontemps F, DelacauwA, Cardoen S, Louviaux I, Scheiff JM, Gillis E,Leveugle P, Deneys V, Ferrant A, Van denBerghe G. Potentiation of antitumor effects ofcyclophosphamide derivatives in B-chronic lymphocyticleukemia cells by 2-chloro-2’-deoxyadenosine.Leukemia 1999;13:918-25.

11. Van Den Neste E, Cardoen S, Husson B,Rosier JF, Delacauw A, Ferrant A, Van denBerghe G, Bontemps F. 2-Chloro-2’-deoxyadenosineinhibits DNA repair synthesis and potentiates UVCcytotoxicity in chronic lymphocytic leukemia Blymphocytes. Leukemia 2002;16: 36-43.

12. Van Den Neste E, Louviaux I, Michaux JL,Delannoy A, Michaux L, Sonet A, Bosly A,Doyen C, Mineur P, André M, Straetmans N,Coche E, Venet C, Duprez T, Ferrant A. PhaseI/II study of 2-chloro-2'-deoxyadenosine withcyclophosphamide in patients with pretreated B cell chroniclymphocytic leukemia and indolent non-Hodgkin'slymphoma. Leukemia 2000;14:1136-1142.

19

TISSUE-SPECIFIC TRANSCRIPTION FACTORS INDEVELOPMENT

Guy G. ROUSSEAU, MemberFrédéric P. LEMAIGRE, Member

Patrick JACQUEMIN, Associate MemberFrédéric CLOTMAN, Assistant MemberChristophe PIERREUX, Assistant MemberJean-Bernard BEAUDRY, Graduate StudentSabrina MARGAGLIOTTI, Graduate StudentNicolas PLUMB, Graduate StudentAurélie POLL, Graduate StudentJonathan van EYLL, Graduate StudentVinciane VANHORENBEECK, Graduate StudentSabine CORDI, TechnicianJean-François CORNUT, TechnicianMarie-Agnès GUEUNING, Technician (half-time)Vivien O'CONNOR, Secretary (half-time)

We are studying the role of the Onecut (OC) transcription factors in tissue-specific gene expression and indevelopment. These factors, discovered in the laboratory, define a new class (1) of conserved homeoproteins, withthree members in mammals : HNF-6 or OC-1, OC-2 and OC-3. They display a restricted tissue distributionand the classes of genes that they regulate overlap, but do not superimpose. HNF-6 (Hepatocyte Nuclear Factor-6), the OC prototype, was identified (2) as a factor that controls the liver-specific transcription ofPFK2/FBPase2. HNF-6 stimulates transcription by at least two different mechanisms depending on thenucleotide sequence of the DNA binding site (3). By interacting with DNA-bound glucocorticoid receptor, HNF-6 can also inhibit glucocorticoid-induced gene transcription in a target-specific way. In adult liver, HNF-6stimulates the transcription of genes coding for enzymes of glucose metabolism, for P450 cytochromes and forsecreted proteins. In the embryo, HNF-6 is expressed in the liver and pancreas where it regulates genes coding forother transcription factors. The role of HNF-6 in development has been addressed by studying Hnf6 knockoutmice generated in our laboratory. The aim of our current work is to determine the mechanisms by which the OCtranscription factors control development of the pancreas, liver and gastrointestinal tract, and to investigate the roleof OC factors in diseases of these organs.

Figure 1. Stucture of human Onecut proteins. The DNA-binding domain consists of a cut domain and ahomeodomain. The cut domain contains a LXXLL motif (LSDLL) that contacts transcriptional co-activators. Thehomeodomain diverges from other homeodomains by the presence of a phenylalanine at position 48 and amethionine at position 50. These two amino acids are important for recruitment of co-activators. The TP, His, Glyand Ser domains are rich in Threonine/Proline, Histidine, Glycine and Serine, respectively. The TP domain isinvolved in transcriptional activation, the His and Ser domains in transcriptional repression.

20

Control and molecular mode ofaction of HNF-6

N. Plumb, A. Poll, and J.-B. Beaudry

OC proteins contain a divergent homeodomainin which residue 48 is a Phe instead of thecanonical Trp and residue 50 is a Met, an aminoacid never found at this conserved position inclassical homeodomains. They also contain asingle cut domain, as opposed to the Drosophilacut protein and its mammalian homologues,which have a classical homeodomain and threecut domains. In addition to the homeo and cutdomains, which bind DNA on the sequenceDRTCMATND, OC proteins contain otherOC-specific conserved regions that modulatetheir activity (Fig. 1).The Hnf6 gene is made ofthree exons separated by more than 10 kb. Thecut domain is coded by exon 1, and thehomeodomain by exon 3. Exon 2 (78 bp) codesfor a spacer between these two DNA-bindingdomains. Differential splicing yields twoisoforms with distinct DNA-binding properties,HNF-6a (no spacer) and HNF-6b (with spacer)(1). The study of Hnf6 gene regulatorysequences suggest that the gene contains morethan one promoter. We showed that the liverpromoter is a target of STAT5 and in this wayHNF-6 plays a role in the effects of growthhormone in the liver. Our current objective is toidentify the promoter and regulatory sequencesthat are active in the pancreas and liver, and thetranscription factors that bind to them. Theseregulatory sequences have been cloned upstreamof the b -gal reporter gene and their tissue-specific activity is being tested in transgenic miceand in the endoderm of developing embryos.

By stimulating transcription of the Hnf1b,Hnf4 and Hnf3b genes and by being itselfcontrolled by C/EBPa, HNF-6 participates tothe network of liver transcription factors. We arefurther studying this network, as we haveevidence that HNF-6 inhibits the stimulation ofthe Hnf3a gene by TGFb.

Stimulation of gene transcription by HNF-6involves co-activators and depends on theintegrity of the N-terminal region of HNF-6 (3).We are characterising the transcriptional partnersof HNF-6 that are recruited by this region inview of investigating their role in the mechanismof action of HNF-6.

Control of pancreas development byHNF-6

P. Jacquemin, C. Pierreux, J. van Eyll

Hnf6 knockout mice have an hypoplasticpancreas devoid of islets of Langerhans (Fig. 2).We have shown that HNF-6 is required for thedifferentiation of pluripotent prepancreatic cellsinto endocrine precursors. This involves thestimulation by HNF-6 of the proendocrinetranscription factor NGN-3 (4). At birth, theknockout mice have too few pancreaticendocrine cells and they become diabetic. Wehave now found that this glucose intolerancealso results from a defective liver expression ofthe glucokinase gene, which is a target of HNF-6. Type II diabetic patients were screened formutations in the Hnf6 gene, but none has beenidentified so far.

Embryonic stem cells, when grown asembryoid bodies, spontaneously generateinsulin-producing cells that could be used intherapy of diabetes mellitus. Does differentiationof cells in embryoid bodies mimic that ofpancreatic b cells in embryos!? To address thisquestion we verified if the differentiation of theinsulin-producing cells in embryoid bodiesrequires HNF-6. No difference was observed inthe expression of insulin between wild-type andHnf6-/- embryoid bodies. In both cases insulinwas expressed in the outer layer of cells, which issimilar to the visceral endoderm. In wild-typeembryoid bodies HNF-6 was transientlyexpressed in the outer layer of cells, but was notco-expressed with insulin. The expression ofgenes that are targets of HNF-6 in developingpancreas was unaffected in Hnf6-/- embryoidbodies. Thus, the differentiation mechanism ofinsulin-producing cells in embryoid bodiesdiffers from that of the b cells and it is likely toresemble that of insulin-producing cells in thevisceral endoderm.

As to the pancreas hypoplasia of Hnf6knockout mice, our data point to a decrease inthe number of endodermal precursors specifiedto a pancreatic fate by the transcription factorPdx-1. Indeed, in the foetal foregut endoderm,where it is expressed, HNF-6 initiates theexpression of Pdx-1, thereby controlling thetiming of pancreas specification (5) (Fig. 3).

As Pdx-1-expressing cells eventually appearin the Hnf6-/- embryos, we are investigating therole of other genes in pancreas development. In

21

the course of these studies we have discovered,using cultured pancreatic explants, thatembryonic pancreas can be differentiated, via a

Shh-dependent mechanism, into intestinal tissueby activin A (6).

Figure 2. Abnormal development of the endocrine pancreas in HNF-6 knockout mice.Immunohistochemistry of tissue sections four days after birth shows that only a few insulin-producing cells(brown) are found near pancreatic ducts in Hnf6-/- (right panel), instead of being organized in islets as in Hnf6+/+ littermates (left panel)

Control of liver development byHNF-6

F. Clotman, S. Margagliotti, N. Plumb

During liver development, hepatoblastsdifferentiate into hepatocytes or into biliaryepithelial cells (BEC) which delineate theintrahepatic and extrahepatic bile ducts and thegallbladder. We have shown that HNF-6 isexpressed not only in hepatoblasts, but also inBEC. In Hnf6-/- mice, the gallbladder does notdevelop and the extrahepatic bile ducts areabnormal. The mice suffer from cholestasis anddisplay a phenotype that resembles humanbil iary diseases cal led "ductal platemalformations", which are related to biliaryatresia. Indeed, the differentiation andmorphogenesis of the intrahepatic bile ducts areperturbed (Fig. 4). These disorders involve thetranscription factor HNF-1b, which we showedto be a target of HNF-6 and whose expression isdownregulated in Hnf6-/- fetal liver (7).

Moreover, this primary defect in liverdevelopment was associated with anomalies ofthe hepatic artery (8), providing a model forstudying such anomalies in humans, where theyare often associated with ductal platemalformations.

We are further studying the mechanisms ofthis biliary phenotype by microarray analysis andby exploiting an ex vivo model, the immortalisedbipotential mouse embryonic liver (BMEL) cells,which can differentiate into either BEC orhepatocytes.

The developing liver harbours haematopoieticprecursors which differentiate into the differentblood cell lineages in response to signals sent byhepatoblasts. We have discovered a severe Blymphopenia in young Hnf6-/- mice and haveshown that this results from a liver defect,thereby identifying HNF-6 as the first non cell-intrinsic transcription factor known to control Blymphopoiesis specifically in fetal liver.

22

Figure 3. Role of HNF-6 in pancreas development. In themouse, pancreas starts developing at embryonic day (e) 8.5 inthe foregut endoderm, which becomes specified to a pancreaticfate by the expression of the transcription factor Pdx-1. Cells inthis epithelium become precursors that are committed to aductal, exocrine or endocrine fate. These precursorsdifferentiate into ducts, acini, or the four types of hormone-producing cells which form the islets of Langerhans just beforebirth. Our analysis of Hnf6-/- embryos shows that HNF-6controls the timing of pancreas specification by inducing Pdx-1in the endoderm and that, later on, HNF-6 regulates thedevelopment of the endocrine lineage by controlling thetranscription factor Ngn3.

Figure 4. Abnormal development of the intrahepatic bile ducts in HNF-6 knockout mice.Immunohistochemistry of liver sections 10 days after birth shows in normal mice (left panel) a typical bile duct (bd)delineated by biliary cells (stained red) embedded in mesenchyme. In contrast, biliary cells do not form bile ducts inHnf6-/- mice (right panel) and remain dispersed as a layer around a branch of the portal vein (pv).

Role of OC-2 and OC-3 in thedevelopment of endodermderivatives

P. Jacquemin, V. Vanhorenbeeck, C. Pierreux

We have cloned OC-2 (9) and OC-3 (10) andhave shown they have properties similar to thoseof HNF-6 in terms of DNA-binding andtranscriptional activation. In the adult mouse,the patterns of expression of these paraloguesare tissue-restricted and partially overlappingwith those of HNF-6 (10). While they are allexpressed in brain and foregut endoderm, HNF-6 is also expressed in the testis, liver andpancreas, OC-2 in the liver, gut and stomach,and OC-3 in the gut and stomach. In addition,OC-2 is expressed in skin melanocytes where itcontrols differentiation by regulating themicrophthalmia-associated transcription factor(MITF) gene. We have now generated OC-2 and

OC-3 knockout mice and are studying theirphenotype.

As the three mammalian Onecut genes areexpressed in the foregut endoderm, we aresetting up cultures of this tissue from normal orknockout mouse embryos in view of exploringex vivo the respective roles of the Onecut factorsin pancreas, liver, stomach and gut development.


1. Lannoy VJ, Burglin TR, Rousseau GG,Lemaigre FP. Isoforms of hepatocyte nuclear factor-6differ in DNA-binding properties, contain a bifunctionalhomeodomain, and define the new ONECUT class ofhomeodomain proteins. J Biol Chem 1998;273:13552-62.

2. Lemaigre FP, Durviaux SM, Truong O,Lannoy VJ, Hsuan JJ, Rousseau GG. Hepatocyte

23

nuclear factor 6, a transcription factor that contains anovel type of homeodomain and a single cut domain.Proc Natl Acad Sci USA 1996;93:9460-4.

3. Lannoy VJ, Rodolosse A, Pierreux CE,Rousseau GG, Lemaigre FP. Transcriptionalstimulation by hepatocyte nuclear factor-6. Target-specificrecruitment of either CREB-binding protein (CBP) orp300/CBP-associated factor (p/CAF). J Biol Chem2000;275:22098-103.

4. Jacquemin P, Durviaux SM, Jensen J,Godfraind C, Gradwohl G, Guillemot F,Madsen OD, Carmeliet P, Dewerchin M, CollenD, Rousseau GG, Lemaigre FP. Transcriptionfactor hepatocyte nuclear factor 6 regulates pancreaticendocrine cell differentiation and controls expression of theproendocrine gene ngn3. Mol Cell Biol 2000;20:4445-54.

5. Jacquemin P, Lemaigre FP, Rousseau GG. Theonecut transcription factor HNF-6 (OC-1) is requiredfor timely specification of the pancreas and acts upstreamof Pdx-1 in the specification cascade. Dev Biol 2003;258:105-116.

6. Van Eyll JM, Pierreux CE, Lemaigre FP,Rousseau GG. Shh-dependent differentiation of

intestinal tissue from embryonic pancreas by activin A JCell Sci 2004;117:2077-86.

7. Clotman F, Lannoy VJ, Reber M, Cereghini S,Cassiman D, Jacquemin P, Roskams T,Rousseau GG, Lemaigre FP. The onecuttranscription factor HNF6 is required for normaldevelopment of the biliary tract. Development 2002;129:1819-28.

8. Clotman F, Libbrecht L, Gresh L, Yaniv M,Roskams T, Rousseau GG, Lemaigre FP. Hepaticartery malformations associated with a primary defect inintrahepatic bile duct development. J Hepatol 2003;39:686-92.

9. Jacquemin P, Lannoy VJ, Rousseau GG,Lemaigre FP. OC-2, a novel mammalian member ofthe ONECUT class of homeodomain transcriptionfactors whose function in liver partially overlaps with thatof hepatocyte nuclear factor-6. J Biol Chem 1999;274:2665-71.

10. Vanhorenbeeck V, Jacquemin P,Lemaigre FP, Rousseau GG. OC-3, a novelmammalian member of the ONECUT class oftranscription factors. Biochem Biophys ResCommun 2002;292:848-54.

24

SIGNAL TRANSDUCTION AND PROTEINPHOSPHORYLATIONLouis HUE, MemberMark H. RIDER, Member

Isabelle DE POTTER, Assistant MemberBruno GUIGAS, Assistant MemberSandrine HORMAN, Assistant MemberDidier VERTOMMEN, Assistant MemberNicole EL-NAJJAR, Graduate StudentDaniel LEROY, Graduate StudentVéronique MOUTON, Graduate StudentMartine DE CLOEDT, Technician (half-time)Nusrat HUSSAIN, TechnicianLiliane MAISIN, Technician (part-time)Vivien O'CONNOR, Secretary (half-time)Freddy ABRASSART, Technical Staff (part-time)

Our research concerns the role of protein phosphorylation in the control of metabolism by nutrients, hormones andvarious stresses. As a model system, we started out by studying 6-phosphofructo-2-kinase (PFK-2) /fructose-2,6-bisphosphatase (FBPase-2). This bifunctional enzyme catalyzes the synthesis and degradation of fructose 2,6-bisphosphate, a potent stimulator of glycolysis. Fructose 2,6-bisphosphate was discovered in this Institute byVan!Schaftingen, Hue and Hers in 1980 and is the most potent stimulator of 6-phosphofructo-1-kinase (PFK-1), a key enzyme of glycolysis. Fructose 2,6-bisphosphate is synthesised from fructose 6-phosphate and ATP by 6-phosphofructo-2-kinase (PFK-2). Its hydrolysis to fructose 6-phosphate and Pi is catalysed by FBPase-2. Thesetwo activities are catalysed at separate sites of a bifunctional enzyme (PFK-2/FBPase-2) composed of two identicalsubunits.

We characterised several PFK-2/FBPase-2 isoforms in mammalian tissues and cloned the correspondingmRNAs, showing that they originate from at least two genes (1). The isoforms differ in PFK-2/FBPase-2activity ratio, kinetic properties and response to phosphorylation by protein kinases. The C-terminus of the heart(H) isozyme, contains phosphorylation sites for several protein kinases. These sites are not present in the otherisozymes, such as the liver (L) isozyme, which, by contrast, contains a single phosphorylation site for the cyclicAMP-dependent protein kinase (PKA) at the N-terminus. The concentration of fructose 2,6-bisphosphate changesin response to metabolites, hormones, growth factors, and oncogene activation (1).

Over recent years, we made a detailed study of the molecular mechanisms responsible for the activation of heartPFK-2 by insulin and showed that protein kinase B was probably not necessary. More recently, we studied theAMP-activated protein kinase (AMPK) which phosphorylates and activates heart PFK-2 in ischaemia, providinga new explanation for the Pasteur effect. We characterised new actions of AMPK on cellular processes. Itsactivation not only leads to the stimulation of glycolysis in heart and monocytes, but also induces apoptosis incertain cells, inhibits mTOR signalling by amino acids and inhibits protein synthesis at elongation. The latter canbe explained by the phosphorylation-induced inactivation of eukaryotic elongation factor-2 (eEF2). The target ofAMPK is not eEF2 itself but its upstream eEF2 kinase which is phosphorylated and activated by AMPK. Weare currently investigating the upstream regulation of AMPK and novel downstream targets.

25

Insulin signalling

V. Mouton, D. Vertommen, L. Hue, M.H. Rider, incollaboration with L. Bertrand, UCL, Brussels, C.Erneux and D. Bléro, ULB, Brussels, and D. Alessi,Dundee

Activation of heart PFK-2 by insulin

Insulin stimulates heart glycolysis by increasingglucose transport and by activating PFK-2. Thisin turn leads to a rise in fructose 2,6-bisphosphate. The mechanism involved in thisinsulin-induced activation of heart PFK-2 isbeing studied both in vitro and in intact cells. Therecombinant heart PFK-2 isozyme is a substrateof several protein kinases, especially proteinkinases of the insulin signalling pathways (2),such as protein kinase B (PKB), also known asAkt, which is believed to mediate mostmetabolic effects of insulin. In mice lacking 3-phosphoinositide-dependent protein kinase-1(PDK1) in cardiac muscle, insulin did notactivate PKB and PFK-2, nor did it increasefructose 2,6-bisphosphate (3), consistent withPDK1 mediating these processes. We also testedthe role of PKB in the activation of PFK-2 byinsulin using a 'dominant-negative' construct andfound that the activation of PFK-2 by insulindid not require PKB, but was mediated byanother protein kinase located downstream ofPDK1 (4). We purified a wortmannin-sensitiveand insulin-stimulated protein kinase (WISK).WISK phosphorylates heart PFK-2 mainly onSer466 leading to its activation (5). Our recentwork indicated that WISK contains proteinkinase C zeta (PKCz), which could participate inthe insulin-induced activation of heart PFK-2.

Disruption of the gene for the 5’-phosphatase acting on phosphatidylinositol-3,4,5-trisphosphate (PtdIns(3,4,5)P3) known asSHIP2, increases insulin sensitivity (6). Whenco-transfected with heart PFK-2 in CHO-IRcells, SHIP2 blunted insulin-induced PFK-2activation, emphasizing the requirement ofPtdIns(3,4,5)P3 for WISK activation.

AMP-activated protein kinase

S. Horman, D. Vertommen, B. Guigas, D. Leroy, N.El-Najjar, M.H. Rider, L. Hue, in collaboration withJ.-L. Vanoverschelde and L. Bertrand, UCL, Brussels,D. Carling and A. Woods, London, D.G. Hardie,

Dundee, P. Ferré and F. Foufelle, Paris, and T.Walliman and U. Schlattner, Zurich

The AMP-activated protein kinase (AMPK) actsas a sensor of cellular energy status. AMPK isactivated by an increase in the AMP/ATP ratioas it occurs when the oxygen supply is restrictedor after exposure of cells to inhibitors of themitochondrial respiratory chain, such asoligomycin. In certain cells, AMPK can also beactivated by 5-aminoimidazole-4-carboxamide(AICA)-riboside, which enters cells to bephosphorylated into ZMP, an analogue of AMP.AMPK switches off energy-consumingbiosynthetic pathways, thereby conserving ATP.

Stimulation of heart glycolysis by ischemia.

Ischemia or anoxia stimulates glycolysis (PasteurEffect) which involves increased glucosetransport and PFK-2 activation in heart. Weinvestigated whether AMPK could mediate thisphenomenon. AMPK phosphorylated heartPFK-2 on Ser 466 which led to its activation (7).In perfused hearts, ischemia induced anactivation of AMPK, which correlated withPFK-2 activation and with an increase in Fru-2,6-P2 concentration. In cultured HEK-293 cellstransfected with heart PFK-2, a dominantnegative construct of AMPK abolished both thephosphorylation and activation of transfectedPFK-2 induced by oligomycin, an inhibitor ofoxidative phosphorylation (7). Therefore, heartPFK-2 is a new substrate of AMPK and itsactivation is involved in the Pasteur Effect.

AMPK activation inhibits protein synthesis

Protein synthesis, in particular peptide chainelongation, consumes a large proportion ofintracellular ATP. Therefore, we investigatedwhether AMPK activation could inhibit proteinsynthesis via the phosphorylation of regulatorycomponents of the translation machinery. Inanoxic rat hepatocytes or in hepatocytes treatedwith AICA-riboside, AMPK was activated andprotein synthesis was inhibited. The inhibition ofprotein synthesis was associated with thephosphorylation of eEF2. In HEK-293 cells,transfection of a dominant negative AMPKconstruct abolished the oligomycin-inducedinhibition of protein synthesis and eEF2phosphorylation. Lastly, eEF2 kinase (eEF2K),the kinase that phosphorylates eEF2, wasactivated in anoxic or AICA riboside-treatedhepatocytes and in ischaemic hearts. Moreover,phosphorylation of recombinant eEF2K by

26

AMPK correlated with eEF2K activation,providing a novel mechanism for the inhibitionof protein synthesis (8,9).

Incubation of hepatocytes with amino acids,such as glutamine and leucine, leads to anactivation of biosynthetic pathways, such asglycogen synthesis, lipogenesis and proteinsynthesis. Under these conditions, p70ribosomal S6 kinase (p70S6K), a protein kinasethat participates in the control of proteinsynthesis and is activated in response tohormones, mitogens and nutrients, becomesactivated via activation of the mammalian targetof rapamycin (mTOR) by an unknownmechanism.

Pretreatment of hepatocytes withphosphorylating target(s) in the mTOR AICAriboside prevented the activation andphosphorylation of p70S6K (10). Therefore, it islikely that AMPK inhibits p70S6K activation bysignalling pathway.

Sustained activation of AMPK triggersapoptosis in liver cells

We studied the effect of long-term AMPKactivation on liver cell survival. AMPKactivation was maintained in FTO2B cellstreated with AICA riboside or by adenoviraltransfection of hepatocytes with constitutivelyactive AMPK. Sustained AMPK activationtriggered apoptosis through an activationpathway involving c-Jun kinase and caspase-3(11).

Figure 1. Mechanism of activation and targets of AMP-activated protein kinase (AMPK). AMPK isphosphorylated and activated by AMPKK when the AMP/ATP ratio increases as a result of metabolic stresses. Itcan also be activated by an AMP analogue, ZMP, which is formed from AICA-riboside. Insulin antagonises AMPKactivation by a mechanism which might involve phosphorylation of regulatory sites in the catalytic a-subunits byPKB. The targets of AMPK, that we discovered and which are responsible for some of its short-and long-termeffects, are indicated. AMP, adenosine monophosphate; ATP, adenosine triphosphate; AICA riboside, 5-aminoimidazole-4-carboxamide riboside; ZMP, AICA ribotide; AMPK, AMP-activated protein kinase, AMPKK,AMPK-kinase; eEF2K, eukaryotic elongation factor-2-kinase; p70S6K, p70 ribosomal protein S6 kinase; mTOR,mammalian target of rapamycin; ACC, acetyl-CoA carboxylase; PFK-2, 6-phosphofructo-2-kinase; JNK, c-junkinase.

↑AMP/ATP

LKB1 PKB (?)AMPK

Nutrients ÆStress ¨Exercise

Ø Anabolism

• Protein synthesis

Elongation (eEF2-K) Initiation (p70S6K, mTOR)

• Fatty acids synthesis(ACC)

↑ Catabolism

• Glycolysis (PFK-2)

• Glucose transport

• Fatty acids oxidation (ACC)

Short-term effects Long-term effects

?

insulin

↑ Apoptosis

(JNK)

AICAriboside

ZMP

27

AMPK activation by upstream kinases

AMPK activation requires phosphorylation inthe activation loop at Thr172 of its catalytic a-subunit by an upstream kinase, AMPK-kinase(AMPKK). An upstream kinase phosphorylatingThr172 for AMPK activation was identifiedindependently by the groups of Carling andHardie as the Peutz-Jegher’s protein, LKB1.Using partial ly purified AMPKK tophosphorylate bacterially expressed AMPKheterotrimers, we have identified upstreamkinase phosphorylation sites in AMPK by massspectrometry. In addition to confirming thephosphorylation of Thr 172, new sites (Thr258and Ser485/Ser491) have been identified in thea1/2 catalytic subunits (12). Mutagenesis studiesshowed that phosphorylation at the novel sites isnot essential for AMPK activation but mightaffect subcellular localisation or play amodulatory role. We are currently trying toidentify the AMPKK(s) responsible forphosphorylation at the Thr258 andSer485/Ser491 sites.

We studied the effect of insulin on AMPKactivation. We observed that AMPK activationin ischaemic hearts was antagonised by a pre-treatment of the hearts with insulin. The effectof insulin was blocked by wortmannin, aninhibitor of PI 3-kinase and resulted in adecreased phosphorylation state of Thr172 inAMPK (13). In addition, the insulin effect wasunrelated to changes in the AMP/ATP ratio,thus demonstrating that AMPK activity could bemodified by a mechanism independent of theAMP/ATP ratio in cardiomyocytes. Our recentstudies suggest that PKB might mediate theantagonism of AMPK by phosphorylation of thenovel regulatory sites (Ser485/Ser491) in thecatalytic a-1/2 subunits.

Control of smooth musclecontraction

D. Vertommen and M.H. Rider in collaboration withE. Waelkens, KUL, P. Gailly, UCL, Brussels and M.Walsh, Calgary, Canada

In earlier work, we used nanoelectrospray- andon-line capillary- electrospray ionisation massspectrometry (ESI-MS) to ident i fyautophosphory la t ion and regu la toryphosphorylation sites in the protein kinase D(PKD)(14). However, there are very fewphysiological substrates recognised for this

protein kinase. We have shown that in varioussmooth musc l e s s t imu l a t ed w i thvasoconstrictors, such as vasopressin, PKDbecomes activated. Moreover, we have foundthat a thin filament regulatory protein is a new invitro substrate for PKD. Its phosphorylationcould be implicated in the prolonged phase ofmuscle contraction when Ca2+-dependent P-lightchain phosphorylation of myosin is not involvedin force generation. Using mass spectrometry,we are looking whether the phosphorylationsites phosphorylated by PKD in vitro arephosphorylated in response to agonists in vivo.This work has potential for the understanding ofhypertension and could lead to new treatmentsfor this condition.


1. Rider MH, Bertrand L, Vertommen D,Michels PA, Rousseau GG, Hue L. 6 -Phosphofructo-2-kinase/fructose-2,6-bispphosphatase:head-to-head with a bifunctional enzyme that controlsglycolysis. Biochem J 2004;381:in press.

2. Deprez J, Vertommen D, Alessi DR, Hue L,Rider MH. Phosphorylation and activation of heart 6-phosphofructo-2-kinase by protein kinase B and otherprotein kinases of the insulin signaling cascades. J BiolChem 1997;272:17269-75.

3. Mora A, Davies AM, Bertrand L, Sharif I,Budas GR, Jovanovic S, Mouton V, Kahn CR,Lucocq JM, Gray GA, Jovanovic A, Alessi DR.Deficiency of PDK1 in cardiac muscle results in heartfailure and increased sensitivity to hypoxia. EMBO J2003;22:4666-76.

4. Bertrand L, Alessi DR, Deprez J, Deak M,Viaene E, Rider MH, Hue L. Heart 6-phosphofructo-2-kinase activation by insulin results fromSer-466 and Ser-483 phosphorylation and requires 3-phosphoinositide-dependent kinase-1, but not proteinkinase B. J Biol Chem 1999;274:30927-33.

5. Deprez J, Bertrand L, Alessi DR, Krause U,Hue L, Rider MH. Partial purification andcharacterization of a wortmannin-sensitive and insulin-stimulated protein kinase that activates heart 6-phosphofructo-2-kinase. Biochem J 2000;347:305-12.

6. Clement S, Krause U, Desmedt F, Tanti JF,Behrends J, Pesesse X, Sasaki T, Penninger J,

28

Doherty M, Malaisse W, Dumont JE, LeMarchand-Brustel Y,Erneux C, Hue L,Schurmans S. The lipid phosphatase SHIP2 controlsinsulin sensitivity. Nature 2001;409:92-7.

7. Marsin AS, Bertrand L, Rider MH, Deprez J,Beauloye C, Vincent MF, Van den Berghe G,Carling D, Hue L. Phosphorylation and activation ofheart PFK-2 by AMPK has a role in the stimulation ofglycolysis during ischaemia. Curr Biol 2000;10:1247-55.

8. Horman S, Browne G, Krause U, Patel J,Vertommen D, Bertrand L, Lavoinne A, Hue L,Proud C, Rider MH. Activation of AMP-activatedprotein kinase leads to the phosphorylation of elongationfactor 2 and an inhibition of protein synthesis. CurrBiol 2002;12:1419-23.

9. Horman S, Beauloye C, Vertommen D,Vanoverschelde JL, Hue L, Rider MH. Myocardialischemia and increased heart work modulate thephosphorylation state of eukaryotic elongation factor-2. JBiol Chem 2003;278:41970-76.

10. Krause U, Bertrand L, Hue L. Control of p70ribosomal protein S6 kinase and acetyl-CoA carboxylaseby AMP-activated protein kinase and proteinphosphatases in isolated hepatocytes. Eur J Biochem.2002;269:3751-9.

11. Meisse D, Van de Casteele M, Beauloye C,Hainault I, Kefas BA, Rider MH, Foufelle F,Hue L. Sustained activation of AMP-activated proteinkinase induces c-Jun N-terminal kinase activation andapoptosis in liver cells. FEBS Lett 2002;526:38-42.

12. Woods A, Vertommen D, Neumann D,Türk R, Bayliss J, Schlattner U, Wallimann T,Carling D, Rider MH. Identification ofphosphorylation sites in AMP-activated protein kinase(AMPK) for upstream AMPK kinases and study oftheir roles by site-directed mutagenesis. J Biol Chem2003;278: 28434-42.

13. Beauloye C, Marsin AS, Bertrand L, KrauseU, Hardie DG, Vanoverschelde JL, Hue L.Insulin antagonizes AMP-activated protein kinaseactivation by ischemia or anoxia in rat hearts, withoutaffecting total adenine nucleotides. FEBS Lett 2001;505:348-52.

14. Vertommen D, Rider MH, Ni Y, WaelkensE, Merlevede W, Vandenheede JR, Van Lint J.Regulation of protein kinase D by multisitephosphorylation. Identification of phosphorylation sites bymass spectrometry and characterization by site-directedmutagenesis. J Biol Chem 2000;275:19567-76.

29

ENDOCYTOSISPierre J. COURTOY, Member (right)Marie-France VANDENBROUCKE, Member (left)

Patrick VAN DER SMISSEN, Assistant MemberCéline AUZANNEAU, Postdoctoral FellowKarine CROIZET, Postdoctoral FellowPhilippe de DIESBACH, Postdoctoral FellowDonatienne TYTECA, Postdoctoral FellowMarcel METTLEN, Graduate StudentAnna PLATEK, Graduate StudentGeneviève DOM, Research AssociateThanh LAC, Technician (half-time),Michèle LERUTH, TechnicianOlga MEERT, Technician (half-time)Francisca N'KULI, TechnicianYves MARCHAND, Secretary

Endocytosis is a central activity of all eukaryotic cells, that allows for cell nutrition, regulates the composition of thecell surface and controls transfer of macromolecules across epithelial barriers. The role of endocytosis in signalling isalso increasingly recognized. This research group has made significant contributions in the dissection of endocyticpathways (1) and in unravelling its contribution to physiopathology (6,9), parasitology (2,4) as well aspharmacology (7,8). We are currently addressing the molecular machineries controlling the endocytic activity at theapical surface of epithelia and upon malignant transformation. Recent achievements include the elucidation of thesignalling cascade whereby the paradigmatic oncogenes, v-Src and K-Ras, control the actin cytocortex (3,5,10),specifically at the apical domain of polarized MDCK cells; the role of apical endocytosis in the regulated productionof thyroid hormones (6); and the elucidation of a deficit of apical endocytosis in a genetic form of kidney stones (9).

Oncogene-inducedmacropinocytosis in fibroblasts

M. Amyere, M. Mettlen, Ph. de Diesbach and P.J.Courtoy

We have originally reported that v-Src and K-Ras cause a profound remodelling of actincytoskeleton in Rat-1 fibroblasts, resulting instress fibre disappearance, cortical actinpolymerisation, ruffling and macropinocytosis(3). These alterations were found to depend onthe constitutive sequential activation ofphosphoinositide 3-kinase (PI3K) andphospholipase C (PLC). Noticeably, there wasno effect of v-Src, K-Ras and overall activationof PI3K and PLC on the receptor-mediatedendocytosis of transferrin by fibroblasts,underscoring the difference in the molecularmachinery supporting micropinocytosis viaclathrin-coated pits (i.e. receptor-mediatedendocytosis of transferrin) and macropinocytosis

upon closure of membrane ruffles (5). Recentinvestigations further involve phospholipase D(PLD) in macropinocytosis, downstream ofPI3K. Therefore, the role of the small GTPase,Arf-6, is also analyzed.

Regulation of endocytosis by v-Srcin polarized cells

M. Mettlen, A. Platek, P. Van Der Smissen, Ph. deDiesbach, D. Tyteca and P.J. Courtoy

Since most cancers are of epithelial origin, andsince apical endocytosis depends on actin, weexamined whether v-Src would similarly triggerfluid-phase endocytosis in MDCK cells andwhether apical endocytosis would be selectivelyaffected. However, because transformationcauses a rapid loss of epithelial polarity, weresorted to MDCK cells bearing athermosensitive v-Src kinase (MDCK/tsLA31cell line). When MDCK/tsLA31 cells were

30

Figure 1. Pathways of endocytosis. This scheme represents four possible modes of vesicular entry of solutes intocells : (a) caveolae; (b) clathrin- and dynamin-associated pits, or "coated pits"; (c) clathrin- and dynamin-independentmicropinocytic pits; and (d) macropinocytosis. Crosses represent cortical actin. It further emphasises the endosomesas sorting organelles after micropinocytosis and outlines the four endocytic routes inside the cell : (1) the degradativepathway to lysosomes; (2) the recycling pathway, back to the plasma membrane; (3) transcytosis to the oppositemembrane domain (here illustrated from basolateral to apical); and (4) sequestration into slowly recyclingendosomes.

plated at high density on a permeable supportand cultured at 40°C, a polarized epithelialmonolayer could be established, with closelyapposed tight junctions, and ~ 90 % oftransferrin-receptors being exposed at thebasolateral surface. This epithelium showed a 2-fold faster rate of fluid-phase endocytosis at thebasolateral than at the apical surface, and asimilar rate of receptor-mediated endocytosis oftransferrin at both membrane domains. Shiftingfrom the non-permissive temperature (40°C) tothe permissive temperature for v-Src kinase(34°C) dramatically accelerated fluid-phaseendocytosis at the apical surface (up to 6-fold),but basolateral endocytosis was not affected.This was accompanied by the induction of asingle, huge (> 5 µm) apical endocytic vacuole,presumably due to coalescence or swelling of thesubapical compartment. The selectiveacceleration of apical endocytosis and theappearance of the apical vacuole depended onPI3K, PLC and PLD, as shown by inhibition bywortmannin, NCDC and 1-butanol, respectively.

v-Src activation also induced a PI3K-, PLC-, andPLD-dependent apical ruffling. These data showthat v-Src selectively affects the dynamics of theapical plasma membrane, where microdomainsknown as “lipid rafts” are abundant. Currentinvestigations address the interaction between v-Src and “lipid rafts”, as well as the effect of v-Srcon polarized membrane lipid trafficking.

Relation between endocytosis andcell motility

A. Platek, M. Mettlen and P.J. Courtoy

It has been proposed that endocytosiscontributes to cell motility by allowing forselective removal of plasma membraneconstituents from the trailing edge and theirrecycling to the leading edge. Since v-Srcaccelerates both motility and endocytosis, weexamine whether these two processes are linkedand depend on the same regulatory machinery.To this aim, we made use of fibroblasts and

micropinocytosis macropinocytosis

?

earlyendosome

?

macropinosome

slowly recyclingcompartment

lateendosome

rapid recyclingand regurgitation

lysosomeapicalendosome

transcytoticendosome

(a) (b) (c) (d)

(2)

(1)

(4)

(3)

31

MDCK cells harbouring thermosensitive v-Srckinase (Rat-1/tsLA29 and MDCK/tsLA31cells). In both cell lines, v-Src activation led to a2-fold acceleration of cell motility, as evidencedby the population-based wound healing assayand by single cell recording in Dunn chambers.Accelerated motility was selectively abrogated byPI3K, PLC and PLD inhibitors. Theseobservations suggest a link between acceleratedmotility and endocytosis. Surprisingly, v-Srcactivation abrogated directionality of cell motilityin response to chemotactic growth factor (GF)gradients. v-Src down-regulated GF-receptorsby about two-fold, but half receptor occupancyin non-transformed cells was sufficient to inducedirectional motility. We currently test whetherthis loss of a polarized actin response resultsfrom diffuse activation of PI3K, PLC and/orPLD (“excess noise”) (10).

Regulation of transcytosis

P.J. Courtoy, in collaboration with J.P. Vaerman,MEXP

We have continued our long-lastingcollaboration with Dr. J.P. Vaerman to addressthe regulation of transcytosis in transfectedMDCK cells bearing the rabbit polymeric Ig-receptor. Receptor occupancy by polymeric IgA,which accelerates transcytosis, was found toactivate PLC, as shown by an increased level ofinositol 3,4,5-trisphosphate. In the absence ofadded polymeric IgA, transcytosis of polymericIg-receptor was also accelerated by the co-operative effect of the two downstream arms ofthe PLC response, i.e. by increasing cytosoliccalcium concentration with ionomycin, andactivation of protein kinase C via phorbol esters.Thus, PLC is involved in signalling, not only ofv-Src-induced apical endocytosis (see above), butalso in a presumably late step of basolateral-to-apical transcytosis of the polymeric Ig-receptor.It is likely that both effects are mediated by theactin cytoskeleton.

Apical endocytosis regulatesthyroid hormone production inhuman disease; moleculardissection in vitro

K. Croizet, P.J. Courtoy and M.F. van den Hove

The production of thyroid hormones bythyrocytes results from apical endocytosis of

thyroglobulin stored in the colloid, followed byintracellular proteolysis. Since both substratesand hydrolytic enzymes are in vast excess, wehypothesised that the production of thyroidhormones is regulated by their encounter, i.e.depends on rate-limiting endocytic catalysts. Totest this hypothesis, we have followed twoapproaches. First, we found that the increasedlevel of expression of the rate-limiting endocyticcatalysts, Rab5a and Rab7, in autonomoushyperactive adenomas, closely correlates with (i)a decrease in residual thyroglobulin content; and(ii) an increased recovery of particulate iodinetowards more distal compartments (i.e. the mostactive proteolytic organelles of the degradativepathway). Second, having established polarizedhuman thyrocytes that are competent forselective basolateral delivery of thyroidhormone, we found that TSH stimulation, ordirect activation of the cAMP cascade, aresufficient to increase Rab5a and Rab7 expression(6). Rabs membrane recruitment was alsoenhanced, indicating their further activation.Accordingly, we currently investigate thepossible effect of the cAMP cascade on theexpression of appropriate Rab GEFs and GAPsin this physiopathological system.

Alterations in the endocyticapparatus of mice lacking renalchloride channel, ClC-5, account forproteinuria in Dent’s disease

C. Auzanneau and P.J. Courtoy, in collaboration withO. Devuyst, NEFR

To elucidate the molecular basis of Dent’sdisease, an X-linked familial form ofnephrolithiasis linked to low-molecular weightproteinuria, we have analysed the endocyticapparatus in knock-out (KO) mice for thekidney-specific chloride channel, ClC-5. Theseinvestigations are carried out in collaborationwith Dr. W. Guggino (John Hopkins, Baltimore,MD, USA) and Dr. E.I. Christensen (Aarhus,DK). The comparison of the steady-state level ofseveral major constituents and regulators of theap ica l endocy t i c appara tus , the i rimmunolocalisation at the ultrastructural level,and the follow-up by analytical subcellularfractionation of an apical endocytic tracer, atvarious intervals of uptake, showed thatdefective ClC-5 in Dent’s disease patients andKO mice leads to a major trafficking defect ofthe low-molecular weight protein receptors,megalin and cubilin (9). To address this defect,

32

we currently examine the role of ClC-5 in theacidification of apical endosomes. Incollaboration with Dr. H. de Jonge (Amsterdam,NL) and Dr. J.J. Cassiman (KUL, Belgium), wealso compare, for their role and partners, ClC-5with the Cystic Fibrosis Transmembraneconductance Regulator (CFTR) in KO andDF508 transgenic mice.


1. Cupers P, Veithen A, Kiss A, Baudhuin P,Courtoy PJ. Clathrin polymerization is not requiredfor bulk-phase endocytosis in rat fetal fibroblasts. J CellBiol 1994;127:725-35.

2. Coppens I, Levade T, Courtoy PJ. Host plasmalow density lipoprotein particles as an essential source oflipids for the bloodstream forms of Trypanosoma brucei.J Biol Chem 1995;270:5736-41.

3. Veithen A, Cupers P, Baudhuin P, Courtoy PJ.v-Src induces constitutive macropinocytosis in ratfibroblasts. J Cell Sci 1996;109:2005-12.

4. Coppens I, Courtoy PJ. The adaptativemechanisms of Trypanosoma brucei for sterol homeostasisin its different life-cycle environments. Annu RevMicrobiol 2000;54:129-56.

5. Amyere M, Payrastre B, Krause U, Van DerSmissen P, Veithen A, Courtoy PJ. Constitutivemacropinocytosis in oncogene-transformed fibroblastsdepends on sequential permanent activation ofphosphoinositide 3-kinase and phospholipase C. MolBiol Cell 2000;11:3453-67.

6. Croizet-Berger K, Daumerie C, Couvreur M,Courtoy PJ, van den Hove MF. The endocyticcatalysts, Rab5a and Rab7, are tandem regulators ofthyroid hormone production. Proc Natl Acad SciUSA 2002;99:8277-82.

7. Tyteca D, Van Der Smissen P, Mettlen M,Van Bambeke F, Tulkens PM, Mingeot-LeclercqM-P, Courtoy PJ. Azithromycin, a lysosomotropicantibiotic, has distinct effects on fluid-phase and receptor-mediated endocytosis, but does not impair phagocytosis inJ774 macrophages. Exp Cell Res 2002;281:86-100.

8. de Diesbach Ph, N’Kuli F, Delmée M,Courtoy PJ. Infection by Mycoplasma hyorhinisstrongly enhances uptake of antisense oligonucleotides: areassessment of receptor-mediated endocytosis in theHepG2 cell line. Nucleic Acids Res 2003;31:886-92.

9. Christensen EI, Devuyst O, Dom G, NielsenR, Van Der Smissen P, Verroust P, Leruth M,Guggino WB, Courtoy PJ. Loss of chloride channelClC-5 impairs endocytosis by defective trafficking ofmegalin and cubilin in kidney proximal tubules. ProcNatl Acad Sci USA 2003;100:8472-77.

10. Platek A, Mettlen M, Camby I, Kiss R,Amyere M, Courtoy PJ. V-Src acceleratesspontaneous motility via phosphoinositide 3-kinase,phospholipase C and phospholipase D, but abrogateschemotaxis in Rat-1 and MDCK cells. J Cell Sci2004; in press.

33

EXTRACELLULAR MATRIX BREAKDOWNPierre J. COURTOY, Member (left)Yves EECKHOUT, Member (right)Etienne MARBAIX, Member (middle)

Patrick HENRIET, MemberHervé EMONARD, Guest InvestigatorChristine GALANT, Assistant MemberAlix BERTON, E.U. Marie Curie FellowPatricia CORNET, Graduate StudentVassil VASSILEV, Graduate StudentDenis DELVAUX, TechnicianPascale LEMOINE, TechnicianOlga MEERT, TechnicianYves MARCHAND, Secretary

The extracellular matrix (ECM) plays a central role in the structural and functional organization of tissues andorgans. ECM constituents, in particular fibrillar collagens, are the most abundant proteins of the human body.Physiological and pathological breakdown of ECM is predominantly achieved by a family of neutralmetalloproteinases, called matrix metalloproteinases (MMPs). Our group has a long-standing expertise in thebiochemistry and molecular biology of collagenase and related MMPs (1, 2). We have demonstrated thatmenstrual bleeding in women is due to the expression and activation of some MMPs (3). This seminal observationled us to : (i) exploit this system as a human model to study the regulation of MMPs, in particular cellularinteractions that integrate overall hormonal impregnation (4) with local environmental changes (5, 6); and (ii)explore whether this basic knowledge can lead to a better understanding and a rational treatment of abnormaluterine bleeding, a major health problem (7). Recently, our group has entered a new field of research, investigatinghow local MMP activity may be controlled by cells through plasma membrane binding or endocytosis.

Regulation of the expression ofendometrial MMPs and cytokines

P. Henriet, P. Cornet, V. Vassilev, P.J. Courtoy, Y .Eeckhout and E. Marbaix

Both endocrine and paracrine factors participatein controlling the expression and activity ofMMPs involved in menstrual breakdown of thehuman endometrium (4, 6). Several genesencoding MMPs and cytokines present maximalendometrial mRNA concentrations aroundmenstruation (reviewed in 1). They substantiallydiverge, however, in their expression profileduring the other phases of the cycle, suggestingdifferential regulation by estradiol andprogesterone.

To directly measure the effect of the ovariansteroids, mRNA amounts of selected genes werequantified in a large collection of endometrialsamples collected throughout the menstrualcycle and stored before or after their culture as

explants. We first focused on gelatinaseB/MMP-9 and on the TGFb-related cytokine,LEFTY-A/EBAF (8). A major upregulation inMMP-9 mRNA expression occurred in vivo inendometria showing signs of menstrualbreakdown; this followed a larger increase inLEFTY-A mRNA, pointing to a regulatory roleof this cytokine. Differential response to theovarian steroids was observed in explant culture.LEFTY-A mRNA proved more sensitive thanMMP-9 mRNA to the presence of endogenousas well as exogenous progesterone. Moreover,addition of recombinant LEFTY-A to explantculture further increased the amounts of selectedmenstrual MMPs, including MMP-9, a responsethat was blocked by the ovarian steroids.Altogether, these observations suggest that, invivo, different pathways finely tune in space, timeand amplitude, the global control of estradioland progesterone on the expression of genesrequired for menstrual ECM breakdown. Theapproach will be extended to other relevantmenstrual MMPs and cytokines

34

Figure 1. Regulation of MMPs in the human endometrium : current model. Upon ovarian steroids withdrawal,LEFTY-A and epithelial IL-1a are released and trigger the production by adjacent fibroblasts of various proMMPs,including interstitial procollagenase-1/proMMP-1, progelatinase B/proMMP-9 and prostromelysin-1/proMMP-3.Fibroblast activation involves interleukin-1 receptor (IL-1RI) and is opposed by various soluble factors, such asinterleukin-1 receptor antagonist (IL-1Ra). TGF-b inhibits MMPs production, but this brake is relieved by LEFTY-A. Expression of proMMPs by stimulated fibroblasts is further blocked downstream by ovarian steroids (dual brake).Secreted proMMPs need activation by proteinases, including MMP-3. Active MMP-1, if not neutralised by tissueinhibitors (TIMPs), cleaves fibrillar collagens at position 3/4 of the distance from the amino-terminus. Thefragments and other matrix proteins are then further degraded by other proteinases, including MMP-9 and MMP-3.

Selective binding of active MMP-7 tohuman epithelial cell membrane

A. Berton, H. Emonard, P.J. Courtoy and E. Marbaix

Most normal and cancerous epithelial cellsexpress matrilysin-1/MMP-7. In addition todegrading extracellular matrix substrates,activating other MMPs and inactivating enzymeinhibitors, MMP-7 also cleaves cell surfaceproteins such as E-cadherin, b 4 integrin,proTNF-a and Fas ligand. Whereas proMMP-7tightly binds to heparan sulfate proteoglycans atthe surface of rat endometrial epithelial cells,only active MMP-7 binds to colon cancer cells.In our model of endometrial cultured explants,we observed a preferential accumulation ofMMP-7 in the tissue whereas proMMP-7 wasthe major form detected in the conditionedmedium. In cryosections of humanendometrium, proMMP-7 was immunolocalizedin the apical cytoplasm of all epithelial cells,whereas MMP-7 was detected in focallydistributed epithelial cells, showing a diffuse

cytoplasmic staining with a strongly enhancedperipheral signal, which suggested associationwith the plasma membrane. Recombinant MMP-7, but not proMMP-7, showed significantbinding to endometrial epithelial cells, through asingle class of receptors. Cell surface-associatedMMP-7 was found to be functionally active.Targeting of active MMP-7 to the plasmamembrane could represent a relevant mechanismto regulate various cellular functions byactivating, inactivating or releasing structural orsignaling membrane proteins.

Receptor-mediated endocyticclearance of proMMP-2:TIMP-2complexes

H. Emonard, A. Berton, and P.J. Courtoy, incollaboration with CNRS, Reims, France

Activity of MMPs is controlled at differentlevels. Once synthesized as proenzyme species,proMMPs must be activated to exert fullproteolytic activity. MMP activation may depend

activatingproteinase

TIMPs

proMMP-1

MMP-1

fibrillar collagen 3/4 and 1/4fragments

Fibroblasts

IL-1RI

IL-1Ra

Epithelial cells

X ?

priming ?

IL-1a

(release ?)

ovariansteroids

ovariansteroids

TGF-b LEFTY-A

proMMP-9

MMP-9

proMMP-3

MMP-3

degraded collagen

35

on other MMPs, as illustrated for proMMP-9activation in the endometrium tissue (9). Besidestissue inhibitors of metalloproteinases (TIMPs),MMPs may be inhibited by various naturalcompounds, including unsaturated long-chainfatty acids (10). Plasma membrane binding doesnot only contribute to MMPs concentration ondiscrete spots at the cell surface, as illustratedabove for MMP-7, but can as well silence MMPsactivities by promoting their intracellulardegradation. ProMMP-2/gelatinase A ispredominantly secreted in association withTIMP-2. The low-density lipoprotein receptor-related protein (LRP), a scavenger receptor,mediates the endocytic clearance of variousproteins, including proteinases and proteinase-inhibitor complexes. Upon treatment with RAP,a natural LRP ligand antagonist, we showed thatHT1080 human fibrosarcoma cells accumulatedboth proMMP-2 and TIMP-2. We furtherdemonstrated that RAP inhibited endocytosisand lysosomal degradation of 125I-proMMP-2:TIMP-2 complexes, but had no effect onbinding of 125I-proMMP-2:TIMP-2 complexes tothe cell surface. These data indicate thatclearance of proMMP-2:TIMP-2 complexes is atwo-step process, involving initial binding to thecell membrane in a RAP-independent manner,and subsequent LRP-mediated internalisationand degradation.

Role of matrix metalloproteinases inabnormal endometrial bleeding

C. Galant, Y. Eeckhout, P.J. Courtoy and E.Marbaix, in collaboration with J.L. Brun, Bordeaux,France

Since matrix metalloproteinases (MMPs) play akey role in initiating normal menstrualbreakdown, we looked for their contribution inpathological conditions characterized byexcessive, prolonged or irregular bleedingwithout organic lesion. Up to one fourth ofhysterectomies are due to such functionalmenstrual disorders. Endometrial biopsies frompatients irregularly bleeding at the time ofsampling were compared to biopsies sampledduring non-bleeding intervals and to controlbiopsies from normally menstruating women.Irregular bleeding was clearly associated withmenstrual-like stromal breakdown in focicontaining low levels of ovarian steroidsreceptors, and with increased expression andactivation of several MMPs, together withdecreased production of TIMP-1 (11). These

results confirmed in vivo and in various hormonalconditions our previous findings obtained withcultured explants from patients on long-termprogestinic contraception (7). A recentcollaboration with Dr. J.L. Brun addresseswhether excessive menstrual bleeding andrecurrence of bleeding after surgical treatmentcould be linked to the same molecularmechanisms. To clarify the reason for these localdisorders, we currently look for alteredexpression or regulation of ovarian steroidreceptors and appropriate cytokines.


1. Henriet P, Cornet PB, Lemoine P, Galant C,Singer CF, Courtoy PJ, Eeckhout Y, Marbaix E.Circulating ovarian steroids and endometrial matrixmetalloproteinases (MMPs). Ann N Y Acad Sci2002;955:119-38 (review).

2. Krane SM, Byrne MH, Lemaitre V, Henriet P,Jeffrey JJ, Witter JP, Liu X, Wu H, Jaenisch R,Eeckhout Y. Different collagenase gene products havedifferent roles in degradation of type I collagen. J BiolChem 1996;271:28509-15.

3. Marbaix E, Kokorine I, Moulin P, Donnez J,Eeckhout Y, Courtoy PJ. Menstrual breakdown ofhuman endometrium can be mimicked in vitro and isselectively and reversibly blocked by inhibitors of matrixmetalloproteinases. Proc Natl Acad Sci USA1996;93:9120-5.

4. Marbaix E, Kokorine I, Henriet P, Donnez J,Courtoy PJ, Eeckhout Y. The expression ofinterstitial collagenase in human endometrium iscontrolled by progesterone and by oestradiol and is relatedto menstruation. Biochem J 1995;305:1027-30.

5. Kokorine I, Marbaix E, Henriet P, Okada Y,Donnez J, Eeckhout Y, Courtoy PJ. Focal cellularorigin and regulation of interstitial collagenase (matrixmetalloproteinase-1) are related to menstrual breakdownin the human endometrium. J Cell Sci 1996;109:2151-60.

6. Singer CF, Marbaix E, Kokorine I, LemoineP, Donnez J, Eeckhout Y, Courtoy PJ. Paracrinestimulation of interstitial collagenase (MMP-1) in thehuman endometrium by interleukin 1alpha and its dualblock by ovarian steroids. Proc Natl Acad SciUSA 1997;94:10341-5.

36

7. Galant C, Vekemans M, Lemoine P, KokorineI, Twagirayezu P, Henriet P, Picquet C, Rigot V,Eeckhout Y, Courtoy PJ, Marbaix E. Temporaland spatial association of matrix metalloproteinases withfocal endometrial breakdown and bleeding upon progestin-only contraception. J Clin Endocrinol Metab2000;85:4827-34.

8. Cornet PB, Picquet C, Lemoine P, OsteenKG, Bruner-Tran KL, Tabibzadeh S, CourtoyPJ, Eeckhout Y, Marbaix E, Henriet P.Regulation and function of LEFTY-A/EBAF in thehuman endometrium. mRNA expression during themenstrual cycle, control by progesterone, and effect onmatrix metalloprotineases. J Biol Chem 2002;277:42496-504.

9. Rigot V, Marbaix E, Lemoine P, Courtoy PJ,Eeckhout Y. In vivo perimenstrual activation of

progelatinase B (proMMP-9) in the human endometriumand its dependence on stromelysin 1 (MMP-3) ex vivo.Biochem J 2001;358:275-80.

10. Berton A, Rigot V, Huet E, Decarme M,Eeckhout Y, Patthy L, Godeau G, HornebeckW, Bellon G, Emonard H. Involvement of fibronectintype II repeats in the efficient inhibition of gelatinases Aand B by long-chain unsaturated fatty acids. J BiolChem 2001;276:20458-65.

11. Galant C, Berlière M, Dubois D,Verougstraete J-C, Charles A, Lemoine P,Kokorine I, Eeckhout Y, Courtoy PJ, MarbaixE. Focal expression and final activity of matrixmetalloproteinases may explain irregular dysfunctionalendometrial bleeding. Am J Pathol 2004; 165: 83-94.

37

CONNECTIVE TISSUE AND ARTHRITISDaniel MANICOURT, Member

Hafida EL AJJAJI, PostgraduateAnnette MARCELIS, TechnicianAnne van EGEREN, Technician

The major research efforts of the group are focused on the pathophysiology of osteoarthritis (OA), the most prevalentof joint disorders. In this disease process, the anatomy and composition of both articular cartilage and subchondralbone are altered by a complex combination of degradative and reparative processes which depend upon an interplaybetween, on the one hand, local biomechanical factors acting on the joint and, on the other hand, generalized factorsmaking a predisposition to the disease. However, relatively little is known about which of the disease processes andetiological factors control progression of the disease process to OA. Further, although it is clear that the initiationof cartilage damage and the progression to full cartilage loss may involve separate pathophysiological mechanisms, itis unclear whether OA changes occur first in cartilage, in bone or are concurrent.

Effects of nonsteroidal anti-inflammatory drugs on the overallmetabolism of articular cartilage

Because they inhibit cyclo-oxygenase (COX),and hence the production of prostaglandins(PGs), nonsteroidal anti-inflammatory drugs(NSAIDs) are widely prescribed in patientssuffering from arthritides. Two isoforms ofCOX have been identified thus far: COX-1,which is constitutively expressed in most tissuesand COX-2, which is highly inducible inresponse to proinflammatory cytokines andmitogens. It is generally believed that thebeneficial effects of NSAIDs are related to theirability to inhibit COX-2 whereas thegastrointestinal and renal toxicity of these drugsresults from their inhibition of COX-1, acontention that has provided the basis for thedevelopment of highly selective COX-2inhibitors. It should be however stressed thatCOX-1-derived PGs can contribute to theinflammatory response and that COX-2-derivedPGs perform physiologically important roles

such as the maintenance of normal renalfunction. Furthermore, COX-2-derived PGs, inaddition to their anti-inflammatory properties,have been implicated in the protection of thegastrointestinal tract from injury.Although NSAIDs undeniably produce relief ofpain and improvement of joint mobility inpatients suffering from arthritides, ex vivo and invivo studies have shown that some NSAIDsinhibit the synthesis of cartilage proteoglycanswhereas others do not. This differential effect ofNSAIDs on cartilage metabolism is mostrelevant to clinical practice since any drug, thatsuppresses proteoglycan synthesis and impairsthe chondrocyte to repair its already damagedextracellular matrix, could potentially acceleratethe breakdown of the cartilage tissue. On theother hand, although hyaluronan (HA) plays acentral structural role in the supramolecularorganization of proteoglycans and, hence on thebiomechanical properties of articular cartilage,the possible effects of NSAIDs on themetabolism of this glycosaminoglycan has so farfocused little investigative attention.

38

We therefore investigated the action ofcelecoxib (a strong selective COX-2 inhibitor),on the metabolism of newly synthesized HA andproteoglycan molecules in explants from humanOA cartilage (1). In contrast to classicalNSAIDs, this COX-2 selective inhibitor had apositive effect on the overall metabolism of bothproteoglycans and hyaluronan, two majorcomponents of the extracellular matrix ofcartilage. This effect, which is independent ofthe inhibition of prostaglandin production, isunder investigation as it might be of greatbiological and therapeutic significance inarthritides.

Markers of connective tissuemetabolism in health and disease

(In collaboration with E. Thonar, Rush-Presbyterian-StLuke’s Medical Center, Chicago, USA).

Nowadays, several biochemical moleculesderived from the joint components can bequantified in body fluids (joint fluid, blood andurine). Theses molecules termed “metabolicmarkers” or simply “markers” appear asimportant tools to disclose in vivo importantchanges occurring during both the preclinicaland clinical stages of various joint diseases,including osteoarthritis. There is also evidencethat these markers may prove helpful indetermining whether a therapeutic regimen iseffective or not, and this in a relatively shortperiod of time. Indeed, in the absence ofmarkers, the efficacy of treatment in jointdisorders relies mainly on radiographic changes,an approach that takes years before one canreach meaningful results.

The markers that are most currently used arehyaluronan, a marker of synovial proliferationand inflammation, antigenic keratan sulfate, amarker of proteoglycan metabolism, cartilageoligo-matrix protein, a marker of cartilage matrixremodeling, and the telopeptides of type IIcollagen, a marker of the breakdown of cartilagecollagen. There is indeed good agreement thatthis panel of markers helps diagnose, monitor orprognosticate osteoarthritic changes.

Role of the subchondral bone in theinitiation and progression of theosteoarthritic disease process

So far, the possible role of subchondral bone inthe initiation and/or progression ofosteoarthritis (OA) has focused littleinvestigative attention. We have thereforeexplored this topic in an animal model ofosteoarthritis. In this model, severing of theanterior cruciate ligament of the knee increasesdramatically the biomechanical forces applied tothe internal compartment of the knee joint andresults in the progressive apparition of OAlesions in the operated joint which closelyresemble those seen in human OA.

During the first weeks following jointdestabilization, we have observed a dramaticdecrease in the density and volume of thetrabecular subchondral bone. These changesincreased with time post-surgery and wererestricted to the internal compartment of theoperated joint whereas no significant changes inbone density and volume could be disclosed inthe external compartment of the unstable joint.Obviously, theses changes reflect an adaptationof the bone to absorb the enhancedbiomechanical forces imposed upon it. On theother hand, these changes concomitantly inducea dramatic increase in the tensile and shearingforces upon the overlying articular cartilage and,in so doing, contribute to the degradation of thecartilage tissue. Our working hypothesis issupported by the finding that animals receivingdrugs known to inhibit bone resorption do notshow up any change in the volume and densityof the trabecular subchondral bone of theoperated knee and, more importantly, exhibit adramatic decrease in the severity of cartilage OAlesions (2).

These findings open a new approach in thetherapeutic regimen of OA and studies arecurrently conducted in human OA.

Towards a better understanding ofthe metabolism of hyaluronan inconnective tissues

Research efforts are also devoted to theregulation of hyaluronan metabolism both inhealth and disease. In skin, which contains 50 %of total body hyaluronan, the half-life ofhyaluronan is about one day, and even in asseemingly inert tissue as cartilage, hyaluronan

39

turns over with a half-life of one to three weeks.In the blood stream, the half-life of hyaluronanis two to five minutes. All such catabolism ispresumably a result of hyaluronidases. What isthe nature of the control mechanisms thatorchestrate such vastly different rates ofturnover? The hyaluronan of vertebrateorganisms can exist in many states, in a varietyof sizes, in extracellular forms, free in thecirculation, loosely associated with cells andtissues, tightly intercalated within proteoglycan-rich matrices such as that of cartilage, bound byreceptors to cell surfaces, or even in severalintracellular locations. Superimposed on thesemany states are the panoply of binding proteins,or hyaladherins, that decorate the hyaluronanmolecule. How do mechanisms of catalysis differamong this wide range of physical and chemicalstates of the hyaluronan substrate? It is unlikelythat hyaluronidase activity is retained in vivo in anactive form within the extracellular matrix whereit could cause great havoc. If it is found withinthe extracellular matrix, it may be in an inactiveor suppressed form, perhaps bound to aninhibitor. Such a situation would parallel therelationship between the metalloproteinases andthe tissue inhibitors of metalloproteinases orTIMPs that exert exquisite control overmetalloproteinase activity.


1. El Hajjaji H, Marcelis A, Devogelaer, and J-Pand Manicourt DH. Celecoxib has a positive effect onthe metabolism of proteoglycans and hyaluronan inhuman osteoarthritic cartilage. J Rheumatol, 2003;30:2444-51

2. Behets C, Williams JM, Chappard D,Devogelaer JP, Manicourt DH. Effects of calcitonin

on trabecular bone changes and on osteoarthritic cartilagelesions after acute anterior cruciate ligament deficiency. JBone Miner Res 2004;19(11):1821-6.

3. Manicourt DH, Druetz-Van Egeren A,Haazen L, Nagant de Deuxchaisnes C. Effects oftenoxicam and aspirin on the metabolism of proteoglycansand hyaluronan in normal and osteoarthritic humanarticular cartilage. Br J Pharmacol 1994;113:1113-20.

4. Blot L, Marcelis A, Devogelaer JP, ManicourtDH. Effects of diclofenac, aceclofenac and meloxicam onthe metabolism of proteoglycans and hyaluronan inosteoarthritic human cartilage. Br J Pharmacol2000;131:1413-21.

5. Manicourt DH, Poilvache P, Van Egeren A,Devogelaer JP, Lenz ME, Thonar EJ.Synovial fluid levels of tumor necrosis factor alpha andoncostatin M correlate with levels of markers of thedegradation of crosslinked collagen and cartilage aggrecanin rheumatoid arthritis but not in osteoarthritis.Arthritis Rheum 2000;43:281-8.

6. Manicourt DH, Poilvache P, Nzeusseu A, vanEgeren A, Devogelaer JP, Lenz ME, Thonar EJ.Serum levels of hyaluronan, antigenic keratan sulfate,matrix metalloproteinase 3, and tissue inhibitor ofmetalloproteinases 1 change predictably in rheumatoidarthritis patients who have begun activity after a night ofbed rest. Arthritis Rheum 1999;42:1861-9.

7. Thonar EJM, Lenz ME, Masuda K andManicourt DH: Body fluid markers of cartilagemetabolism. In Dynamics of bone and cartilagemetabolism. (M.J. Seibel, S. P. Robins and J. P.Bilezikian, eds) Academic Press, San Diego,1999,453-64.

40

METABOLIC COMPARTMENTATION IN TRYPANOSOMES

Fred R. OPPERDOES, MemberPaul A.M. MICHELS, Member

Véronique HANNAERT, Assistant MemberJean-Pierre SZIKORA, Assistant MemberJohan DE JONCKHERE, Visiting ScientistDelphine GERBOD, Société Générale deBelgique FellowWilfredo QUINONES, Post-Doctoral FellowMarie-Astrid ALBERT, Graduate StudentDaniel GUERRA, Graduate StudentMurielle HERMAN, Graduate StudentHanane KRAZY, Graduate StudentJuliette MOYERSOEN, Graduate StudentCédric YERNAUX, Graduate StudentFreddy ABRASSART, TechnicianNathalie CHEVALIER, TechnicianDominique COTTEM, TechnicianChristian VAN LANGENHOVE, TechnicianJoris VAN ROY, TechnicianFrançoise MYLLE, Secretary

Trypanosomes are responsible for human sleeping sickness in tropical Africa and for a similar disease called‘nagana’ in cattle. These are very serious diseases, with fatal outcome if left untreated. The presently available drugsare not very efficient and cause serious side effects. Moreover, development of drug resistant parasites is becoming amajor problem. Therefore new drugs are badly needed.

Trypanosoma brucei, when it resides in the mammalian bloodstream, relies entirely on glycolysis for its ATPsupply. Moreover, the parasite is characterized by a unique form of metabolic compartmentation; the majority of theenzymes of the glycolytic pathway is sequestered in peroxisome-like organelles called glycosomes (1). For the abovereasons the glycolytic pathway is considered a validated and promising target for new drugs to be designed. Sincemany years we study the kinetic and structural properties of the glycolytic enzymes of T. brucei and closely relatedparasites such as Trypanosoma cruzi and Leishmania mexicana, and use the collected information for the design ofeffective and selective inhibitors by structure-based and catalytic mechanism-based approaches (2).

In addition our research aims at understanding what controls the glycolytic flux in vivo. The flux control isbeing studied in a quantitative manner by using a mathematical model prepared on the basis of the experimentallydetermined kinetic properties of all enzymes constituting the pathway, and by in vivo experiments in which theactivity of different enzymes of the pathway is varied by either biochemical or genetic means. Such experiments couldprovide both insight into the consequences of the compartmentation of the pathway and information as to whichenzymes of the pathway are the best targets for drugs.

Several enzymes of another pathway of carbohydrate metabolism: the hexose monophosphate pathway, involvedin the generation of intermediates essential for cell growth, cell division and protection against oxidative stress, areassociated with the glycosomes as well. This triggered our interest in their function as glycosomal proteins. Not onlythe role of the glycosome in trypanosomatid metabolism is the topic of our research, but also the assembly of theorganelle. We are studying the proteins, called peroxins, involved in glycosome biogenesis and particularly themechanism by which they accomplish the import of matrix proteins.

41

Enzymes of carbohydratemetabolism

Cloning and characterization of genes forglycolytic enzymes, expression andcharacterization of recombinant enzymesand inhibitor development.

V. Hannaert, D. Guerra, M.-A. Albert, F.!Opperdoesand P. Michels

In previous years, genes of all enzymes of thepathway have been cloned and characterized inour laboratory (3). In addition, we have clonedand sequenced the genes for two isoenzymes ofthe 6-phosphofructo-2-kinase/fructose2,6-bisphosphatase (PFK2/FBPase2), enzymesresponsible for the synthesis and hydrolysis offructose 2,6-bisphosphate, a potent allostericeffector of pyruvate kinase (PYK) and possiblythe major regulator of glycolysis intrypanosomatids. Also genes for a number ofglycolytic enzymes of Leishmania species and T.cruzi have been cloned and sequenced; in 2001the genes for L. mexicana phosphoglyceratemutase (PGAM) and L. donovaniphosphofructokinase (PFK) were characterized(4). The predicted primary structure of almost alltrypanosomatid glycolytic enzymes appearedquite different from their mammaliancounterparts and some enzymes are even nothomologous, rendering several of these enzymeshighly promising targets for anti-parasite drugs.Most of the T. brucei glycolytic enzymes havebeen expressed in Escherichia coli and purified andtheir kinetic properties have been determined.

In collaboration with C. López and J.L.Ramirez (Universad Central de Venezuela,Caracas) the gene encoding L donovani PFK wascloned and sequenced and its bacteriallyexpressed gene product studied (4). The deducedpolypeptide contains a C-terminal type 1peroxisome-targeting signal (PTS1), -SKV. Likethat of the previously characterized T. brucei PFK(70 % identity), the sequence showed the highestsimilarity to inorganic pyrophosphate (PPi)-dependent PFKs, despite being ATP-dependent.Its kinetic properties were similar to those of theT. brucei enzyme. Modeling studies and site-directed mutagenesis were employed to shedlight on the structural basis for the unique AMPeffector specificity and on ATP/PPi specificityamong PFKs.

The predicted amino-acid sequence of L.mexicana PGAM is 74 % identical to that of theT. brucei mutase. Both trypanosomatid PGAMsbelong to the c lass of cofactor2,3-bisphosphoglycerate independent mutases,contrary to the non-homologous cofactor-dependent enzyme of humans. The parasiteenzymes have been expressed in E. coli, and,upon purification, their kinetic properties havebeen determined.

During previous years, Prof. W. Hol(University of Washington, Seattle, USA) andDr. L. Gilmore (University of Edinburgh, UK)and their coworkers have, in collaboration withus, established the crystal structure of severaltrypanosomatid glycolytic enzymes: aldolase(ALD), triosephosphate isomerase (TIM),glyceraldehyde-3-phosphate dehydrogenase(GAPDH), 3-phosphoglycerate kinase (PGK),glycerol-3-phosphate dehydrogenase (GPDH)and pyruvate kinase (PYK). Based on thesecrystal structures, and on insight in the kineticproperties and catalytic mechanisms, inhibitorsof each of these enzymes have been designedand synthesised by our colleagues Prof. J. Périé(Université Paul Sabatier, Toulouse, France) andProf. M. Gelb (University of Washington,Seattle, USA). Promising inhibitors, selective forparasite GAPDH and ALD, with Ki’s in thenanomolar range and inhibiting growth ofparasites in culture without affecting growth ofcultured human cells have already been obtained.

In collaboration with Dr. M. Willson andProf. J. Périé (Université Paul Sabatier) we alsosynthesized a number of glucosamine analogueswhich inhibit both the yeast and the T. bruceihexokinase (HXK) (5). The most potentinhibitor of the trypanosome HXK,m-bromophenyl glucosamide, did not affect theactivity of its yeast counterpart. This compoundalso effectively inhibited growth of in vitrocultured trypanosomes. The structure of the T.b r u c e i enzyme, in complex with its m -bromophenyl glucosamide inhibitor, wasmodelled using the crystal structure of theSchistosoma mansoni HXK - glucose complex. Thismodel allowed us to explain the mode of actionof this inhibitor on the trypanosome HXK.

Also in collaboration with our colleagues inToulouse, the kinetic mechanism of T. bruceiPFK has been studied in detail, and its active sitehas been explored by using a variety of inhibitorsderived from the fructose 6-phosphate analogue

42

2,5-anhydromannitol (6). The best inhibitor wasa compound with an electrophilic isothiocyanategroup at position 1; it displayed an irreversibleinactivation pattern with a Ki value of 133 mM.The residue involved in the specific inactivationof the parasite enzyme was identified by site-directed mutagenesis, Lys227. Based on thispromising result, other compounds are nowbeing developed.

The cloning of the T. brucei enolase (ENO)gene was reported last year. The encodedpolypeptide has 59 – 62 % identity with thedifferent human enolases.

The kinetic properties of bacteriallyexpressed T. brucei ENO are very similar to thoseof the mammalian enzymes. Furthermore,structure modelling (in collaboration with Dr. D.Rigden, (CENARGEN/EMPRAPA, Brasilia,Brazil) indicated that the overall conformation ofthe active site of the trypanosomal enzyme isvery similar to those of the enzyme from yeastand lobster for which crystal structures areavailable. However, there are some atypicalresidues (one Lys and two Cys residues) close tothe T. brucei active site. These residues couldpossibly be exploited for the irreversible bindingof selective inhibitors. The accessibility of theseresidues for inhibitors is currently being studiedin collaboration with Dr. D. Vertommen(HORM Unit).

The activity of trypanosomatid PYK isallosterically regulated by fructose 2,6-bisphosphate (F-2,6-P2), contrary to the PYKsfrom other eukaryotes that are usually stimulatedby fructose 1,6-bisphosphate (F-1,6-P2). Themolecular basis of the specificity for theallosteric effector was studied in more detail incollaboration with Drs. D. Rigden (Brasilia) andL. Gilmore (Edinburgh)(7). Based on thecomparison of the three-dimensional structureof Saccharomyces cerevisiae PYK crystallized withF-1,6-P2 present at its effector site (R-state) andthe L. mexicana enzyme crystallized in the T-state, two residues (Lys453 and His480) wereproposed to bind the 2-phospho group of theeffector. This hypothesis was tested by site-directed mutagenesis. The allosteric activation byF-2,6-P2 appeared to be entirely abrogated in themutated enzymes confirming our predictions. Inaddition, we have prepared two mutants for useas tools to screen the large number ofcompounds we anticipate from ligand docking,database mining and combinatorial chemistry.Wild-type trypanosomatid PYK has notryptophan residues, and we have introducedthis residue into two different positions near theeffector site (F442W and E451W). Both mutantsshow fluorescence quenching in response tosubstrates and effectors, and will thus play animportant role in screening combinatoriallibraries.

Figure 1. Proposed scheme for the first two steps of the HMP. Glucose 6-phosphate (G-6-P) oxidation byG6PDH leads to the formation of d-6-phosphogluconolactone (d-6-P-G-L). The two 6-P-G-Ls are in exchange,characterized by ka, the rate constant of the conversion of d-6-P-G-L into g-6-P-G-L, and kb, for the reversereaction. No spontaneous hydrolysis of g-6-P-G-L was measured.

ka= 0.028 min-1

kb= 0.021 min-1

OH OH O

OH OH

OH

-O3PO 12

34

5

6

OH

HO

HOO

O

OPO3-

65

4

3

2

1

6

5

4

32

1OH

O

-O3PO

HO

HOHO

6

5

32

1

O

-O3PO

HO

HOHO

O

6

5

4

32

1

OH

O

-O3PO

HO

HOHO

k2= 0.010 min-1 k1= 0 min-1

aG-6-P

bG-6-P d6PGL

6PGL

g6PGL

6PGA

NADP+ NADPH+H+

G6PDH 4

43

Enzymes of the hexosemonophosphate pathway (HPM)

Glucose-6-phosphate dehydrogenase and 6-phosphogluconolactonase

E. Saavedra, P. Michels and F. Opperdoes, incollaboration with F. Duffieux, Université de Paris Sud,France

Previously, we reported the cloning andcharacterization of T. brucei glucose-6-phosphatedehydrogenase (G6PDH) and 6 -phosphogluconolactonase (6PGL). We havenow also cloned, characterized and expressedthe corresponding genes of L. mexicana.

In collaboration with Dr. Duffieux (Paris) thethree-dimensional structure of the T. brucei6PGL is being solved using the technique ofnuclear magnetic resonance. So far, the role ofthis enzyme in metabolism was still questionable,because 6-phosphogluconolactones werebelieved to undergo rapid spontaneoushydrolysis. By using both 13C and 31P-nuclearmagnetic resonance spectroscopy we havecharacterized the chemical scheme and kineticfeatures of the oxidative branch of this pathway(Fig. 1) (8). The d form of the lactone is the onlyproduct of glucose 6-phosphate oxidation. Itleads to the spontaneous formation of the gform by intramolecular rearrangement.However, only the d form undergoesspontaneous hydrolysis, the g form being a“dead end” of this branch. The d form is theonly substrate for 6PGL. Therefore, the activityof this enzyme accelerates hydrolysis of the dform, thus preventing its conversion into the gform. Furthermore, 6PGL guards against theaccumulation of d-6-phosphogluconolactone,which may be toxic through its reaction withendogenous cellular nucleophiles.

The presence of plant traits in theTrypanosomatidae

V. Hannaert, J.-P. Szikora, P. Michels andF.!Opperdoes, in collaboration with D. Rigden,CENARGEN/EMPRAPA, Brasilia, Brazil

While searching for fructose-1,6-bisphosphatasein trypanosomatid genome databases, we haverecently identified in T. brucei a complete open-reading frame encoding a homologue of are l a ted enzyme , sedoheptu lose -1 ,7 -

bisphosphatase (SBPase). The gene has beencloned and sequenced and contains a PTS1,making it a glycosomal protein. SBPase is anenzyme typical of the Calvin cycle ofphotosynthetic organisms and hence (so far)only encountered in the chloroplasts of greenalgae and plants. Phylogenetic analysis showsthat the closest affiliation of the trypanosomeenzyme is with that of the chlorophyteChlamydomonas reinhardtii.

Our recent observation that the glycosomalfructose-bisphosphate aldolase is also closelyrelated to its homologues from plants, which allhave a broad substrate specificity and are able tosynthesize (and cleave) sedoheptulose 1,7-bisphosphate, suggests to us that these twoenzymes must function in tandem in thetrypanosomatid HMP. A closer inspection ofother genes available in the trypanosomegenome database revealed many more sequenceswith either plant or chloroplast/cyanobacterialaffiliation. Most of these enzymes fulfil usuallyfunctions in either the Calvin cycle, in glycolysisor in the HMP. We hypothesize that these genesof carbohydrate metabolism probably enteredthe trypanosomatid ancestor from an algalendosymbiont and that their gene products havelater been relocated from the endosymbiont tothe host, often to its peroxisomes, after whichthe remainder of the endosymbiont was lost (9).This may explain the enigmatic presence ofglycosomes (i.e. peroxisomes specialized incarbohydrate metabolism) in these organisms.

Eubacteria

Animals

Fungi

Eukaryota

Plants

ProtistsKinetoplastidaEuglena

Archaebacteria

AlgaeCilates

Acquisition of algal endosymbiont

Loss of chloroplast

Figure 2. The "tree of life" based on 16Sribosomal RNA sequences, as modified fromSogin.Indicated are the supposed acquisition of an algalendosymbiont by an organism ancestral to botheuglenoids and kinetoplastids. The subsequent loss ofthis endosymbiont (and its chloroplast) from thekinetoplastid (Trypanosoma and Leishmania) lineagestook place after their separation from the euglenoids,which still have chloroplasts.

44

Analysis of the control of theglycolytic flux

M.-A. Albert and P. Michels in collaboration with B.Bakker and H. Westerhoff, Vrije UniversiteitAmsterdam, The Netherlands, and S. Helfert andC.!Clayton, Universität Heidelberg, Germany

Previously, a mathematical model oftrypanosome glycolysis was developed based onthe kinetic data available for the enzymesinvolved. This model was able to predictsuccessfully the experimentally determinedfluxes and metabolite concentrations intrypanosomes. Our present experiments focuson the experimental determination of the fluxcontrol coefficients of the various steps ofglycolysis either by their titration with inhibitors,or by the regulation of the expression of thegenes coding for the respective glycolyticenzymes. To this end T. brucei cell lines havebeen created in which the expression of HXK,PFK, TIM and glycerol-3-phosphate oxidase(GPO), the mitochondrial enzyme responsiblefor oxidising the glycolytically produced NADH,can be decreased by RNA interference (RNAi).For that purpose, double-stranded RNAcorresponding to the mRNA is synthesised froma transgene under the control of a promoter thatis regulated by exogenously added tetracycline.Using these cell lines, it was shown that thegrowth rate is halved when the TIM level isdecreased to 15%; lower TIM levels are lethal.Indeed, simulation by the model predicted thatthe flux decreases when the enzyme activitydrops to about 30%. Upon lowering the level ofGPO mRNA, the oxygen consumption wasreduced 4-fold and the rate of trypanosomegrowth was halved. Similarly, reducing HXKexpression by RNAi leads to reduction of theglycolytic flux and the growth rate of the cells.Currently, experiments are in progress toestablish the quantitative relationship betweenHXK and PFK expression on the one hand, andthe flux and growth rate on the other hand.Similar experiments are being performed withconditional knockout cells which have beencreated by disruption of the endogenous HXKand PFK genes after the introduction of annewly introduced additional gene copy under thecontrol of an inducible promoter. Theseexperiments will also reveal if the in vitrodetermined kinetic properties of these enzymesapply also to in vivo conditions, and/or ifadditional regulatory mechanisms occurring inthe intact cell have to be invoked.

Biogenesis of glycosomes

J. Moyersoen, H. Krazy, V. Hannaert and P. Michels

Glycosomal matrix proteins are synthesized inthe cytosol and imported post-translationally.The translocation of these matrix proteins acrossthe peroxisomal membrane involves a variety ofperoxins. Inhibitors interfering with peroxininteractions in trypanosomatids are expected toprevent the synthesis of functional glycosomesand thus kill the parasites. The design ofselective inhibitors seems feasible because of thevery low level of conservation of peroxins.Previously, we reported the cloning andcharacterization of two cytosolic peroxins, Pex5and Pex7. These peroxins act as receptors forproteins to be imported into the glycosomalmatrix. Pex5 recognizes the PTS1, a signalspecified by the three C-terminal amino acids(10), and Pex7 interacts with the PTS2, anonapeptide motif close to the N-terminus.Both T. brucei peroxins, have been expressed asrecombinant proteins in E. coli. The functionalidentity of the 70 kDa T. brucei Pex5 has beenestablished in vitro; the purified proteinrecognized glycosomal PGK with high affinity.

We have now also cloned and sequenced theT. brucei homologues of Pex6, 10, 12 and 14.Their sequences have 32, 21, 25 and 26%identity with the corresponding human peroxins.Pex14 is part of the receptor-docking complex atthe glycosomal membrane. Indeed, we couldshow, by in-v i t ro experiments, its specificinteraction with Pex5. The vital importance ofPex14 for bloodstream-form T. brucei wasdemonstrated by RNAi; induced expression ofdouble-stranded Pex14 mRNA resulted ingrowth arrest of the cells. Pex10 and Pex12 arepossibly involved in the translocation process ofthe receptors Pex5 and Pex7, charged with theirligand, across the organellar membrane, and/orthe dissociation of the receptor-cargo complexesat the matrix site of the membrane. Theseperoxins contain a C-terminal zinc-bindingRING domain known to be involved in protein-protein interactions. Pex6 belongs to the familyof AAA-proteins (ATPases Associated with avariety of cellular Activities). It is a large (>100kDa) protein thought to be either involved inmembrane fusion processes which lead to theformation of mature organelles or in thereceptor recycling from the organellar matrix tothe cytosol. So far, we have expressed in E. colithe separate ATPase domain of T. brucei Pex6.

45

Crystallization trials are currently beingperformed with this trypanosome Pex6 domainand with various forms of Pex14 and Pex5, andwith combinatons of these latter two peroxins.(J. Choe and Prof. W. Hol).

Analysis of glycosomal membranesolute

Transporters

C. Yernaux and P. Michels

We have started an investigation of glycosomalmembrane proteins that might be involved inthe transport of glycolytic intermediates or othersolutes across the membrane. We have clonedand sequenced two genes coding for putative T.brucei glycosomal membrane transporters(TbGAT1 and 2). The amino-acid sequencesencoded by these genes are only 30% identical toeach other. They are so-called half ABCtransporters, containing only a single ATP-binding cassette in their C-terminal half. Theyare homologous to peroxisomal ABCtransporters. Segments of both polypeptideshave been expressed in E. coli and are beingpurified; they will be used for antiseraproduction. Investigation of the subcellularlocalisation and topology of the transporters is inprogress. These studies involve transfection oftrypanosomes with constructs encoding fusionsof segments of the transporters and fluorescentreporter proteins.


1. Michels PA, Hannaert V, Bringaud F. Metabolicaspects of glycosomes in trypanosomatidae - new data andviews.Parasitol Today 2000;16:482-9.

2. Verlinde CL, Hannaert V, Blonski C, WillsonM, Perie JJ, Fothergill-Gilmore LA, OpperdoesFR, Gelb MH, Hol WG, Michels PA. Glycolysis asa target for the design of new anti-trypanosome drugs.Drug Resist Updat 2001;4:50-65.

3. Opperdoes FR, Michels PA. Enzymes ofcarbohydrate metabolism as potential drug targets. Int JParasitol 2001;31(5-6):482-90.

4. Lopez C, Chevalier N, Hannaert V, RigdenDJ, Michels PA, Ramirez JL. Le i shman iadonovani phosphofructokinase. Gene characterization,biochemical properties and structure-modeling studies.Eur J Biochem 2002;269:3978-89.

5. Willson M, Sanejouand YH, Perie J, HannaertV, Opperdoes F. Sequencing, modeling, and selectiveinhibition of Trypanosoma brucei hexokinase.Chem Biol 2002;9:839-47.

6. Claustre S, Denier C, Lakhdar-Ghazal F,Lougare A, Lopez C, Chevalier N, Michels PA,Perie J, Willson M. Exploring the active site ofTrypanosoma brucei phosphofructokinase byinhibition studies: specific irreversible inhibition.Biochemistry 2002;41:10183-93.

7. Hannaert V, Yernaux C, Rigden DJ,Fothergill-Gilmore LA, Opperdoes FR, MichelsPA. The putative effector-binding site of Leishmaniamexicana pyruvate kinase studied by site-directedmutagenesis. FEBS Lett 2002;514:255-9.

8. Miclet E, Stoven V, Michels PA, OpperdoesFR, Lallemand JY, Duffieux F. NMR spectroscopicanalysis of the first two steps of the pentose-phosphatepathway elucidates the role of 6-phosphogluconolactonase.J Biol Chem 2001;276:34840-6.

9. Hannaert V, Saavedra E, Duffieux F, SzikoraJP, Rigden DJ, Michels PA, Opperdoes FR.Plant-like traits associated with metabolism ofTrypanosoma parasites. Proc Natl Acad SciUSA 2003;100:1067-71.

10. Kumar A, Roach C, Hirsh IS, Turley S,deWalque S, Michels PA, Hol WG. An unexpectedextended conformation for the third TPR motif of theperoxin PEX5 from Trypanosoma brucei. J MolBiol 2001;307:271-82.

46

GENETICS OF HUMAN CARDIOVASCULAR ANOMALIES,SKELETAL DISORDERS AND CEREBRAL TUMORS

Miikka VIKKULA, Member

Laurence BOON, Plastic Surgeon,Part time Research Associate

Mustapha AMYERE, Postdoctoral FellowPascal BROUILLARD, Postdoctoral FellowNicole REVENCU, Pediatrician, Genetics ResidentVirginie AERTS, Graduate StudentArash GHALAMKARPOUR, M.D., Graduate StudentMichella GHASSIBE, Graduate StudentIlse GUTIERREZ-ROELENS, Graduate StudentBrendan MCINTYRE, Graduate StudentThomas PALM, Graduate StudentVinciane WOUTERS, Graduate StudentAnne Van Egeren, TechnicianDominique COTTEM, TechnicianJean-Claude SIBILLE, Part-time Technical AssistantClaude MAUYEN, Part-time Technical AssistantLiliana NICULESCU, Part-time Secretary

The basic aim of our research is to get insights into the molecular mechanisms behind certain human diseases, andespecially to evaluate the importance of genetic variation in disease development. For many disorders, the cause isunknown, and therefore current treatments are aimed at alleviating symptoms. Identification of the primary causesas well as the modulating factors would allow to develop treatments that are more “curative” and more specific. Tothis end, we use a genetic approach. We are interested in disorders affecting the cardiovascular and the skeletalsystems, and certain cancers of the nervous system. As this research is based on human DNA extracted from bloodand tissue samples obtained from patients, the group works tightly together with several clinicians andmultidiciplinary centers worldwide (e.g. Centre des Malformations Vasculaires, Cliniques Universitaires St-Luc;Vascular Anomalies Center, Children’s Hospital, Boston, USA, Consultation des Angiomes, HôpitalLariboisière, Paris, and Centre labiopalatin, Cliniques Universitaires St-Luc).

Venous malformations,glomuvenous malformations(“glomangiomas”) and Maffuccisyndrome

P. Brouillard, M. Amyere, B. McIntyre, V. Aerts, V.Wouters, L.M. Boon and M. Vikkula, in collaborationwith B.R. Olsen, Harvard Medical School, Boston,USA; J.B. Mulliken and S. Fishman, Children’sHospital, Boston, USA and O. Enjolras, HôpitalLariboisière, Paris, France

Venous malformations (VM) are bluish-purplecutaneous and mucosal lesions. They are oftencongenital, but can appear later in life. They have

a tendency to grow slowly with the growth ofthe child. Glomuvenous malformations (GVM,“glomangiomas”) are a special subtype ofvenous anomalies. They are clinically similar toVMs, yet our recent study has allowed clinicaldifferentiation (15).

We have previously identified that hereditaryvenous malformations can be caused by anactivating mutation in the receptor tyrosinekinase TIE2/TEK (1). In contrast to inheritedVMs, inherited glomuvenous malformations donot link to the TIE2/TEK gene. Instead, theylink to VMGLOM on chromosome 1p21 (2).Characterization of the positional candidategenes led to the identification of the mutated

47

gene that we named “glomulin” (3). By screeningseveral families in which GVMs are inherited, wehave discovered that about 70% of theindividuals with inherited GVM show one offour common glomulin mutations (16). Thisallows an easly genetic diagnosis for individualsthought to have GVM. The rest, 30%, have aunique mutation.

Glomulin does not have sequence identitiesto known proteins, nor does it contain knownfunctional domains. Thus, its molecular functionis unknown. To unravel at least partially thefunction of glomulin, we have studied itsexpression. Glomulin was found in almost alltissues (4), but almost exclusively in vascularsmooth muscle cells (14).

As most of the identified mutations causepremature STOP codons in the coding sequenceof glomulin (Fig. 1), loss-of-function is the mostlikely mechanism causing GVMs (4).Furthermore, we hypothesized that as the lesionsare localized, a somatic second hit might beneeded in the normal allele of the glomulin gene,for lesions to develop. We have obtained prooffor this from one lesion (4). To further studyglomulin function, we have cloned 20kb of themurine glomulin gene. This has been used tocreate a construct for inactivating glomulin byhomologous recombination in murineembryonic stem cells. As the construct containsloxP sites within the glomulin sequence, we candecide the time and tissue for glomulindeficiency. Murine embryos containing suchdeficiency would be used to study the role ofglomulin in development and angiogenesis.

Maffucci syndrome is a rare non-hereditarydisorder characterized by venous-like cutaneouslesions associated with enchondromas andincreased risk of cancer. With world-widecollaborations we have collected samples from aseries of patients with the goal of identifyingpossible underlying genetic defects.

Lymphedema

A. Ghalamkarpour, L.M. Boon and M. Vikkula incollaboration with K. Devriendt, KUL, D. Chitayat,Hospital for Sick Children, Toronto, Canada

Primary lymphedema can occur at birth (Meige’sdisease) or at puberty (Milroy disease). It isextremely difficult to treat and the patients have

a life-time problem with progressive swelling ofextremities. To understand the pathophysiology,we have initiated genetic studies. We identified afamily in which primary lymphedema waspresent at birth in several family members.Genetic studies confirmed linkage to 5q33-34and led to the identification of a mutation in theVEGFR3 gene (5). In vitro expression studiesdemonstrated that the mutated receptor has lostits autophosphorylation capacity (5). Thecontinued studies have led to the identificationof a transcription factor gene, SOX18, to bemutated in a family with recessively inheritedcongenital lymphedema. A dominant SOX18mutation was identified in two other families,with individuals having varying degree oflymphedema. All individuals with a SOX18mutation had also a hypothrichosis. This studydescribed a new target for lymphedema therapy,the third human gene known to causelymphedema, a disorder currently without cure.

Vascular anomalies affectingcapillaries

N. Revencu, I. Eerola, L.M. Boon and M. Vikkula incollaboration with J.B. Mulliken, Children’s Hospital,Boston, USA, S. Watanabe, Showa University Schoolof Medicine, Tokyo, Japan, A. Dompmartin, CHU deCaen, France and Virginia Sybert, WashingtonUniversity, Seattle, USA

Capillaries, the smallest blood vessels thatconnect arterioles to venules, can give rise tovarious anomalies, two of which are verycommon: 1) hemangioma, a benign, localizedovergrowth of capillary-like vessels, and 2)capillary malformation (CM; commonly knownas portwine stain), a localized maldevelopmentof capillary like vessels. Hemangiomas have afrequency of up to 12 % in 1-year-old children,and CMs occur in 0,3% of newborns. Whereashemangiomas usually disappear spontaneously,capillary malformations stay throughout life, ifnot treated. Other types of cutaneous capillaryanomalies also exist. In addition, some can affectother organs, like the brain, in case of CCMs,cerebral capillary malformations.

48

Figure 1. Schematic representation of the glomulin gene summarizing all known mutations (15). Mutationsfound in more than one family are underlined. Somatic second hit is boxed.

As the molecular mechanisms leading tothese localized capillary lesions are unknown, wehave collected clinical information and samplesfrom families in which more than twoindividuals are affected. These studies led to thediscovery that inherited hyperkeratoticcutaneous capillary-venous malformations(HCCVM) associated with cerebral capillarymalformations are caused by a mutation in theKRIT1 (Krev interaction trapped 1) gene (6).This suggests that KRIT1, a possible intracellularsignaling molecule, is important not only forcerebral but also for cutaneous vasculature. Inaddition, a genome-wide linkage mapping onfamilies with inherited capillary malformationsidentified a linked locus CMC1 (7). Screening ofpositional functional candidate genes led to theidentification of mutations in the RASA1 gene, amodifier of Ras signaling pathway (9). Thisimplies that RAS signaling pathway modulatorsmay serve as a novel therapy for these patients inthe future.

Cardiopathies

I. Gutierrez-Roelens, A. Irrthum and M. Vikkula, incollaboration with T. Sluysmans, C. Ovaert, St-Luc,UCL and M. Gewillig and K. Devriendt, KUL

The cardiovascular system may also encounterdevelopmental problems affecting the heart.

These cardiac defects, cardiopathies, vary fromphysiological septal defects to life-threateningcomplex malformations. To get insight into themolecular mechanisms behind these phenotypes,we are collecting samples from families withpossibly hereditary cardiopathies. In twofamilies, in which atrial septal defect is associatedwith progressive atrioventricular conductiondefect, we identified two novel mutations in theC S X / N k x 2 . 5 gene (8), an importanttranscription factor for cardiac development.Identification of mutation carriers is crucial, as inthe few studied families the first “symptom” hassometimes been sudden death. Identification ofmutations allows genetic testing in the respectivefamilies, enabling tight follow-up and preventivepacemaker implantation.

More recently, we have used the DNA-chipbased approach to perform whole-genomelinkage analysis, and identified a possible locusfor a gene causing heterotaxia, situs inversus(Gutierrez et al., unpublished).

Cleft lip and palate

M. Ghassibé, N. Revencu, M. Vikkula, incollaboration with D. Manicourt, B. Bayet, R.Vanwijck, Ch. Verellen-Dumoulin, St-Luc, UCL

Our main project in collaboration with Centrelabio-palatin, St Luc, is to unravel the molecular

31delAA

157delAAGAA

108CÆA

107insG

IVS6+4delA

554delA+556delCCT

421insT

842delAGTT

1179delCAA

1470delTCAA

1355delT

1547CÆG

1711delGT

980delCAGAA

gatca:tagca taaat:ttgaa

gatcaGGttgaa

(IVS7-2884)-(IVS13+255)del8.4kbp+insGG

1150delAG 1292delA+1295delAAA

738insT

Glomulin

49

background of syndromic and non-syndromiccleft lip and/or palate. Numerous blood samplesof affected individuals, and their parents andsiblings have been collected. They are used forassociation studies. In addition, collaborationwith the cleft lip and palate center of the CHRULille (Prof. Ph. Pellerin) has been initiated. Thesestudies have recently led to the identification ofIRF6 mutations causing Van der Woudesyndrome (Fig. 2) (13). Based on this, we haveinitiated a collaboration with Prof. J. Murray,University of Yowa, to study IRF6 association incleft lip and palate.

Figure 2. Electropherograms of detected IRF6heterozygous mutations.Arrow, site of mutation (12).

Other cutaneous disorders

M. Amyere, L.M. Boon, M. Vikkula, in collaborationwith B. Olsen, Harvard Medical School and Th. Vogt,University of Regensburg, Germany

A peculiar pigmentation problem in someindividuals from the town of Teublitz wasrecently identified by Dr Th. Vogt, Germany. Asthis progressive hereditary hyperpigmentationaffects some individuals from the same smallvillage, and is extremely rare, it is plausible thatall the individuals carry the same ancestralmutation (12). Based on this hypothesis, we haveperformed a whole genome DNA-chip basedpolymorphic marker analysis and identified apossible locus for the mutated gene (manuscriptin preparation). This is the first step towardsidentification of the causative gene and

unravelling the pathway involved in this!mirrow-image causing pigmentation disorder.

Another clinically important cutaneousproblem is the formation of keloids in someindividuals after wounding. The tendancy tomake grossly hypertrophic scars is sometimeshereditary, and in collaboration with Prof. B.Olsen we try to identify the mutated genes tounderstand their pathophysiology.

Cerebral tumors

Th. Palm and M. Vikkula, in collaboration with C.Godfraind, Laboratory of Neuropathology, St-Luc,UCL

Morphological characterization and classificationof tumors is not always clear. Thus, better(molecular) criteria are needed. In addition, thecausative genes are often unknown. We areespecially interested in two types of cerebraltumors: oligodendrogliomas and ependymaltumours. Using DNA, extracted from formalin-fixed and paraffin-embedded tissues, we haveperformed loss-of-heterozygosity testing. Arestricted screening was performed in a numberof oligodendroglial tumours as well as in a largeseries of ependymal tumours. Foroligodendrogliomas, this allowed us to identifyand define specific histological characteristics fortumors that have lost chromosome 1p and 19qand that are known to have a preferableresponse to chemotherapy (10). This data haddirect clinical relevance. In addition, weidentif ied methylation differences inependymomas depending on their age andlocation (11).

Conclusions

With the genetic approach described, our grouphas unravelled genetic defects behind severalhuman disorders. These discoveries have createdinternational and national collaboration tounderstand the way these genetic defects alterthe function of the defective gene and therebydevelopment and organ function. Already now,this data has, in some cases, made more preciseclinical diagnosis possible, thus, directly aiding intreating better. In the more distant future, wehope all this knowledge will help develop noveland better therapies.

50


1. Vikkula M, Boon LM, Carraway KL 3rd,Calvert JT, Diamonti AJ, Goumnerov B, PasykKA, Marchuk DA, Warman ML, Cantley LC,Mulliken JB, Olsen BR. Vascular dysmorphogenesiscaused by an activating mutation in the receptor tyrosinekinase TIE2. Cell 1996;87:1181-90.

2. Boon LM, Brouillard P, Irrthum A, KarttunenL, Warman ML, Rudolph R, Mulliken JB, OlsenBR, Vikkula M. A gene for inherited cutaneous venousanomalies ("glomangiomas") localizes to chromosome1p21-22. Am J Hum Genet 1999;65:125-33

3. Brouillard P, Boon LM, Mulliken JB,Ghassibé M, Warman ML, Tan OT, Olsen BR,Vikkula M. Mutations in a novel factor glomulin areresponsible for glomuvenous malformations(«!glomangiomas!»). Am J Hum Genet 2002;7:866-874

4. Irrthum A, Karkkainen MJ, Devriendt K,Alitalo K, Vikkula M. Congenital hereditarylymphedema caused by a mutation that inactivatesVEGFR3 tyrosine kinase. Am J Hum Genet2000;67:295-301.

5. Eerola I, Plate KH, Spiegel R, Boon LM,Mulliken JB, Vikkula M. KRIT1 is mutated inhyperkeratotic cutaneous capillary-venous malformationassociated with cerebral capillary malformation. HumMol Genet 2000;9:1351-5.

6. Eerola I, Boon LM, Watanabe S, Grynberg H,Mulliken JB, Vikkula M. Locus for susceptibility forfamilial capillary malformations («!port-wine stain!»)maps to 5q. Eur J Hum Genet 2002;10:375-380.

7. Gutierrez-Roelens I, Sluysmans T, DevriendtK, Gewillig M, Vikkula M. Identification of two novelmissense mutations in the CSX/NKX2-5 gene encodinga cardiac specific transcription factor. Hum Mutation2002;20:75-76.

8. Eerola I, Boon LM, Mulliken JB, Burrows PE,Vanwijck R, Vikkula M. Capillary Malformation-Arteriovenous Malformation, a heretofore undescribedclinical and genetic entity, is caused by RASA1mutations. Am J Hum Genet 2003!; 73!:1240-

1249.

9. Godfraind C, Rousseau E, Ruchoux MM,Scaravilli F, Vikkula M. Tumour necrosis andmicrovascular proliferation are associated with 9p deletionand CDKN2A alterations in 1p/19q-deletedoligodendrogliomas. Neuropathology and AppliedNeurology 2003;29:462-471.

10. Rousseau E, Ruchoux MM, Scaravilli F,Chapon F, Vinchon M, De Smet C, GodfraindC, Vikkula M. CDKN2A, CDKN2B and p14ARF

are frequently and differentially methylated in ependymaltumors . Neuropathology and AppliedNeurology 2003;29:574-583.

11. Vogt Th, Zanardo L, Schmitz G, Orsó E,Kaminski W, Barlage S, Landthaler M, VikkulaM, Stolz W. Progressive hyperpigmentation, generalizedlentiginosis, no associated symptoms!: a rare hereditarypigmentation disorder in south-east Germany (Teublitzsyndrome). Acta Derm Venereol 2003;83:1-4.

12. Ghassibé M, Revencu N, Bayet B, Gillerot Y,Vanwijck R, Verellen-Dumoulin C, Vikkula M.Six Families with Van der Woude and/or PoplitealPterygium Syndrome: all with a mutation in the IRF6gene. J Med Genet 2004;41:e15.

13. McIntyre BAS, Brouillard P, Aerts V,Gutierrez-Roelens I, Vikkula M. Glomulin ispredominantly expressed in vascular smooth muscle cellsin the embryonic and adult mouse. GeneExpression Patterns 2004;4:351-358.

14. Boon LM, Mulliken JB, Enjolras O, VikkulaM. Glomuvenous malformation (“glomangioma”) andvenous malformation are distinct clinicopathologic andgenetic entities. Arch Dermatol, in press.

15. Brouillard P, Ghassibé M, Penington A,Boon LM, Dompmartin A, Temple K, CordiscoM, Adams D, Piette F, Harper J, Syed S,Boralevi F, Taïeb A, Danda S, Baselga E,Enjolras O, Mulliken JB, Vikkula M. Four commonglomulin mutations cause two-thirds of glomuvenousmalformations (“familial glomangiomas”): evidence for afounder effect. (submitted)

51

THEILER'S ENCEPHALOMYELITIS VIRUSThomas MICHIELS, Associate Member

Sophie DELHAYE, Graduate studentCaroline SOMMEREYNS, Graduate studentPierre RENSONNET, TechnicianMuriel MINET, Technician

Theiler’s murine encephalomyelitis virus (TMEV or Theiler’s virus) is a murine picornavirus responsible forinfections of the central nervous system. Strains of Theiler's virus have a striking ability to persist in the centralnervous system in spite of a specific cellular and humoral immune response. Persistence of the virus is associatedwith a strong inflammatory response and with lesions of primary demyelination reminiscent of those found inhuman multiple sclerosis. The genome of Theiler's virus is an 8 kb-long positive strand RNA molecule (Fig. 1).

Theiler's virus is an outstanding model to analyze the basic mechanisms of viral persistence and demyelination.i) It replicates and persists in the central nervous system in the face of a specific humoral and cellular immuneresponse. ii) It induces chronic demyelination in mice. iii) It is a natural pathogen of the mouse and allows theexperimental analysis of a natural host-pathogen interaction. iv) Its genome is cloned and can be manipulated bythe tools of molecular biology.

Our work aims at understanding how a virus can persist in the central nervous system of an immunocompetenthost, thus evading the immune response. We analyze viral and cellular determinants of tropism and persistence,notably the interferon a/b response and the inhibition of this response by Theiler's virus, in the context of thecentral nervous system.

Infection ofmacrophages

L VP

4 VP2 VP3 VP1 3A2A 2B 2C

"VPg"3B

3C 3DL*

8 100 nt

polymerase

proteasecapsidreplication

?IRES

CRE

Inhibition ofinterferon production

Figure 1. Genome of Theiler's virus. A large open reading frame encodes a 2000 amino acid-long polyprotein thatis cleaved, by autoproteolytic activity, into 12 mature proteins. An additional protein (L*) is encoded by analternative open reading frame. Translation of both ORFs is driven by an Internal Ribosome Entry Site (IRES)present in the 5' non-coding region of the genome (1). Protein L* was shown to facilitate the infection ofmacrophages and viral persistence (2). Protein L inhibits type-I interferon production (3). The role of protein 2A isunknown (4). 60 copies of proteins VP1 to VP4 assemble to form the viral capsid. 3B (also termed VPg) iscovalently linked to the 5' end of the RNA molecule during encapsidation and replication. 3C is the proteaseresponsible for most of the cleavages occurring during polyprotein processing. 3D is the RNA-dependent RNApolymerase. Proteins 2B, 2C, 3A participate in the replication complex. A replication signal has been discovered inthe VP2 coding sequence and is denoted CRE for "cis-acting replication element" (5).

52

Analysis of viral proteins involved inTheiler's virus escape of the hostimmune response

S. Delhaye, C. Sommereyns, and T. Michiels

The viral capsid

The viral capsid was found to play an importantrole in determining the pathology induced by thevirus. For instance, exchanging the capsid ofneurovirulent and persistent Theiler's virusstrains resulted in phenotype swapping, thoughnot complete.

The isolation of cellular mutants that becameresistant to the neurovirulent GDVII virus, butthat retained susceptibility to the persistent DAvirus, showed that the capsids of neurovirulentand persistent strains interacted with differentreceptors or co-receptors on the target cell (6).

Despite many efforts, the cellular receptorfor Theiler's virus has not yet been identified.Persistent strains but not neurovirulent strains ofTheiler's virus were found to bind sialic acid.Interaction with sialic acid involves proteinloops exposed at the surface of the viral capsid,which were found to modulate viral tropism aswell as antigenicity and pathogenesis (7).

Proteins L and L*

Two viral proteins, namely L and L* were foundto be crucial for persistence of the virus in thecentral nervous system though they were notrequired for replication of the virus in cellculture. Hence, these proteins are believed tointeract with host factors in vivo and tocounteract the host immune defenses.

Inhibition of type-I interferonproduction by the leader protein

S. Delhaye, C. Sommereyns, and T. Michiels

The leader (L) protein encoded by Theiler's virusis a 76 amino acid-long peptide containing azinc-binding motif. We showed previously thatthe L protein could inhibit production of type-Iinterferons (IFNs) by infected cells (3). Mutationof the zinc-finger was sufficient to abolish theanti-IFN activity of the L protein in vitro and todramatically impair viral persistence in the

central nervous system of SJL/J mice. However,IFN production inhibition was not complete invivo. Modulation rather than blockade of theIFN response might be viewed as a better viralstrategy toward long-term persistence in thehost.

Inhibition of IFN production was found tooccur at the transcriptional level (8). Thus, apotential target of the leader protein is IRF-3, afactor known to be required for transcritpionalactivation of IFN genes. IRF-3 is present in thecytoplasm of non-infected cells. Upon viralinfection, IRF-3 is activated and translocated tothe nucleus where it activates the transcriptionof the IFN genes.

We recently showed that the leader proteininterfered with nucleo-cytoplasmic trafficking ofhost cell proteins, and notably of IRF-3 (9).Perturbaation of nucleo-cytoplasmic traffickingby viruses could be viewed as a fast way toinhibit early cell defense mechanisms, includingthe production of cytokines.

Influence of the L* protein onmacrophage infection and viralpersistence

Persistent strains of Theiler’s virus produce an18 KDal protein called L*, which is encoded byan open reading frame (ORF) overlapping theORF coding the viral polyprotein (see Fig. 1).This protein was shown to enhance the infectionof macrophage cell lines and to be critical forpersistence of the virus in the central nervoussystem of the mouse (2).

We analyzed the phenotype of L* mutantviruses carrying either an AUG to ACGmutation of the initiation codon or a stop codonmutation introduced in the L* ORF. Our datasuggest that L* can be expressed from an ACGinitiation codon. Thus, neurovirulent strainswhich have ACG instead of AUG, can alsoproduce the L* protein. This would be the firstexample of picornavirus IRES-driven non-AUGtranslation initiation (2).

Current efforts aim at characterizing thefunction of the L* protein in the infection ofmacrophages. Macrophages are indeed keyplayers in the demyelinating disease induced byTheiler's virus, being simultaneously effectors of

53

the immune response and targets of viralinfection.

Characterization of the type-I IFNresponse

C. Sommereyns, S. Delhaye, and T. Michiels

Mouse and human genomes carry more than adozen of genes coding for closely relatedinterferon-alpha (IFN-a ) subtypes. Theseinterferons as well as other type-I interferons,like IFN-b , IFN-k , IFN-e , and limitin, arethought to bind the same receptor, raising thequestion of whether they possess specificfunctions.

As some confusion existed in the identity andcharacteristics of mouse IFN-a subtypes,availability of data from the mouse genomesequence prompted us to characterize themurine IFN-a family. A total of 14 IFN-a geneswere detected in the mouse genome, in additionto three IFN-a pseudogenes (10).

All IFN-a subtypes were found to be stableat pH2 and to exhibit antiviral activity.Interestingly, some IFN subtypes showed higherbiological activity than others. Most murineI F N - a turned out to be N-glycosylated.However, no correlation was found between N-glycosylation and activity.

The various IFN-a subtypes displayed agood correlation between their antiviral andantiproliferative potencies, suggesting that IFN-a subtypes did not diverge primarily to acquirespecific biological activities, but probablyevolved to acquire specific expression patterns.However, the same set of IFN genes wasactivated in L929 cells in response to differentstimuli such as poly(I-C) transfection and viralinfection (10).


1. Shaw-Jackson C, Michiels T. Absence of internalribosome entry site-mediated tissue specificity in thetranslation of a bicistronic transgene. J Virol 1999;73:2729-38.

2. van Eyll O, Michiels T. Non-AUG-initiatedinternal translation of the L* protein of Theiler's virusand importance of this protein for viral persistence. JVirol 2002;76:10665-73.

3. van Pesch V, van Eyll O, Michiels T. The leaderprotein of Theiler's virus inhibits immediate-earlyalpha/beta interferon production. J Virol 2001;75:7811-7.

4. Michiels T, Dejong V, Rodrigus R, Shaw-Jackson C. Protein 2A is not required for Theiler'svirus replication. J Virol 1997;71:9549-56.

5. Lobert PE, Escriou N, Ruelle J, Michiels T. Acoding RNA sequence acts as a replication signal incardioviruses. Proc Natl Acad Sci USA . 1999;96:11560-5.

6. Jnaoui K, Michiels T. Analysis of cellular mutantsresistant to Theiler's virus infection: differential infectionof L929 cells by persistent and neurovirulent strains. JVirol 1999;73:7248-54.

7. Jnaoui K, Minet M, Michiels T. Mutations thataffect the tropism of DA and GDVII strains ofTheiler's virus in vitro influence sialic acid binding andpathogenicity. J Virol 2002;76:8138-47

8. van Pesch V, Michiels T. Characterization ofinterferon a13, a novel constitutive murine interferonsubtype. J Biol Chem 2003;278:46321-28

9. Delhaye D, van Pesch V, Michiels T. The leaderprotein of Theiler's virus interferes with nucleo-cytoplasmictrafficking of cellular proteins. J Virol 2004;78:4357-62.

10. van Pesch V, Lanaya H, Renauld JC, MichielsT. Characterization of the murine alpha interferon genefamily. J Virol 2004;78: in press

54

ANTIVIRAL IMMUNITY AND PATHOGENESISJean-Paul COUTELIER, Associate Member

Andrei MUSAJI, Graduate StudentThao LE THI PHUONG, Graduate StudentGaëtan THIRION, Graduate Student (from october)Maria Dolores GONZALEZ, Technician (untillnovember)Thérèse BRIET, Technician (part time)Nadia OULED ADDOU, Technician (from november)Anick DELCHAMBRE, SecretaryClaude MAUYEN, Technical Staff (part time)

The possibility for evoluted organisms to survive viral infections depends on the ability of their immune system toeliminate the infectious agent. Therefore, numerous mechanisms, involving different types of immune cells such ascytolytic lymphocytes, T helper and B lymphocytes and macrophages, the molecules that allow those cells tocommunicate, namely the lymphokines, and the products of those interactions, including antibodies, have beenelaborated. On the other hand, viruses have developed strategies to escape the immune system of their hosts, such asfrequent mutations or latency, or even to impair this system, which often leads to diseases such as autoimmunity orimmune deficiencies. Our project is to investigate, in murine models, some aspects of these relations between virusesand the immune system.

Viral infections result in a dramaticincrease in the proportion of IgG2a

Of particular interest is the fact that all antibodyresponses are not equal. Indeed, depending ontheir isotype, immunoglobulins display variousproperties. For example, IgG1, one of the majorIgG subclass in mice, cannot activate thecomplement system, in contrast to IgG2a,another major i sotype of murineimmunoglobulins. Such a difference may lead todramatic variations in the functional effect ofantibodies, as their ability to lyse cells they havebound. During the last few years, we found thatthe isotype of antibody responses was influencedby concomitant viral infections. The effect ofthe virus resulted in a dramatic increase in theproportion of IgG2a, not only in antiviralantibodies, but also in immunoglobulins with anantigenic target unrelated to viral proteins. Adual regulation of antibody responses bygamma-interferon and interleukin-6 explains thisisotypic bias (1, 2). In the case of antiviralantibodies, a possible explanation for thisphenomenon could be the selection by theinfected host of the most appropriate responseagainst the virus. Using a model of infectionwith lactate dehydrogenase-elevating virus(LDV), we could demonstrate that IgG2a

antiviral antibodies are indeed more efficientthan other isotypes to protect mice against afatal polioencephalomyelitis induced by the virus(3, 4). The advantage for the host to selectIgG2a in non-antiviral responses is moredifficult to understand. In addition, themodification of the isotype of antibodiesreacting with self-antigens could potentially leadto more deleterious autoimmune reactions. Thisproperty of viruses to enhance selectively theproduction of one immunoglobulin isotypecould depend on the preferential activation of asubset of T helper lymphocytes (5, 6). Indeed,different subpopulations of those cells, calledTh1 and Th2, respectively, are distinguished inparticular by their capability of producingselectively interferon-g (IFN-g) or interleukin-4,which can selectively trigger B lymphocytes toproduce IgG2a or IgG1, respectively.

Activation of natural killer cells

Many of the influences that viruses may have ondiverse immune responses can be explained bythe production of pro-inflammatory cytokines(7), including IFN-g. Therefore, our analysis ofthe relationship between viruses and the immunesystem has focused on the activation, by LDV,of cells from the innate immune system that areable to secrete this cytokine, namely the natural

55

killer (NK) cells. Within a few days afterinfection, a strong and transient NK cellactivation, characterized by accumulation of thiscell population in the spleen, by enhanced IFN-g message expression and production, as well asby cytolysis of target cell lines was observed.Because NK cells and IFN-g may participate inthe defense against viral infection, we analyzedtheir possible role in the control of LDV titers,with a new agglutination assay. Our resultsindicate that neither the cytolytic activity of NKcells nor the IFN-g secretion affects the earlyand rapid viral replication that follows LDVinoculation (8).

Activation of macrophages

Activation of cells of the innate immune systemincludes also macrophages and leads to anenhanced phagocytic activity, with potentialdetrimental consequences for ongoingautoimmune diseases (9). We have thus analysedwhether it was possible to modulate such anactivation by treating infected mice withc l o d r o n a t e - c o n t a i n i n g l i p o s o m e s .Administration of anti-erythrocyte monoclonalautoantibody to mice resulted in thedevelopment of a transient hemolytic anemia.Infection with LDV simultaneously withautoantibody injection was followed by adramatic enhancement of the anemia, leading tothe death of most animals. This viral infectioninduced an increase in the ability ofmacrophages to phagocytose in vitroautoantibody-coated red cells, and anenhancement of erythrophagocytosis in the liver.Treatment with total immunoglobulin G (IVIg)attenuated the autoantibody-induced disease inuninfected mice, but not in LDV-infectedanimals (10). In contrast, administration ofclodronate-containing liposomes resulted in adelay and a decrease of anemia in LDV-infectedmice. This treatment decreased also the in vitrophagocytosis of autoantibody-coated red cells bymacrophages from LDV-infected animals.Thus, regulation of macrophage activationresults in modulation of autoantibody-mediatedanaemia and may be considered as a possibletreatment for autoimmune diseases that involvephagocytosis as a pathogenic mechanism.

Autoimmune thrombocytopenia

To further analyse this effect of viral infectionsthat could be relevant to human pathology, anew model of autoimmune disease, againstplatelets, was developped (11). Immunization ofCBA/Ht mice with rat platelets was followed bya transient thrombocytopenia. Platelets wereopsonized by autoantibodies that recognizedboth rat and mouse normal platelets.Absorption experiments indicated that theseautoantibodies reacted with epitope(s) shared byrat and mouse platelets. In contrast, BALB/Cmice similarly immunized with rat platelets didnot develop thrombocytopenia. The ability ofBALB/C mice to produce anti-rat plateletantibodies and to eliminate antibody-coatedplatelets was comparable with that of CBA/Htanimals. However, the specificity of theantibody response elicited in these two mousestrains differed markedly, with a 145 kDa mouseplatelet antigen recognized in CBA/Ht, but notin BALB/C animals. This immunizationprotocol may serve as a model of anti-plateletautoimmune response, and especially of post-transfusion purpura.


1. Markine-Goriaynoff D, van der Logt JTM,Truyens C, Nguyen TD, Heessen FWA,Bigaignon G, Carlier Y and Coutelier J-P IFN-g-independent IgG2a production in mice infected withviruses and parasites. Intern Immunol 2000;12:223-30.

2. Markine-Goriaynoff D, Nguyen TD,Bigaignon G, Van Snick J and Coutelier J-PDistinct requirements for IL-6 in polyclonal and specificIg production induced by microorganisms. InternImmunol 2001;13:1185-92.

3. Monteyne P, Coulie PG and Coutelier J-PAnalysis of the Fv1 alleles involved in the susceptibility ofmice to lactate dehydrogenase-elevating virus-inducedpolioencephalomyelitis. J Neurovirol 2000;6: 89-93.

4. Markine-Goriaynoff D and Coutelier J-PIncreased efficacy of the immunoglobulin G2a subclass inantibody-mediated protection against lactated e h y d r o g e n a s e - e l e v a t i n g v i r u s - i n d u c e dpolioencephalomyelitis revealed with switch mutants. JVirol 2002;76:432-35.

56

5. Monteyne P, Van Broeck J, Van Snick J andCoutelier J-P. Inhibition by lactate dehydrogenase-elevating virus of in vivo interleukin-4 production duringimmunization with keyhole limpet haemocyanin.Cytokine 1993;5:394-97.

6. Monteyne P, Renauld J-C, Van Broeck J,Dunne DW, Brombacher F and Coutelier J-PIL-4-independent regulation of in vivo IL-9 expression. JImmunol 1997;159: 2616-23.

7. Coutelier J-P, Van Broeck J and Wolf SFInterleukin-12 gene expression after viral infection in themouse. J Virol 1995;69:1955-58.

8. Markine-Goriaynoff D, Hulhoven X,Cambiaso CL, Monteyne P, Briet T, GonzalezM-D, Coulie P and Coutelier J-P. Natural killer

cell activation after infection with lactate dehydrogenase-elevating virus. J Gen Virol 2002;83:2709-16.

9. Meite M, Léonard S, El Azami El Idrissi M,Izui S, Masson PL and Coutelier J-P. Exacerbationof autoantibody-mediated hemolytic anemia by viralinfection. J Virol 2000;74:6045-49.

10. Pottier Y, Pierard I, Barclay A, Masson PL.and Coutelier J-P. The mode of action of treatment byIgG of haemolytic anaemia induced by an anti-erythrocytemonoclonal antibody. Clin Exp Immunol 1996;106:103-07.

11. Musaji A, Vanhoorelbeke K, Deckmyn Hand Coutelier J-P. New model of transient strain-dependent autoimmune thrombocytopenia in miceimmunized with rat platelets. Exp Hematol 2004;32:87-94.

57

SIGNAL TRANSDUCTION AND GENE REGULATION BYPLATELET-DERIVED GROWTH FACTORSJean-Baptiste DEMOULIN, Assistant Member

Anders KALLIN, Post-doctoral Fellow (Lacroixfellowship)Catherine MARBEHANT, StudentAnick DELCHAMBRE, Secretary

Platelet-derived growth factors (PDGF) are hormone-like proteins that stimulate cell proliferation and migrationduring wound healing and the development of the embryo and. PDGF acts through two specific receptor tyrosinekinases, the PDGF a- and b-receptors, which are targeted by several anticancer drugs. Uncontrolled PDGFproduction or receptor activation is indeed associated with the development of cancers such as dermatofibrosarcoma,glioblastoma and certain types of leukemia.

The growth factor group, which started in collaboration with Prof. Carl-Henrik Heldin at the LudwigInstitute for Cancer Research (Uppsala, Sweden), focuses on the mechanism of action of PDGF. The cytoplasmicdomain of the PDGF receptors contains a tyrosine kinase domain that is activated upon PDGF binding andphosphorylates signal transduction molecules. Our present lines of research are : (1) the role of SREBPtranscription factors in gene regulation by PDGF in fibroblasts and tumor cells, (2) the regulation of the PDGFb-receptor activity by its C-terminal tail and the NHERF protein, and (3) the role of Gab1 in PDGF signaltransduction and the reorganization of the cell cytoskeleton, as illustrated below.

58

Gene regulation by PDGF

We analyzed the transcriptional program elicitedby stimulation of normal human fibroblasts withPDGF using cDNA microarrays produced bythe Sanger/Ludwig/CRUK consortium. Weidentified 103 significantly regulated transcriptsthat had not been previously linked to PDGFsignaling. Among them, a cluster of genesinvolved in fatty acid and cholesterolbiosynthesis, including stearoyl-CoA desaturase(SCD), fatty acid synthase and hydroxy-methylglutaryl-CoA synthase (HMGCS), wasup-regulated by PDGF after 24 h of treatment,and their expression correlated with increasedmembrane lipid production. All these genes areknown to be controlled by sterol regulatoryelement-binding proteins (SREBP). PDGFincreased the amount of mature SREBP-1, andregulated the promoters of SCD and HMGCS ina SREBP-dependent manner. In line with theseresults, blocking SREBP processing by additionof 25-hydroxycholesterol blunted the effects ofPDGF on lipogenic enzymes. SREBP activationwas dependent on the phosphatidylinositol 3-kinase (PI3K) pathway, as judged from theeffects of the inhibitor LY294002 and mutationof the PDGF b-receptor tyrosines that bindp85. Fibroblast growth factors (FGF-2 andFGF-4) mimicked the effects of PDGF onNIH3T3 and human fibroblasts. In conclusion,our results suggest that growth factors inducemembrane lipid synthesis via the activationSREBP and PI3K (1). The role of SREBP ingrowth factor signaling and tumour developmentwill be further analyzed in the near future.

Regulation of the PDGF receptorkinase activity by its C-terminal tailand NHERF

In an effort to better understand the activationof the PDGF receptor tyrosine kinase, weinvestigated the role of the C-terminal tail of thePDGF b-receptor in the control of the receptorkinase activity. First, we showed that NHERF isrecruited to the C-terminal end of the receptor(2). By contrast to published data, NHERFbinding did not modify the receptor activity.Using a panel of PDGF b-receptor mutants withprogressive C-terminal truncations, we observedthat deletion of the last 46 residues, whichcontain a proline- and glutamic acid-rich motif,increased the activity of the receptor in the

absence of ligand, compared to wild-typereceptors (3). By contrast, the kinase activity ofmutant and wild-type receptors that were pre-activated by treatment with PDGF wascomparable. Using a conformation-sensitiveantibody, we found that truncated receptorspresented an active conformation even in theabsence of PDGF. A soluble peptide containingthe Pro/Glu-rich motif specifically inhibited thePDGF b -receptor kinase activity. Whereasdeletion of this motif was not enough to conferligand-independent transforming ability to thereceptor, it dramatically enhanced the effect ofthe weakly activating D850N mutation in a focusformation assay. These findings indicate thatallosteric inhibition of the PDGF b-receptor byits C-terminal tail is one of the mechanismsinvolved in keeping the receptor inactive in theabsence of ligand.

Role of Gab1 in PDGF signaltransduction

Gab1 is a scaffolding/docking protein that hasbeen suggested to play a role in signaltransduction downstream of platelet-derivedgrowth factor (PDGF) receptors. We found thatPDGF induced a rapid Gab1 phosphorylation,which depended on the recruitment of Grb2,indicating that Grb2 acts as a bridge betweenGab1 and the PDGF b-receptor. PDGF alsoenhanced the binding of Gab1 to thephosphatase SHP-2, but not to p85. To furtherstudy the role of Gab1 in PDGF signaling, wetransfected porcine aortic endothelial cells with adoxycyclin-inducible Gab1 construct. IncreasedGab1 expression enhanced the recruitment andactivation of SHP-2, as well as thephosphorylation of the mitogen-activatedprotein kinases Erk and p38 by PDGF. Gab1expression also enhanced the formation oflamellipodia and cellular protrusions. In Gab1-deficient mouse embryonic fibroblasts, the samephenotype was induced by restoring theexpression of wild-type Gab1, but not a mutantGab1 that was unable to associate with SHP-2.Finally, Gab1-deficient fibroblasts showed adecreased chemotactic response towardsgradients of PDGF as compared to wild-typecells. In conclusion, Gab1 plays a selective rolein regulation of the mitogen-activated proteinkinases Erk and p38 downstream of the PDGFb -receptor, and contributes to cytoskeletalreorganization and chemotaxis in response toPDGF (4).

59

Selected publications:

1. Demoulin JB, Ericsson J, Kallin A, RorsmanC, Rönnstrand L and Heldin CH. Platelet-derivedgrowth factor stimulates membrane lipid synthesis throughactivation of phosphatidylinositol 3-kinase and sterolregulatory element-binding proteins. J Biol Chem2004; In press and published online.

2. Demoulin JB, Seo J, Ekman S, GrapengiesserE, Hellman U, Ronnstrand L, Heldin CH.Ligand-induced recruitment of Na+/H+-exchangerregulatory factor to the PDGF (platelet-derived growthfactor) receptor regulates actin cytoskeleton reorganizationby PDGF. Biochem J 2003;376:505-10.

3. Chiara F, Heldin CH and Demoulin JB.Autoinhibition of the PDGF b-receptor tyrosine kinaseby its C-terminal tail. J Biol Chem 2004;279:19732-8.

4. Kallin A, Demoulin JB, Nishida K, Hirano T,Ronnstrand L, Heldin CH. Gab1 contributes tocytoskeletal reorganization and chemotaxis in response toplatelet-derived growth factor. J Biol Chem 2004;279:17897-9

60

HUMAN TUMOR IMMUNOLOGY*Pierre G. COULIE, Member

Wolfram FINK, Post-doctoral fellowTomoko SO, Post-doctoral fellowThierry CONNEROTTE, Graduate studentVéronique CORBIERE, Graduate studentYgierne DODOO, Graduate studentThérèse AERTS, TechnicianGérald HAMES, TechnicianCatherine MULLER, TechnicianSuzanne DEPELCHIN, Secretary

In association with the Ludwig Institute!:see Analysis of T cell responses of vaccinated cancerpatients (research at LICR, Brussels)

Human tumor cells bear antigens that are not present on normal cells and that can be specifically recognized byautologous CD4 or CD8 T lymphocytes. We contributed to the identification of several tumor-specific antigens,present on melanoma or lung carcinoma cells (1-5). Tumor-specific antigens, such as those encoded by the MAGEgenes, have been used to vaccinate melanoma patients with detectable disease. About 20 % of the vaccinatedpatients display a tumor regression, a frequency that appears well above the level reported for spontaneousmelanoma regressions. Nevertheless, the treatment fails in most patients, and this can probably only be improved ifa better understanding of the immune responses of the patients is acquired. Our group focuses on such analyses.

Correlation between tumorregression and T cell responses inmelanoma patients vaccinated witha MAGE antigen

V. Corbière, T. Connerotte, Y. Dodoo in collaborationwith P. van der Bruggen, D. Colau, D. Godelaine, A.Van Pel, and N. van Baren, Brussels branch of theLudwig Institute for Cancer Research, K. Thielemans,Laboratory of Physiology, Vrije Universiteit Brussel, andG. Schuler, Department of Dermatology, UniversityHospital of Erlangen, Germany.

Shared tumor-specific antigens encoded bycancer-germline genes such as those of theMAGE family have been used for therapeuticvaccination of cancer patients. A number ofsmall clinical trials on metastatic melanomapatients have been performed with the MAGE-3antigenic peptide EVDPIGHLY, which ispresented by HLA-A1. Evidence of tumorregression was observed in about 20% of thepatients, but clinical benefit was limited to about10% of the patients. As these vaccination trialshave not included randomization with an

untreated arm, one cannot rigorously exclude thepossibility that the regressions observed are notdue to the vaccine. Occasional spontaneousregressions of melanoma have been observed,but the frequency reported is at least 20 timeslower than the frequency observed in thevaccination trials.

Our initial work suggested that in mostvaccinated patients, even in those who displayedtumor regression, it was difficult to ascertain theexistence of an anti-vaccine T cell response. Wenevertheless felt that it was crucial to knowwhether or not low-level responses had occurredand whether such cytolytic T lymphocytes (CTL)responses showed a correlation with tumorregression, in order to understand why mostpatients failed to show any evidence ofregression. We therefore developed a sensitiveapproach based on in vitro restimulation ofblood lymphocytes with the antigenic peptideover two weeks, followed by labeling withtetramers. To evaluate precursor frequencies,these mixed lymphocyte-peptide cultures wereconducted under limiting dilution conditions.

61

Cells that were labeled with the tetramer werecloned, the lytic specificity of the clones was

verified, and their diversity was analyzed by Tcell receptor (TCR) sequencing (3, 6-8).

PBMC+ MAGE-3.A1 peptide+ IL-2 + IL-4 + IL-7

+ MAGE-3.A1 peptide+ IL-2 + IL-4 + IL-7

day 0 day 7 day 14

HLA-A1 + EBV-B

EBV-B + peptide K562

MZ2-MEL+ HLA-A1+ MAGE-3

80

40

0

1 3 10 30 1 3 10 30 1 3 10 30 3 10 301

% S

pec

ific

lysi

sEffector to target ratio

A1 / MAGE-3 tetramer

Con

trol

A1

tetr

amer

Number ofnegative

microcultures

Number ofpositive

microcultures

frequency

T cell receptor gene sequencing

presence of repeated clonotypes

1 cell/well

tetramer test

lysis assay

Figure 1. Overview of the procedure to evaluate CTL responses of vaccinated patients. Typically,250,000 peripheral blood mononuclear cells (PBMC) were distributed in each microwell.

As the interpretation of these analyses wasbased both on the frequency and on the diversityof the anti-MAGE-3.A1 CTL clones, it wasnecessary to evaluate these parameters in a non-cancerous individual. Such a study wasundertaken by the group of P. van der Bruggenand the results indicate that the frequency of thenaïve anti-MAGE-3.A1 CTL precursors is about4x10-7 of the blood CD8 T lymphocytes (9).

To evaluate the diversity of this naïve anti-MAGE-3.A1 repertoire, we examined the TCRsequences of the 23 anti-MAGE-3.A1 CTLclones obtained in this individual. Among thetotal of 23 clones, only two had the same TCRαand β sequences, indicating that the size of theanti-MAGE-3.A1 repertoire is very likely to beabove 50, with a maximum likehood at 250. Onthe basis of this evaluation of the diversity of theanti-MAGE-3.A1 naïve TCR repertoire, it ispossible to ascertain low level CTL responses,provided TCR sequence analysis indicates thatthe same clonotypes are obtained in severalindependent microcultures.

Table 1 presents a summary of the anti-MAGE-3.A1 CTL responses that we observedin patients who showed evidence of tumorregression and in patients who did not. Amongpatients vaccinated with ALVAC-MAGE, arecombinant poxvirus containing a minigeneencoding the MAGE-3.A1 peptide, a CTLresponse was found in 3/4 patients who showedregression and 1/11 patients who did not.Among patients vaccinated with dendritic cellsloaded with peptide MAGE-3.A1 (G. Schuler,Erlangen), a CTL response was found in 3/3patients who showed regression and 0/3 patientswho did not. These are statistically significantcorrelations. Only one out of seven patientsvaccinated with the MAGE-3.A1 peptide,showed a CTL response. The correlationbetween CTL responses and tumor regressionobserved in the ALVAC and dendritic cell trialssupports the notion that the rejection is causedby the vaccine.

62

Table 1. Summary of anti-MAGE3.A1 CTL responses in vaccinated melanoma patients.

CTL response in patients with

Vaccination mode evidence oftumor regression

no evidence oftumor regression

ALVAC-MAGE 3/4 1/11

Dendritic cells + peptide MAGE-3.A1 3/3 0/3

Peptide MAGE-3.A1 1/7 0/13

Tumor regressions observed aftervaccination: a possible role fortumor-specific cytolytic Tlymphocytes that do not recognizethe vaccine antigens

C. Germeau, in collaboration with W. Ma, C. Lurquin,N. Vigneron, F. Brasseur, B. Lethé, E De Plaen,Brussels branch of the Ludwig Institute for CancerResearch, and T. Velu, Medical Oncology, ErasmeHospital, Brussels.

Even among those vaccinated patients whoshowed a CTL response, most had a lowfrequency of anti-MAGE-3.A1 CTL in theblood, ranging between 10-6 and 10-5 of CD8 Tcells. Because we felt that such a level of CTLmight be insufficient to produce on its own theobserved tumor regressions, we examined thepossibility that CTL directed against otherantigens present on the tumor might contributeto the regression.

We selected five metastatic melanomapatients who had been vaccinated with MAGEantigens. They were selected because it had beenpossible to derive a permanent cell line fromtheir tumor cells. We set up mixed lymphocyte-tumor cell cultures (MLTC), to estimate theblood frequencies of their CTL directedspecifically against the tumor. We will refer tothe lytic effectors detected in these MLTC as‘anti-tumor’ CTL to distinguish them from the‘anti-vaccine’ CTL, which recognize the vaccineantigen. For all five patients, anti-tumor CTLwere found at high frequencies, i.e. from 10-4 to3 x 10-3 of the CD8 T cells, in the blood aftervaccination. Unexpectedly, they were alreadypresent at similar high frequencies beforevaccination. The frequency of anti-tumor CTLobserved after vaccination was considerably

higher than that of the anti-vaccine CTL, rangingfrom 12 fold to 20,000 fold higher.

Because T cells directed at other tumorantigens than the vaccine antigen could make animportant contribution to the tumor regressionsthat are observed occasionally followingvaccination, we felt that it was necessary todefine the precise nature of their target antigens,particularly with regard to their tumor specificity.We focused our effort on patient EB81, whohad shown complete regression of a largenumber of cutaneous metastases followingvaccination with ALVAC-MAGE. The antigensrecognized by the anti-tumor CTL of patientEB81 turned out to belong to the main classesof tumor antigens that have been definedpreviously. A majority of anti-tumor CTL clonesrecognized antigens encoded by MAGE-C2, acancer-germline gene (10). Others recognized anantigen encoded by gp100 , a melanocyticdifferentiation gene. Another anti-tumor CTLclone of patient EB81 recognized an antigenencoded by an ubiquitously expressed genewhich had undergone a point mutation in thetumor. In conclusion we are facing a paradoxicalsituation where the melanoma patients that arebeing vaccinated, have already mounted a highspontaneous response against the types ofantigens used in the vaccines. At the time ofvaccination this spontaneous T cell response isclearly ineffective in halting tumor progression.

To evaluate the potential contribution of the“anti-tumor” T cells to the occasional tumorrejections that occur following vaccination, wemeasured the frequency of the anti-vaccine andanti-tumor T cells in metastases of patient EB81.The frequency of anti-MAGE-3.A1 T cells was2.5!x!10-6 of CD8 T cells in the blood and it was6-fold higher in an invaded lymph node. An anti-tumor CTL recognizing an antigen encoded by

63

M A G E - C 2 showed a considerably higherenrichment: whereas in the blood the frequencyof this CTL was 9!x!10-5, it was about 1,000times higher in the invaded lymph node. Severalother anti-tumor T cell clonotypes also hadfrequencies above 1% and appeared toconstitute the majority of the T cells present inthis site. Similar findings were made on aregressing cutaneous metastasis.

These results suggest that the anti-vaccineCTL may not be the principal effectors that killthe bulk of the tumor cells. They may exert theireffect mainly by an interaction with the tumorthat creates conditions enabling the stimulationof large numbers of CTL directed against othertumor antigens, which then proceed to destroythe tumor cells. New naive T cells may bestimulated in the course of this process, as weobserved that new anti-tumor CTL clonotypesemerged following vaccination and were presentin the metastases at a very high frequency. Animplication of this wide T cell response triggeredby the vaccine is that loss of the vaccine antigenby a number of tumor cells would not ensuretumor escape.


1. Baurain JF, Colau D, van Baren N, Landry C,Martelange V, Vikkula M, Boon T, Coulie PG.High frequency of autologous anti-melanoma CTLdirected against an antigen generated by a point mutationin a new helicase gene. J Immunol 2000;164:6057-66.

2. Chiari R, Hames G, Stroobant V, Texier C,Maillere B, Boon T, Coulie PG. Identification of atumor-specific shared antigen derived from an Ephreceptor and presented to CD4 T cells on HLA class IImolecules. Cancer Res 2000;60:4855-63.

3. Coulie PG, Karanikas V, Colau D, Lurquin C,Landry C, Marchand M, Dorval T, Brichard V,Boon T. A monoclonal cytolytic T-lymphocyte responseobserved in a melanoma patient vaccinated with a tumor-specific antigenic peptide encoded by gene MAGE-3.Proc Natl Acad Sci USA 2001;98:10290-5.

4. Schiavetti F, Thonnard J, Colau D, Boon T,Coulie PG. A human endogenous retroviral sequenceencoding an antigen recognized on melanoma by cytolyticT lymphocytes. Cancer Res 2002;62:5510-6.

5. Karanikas V, Colau D, Baurain JF, Chiari R,Thonnard J, Gutierrez-Roelens I, Goffinet C,Van Schaftingen EV, Weynants P, Boon T,Coulie PG. High frequency of cytolytic T lymphocytesdirected against a tumor-specific mutated antigendetectable with HLA tetramers in the blood of a lungcarcinoma patient with long survival. Cancer Res2001;61:3718-24.

6. Coulie PG, Karanikas V, Lurquin C, Colau D,Connerotte T, Hanagiri T, Van Pel A, Lucas S,Godelaine D, Lonchay C, Marchand M, VanBaren N, Boon T. Cytolytic T-cell responses of cancerpatients vaccinated with a MAGE antigen. ImmunolRev 2002;188:33-42.

7. Coulie PG, Bruggen P. T-cell responses ofvaccinated cancer patients. Curr Opin Immunol2003;15:131-76.

8. Karanikas V, Lurquin C, Colau D, van BarenN, De Smet C, Lethé B, Connerotte T, CorbièreV, Demoitié M-A, Liénard D, Dréno B, Velu T,Boon T, and Coulie PG. Monoclonal anti-MAGE-3 CTL responses in melanoma patients displaying tumorregression after vaccination with a recombinant canarypoxvirus. J Immunol 2003;171:4898-4904.

9. Lonchay C, van der Bruggen P, Connerotte T,Hanagiri T, Coulie PG, Colau D, Lucas S, VanPel A, Thielemans K, van Baren N, Boon T.Correlation between tumor regression and T cell responsesin melanoma patients vaccinated with a MAGE antigen.Proc Natl Acad Sci USA 2004;101:14631-14638.

10. Ma W, Germeau C, Vigneron N, MaernoudtA-S, Morel S, Boon T, Coulie PG, Van denEynde B. Two new tumor-specific antigenic peptidesencoded by gene MAGE-C2 and presented to cytolytic Tlymphocytes by HLA-A2. Int J Cancer 2004;109:698-702.

65

Ludwig Institute for Cancer Research (LICR),Brussels branch

67

The Ludwig Institute for Cancer Research

Cancer is a major concern in human

health. The prospects for bringing cancer

under control require linked innovative basic

and clinical research. In this view, Daniel K.

Ludwig created in 1971 the Ludwig Institute

for Cancer Research, an international

organization bringing together scientists and

clinicians from around the world. Ludwig

investigators are recognized leaders in many

areas of science, involving genetics,

bioinformatics, immunology, virology, cell

biology and signal transduction.

Thierry Boon

Faithful to the organizing principles laid down by Mr Ludwig, the Institute conducts its research

through ten Branches, located in seven countries. The Branch structure allows the Institute to interact

with a number of different research and clinical environments. Each Branch is focused on a research

program defined by the Branch Director in relation with the overall objectives of the Institute. The

Branches are established in association with University Hospitals, to stimulate close collaborations

between research laboratories and the clinic. By organizing and controlling its own clinical trials

programs, the Institute has indeed created a continuum that integrates laboratory and clinical research.

The biological properties of any given cancer cell constantly change, allowing tumors to spread and

become more aggressive. To overcome these obstacles, the Ludwig Institute has developed a broad-based

discovery program that seeks to understand the full complexity of cancer. Research is organized

according to the four major programmatic themes that define the Institute!: genetics, cell biology, cell

signalling and immunology.

Branch staffs vary in size from 30 to over 90, and internationally the Institute employs some 600

scientists, clinicians and support personnel. The quality of the research is monitored on an ongoing basis

by the Institute’s Scientific Committee and by an external peer review process.

The Brussels Branch of the Institute was created in 1978. It is composed of 84 members and is headed

by Thierry Boon, Branch Director.

68

Administration and general services

Administration and secretariat

Dario FLOREAN, Administrator

Aline BIEN, Accounting and Administrative Assistant

Sylvie HOPPE, Scientific Secretary

Nathalie KRACK, Librarian

Carine LEMOINE, Scientific and Administrative Assistant

Arnaud ROKS, Accounting Assistant

Geneviève SCHOONHEYDT, Administrative Assistant

Valérie WINAND, Scientific Secretary

Contact!:

UCL 7459, Avenue Hippocrate 74,1200 Bruxelles, Belgium

Tel!: +32 (2) 764 7459Fax!: +32 (2) 764 9405

Pre-clinical Research Resources Group:

Guy WARNIER, Veterinarian and Group Head

Jean -Pierre MATTHEWS, Technical Assistant

Phone SIRICHANTO, Technical Assistant

Kéota SOUVANNARAJ, Technical Assistant

Keoprasith SOUVANNARAJ, Technical Assistant

Technical support

Jo LEJEUNE, Head of Computer Systems

Paul WAUTERS, Laboratory Manager

André TONON, FACS Operator

Jacques VAN CLEVE, Research Assistant in Bioinformatics

Jean-Luc GERMIS, Technician

Manuel ORVIZ, Senior Technical Assistant

Antoinette VAN GOETSENHOVEN, Technical Assistant

Jean-Claude VERHELST, Administrative and Technical Assistant

M. AHLOUMOU, Laboratory Helper

Mustafa HAMED, Laboratory Helper

Samira JEBARI, Laboratory Helper

Pascale MAHIEU-GAUDY, Laboratory Helper

Encarnacion MARIANINI, Laboratory Helper

Rafaëlla REYES PRIETO, Laboratory Helper

69

TUMOR IMMUNOLOGY AND ANTIGEN PROCESSINGBenoît VAN DEN EYNDE, Member

Bernard LAUWERYS, Senior InvestigatorCatherine UYTTENHOVE, Senior InvestigatorGuy WARNIER, Veterinarian, Associate InvestigatorFanny PIETTE, Research AssistantLuc PILOTTE, Research AssistantJacques CHAPIRO, Postdoctoral FellowWenbin MA, Postdoctoral Fellow, until AprilIvan THEATE, Postdoctoral FellowEdus WARREN, Visiting Scholar, until JanuaryBenoît GUILLAUME, StudentIvo HUIJBERS, StudentTanja NASLUND, Student, until FebruaryNicolas PARMENTIER, StudentNathalie VIGNERON, StudentChristine CORBIAU, TechnicianDominique DONCKERS, Technician

The central research theme of our group is the study of tumor antigens recognized by T lymphocytes. Besides ourcontinued effort to identify additional antigens of interest, we mainly want to address a number of fundamental ormechanistic issues that have a direct impact on the utilization of such antigens as cancer vaccines in humanpatients. These antigens consist of peptides that are presented by MHC molecules at the cell surface and derivefrom intracellular proteins that are degraded by the proteasome. The intracellular pathway leading from the proteinto the peptide/MHC complex is known as “antigen processing”. We are currently studying the processing ofseveral human tumor antigens by the proteasome, and we are particularly interested by the processing differences wehave observed between the standard proteasome, which is present in most cells, and the immunoproteasome which isfound in some dendritic cells and in cells exposed to interferon-gamma.

We have also recently described a new aspect of the proteasome function in antigen processing, which is theproduction of antigenic peptides made of two non-contiguous segments. The production of such peptides involves theexcision of a peptidic segment and the splicing of the two fragments. We showed that this excision and splicingprocess is done by the proteasome, presumably by a mechanism of transpeptidation. The study of additionalpeptides produced by splicing is ongoing.

We are also studying a mouse preclinical model of cancer immunotherapy, where we try to define the optimalconditions to induce effective anti-tumor responses by various vaccination approaches against defined antigens. Thisled us to uncover a powerful mechanism of tumor resistance which is based on tryptophan catabolism byindoleamine-2,3 dioxygenase, an enzyme that we found frequently expressed in tumors. The resulting localtryptophan shortage appears to prevent the proliferation of lymphocytes at the tumor site. Treatment of mice withinhibitors of indoleamine-2,3 dioxygenase can counteract this tumor resistance mechanism. These results suggestthat a treatment based on the inhibition of this enzyme could be combined with approaches of cancerimmunotherapy in order to increase the tumor response rate.

To obtain the most relevant information from such preclinical models, we are building a new mouse melanomamodel where tumors expressing a given antigen can be induced, using a transgenic system based on Cre-loxrecombination. This should recapitulate the long-term host/tumor relationship that occurs in humans when atumor slowly develops within a normal tissue.

70

Differential processing of tumorantigens by the standardproteasome and theimmunoproteasome

J. Chapiro, V. Stroobant, B. Guillaume, W. Ma, F.Piette

Antigens recognized by cytolytic T lymphocytes(CTL), such as viral or tumor antigens, usuallyconsist of peptides of 8-10 amino acids inlength, which are presented by MHC class Imolecules at the cell surface. Because suchpeptides derive from intracellular proteins, aprocessing step is required before they can beexposed to the cell surface in association withMHC molecules. Firstly, the peptide isproduced, as a result of the degradation of theparental protein by the proteasome. Secondly, itis taken up by a dedicated transporter namedTAP, and translocated inside the endoplasmicreticulum where it meets and associates withnewly synthesized MHC molecules. The firststep of cleavage by the proteasome is crucial inthat cleavage location determines the precisesequence of the final antigenic peptide. We haveobserved that this cleavage may occur differentlyin some cells, depending on their proteasomecontent. The proteasome comes in two forms:the standard proteasome, which is found in mostcells, and the immunoproteasome, which isexpressed by mature dendritic cells and by cellsexposed to interferon-gamma (IFNg).

We previously reported that a class-Irestricted antigenic peptide derived from anubiquitous human protein was processedefficiently by the standard proteasome but notby the immunoproteasome. As a result, therelevant epitope is not presented efficiently bymature dendritic cells, which containimmunoproteasomes (1). This could explainhow certain potentially autoreactive CTL canescape tolerance induction in the thymus and failto be activated in the periphery. We have nowextended those observations to several antigenicpeptides of interest for cancer immunotherapy,including HLA-A2-restricted epitopes derivedfrom tyrosinase, Melan-AMART1 and gp100. Onthe contrary, we showed that other tumorepitopes, which are derived from MAGE-3 andMAGE-C2, are processed by theimmunoproteasome but not by the standardproteasome and therefore are presented to CTL

only by tumor cells pre-treated with IFNg (2).By analysing the peptidic fragments producedafter in vitro digestion with the two proteasometypes, we found that the differential processingcan result from two mechanisms. In some cases,one of the proteasome types predominantlycleaves within the sequence of the epitope,resulting in its destruction. In other cases, thedifference lies in the efficiency of cleavage at theC-terminal end of the antigenic peptide.

Our observations indicate that the pool ofantigenic peptides presented at the cell surfacemay differ substantially according to theproteasome type that is predominant in the cell.This may have major implications forimmunotherapy, particularly for cancerimmunotherapy, as it means that the peptiderepertoire presented by tumor cells may differfrom the repertoire presented by antigen-presenting cells (3). The peptide repertoire oftumor cells themselves may vary according tothe localization of the tumor (e.g. primary tumorversus lymph node metastasis) and its level ofexposure to IFNg . Therefore, it appearsessential for the success of cancerimmunotherapy to study those processingdifferences in detail, so as to define the mosteffective vaccination strategy for each epitopeand to use the appropriate combination ofantigens in order to minimize the risk of tumorescape by proteasome switching.

An antigenic peptide produced bypeptide splicing in the proteasome

N. Vigneron, V. Stroobant, J. Chapiro (incollaboration with A. Ooms and G. Degiovanni,Université de Liège, Belgium)

By stimulating blood lymphocytes of amelanoma patient with autologous tumor cells,we isolated a clone of CD8 cytolytic Tlymphocytes that recognizes an antigenic peptidepresented by HLA-A32 and derived fromglycoprotein gp100. This peptide, whosesequence is RTKQLYPEW, is present onseveral melanoma lines and was found to becomposed of two non-contiguous segments ofthe gp100 protein. Its production requires theexcision of four amino acids and splicing of theresidual fragments (4). The antigenic peptidecould be produced in vivo by electroporating the13-amino acid precursor RTKAWNRQLYPEWinto EBV-B cells. Proteasome inhibitors

71

lactacystin and epoxomicin prevented therecognition of target cells electroporated withthe 13-amino acid peptide, indicating thatproteasome activity was required for the splicingof this precursor. Moreover, the digestsproduced after incubation of the 13-amino acidprecursor with highly purified proteasomesstrongly stimulated IFNg production by theCTL. Analysis of these digests by tandem massspectrometry clearly demonstrated the presenceof the spliced peptide in the reaction mixture.Thus, the proteasome appears to produce theantigenic peptide RTKQLYPEW by excisionand splicing. By incubating proteasomes withsets of two distinct fragments each containing adifferent portion of the precursor peptide, wecould show that the energy required for thecreation of the new peptide bond was recycledfrom one of the bonds that are cleaved duringthe excision process. These data suggest that thesplicing occurs via a transpeptidation mechanisminvolving an acyl-enzyme intermediate. Ourresults reveal an unanticipated aspect of theproteasome function, which increases thediversity of antigenic peptides presented to Tcells.

We are currently studying in more details thisnovel activity of the proteasome. We are alsoextending our observation to additional peptidesthat are also produced by splicing. This is donein collaboration with two research groups fromUSA.

Identification of new antigensrecognized by autologous CTL onhuman melanoma

W. Ma, N. Vigneron (in collaboration with P. Coulie)

Melanoma patient EB81 was vaccinated with aMAGE-type antigen and showed regression ofall cutaneous metastases. Blood lymphocytescollected after the regression were stimulatedwith autologous tumor cells, and CTL cloneswere obtained. Surprisingly, none of theseclones was directed against the antigen used forvaccination. Using a cDNA expression cloningapproach, we identified the antigens recognizedby three of them. These antigens correspond tothree distinct peptides all derived from MAGE-C2, a gene with a cancer-germ-line expressionpattern, which is expressed in about 40% ofmelanomas and 30% of bladder carcinomas.Two peptides are presented by HLA-A2, andone by HLA-B57 (5). Because of their strict

tumor-specificity and their wide expression intumors, these new antigens represent promisingtargets for cancer immunotherapy. Theprocessing of two of these peptides is dependenton the immunoproteasome.

Melanoma line LG2-MEL also expressesseveral antigens recognized by autologous CTLs.We previously reported the molecular definitionof three distinct antigens recognized by some ofthese CTL clones. One of them consists of apeptide derived from tyrosinase and presentedby HLA-B35. Another is a peptidecorresponding to a mutation in a gene expressedubiquitously, named OS9 (6). A third is apeptide produced by splicing of two non-contiguous fragments of melanocytic proteingp100 (4) (see above). We have now identified afourth antigenic peptide expressed by thismelanoma line and recognized by a CTL clonerestricted by HLA-B35. The antigenic peptide,which is 9-amino acid long, has the sequenceLPHSSSHWL and is also derived frommelanocyte differentiation protein gp100.

A novel tumor immune escapemechanism based on tryptophandegradation by indoleamine 2,3dioxygenase

C. Uyttenhove, L. Pilotte, I. Théate, D. Donckers, N.Parmentier, V. Stroobant, D. Colau

Indoleamine 2,3-dioxygenase (IDO) is anintracellular enzyme that catalyses rapidtryptophan degradation. Because tryptophancan freely cross the plasma membrane, IDOexpression results in a local depletion oftryptophan in the extracellular mediumsurrounding the expressing cell. Tryptophandepletion was shown to impair T lymphocyteproliferation, and therefore IDO expressionrepresents of powerful immunosuppressivemechanism that accounts, for example, formaternal tolerance to allogeneic fetuses, whereIDO expression by placenta was found to playan essential role. Expression of IDO can beinduced by interferon-gamma in many cellulartypes, including macrophages and dendritic cells,and appears to play a prominent role in immuneregulation.

We have observed that many human tumorsexpress IDO in a constitutive manner (7). Todetermine whether IDO expression providestumor cells with a survival advantage by allowing

72

their escape from immune rejection in vivo, weused the well-characterized model system ofmouse tumor P815, where the antigen encodedby gene P1A is the major target of the rejectionresponse. We observed that expression of IDOby P815 tumor cells prevents their rejection bypre-immunized mice. This effect can be partlyreverted by systemic treatment of mice with aninhibitor of IDO, in the absence of noticeabletoxicity. These results suggest that the efficacyof therapeutic vaccination of cancer patientscould be improved by concomitantadministration of an IDO inhibitor.

Development of an inducible mousem e l a n o m a m o d e l f o rimmunotherapy

I. Huijbers (in collaboration with P. Krimperfort (NKI,Amsterdam) and A.-M. Schmitt-Verhulst (CIML,Marseille))

Immunotherapy represents an attractiveapproach for the treatment of cancer, inparticular melanoma. Preclinical studies ofvarious strategies of immunotherapy rely onmodel systems where tumor cells grown in vitroare inoculated into syngeneic mice. However,this does not recapitulate the long-term host-tumor relationship that occurs in patients duringtumor development. In order to have a modelsystem more relevant to the human situation, weare developing a mouse strain in which we caninduce melanomas expressing a tumor antigen ofinterest.

One of the most common site of geneticlesions in human melanoma is the INK4A/ARFlocus, which encodes two distinct tumorsuppressor proteins p16INK4A and p14ARF.Genetic disruption of this locus predisposesmice to the formation of various tumor types,but is not sufficient to induce melanoma unlessthe Ras pathway is specifically activated inmelanocytes. In order to have a fully controlledmodel system for melanoma, we have generatedtransgenic mice in which the deletion of theInk4a/Arf genes and the melanocyte-specificexpression of both activated Harvey-RasG12V anda well characterized antigen is spatially andtemporally regulated by a fusion protein betweenthe Cre-recombinase and the tamoxifenresponsive hormone-binding domain of theestrogen receptor (CreERDD). The antigen isencoded by P1A , a gene expressed in severaltumors but silent in normal tissues except testis

and placenta. The tumor induction in these micewill be performed by topical administration oftamoxifen, which should be sufficient to inducethe essential genetic rearrangements inmelanocytes necessary to establish neoplastictransformation.

Eighteen transgenic lines were generated bythree different strategies based on the use of aconstruct containing the tyrosinase promoter, aCreERDD fusion gene flanked by loxP sites, aV12 mutated H-Ras gene, an internal ribosomalentry site and the P815A-antigen encoding gene,P1A. After crossing with a mouse straincontaining a conditional INK4A/ARF locusflanked by loxP sites, one transgenic line wasfound to develop cutaneous melanomas aftersubcutaneous injection of tamoxifen, in 8 out of33 treated animals. The induced tumors aredeleted for INK4A/ARF, express the activatedRas and express P1A.

This line is now being backcrossed to the H-2d background in order to be used as a modelfor immunotherapy. In parallel, we havedeveloped a strain of mice transgenic for theP1A-specific T cell receptor, which will be usefulfor such studies.

Antigen presentation by dendriticcel ls in Systemic LupusErythematosus

B. Lauwerys (in collaboration with F. Houssiau, Unitéde Rhumatologie)

Systemic lupus erythematosus (SLE) is anautoimmune disorder that is characterized byovert polyclonal B cell activation and T cell-driven autoantibody production against nuclearantigens. We are investigating the involvementof dendritic cells (DC) in the impaired peripheraltolerance leading to the activation ofautoreactive CD4 T cells in SLE. Studyingstrains of mice that are congenic for differentSLE susceptibility loci and were developed byE.K. Wakeland (University of Texas,Southwestern Medical Center at Dallas), weobserved that DC from one of these strains(Sle3) significantly impact CD4 T cell activationand apoptosis as compared to control DC.Further work will focus on the identification ofthe genetic mechanisms underlying theseobservations, taking advantage of the availabilityof newly produced subcongenic strains, carryingsmaller intervals of the Sle3 region.

73

In parallel, the availability of large amounts ofPBMC collected from untreated patients withactive disease gives us the opportunity to carryon experimental procedures aiming at theidentification of new physiopathological targetsin SLE, using high-density cDNA microarrays(Affymetrix). In order to increase the sensibilityof that procedure, PBMC from SLE patients aresorted by flow cytometry in lymphocyte subsets,prior to RNA extraction and hybridization of theslides. Interpretation of the results will befacilitated by the knowledge of numerous SLEsusceptibility regions in the human genome thatcontain genes of particular interest.

Finally, we recently identified a new geneencoding a potential decoy receptor for MIF(Macrophage migration Inhibitory Factor), acytokine that plays important roles ininflammatory and tumoral responses. Wecurrently investigate the functions of that novelmolecule, targeting its potential inhibitoryactivity of MIF actions in inflammation andtumor growth.


1. Morel S, Levy F, Burlet-Schiltz O, Brasseur F,Probst-Kepper M, Peitrequin AL, Monsarrat B,Van Velthoven R, Cerottini JC, Boon T, GairinJE, Van den Eynde BJ. Processing of some antigens bythe standard proteasome but not by theimmunoproteasome results in poor presentation bydendritic cells. Immunity 2000;12:107-17.

2. Schultz ES, Chapiro J, Lurquin C, Claverol S,Burlet-Schiltz O, Warnier G, Russo V, Morel S,Levy F, Boon T, Van den Eynde BJ, van derBruggen P. The production of a new MAGE-3 peptidepresented to cytolytic T lymphocytes by HLA-B40requires the immunoproteasome. J Exp Med2002;195:391-9.

3. Van den Eynde BJ, Morel S. Differentialprocessing of class-I-restricted epitopes by the standardproteasome and the immunoproteasome. Curr OpinImmunol 2001;13:147-53.

4. Vigneron N, Stroobant V, Chapiro J, OomsA, Degiovanni G, Morel S, van der Bruggen P,Boon T, Van den Eynde B. An antigenic peptide

produced by peptide splicing in the proteasome. Science2004;304:587-90.

5. Ma W, Germeau C, Vigneron N, MaernoudtA-S, Morel S, Boon T, Coulie PG, Van denEynde BJ. Two new tumor-specific antigenic peptidesencoded by gene MAGE-C2 and presented to cytolytic Tlymphocytes by HLA-A2. Int J Cancer 2004;109:698-702.

6. Nathalie Vigneron, Annie Ooms, SandraMorel, Gérard Degiovanni and Benoît J. Vanden Eynde. Identification of a new peptide recognizedby autologous cytolytic T lymphocytes on a humanmelanoma Cancer Immunity 2002;2:9.

7. Uyttenhove C, Pilotte L, Theate I, StroobantV, Colau D, Parmentier N, Boon T, Van denEynde BJ. Evidence for a tumoral immune resistancemechanism based on tryptophan degradation byindoleamine 2,3-dioxygenase. Nat Med 2003;9:1269-74.

8. Bilsborough J, Uyttenhove C, Colau D,Bousso P, Libert C, Weynand B, Boon T, vanden Eynde BJ. TNF-mediated toxicity after massiveinduction of specific CD8+ T cells followingimmunization of mice with a tumor-specific peptide. JImmunol 2002;169:3053-60.

9. Probst-Kepper M, Stroobant V, Kridel R,Gaugler B, Landry C, Brasseur F, Cosyns JP,Weynand B, Boon T, Van Den Eynde BJ. Analternative open reading frame of the human macrophagecolony-stimulating factor gene is independently translatedand codes for an antigenic peptide of 14 amino acidsrecognized by tumor-infiltrating CD8 T lymphocytes. JExp Med 2001;193:1189-98.

10. Van Den Eynde BJ, Gaugler B, Probst-Kepper M, Michaux L, Devuyst O, Lorge F,Weynants P, Boon T. A new antigen recognized bycytolytic T lymphocytes on a human kidney tumor resultsfrom reverse strand transcription. J Exp Med 1999;190:1793-800.

11. Brandle D, Brasseur F, Weynants P, Boon T,Van den Eynde B. A mutated HLA-A2 moleculerecognized by autologous cytotoxic T lymphocytes on ahuman renal cell carcinoma. J Exp Med 1996;183:2501-8.

74

GENES EXPRESSED IN CANCER AND GERMLINE CELLS

Etienne De PLAEN, Assistant memberCharles DE SMET, Assistant member

Axelle LORIOT, StudentOlga KHOLMANSKIKH, StudentClaudine BLONDIAUX, Technician

Human tumors express specific antigens arising from the activation of genes, such as MAGE genes, that arenormally expressed only in germ cells. As germ cells are not subject to scrutiny by the immune system, antigensencoded by these genes are strictly tumor-specific. We are trying to identify new genes that present the same patternof expression as MAGE genes. Additionally, efforts are devoted to determining both the function of "cancer-germline" genes, and the mechanisms leading to their activation in tumors.

Isolation and characterization ofnew cancer-germline genes

Screening procedures based on differentialexpression profiling allowed the isolation ofseveral genes with cancer and germline specificexpression (1-4). Most of these genes arelocated on the X chromosome, and have theirnormal site of expression in spermatogonia, thepre-meiotic stage of sperm development. Thisrestricted expression is likely related to thechromosomal location of these genes, as the Xchromosome is inactivated in post-meioticsperm cells.

MAGE-A1 acts as a transcriptionalrepressor

MAGE-A1 , the first characterized cancer-germline gene, belongs to a family of twelvegenes located on the X chromosome in regionq28 (5,6). To analyze the functions of MAGE-A1, we searched to identify protein partners ofthis protein. Using yeast two-hybrid screening,we found interaction between MAGE-A1 andtranscriptional regulator SKIP. SKIP is anadaptor protein that connects DNA-bindingproteins such as Smad3, the Vitamin D receptor,

CBF1 or MyoD, to proteins that activate orrepress transcription. A repression complexincluding histone deacetylases (HDAC) is knownto bind to SKIP and CBF1. In the presence ofthe intracellular part of Notch1 (Notch1-IC), therepression complex is detached from SKIP byNotch1-IC and recruitment of an activationcomplex including histone acetyltransferases(HAT) is facilitated by SKIP. To examinewhether MAGE-A1 could interfere with asignalling pathway involving SKIP, we expressedMAGE-A1 in mammalian cells in whichNotch1-IC binds to SKIP fused to the Gal4DNA-binding domain and activates a C A Treporter gene containing Gal4-binding sites nearits promoter (Fig. 1).

Figure 1. MAGE-A1 counteracts Notch1-ICtransactivation and recruits histone deacetylases

tk promoterGal4 UASCATCAT

Gal4 bd

HDAC

-OFF

SKIP

MAGE-A1

CAT

Notch1-IC

SKIP

tk promoterGal4 UASCATCAT

ONGal4 bd

HDACcomplexSMRT

Notch1-IC

+

HAT

Mam

75

We found that MAGE-A1 inhibited theNotch-1/SKIP transcriptional activation.Deletion analysis indicate that binding to SKIPis required to observe MAGE-A1-mediatedrepression of Notch1-IC transactivation.Moreover, MAGE-A1 was found to activelyrepress transcription by binding and recruitinghistone deacetylase 1. Our results suggest that bybinding to SKIP and by recruiting histonedeacetylases, MAGE-A1 protein present in thenucleus could repress genes implicated indevelopment and spermatogenesis. We are nowtrying to identify the genes that are regulated byMAGE-1 by using an inducible transfectedMAGE-A1 gene and the microarray technology.

Genome hypomethylation in theactivation of cancer-germline genesin tumors

Studies on the transcriptional regulation ofcancer-germline genes, such as MAGE-A1,showed that CpG methylation is an essentialcomponent of their repression in normalsomatic tissues (7,8). Activation of these genes intumors is associated with the genome-wideDNA demethylation process which oftenaccompanies tumorigenesis (9). Recently, wefound that hypomethylated CpGs in tumors arenot randomly distributed but are clustered intospecific regions such as the promoter region ofM A G E - A 1 (10). This site-specifichypomethylation is not due to a persistenttargeted demethylating activity, but appears toresult from a transient overall demethylationprocess followed by a permanent local inhibitionof remethylation due to the binding oftranscriptional activators (Fig. 2).

Using the antisense oligonucleotidetechnology, we are currently trying to identifythe enzymes that are normally involved inmaintaining CpG methylation within cancer-germline genes, with a special emphasis on theDNA methyltransferases DNMT1, DNMT3aand DNMT3b.

These studies should bring insight into theprocesses that lead to DNA methylation changesin cancer. Moreover, understanding themechanisms that activate cancer-germline genesmay help designing procedures to induce theexpression of specific antigens on tumors,thereby facilitating their elimination by theimmune system.

Transientglobal CpG

demethylation

Binding oftranscription factors

Methylation activityrestored, transcriptionfactors prevent local

remethylation

Normal

Tumor

MAGEA1

transcriptionfactors

meCpG

CpG

gene activated

Figure 2. Model for the stable activation ofMAGE-A1 in tumors.


1. Lucas S, De Smet C, Arden KC, Viars CS,Lethe B, Lurquin C, Boon T. Identification of a newMAGE gene with tumor-specific expression byrepresentational difference analysis. Cancer Res 1998;58:743-52.

2. Lethé B, Lucas S, Michaux L, De Smet C,Godelaine D, Serrano A, De Plaen E, Boon T.LAGE-1, a new gene with tumor specificity. Int JCancer 1998;76:903-8.

3. Martelange V, De Smet C, De Plaen E,Lurquin C, Boon T. Identification on a humansarcoma of two new genes with tumor-specific expression.Cancer Res 2000;60:3848-55.

4. Loriot A, Boon T, De Smet C. Five new humancancer-germline genes identified among 12 genes expressedin spermatogonia. Int J Cancer 2003;105:371-376.

5. De Plaen E, Arden K, Traversari C, GaforioJJ, Szikora JP, De Smet C, Brasseur F, van derBruggen P, Lethé B, Lurquin C, Brasseur R,Chomez P, De Backer O, Cavenee W, Boon T.Structure, chromosomal localization, and expression of12 genes of the MAGE family. Immunogenetics1994;40:360-9.

6. Serrano A, Lethe B, Delroisse JM, Lurquin C,De Plaen E, Brasseur F, Rimoldi D, Boon T.

76

Quantitative evaluation of the expression of MAGEgenes in tumors by limiting dilution of cDNA libraries.Int J Cancer. 1999 Nov 26;83(5):664-9.

7. De Smet C, Courtois SJ, Faraoni I, Lurquin C,Szikora JP, De Backer O, Boon T. Involvement oftwo Ets binding sites in the transcriptional activation ofthe MAGE1 gene. Immunogenetics 1995;42:282-90.

8. De Smet C, Lurquin C, Lethe B, MartelangeV, Boon T. DNA methylation is the primary silencingmechanism for a set of germ line- and tumor-specific genes

with a CpG-rich promoter. Mol Cell Biol 1999;19:7327-35.

9. De Smet C, De Backer O, Faraoni I, LurquinC, Brasseur F, Boon T. The activation of human geneMAGE-1 in tumor cells is correlated with genome-widedemethylation. Proc Natl Acad Sci USA 1996;93:7149-53.

10. De Smet C, Loriot A, Boon T. Promoter-dependent mechanism leading to selective hypomethylationwithin the 5' region of gene MAGE-A1 in tumor cells.Mol Cell Biol 2004;24:4781-4790.

77

IDENTIFICATION OF HUMAN TUMOR ANTIGENSPierre van der BRUGGEN, Member

Vincent STROOBANT, Postdoctoral FellowNicolina RENKVIST, Postdoctoral FellowZhaojun SUN, Postdoctoral FellowYi ZHANG, Visiting ScientistClaude WILDMANN, Research AssociateNathalie DEMOTTE, Ph.D. StudentHugues NICOLAY, Ph.D. Student (to February 2004)Sabrina OTTAVIANI, TechnicianBénédicte TOLLET, Technician

The group led by Pierre van der Bruggen is defining antigenic peptides encoded by genes such as those of theMAGE family. Therapeutic vaccination of cancer patients with MAGE peptides is now in progress, and theidentification of additional antigenic peptides is important to increase the range of patients eligible for therapy withpeptides and to provide tools for a reliable monitoring of the immune response. The monitoring of the anti-vaccine Tcell frequencies in patients vaccinated with protein is even more complex. The responses could be directed against alarge number of HLA-peptide combinations, including many that are presently unknown. The group hasundertaken a systematic effort to develop a reproducible monitoring approach with high specificity and sensitivity.The group is also involved in the study of functional defects of T cells.

New MAGE antigens recognized byCD8+ and CD4+ T cells

Y. Zhang, V. Stroobant, Z. Sun, S. Ottaviani

“Cancer germline” genes such as those of theMAGE family are expressed in many tumorsand in male germline cells, but are silent innormal tissues. They encode shared tumorspecific antigens, which have been used intherapeutic vaccination trials of cancer patients.The first antigens and genes that code for theseantigens were identified with anti-tumor cytolyticT lymphocytes obtained from cancer patients(1). A few HLA class I-restricted antigenicpeptides were identified by this “directapproach”. A large set of additional cancer-germline genes have now been identified bypurely genetic approaches (2). As a result, a vastnumber of sequences are known that can codefor tumor-specific shared antigens, but most ofthe encoded antigenic peptides have not beenidentified yet. The identification of a largenumber of antigenic peptides presented by HLAclass I and class II molecules is likely to beimportant for the future of clinical trials withdefined antigenic peptides. A large set ofpeptide/HLA combinations will alleviate HLArestriction and widen the set of eligible patients.

It will also facilitate the design of concurrentimmunizations against several antigens. Suchimmunizations could increase the primary anti-tumor efficacy of the vaccine and also decreasethe risk of tumor escape by loss of antigenexpression.

We have used approaches that we haveloosely named "reverse immunology" (3). Theyaim at identifying antigenic peptides recognizedby T cells, using gene sequences as startingpoint. We have focused this search on thecancer-germline genes, which are expected tocode for tumor-specific shared antigens on thebasis of their pattern of expression.

Search for antigenic peptides recognized byCD8+ T cells

We stimulated CD8+ T lymphocytes withdendritic cells transduced with viral vectorscontaining complete MAGE-coding sequences.As this requires the processing of the antigen bythe dendritic cells, we surmised that the peptidesthat would be identified would also be processedin the tumors expressing the MAGE genes. Adifficulty of the use of recombinant viruses is theactivation of CTL precursors directed againstviral antigens. We circumvented this problem byusing different vectors for the stimulation of the

78

microcultures, for the lytic assay with theresponder T cells, and for the cloning step. Thisprocedure is summarized in Figure 1. Dendriticcells were infected with either an adenovirus, acanarypoxvirus or a lentivirus, and they wereused to stimulate microcultures of autologousCD8+ T lymphocytes (4, 5). After three weeklystimulations, the responder cells were tested forlysis on autologous EBV-B cells infected withvaccinia-MAGE. Positive microcultures werecloned. To identify the antigenic peptide, theresulting CTL clone was tested for lysis of

autologous EBV-B cells pulsed with a completeset of peptides of 16 amino acids that overlap by12. When a peptide scored positive, shorterpeptides were synthesized to identify theshortest optimal peptide. To identify the HLApresenting molecule, the CTL clones were testedfor stimulation by cells transfected with theMAGE cDNA together with cDNAs coding forthe possible HLA presenting molecules. Finally,relevant tumor targets were used in a lysis assayto ascertain that the antigen was also processedby tumor cells.

Figure 1. Overview of the procedure to obtain anti-MAGE CD8+ CTL clones by stimulation with dendriticcells infected with viral vectors carrying a MAGE coding sequence.

We have listed in a database class I-restrictedantigenic peptides that are encoded by cancer-g e r m l i n e g e n e s(http://www.cancerimmunity.org/peptidedatabase/Tcellepitopes.htm). MAGE-1 and MAGE-3antigenic peptides are available for more than90% of Caucasians.

We have identified a MAGE-3.B40 antigenwhich is the first example of a tumor-specificantigen exclusively presented by tumor cellsexpressing the immunoproteasome (6). Thiswork was done in collaboration with the groupof Benoît Van den Eynde.

We have found that MAGE-1 peptideSAYGEPRKL is recognized by CTL clones thatare restricted either by HLA-Cw3, Cw6 or Cw16

(3). The presentation of the same peptide bydifferent HLA molecules may be frequent forHLA-C molecules, because they are more closelyrelated to each other in the peptide-bindingregion than HLA-A and B molecules. But forHLA-A and B also this may occur morefrequently than usually thought. MAGE-3peptide MEVDPIGLY is presented to differentCTL by HLA-B*4402, B*4403, and B*1801 (3).MAGE-1 peptide EADPTGHSY was found tobe recognized by different CTL on HLA-A1 andB35 molecules, and to bind to HLA-A29 (3).The same was found for the MAGE-3homologous peptide EVDPIGHLY.

These results have consequences for themonitoring of the immune response of patientsvaccinated with such tumor-specific sharedpeptides. A number of HLA-A1 patients were

79

injected with MAGE-3.A1 pept ideEVDPIGHLY, at a time when we did not knowthat it could be presented by B35 and A29 (3).The immune response was evaluated with HLA-A1 tetramers folded with the MAGE-3 peptide.Thus, A29 or B35-restricted responses againstthe peptide may have been missed.

Search for antigenic peptides recognized byCD4+ T cells

Studies in several animal models havedemonstrated an important role for CD4+ Tcells in inducing and maintaining anti-tumor

immunity. It is therefore possible that theaddition of antigenic peptides presented by classII to those presented by class I will improve theefficacy of therapeutic anti-tumor vaccination.To identify new HLA-peptide combinations, weused dendritic cells loaded with a recombinantMAGE protein to stimulate autologous CD4+ Tlymphocytes (7). After four weekly stimulations,the responder cells were tested for their ability tosecrete IFN-g upon stimulation with the antigen,and the positive microcultures were cloned. Theprocedure is summarized in Figure 2.

Figure 2. Overview of the procedure to obtain anti-MAGE CD4+ T cell clones by stimulation with dendriticcells loaded with a whole protein.

To identify the antigenic peptide, the positiveclones were stimulated with a set of peptides of16 amino acids that overlapped by 12 andcovered the entire MAGE protein sequence. Thepositive peptide was then tested for recognitionon several Epstein-Barr virus immortalized Bcell lines (EBV-B cell lines) to identify the HLApresenting molecule. Because a large number ofthe CD4+ T cells that were obtained in our firstexperiments appeared to be directed againstbacterial contaminants, we chose to alternate thesources of protein used at the various stages ofthe procedure. For example, to stimulate thelymphocytes, we used a MAGE proteinproduced in insect cells, and to test thespecificity of the responder lymphocytes, weused a protein produced in bacteria.Microcultures that specifically produced IFN-g

after stimulation with the MAGE protein werecloned by limiting dilution using autologousEBV-B stimulator cells either loaded with theMAGE protein used during the stimulationstep, or transduced with a retroviral constructencoding a truncated human invariant chain (Ii)fused with the MAGE protein.

MAGE-1 and MAGE-3 antigenic peptidesidentified by this procedure are listed in adatabase(http://www.cancerimmunity.org/peptidedatabase/Tcellepitopes.htm) (8). They include aMAGE-3 peptide presented by HLA-DP4,which is expressed by more than 70% ofCaucasians. This peptide could not have beenfound by a peptide stimulation approach because

80

no consensus anchor residue was known forHLA-DP4.

A novel approach to identifyantigens recognized by CD4 T cellsusing complement-opsonizedbacteria expressing a cDNA library

P. van der Bruggen (in collaboration with L. van deCorput from Leiden University, and P. Chaux)

We propose a new sensitive and rapid approachusing the exogenous pathway to take up andprocess proteins encoded by a cDNA libraryexpressed in bacteria. We hypothesized that,after opsonization with complement,recombinant bacteria can be endocytosed viareceptor-mediated uptake by Epstein-Barr virusimmortalized B cells to allow protein processingand presentation. To validate this approach, wemade use of a minor histocompatibility antigenencoded by the human male-specific gene DBY.A recombinant bacteria libray was constructedand screened with a DBY-specific CD4 T cellclone. We were able to identify bacteriaexpressing DBY diluted into a 300-fold excess ofbacteria expressing a non-relevant gene.Screening of a bacterial library using a DBY-specific CD4 T cell clone resulted in theisolation of several DBY cDNAs (6).

The normal anti-MAGE-3.A1repertoire

S. Ottaviani, C. Wildmann (a project started by C.Lonchay)

Cancer-germline gene MAGE-3 codes fortumor-specific antigens recognized on manytumors by T lymphocytes. A MAGE-3 antigenpresented by HLA-A1 has been used in severalvaccination trials on metastatic melanomapatients. Only a small minority of patients haveshown evidence of tumor regression. Attemptsto correlate the tumor rejections with the CTLresponse against the vaccine have beenhampered by the low level of these responses.To estimate not only the frequency but also theT cell receptor diversity of the naive precursorsdirected against MAGE-3.A1, we developed anapproach aimed at obtaining a large number ofCTL clones from very low frequency precursors.More than three billion peripheral bloodmononuclear cells (PBMC) were obtained fromseveral blood samples of a non cancerous

individual. An HLA-A1/MAGE-3 tetramer wasused to label large numbers of T cells and thepositive cells were enriched by magnetic sorting.The selected fraction was then passed through aflow cytometer to further enrich theA1/MAGE-3 tetramer-positive cells among theCD8 cells. A number of positive cells werecloned. A fraction of the proliferating cloneswere tetramer-positive and showed lytic activityon cells pulsed with the peptide. These CTLclones invariably lysed the melanoma cells also,indicating that their avidity was high enough torecognize the level of antigen expressed bytumor cells. This led to an estimate of 4x10-7 ofthe CD8 T cells for the anti-MAGE-3.A1 CTLpin this individual (9). To evaluate the diversity ofthe naive anti-MAGE-3.A1 repertoire, weexamined the TCR sequences of the 24 lytic anti-MAGE-3.A1 CTL clones obtained in thisindividual. Among the total of 23 clones, onlytwo had the same TCRa and b sequences. Theresults indicate that it is very likely that therepertoire comprises more than 100 clonotypes.On this basis it is possible to use not only thefrequency of CTL precursors in the blood butalso the presence of dominant clonotypes toascertain in patients the existence of anti-MAGE-3.A1 responses as low as 10-6 of CD8.

A reversible functional defect ofCD8+ T lymphocytes involving lossof tetramer labeling

N. Demotte, S. Ottaviani, C. Wildmann

We have observed that human CTL clones losetheir specific cytolytic activity and cytokineproduction under certain stimulation conditions,while retaining an antigen-dependent growthpattern. These inactive CTL simultaneously losetheir labeling by an HLA-peptide tetramer, eventhough the amount of TCR-CD3 at their surfaceis not reduced (10). The tetramer-negative cellsrecover tetramer staining and cytolytic activityafter stimulation with tumor cells in the presenceof a supernatant of activated lymphocytes. Ourresults suggest the existence of a new type offunctional defect of CTL. They also indicate thattetramers may fail to reveal some CTL bearingthe relevant TCR, even when such functionallyarrested CTL retain the potential to participatein immune responses because their defect isreversible. We will analyze the ability of CTL tobe labelled by tetramer during the days following

81

the antigenic stimulation. We also plan toanalyze the composition of lipid rafts oftetramer-positive and tetramer-negative cells. Inaddition, we will analyze tetramer-positive andtetramer-negative cells for differential expressionof genes by microarray analyzes.

Detection of anti-vaccine CD4 T cellresponse in vaccinated patients

Y. Zhang, N. Renkvist, Z. Sun, H. Nicolay, S.Ottaviani, in collaboration with D. Colau

Patients injected with class II-restrictedpeptides

For therapeutic vaccination trials, the vaccinecan consist of defined antigenic peptides. Thisapproach greatly facilitates the monitoring of theT cell response, because the presumed target ofthe T cells is completely defined. It allows theuse of HLA-peptide tetramers to detect T cellresponses in patients. HLA class II tetramershave been more difficult to obtain than HLAclass I tetramers. Didier Colau has recentlysucceeded in obtaining a DP4.MAGE-3tetramer, which was produced in insect cells. Itstained specifically relevant CD4 clones. Wehave evaluated the anti-vaccine T cellfrequencies in melanoma patients, who havebeen injected with dendritic cells pulsed with theMAGE-3 peptide presented by HLA-DP4.Blood cells from vaccinated patients werelabeled ex vivo with the multimers and themultimer-positive CD4 T cells were sorted byflow cytometry, multiplied in clonal conditionsand tested for their specific recognition of theMAGE-3 antigen. In one patient, the frequencyof anti-MAGE-3.DP4 T cells peaked at 1/1,400of the CD4 T cells, representing at least anincrease of 3,000-fold of the frequency foundbefore immunization. TCR analysis of the anti-MAGE-3.DP4 clones from this patientestablished that the response was highlypolyclonal.

Patients injected with a protein

Immunizing patients with a MAGE-3recombinant protein ought to induce T cellresponses against several MAGE-3 peptides,including peptides recognized by CD4 T cells,and this might result in a more effective anti-tumor response. Moreover, protein vaccinationalleviates the need to select patients according totheir HLA, as many peptides presented by

various HLA alleles are expected to bepresented. The monitoring of the anti-vaccine Tcell frequencies in patients vaccinated withprotein is complex. The responses could bedirected against a large number of HLA-peptidecombinations, including many that are presentlyunknown. Therefore, the tetramer approachused for the detection of CD8 or CD4 is notappropriate. We have undertaken a systematiceffort to develop a reproducible monitoringapproach with high specificity and sensitivity. Itcombines sorting of living blood T cellsproducing IFN-g after a short antigenicstimulation and detailed functional analyses ofthe cells amplified in clonal conditions. Usingthis approach, we have estimated the anti-vaccine CD4 T cell frequencies in the fivepatients who showed tumor regressions afterinjection of a MAGE-3 protein withoutadjuvant. Blood cells were stimulated overnightwith autologous dendritic cells loaded withprotein MAGE-3. The IFN-g-secreting CD4 Tcells were sorted by flow cytometry, multipliedin clonal conditions and tested for their specificrecognition of MAGE-3. Anti-MAGE-3 CD4 Tcells were detected in one patient aftervaccination. Surprisingly the 13 anti-MAGE-3clones, which corresponded to five differentTCR clonotypes, recognized the same peptidepresented by HLA-DR1. The frequency of theanti-MAGE-3 CD4 T cells was estimated at1/60,000 of the CD4 T cells in post-vaccinationblood samples, representing at least an 80-foldincrease of the frequency found beforeimmunization. The frequencies were confirmedin experiments where, instead of the protein, thepeptide was used to stimulate the blood cells.The frequency of the dominant CD4 clonotypewas also confirmed by a direct PCR evaluationon cDNA extracted from blood cells. The bloodcells of the four other clinical responders werescreened using our specific and sensitivemonitoring approach, but no anti-MAGE-3 Tcell was detected. We are now defied to explainwhy several patients with evidence of tumorregression following vaccination have nodetectable anti-vaccine T cells in their blood.


1. van der Bruggen P, Traversari C, Chomez P,Lurquin C, De Plaen E, Van den Eynde B,Knuth A, Boon T. A gene encoding an antigen

82

recognized by cytolytic T lymphocytes on a humanmelanoma. Science 1991;254:1643-7

2. Boël P, Wildmann C, Sensi ML, Brasseur R,Renauld JC, Coulie P, Boon T, van der BruggenP. BAGE: a new gene encoding an antigen recognized onhuman melanomas by cytolytic T lymphocytes.Immunity 1995;2:167-75.

3. van der Bruggen P, Zhang Y, Chaux P,Stroobant V, Panichelli C, Schultz ES, Chapiro J,Van den Eynde BJ, Brasseur F, Boon T. Tumor-specific shared antigenic peptides recognized by human Tcells. Immunol Rev 2002;188:51-64.

4. Chaux P, Luiten R, Demotte N, VantommeV, Stroobant V, Traversari C, Russo V, SchultzE, Cornelis GR, Boon T, van der Bruggen P.Identification of five MAGE-A1 epitopes recognized bycytolytic T lymphocytes obtained by in vitro stimulationwith dendritic cells transduced with MAGE-A1. JImmunol 1999;163:2928-36.

5. Breckpot K, Heirman C, De Greef C, van derBruggen P, Thielemans K. Identification of newantigenic peptide presented by HLA-Cw7 and encodedby several MAGE genes using dendritic cells transducedwith lentiviruses. J Immunol 2004;172:2232-7.

6. van de Corput L, Chaux P, van der MeijdenED, De Plaen E, Falkenburg JHF, van derBruggen P. A novel approach to identify antigensrecognized by CD4 T cells using complement-opsonized

bacteria expressing a cDNA library. Leukemia2004;in press.

7. Chaux P, Vantomme V, Stroobant V,Thielemans K, Corthals J, Luiten R, EggermontAM, Boon T, van der Bruggen P. Identification ofMAGE-3 epitopes presented by HLA-DR molecules toCD4(+) T lymphocytes. J Exp Med 1999;189:767-78.

8. Zhang Y, Chaux P, Stroobant V, EggermontAM, Corthals J, Maillère B, Thielemans K,Marchand M, Boon T, van der Bruggen P. AMAGE-3 peptide presented by HLA-DR1 to CD4+

T cells that were isolated from a melanoma patientvaccinated with a MAGE-3 protein. J Immunol2003;171:219-225.

9. Lonchay C, van der Bruggen P, ConnerotteT, Hanagiri T, Coulie PG, Colau D, Lucas S,Van Pel A, Thielemans K, van Baren N, BoonT. Correlation between tumor regression and T cellresponses in melanoma patients vaccinated with aMAGE antigen. Proc Natl Acad Sci USA 2004;101:14631-14638.

10. Demotte N, Colau D, Ottaviani S, GodelaineD, Van Pel A, Boon T, van der Bruggen P. Areversible functional defect of CD8+ T lymphocytesinvolving loss of tetramer labeling. Eur J Immunol2002;32:1688-97.

83

THERAPEUTIC VACCINATION OF CANCER PATIENTS WITHTUMOR-SPECIFIC ANTIGENSMarie MARCHAND, Senior Clinical Investigator

Nicolas van BAREN, Clinical InvestigatorJérôme DEGUELDRE, Clinical Research AssociateChristian VERFAILLE, Project ManagerSylvie HOPPE, Secretary

Tumor cells carry antigens such as MAGE antigens that are absent from normal tissues, and that can be targetedby cytolytic T lymphocytes (CTL) (1). While it is possible to make such CTL recognize and kill autologous tumorcells in vitro, the precise way to induce an effective CTL response against a MAGE antigen in cancer patients isnot known yet. In clinical vaccination trials, patients with a MAGE expressing cancer, often melanoma, aretreated repeatedly with a MAGE vaccine. These trials have two main objectives. First, the effectiveness of variousvaccination modalities can be assessed by following the clinical evolution of the tumor, by analyzing whether aspecific CTL response to the vaccine antigen occurred, and by determining whether immunological and clinicalresponses are correlated. Secondly, T lymphocytes and tumor samples collected at different timepoints duringvaccination can be analyzed in detail, which improves our understanding on what happens in patients whoexperience regression of metastatic lesions, and which may explain why this does not happen in the majority ofpatients with overall disease progression. This knowledge can then be used to design new vaccination modalities.

Therapeutic vaccination with MAGEtumor antigens

We have set up small-scale clinical trials aimed atevaluating the toxicity, the antitumoraleffectiveness and the immunological response incancer patients immunized with MAGE vaccinesinvolving either peptides, a recombinant proteinor a recombinant viral vector. A total of about330 patients have been included in thesemulticentric trials.

Clinical trials with the MAGE-3.A1 peptide

In a pilot study, the synthetic Mage-3.A1 peptidewas administered to 45 HLA-A1 patients withM A G E - 3 expressing melanoma, bysubcutaneous (s.c.) and intradermal (i.d.)injections of 100 or 300 mg of peptide on threeoccasions at monthly intervals. No significant

toxicity was reported. Of the 25 melanomapatients with measurable disease who received all3 immunizations, seven displayed tumorregressions. We observed 3 complete responses,1 partial response and 3 mixed responses (2).

Other vaccination modalities involving thesame peptide were investigated in melanomapatients with measurable disease (3). Thispeptide was mixed with the immunologicaladjuvant MPL + QS21 and injectedintramuscularly on a 4 week basis to 5 patients,without any evidence of melanoma regression. Acombination of the MAGE-3.A1 and MAGE-1.A1 peptides was administered s.c. and i.d.every 3 weeks to 11 patients. Two of themexperienced tumor regression (1 CR, 1 MxR).MAGE-3.A1 was injected s.c. and i.d. every 10-11 days instead of every 3-4 weeks to analyzewhether vaccination at higher frequency couldimprove the clinical response rate. Among 21patients treated, three had regressions of tumor

84

lesions (3 MxR). The same peptide wasassociated with the HLA class II-restrictedMAGE-3.DP4 peptide, in order to induce bothCD8 and CD4 T cell responses, hoping for animproved immunological and antitumoraleffectiveness. None of the 7 patients evaluableafter 9 i.d. and s.c. vaccinations given every 10-11 days had tumor regression.

Initial assessment of the CTL responsesinduced by vaccination with the MAGE-3.A1peptide was hampered by the lack of sensitivityof available CTL monitoring techniques. Morerecently, a new approach with improvedsensitivity, involving lymphocyte-peptide cultureand the use of HLA/peptide tetramers, was usedto document a significant increase in CTLpfrequency in a patient who showed tumorregression following vaccination with thispeptide at high frequency. This method alsoshowed that the CTL response was monoclonal.It was extended to 19 other patients whoreceived this peptide without adjuvant. Nonehad a detectable CTL response, indicating thatthis vaccine is weakly immunogenic (4).

In another study, patients with completelyresected primary or regional metastaticmelanoma with a high risk of relapse have beenvaccinated with the MAGE-3.A1 peptideinjected i.d. and s.c. every 2 weeks on 6occasions. The purpose was to analyze whethervaccination of melanoma patients with lessadvanced disease in the adjuvant setting wouldimprove the immunological response to apeptide vaccine. No CTL response was detectedby our tetramer assay in the 6 patients who havereceived the complete treatment, including 3patients with a resected tumor that did notexpress the appropriate antigen and who areassumed to be immunologically naive.

In a future clinical trial, the MAGE-3.A1pept ide wi l l be mixed wi th ani m m u n o s t i m u l a t o r y C p G - c o n t a i n i n goligonucleotide to try to increase itsimmunogenicity. This new and promisingadjuvant activates antigen presenting cells afterbinding to Toll-like receptor 9, and is thought toenhance CTL responses.

Clinical trials with the MAGE-3 protein

In a phase I/II trial, the recombinant Mage-3protein was tested as a vaccine in patients withMAGE-3 expressing cancer, mainly melanoma.The patients received either 30, 100 or 300 µg of

the protein, with or without the immunologicaladjuvants MPL and QS21, repeatedly byintramuscular injection. No severe toxicity wasreported. Among 33 evaluable melanomapatients, four experienced regressions ofmetastatic lesions, 2 partial and 2 mixedresponses. A partial response was also observedin a patient with metastatic bladder cancer (5).

The clinical efficacy of the MAGE-3 proteininjected i.d. and s.c. without adjuvant in non-visceral melanoma patients was tested in studyLUD 99-003. Patients received 300 µg ofMAGE-3 protein on 6 occasions at 3-weekintervals. Five out of 26 evaluable patients haveshown regressions, including 1 completeresponse lasting for more than 1 year. Thus thisvaccine does not seem to induce moreregressions than the MAGE-3.A1 peptide, but itdoes not require that the patient carries a specificHLA type. We will now mix this recombinantprotein with adjuvant AS15 containing animmunostimulatory CpG nucleotide, andcombine these i.m. injections with theadministration of selected class I or class IIpeptides by i.d. and s.c. routes, which may resultin the simultaneous activation of both CD8+and CD4+ specific T lymphocytes (Study LUD02-002).

Clinical trial with an ALVAC-MAGE virus

40 patients with advanced cancer, including 37with melanoma, were vaccinated with arecombinant canarypox (ALVAC) viruscontaining a minigene that encodes the MAGE-1.A1 and MAGE-3.A1 antigens, followed bybooster vaccinations with the 2 correspondingpeptides. The treatment comprised 4 ALVACinjections followed by 3 peptides injections, alli.d. and s.c., separated by 3 weeks each. Localinflammatory reactions at the sites of ALVACinjection were common, but were moderate inintensity and transient in duration. Among the30 melanoma patients who received at least 4ALVAC vaccinations, six experienced regressionof one or more melanoma metastases.Significant CTL responses were detected in 3 of4 patients with regressions, and in only one of 11patients with disease progression, whichindicates a significant correlation betweenimmune and antitumor responses. We plan toinvestigate in a new trial whether increasing thedose of ALVAC would result in improvedimmunological and clinical responses.

85

Summary of relevant observations andperspectives

Immunization with MAGE peptides, theMAGE-3 recombinant protein or the ALVACrecombinant viral vector, is devoid of significanttoxicity. A minority of melanoma patients (about20 %) show regression of metastatic lesionsfollowing immunization, whatever the MAGEvaccine used. About 10% of the patients showcomplete or partial clinical responses. Some ofthese lasted for several years. This frequency isfar beyond the reported incidence ofspontaneous regressions of melanomametastases, estimated at 0.2-0.3%, indicating thatthese regressions are linked to the vaccinations.CTL responses can be detected in a minority ofpatients vaccinated either with peptide orALVAC virus. The responses appear to be weakand are mainly monoclonal. The relativefrequency of CTL responders versus non-responders is higher in patients who had tumorregressions (4).


1. Boon T, Coulie PG, van der Bruggen P, vanBaren N. Immunology of the cancer cell: T lymphocyte

responses. In: Clinical Oncology, 2nd Edition.Eds: MD Abeloff, JO Armitage, AS Lichter, JE

Niederhuber. Churchill Livingstone, New York,2000.

2. Marchand M, van Baren N, Weynants P,Brichard V, Dréno B, Tessier M-H, Rankin E,Parmiani G, Arienti F, Humblet Y, Bourlond A.,Vanwijck R, Liénard D, Beauduin M, Dietrich P-Y, Russo V, Kerger J, Masucci G, Jäger E, DeGreve J, Atzpodien J, Brasseur F, Coulie PG,van der Bruggen P, and Boon T. Tumorregressions observed in patients with metastatic melanomatreated with an antigenic peptide encoded by geneMAGE-3 and presented by HLA-A1. Int J Cancer1999;80:219-230.

3. Machiels JP, van Baren N, Marchand M.Peptide-based Cancer Vaccines. Semin Oncol2002;29:494-502.


5. Marchand M, Punt CJ, Aamdal S, Escudier B,Kruit WH, Keilholz U, Hakansson L, van BarenN, Humblet Y, Mulders P, Avril MF, EggermontAM, Scheibenbogen C, Uiters J, Wanders J,Delire M, Boon T, Stoter G. Immunisation ofmetastatic cancer patients with MAGE-3 proteincombined with adjuvant SBAS-2: a clinical report. EurJ Cancer 2003;39:70-7.

86

ANALYSIS OF T CELL RESPONSES OF VACCINATEDCANCER PATIENTS *Aline VAN PEL, Associate Member

Danièle GODELAINE, Senior InvestigatorDidier COLAU, Senior InvestigatorJavier CARRASCO, StudentDavid BASTIN, TechnicianVinh HA THI, TechnicianMaria PANAGIOTAKOPOULOS, Technician

Christophe LURQUIN, Associate InvestigatorBernard LETHE, Associate InvestigatorFrancis BRASSEUR, Associate InvestigatorMarie-Claire LETELLIER, TechnicianMadeleine SWINARSKA, Technician

* In association with the group of Pierre Coulie, ICP: see Human tumor immunology (research atICP).

The identification of antigens recognized on human tumors by autologous T lymphocytes has opened the way fortherapeutic vaccination strategies involving defined tumor antigens such as the MAGE antigens (1, 2). Tumorregressions have been observed in a minority of treated melanoma patients. Such clinical responder patients havebeen found following immunization with peptides and recombinant ALVAC viruses. In some patients, amonoclonal T cell response was observed and the level of the response appeared to be stable during the course of thevaccination protocol (3, 4). Since detectable T cell responses occured more frequently in patients who show signs oftumor regression than in those who did not, we consider the possibility that the limiting factor for the anti-tumoreffect of the vaccine is the intensity of the CTL response to this vaccine (5, 6). Improving the efficacy of such vaccineswould thus critically depend on their capacity to trigger a robust immune response. A novel approach to vaccinationis to exploit the potentiality of dendritic cells that are widely accepted to be particularly effective in presentingantigens to T cells and immunize cancer patients with a sample of their autologous dendritic cells charged withtumor antigens. We initiated a collaboration with G. Schuler and B. Schuler-Thurner at the University ofErlangen (Germany), who vaccinated advanced stage IV melanoma patients with mature, monocyte-deriveddendritic cells pulsed with MAGE peptides and observed regression of some metastases(7). As shown in ourprevious report, an immune response was found in a few regressing patients(8). These results prompted us toinitiate a new collaboration with K. Thielemans ( VUB, Belgium ), who vaccinated less advanced patients with thesame protocol.

Methods for evaluation of T-cellresponses in vaccinated cancerpatients

To establish whether there is a correlationbetween tumoral regressions and T-cellresponses against the vaccine antigen, weevaluated the responses of patients vaccinatedwith a MAGE-3 antigenic peptide, with arecombinant virus coding for this peptide orwith MAGE peptide-pulsed dendritic cells.

The recent development of tetramericpeptide- MHC class I and class II complexes(tetramers) allows the direct identification ofantigen-specific T-cells. The technology hasbeen introduced and is still being developed byDidier Colau.

To detect low-level responses, bloodlymphocyte microcultures were stimulated withthe antigenic peptide in limiting dilutionconditions (Mixed Lymphocyte Peptide Culture,MLPC), followed by tetramer analysis, cloningof the tetramer-positive cells, and T-cell receptor

87

(TCR) sequence analysis of the cytolytic Tlymphocyte (CTL) clones that showed strictspecificity for the vaccine antigen (for method,see P. Coulie, ICP report).

Christophe Lurquin and Bernard Lethéfocused their efforts on detailed analysis offrequencies of characterized T-cell clones inblood, metastases and non tumoral tissuesamples, using ‘clonotypic’ polymerase chainreaction (PCR) amplifications specific for theVa and Vb rearrangements of relevant TCR.These PCR amplifications on cDNA weresensitive enough to detect one CTL expressing agiven TCR mixed with 3x107 PBMC of normaldonor and they were highly specific for the givenTCR insofar as no product was amplified withcDNA prepared from 3x107 PBMC of 5unrelated donors.

Previous results showed that a monoclonalCTL response against a MAGE-3 antigenpresented by HLA-A1 was observed by the invitro tetramer analysis in a melanoma patient whoshowed partial rejection of a large metastasisafter treatment with a vaccine containing onlythe tumor-specific antigenic peptide. Our resultsproved that the vaccination induced at least a100-fold amplification of an anti-MAGE-3.A1CTL and that vaccines containing only a tumor-specific antigenic peptide can elicit a CTLresponse (3)

A melanoma patient vaccinated witha MAGE-3 recombinant canarypoxvirus

Patient EB81 had about 70 cutaneous metastaseswhen she was vaccinated with a recombinantcanarypox virus of the ALVAC type containing aminigene encoding the MAGE-3.A1 peptide.Repeated injections of ALVAC were followed byvaccinations with the MAGE-3 peptide. Tenmonths after the first vaccine, all the metastaseshad become undetectable except an enlargedlymph node which was resected. A similarvaccination protocol alternating virus andpeptide injections was carried on until now.

Blood lymphocytes collected beforevaccination were analyzed with the MLPC-tetramer method. No anti-MAGE-3 CTL couldbe detected among 107 CD8+ cells, suggesting afrequency similar to that found in normal donors(4). Two different anti-MAGE-3.A1 CTL clones

were identified in postimmune blood!. One,CTL 35, was found in more than 95% of theanalyzed independent microcultures, suggestingthat this MAGE-3.A1 response was essentiallymonoclonal.

PCR amplifications specific for the TCR 35Va and Vb rearrangements were applied directlyto cDNA obtained from groups of PBMC. Inthe first post-ALVAC sample, the frequency ofCTL C35 rose to 3.6 x 10-6 of the CD8+,suggesting that the ALVAC vaccination inducedat least a 30 times amplification of anti-MAGE-3.A1 T cells. This frequency, that remainedstable during the peptide vaccinations, rose to1.1x10-5 after a boost of ALVAC vaccinationsgiven one year later and remained stablethereafter.

By stimulating blood lymphocytes frommelanoma patient EB81 with autologous tumorcells (MLTC), a series of CTL clones thatspecifically lysed autologous melanoma cellswere isolated. Some of these CTL clonesrecognized peptides presented by HLA-A2 andencoded by gene MAGE-C2, another cancer-germline gene which is expressed at high level bythe melanoma cells of the patient. We analyzedthe TCR Vb gene expression of this set of anti-MAGE-C2 CTL clones. One of them, namedCTL 16, seemed to be amplified aftervaccination since it was retrieved many times asindependent clones from the postimmune bloodbut was not found in preimmune blood samples.Clonotypic RT-PCR amplifications of TCR 16Va and Vb rearrangements indicated an averagefrequency of the corresponding CTL in bloodthroughout the whole time of vaccination of~3.5 x 10-5 among the CD8+.

To investigate the involvement of theMAGE-C2-specific CTL in the tumor regressionprocess observed in this patient after vaccinationagainst a MAGE-3 antigen, we have analyzed thefrequencies of the anti-MAGE-3.A1 CTL 35 andthe anti-MAGE-C2.A2 CTL 16 in the resectedmetastatic lymph node which presentedhistological signs of regression. Sections of 7mm thick and ~80 mm2, fragments of suchsections and groups of cells excised fromsections with laser microdissection were testedby RT-PCR for the presence of TCR 35 andTCR 16. We obtained a frequency of~1/32,500 CD8 for anti-MAGE-3 CTL, whichrepresents a 20-fold enrichment relative to bloodfrequency at the time of the metastasis resection.

88

Moreover, the anti-MAGE-C2 CTL was at least275-times enriched in the resected sample with afrequency higher than 1/100 CD8 in the lymphnode tissue and a frequency >1/5 CD8 in tissueregions strongly invaded by tumoral cells.

During this year, we confirmed and extendedthese results by analyzing TCRß cDNA librariesperformed on the RNA of various metastases ofpatient EB81. These assays allowed theestimation of the frequencies of various anti-tumor CTL within metastases. They also enabledus to considerably enlarge the diversity ofactivated T lymphocyte clones repeatedly foundin either the same metastasis, or in othermetastases resected before and after vaccination.T lymphocyte clones above a frequency of 1%of all the T cells were not rare. Some of theseclonotypes were not found before vaccinationneither in blood nor in tumor. Among thefrequent clones, one appeared to be directedagainst a mutated antigen, but most of themwere directed against various MAGE-C2epitopes, accounting for up to 20% of the T cellspresent in the metastasis. In this patient who isstill disease-free four years after the onset ofvaccinations, some of the intra-tumor clonotypeswere enriched up to 1000-fold relative to theirfrequencies in the blood.

A melanoma patient vaccinated withdendritic cells pulsed with a MAGE-3 peptide presented by HLA-A1

Melanoma patient MMB02 had about 20cutaneous metastases on her right leg when shewas vaccinated by K. Thielemans and hiscoworkers at the VUB with mature, monocyte-derived dendritic cells pulsed with MAGE-3.A1peptide according to a protocol set up byGerold Schuler at Erlangen (Germany). At theend of the first cycle of 6 vaccinations, somemetastases started to regress while new onesappeared. Vaccinations were continued and thepatient kept showing evidence of a mixedresponse. To improve our understanding of theprocess leading to occasional tumor regressionsthat occur following vaccination, we investigatedthe presence of anti-vaccine and anti-tumor CTLin the blood and inside metastases collected atvarious time points.

In the blood, the frequency of anti-MAGE-3.A1 T cells among CD8 cells raised from lessthan 3.4x10-7 before vaccination to 9.3x10-7 after

3 vaccinations, to 2.6x10-6 after 6 vaccinationsand 5.4x10-6 after another cycle of 6vaccinations. At this time point, only one anti-MAGE-3.A1 clonotype was detected, amonoclonality that contrasts with thepolyclonality observed in Schuler’s patients (8).On the other hand, the frequency of anti-tumorCTL, i.e. lytic effectors that recognize theautologous melanoma cells but not autologous Bcells nor NK target K562, was 2.2x10-4 beforeand 2x10-4 after 6 vaccinations. Thus, in thesame patient, the frequency of blood anti-tumorCTL was more than two orders of magnitudeabove the frequency of anti-vaccine CTL andwas already present before vaccination.

Cutaneous metastases removed after thesecond cycle of vaccinations allowed theisolation of a few TIL clones able to specificallyrecognize the autologous tumor cells. None ofthese clones was directed against the vaccineantigen. One of them was shown to be directedagainst a new tumor-specific antigen encoded bygene MAGE-C2 and presented to CTL by HLA-B44: peptide SESIKKKVL. This anti-MAGE-C2.B44 CTL was also found in the blood takenafter 6 vaccinations. RT-PCR specific for TCRsequences of that anti-MAGE-C2-B44 CTLclone is being set up and will allow estimation ofthe frequency of that clone in the blood samplestaken before and after immunization, as well asin cutaneous metastases resected before andafter vaccination. This should help to evaluatethe potential contribution of that clone to thetumor regressions that are observed occasionallyafter vaccination.

If the frequency and the diversity of the anti-vaccinal and anti-tumoral CTL are importantfeatures to understand the mechanismsunderlying tumor regressions, the functionalactivity of these CTL is another important aspectto be analyzed. In this prospect, we haveundertaken the analysis of anti-MAGE-3.A1CTL from vaccinated patients for expression ofsome phenotypic markers (CD25, CD69, CD28,CD27, CCR7, CD45…) that have beenassociated to the CTL activation status orparticular levels of maturation, in order to searchfor a correlation with CTL expansion capacities,lysis and secretion profile of cytokines.

89


1. Boon T, Cerottini JC, Van den Eynde B, vander Bruggen P, Van Pel A. Tumor antigensrecognized by T lymphocytes. Annu Rev Immunol1994;12:337-65.

2. Van Pel A, van der Bruggen P, Coulie PG,Brichard VG, Lethé B, van den Eynde B,Uyttenhove C, Renauld JC, Boon T. Genes codingfor tumor antigens recognized by cytolytic T lymphocytes.Immunol Rev 1995;145:229-50.

3. Coulie PG, Karanikas V, Colau D, Lurquin C,Landry C, Marchand M, Dorval T, Brichard V,Boon T. A monoclonal cytolytic T-lymphocyte responseobserved in a melanoma patient vaccinated with a tumor-specific antigenic peptide encoded by gene MAGE-3.Proc Natl Acad Sci USA 2001;98:10290-10295.

4. Karanikas V, Lurquin C, Colau D, van BarenN, De Smet C, Lethé B, Connerotte T, CorbièreV, Demoitié MA, Liénard D, Dréno B, Velu T,Boon T, Coulie PG. Monoclonal anti-MAGE-3CTL responses in melanoma patients displaying tumorregression after vaccination with a recombinant canarypoxvirus. J Immunol 2003; 171:4898-4904.


6. Demotte N, Colau D, Ottaviani S, GodelaineD, Van Pel A, Boon T, van der Bruggen P. Areversible functional defect of CD8+ T lymphocytesinvolving loss of tetramer labeling. Eur J Immunol2002;32:1688-97.

7. Thurner B, Haendle I, Roder C, DieckmannD, Keikavoussi P, Jonuleit H, Bender A, MaczekC, Schreiner D, von den Driesch P, Brocker EB,Steinman RM, Enk A, Kampgen E, Schuler G.Vaccination with mage-3A1 peptide-pulsed mature,monocyte-derived dendritic cells expands specific cytotoxicT cells and induces regression of some metastases inadvanced stage IV melanoma. J Exp Med 1999;190:1669-78.

8. Godelaine ., Carrasco J, Lucas S, Karanikas V,Schuler-Thurner B, Coulie PG, Schuler G.,Boon T, Van Pel A. Polyclonal cytolytic T lymphocyteresponses observed in melanoma patients vaccinated withdendritic cells pulsed with a MAGE-3.A1 peptide. JImmunol 2003;171:4893-97.

90

CYTOKINES IN IMMUNITY AND INFLAMMATION

Jean-Christophe RENAULD, Member

Jacques VAN SNICK, Member, GuestInvestigatorLaure DUMOUTIER, Postdoctoral FellowJamila LOUAHED, Postdoctoral FellowLaurent KNOOPS, StudentValérie STEENWINCKEL, StudentAmel TOUNSI, StudentBrigitte de LESTRE, TechnicianMonique STEVENS, TechnicianEmiel VAN ROOST, TechnicianFrancine WELLEKENS, Technician

The cytokine group studies the biology of Interleukin-9 (IL-9) and IL-22, two cytokines discovered at the Branch.IL-9 is a TH2 cytokine that plays a role in immune responses against intestinal parasites and asthma. IL-22,originally identified as a gene induced by IL-9 in T lymphocytes, upregulates the production of acute phase reagentsin the liver. Its activity in inflammatory responses is modulated by a specific antagonist, the IL-22 binding protein(IL-22BP). The role of IL-9 and IL-22 in inflammation is currently being investigated using transgenic and gene-targeted mice for these cytokines and their receptors.

Interleukin 9

J.-C. Renauld, J. Louahed, L. Knoops, V.Steenwinckel, B. de Lestré, M. Stevens, E. Van Roost

Interleukin-9 (IL-9) was discovered in our groupin 1989, through its ability to sustain antigen-independent growth of certain murine T helperclones. We further identified human IL-9 bycross-hybridization with the mouse gene.Although IL-9 did not turn out to be a T cellgrowth factor for freshly isolated T cells, it wasfound particularly potent on T cell lymphomas,as an anti-apoptotic agent. To determine thebiological activities of this factor, we generatedtransgenic mice overexpressing this cytokine.Analysis of these animals disclosed threeessential properties of IL-9: its tumorigenicpotential in T lymphocytes, its stimulatoryactivity on a particular subset of B lymphocytes,and its activity on mast cells and eosinophilswith consecutive implications in asthma.

IL-9-transgenic mice : T cell lymphomas

IL-9 transgenic animals showed normal T celldevelopment and T cell numbers but

spontaneously developed thymic lymphomas atlow frequency (5%) when maintained in aconventional environment. Two lines ofevidence indicate that IL-9 is not a conventionaloncogene but rather favors tumor developmentin response to exogenous stimuli. First, thetumor incidence was significantly lower whenmice were maintained under pathogen-freeconditions. Secondly, all IL-9 transgenic micedeveloped T cell lymphomas when exposed tosubliminal doses of a chemical carcinogen or toirradiation, that were innocuous in wild typemice (1). The above mentioned anti-apoptoticactivity of IL-9 provides an attractiveexplanation for these observations, namely thatIL-9 could lead to increased survival ofabnormal cells generated by exposure to minimaldoses of oncogenic stimuli. The potentialimplication of IL-9 in oncology was alsoconfirmed in human systems by its constitutiveexpression in Hodgkin lymphomas.

IL-9-transgenic mice : B1 cell expansion

Further analysis of these IL-9-transgenic miceshowed that a particular B lymphocytepopulation, called B-1 lymphocytes and usuallyrestricted to the peritoneal and pleuropericardial

91

cavities, were dramatically expanded in responseto IL-9 overproduction. In addition, such cellswere also found in the blood and in the lungs ofIL-9 transgenic mice. This observation isreminiscent of mice that are prone to thedevelopment of diseases that are characterizedby the production of autoantibodies, such asSystemic Lupus Erythematosus, and suggeststhat IL-9 might play a role in some autoimmuneprocesses (2).

IL-9-transgenic mice : parasite infectionsand asthma

In addition, IL-9 transgenic mice were found toharbor increased numbers of mast cells in theintestinal and respiratory epithelia, and were alsocharacterized by a general hypereosinophilia.This phenotypic characteristic was found toincrease the capacity of these animals to expelnematodes like Trichinella spiralis or Trichurismuris., suggesting that IL-9 administration couldprotect susceptible hosts against these parasites.This was confirmed by taking advantage of anew strategy of anti-cytokine vaccination: micevaccinated against their own IL-9 failed to expelT.muris parasites and had a decreasedeosinophilic response against the parasite (3).

The other side of the coin was the discoverythat IL-9 overexpression such as thatcharacterizing the IL-9 transgenic animalsresulted in bronchial hyperresponsiveness uponexposure to various allergens. Current studiesindicate that IL-9 promotes sthma through bothIL-13-dependent and IL-13-independentpathways. The potential aggravating role of IL-9in asthma was confirmed by genetic analysesperformed by others and pointing to both IL-9and the IL-9 receptor genes as major candidategenes for human asthma. Phase I clinical trialsusing anti-IL-9 antibodies produced in ourlaboratory have been initiated in collaborationwith Medimmune.

IL-9 receptor and s ignaltransduction

J.-C. Renauld, L. Knoops, M. Stevens, E. Van Roost

Analysis of the mode of action of IL-9 at themolecular level was initiated in 1992 by thecloning of the murine and human IL-9 receptor(IL-9R) cDNAs (4). By further dissecting thesignal transduction cascade triggered by IL-9, weshowed that, upon IL-9 binding, the IL-9R

associates with a co-receptor protein called gc.This induces the phosphorylation of the JAK1and JAK3 tyrosine kinases, which are associatedwith IL-9R and gc, respectively. A single tyrosineresidue of the IL-9R is then phosphorylated andacts as a docking site for 3 transcription factorsof the STAT family, STAT-1, -3 and -5, whichbecome phosphorylated and migrate to thenucleus, where they activate the transcription ofa number of genes. This pathway is common tomany cytokines but is often dispensable for theirbiological activities. For IL-9, our groupdemonstrated that activation of the STATtranscription factors is crucial for all the effectsof IL-9 studied on various cell lines, includingpositive and negative regulation of cellproliferation, as well as inhibition of corticoid-induced apoptosis in T cell lymphomas. Furtheranalysis demonstrated that STAT-1, -3 and -5play specific, redundant and synergistic roles inthe different activities of IL-9 in vitro (5).

The pathways responsible for IL-9-inducedproliferation were studied in details, and thisprocess was found to depend mainly on theactivation of STAT-5, on the recruitment of theIRS-1 adaptor, and on the activation of the ErkMAP-Kinase pathway.

The signal transduction pathway downstreamthe IL-9 receptor is illustrated in Fig. 1.

Figure 1.

92

Anti-apoptotic activity ofI-309receptor and signaltransduction

J.-C. Renauld, A. Tounsi, J. Van Snick, J. Louahed

Incidentally, our studies of this particular modelof the regulation of cell death by cytokines, leadthem to purify another protein called I-309,originally described as a human chemotacticfactor, and that turned out to exert a significantanti-apoptotic activity for thymic lymphomas (6).However, I-309 and IL-9 trigger completelydifferent pathways and it was shown that the I-309 anti-apoptotic activity was dependent on theact ivat ion of G-prote ins and theRas/MAPKinase pathway, whereas the IL-9-mediated effect was not. More recently, weshowed that a viral protein related to humanchemotactic factors (vMIP-I), and isolated fromHerpes viruses that induce T cell tumors, has thesame anti-apoptotic activity by binding to the I-309 receptor

IL-9-induced genes

J.-C. Renauld, J. Van Snick, L. Dumoutier, J.Louahed, L. Knoops, A. Tounsi, M. Stevens, E. VanRoost

To further characterize the mechanisms involvedin the anti-apoptotic activity of IL-9 in thisexperimental model, we sought to identify genesinduced by IL-9 in T cell lymphomas. Amongthe genes we identified, three open unexpectedperspectives: BCL3, M-Ras and IL-TIF/IL-22.

BCL3 : indirect modulation of NF-kB

BCL3 is a gene originally identified at thebreakpoint of translocations found in B cellleukemia, resulting in its transcriptionalactivation. The BCL3 protein interacts with NF-kB transcription factors and its induction by IL-9 represents a novel mechanism of NF-kBregulation by cytokines, and a new crosstalkbetween the JAK/STAT and NF-kB signaltransduction pathway (7). BCL3 induction mightplay a role in the antiapoptotic activity ofcytokines such as IL-4 and IL-9.

M-Ras : transcriptional regulation of theRas-MAPKinase pathway

M-Ras is a new member of the Ras oncogenesuperfamily. The Ras proteins are known to

regulate various cellular processes such asproliferation and apoptosis, when they are intheir activated form, in association with a GTPmolecule. Contrasting with the potentupregulation of M-Ras expression, M-Ras wasnot activated by IL-9 at the level of GTPbinding. However, other cytokines such as IL-3increased GTP binding to M-Ras, suggestingthat M-Ras induction might represent a newmechanism of cooperativity between cytokines.Constitutively activated M-Ras mutants triggerthe MAP Kinase pathway and induceproliferation of cytokine-dependent cells (8).

IL-TIF/IL-22 : a new cytokine structurallyrelated to IL-10

IL-TIF is a new gene that turned out to encode a179 amino acid long protein, including apotential signal peptide, and showing a weak butsignificant sequence homology with IL-10. Thisprotein, originally designated IL-TIF for IL-10-related T-cell derived Inducible Factor, was laterrenamed IL-22. Its expression is induced by IL-9in thymic lymphomas, T cells and mast cells andby lectins in freshly isolated spleen cells. Inaddition, constitutive expression of IL-22 wasdetected by RT-PCR in thymus and brain,suggesting that the role of this new factor is notrestricted to the immune system. Preliminaryexperiments showed that IL-22 induces STATactivation in various cell lines, suggesting thatthis factor might mediate some of the activitiesof IL-9. Biological activities of IL- 22 include theinduction of acute phase proteins in liver (9) andprotection against experimental heatitis (L.Dumoutier, unpublished results). Recombinanthuman IL-22 was produced (with D. Colau,LICR) and its crystallograhic structure solved.Despite its structural homology with IL-10, IL-22 fails to recapitulate any of IL-10 biologicalactivities.

Analysis of the genome databases leads to theidentification of a new receptor belonging to theIL-10 receptor family (10). This gene is locatedin the chromosome 6q24, at 24 kb from theIFNGR1 gene and at 152 kb from the IL-20R. Itencodes a protein of 231 amino acid, showing 33% and 34 % amino acid identity with theextracellular domains of the IL-22R and the IL-20R, respectively, but no cytoplasmic nortransmembrane domains were found. IL-22BPis highly expressed in the placenta, in the breast,in the mammary gland and in the skin. A specificinteraction was demonstrated between

93

insolubilized IL-22 and an IL-22BP-Ig fusionprotein. Moreover, recombinant IL-22BP couldblock IL-22 biological activity demonstratingthat this protein can act as an IL-22 antagonist.

Although IL-22 does not share any biologicalactivity with IL-10, these 2 cytokines share acommon component of their respective receptorcomplex, IL-10Rß. Anti-IL-10Rß antibodiesindeed block the IL-22-induced acute phaseresponse in HepG2 cells (9). All receptorcomplexes for IL-10-related cytokines include along chain and a short chain, based on the lengthof the cytoplasmic domain of thesetransmembrane proteins. IL-10Rß is a typicalshort chain component, with only 76 aminoacids in the cytoplasmic domain, whose mainfunction seems to consist in recruiting the Tyk2tyrosine kinase. In addition to IL-10R ß, IL-22signalling requires the expression of a long chainprotein, called IL-22R and comprising a 319amino acid long cytoplasmic domain. This chainassociates with Jak1, and is responsible for theactivation of cytoplasmic signalling cascadessuch as the JAK/STAT, ERK, JNK and p38MAP kinase pathways.

Figure 2.

In addition to its role in IL-22 binding andsignalling, the IL-22R chain also forms afunctional heterodimeric receptor complex byassociating with IL-20R ß, the second shortchain member of the IL-10R-related receptorfamily. This complex mediates STAT-1 and –3activation by IL-20 and IL-24, but not by IL-22(11). In addition, IL-20 and IL-24 can also bindto other complexes consisting of IL-20Ra andIL-20Rß. This promiscuity in cytokine receptorusage is illustrated in Fig 2 (see also ref. 12 for areview of this new cytokine family).

LICR2: a new cytokine receptormediating antiviral activities

J.-C. Renauld, L. Dumoutier

Type II cytokine receptors include receptors fortype I and Ii interferons (IFNs) and for IL-10-related cytokines. These transmembrane proteinsare almost exclusively related by theirextracellular part, which consists of tandemfibronectin type II domains, whereas thecytoplasmic domain is associated with a tyrosinekinase of the Janus Kinase (JAK family).Byscreening genomic databases for similarity withthe extracellular domain of these receptors, weidentified a new receptor that we called LICR2(Likely Interleukin or Cytokine receptor 2). Thisreceptor binds new cytokines designated IFN-l1-3, and mediates the same activities as thosemediated by the receptors for IFN-a and ß,including antiviral and antiproliferative activities(13), raising the possibility of therapeuticapplications in viral infections and cancer.


1. Renauld JC, van der Lugt N, Vink A, vanRoon M, Godfraind C, Warnier G, Merz H,Feller A, Berns A, Van Snick J. Thymic lymphomasin interleukin 9 transgenic mice. Oncogene 1994;9:1327-32.

2. Vink A, Warnier G, Brombacher F, RenauldJC. Interleukin 9-induced in vivo expansion of the B-1lymphocyte population. J Exp Med 1999;189:1413-23.

3. Richard M, Grencis RK, Humphreys NE,Renauld JC, Van Snick J. Anti-IL-9 vaccinationprevents worm expulsion and blood eosinophilia inTrichuris muris-infected mice. Proc Natl Acad SciUSA 2000;97:767-72.

4. Renauld JC, Druez C, Kermouni A, HoussiauF, Uyttenhove C, Van Roost E, Van Snick J.Expression cloning of the murine and human interleukin9 receptor cDNAs. Proc Natl Acad Sci USA1992;89:5690-4.

5. Demoulin JB, Van Roost E, Stevens M,Groner B, Renauld JC. Distinct roles for STAT1,STAT3, and STAT5 in differentiation gene inductionand apoptosis inhibition by interleukin-9. J BiolChem 1999;274:25855-61.

94

6. Van Snick J, Houssiau F, Proost P, VanDamme J, Renauld JC. I-309/T cell activation gene-3 chemokine protects murine T cell lymphomas againstdexamethasone-induced apoptosis. J Immunol 1996;157:2570-6.

7. Richard M, Louahed J, Demoulin JB, RenauldJC. Interleukin-9 regulates NF-kappaB activity throughBCL3 gene induction. Blood 1999;93:4318-27.

8. Louahed J, Grasso L, De Smet C, Van RoostE, Wildmann C, Nicolaides NC, Levitt RC,Renauld JC. Interleukin-9-induced expression of M-Ras/R-Ras3 oncogene in T-helper clones. Blood 1999;94:1701-10.

9. Dumoutier L, Van Roost E, Colau D, RenauldJC. Human interleukin-10-related T cell-derivedinducible factor: molecular cloning and functionalcharacterization as an hepatocyte-stimulating factor.Proc Natl Acad Sci USA 2000;97:10144-9.

10. Dumoutier L, Lejeune D, Colau D, RenauldJC. Cloning and characterization of IL-22 bindingprotein, a natural antagonist of IL-10-related T cell-derived inducible factor/IL-22. J Immunol 2001;166:7090-5.

11. Dumoutier L, Leemans C, Lejeune D,Kotenko SV, Renauld JC. Cutting edge: STATactivation by IL-19, IL-20 and mda-7 through IL-20receptor complexes of two types. J Immunol 2001;167:3545-9.

12. Renauld JC. Class II cytokine receptors and theirligands: key antiviral and inflammatory modulators.Nature Rev Immunol 2003; 3:667-676.

13. Dumoutier L, Tounsi A, Michiels T,Sommereyns C, Kotenko SV,and Renauld JC.Role of the Interleukin-28 Receptor tyrosine residues forantiviral and antiproliferative activity of IL-29/IFN-lambda 1 : Similarities with type I Interferon signalling.J Biol Chem 2004;279:32269-32274.

95

SIGNAL TRANSDUCTION GROUP: STRUCTURE ANDFUNCTION OF CYTOKINE RECEPTORS

Stefan CONSTANTINESCU, Associate Member

Katharina KUBATZKY, Postdoctoral Fellow (Delori Fellow)Virginie MOUCADEL, Postdoctoral Fellow (Marie Curie Fellow)Yohan ROYER, PhD StudentJudith STAERK, PhD StudentAlexandra DUSA, PhD StudentNadine SEUBERT, Student, (Univ. Würzburg, Germany)Mingli LI, Research AssistantYan YIN, Research AssistantLoic GARCON, Marie Curie Visiting Student, Institut Gustave Roussy,

Villejuif, FranceYaniv MALKA, Marie Curie Visiting Student, Tel-Aviv University, IsraelSylvie HOPPE, Administrative Assistant

Cytokines and their receptors are critical for the formation of mature blood cells and for the function of the immunesystem. Signaling by cytokine receptors is triggered by ligand-induced changes in receptordimerization/oligomerization, which induces the activation of cytosolic Janus tyrosine kinases (JAK). Thesekinases phosphorylate downstream proteins, like receptors themselves, signal transducers and activators oftranscription (STAT) proteins and a variety of other signaling proteins. We study the signal transductionmechanisms and biologic functions of cytokine receptors such as the receptors for erythropoietin (Epo),thrombopoietin (Tpo), interleukin 2 (IL2) and interleukin 9 (IL9) and their involvement in diseases such asPolycythemia Vera or leukemias. The assembly of cell-surface receptor complexes, the structure and orientation ofthe transmembrane (TM) and cytosolic juxtamembrane (JM) domains, and the regulation by JAK kinases ofreceptor traffic are major focuses. We also study the mechanisms by which STAT proteins become constitutivelyactivated and how they function in transformed hematopoietic or patient-derived leukemia cells.

Determination of the interface andorientation of the activatederythropoietin receptor dimer

N. Seubert, Y. Royer, J. Staerk, K. Kubatzky, V.Moucadel

The determination of the interface of the activeerythropoietin receptor (EpoR) dimer has been apriority during the last year (1). Epo binding tothe erythropoietin receptor (EpoR) results insurvival, proliferation and differentiation oferythroid progenitors into mature red bloodcells. We have shown that, in the absence ofEpo, the cell-surface EpoR is dimerized in aninactive conformation, which is stabilized byinteractions between the TM sequences (2). Epobinding to the extracellular EpoR domaininduces a conformational change of the receptor,

which results in the activation of cytosolic JAK2proteins (Figure 1A and B). The a -helicalorientation of the (TM) and cytosolic JMdomains is crucial for receptor activation (3). Incollaboration with Lily Huang and HarveyLodish, Whitehead Institute, Cambridge, MA,USA, we have identified a number of keyresidues in the EpoR cytosolic JM domain,which are required for switching on the activityof JAK2 and initiate signaling (4).

To identify the residues that form theinterface between the receptor monomers in theactivated EpoR dimer we have replaced theEpoR extracellular domain with a coiled-coildimer of a-helices (1).Because coiled-coils have a characteristic heptadrepeat with hydrophobic residues at positions a(one), d (four), the register of the coiled-coil a-helices is imposed on the downstream TM a-

96

helix and intracellular domain. This allows theprediction of the positions (on an a-helix) of theresidues that will be in the interface of theactivated EpoR dimer. We have generatedseven different constructs where all sevenpossible orientations were imposed by thecoiled-coil on the fused TM and intracellulardomain of the EpoR. All seven fusion proteinsare expressed when transduced in cytokine-dependent cell lines and reach the cell-surface.However, only two of the seven fusion proteinsshowed activity represented by stimulation ofproliferation of cytokine-dependent cell lines

and erythroid differentiation of primary fetalliver cells (1). The predicted dimeric interfaces ofthe two active fusion proteins are very close,emphasizing the notion that a unique dimericEpoR conformation is required for activation ofsignaling. In this active conformation TMresidues L241 and L244 and JM residue W258are predicted to be in the interface. Thisapproach of exploring orientation-dependentsignaling is now applied in our group for thedetermination of the active interfaces of thethrombopoietin and G-CSF receptors.

(A) The erythropoietin receptor (EpoR) is an inactive dimer on the cell-surface in the absence of ligand due tointeractions between the transmembrane (TM) domains (interrupted line). Cytosolic Janus kinase 2 (JAK2) is boundto the receptor juxtamembrane domain and stimulates receptor folding and traffic to the cell-surface.(B) Erythropoietin (Epo) binding induces a conformational change in the extracellular domain which is transmittedvia the a-helical TM domain to receptor residues in the juxtamembrane domain that contact JAK2 and switch itsactivity on. Activated JAK2 phosphorylates (-P) itself, the receptor cytosolic domain and many other signalingproteins leading to survival, proliferation and differentiation of erythroid progenitors into mature red blood cells.(C) Co-expression of the EpoR with the gp55-A envelope protein of the Spleen Focus Forming Virus results in cell-surface complex formation due to specific interactions between the TM domains. The EpoR transmembrane dimer,which maintains the EpoR inactive, is disrupted by the interaction with gp55-A. The receptor acquires aconformation, which allows weak constitutive activation resulting in erythroleukemia with low numbers of red bloodcells (anemia).(D) Co-expression of the EpoR with the viral gp55-P envelope protein results in cell-surface complex formation dueto specific interactions between the TM domains. In this complex the EpoR acquires a dimeric conformation verysimilar to that induced by Epo, which results in strong constitutive activation of EpoR signaling leading toerythroleukemia and massive production of mature red blood cells (polycythemia).

97

Structural studies on thetransmembrane and juxtamembranecytosolic sequences of the EpoR

K. Kubatzky, J. Staerk, M. Li, A. Dusa

To define the interfaces of the active andinactive EpoR dimers we performed cysteinescanning mutagenesis of the extracellularjuxtamembrane and TM domains (5). Weisolated three constitutively active novel mutantsof the EpoR where residues L223, L226 or I227were mutated to cysteine. These three mutantsas well as cysteine mutants of residues 220-230formed disulfide-bonded dimers. Cysteine-mediated maleimidyl crosslinking indicated thatthe first five TM residues are not helical and thatthe interface of the active EpoR dimer containsresidues L241 and L244 (5). The same residueswere found to be in the interface of the activecoiled-coil-EpoR fusion proteins (1). These datashow that in the native structure the TMdomains of the EpoR are closely interacting witheach other.

The structure of the cytosolic domains ofcytokine receptors remains a mystery. Due toour previous work showing that the junctionbetween the EpoR TM and intracellular domainis rigid (3) we hypothesized that the importantstructured segments of the cytosolic domainmust contain the cytosolic JM segment. We havecloned the cDNAs coding for the seven coiled-coil-EpoR fusion proteins in the pET31b vectorin order to produce recombinant fusion proteinsin quantities amenable for biophysical andstructural studies. In collaboration with thegroup of Steven Smith, SUNY, Stony Brook,NY, USA, we will determine the NMR structureof the EpoR TM and cytosolic JM domains inthe active and inactive coiled-coil-EpoR fusionproteins.

Signaling by the thrombopoietinreceptor

J. Staerk

The thrombopoietin receptor (TpoR) is essentialfor formation of platelets, for renewinghematopoietic stem cells and for expandingmyeloid progenitors. Like the EpoR, the TpoR isthought to signal by activation of JAK2, ofseveral STATs (STAT1, 3 and 5) as well as ofMAP-kinase, PI-3-kinase and AktB. However,

TpoR and EpoR signal quite differently sinceonly TpoR can induce hematopoieticdifferentiation of embryonic stem cells orstimulate the earliest stages of hematopoiesis inimmature hematopoietic cells. In contrast, onlyEpoR can support efficient formation of maturered cells. Since both EpoR and TpoR aremembers of the homodimeric class of cytokinereceptors we have started to compare signalingand gene expression induced by these tworeceptors as well as the orientation of theirintracellular domains in the activated state. Wehave already identified a major differencebetween TpoR and the EpoR, namely thatseveral orientations of the TpoR are compatiblewith inducing cell proliferation but only arestricted dimeric conformation is compatiblewith the induction of intercellular adhesion. Byconstructing chimeric receptors we attempt toidentify the relevant sequences which confer tothe TpoR and EpoR their specific biologicactivities in immature hematopoietic progenitorsand committed erythroid progenitors,respectively

Signaling by the IL9R via thecommon g chain (gc) and traffic ofcytokine receptors to the cell-surface

Y. Royer

gc is a cytokine receptor that is shared by thereceptor complexes of several cytokines, such asIL2, IL4, IL7, IL9, IL15 and IL21. gc binds andactivate JAK3. Humans that lack the gc or havemutations in JAK3 develop severe combinedimmunodeficiency (SCID). We study the regionsimportant in the IL9R and the gc for JAK1 andJAK3 interactions. For both of them, the regioncomprising Box1 and Box2 is necessary for theJAKs binding. In contrast, the region betweenthe juxtamembrane domain and Box1 has quitedifferent sequence requirements. While IL9R israther similar to the EpoR (3), g c does notrequire hydrophobic residues before Box1.Finally, by confocal microscopy, the distributionof the gc on the cell surface is totally differentthan the one of the IL9R or the EpoR. The gc

forms big patches whereas the IL9R and theEpoR exhibit a normal diffuse distribution, aswe have shown in collaboration with PierreCourtoy. Paraformaldehyde fixation wasrequired for the appearance of the big patches inthe case of the gc, suggesting that the receptor is

98

localized in membrane regions whereformaldehylde can not block diffusion andaggregation of membrane proteins. Thisobservat ion could revea l d i f ferentcompartmentalization of receptors on the cell-surface.

Traffic and cell-surface expression of theEpoR critically depend on the ability of EpoR tobind JAK2 intracellularly (4). In the absence ofJAK2, the receptor does not get transportedfrom the endoplasmic reticulum (ER) to theGolgi apparatus and does not acquire EndoHresistance. We have observed that, inhematopoietic cells, overexpressing cognate JAKproteins, leads to enhanced cell-surfacelocalization of several cytokine receptors (i.e.TpoR, IL9R, IL2R, gc). The IL9Ra whichrequires JAK1 for signaling is expressed athigher levels on the cell-surface when JAK1 butnot JAK2 or JAK3 is overexpressed. Ourworking hypothesis is that the N-terminusFERM domain of JAK proteins exerts a genericpro-folding effect on cytosolic domains ofcytokine receptors. We are testing thishypothesis on several different cytokinereceptors and are investigating the link betweenproper folding in the ER and transport to thecell-surface.

Sequence–specific interactionsbetween transmembrane domains

A. Dusa

Two transmembrane viral envelope proteins(gp55-P and gp55-A) belonging to thepolycythemic (P) and anemic (A) Spleen FocusForming Virus (SFFV) strains, can activate theEpoR when co-expressed in the same cell (6).In collaboration with Yoav Henis, Tel-AvivUniversity, Israel, we have shown that both thegp55-A and gp55-P TM domains specificallyinteract with the TM domain of the EpoR(Figure 1C and D)(7). gp55-A weakly activatesthe receptor leading to erythroleukemia with lownumber of red blood cells (anemia). gp55-P fullyactivates the EpoR inducing erythroleukemiawith elevated levels of red cells (polycythemia). Thebasis for this difference between gp55-P andgp55-A is represented by differences in specificbinding of the TM domains to the TM domainof the EpoR (8). Taking advantage of thisspecific interaction we are constructing a geneticsystem where the TM sequence of gp55-P israndomized and tested for the ability to bind and

activate the EpoR. In this system activation ofEpoR signaling will result in cell survival andproliferation, which represent a powerfulselection.

Constitutive activation of JAK-STATsignaling pathways and genestargeted by STAT5 in transformedhematopoietic and patient-derivedleukemia cells

V. Moucadel, J. Staerk

Cytokine stimulation of cytokine receptors,induces transient activation of the JAK-STATpathway. In contrast several mutant forms ofcytokine receptors have been isolated that signalconstitutively (i.e. EpoR R129C or TpoRS498N) [reviewed in 9]. Such receptors arepermanently dimerized in an activated state andinduce the biologic effects of the wild typereceptors as well as leukemic celltransformation. In cultured cells this process isstudied by expressing oncogenic forms ofcytokine receptors in cytokine-dependent cellsand assaying for their transformation into cellsthat grow autonomously, in the absence of anycytokine. In the transformed cells many of thetransient signaling events induced by cytokinesare detectable permanently, i.e. ligand-independent phosphorylation of JAK and STATproteins or high levels of nuclear activatedSTATs especially STAT5 and STAT3. A similarpicture has been noted in patient derivedleukemia cells. The critical questions we wouldlike to answer concern the mechanisms by whichthe JAK-STAT remain permanently activated intransformed cells and which genes are regulatedby constitutively active STAT proteins inleukemic cells. Using chromatin immuno-precipitation and sequencing of nativepromoters bound by STAT5 we noted that intransformed cells STAT5 can also bind to lowaffinity N4 sites not only to N3 sites, which ischaracteristic of ligand-activated STAT5 (12).Furthermore, while cytokines such as Epoactivate both STAT5A and STAT5B, we haverecently observed that in transformedhematopoietic cells it is mainly STAT5B that isconstitutively active (12).

We are attempting to identify the promotersactually bound by STAT proteins in living cellsin physiologic and pathologic situations. We usea modified version of the chromatin

99

immunoprecipitation assay pioneered by AlexVarshavsky in conjunction with DNAmicroarray gene profiling. The isolated genomicfragments are screened for the presence ofSTAT-binding sites and tested for the ability toregulate transcription of reporter genes. Newlyidentified genes regulated by such genomicsequences will be expressed in bicistronicretroviral vectors that allow wide expression ofcDNAs at physiologic levels.


1. Seubert N, Royer R, Staerk J, Kubatzky, KF,Moucadel V, Krishnakumar S, Smith SO,Constantinescu SN. Active and inactive orientationsof the transmembrane and cytosine domains of theerythropoietin receptor dimer. Mol Cell 2003 ;12:1239-50.

2. Constantinescu SN, Keren T, Socolovsky M,Nam H, Henis YI, Lodish HF. Ligand-independentoligomerization of cell-surface erythropoietin receptor ismediated by the transmembrane domain. Proc NatlAcad Sci USA 2001;98:4379-84.

3. Constantinescu SN, Huang LJ, Nam H,Lodish HF. The erythropoietin receptor cytosolicjuxtamembrane domain contains an essential, preciselyoriented, hydrophobic motif. Mol Cell 2001;7:377-85.

4. Huang LJ, Constantinescu SN, Lodish HF.The N-terminal domain of Janus kinase 2 is required forGolgi processing and cell surface expression oferythropoietin receptor. Mol Cell 2001;8:1327-38.

5. Kubatzky, KF, Liu W, Li M, Royer Y,Simmerling C, Smith SO, Constantinescu SN.Structure of the extracellular juxtamembrane region ofthe erythropoietin receptor!: novel constitutive activereceptor mutants. Submitted EMBO J, 2004.

6. Constantinescu SN, Wu H, Liu X, Beyer W,Fallon A, Lodish HF. The anemic Friend virus gp55envelope protein induces erythroid differentiation in fetalliver colony-forming units-erythroid. Blood 1998;91:1163-72.

7. Constantinescu SN, Keren T, Russ, W,Ubarretxena-Belandia I, Malka Y, Kubatzky, K,Engelman DM, Lodish, HF, Henis YI. The Eporeceptor transmembrane protein modulates complexformation with viral anemic and polycythemic gp55proteins. J Biol Chem 2003;278:43755-63.

8. Constantinescu SN, Liu X, Beyer W, Fallon A,Shekar S, Henis YI, Smith SO, Lodish HF.Activation of the erythropoietin receptor by the gp55-Pviral envelope protein is determined by a single amino acidin its transmembrane domain. EMBO J 1999;18:3334-47.

9. Constantinescu SN, Moucadel V. STATsignaling by erythropoietin. In Signal Transducersand Activators of Transcription (STATs)!:Activation and Biology (Editors PB Sehgal, THirano, DE Levy) 2003;12:575-593, KluwerAcademic Publishers, New York.

10. Royer Y, Menu C, Liu X, ConstantinescuSN. High throughput Gateway bicistronic retroviralvectors for stable expression in mammalian cells!:exploring the biologic effects of STAT5 overexpression.DNA and Cell Biol 2004;23:355-65.

Research at the Christian de Duve Institute of Cellular ... Report DDU… · Structural studies on...

Documents

Transcript of Research at the Christian de Duve Institute of Cellular ... Report DDU… · Structural studies on...