Intracellular localization of the glycosyl-phosphatidylinositol-specific

phospholipase C of Trypanosoma brucei

ROLAND BtJLOW1'*, GARETH GRIFFITHS2, PAUL WEBSTER3, YORK-DIETER STIERHOF4, FRED

R. OPPERDOES5 and PETER OVERATH1'!

lMax-Planck-Institut fiir Biologie, Corrensstrasse 38, D7400 Tubingen, Federal Republic of Germany^European Molecular Biology Laboratory, Poslfach 10.2209, D6900 Heidelberg, Federal Republic of Germany^ International Laboratory for Research on Animal Diseases, Nairobi, Kenya*Max-Planck-Instttut fiir Entwicklungsbiologie, Spemannstr. 35, D74O0 Tubingen, Federal Republic of GermanyJ'International Institute of Cellular and Molecular Pathology, University Catholique de Louvain, B1200 Brussels, Belgium

•Present address: Department of Medical Microbiology, Sherman Fairchild Science Building, Stanford, CA, USAf Author for correspondence

Summary

Glycosyl-phosphatidylinositol-specific phospholi-pase C (GPI-PLC) is a membrane-bound enzyme ofbloodstream forms of Trypanosoma brucei, whichcleaves the GPI-membrane anchor of the variantsurface glycoprotein forming diacylglycerol and1,2-cyclic phosphate on the inositol ring. The cellu-lar localization of the enzyme was studied byfractionation of sub-cellular organelles and immu-nofluorescence microscopy and was found to beprimarily cytoplasmic. This was confirmed byimmuno-electron microscopy using cryo-sections,

which showed that the labelling was predominantlyon the cytoplasmic side of intracellular membranesbut was absent from the plasma membrane includ-ing the region lining the flagellar pocket. Thesignificance of these results for the possible func-tion of the phospholipase is discussed.

Key words: glycosyl-phosphatidylinositol-specificphospholipase C, variant surface glycoprotein, Trypanosomabrucei, sub-cellular fractionation, cryo-section immuno-electron microscopy.

Introduction

Trypanosoma brucei contains a glycosyl-phosphatidyl-inositol (GPI)-specific phospholipase C (PLC), whichcleaves the COOH-terminal GPI anchor of the variantsurface glycoprotein (VSG), forming diacylglycerol and1,2-cyclic phosphate on the inositol ring (Cardoso deAlmeida & Turner, 1983; Ferguson & Cross, 1984;Ferguson et al. 1985, 1988). The enzyme is present onlyat those stages of the parasite's life cycle in which theorganism is covered with a VSG coat: the bloodstreamforms in the mammalian host and the metacyclic forms inthe salivary glands of the tsetse fly (Lamont et al. 1987;Ross et al. 1987; Grab et al. 1987). Bloodstream formsdifferentiate into coatless procyclic forms after ingestionby the vector, or in suitable culture systems. Procyclicforms contain no detectable enzyme activity (Biilow &Overath, 1985; Turner et al. 1985; Grab etal. 1987), andvery low levels of GPI-PLC-specific mRNA (Carringtonet al. 1989). In cell lysates of bloodstream forms, theenzyme is found in the particulate fraction and solubiliz-ation requires the use of detergents (Hereld et al. 1986;Fox et al. 1986; Bulow & Overath, 1986). After phase

Journal of Cell Science 93, 233-240 (1989)Printed in Great Britain © The Company of Biologists Limited 1989

separation of a Triton X-114 solution, GPI-PLC par-titions into the detergent phase (Fox et al. 1986).Moreover, the purified enzyme can be effectively andstably incorporated into liposomes (Billow & Overath,1986). By these criteria, GPI-PLC is considered to be anintegral membrane protein. In contrast, the primarysequence of the protein does not contain any obviousfeatures that would indicate how the enzyme could beassociated with a membrane (Hereld et al. 1988; Carr-ington et al. 1989). There is no easily recognizableN-terminal signal sequence, so the protein would not beexpected to be translocated across the membrane of theendoplasmic reticulum. Furthermore, the amino acidsequence contains no extended runs of hydrophobicamino acids. Therefore, the protein is likely to besynthesized on cytoplasmic ribosomes and, presumably,some part of the sequence must fold in such a way thatallows it to penetrate the lipid bilayer. Alternatively, theprotein may be covalently modified by a hydrophobiccomponent.

GPI-PLC is the only enzyme present in bloodstreamtrypanosomes that converts the membrane-form VSG(mfVSG) to the soluble form (sVSG, cf. Biilow &

233

Overath, 1986). Therefore, the enzyme is undoubtedlyresponsible for the rapid cleavage of mfVSG when cellsare lysed by hypotonic shock or detergents. However, thefunction of the enzyme under physiological conditions isnot known. In particular, it is unclear whether thephospholipase plays an obligatory role in the release ofthe surface VSG during the differentiation of blood-stream to procyclic forms (Billow et al. 1989).

In order to define the function of the GPI-PLC it isimportant to know its intracellular localization. On thebasis of cell fractionation experiments, Grab et al. (1987)concluded that the enzyme resides in the part of theplasma membrane lining the flagellar pocket as well asintracellular membranes such as the Golgi apparatus,coated vesicles and endocytic organelles. In the presentstudy we have used three different approaches to eluci-date in greater detail the intracellular localization of GPI-PLC. The results all show that this enzyme is notdetectable on the plasma membrane and suggest insteadthat it is associated with the cytoplasmic side of intra-cellular membranes. A preliminary account of part of thiswork has been published (Overath et al. 1987).

Materials and methods

TrypanosotnesThe monomorphic T. brucei variant clones MITat 1.4 andMITat 1.2 were obtained from Dr G. A. M. Cross (RockefellerUniversity, New York). Bloodstream forms were harvestedfrom the blood of infected mice or rats. Established procyclicforms from variant MITat 1.4 were grown in DTM medium(Overath et al. 1986).

Cell fractionationBloodstream trypomastigotes were disrupted in homogeniz-ation buffers containing either 0-25 M-sucrose, 1 mM-imidazole-HC1 and 1 mM-EGTA (pH70) or 025 M-sucrose, 25 mM-Tris-HCl and 1 mM-EGTA (pH 7-2), and a cytoplasmic extractdevoid of nuclei and cells was prepared (Opperdoes & Steiger,1981). Marker enzymes in cell fractions after density equili-bration were determined according to Steiger et al. (1980),Opperdoes & Steiger (1981) and Gbenlee/a/. (1986), and GPI-PLC activity was estimated as described by Biilow & Overath(1986). The distribution pattern of enzymes is presented asdescribed by Steiger et al. (1980).

Trypsin treatment of trypanosomesBloodstream forms of the variant clone MITat 1.2 harvestedfrom mouse blood were washed twice and suspended in 1 ml ofmedium B ( lx 10s cells ml"1, cf. Brun et al. 1981). Alterna-tively, 1X108 cells were first lysed in 200 jul water and thenmixed with 800 ̂ tl medium B. Samples were incubated at 37°Cand 10/ig trypsin (Sigma Chemical Co., St Louis, MO; Cat.no. T8642) was added. After various times, 200/il samples weremixed with 800^1 phosphate-buffered saline (PBS) containing1 mM-tosyl-L-lysine chloromethyl ketone, 1 mM-phenylmethyl-sulfonyl fluoride, lOmgmr 1 Triton X-100 and 10% inacti-vated sheep serum. Lysates were subsequently processed for thedetermination of GPI-PLC activity.

ImmunofluorescenceBloodstream or procyclic trypanosomes were fixed in 2%

formaldehyde in medium B for lOmin at room temperature.After washing the cells in PBS the cells were mounted on glassslides previously treated with l m g m l " polylysine and thentreated sequentially with PBS/10% sheep serum (30min),monoclonal anti-GPI-PLC antibody mAT3 (Biilow & Overath,1986) for 60min, biotinylated sheep anti-mouse Ig antibody(1:200, 15min, Amersham Buchler, Braunschweig, FRG, Cat.no. RPN 1001) and FITC-streptavidin (1:200, 15min, RPN1232). The slides were mounted in 50% glycerol, 25mgml~1

1,4 diazabicyclo(2,2,2)-octane (Aldrich Chemie, Steinheim,FRG) and inspected in a Leitz Ortholux fluorescence micro-scope. A polyclonal rabbit anti-GPI-PLC antiserum was raisedby multiple injections of immuno-affinity-purified enzymeinitially in complete, and subsequently in incomplete, Freund'sadjuvant.

hnmuno-electron microscopyThe trypanosome preparations were centrifuged for 2min at1000 g and fixed with a solution of 4% paraformaldehyde in200 mM-Hepes buffer, pH 7-4, for 5 min, followed by a solutionof 8 % paraformaldehyde in the same buffer for times rangingfrom <1 to 24h. The preparation of cryo-sections and theirlabeling with antibodies and gold was done as described(Griffiths et al. 1983, 1984). Since the anti-phospholipaseantibodies are monoclonals that bind protein A with highaffinity, they could be visualized directly using protein A-gold.Double labeling was done by the method of Geuze et al. (1981).Controls involved the use of a monoclonal antibody against thecytoplasmic domain of the vesicular stomatitis virus G (VSV-G) protein, a gift from T. Kreis (Kreis, 1986). For labeling ofVSG a polyclonal rabbit anti-MITat 1.4 VSG serum was used.

For the quantification of anti-GPI-PLC labeling, randommicrographs of cryo-sections showing good presentation, andwhich were relatively thin, were taken at a primary magnifi-cation of x 13 000. The thickness can be roughly estimated fromthe electron density at low magnification and we have previouslycorrelated this thickness to mass-thickness measurements onhydrated cryo-sections. Our routine sections are 80-110 nmthick (Griffiths et al. 1984). The sections were labeled eitherwith anti-phospholipase, or anti-VSV-G protein as a control,followed by rabbit anti-mouse antibody and 9 nm proteinA-gold. The micrographs were enlarged by a factor X41 on anenlarger built at EMBL. For the plasma membrane all goldparticles within about 20 nm were counted as well as intersec-tions of the lines of a lattice grid as described by Griffiths &Hoppeler (1986). A simple formula described in the latter paperenables the number of gold particles per f.im of plasmamembrane to be estimated. For the cytoplasmic labeling all goldparticles over the total cytoplasm, including those closelyassociated with membrane structures (within 20nm), werecounted. All the particles associated with the lumen of thenucleus, Golgi complex, the matrix of glycosome-like structuresor mitochondria, and with all endosome-lysosome type struc-tures, were ignored because their labeling was not abovebackground. The number of gold particles was related to thenumber of points on a lattice grid and converted to goldparticles per /lm2 of cytoplasm as described by Griffiths &Hoppeler (1986).

Results

Sub-cellular fractionation

In the first approach, we determined the activity of GPI-PLC in sub-cellular organelle fractions obtained byisopycnic sucrose-gradient centrifugation (Opperdoes &

234 R. Biilow et al.

Borst, 1977; Steiger et al. 1980; Opperdoes & Steiger,1981). The distribution profiles of constituents fromcytoplasmic extracts are shown in Fig. 1. GPI-PLC was•entirely membrane-associated, since no activity wasfound associated with a soluble marker enzyme, alanine

NADP-isocitratedehydrogenase

111 1-25 1-11Density (gcm~3)

1-25

F ig . 1. Distribution profiles of bloodstream T. bruceienzymes and organelle markers after isopycnic centrifugationof a cytoplasmic extract. Centrifugation was performed in avertical rotor in two otherwise identical linear sucrosegradients. One contained 1 mM-imidazole buffer + 1 mM-EGTA, p H 7 - 0 ( ); the other 25 mM-Tris-HCl, 1 mM-EGTA, p H 7 - 0 ( ).

aminotransferase. Furthermore, its distribution wasclearly different from that of two marker enzymes for theplasma membrane, tf-D-glucosidase and 3'-nucleotidase.Importantly, the phospholipase did not show the charac-teristic density shift observed for tf-D-glucosidase or 3'-nucleotidase (not shown), when cells were homogenizedin 25 mM-Tris-HCl instead of 1 mM-imidazole (comparebroken and continuous curves in fig. 1 cf. Opperdoes &Steiger, 1981). This decrease in density has been at-tributed to the dissociation of sub-pellicular micro-tubules. By these criteria, an association of GPI-PLCwith the plasma membrane could be excluded. Thebimodal distribution profile of this enzyme did notcoincide with that of any marker protein. Part of theactivity (p= l-25gcm~ ) corresponded to the density ofglycosomes, exemplified by hexokinase as a marker. Thebroad peak of phospholipase at lower density(p = l - l l - l -18gcm~ ) corresponded most closely to thedistribution of acid phosphatase. The latter enzyme,thought to be associated with the membrane lining theflagellar pocket as well as intracellular membranes (Lan-greth & Balber, 1975; Steiger etal. 1980), did not exhibitthe density shift characteristic of the plasma membrane-associated enzymes discussed above. Finally, the densitycorresponding to the broad phospholipase peak over-lapped with the part of the gradient where the lysosomes(cr-D-mannosidase) and mitochondrial membranes (iso-citrate dehydrogenase) were concentrated. In summary,these cell fractionation experiments suggested that GPI-PLC was associated with membranes of intracellularorganelles and, possibly, the specialized part of theplasma membrane forming the flagellar pocket, and thatit was absent from the bulk of the plasma membrane.

Trypsin treatmentThe following experiment appeared to exclude the possi-bility that GPI-PLC was located at the cell surface. Intactcells of the variant clone MITat 1.2 were treated withtrypsin. Within a few minutes, this protease degraded themfVSG, releasing a characteristic fragment into themedium (data not shown). The cells remained motile andretained their elongated shape for about 20min, thenrounded up assuming a tadpole-like morphology with theflagellum still motile but, after a further 10 min, started todisintegrate. As shown in Fig. 2, trypsin treatment didnot affect the phospholipase activity for about 20 min; thesubsequent decline of the activity correlated with celllysis. In contrast, the major part of the phospholipase in acell lysate was trypsin-sensitive.

Imtnunofluorescence studiesA number of immunofluorescence experiments weredone in order to determine the localization of the GPI-PLC at the whole-cell level. Fig. 3 shows trypanosomesfixed for only 10 min with formaldehyde and then labeledwith the anti-GPI-PLC-specific monoclonal antibodymAT3 and biotinylated anti-Ig/FITC-streptavidin. Flu-orescence was largely confined to intracellular vesicles,some in close association with the plasma membrane.Procyclic cells could not be labeled under these con-ditions (data not shown). Fixation of bloodstream trypa-

Intracellular phospholipase C in Trypanosoma 235

nosomes with formaldehyde or glutaraldehyde for severalhours abolished the detection of the enzyme by mono-clonal antibodies or a polyvalent anti-GPI-PLC anti-serum. More drastic fixation probably prohibited theaccess of the antibodies to the intracellular proteinbecause antigenicity was retained when purified GPI-PLC adsorbed to nitrocellulose sheets was treated withfixatives under the same conditions (data not shown) aswell as in the immuno-electron microscopy experiments(below). Finally, attempts to label live bloodstreamforms were consistently negative.

Immuno-electron microscopyThe association of GPI-PLC preferentially with thecytoplasmic side of intracellular membranes could bedemonstrated by cryo-section immuno-electron mi-croscopy using monoclonal antibodies and proteinA-gold particles. Fig. 4 and Fig. 5A show representativesections; the evaluation of a large number of sections is

lOU»s-°—o

g 80

1 60

Uo.EuO

40

20

\

10 20 30 40Time (min)

50 60

Fig. 2. GPI-PLC activity after treatment of live bloodstreamforms of variant MITat 1.2 (O O) or a corresponding celllysate ( • • ) with trypsin.

summarized in Table 1. Although low in density, thelabeling of bloodstream forms was both specific andhighly significant when compared with that over sectionslabeled with a control antibody (cf. Table 1, 'cytoplasm')or of procyclic forms (Table 1 and Fig. 5B). Most of thelabel (about 70 %) was associated with the cytoplasmicside of intracellular membranes or over electron-denseregions of cytoplasm (Figs 4E and 5A). The preciseidentity of these membranes/structures was not clear.Often, the label was closely associated with glycosome-like structures (Fig. 4A, compare also with Fig. 1).When the sections were double-labeled with both anti-VSG and anti-phospholipase antibodies, structures thatlabeled with anti-GPI-PLC were not, in general, labeledwith anti-VSG (Figs 4E and 5A). Conversely, the plasmamembrane, flagellar pocket and Golgi complex, whichwere labeled significantly with anti-VSG, were notlabeled with anti-phospholipase antibodies. The intenselabeling of the surface, the flagellar pocket and intracellu-lar vesicles with anti-VSG antibodies compared with thefew gold particles associated with GPI-PLC is a reflec-tion of the molar ratio of 300 to 400 for these antigens;one cell contains about 107 molecules of VSG (Cross,1975) but only about 30000 molecules of GPI-PLC(Biilow & Overath, 1986; Hereld et al. 1986). Quantifi-cation of the VSG versus GPI-PLC labeling gave anaverage ratio of 258 to 1, in good agreement with therelative abundance of these molecules.

Discussion

Taken together, the results of the experiments presentedin this paper provide a consistent picture of the localiz-ation of GPI-PLC in T. brucei bloodstream forms. Theenzyme is not associated with the inner or outer face ofthe plasma membrane, including the specialized regionforming the flagellar pocket. The suggestion by Grab et

Fig. 3. Detection of GPI-PLC byimmunofluorescence inbloodstream forms fixed for 10 minat room temperature with 2 %formaldehyde, using themonoclonal antibody mAT3.

236 R. Biilow et al.

4A

B

•r

c

Fig. 4. Images of cryo-sections of bloodstream form trypanosomes labeled with monoclonal antibody against GPl-PLC followedby rabbit anti-mouse igG and protein A-gold. In A-D the arrows indicate the gold particles. A. A low magnification image inwhich labeling is apparent in the vicinity of glycosome-like structures (arrowhead); B-D, show evidence of labeling nearmembrane structures in the vicinity of the flagellar pocket (/). A coated pit structure is indicated in D (arrowhead). In E thesection was double labeled with anti-phospholipase (9nm gold, arrows) and rabbit anti-VSG (6nm gold). The plasmamembrane (/>), indicated in A, C and E, is characteristically unlabeled with the anti-phospholipase antibody. A, X 40 000; B,X77 000; C, X60000; D, X86000; E X81000.

Intracellular phospholipase C in Trypanosoma 237

1

5A

*

B

Fig. 5. A shows a bloodstream form trypanosome double labeled with anti-GPI-PLC (9nm gold, arrows) and anti-VSG (6nmgold). An area in the vicinity of the flagellar pocket (f) and Golgi complex (G) is shown. The precise location of thephospholipase is not clear in this image. B shows a section through procyclic forms labeled with anti-phospholipase. No goldparticles are evident in this image, p, plasma membrane; n, nucleus. A, X80000; B, X32000.

238 R. Bulotu et al.

Table 1. Quantification oj GPI-PLC from cryo-sections

No. of gold particles ftm~

mATl mAT3 Control

'Cytoplasm'Bloodstream form

Procyclic form

Plasma membraneBloodstream form

Procyclic form

NucleusBloodstream form

Procyclic form

0-64±0-10n = 32

0-1010-03n = 28

0-002 ± 0 0 0 2« = 32

0-002 ±0-002n = 28

0-27 ±0-11n = 32

0-16±0-10

0-45 ±0-06H = 40

0-22 ±0-04

0-0050 ± 00030« = 40

0-0O65 ± 0-0033

0-18 ±0-09o = 40

0-13±0-06H = 4 4

0-15 ±0-05» = 28

0-18±0-06n = 30

0-0050 ± 0-0025

0-0040 ± 0-0020n = 30

0-12±0-08n = 28

0-11 ±0-06

mATl and mAT3 refer to two independently isolated monoclonal antibodies against GPI-PLC (Bulow & Overath, 1986). Control sectionswere labeled with an antibody to the vesicular stomatitis virus protein G. 'Cytoplasm' refers to all areas of cytoplasm including the periphery ofvesicular structures but excluding identifiable organelles such as the nucleus, mitochondria, the matrix of glycosome-like structures, lysosome-like structures and Golgi complex, it refers to the number of micrographs evaluated. For further details see Materials and methods.

al. (1987) that part of the phospholipase is bound to theflagellar pocket membrane, which comprises only 0-2%of the total cellular membrane area (Coppens et al. 1987),is mainly based on its co-localization with acid phospha-tase activity in subcellular fractions. This observationwas confirmed in the present study (Fig. 1). There are atleast two possibilities for resolving the apparent discrep-ancy between the immuno-electron microscopic data andthe results of the sub-cellular fractionation experiments.First, disruption of parasites may lead to fusion of GPI-PLC-containing membranes with the flagellar pocketmembrane; thus, the co-localization of the two enzymesin sub-cellular fractions may be an artifact. Second,rather than residing predominantly in the flagellar pocketmembrane, acid phosphatase may indeed be mainlyassociated with similar intracellular organelles like GPI-PLC is. Although the cytochemical staining experimentsof Langreth & Balber (1975) and the hydrolysis ofphosphatase substrates in live cells (Steiger et al. 1980)suggest that some enzyme activity is located in the lumenor at the surface of the pocket, a quantitative evaluation ofthese experiments in terms of the molar ratio of surfaceversus intracellular phosphatase is not possible until thepermeability of the substrates employed has been inde-pendently assessed.

While both the immunofluorescence and the immuno-electron-microscopic results suggest that the phospholi-pase is associated preferentially with the peripheral faceof intracellular vesicles, the functional identity of theseorganelles remains obscure. Since the enzyme is likely tobe formed on cytoplasmic ribosomes (Carrington et al.1989), it may insert post-translationally into the mem-brane of several organelles such as glycosomes or part ofthe endocytic pathway. A more precise localization maybe possible in future double-labeling studies when anti-bodies to defined trypanosome compartments becomeavailable.

Our experiments suggest that the phospholipase isseparated from its major substrate, the mfVSG. Thelatter protein is translocated during synthesis into thelumen of the endoplasmic reticulum, rapidly providedwith a GPI anchor (Bangs et al. 1985; Ferguson et at.1986) and subsequently transported along the conven-tional secretory pathway to the flagellar pocket (Dus-zenkoeio/. 1988). Therefore, mfVSG and phospholipasewill always be on opposite sides of membrane-enclosedorganelles, in agreement with the observation that therelease of sVSG in cultured bloodstream forms is slow(Billow et al. 1989). How enzyme and substrate cometogether on the same side of a membrane during hypo-tonic cell lysis or, possibly, during coat release whileundergoing differentiation to procyclic forms remains tobe elucidated.

We thank Christina NonnengUsser, Margret Quinten (Tilb-ingen) and Ruth Back (EMBL) for expert technical assistance,Heinz Schwarz for helpful discussions, Robert Etges forcorrections of the manuscript, Thomas Kreis for the gift of theanti-VSV G antibody, and the Fond der Chemischen Industriefor support.

References

BANGS, J. D., HERELD, D., KRAKOW, J. L., HART, G. W. &ENGLUND, P. T. (1985). Rapid processing of the carboxylterminus of a trypanosome variant surface glycoprotein. Proc.natn.Acad. Sci. USA. 82, 3207-3211.

BRUN, R., JENNI, L., SCHONENBERGER, M. & SCHELL, K.-F. (1981).In vitro cultivation of bloodstream forms of Trypanosoma bnicei,Trypanosoma rhodesiense or Trypaiiosoma gwiibiense.J. Prvtozool.28, 470-479.

BOLOW, R., NONNENGASSER, C. & OVERATH, P. (1989). Release ofthe variant surface glycoprotein during differentiation ofbloodstream to procyclic forms of Trypaiiosoma bnicei. Molec.biochem. Parasitol. 32, 85-92.

BOLOW, R. & OVERATH, P. (1985). Synthesis of a hydrolase for the

Intracellular phospholipase C in Trypanosoma 239

membrane-form variant surface glycoprotein is repressed duringtransformation of Trypanosoma brucei. FEBS Lett. 187, 105—110.

BOLOW, R. & OVERATH, P. (1986). Purification and characterizationof the membrane-form variant surface glycoprotein hydrolase ofTrypanosoma bnicei.J. biol Chem. 261, 11918-11923.

CARDOSO DE ALMEIDA, M. L. & TURNER, M. J. (1983). The

membrane form of variant surface glycoproteins of Trypanosomabnicei. Nature, Land. 302, 349-352.

CARRINGTON, M., BOLOW, R., REINKE, H. & OVERATH, P. (1989).

Sequence of the glycosyl-phosphatidylinositol-specificphospholipase C of Trypanosoma brucei. Mot. Biochem. Parasitol.33, 289-296.

COPPENS, I., OPPERDOES, F. R., COURTOY, P. J. & BANDHUIN, P.

(1987). Receptor-mediated endocytosis in the bloodstream form ofTrypanosoma brucei. J. Protozool. 34, 465-473.

CROSS, G. A. M. (1975). Identification, purification and properties ofclone-specific glycoprotein antigens constituting the surface coat ofTrypanosoma bnicei. Parasitology 67, 393-417.

DUSZENKO, M. , IVANOV, I. E., FERGUSON, M. A. J., PLESKEN, H . &CROSS, G. A. M. (1988). Intracellular transport of a variantsurface glycoprotein in Trypanosoma brucei. J. Cell Biol. 106,77-86.

FERGUSON, M. A. J. & CROSS, G. A. M. (1984). Myristylation of themembrane form of a Trypanosoma brucei variant surfaceglycoprotein. J . biol. Chem. 259, 3011-3015.

FERGUSON, M. A. J., DUSZENKO, M., LAMONT, G. S., OVERATH, P.& CROSS, G. A. M. (1986). Biosynthesis of Trypanosoma bruceivariant surface glycoproteins. A'-glycosylation and addition of aphosphatidylinositol membrane anchor. J. biol. Chem. 261,356-362.

FERGUSON, M. A. J., HOMANS, S. W., DWEK, R. A. &

RADEMACHER, T. W. (1988). Glycosylphosphatidylinositol moietythat anchors Trypanosoma brucei variant surface glycoprotein tothe membrane. Science 239, 753-759.

FERGUSON, M. A. J., Low, M. G. & CROSS, G. A. M. (1985).Glycosyl-s(i-l,2-dimyristylphosphatidylinositol is covalently linkedto Trypanosoma brucei variant surface glycoprotein. J. biol. Chem.260, "14547-14555.

Fox, J. A., DUSZENKO, M., FERGUSON, M. A. J., Low, M. G. &CROSS, G. A. M. (1986). Purification and characterization of anovel glycan-phosphatidylinositol-specific phospholipase C fromTrypanosoma brucei. J. biol. Chem. 261, 15 767-15 771.

GBENLE, G. O., OPPERDOES, F. R. & VAN ROY, J. (1986).Localization of 3'-nucleotidase and calcium-dependentendonbonuclease in the plasma-membrane of Trypanosoma brucei.Ada tropica 43, 295-305.

GEUZE, H. J., SLOT, J. W., VAN DER LEY, P. A., SCHEFFER, R. T.C. & GRIFFITH, J. M. (1981). Use of colloidal gold particles indouble-labelling immunoelectron microscopy of ultrathin tissuesections. 7. Cell Biol. 89, 653-665.

GRAB, D. J., WEBSTER, P., ITO, S., FISH, W. R., VERJEE, Y. &LONSDALE-ECCLES, J. D. (1987). Subcellular localization of avariable surface glycoprotein phosphatidylinositol-specificphospholipase-C in African trypanosomes. J. Cell Biol. 105,737-746.

GRIFFITHS, G. & HOPPELER, H. (1986). Quantitation inimmunocytochemistry: correlation of immunogold labeling to

absolute number of membrane antigens. J. Histochem. Cvtochem.34, 1389-1398.

GRIFFITHS, G., MCDOWALL, A., BACK, R. & DUBOCHET, J. (1984).

On the preparation of cryosections for immunocytochemistry. J.Ultrastruct. Res. 89, 65-78.

GRIFFITHS, G., SIMONS, K., WARREN, G. & TOKUYASU, K. T.

(1983). Immunoelectron microscopy using thin, frozen sections:application to studies of the intracellular transport of SemlikiForest virus spike glycoproteins. Meth. Enzytn. 96, 466-485.

HERELD, D., HART, G. W. & ENGLUND, P. (1988). cDNA encoding

the glycosyl-phosphatidylinositol-specific phospholipase C ofTrypanosoma brucei. Proc. natn. Acad. Sci. U.SA. 85, 8914-8918.

HERELD, D., KRAKOW, J. L., BANGS, J. D., HART, G. W. &

ENGLUND, P. T. (1986). A phospholipase C from Trypanosomabrucei which selectively cleaves the glycolipid on the variantsurface glycoprotein. 7 biol. Chem. 261, 13 813-13 819.

KREIS, T. E. (1986). Microinjected antibodies against thecytoplasmic domain of vesicular stomatitis virus glycoprotein blockits transport to the cell surface. EMBOJ. 5, 931-941.

LAMONT, G. S., FOX, J. A. & CROSS, G. A. M. (1987). Glycosyl-s;;-

1,2-dimyristylphosphatidylinositol is the membrane anchor forTrypanosoma equiperdum and T. (Nannomonas) congolense variantsurface glycoproteins. Molec. biochem. Parasitol. 24, 131-136.

LANGRETH, S. G. & BALBER, A. E. (1975). Protein uptake anddigestion in bloodstream and culture forms of Trvpanosoma brucei.J. Protozool. 22, 40-53.

OPPERDOES, F. R. & BORST, P. (1977). Localization of nine glycolyticenzymes in a microbody-like organelle in Trypanosoma brucei: Theglycosome. FEBS Lett. 80, 360-364.

OPPERDOES, F. R. & STEIGER, R. F. (1981). Localization of

hydrolases in cultured procyclic trypomastigotes of Trypanosomabrucei. Molec. biochem. Parasitol. 4, 311-323.

OVERATH, P., BOLOW, R., CZICHOS, J. & EHLERS, B. (1987).

Differentiation of Trypanosoma brucei from bloodstream toprocyclic trypomastigotes. In Host-Parasite Cellular andMolecular Interactions in Protozoal Infections (Chang, K.-P. &Snary, D., eds), NATO ASI Series H, vol. 11, pp. 41-49,Springer Verlag, Berlin.

OVERATH, P., CZICHOS, J. & HAAS, C. (1986). The effect of

citrate/cis-aconitate on oxidative metabolism during transformationof Trypanosoma brucei. Eur.J. Biochem. 160, 175-182.

Ross, C. A., CARDOSO DE ALMEIDA, M. L. & TURNER, M. J. (1987).

Variant surface glycoproteins of Trypanosoma congolensebloodstream and metacyclic forms are anchored by a glycolipid tail.Molec biochem. Parasitol. 22, 153-158.

STEIGER, R. F., OPPERDOES, F. R. & BONTEMPS, J. (1980).

Subcellular fractionation of Trypanosoma brucei bloodstream formswith special reference to hydrolases. Eur.J. Biochem. 105,163-175.

TURNER, M. J., CARDOSO DE ALMEIDA, M. L., GURNETT, A. M.,

RAPER, J. & WARD, J. (1985). Biosynthesis, attachment and releaseof variant surface glycoproteins of the African trypanosome. Curr.Top. Microbiol. Immun. 117, 23-55.

(Received 5 December 1988 -Accepted, in revised fonn,16 March 1989)

240 R. Billow et al.