Glycosyl-Phosphatidylinositol-Anchored Membrane...

Journal of the American Society of Nephrology 895

Glycosyl-Phosphatidylinositol-Anchored MembraneProteins1

Dennis Brown2 and Gerald L. Waneck

D. Brown, Renal Unit and Department of Pathology.

Massachusetts General Hospital and Harvard UniversityMedical School. Boston, MA

G. Waneck, Surgical Services and Massachusetts Gen-eral Hospital Cancer Center, Massachusetts GeneralHospital. and Department of Surgery. Harvard Medical

School, Boston, MA

(J. Am. Soc. Nephrol. 1992; 3:895-906)

ABSTRACTMany proteins of eukaryotic cells are anchored tomembranes by covalent linkage to glycosyl-phos-phatidylinositol (GPI). These proteins lack a trans-membrane domain, have no cyfoplasmic tail, andare, therefore, located exclusively on the extracel-lular side of the plasma membrane. GPI-anchoredproteins form a diverse family of molecules that in-cludes membrane-associated enzymes, adhesionmolecules, activation antigens, differentiation mark-ers, protozoan coat components, and other miscel-Ianeous glycoproteins. In the kidney, several GPI-anchored proteins have been identified, includinguromodulin (Tamm-Horsfall glycoprotein), carbonic

anhydrase type IV, alkaline phosphatase, Thy-I, BP-3, aminopeptidase P. and dipeptidylpeptidase. GPI-anchored proteins can be released from membraneswith specific phospholipases and can be recoveredfrom the detergent-insoluble pellet after Triton X-I 14treatment of membranes. All GPI-anchored proteinsare initially synthesized with a transmembrane an-chor, but after translocation across the membraneof the endoplasmic reticulum, the ecto-domain ofthe protein is cleaved and covalently linked to apreformed GPI anchor by a specific transamidaseenzyme. Although it remains obscure why so manyproteins are endowed with a GPI anchor, the pres-ence of a GPI anchor does confer some functional

‘Received March 17. 1992. Accepted April 6, 1992.2Correspondence to Dr. D. Brown. Renal unit. Massachusetts General Hospital

East, 149 13th Street. Charlestown, MA 02129.

I 046-6673/0304-0895$03.00/0Journal of the American Society of NephrologyCopyright C 1992 by the American Society of Nephrology

characteristics to proteins: ( 1) it is a strong apical

targeting signal in polarized epithelial cells; (2) GPI-

anchored proteins do not cluster into clathrin-coatedpits but instead ar� concentrated into specializedlipid domains in the membrane, including so-called

smooth pinocytotic vesicles, or caveoli; (3) GPI-an-chored proteins can act as activation antigens in theimmune system; (4) when the GPI anchor is cleavedby PI-phospholipase C or Pl-phospholipase D, sec-ond messengers for signal transduction may be gen-erated; (5) the GPI anchor can modulate antigenpresentation by major histocompatibility complexmolecules. Finally, at least one human disease, par-oxysmal nocturnal hemoglobinuria, is a result of de-fective GPI anchor addition to plasma membrane

proteins.

Key Words: Cell polarity, kidney. endocytosis. antigen pres-

entation. phosphoilpases

A n extensive body of work over the last few yearshas established that many proteins of eukary-

otic cells are anchored in the plasma membrane, aswell as to the membrane of intracellular organelles.by covalent linkage to glycosyl-phosphatidylinositol

(GPI) (1-3). In contrast to proteins that are anchoredby a conventional membrane-spanning domain thatconsists of a short sequence of principally hydropho-bic amino acids, GPI-anchored proteins have none of

their sequence embedded in the lipid bilayer. Conse-

quently, they lack a cytoplasmic tail that is of consid-erable functional significance to many membrane

proteins, such as hormone and growth factor recep-tors and ion channels. GPI-linked proteins are, there-fore, located exclusively on the extracellular leafletof the plasma membrane and are oriented facing the

lumenal side of intracellular compartments.

GPI-linked proteins are a diverse family of mole-cules: there are no obvious shared functional char-acteristics that provide clues as to why nature has

chosen to use this particular strategy to attach them

to their respective membrane systems. Thus, amongGPI-anchored proteins are membrane-bound en-zymes, adhesion molecules, activation antigens. dif-ferentiation markers, protozoan coat components,and other miscellaneous glycoproteins. Over 100 GPI-anchored proteins have now been identified: a partial

list of these proteins is shown in Table 1.

TABLE I. Partial list of known GPI-anchoredproteins0

Cell Surface HydrolasesAcetylcholinesterase

Alkaline phosphatase

Aminopeptidase P

CA IVDipeptidylpeptidase

Lipoprotein lipase5’ nucleotidasep76 proteinase

TrehalaseOther Mammalian Proteins

Blast ICarcinoembryonic anti-

genCDI4CD48Folate binding proteinLy-6Mo3MRC OX-45MEM 43Qa2 (MHC-l)Thy-I

Thy-3RT-6Uromodulin (Tamm-Horsfall

protein)Zymogen granule GP-2

N-terminal

ule

Dictyostelium discoideumChicken brainRat hepatocytes

Human blood cellsMammalian and chicken

brain, and muscleGuinea pig sperm

Acylation of inositol ring bypalmitic add residue inanchors of some proteins (e.g.AChE) renders them resistantto P1-PLC (but not PLD)

PLD

PLC

Membrane bilayer

GPI-Anchored Membrane Proteins

896 Volume 3’ Number 4’ 1992

Protein Location

Torpedo electric organ, in-sect brain, mammalianblood cells, Schisfosoma

Mammalian tissues, Schis-tosoma, salamander

Pig and human kidneyRat and human kidneyMammalian kidney, sheep

lung3T3-LI adipocytesMammalian tissuesPlasmodium falciparum

merozoitesMammalian tissues

Human lymphocytesHuman tumor cells

Human monocytesHuman leukocytesMammalian tissuesMouse lymphocytesHuman monocytesRat lymphocytesHuman leukocytesMouse lymphocytesMammalian brain and lym-

phocytes, kidney mes-angial cells

Mouse lymphocytesRat lymphocytesMammalian kidney

Pancreatic secretory gran-

O DerIved from data in reviews by Cross (3), Ferguson and Williams (2).

and Low (1), as well as from other literature cited in this review.

The purpose of this review is to provide some fun-

damental background information concerning GPI-anchored proteins and to highlight some specific ex-amples of recent progress that is relevant to the cellbiology of these proteins in epithelia, with specialemphasis on the kidney. A number of more detailedand comprehensive reviews are also available (1-3).

Cell Adhesion MoleculesContact site AFilHeparan sulfate proteo-

glycanLFA-3Neural cell adhesion mol-

eculePH-20

The reader should consult these for additional infor-mation that is beyond the scope of this article.

IDENTIFICATION OF GPI-ANCHORED PROTEINS

The classification of a protein as a member of the

GPI-linked family depends initially upon the libera-tion of the protein from the cell surface by phospha-tidylinositol-specific phospholipase C (P1-PLC) (4).

Treatment of cells with purified P1-PLC results in the

extremely rapid and specific release of P1-linked pro-

teins from the cell surface. This technique was usedto identify two proteins that were among the first tobe recognized as having GPI anchors: alkaline phos-phatase (which is abundant on the brush border ofthe proximal tubule) and acetylcholinesterase. Itshould be noted, however, that failure to release a

protein by P1-PLC treatment does not necessarilyindicate that the protein is of the transmembrane

type. Cell-specific modifications of GPI anchors mayresult in resistance to some (5) or all (6) P1-PLC fromdifferent sources. These P1-PLC-resistant proteins

nevertheless can be labeled with radioactive precur-sors that are incorporated into GPI anchors, on the

basis of common structural features (Figure 1). Inaddition, it has been shown that a phospholipase D(PLD) present in mammalian serum can release GPI-

anchored proteins that are not affected by P1-PLCtreatment (7-9).

The diagnostic process may also be complicated bythe ability of some proteins to be expressed as either

transmembrane or GPI-anchored forms, either as aresult of alternative splicing or as a result of the

CH2- OH - cH2 -

Fatty acids

Figure I. Diagram showing structure of a GPI anchor usedto attach GPI-linked proteins to the membrane. Sites ofcleavage by PLC and PLD are indicated. In, inositol; Eth,ethanolamine; M, mannose; P. phosphate; Gl, glucosa-mine; AChE, acetylcholinesterase.

Brown and Waneck


expression of distinct but related genes. Some poten-

tial implications of this alternative anchoring to thecellular localization and function of these proteinswill be addressed in a later section.

An additional method by which GPI-anchored pro-

teins can be distinguished from proteins with a mem-brane-spanning polypeptide and from soluble pro-

teins relies on differential solubilization and temper-ature-induced phase separation in Triton X-1 14, amethod Introduced by Bordier (1 0) to separate inte-gral membrane proteins from soluble proteins. Whenmicrovillar membranes from kidney were treated

with the nonionic detergent Triton X- 1 1 4 at 0#{176}C,followed by low-speed centrifugatlon, a detergent-Insoluble pellet and a detergent-soluble supernatantwere formed ( 1 1 ). The supernatant was separated

into a detergent rich phase and an aqueous phase byphase separation at 30#{176}C.The GPI-anchored ecto-

enzymes were recovered predominantly in the initialdetergent-insoluble pellet, whereas membrane-span-ning proteins were recovered in the detergent-rich

phase after 30#{176}Ctreatment of the detergent-soluble

supernatant. Removal of the hydrophobic anchoring

domain from both classes of membrane-anchored

proteins resulted in their preferential appearance inthe aqueous phase.

Finally, it is important to keep in mind that al-though the methods detailed above can clearly dem-onstrate the presence of a GPI anchor, negative datapointing to the apparent use of a transmembrane-anchoring domain must always be interpreted with

caution.

STRUCTURE OF THE GPI-ANCHOR

In contrast to the structural and functional heter-ogenelty of GPI-linked proteins, GPI anchors shareseveral conserved features (Figure 1 ). All GPI anchorsthat have been characterized, from protozoans to

mammals ( 1 2- 1 4), have a highly-conserved corestructure consisting of ethanolamine-PO4-Man a 1-2Man al-6 Man al-4 G1cNH2-PI. The COOH terminusof the protein is amide linked to the ethanolamine.

Cell-specific and/or species-specific heterogeneity inGPI anchors have, however, been found in threegeneral areas: (1) the core may contain additional

mannose and phosphoethanolamine, as well as ga-lactose and N-acetylgalactosamine; (2) the fatty acidsmay be alkyl and/or acyl linked and may vary in

carbon chain length and degree of saturation: (3) thehydroxyl on the inositol ring may be substituted withan additional fatty acid. All of the mammalian GPI

anchor precursors that contain mannose have thissubstitution (15), and this may be an important reg-

ulatory point in GPI anchor biosynthesis. The substi-tuted Inositol imparts resistance to all P1-PLC en-

zymes that have been tested, although the anchorcan still be hydrolyzed by PLD (7,15,16). In general,

the functional consequences of this heterogeneity arenot understood.

ANCHOR SYNTHESIS AND ATTACHMENT

GPI-anchored proteins are similar to type I trans-membrane and secreted proteins in that they are

synthesized with an NH2-terminal hydrophobic se-

quence that directs the nascent polypeptide into thelumen of the endoplasmic reticulum and is then

cleaved by signal peptidase. Type I transmembraneproteins also have a COOH-terminal hydrophobic se-quence that stops translocation of the polypeptideand serves to anchor the protein in the membrane.These transmembrane anchors are heterogeneous insequence but have a requirement for minimum

length, hydrophobicity, and the presence of adjacentCOOH-terminal basic amino acids that interact with

acidic headgroups of phospholipids on the cyto-plasmic face of the membrane bilayer (17,18). In theabsence of an anchor sequence, the polypeptide can

be secreted.

The peptide segment that signals GPI anchor at-

tachment resembles the type I membrane anchor inthat it is somewhat hydrophobic and is located at theCOOH-terminal end of the polypeptide. Alignment ofCOOH-terminal regions predicted from the cDNA se-quences of a number of GPI-anchored proteins (Table2) (2,3) has revealed that these segments vary in

length and contain no specific amino acid sequencesuch as that required for RER retention (KDEL) (19),coated pit clustering and internalization (NPXY) (20),and isoprenylation (CaaX) (2 1 ). GPI signals do sharecommon features, however, that give them the ap-pearance of being defective or modified transmem-brane anchors. They may contain polar residues inthe otherwise hydrophobic segment (22): they may

be shorter than the minimum length required for atransmembrane domain: and/or they usually containno basic residues in the predicted cytoplasmic tail(2,3).

Another requirement for GPI modification of a pro-tein is that the signal be cleaved before attachment

of the GPI anchor (23,24). This cleavage occurs dur-ing or immediately after translation of the polypep-tide and appears to be coupled to the covalent linkage

of the new COOH terminus to ethanolamine, in atransamidation reaction. The new residue at theCOOH terminus has been designated the w site, andamino acids adjacent and COOH-terminal to the w

site have been designated w + 1 and w + 2, etc. (24).Molecular analysis with site-directed mutagenesishas demonstrated a requirement for specific aminoacids at the w and w + 2 sites: in general, residueswith hydrophobic or bulky side chains are not per-

mitted (23,24). These experimental analyses con-form, for the most part, with the observed residuesoccurring naturally at these positions (2,3). This w, w


898 Volume 3 ‘ Number 4’ 1992

TABLE 2. Sequences of some mammalian proteins encompassing known GPI anchor attachment sites#{176}

Protein COOH Terminal Sequence Not Present in Mature Protein

Alkaline phosphatase PPAGTTD AAHPGRSVVPALLPLLAGTLLLLETATPHuman THY-I RDKLVKC EGISLLAQNTSWLLLLLLSLSLLQATDFMSL5’ Nucleotidase EGRIKFS AASHYQGSFPLIILSFWAVILVLYQ

Folate-binding protein NPNEEVA RFYAAAMSGAGPWAAWPFLLSLALMLLWLLS

DAF KGSGTTS GTTRLLSGHTCFTLTGLLGTLVTMGLLTAChE (Torpedo) LLNATAC DGELSSSGTSSSKGIIFYVLFSILYLIFYCEA KSITVSA SGTSPGLSAGATVGIMIGVLVGVALI

#{176}DAF.decay accelerating factor; AChE. acetylcholinesterase; CEA. carcinoembryonic antigen. Data are compiled from reviews by Cross (3) and

Ferguson and Williams (2).

+ 2 rule is similar to the rule predicting cleavage by

NH2-terminal signal peptidase (25). It appears, there-fore, that GPI attachment signals and leader se-quences represent the cleavable counterparts of typeI and type II transmembrane protein anchors, respec-

tively.

Experiments carried out with mutant cell lines de-ficient in the expression of GPI-anchored proteins

(26,27) and with systems that permit cell-free proc-essing (28-3 1 ) have demonstrated that the completeGPI anchor must be synthesized en bloc and be avail-able as a substrate in order for the transamidationreaction (and thus protein cleavage) to occur. In the

presence of surrogate anchors that occur endoge-nously in some mutant cell lines (32) or that are added

exogenously as chemical inhibitors of anchor attach-ment, such as mannosamine (33) or fluoroglucose

(34), the protein signal is cleaved but is not attachedto a glycolipid anchor. In these cases, the protein issecreted. The biochemical pathways and enzymaticsteps involved in this process are, therefore, complexand are defined by a number of genetic complemen-tation groups determined by somatic hybridization ofmutant cells and by analysis of GPI anchor precur-sors (35,36). One of the steps in GPI anchor synthesisinvolves dolichol-phosphate-mannose (Dol-P-Man), adonor for mannose in the GPI anchor core, and aprecursor for the synthesis of N-linked high mannosecarbohydrate side chains of glycoproteins. Indeed,

GPI-anchored protein expression was restored bytransfection of a cell line defective in Dol-P-Man

synthesis with the yeast gene for Dol-P-Man syn-

thase (37).

Defective GPI Anchor Biosynthesis in ParoxysmalNocturnal Hemoglobinuria

Defects in GPI anchor biosynthesis have been im-plicated in the pathogenesis of at least one human

disease, paroxysmal nocturnal hemoglobinuria (38).In particular, defective expression of a GPI-anchored

protein, decay accelerating factor, on the surface of

erythrocytes appears to render them more suscepti-

ble to complement-induced lysis (39). Thus, a betterunderstanding of the pathways involved in GPI an-

chor biosynthesis, together with the molecular don-ing of enzymes that appear to be defective in parox-ysmal nocturnal hemoglobinuria patients (40), may

ultimately permit gene therapy to correct this hema-

topoietic stem cell disorder.

CELL BIOLOGY OF GLYCOPHOSPHOLIPID

ANCHORS

Apical Targeting of GPI-Anchored Proteins

By comparing the list of GPI-anchored proteinswith their membrane localization pattern, it was dis-

covered that these proteins appear to be targettedexclusively to the apical plasma membrane domainin polarized epithelial cells. Lisanti et at. (41,42)

showed that several endogenous GPI-anchored pro-

teins are restricted to the apical surface of a varietyof cultured epithelial cells, including MDCK, LLC-PK1 , Caco-2, and SK-CO15 cells, and conversion to aGPI-anchored form of a viral glycoprotein (herpessimplex glycoprotein gD- 1 ) that is normally targettedto the basolateral membrane resulted in its delivery

to the apical plasma membrane of MDCK cells (43).The reciprocal experiments have also been performedand show that when GPI-anchored proteins such asalkaline phosphatase are given a transmembraneanchor, they can be inserted into the basolateralplasma membrane (44). Work from our laboratorywith transfected LLC-PK1 epithelial cells (see below)has shown that the polarized targeting of GPI-an-chored and transmembrane-anchored major histo-

compatibility complex class I (MHC I) molecules withthe same ecto-domain also depends upon the natureof the membrane anchor.

Ithas been proposed, therefore, that the GPI anchoris a strong targeting signal that directs proteins tothe apical plasma membrane in epithelial cells. How-ever, because a variety of proteins with a conven-tional transmembrane anchor are also delivered to

the apical surface, the GPI anchor appears to be

Brown and Waneck


sufficient, but not necessary, to ensure the apicallocalization of a protein.

Sorting Signals in GPI-Anchored Proteins

A typical transmembrane protein can be subdi-vided into three distinct regions that potentially reg-

ulate its trafficking: the extracellular domains,

which may interact with other lumenal proteins, in-cluding specific receptors: the transmembrane an-

chor, which may interact with other transmembranesegments without involvement of other domains: andthe cytoplasmic tail, which may interact with cyto-

skeletal elements including microfilaments and mi-crotubules, adaptor molecules in clathrin-coated pits,and molecules involved in signal transduction prod-esses. Secretory proteins lack transmembrane seg-ments and cytoplasmic tails, so that sorting of theseproteins depends primarily on their interaction with

other lumenal proteins. GPI-anchored proteins re-semble secretory proteins in that they also lack trans-

membrane and cytoplasmic domains. The extracel-

lular domain of these molecules may, therefore, be amajor factor in determining their membrane target-ing properties. This is supported by experiments inwhich a truncated form of Thy- 1 (a GPI-anchoredprotein identified first on the plasma membrane of

murine thymocytes), lacking a COOH-terminal se-quence that is required for attachment of the GPIanchor, was constructed and expressed in MDCKcells. This truncated molecule was secreted at the

apical pole of these cells, indicating that the proteinecto-domain contains apical targeting information

(45). On the other hand, GPI-anchored proteins arelinked to the external leaflet of the lipid bilayer, a

feature that distinguishes them from secreted pro-teins. Some evidence now suggests that the GPI an-chor may cause them to be segregated into specificregions of the membrane bilayer that are rich incholesterol (46) and that they may be subject to thesame factors that regulate the flow of specific lipidsto the apical plasma membrane in epithelial cells(47). This Idea is supported by data showing that GPI-anchored proteins can diffuse laterally in the planeof the plasma membrane at a rate that is in some

cases comparable to that of membrane lipids, al-though the diffusion coefficients for individual pro-teins are highly variable (48). Thus, the presence ofa GPI anchor does not intrinsically confer increasedlateral mobility, and interaction with other mem-brane proteins may occur that restricts movement inthe plane of the lipid bilayer.

GPI-Anchored Proteins Are Excluded FromClathrin-Coated Pits

Studies in which the subcellular and plasma mem-brane distribution of GPI-anchored proteins have

been examined have provided some clues concerningthe potential effects of having a GPI anchor, ratherthan a conventional transmembrane anchor. In par-

ticular, those GPI-anchored proteins that have so far

been examined seem to be excluded from clathrin-coated pits at the cell surface. This ability of clathrin-

coated pits to sort and concentrate specific mem-brane proteins destined for endocytosis was first

shown by Bretscher et at. (49), who described coated

pits as “molecular filters” on the cell surface. Theprotein that they examined that was specifically ex-

cluded from these membrane microdomains wasThy- 1 , which is now known to be GPI anchored. At

the time their work was published, this fact had notbeen established.

The exclusion of GPI-linked proteins from clathrin-

coated pits can be readily demonstrated in proximaltubule epithelial cells of the kidney. Figure 2A showsthat the GPI-linked, membrane-associated form of

carbonic anhydrase (CA IV) in rat kidney proximaltubules (50,51) is localized on the brush border mi-

crovilli but is not concentrated in clathrin-coatedmembrane domains at the base of the microvilli. Incontrast, gp330, an apical membrane glycoprotein

with a transmembrane peptide anchor, is concen-trated in the clathrin-coated regions. as well as beingdiffusely present on the microvilli (52-54).

It has been proposed, based in part on this failure

to cluster into clathrin-coated pits, that proteins mayuse a GPI anchor as a means to prolong their half-life on the cell surface (55). This is supported by theobservation that a glycolipid-anchored fusion protein

expressed in MDCK cells shows a slow rate of endo-cytosis from the apical membrane (43). Because

many receptors and other proteins are rapidly recy-cled between intracellular compartments and the cell

surface by a mechanism involving clathrln-coatedpits for the endocytic step, proteins that avoid being

entrapped in coated pits would escape a major meansof removal from the cell surface. This may be ofparticular Importance in cell types that have a highlyendocytotic surface, such as the proximal tubule ep-ithelial cell.

The Folate Receptor is GPI Anchored-theRole of Potocytosis in Ligand Uptake

It has been demonstrated that the receptor for fo-late is GPI linked to membranes (56), yet this receptor

clearly has a functional need to internalize its ligand.How is this consistent with the presence of a mem-brane anchor that excludes the receptor from coated

pits? Studies by Anderson and his colleagues haveproposed that potocytosis, a novel internalization

mechanism for small ligands, is involved in the up-take of folate into cells (57).

GPI-Anchored Membrane Proteins � � �


It has long been known that smooth invaginations,or caveoli, are present on the plasma membranes of

many cell types. The role of these structures In en-docytosis is controversial, because it has never beenconvincingly demonstrated that the vesicles actually

detach from the membrane during physiologicallyrelevant conditions (58). Despite this, they appear to

be involved in the uptake of several ligands. includingsome bacterial toxins such as cholera toxin. Labelingstudies with cholera toxin coupled to colloidal goldparticles have clearly shown that cholera toxin local-izes preferentially to smooth invaginations on thecell surface and is excluded from clathrin-coated pits(59,60). Similarly, GPI-linked folate receptors seem

to be clustered into smooth membrane invaginatlons(6 1 ). These regions are cholesterol rich, as shown bya heavy labeling with the sterol-specific antibiotic,filipin, and the cell surface distribution of folate

receptors can be greatly modified by agents that dis-rupt the pattern of membrane cholesterol distribu-

tion in cultured cells (46).Because cholera toxin receptor is a ganglioside. a

glycolipid. and because the folate receptor that also

localizes in smooth caveoli is GPI anchored, these

structures may have a specific role to play in theendocytosis of llgands that are attached to receptorslacking a transmembrane domain. It is now pro-posed, but not definitively proven (62), that the

smooth caveoli Invaginate into the cell but never

physically detach from the surface plasma mem-

brane. Instead, a tiny neck portion of the caveolusremains attached but forms a seal between the ex-ternal milieu and the lumen of the Invaginatingstructure. This allows proton pumps on the mem-brane of the caveolus to acidify its lumen and causereceptor-ligand dissociation to occur, just as in con-ventional endosomes (63). The dissociated ligand caneither diffuse or be transported across the membraneof the vesicle Into the cell cytoplasm: the neck of thecaveolus then opens up to accept more ligand at thecell surface.

Figure 2. Ultrathin cryostat sections of rat kidney cortex,fixed in 3% paraformaldehyde-O. 1% glutaraldehyde. CAIV antigenic sites were localized by the protein A-goldtechnique. Panel A shows that the brush border is exten-sively labeled with gold particles, which are localized onthe external surface of the plasma membrane (small ar-rows). The clathrin-coated apical invaginations (arrow-heads) and subapical vesicles are virtually unlabeled, how-ever, indicating that this GPI-anchored protein does notconcentrate in clathrin-coated domains that are extensivein this segment of the urinary tubule. A small amount oflabel is found on the lateral plasma membrane (largearrows). The labeling intensity of the basal plasma mem-brane is greater than that of the lateral membrane. Bar, 0.5�m.

Caveolin-a Molecular Component of theCoat of Caveoli

As detailed above, caveoli have been identified asa compartment that is involved in the internalizationof folate and these structures also internalize otherligands that bind to lipidic receptors. Previously re-ferred to as “smooth” or “uncoated” invaginations onthe cell surface, It has now been established that

these structures have a characteristic striated coat-ing material on their cytoplasmlc surface, as revealed

by rapid-freeze, deep-etch electron microscopy (62).One of the components of this coating material is

recognized and labeled by antIbodIes raised against

Brown and Waneck


a 22-kd substrate for v-src tyrosine kinase in Rous

sarcoma virus-transformed chick embryo fibro-blasts. Immunogold labeling shows that this antibodybinds to the filaments that form the coat. The proteinpresent in the filaments has been named caveolin

(62). Whether caveolin interacts with the src-related

tyrosine klnase pS6”�, which is complexed to GPI-anchored molecules (64), remains to be determined.

Caveolin is, therefore, thefourth distinctive coat-ing material that has been associated with different

veslcles involved in transport processes in cells. Theother three types are clathrln-coated pits and vesicles

(65), nonclathrln-coated transporting vesicles thatare intermediates In intra-Golgl transport (66), andH�-ATPase-coated vesicles that are involved in therapid recycling of proton pumps to and from the

plasma membrane of kidney intercalated cells andrelated proton-secreting cells in other epithelia (67).

Although these other forms of vesicles are involvedin intracellular trafficking of proteins with a trans-membrane anchor, caveoli seem to be specializedstructures that concentrate proteins with glycolipidanchors. In other types of vesicles, the cytoplasmicdomain of the transported protein is accessible fordirect or Indirect interaction with the coat material,whereas this cannot be the case for receptors or otherGPI-anchored molecules that are located in caveoll.The means by which these proteins cluster in caveoli

is unknown, although Rothberg et at. have proposeda mechanism that is dependent on the specific lipidcontent of these membrane domains (46).

GPI-Anchored Proteins Are Found in Glycolipid-Enriched Membrane Domains

Evidence in favor of the hypothesis that GPI-an-chored proteins reside in specialized lipid-rich mem-

brane domains has been obtained from studies by D.A. Brown and J. K. Rose (68). These workers haveshown that proteins with a GPI anchor, includingplacental alkaline phosphatase, can be recovered

from lysates of MDCK cells in a low-density, deter-gent-Insoluble form. In addition to containing GPI-anchored proteins, these extracted membrane do-mains are enriched in glycosphingolipids but are de-pleted of basolateral protein markers. This work sup-

ports that of Simons and Van Meer (47), who hadpreviously proposed that apical membrane proteinsmay be sorted together with glycosphingolipids into

specialized membrane patches at the level of the

Golgl apparatus. Whether there is a segregation ofGPI-anchored and transmembrane-anchored pro-teIns Into different vesicles during transport to theapical pole of eplthellal cells remains to be estab-

lished.

FUNCTIONAL CONSEQUENCES OF GPIANCHORAGE

Table 1 shows that GPI-anchored proteins sub-

serve a variety of cellular functions. However, thereason why these proteins are GPI anchored, ratherthan being transmembrane anchored, remains ob-scure. There are certain features that the GPI anchor

imparts on a protein that distinguish them from othercell surface proteins, but so far, there are only alimited number of examples relating the importanceof a GPI anchor to the specific biologic activity of aprotein. Most of these are related to roles in the

immune system, and the following section gives ex-amples of how GPI anchors can affect T-cell activa-tion and antigen presentation.

GPI-Anchored Proteins and 1-Cell Activation

A common property of many GPI-anchored proteins

expressed on resting T lymphocytes is their ability to

activate cell proliferation when they bind monoclonalantibodies that are subsequently cross-linked withsecondary antibodies, in the presence of phorbol es-

ters (69-7 1 ). Indeed, the classification of a numberof GPI-anchored proteins as “activation antigens” isbased on this empirical property. Because the only

apparent similarity between these activation pro-

teins is their GPI anchor, it has been proposed thatany protein attached to the membrane via a GPI

anchor would become an activation antigen. Thishypothesis is supported by experiments in transgenlc

mice expressing either native GPI-anchored Qa-2 ad-tivation protein, native transmembrane H-2D’� trans-

plantation antigen, or reciprocal chimeric moleculesthat had their anchors exchanged by genetic engi-neering (72). Qa-2 and H-2D’� are both murine MHC-1 glycoproteins and are more than 85% identical intheir extracellular domains (73). As predicted, GPI-anchored Qa-2 and H�2Db molecules served as acti-vation proteins for resting T lymphocytes, whereas

their transmembrane counterparts did not (72). In-

dependent experiments in a T-cell helper cloneshowed that the GPI-anchored Ly-6 activation pro-tein ceased to function when linked to the H�2Dbtransmembrane anchor (74).

Conflicting results have been obtained with theYH16.33 T-cell hybridoma line transfected with GPI-anchored and transmembrane H�2Db constructs thatare identical to the ones described above (Chang,Rollins, Yeh, and Waneck, unpublished data).

YH16.33 cells proliferate constitutively but will se-

crete interleukin 2 when activated by monoclonalantibodies to the endogenous GPI-anchored proteins

Thy-i and Ly-6. However, In the same cells, antIbod-ies to the transfected GPI-anchored H-2D’� construct

failed to induce InterleukIn 2 secretion. Thus, it ap-

GPI-Anchored Membrane Proteins ..�<,- � � � �

902 Volume 3 ‘ Number 4’ 1992

pears that some, if not all, GPI-anchored proteinsmay require natural ligands in order to execute cer-tam functions, and having a GPI anchor may be

necessary but not sufficient for these effects.

Role of the GPI Anchor on MHC-l Function inAntigen Presentation

Transmembrane MHC-I molecules function asscavenger receptors that bind a variety of peptidesfrom degraded intracellular proteins and transportthem to the cell surface for surveillance by T lym-phocytes (75). The bound peptide is essential forproper folding and transport of the MHC molecule. If

the peptide is of self-origin, the cell is tolerated bythe immune system; if the peptide is foreign (e.g.,

derived from a viral protein or a cancer antigen) or ifthe MHC-I is foreign (i.e. . a mismatched tissue trans-plant), the cell Is lysed. We Investigated the effects oftransmembrane versus GPI anchors on MHC-I func-

tion in thymoma cells transfected with transmem-brane or GPI-linked H�2Db (Flyer, Brown, Flavell, and

Waneck, submitted for publication). Cells expressingthe transmembrane form functioned normally inmediating lysis by T cells educated to recognize H-

2D’� as a foreign MHC-I or as H�2Db presenting murineleukemia retrovirus antigen on infected cells. GPI-anchored H-2D” could also be recognized normally as

a foreign MHC-I In transfected cells, but in contrast,these cells were defective in presenting leukemiavirus antigen. Because the extracellular domains of

the transmembrane- and GPI-anchored MHC mole-cules are Identical, these differences in ability to

present a retrovirus antigen may correlate with dif-ferences in intracellular compartmentallzatlon andtrafficking between the two forms of the MHC mole-cule that affect pathways for antigen processing and

recognition.

Many diseases either result from or involve a fail-

ure of epithelial cells to maintain a polarized pheno-type. As discussed above, GPI-anchored proteins arepreferentially located on the apical membrane do-main of epithelial cells, whereas transmembraneMHC molecules are normally expressed on the baso-lateral plasma membrane, where retroviruses also

assemble and bud. We have transfected LLC-PK1cells and have demonstrated that, as expected, GPI-

anchored H-2D’� molecules are targeted to the apicalpole of these cells, whereas their transmembrane-linked counterparts are targeted basolaterally (Wa-neck, Rollins, and Brown, unpublished results).Taken together, these studies suggest that abnor-malities or modifications in the pathways by which

MHC-I and processed antigens converge and diverge

can lead to an altered capacity of cells to presentantigen and, consequently, to inapproprIate T-cell

recognition. These abnormalities may play an impor-tant role in the pathogenesls of diseases such as

autoimmune interstitial nephritis (76) and type I di-abetes mellitus (77).

GPI-ANCHORED PROTEINS IN THE KIDNEY

Uromodulin (Tamm-Horsfall Glycoprotein-Uromucoid)

Tamm-Horsfall glycoprotein (uromodulin) is syn-

thesized in the kidney and appears to be restricted inits distribution to distal convoluted tubules and thickascending limbs of Henle. It is found in large amountsin the urine, and although Its function is unknown,it is identical to uromodulin, an immunosuppressiveglycoprotein originally described in the urine of preg-

nant women. Because of its abundance in urine, itwas believed to be a typical secretory protein but it is

now known that uromodulin can be anchored to theplasma membrane by a GPI anchor, as shown byexpressing cDNA encoding this protein in HeLa,

Caco-2, MDCK, COS, and AT-20 cells (78). The ap-pearance of soluble uromodulin in the urine might

then be the result of the action of endogenous phos-pholipases in the kidney or the result of proteaseaction. However, by Immunocytochemistry, uromo-dulin can also be detected on the basolateral plasmamembranes of thick ascending limb cells (79). It is

unknown whether this basolateral form of the pro-tein is also GPI anchored or whether an alternativesplicing event produces a conventionally membrane-linked form in thick ascending limb cells.

Membrane-Associated Carbonic AnhydraseType IV

The existence of a membrane-associated form ofcarbonic anhydrase in some nephron segments, es-pecially the proximal convoluted tubule, has beenrecognized for many years (80). Histochemical stud-

ies that cannot distinguish between different iso-forms of the enzyme revealed dense precipitates ofreaction product in close association with the plasma

membrane, indicating the presence of a membrane-associated enzyme in proximal tubules and thickascending limbs of rat kidney, and physiologic stud-ies also indicated the presence of a carbonic anhy-drase in contact with the lumenal fluid.

Recent immunocytochemical data with specific an-

tibodies against CA IV have shown that it is concen-trated In the S2 segment of the proximal convolutedtubule and is abundant in the thick ascending limbs

of Henle (50). Although the immunoreactivity isstrongest in the apical plasma membrane In both

tubule segments, there is also a significant reactivityof the antibodies with the basolateral membranes(Figure 2). In human kidney, it has been shown that

a large portion of CA IV exists as a GPI-anchoredprotein, because the enzyme can be released from

Brown and Waneck


kidney membranes by P1-specific PLC (5 1). Thesemembranes presumably derive from apical mem-

branes of the urinary tubule eplthellal cells. How-

ever, a transmembrane-linked form of the enzyme isalso detectable in human kidney, Indicating that an

alternative splicing event may occur. Thus, the in-trigulng possibility exists that the apical or basolat-

eral targeting of kidney CA IV may be directed by amodification of the anchor used to attach this protein

to the plasma membrane.Immunocytochemical studies at the electron micro-

scope level, by the Immunogold procedure. haveshown that the apical CA IV in the proximal tubuleis associated only with the microvilli of the brush

border and is not located in the clathrln-coated region

at the base of the microvilli (Figure 2A). This isconsistent with the selective exclusion of GPI-linked

proteins from clathrin-coated pits outlined earlierand presumably results in an increased half-life ofCA IV on the cell surface.

Alkaline Phosphatase

Alkaline phosphatase was one of the first ecto-enzymes identIfied as having a GPI linkage. It isabundant on the apical brush border microvllli ofproximal tubules but is also found on apical mem-

branes of other cells in the urinary tubule, includingcollecting duct epithelial cells (8 1). The physiologicrelevance of this protein is unclear.

Aminopeptidases

Two aminopeptidases of the proximal tubule brushborder are known to be GPI anchored; aminopepti-dase P (82) and renal dipeptidase (83). In addition toits role In the luminal degradation of filtered peptides,dipeptidyl peptidase IV has also been implicated in

some types of autoimmune proximal tubule damageand it can be found in immune deposits and debrisin the lumen of proximal tubules from animals withHeymann nephritis.

5’ Nucleotidase

This is another GPI-anchored ecto-enzyme of the

brush border of the proximal tubule as well as othertubule segments. It is one of a number of ecto-nu-

cleotidases found on the plasma membrane of manyepithelial cell types and may work in concert withbrush border ecto-ATPase (85), whIch has a conven-tional transmembrane anchor, to hydrolyze various

adenine nucleotides to adenosine and Inorganicphosphate.

BP-3 Antigen

BP-3 antigen Is a glycosylated cell surface proteinfound originally on lymphoid and myeloid cells. Re-

cent immunocytochemical studies have revealed this

antigen In other cell types, including collecting ductepithelial cells (86). The function of this protein on

these cells remains unknown, but its presence mdi-cates that these cells are also capable of expressingsurface proteins with a GPI anchor.

Thy-I

Originally described on the plasma membrane ofmurlne thymocytes, this glycoprotein is present onother cell types, including glomerular mesangial cellsand neurons. Although its role is unknown, anti-

Thy- 1 antibodIes Induce mesangial cell damage in

the rat and result in a mesangial proliferative ne-

phritis (87). In these diseased animals, as well as in

cultured mesangial cells, antibody interaction withthe mesangial cells results in increased extracellularmatrix production (88) and glomerular accumulation

of macrophages and platelets. Whether the GPI link-age of the immune target protein Is of significance to

this response is not known. Other autoimmunemodels of glomerulonephrltis, such as Heymann ne-phrltis, develop as a result of antibody interactionwith conventionally anchored protein components of

the glomei-ulus (53,89).Although the list of proteins having a GPI anchor

is expanding rapidly, the most important questionconcerning this group of membrane proteins remainsunanswered, namely, why do they have a GPI anchorand not a conventional membrane anchor? As dis-

cussed above, there are several properties that the

GPI anchor confers on proteins, including localiza-tion to apical membranes of epithellal cells, failure

to cluster in clathrin-coated pits, concentration inspecialized lipid domains of membranes, ability toact as activation antigens, etc. ; however, most ofthese characteristics are shared by at least someproteins that have conventional transmembrane an-

chors. Another theoretical role of GPI-anchored pro-teins Is related to the release of a diacylglycerolmoiety upon P1-PLC cleavage of the anchor or of

phosphatidic acid upon cleavage by PI-PLD. Such aneffect remains to be demonstrated, however. The

question of why so many cell surface proteins havea GPI anchor may eventually be resolved by contin-uing to examine the function of hybrid molecules inwhich conventional and GPI anchors have been ex-changed, by examining the effects of mutations and

treatments that inhibit anchor assembly, and bysearching for pathologic conditions that arise fromdefective GPI anchor attachment.

REFERENCES

1. Low M: The glycosyl-phosphatidylmnositol an-chor of membrane proteins. Biochim BlophysActa 1989:988:427-454.

2. Ferguson MA, Williams AF: Cell-surface an-

GPI-Anchored Membrane Proteins � . . .�*. . , �,. , .� ..�.. ... .. �.


choring of proteins via glycosyl-phosphatidylmno-sitol structures. Annu Rev Biochem i 988:57:285-320.

3. Cross GA: Glycolipid anchoring of plasma mem-brane proteins. Annu Rev Cell Biol 1990:6:1-39.

4. Low M: Biochemistry of the glycosyl-phosphati-dylinositol membrane protein anchors. BiochemJ 1987:244:1-13.

5. Low MG, Stiernberg J, Waneck GL, Flavell RA,Kincaide PW: Cell specific heterogeneity in sen-sitIvity of phosphatidylinositol-anchored mem-brane antigens to release by phospholipase C. JImmunol Methods 1988:113:101-111.

6. Roberts WL, Myher JJ, Kuksis A, Low MG,Rosenberry TL: Lipid analysis of the glycoinosi-tol phospholipid membrane anchor of humanerythrocyte acetylcholmnesterase. Palmitoylationof inositol results in resistance to phosphatidyl-inositol-specific phospholipase C. J Biol Chem1988:263:18766-18775.

7. Toutant JP, Roberts WL, Murray NR, Rosen-berry TL: Conversion of human erythrocyte ac-etylcholinesterase from an amphiphilic to a hy-drophilic form by phosphatidylmnositol-specificphospholipase C and serum phospholipase D.Eur J Biochem 1 989; 180:503-508.

8. Low MG, Huang KS: Factors affecting the abilityof glycosylphosphatidylinositol-specific phos-pholipase D to degrade the membrane anchorsof cell surface proteins. BIochem J 1991:279:483-493.

9. Davitz MA, Horn J, Schenkrnan 5: Purificationof a glycosyl-phosphatidylinositol-specific phos-pholipase D from human plasma. J Biol Chem1989:264:13760-13764.

1 0. Bordier C: Phase separation of integral mem-brane proteins in Triton X-1 14. J Biol Chem1981:256:1604-1607.

1 1 . Hooper NM, Bashir A: Glycosyl-phosphatidymno-sitol-anchored membrane proteins can be distin-guished from transmembrane polypeptide-an-chored proteins by differential solubilization andtemperature-Induced phase separation in TritonX-114. BiochemJ 1991:280:745-751.

12. Ferguson MAJ, Homans SW, Dwek PA, Rade-macher TW: Glycosyl-phosphatidylinositolmoiety that anchors Trypanosoma brucei var-iant surface glycoprotemn to the membrane. Sci-ence 1988:239:753-759.

1 3. Homans SW, Ferguson MAJ, Dwek PA, et at.:Complete structure of the glycosyl phosphatidyl-inositol membrane anchor of rat brain Thy-iglycoprotemn. Nature (Lond) 1 988;333:269-272.

14. Roberts WL, Santikarn S, Reinhold VN, Rosen-berry TL: Structural characterization of the gly-coinositol phospholipid membrane anchor of hu-man erythrocyte acetylcholmnesterase by fastatom bombardment mass spectrometry. J BiolChem 1 988;263: 18776-18784.

15. Urakaze M, Kamitani T, DeGasperi R, et at:Identification of a missing link in glycosylphos-phatidylmnositol anchor biosynthesis in mam-malian cells. J Blol Chem 1992:267:6459-6462.

16. Walter El, Roberts WL, Rosenberry TL, RatnoffWD, Medof ME: Structural basis for variationsin the sensitivity of human decay acceleratingfactor to phosphatidylmnositol-specific phospho-lipase C cleavage [published erratum appears inJ Immunol 1990:144:4072J. J Immunol 1990:

144: 1030-1036.1 7. Wickner WT, Lodish HF: Multiple mechanisms

of protein insertion into and across membranes.Science 1985:230:400-407.

1 8. Boyd D, Beckwith J: The role of charged aminoacids in the localization of secreted and mem-brane proteins. Cell 1990:62:1031-1033.

1 9. Rose JK, Dorns RW: Regulation of protein exportfrom the endoplasmic reticulum. Annu Rev CellBiol 1988:4:257-288.

20. Chen WJ, Goldstein JL, Brown MS: NPXY, asequence often found in cytoplasmic tails, isrequired for coated pit-mediated internalizationof the low density lipoprotein receptor. J BiolChem 1990:265:3116-3123.

2 1 . Lowy DR, Willurnsen BM: New clue to ras lipidglue. Nature (Land) 1989:341:384-385.

22. Waneck GL, Stein ME, Flavell HA: Conversionof a P1-anchored protein to an mntegeral mem-brane protein by a single amino acid mutation.Science 1988:241:697-699.

23. Moran P, Raab H, Kohr WJ, Caras 1W: Glyco-phospholipid membrane anchor attachment.Molecular analysis of the cleavage/attachmentsite. J Biol Chem 1991:266:1250-1257.

24. Gerber LD, Kodukula K, Udenfriend S: Phos-phatidylmnositol-glycan (PI-G) anchored mem-brane proteins: Amino acid requirements adja-cent to the site of cleavage and PI-G attachmentin the COOH-terminal signal peptide. J BiolChem 1992:267:12168-12173.

25. von Heijne G: The signal peptide. J Membr Biol1990:115:195-201.

26. Sugiyarna E, DeGasperi R, Urakaze M, et at.:Identification of defects in glycosylphosphatidyl-inositol anchor biosynthesis in the Thy-iexpression mutants. J Biol Chem 1991:266:12119-12122.

27. Thomas 14, DeGasperi R, Sugiyama E, et at.:Functional analysis of T-cell mutants defectiveIn the biosynthesis of glycosylphosphatldylmnos-itol anchor. Relative importance of glycosyiphos-phatidylinositol anchor versus N-linked glyco-sylation in T-cell activation. J Biol Chem 1991:266:23175-23184.

28. Hirose S, Ravi L, Hazra SV, Medof ME: Assem-bly and deacetylatlon of N-acetylglucosysami-nyl-plasmanylinositol in normal and affectedparoxysmal nocturnal hemoglobmnuria cells.Proc Nati Acad Sd USA 1991:88:3762-3766.

29. Mayor S. Menon AK, Cross GA: Transfer ofglycosyl-phosphatidylmnositol membrane an-chors to polypeptide acceptors in a cell-free sys-tern. J Cell Biol 1991:114:61-71.

30. Stevens VL, Raetz CR: Defective glycosyl phos-phatidylinositol biosynthesis in extracts of threeThy- 1 negatIve lymphoma cell mutants. J BiolChem 1991;266:i0039-i0042.

31. Kodukula K, Amthauer R, Clines D, et at.: Bio-synthesis of phosphatldylmnositol-glycan (PI-G)-anchored membrane proteins in cell-free sys-tems: PI-G Is an obligatory co-substrate forCOOH-terminal processing of nascent proteins.Proc Natl Acad Sci USA 1992:89:4982-4985.

32. Fatemi SH, Tartakoff AM: Hydrophilic anchor-deficient Thy-i is secreted by a class E mutantT lymphoma. Cell 1986:46:653-657.

33. Lisanti MP, Field MC, Caras 1W, Menon AK,Rodriguez-Boulan E: Mannosamine, a novelinhibitor of glycosyiphosphatidylinositol incor-

Brown and Waneck


poration into proteins. EMBO J 1991:10:1969-1977.

34. Takami N, Oda K, Ikehara Y: Aberrant process-ing of alkaline phosphatase precursor caused byblocking the synthesis of glycosylphosphatidyl-inositol. J Biol Chem 1992:267:1042-1047.

35. Lemansky P. Gupta DK, Meyale S. Tucker G,Tartakoff AM: Atypical mannolipids character-ize Thy-i-negative lymphoma mutants. Mol CellBiol 1991:11:3879-3885.

36. Singh N, Singleton D, Tartakoff AM: Anchoringand degradation of glycolipid-anchored mem-brane proteins by L929 versus by LM-TK- mousefibroblasts: Implications for anchor biosyn-thesis. Mol Cell Biol 1991:11:2362-2374.

37. DeGasperi R, Thomas U, Sugiyama E, et at.:Correction of a defect in mammalian GPI anchorbiosynthesis by a transfected yeast gene. Sd-ence 1990:250:988-991.

38. Stafford HA, Tykocinski ML, Lublin DM, et at.:Normal polymorphic variations and transcrip-tion of the decay accelerating factor gene in par-oxysrnal nocturnal hemogl#{243}bmnuria cells . ProcNatI Acad Sd USA 1988:85:880-884.

39. Rosse WF: Paroxysmal nocturnal hemoglobmnu-na and decay accelerating factor. Annu Rev Med1990:41:431-436.

40. Thomas U, Urakaze M, DeGasperi R, et at.:Differential expression of glycophosphatidylmno-sitol-anchored proteins In a murine T cell hy-bridorna mutant producing limiting amounts ofthe glycolipid core. Implications for paroxysmalnocturnal hemoglobinuria. J Clin Invest 1992:89:1 172-i 177.

4 1 . Lisanti MP, Sargiacomo M, Graeve L, SaltielAR, Rodriguez-Boulan E: Polarized apical dis-tribution of glycosyl-phosphatidylmnositol-an-chored proteins In a renal epithelial cell line.Proc Natl Acad Sd USA 1988:85:9557-9561.

42. Lisanti MP, Le-Bivec A, Saltiel AR, Rodriguez-Boulan E: Preferred apical distribution of gly-cosyl-phosphatidylmnositol (GPI) anchored pro-teins: a highly conserved feature of the polarizedepithelial cell phenotype. J Membr Biol 1990;1 13:i55-i67.

43. Lisanti MP, Caras 1W, Gilbert T, Hanzel D, Rod-riguez-Boulan E: Vectorial apical delivery andslow endocytosis of a glycolipid-anchored fusionprotein in transfected MDCK cells. Proc NatlAcad Sd USA 1990:87:7419-7423.

44. Brown DA, Crise B, Rose JK: Mechanism ofmembrane anchoring affects polarized expres-sion of two proteins In MDCK cells. Science1989:245: i499-i501.

45. Powell SK, Lisanti MP, Rodriguez-Boulan EJ:Thy-i expresses two signals for apical localiza-tion in epithelial cells. Am J Physiol 1991:260:C7 1 5-C720.

46. Rothberg KG, Ying Y-S, Kamen BA, AndersonRGW: Cholesterol controls the clustering of theglycophospholipid-anchored membrane receptorfor 5-methyltetrahydrofolate. J Cell Biol 1990:111:2931-2938.

47. Simons K, Van Meer G: Lipid sorting in epIthe-hal cells. Biochemistry 1988:27:6197-6202.

48. Zhang F, Crise B, Su B, et at.: Lateral diffusionof membrane-spanning and glycosylphosphati-dylinositol-hmnked proteins: toward establishingrules governing the lateral mobility of membraneproteins. J Cell Biol i991 :115:75-84.

49. Bretscher MS. Thomson JN, Pearse BMF:Coated pits act as molecular filters. Proc NatlAcad Sd USA i980;77:4i56-4 159.

50. Brown D, Zhu X-L, Sly WS: Localization of mem-brane-associated carbonic anhydrase type IV inkidney epithelial cells. Proc Natl Acad Sci USA1990:87:7457-7461.

5 1 . Zhu XL, Sly WS: Carbonic anhydrase IV fromhuman lung: Purification , characterization , andcomparison with membrane carbonic anhydrasefrom human kidney. J Biol Chem 1990:265:8795-8801.

52. Kerjaschki D, Farquhar MG: Immunocytochem-ical localization of the Heymann nephritis anti-gen (gp330) in glomerular cells of normal Lewisrats. J Exp Med 1983;i57:667-686.

53. Raychowdhury R, Niles JN, McCluskey RT,Smith JA: Autoimmune target in Heymann ne-phritis is a glycoprotein with homology to theLDL-receptor. Science 1 989:244: 1 1 63- 1 165.

54. Gutmann EJ, Niles JL, McCluskey RT, BrownD: Colchicine-induced redistrIbution of an en-dogenous apical membrane glycoprotein (gp 330)In kidney proximal tubule eplthelium. Am JPhysiol 1 989:257:C397-C407.

55. Lemansky P. Fatemi SH, Gorican B, Meyale S.Rossero R, Tartakoff AM: Dynamics and ion-gevity of the glycolipid-anchored membrane pro-temn,Thy-i.JCellBiol i990:ilO:1525-i53i.

56. Lacey SW, Sanders JM, Rothberg KG, Ander-son RGW, Kamen BA: Complementary DNA forthe folate binding protein correctly predicts an-choring to the membrane by glycosyl-phospha-tidylinositol. J Clin Invest 1989:84:715-720.

57. Anderson RGW, Kamen BA, Rothberg KG, La-cey SW: Photocytosis: Sequestration and trans-port of small molecules by caveolae. Science1992:255:410-411.

58. Severs NJ: Caveoli: Static mnpocketlngs of theplasma membrane, dynamic vesicles, or justplain artifact? J Cell Sd 1988:90:341-348.

59. Montesano R, Roth J, Robert A, Orci L: Non-coated membrane Invagmnations are Involved Inbinding and internalization of cholera and teta-nus toxin. Nature (Lond) 1982:296:651-653.

60. Carpentier J-L, Brown D, Jacopetta B, Orci L:Detection of surface-bound ligand by freeze-fracture autoradiography. J Cell Biol 1985:101:887-890.

6 1 . Rothberg KG, Ying Y-S, Koihouse JF, KamenBA, Anderson RGW: The glycophosphohipid-linked folate receptor internalizes folate withoutentering the clathrin-coated pit endocytic path-way. J Cell Biol 1990:110:637-649.

62. Rothberg KG, Heuser JE, Donzell WC, Ying Y-5, Glenney JR. Anderson RGW: Caveolin, a pro-tein component of caveolae membrane coats.Cell 1992:68:673-682.

63. Meilman I, Fuchs R, Helenius A: Acidificationof the endocytic and exocytIc pathways. AnnuRev Biochem 1986:55:663-700.

64. Stefanova I, Horejsi V. Ansotegui IJ, Knapp W,Stockinger H: GPI-anchored cell-surface mole-cules complexed to protein tyrosine kinases. Sci-ence 1991:254:1016-1019.

65. Pearse BMF, Robinson MS: Clathrin, adaptors,and sorting. Annu Rev Cell Biol i990:6:151-171.

66. Orci L, Glick BS, Rothrnan JE: A new type ofcoated vesicular carrier that appears not to con-


906 Volume 3 - Number 4’ 1992

tam clathrin: its possible role in protein trans-port within the Golgi stack. Cell 1986:46:17 1-184.

67. Brown D, Gluck 5, Hartwig J: Structure of thenovel membrane-coating material in proton-se-creting epithelial cells and identification as anH�ATPase. J Cell Biol 1987; i05: 1637-1648.

68. Brown DA, Rose JK: Sorting of GPI-anchoredproteins to glycolipid-enriched membrane sub-domains during transport to the apical cell sur-face. Cell 1992:68:533-544.

69. Robinson PJ: Phosphatidyhinositol membraneanchors and T-cell activation. Immunol Today1991:12:35-41.

70. Presky DH, Low MG, Shevach EM: Role of phos-phatidylinositol-anchored proteins in T cell ac-tivation. J Immunol 1990:144:860-868.

7 1 . Rock KL, Reiser H, Bamezai A, McGrew J,Benacerraf B: The LY-6 locus: A multigene fam-ily encoding phosphatidyhinositol-anchoredmembrane proteins concerned with T-cehl acti-vation. Immunol Rev 1989:1 i 1:195-224.

72. Robinson PJ, Millrain M, Antoniou J, SimpsonE, Mellor AL: A glycophospholipid anchor is re-quired for Qa-2-mediated T cell activation. Na-ture (Lond) 1989:342:85-87.

73. Waneck GL, Sherman DH, Kincade PW, LowMG, Flavell RA: Molecular mapping of signalsin the Qa-2 antigen required for attachment ofthe phosphatidylinositol membrane anchor.Proc Natl Acad Sci USA 1988:85:577-581.

74. Su B, Waneck GL, Flavell PA, Bothwell AL:The glycosyl phosphatidylinositol anchor is crit-ical for Ly-6A/E-mediated T cell activation. JCell Biol 1991:112:377-384.

75. Townsend A, Bodmer H: Antigen recognition byclass I-restricted T lymphocytes. Annu Rev Im-munol 1989:7:601-624.

76. Meyers CM, Kelly CJ: Effector mechanisms inorgan-specific immunity I. Characterization of aCD8+ T cell line that mediates murine intersti-tial nephritis. J Chin Invest 1991:88:408-416.

77. Faustrnan D, Li XP, Lin HY, et at.: Linkage offaulty major histocompatibility complex class Ito autoimmune diabetes. Science 1991:254:1756-1761.

78. Rindler MJ, Naik SS, Li N, Hoops TC, PeraldiM-N: Uromodulin (Tamm-Horsfahl glycoprotein/uromucoid) is a phosphatidyhinositol-hinked

membrane protein. J Biol Chem 1990:265:20784-20789.

79. Friedman J, Hoyer JR, Seiler MW: Formationand clearance of tubulointerstitial immune com-plexes in kidneys of rats immunized with heter-ologous sera to Tamm-Horsfahl Protein. KidneyInt 1982:21:575-582.

80. Wistrand PJ: Properties of membrane-boundcarbonic anhydrase. Ann NY Acad Sd 1984;429:195-206.

81 . Verpooten GF, Nouwen EJ, Hoylaerts MF, Hen-dzix PG, de Broe ME: Segment-specific locahiza-tion of intestinal-type alkaline phosphatase Inhuman kidney. Kidney Int 1989:36:617-625.

82. Hooper NM, Hryszko J, Turner HJ: Purificationand characterization of pig kidney amlnopep-tidase P. A glycosyl-phosphatidylinositol-anchored ectoenzyme. Biochem J i990;267:509-515.

83. Hooper NM, Keen JN, Turner AJ: Characteriza-tion of the glycosyl-phosphatidylinositol-an-chored human renal dipeptidase reveals that itis more extensively glycosylated that the pig en-zyme. Biochem J 1990:265:429-433.

84. Natori Y, Hayakawa I, Shibata S: Identificationof gpio8, a pathogenic antigen of passive Hey-mann nephritis. as dipeptidyl peptidase IV. ChinExp Immunol 1987:70:434-439.

85. Sabolic I, Culic 0, Lin S-H, Brown D: Locahiza-tion of ecto-ATPase in the rat kidney and isolatedrenal cortical vesicles. Am J Physiol 1992:262:F2 1 7-F228.

86. McNagny KM, Bucy RP, Cooper MD: Reticuharcells in peripheral lymphoid tissue express thephosphatidylinositol-linked BP-3 antigen. Eur JImmunol 1991:21:509-515.

87. Yarnamoto T, Wilson CB: Quantitative andqualitative studies of antibody-induced mesan-glal cell damage in the rat. Kidney Int 1987:32:514-525.

88. Floege J, Johnson RJ, Gordon K, et at.: In-creased synthesis of extracehhuhar matrix in mes-anglal proliferative nephritis. Kidney Int 1992;40:477-488.

89. Kerjaschki D, Farquhar MG: The pathogenicantigen of Heymann nephritis is a membraneglycoprotein of the renal proximal tubule brushborder. Proc Natl Acad Sci USA 1982:79:5557-5561.

Glycosyl-Phosphatidylinositol-Anchored Membrane...

Documents

Transcript of Glycosyl-Phosphatidylinositol-Anchored Membrane...