Glycophospholipid membrane anchoring provides clues to the mechanism of protein sorting in polarized epithelial cells

Glycophospholipid membrane anchoring provides clues to the mechanism of protein sorting in polarized epithelial cells

TIBS 15 - MARCH1990 EPITHELIAL CELLS PERFORM a variety of vectorial functions (secretion, absorption, ion transport) and thus their surface is divide...

847KB Sizes 0 Downloads 2 Views

Recommend Documents

No documents
TIBS 15 - MARCH1990

EPITHELIAL CELLS PERFORM a variety of vectorial functions (secretion, absorption, ion transport) and thus their surface is divided into domains of distinct protein and lipid composition ~-3. Comparison of the primary structures of apical or basolateral membrane proteins fails to reveal domain-specific consensus sequences, suggesting that the sorting information is present within three-dimensional motifs formed by non-contiguous protein sequences. From experiments with chimeric constructs of apically and basolaterally targeted viral glycoproteins, the extracellular or ectodomain is thought to contain dominant sorting information3. In agreement with this premise, some constitutive secretory proteins are secreted polarly to the apical or the basolateral medium 4. However, recent evidence suggests the presence of additional sorting signals in the transmembrane and/or cytoplasmic domains. Sometimes the signals in different domains are contradictory, implying the need for mechanisms that ensure dominance of one signal over the other. How does this hierarchy work? The analysis is somewhat simplified if we consider a novel class of plasma membrane glycoproteins built of single ectodomains anchored via covalent linkage to a glycophospholipid, glycosylphosphatidylinositol (GPI; Fig. la) 5~6. The C-terminal amino acid of the extracellular protein moiety is linked via an ethanolamine residue to an oligosaccharide composed of mannose, a variable number of galactose residues and (non N-acetylated) glucosamine, which in turn is attached to the inositolP portion of PI. The diacylglycerol portion of P! is the only site for membrane attachment of these proteins. Since several well known apical ecto-enzymes (alkaline phosphatase, renal dipeptidase, trehalase and 5'nucleotidase) are anchored by GPI, and GPI-anchored Thy-1 was apically polarized when expressed in the kidneys of transgenic mice, Lisanti and colleagues 7 devised a method to survey the polarized diJtribution of all GPI-anchored proteins in a

M.P. Lisanti and E. Rodrlguez-Boulan are at Cornell University Medical College, Department of Cell Biologyand Anatomy, 1300 YorkAve, New York, NY 10021, USA.

The sorting signals that guide proteins to apical and basolateral surfaces of epithelial cells have remained elusive. Current evidence suggests a hierarchy of sorting information with multiple sorting signals (apical and basolateral) present in different domains of a given plasma membrane protein. The observation that covalently attached glycosyl-phosphatidylinositol (GPI) acts as a 'dominant' apical targeting signal is compatible with the involvement of glycolipids in epithelial protein sorting. series of well characterized epithelial cell lines that retain the polarized phenotype.

GPI anchoring correlates with apical localization The procedure to survey the polarized distribution of GPI-anchored proteins (Fig. lb) takes advantage of several conserved structural features of the GPl-anchoring mechanism: (1) cell surface localization; (2) sensitivity to cleavage by a bacterial PI-specific phospholipase C (PI-PLC); and (3) conversion from a hydrophobic to a hydrophilic state upon hydrolysis by PI-PLC, coupled with a biochemical assay for polarity using a biotin analog impermeable to the tight junctional barrier 7,8 that divides the apical and basolateral domains of epithelial cells. Six GPl-anchored proteins were detected in the model MDCK (strain I!) cell line and all appeared apically polarized (Fig. 2, upper panel) 7. Study of other model epithelial cell lines confirmed the same striking correlation, all GPI-anchored proteins were apical (at least 13 different proteins based on distinct molecular weights) 9. These included renal lines, such as high resistance MDCK strain I, LLC-PK1, and intestinal lines, namely Caco-2 and SK-CO15. Two of these proteins were identified as known GPl-anchored proteins: carcinoembryonic antigen (CEA) and decayaccelerating factor (DAF). Since the common feature of this protein class is their unusual membrane anchoring

© 1990,ElsevierSciencePublishersLtd,(UK) 0376-5067/90/$02.00

mechanism, it seemed likely that the GPI-moiety played an additional role as a signal for targeting the attached protein to the apical surface.

Recombinant transfer of GPI to a basolateral or regulated secretory protein leads to apical expression A molecular biological approach was used to study whether attachment to GPI might target basolateral or unsorted secretory proteins to the apical surface. This approach was based on information that recently became available on the mechanism of GP1 attachment. GPI-anchored proteins contain a conventional N-terminal signal sequence which serves in translocation of the protein into the ER lumen. In addition, there is a second cleavable hydrophobic sequence at the C-terminal end of the protein 1°,11which serves as a signal for GPI attachment. In the ER lumen, cotranslationally, this sequence is cleaved and GPI added en bloc by as yet uncharacterized enzymes. Interestingly, this sequence shares properties with the ER signal sequence since it may be functionally replaced by the human growth hormone (hGH) signal sequence 12. MDCK cells were transfected with cDNAs encoding exogenous GPIanchored proteins, such as DAF and fusion proteins containing the DAF GP1attachment signal. These 'cut and paste' experiments are summarized in Table 1. Fusion proteins consisted of the ectodomain of a basolateral antigen (the herpes simplex glycoprotein D, 113

TIBS 15 - MARCH1990

apical surface, but also may re-route basolateral proteins, acting as a 'dominant' apical sorting signal.

(a) kk





Glycan In(~ol







Hydrophobic amino


(b) Aqueous-discard . Sulfo-NHS-biotin or



o ~


(1) Extraction (TX- 114)

(2) Phase separation


Aqueous-collect Detergent-discard

(4) Phase separation

(3) PI- PLC

SDS-PAGE streptavidin-blotting

Figure 1 (a) The three generic forms of membrane proteins are illustrated: peripheral, integral (transmembrane) and GPI-anchored. Note that cell-surface labeling with sulfo-NHS-biotin tags only the protein ectodomains. (b) Polarity assay for GPI-anchored proteins. After apical (solid line) or basolateral (dotted line) labeling with sulfo-NHS-biotin, epithelial monolayers are (1) extracted with Triton X-114 and membrane proteins are (2) subjected to temperature-induced phase separation. Detergent phases (enriched in hydrophobic forms of GPIanchored proteins) are collected, and (3) treated with PI-PLC (arrow) to promote cleavage of the detergent binding domain (i.e. diacylglycerol). Finally, (4) soluble, hydrophilic forms of GPI-anchored proteins are now found in the aqueous phase after an additional phase separation. Biotinylated proteins are detected after SDS-PAGE/transfer to nitrocellulose by 1251-streptavidin blotting and autoradiography7,8.

gD-1) or of a regulated secretory protein (hGH) linked to the C-terminal 37 amino acids of DAFl°,u. Native gD-1 is expressed on the basolateral surface in HSV-infected or in transfected cells ~3, directed in part by sorting signals in the ectodomain, since truncated gD-1 is secreted basolaterally ~s. Transfected hGH was released in significant amounts from both epithelial surfaces ~3, like other exogenous secretory proteins expressed in MDCK cells 4. Strikingly,


GPI-anchored gD-I (Fig. 2, lower panel) and GPloanchored hGH were targeted to the apical surface, as w a s DAF j3. Independent results by Brown, Crise and Rose ~4showed that a fusion protein carrying the ectodomain of basolateral VSV G protein linked to the anchoring sequence of Thy-1 (another GPIanchored protein) was also apically targeted. These observations demonstrate that GPI anchoring not only can direct an unsorted secretory protein to the

A hierarchy of sorting information The experiments discussed above suggest that multiple sorting signals [apical GPl-anchor signal(s) versus basolateral ectodomain signal(s)] may co-exist in different regions of a given membrane protein. Transfection of MDCK cells with cDNAs encoding forms of DAE Thy-1 and placental alkaline phosphatase (PLAP) that lack the GPI anchor results in preferential apical secretion of the gene products ~:~-ls,indicating that additional or 'redundant' apical sorting information exists in the ectodomains of these three GPIanchored proteins. However, a chimeric protein containing the ectodomain of PLAP anchored via the transmembrane and cytoplasmic domains of VSV G protein (a basolateral antigen) is targeted to the basolateral membrane ~4 (see Table I). A similar chimeric molecule consisting of human chorionic gonadotropin (ct-subunit) anchored by the transmembrane and cytoplasmic domains of VSV G protein is sorted basolaterally~E Thus, a hierarchy of sorting information exists within membrane proteins, with different regions of the molecule contributing redundant or contradictory information. Which one dominates over the other may depend on their differential affinities for the sorting mechanism or (perhaps more likely) on the sequence of their operation (see also Ref. 3 for a discussion of this point). Sequential operation of different sorting signals in a given molecule has been clearly shown for mammalian lysosomal hydrolases; N-terminal signal sequences mediate incorporation into the ER, and mannose 6-phosphate groups target them to the lysosome via a specific receptor in the trans-Golgi network (a subcompartment of the distal Goigi apparatus) ~7. In the case of epithelial plasma membrane proteins, we speculate that sorting interactions of the anchors with the bilayer may precede interaction of the ectodomain with sorting elements in the trans-Golgi network. If the off-rate of the first mechanism is low, or if clustering of the anchors into a region of the bilayer prevents interaction with ectodomainrecognition mechanisms in a spatially distinct region of the sorting organelle, then the anchoring mechanism would

TIBS 1 5 - MARCH1990

'dominate' simply by being the first operational mechanism. If the anchor is absent, then the protein arriving to the sorting compartment is free to diffuse and interact with a more distal sorting mechanism which recognizes signals of identical or opposite sign (i.e. apical or basolaterai). An example of this sequential principle for the sorting of plasma membrane proteins is provided by the complex surface delivery pathway of the polymeric-ig receptor. This receptor is initially directed to the basolateral membrane of epithelial cells (where its function is to bind polymeric antibodies [plg] produced by lymphoid cells) by dominant information present in the cytoplasmic domain. It is then endocytosed and delivered to the apical domain (where plg is used to fight infectious agents) by ectodomain signals a. Endocytosis at the basolateral surface requires specific information for coated pit clustering in the cytoplasmic domain, while postendocytic sorting to the apical surface is presumably associated with release of the dominance of the cytoplasmic domain via serine phosphorylation ]7.

apical and basolateral proteins initially use the basolateral pathway but only apical proteins must be relocated 4, implying that the basolateral pathway is not signal mediated. However, there are basolateral proteins that travel first to the apical membrane and are then relocated to the basolateral membrane of intestinal cells ~9 and apical proteins that are vectorially targeted in intestinal cells 2°. Demonstrating what pathway is default is clearly difficult. Direct proof that a given pathway is signal mediated requires (1) a signal that is necessary and sufficient for targeting molecules


into that route (e.g. defined by deletion and transfer to proteins following a different route); (2) a specific sorting receptor; and (3) an energy-dependent concentration step in the sorting compartment. Whereas criteria (2) and (3) have not yet been demonstrated for either pathway, there is evidence for transferable signals that can confer apical or basolateral localization ~3,14,~" (see above). Where are GPl~anchored proteins sorted? As mentioned previously, several apical proteins in liver and intestine appear first on the basolateral sur-


Caco-2 A

.....,-:~-.~-~ IA. Y~l , ~ / B

I.~:.,.~...___,_~:~.--J/~22 ] 4

. a










Is one of the membrane pathways default in epithelial cells? In our discussion, we have assumed that polarized transport of membrane proteins requires both apical and basolateral targeting signals. Alternatively, transport to only one domain may be signal-mediated, while transport to the other occurs by default (i.e. without domain-specific sorting information) (reviewed recently in Refs 1-4). Three arguments, based on circumstantial evidence, support basolateral transport as the default pathway: (1) Basolateral proteins are shared with non-epithelial cells (which transport plasma membrane proteins by default); apical proteins are specific for epithelial cells TM. However, GPl-anchored proteins 5,~ and other apical markers (e.g. aminopeptidase N) ~8, are detected in non-polarized cells. (2) Mis-sorting is greater for apical proteins ~, suggesting signal-mediated diversion from the bulk basolateral membrane flow (like barrels rescued from the edge of a waterfall). However, the higher efficiency of basolateral protein sorting may suggest that it is signal mediated. (3) In intestinal and liver cells, several

Figure 2 Polarized apical distribution of endogenous and transfected GPI-anchored proteins in epithelial cells. Upper panel: Monolayers of the epithelial cell lines MDCK II (left), LLC-PK1 (center) and Caco-2 (right) were labeled with sulfo-NHS-biotin from the apical (A) or from the basolateral (B) side and the GPI-anchored proteins identified with polarity assay described in Fig. 1. Autoradiograms after 1251-streptavidin blotting were scanned densitometrically and the difference between PI-PLC treated and untreated samples obtained by computer subtraction and plotting 7,9. Note that all GPI-anchored proteins are apical in the three cell lines. Lower panel: Immunofluorescence localization of GPIanchored gD-1 (a,b) and native gD-1 (c,d) in transfected MDCK cells. Cells were fixed with formaldehyde and either permeabilized with saponin (b,d) or left intact (a,c) before processing for immunofluorescence with HSV antibodies. GPI-anchored gD-1 appeared in high concentration on the apical surface, note typical punctate pattern of microvilli in (a). Native gD-1 was detected intracellularly in (d) (perinuclear localization) and at the basolateral surface (faint intercellular rings), but was absent from the apical surface in (c). Reproduced, with permission, from Refs 9 and 13. 115


Table I. Polarity of chimeric, native and anchorless GPI-anchored proteins In transfected MDCK cells

their association with a specific class of membrane or secretory proteins, as has been proposed for hormones Protein ectodomain Membrane anchoring mechanism Polarity Refs concentrated in regulated secretory • native, transmembrane basolateral 13,14 Herpes simplex granules 25. Although there is evidence • anchorless, secreted (gD-1 only) basolateral virus gD-1 or demonstrating the ability of certain • fusion protein, GPI-anchored apical VSV G glycolipids to form clusters after exHuman growth • native anchorless, secreted unsorted 13 ceeding a given molar concentration in hormone (hGH) • fusion protein, GPI-anchored a apical the lipid bilayer26, there are no data directly demonstrating the affinity of Decay accelerating • native, GPI-anchored a apical 13-15 apical proteins for glycolipid clusters, factor (DAF), placental • anchorless, secreted apical • fusion protein, with G protein basolateral alkaline phosphatase or even for the existence of these clusor Thy-1 transmembrane and cytoplasmic ters in the trans-Golgi network. On the domains (PLAP only) other hand, FRAP (fluorescence recovaDAF and hGH-DAF fusion proteins (•4DAF and hGH-A4DAF) containing an alternative DAF signal ery after photobleaching) has been used for GPI attachment l l are also apically targeted (Lisanti, Caras, Davitz and Rodriguez-Boulan, unto describe some paradoxical biophysipublished). cal properties of GPi-anchored proteins, such as relatively high diffusion coefficients (D = 2-4 x 10-9 cm ~s -~) and large face before being endocytosed and involving GPl-anchored proteins must immobile fractions (50%)5,~, the latter delivered to the apical surface 4. consider that this protein group suggesting the existence of clusters. Alternatively, proteins may be sorted accounts for only a relatively small frac- Further studies on the biophysical intracellularly in the trans-Golgi net- tion (perhaps less than 5%) of the properties of GPl-anchored proteins work and delivered vectoriaily to the apical protein population. A simple pro- may provide important insights into the respective surface~-4; the envelope gly- posal would be that one (or more) GPI- role of glycolipid clustering as a general coproteins of polar budding viruses and anchored protein(s) act as receptors mechanism of protein sorting in polarseveral endogenous glycoproteins fol- for the targeting of other apical pro- ized epithelial cells and, perhaps, in low this route in MDCK cells2L It is teins. A second model (see, for example, domain formation in other cell types. unlikely that GPl-anchored proteins use Refs 1, 2 and 24) stems from the the first mechanism since it requires observation that glycosphingolipids Structural aspects of GPI anchors that may endocytosis and Thy-1 (a GPI-anchored (sphingomyelin and glucosyl ceramide) be related to apical sorting protein), as well as molecules with are enriched in the apical membranes The fusion protein experiments short cytoplasmic tails (e.g. influenza of polarized cells, where they constitute described above indicate that dominant haemagglutinin), are excluded from almost 100% of the exoplasmic leaflet24. apical sorting information may be transcoated pits and are very poorly endo- Their polarized distribution apparently ferred to the ectodomains of basolatcytosed~7.22; similarly, two GPI-anchored derives from intracellular sorting and eral glycoproteins by addition of 37 proteins, DAF and gD-1-DAF, are not direct apical delivery from the Golgi amino acids of DAF1:~or 53 amirio acids endocytosed at significant rates complex, the site of their synthesis 24. of Thy-l~4. In the case of DAF, the hydro(Lisanti, Caras, Davitz and Rodriguez- This has led to the hypothesis that the phobic sequence is 17 amino acids Boulan, unpublished). A third possibil- sorting of apical glycoproteins might long ~°-~2and 8-10 of the 37 DAF amino ity is that the observed polarized distri- depend on their non-covalent interac- acids remain after signal cleavage and bution of GPI-anchored proteins results tion with glycolipid-rich clusters in the GPI attachment (Caras et al., unpubfrom unpolarized delivery followed by trans-Golgi network, resulting in their lished). Theoretically, sorting informaselective cleavage by a putative baso- incorporation into apical carrier ves- tion may be present in the anchor itself lateral GPI-anchor degrading enzyme icles; basolateral carrier vesicles could or in the DAF sequence that remains. (protease, glycosidase or phospholi- form by glycolipid exclusion24. Since The first possibility appears most likely pase) 23. This mechanism is unlikely, glycolipids far outnumber apical pro- since replacement of the 8-10 DAF however, since shed ectodomains of teins on a molar basis, such a mech- amino acids by upstream DAF seGPI-anchored proteins are not detected anism would provide for efficient, high- quences (thereby creating an alternain the basolateral medium ~3. Thus, sort- capacity segregation of apical and baso- tive signal for GPl-attachment H) does ing at the trans-Golgi network and vec- lateral membrane proteins without the not affect apical targeting of DAF or torial apical delivery appears as the need for a proteinaceous 'sorting recep- a hGH-DAF fusion protein (Lisanti, most likely mechanism for generating tor'. Segregation would be based on the Caras, Davitz and Rodriguez-Boulan, unthe polarized distribution of GPI- existence of different lipid bulk flows to published). anchored proteins in epithelial cells. the apical or the basolaterai surfaces. In the GPI anchor, sorting information Alternatively, proteinaceous apical and may be present in the ethanolamine, Sorting of apical glycoproteins and basolateral sorting receptors (with the glycan group or the PI moiety. The glycolipids: are they linked? affinity for different protein domains or glycan group is presumed to be proThe experiments with GPl-anchored GPI) might be sorted through their in- cessed like the N-linked glycan groups proteins indicate that GPI may act as an trinsic affinity for apical or basolateral of glycoproteins, with addition of cerapical targeting signal in epithelia. At lipid bulk flows. Once segregated, they tain components in the ER and others the moment it is possible only to specu- might enter a distal sorting compart- in the Golgi apparatus. Based on the late on how this glycophospholipid may ment where an environmental change predicted three-dimensional structure contribute to apical sorting. Any model (e.g. a change in pH) could promote of the GPl-anchor, the glycan lies paral116

TIBS15-MARCH1990 lel to the plane of the membrane and occupies about 6 nm 2 of membrane surface area 27. Galactose does not seem to be necessary for apical sorting since Ricinus communis agglutinin-resistant mutants of MDCK cells, with a pleiotropic defect in galactosylation2s,29, still display a polarized distribution of GPl-anchored proteins 9. On the other hand, Concanavalin A-resistant mutant MDCK cells 2s show significant depolarization of two out of their five surface GPl-anchored proteins 9, suggesting that glycosylation may have a sorting role for these proteins. Presumably this mutation results in the production of an altered GPI anchor lacking a putative side group (such as mannose, ethanolamine-P or acylation of the inositol ring) which may contribute to efficient apical localization. Unaffected GPIanchored proteins might be sorted based on 'redundant' apical targeting information present in their ectodomains. Glycosylation also appears important for generating the polarized distribution of sphingolipids, since glucosyl ceramide is five times more concentrated on the apical surface than sphingomyelin24, yet they differ in molecular structure only by a single sugar residue (i.e. glucose replacing choline-P as the polar head group). In further support of this hypothesis, methylation of the glucose moiety of glucosyl ceramide reduces its ability to form clusters to the level observed for sphingomyelin26. Finally, the third possibility is that part (or all) of the glycerophospholipid PI is responsible for sorting. Alternatively, the hydrophobic domain of epithelial GPl-anchored proteins may be composed of ceramide instead of diacylglycerol, a possibility that is enhanced by the observation that inositolphosphate containing glycosphingolipids (glycosyl inositol phosphoryl ceramide) exist in eukaryotic organisms and are analogous to the structure proposed for the GP! anchor 6. Studies with glycosphingolipids suggest that intermolecular hydrogen bonding between hydroxyl and carbonyl groups of the sphingosine moieties or between sugars may be involved in clustering 2G. Although clustering is not established as a mechanism relevant to epithelial protein sorting, its role in the sorting of molecules in the endocytic pathway is well documented 17. It is expected that additional biophysical and biochemical characterization of GPI anchors will provide important clues to the sorting of polarized membrane glycoproteins.

Polarity of GPl-anchored proteins in nonepithelial cell types and membrane microdomains

Neutrophils undergo chemoattractant-induced polarization (in response to fMet-Leu-Phe) and reorganize themOther cell types segregate their sur- selves into locomotor cells with anface proteins into distinct domains. terior and posterior segments; such Typical examples are the surface differ- polarization is also observed in locoentiation of neurons and neuroen- motor lymphocytes. In these cells, GPIdocrine cells into soma and neurite anchored proteins such as the Fc recepextensions (dendrites and axons), of tor, DAF (as defined by C3b binding), sperm cells into head and tail regions and Thy-1 (0-antigen) are localized to (further subdivided into anterior and the anterior pole 34, whereas cell surface posterior segments) and of locomotor immunoglobulins:~5(on B lymphocytes), leukocytes into anterior (leading) and coated pits and most cellular organelles are localized to the posterior pole. posterior (trailing) domains. Like epithelial cells, neuroendocrine Furthermore, early experiments indicell types segregate proteins to distinct cated that Thy-1 behaved differently constitutive or regulated secretory from other membrane glycoproteins. pathways. Upon VSV infection of the Thy-1 is excluded from clathrin-coated neuroendocrine cell line, AtT-20, the pits 22, but is selectively incorporated envelope protein G is initially expressed into the glycoprotein envelopes of budon the cell body and excluded from pro- ding virions 36 and mycoplasma:~7. Its cess tips 2'~.In contrast, endogenous and possession of the novel GPi-anchoring exogenous regulated secretory proteins mechanism may explain why it is 'sort(growth hormone, insulin and ACTH) ed' differently from its transmembrane accumulate in dense-core secretory gran- counterparts by plasma membrane ules at the tips of neurite-iike proces- microdomains. Thus, GPI anchorage ses ready for release by the appropriate may also direct and/or regulate specific stimuli25. Similarly, certain GPl-anchored cellular localization in other polarized proteins (acetylcholinesterase:~° and non-epithelial cell types and in memneuronal-cell-surface protein F33~) are brane micro-domains. localized most prominently to neurites and neurite bundles of neuronal cell Conclusion cultures and (5' nucleotidase:~2)to axons Lipids are asymmetrically distributed and nerve-terminal ramifications in the among the various cellular organelles, Torpedo electric organ. By comparison yet elucidation of the mechanisms that of the localization of membrane pro- determine their sorting have lagged teins (VSV G protein and GPl-anchored behind their protein counterparts. proteins) in neurons and polarized In the case of polarized epithelial epithelia, the neuronal cell body may be cells, glycosphingolipids (e.g. glucosyl analogous to the basolateral surface ceramide) are preferentially enriched in with specialized outgrowths corres- the apical membrane. Since both proponding to the apical membrane. teins and lipids must be segregated and In guinea pig sperm, the cellular dis- transported differentially to the apical tribution and diffusibility of a GPI- and basolateral surfaces, it has been anchored protein, PH-20, varies with suggested that lipid and protein sorting the state of maturation :~3. PH-20 is ran- may occur by a common mechanism domly distributed over the entire cell with glycosphingolipids acting as apical surface in testicular sperm, and is 'im- sorting molecules through self-aggremobile' (D ~10 -11 cm 2 s-~), as measured gation and co-clustering with apical by FRAE In contrast, in epididymal proteins in the trans-Golgi network. The sperm PH-20 is localized to the posterior observations that (1) endogenous GPIhead region and intracellularly at the anchored proteins are apically polarinner acrosomal membrane. After the ized, and (2) recombinant transfer of acrosome reaction, a barrier to diffusion GPI to basolateral or regulated secretory (morphologically not unlike the tight proteins confers apical expression, junctions of epithelial cells) is removed indicate that membrane attachment via and PH-20 redistributes to the anterior this glycophospholipid can act as a head region, while other integral mem- 'dominant' apical transport signal to brane proteins remain posterior. When determine the final destination of a localized to either posterior or anterior, given protein ectodomain. These PH-20 is freely diffusible with diffusion results provide the most suggestive evicoefficients comparable to the mobile dence obtained to date to implicate fractions of other GPl-anchored pro- glycolipid involvement (in this case coteins and lipid probes (D ~ 10-~)cm 2 s-~). valent) in the sorting of apical proteins. 117

TIBS 1 5 - MARCH 1 9 9 0

Acknowledgements We thank Drs Tony Brown, Moses Chao, Sandra Citi, Paul Deutsch, Kathy Perrego and D. Wall for critically reading the manuscript. MPL is the recipient of a Medical Scientist Training grant from the Cornell University MD/PhD program. ERB is the recipient of an Established Investigator Award and a grant in aid from the American Heart Association (New York Branch). This research was supported by grants from the National Institutes of Health (GM34107, HL-37675 and GM-41771) and the American Cancer Society.

References 1 Wandinger-Ness, A. and Simons, K. Intracellular Trafficking of Proteins (Hanover, J. and Steer, C., eds), Cambridge University Press (in press) 2 Rodriguez-Boulan, E. and Nelson, W. J. (1989) Science 245, 718-725 3 Caplan, M. and Matlin, K. S. (1989) in Functional Epithelial Cells in Culture, Modem Cell Biology, Vol. 8 (Matlin, K. S. and Vatentich, J.D., eds), pp. 71-127, Alan R. Liss 4 Bartles, J.R. and Hubbard, A.L. (1988) Trends Biochern. Sci. 13, 181-184 5 Low, M. and Saltiel, A. R. (1988) Science 239, 268-275 6 Ferguson, M. A. J. and Williams, A. F. (1988) Annu. Rev. Biochem. 57, 285-320 7 Lisanti, M.P., Sargiacomo, M., Graeve, L., Saltiel, A. R. and Rodriguez-Boulan, E. (1988)

Proc. Natl Acad. Sci. USA 85, 9557-9561 8 Sargiacomo, M., Lisanti, M. P., Graeve, L., Le Bivic, A. and Rodriguez-Boulan, E. (1989) J. Membr. Biol. 107,277-286 9 Lisanti, M. P., Le Bivic, A., Saltiel, A. R. and Rodriguez-Boulan, E. (1990) J. Membr. Biol. 113, 155-167 10 Caras, I.W., Weddell, G. N., Davitz, M. A., Nussenzweig, V. and Martin, D. W. (1987) Science 238, 1280-1283 11 Caras, I. W., Weddell, G. N. and Williams, S. R. (1989)-1 Cell Biol. 108, 1387-1396 12 Caras, I. W. and Weddell, G. N, (1989) Science 243, 1196-1198 13 Lisanti, M. P., Caras, I. W., Davitz, M. A. and Rodriguez-Boulan, E. (1989) J. Cell Biol. 109, 2145-2156 14 Brown, D. A., Crise, B. and Rose, J. K. (1989) Science 245, 1499-1501 15 Powell, S. and Rodriguez-Boulan, E. (1989) J. Cell Biol. 109, 296a [abstr.] 16 Brown, D. A. and Rose, J. K. (1988) J. Cell Biol. 107,782a [abstr.] 17 Breitfeld, P. P., Casanova, J. E., Simister, N. E., Ross, S. A., McKinnon, W. C. and Mostov, K. E. (1989) Curr. Opinion Cell Biol. 1, 617-623 18 Look, A. T., Ashmun, R. A., Shapiro, L. H. and Peiper, S. C. (1989) J. Clin. Invest. 83, 1299-1307 19 Mellman, I. (1989) Curt. Opinion Immunol. 1, 16-25 20 Le Bivic, A., Real, F. and Rodriguez-Boulan, E. (1989) Proc. Natl Acad. Sci. USA 86, 9313-9317 21 Lisanti, M. P., Le Bivic, A., Sargiacomo, M. and Rodriguez-Boulan, E. (1989) J. Cell Biol. 109, 2117-2127 22 Bretscher, M. S., Thomson, J. N. and Pearse, B. M. F. (1980) Proc. Natl Acad. Sci. USA 77,

4156-4159 23 Lisanti, M. P., Darnell, J. D., Chan, B. L., Rodriguez-Boutan, E. and Saltiel, A. R. (1989) Biochem. Biophys. Res. Commun. 164, 824-832 24 Simons, K. and van Meet, G. (1988) Biochemistry 27, 6197q3202 25 Rivas, R. J. and Moore, H-P. H. (1989) J. Cell Biol. 109, 51-60 26 Thompson, T. E. and Tillack, T. W. (1985) Annu. Rev. Biophys. Biophys. Chem. 14, 361-386 27 Ferguson, M. A. J., Homans, S. W., Dwek, R. A. and Radernacher, T. W. (1988) Biochem. Soc. Trans. 16,265-268 28 Meiss, H. K., Green, R. F. and Rodriguez-Boulan E. (1982) Mol. Ceil. Biol. 2, 1287-1294 29 Brandli, A. W., Hansson, G. C., RodriguezBoulan, E. and Simons, K. (1988) J. Biol. Chem. 263, 16283-16290 30 Rotundo, R. L. and Carbonetto, S. T. (1987) Proc. Natl Acad. Sci. USA 84, 2063-2067 31 Gennarini, G., Cibelli, G., Rougon, G., Mattei, M-G. and Goridis, C. (1989) J. Cell Biol. 109, 775-788 32 Grondal, E. J. M. and Zimmermann, H. (1987) Biochem. J. 245, 805-810 33 Phelps, B. M., Primakoff, P., Koppel, D., Low, M G. and Myles, D. G. (1988) Science 240, 1780-1782 34 Haston, W. S. and Wilkinson, P. C. (1988) Curr. Opinion Immunol. 1, 5-9 35 Schreiner, G. F., Braun, J. and Unanue, E. R. (1974) J. Exp. Med. 144, 1683-1688 36 Calafat, J., Janssen, H., Demant, P., Hilgers, J. and Zavada, J. (1983) J. Gen. Virol. 64, 1241-1253 37 Wise, K. S., Cassell, G. H. and Acton, R. T. (1978) Prec. Natl Acad. Sci. USA 75, 4479-4483

GRAPEVINE Pfizer awards - 1989

Whistler award winner 1990

The research division of the pharmaceutical company Pfizer Ltd. has increased each of its six annual awards to young scientists (first made in 1983) to £4000. This money is to be used personally by the recipients to further their research. To qualify, the entrant's research must have potential application in the search for human and animal health medicines. The awards, in the fields of chemistry, biology and pharmaceutical sciences, are made on the basis of work published in the previous two years (or, in the case of a second consecutive award, in the previous 12 months). The 1989 winners include: Dr G. P. Winter (Medical Research Council, Laboratory of Molecular Biology, Cambridge, UK.) for the application of site-directed mutagenesis and molecular modelling in the analysis of protein structure and function, and Dr P. L. R. Andrews (Department of Physiology, St George's Hospital Medical School, London, UK.) for his studies which have furthered our understanding of the nervous control of gut function.

The Whistler award was established by the International Carbohydrate Organization to honor Roy L. Wl~istler, Professor Emeritus, Purdue University, USA, and to recognize scientists who have made contributions of excellence, with promise of continuing significant contributions in carbohydrate research. it carries a $10 000 cash prize and is made biannually at the International Carbohydrate Symposium. Johannis P. Kamerllng of the Department of Bio-Organic Chemistry, Utrecht University, The Netherlands is the fourth recipient of the Whistler Award in Carbohydrate Chemistry. The presentation will take place at the 15th International Carbohydrate Symposium, 12-17 August 1990, Yokohama, Japan. Dr Kamerling is being recognized for major contributions to modern structural carbohydrate chemistry. He has played an important role in the introduction of physical methods for structural elucidation of polysaccharides and glycoconjugates, methods which have revolutionized the means by which complex carbohydrate structures are


determined and which have provided the basis for our emerging understanding of interactions between oligosaccharide structures and protein molecules.