(Podocyte) Membrane Protein-tyrosine Phosphatase

THEJOURNAL OF BIOLOGICAL CHEMISTRY

Vol. 269, No. 31, Issue of August 5 , pp. 19953-19962, 1994 Printed in U.S.A.

0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

GLEPPl, a Renal Glomerular Epithelial Cell (Podocyte) Membrane Protein-tyrosine Phosphatase IDENTIFICATION, MOLECULAR CLONING, AND CHARACTERIZATION IN RABBIT* (Received forpublication, September 22, 1993, and in revised form, April 4, 1994)

Peedikayil E. Thomas, Bryan L. marram, Meera Goyal,Jocelyn E. Wiggins, Lawrence B. Holzman, and Roger C. Wiggins$

, .

From the DeDartment o f Internal Medicine. NeDhroloev Division, Universityof Michigan, -“ Ann Arbor, kichigan 48109-0364

tion and characterizationof such proteinswould provide useful Podocytes are specialized epithelial cells with delicate interdigitating foot processes which cover the ex- information abouthow podocyte structure andfunction is reguterior basement membrane surface of the glomerular lated. We report the first such protein identified, which is a capillary. They are in part responsible for the extraor- membrane protein-tyrosine phosphatase(PTPase)’ localized to dinary charge and size filtration characteristicsof the the foot processes of the podocyte. glomerulus. To better understand disease processes afThe potential importance of a membrane PTPase in local fecting the glomerular filter, we searched for proteins regulation of foot process structure andfunction is emphasized with relative specificity to the podocyte using a mono-by recent reports defininga role for CD45 in T cell activation clonal antibody strategy. The first such protein charac- via the T cell antigen receptor (TCR). The TCR is a multipro1 tein complex consisting of both variable chains (a and 0) and terized(designatedglomerularepithelialprotein (GLEPP1)) isa membrane protein-tyrosine phosphatase invariant chains (CD3 y, 6, E , and 6 8quence Software Package (GCG, Madison, WI) MOTIFS program indicates that there are15 potential N-linked glycosyla4 5tion sites in the extracellular domain of GLEPPl. Infibronectin type I11 repeat VII, the sequence KKKXKK (amino acid 67429679) where X is a hydrophobic residue might represent a glycosaminoglycan binding site (34). 18Analysis of the Intracellular Domain-The intracellular domain (C-terminal to the transmembrane region) contains a 2 3 0 1 single typical 276-amino acid PTPase domain and PTPase acProtein (ug) tive site aminoacid sequence ((W)HCXAGXXRW")G) (9,351. This PTPase domain begins 45 amino acids C-terminal of the 32P-Myelin Basic E 8 32P-Raytide transmembrane region. Within the PTPase domain are potential regulatory phosphorylation sites for casein kinase I1 (Ser954, Ser-974, Thr-988, Ser-1040) and protein kinase C (Serl1144) as assessed using the MOTIFS program of the GCG x Protein Protein package. Notably, two of five cDNA clones examined lacked 36 bases (nucleotides 2,804-2,839) corresponding to the amino Q I n In m acid sequence SKNGLKKRKLTN in the intracellular region immediately preceding the PTPase domain, suggesting that GST andECD (I: ECD GST and this may be an alternatively spliced region of the GLEPPl Fusion Protein Fusion Protein n molecule. This lysine-rich region contains a potential phosphoN O 2 0 2 01". " 0 1 2 3 " 0 1 2 3 rylation site (Thr-886) for CAMP-dependent protein kinase. Proteln (ug) Protein (us) The aminoacid sequence of GLEPPl wascompared with that of two other membersof the PTPasefamily that are transmembrane receptors and have a single PTPase domain (type I11 membrane PTPase receptors, as defined by Fischer et al. (Ref. 9)). These included human placental PTPP and DPTPlOD (a Drosophila centralnervous system-specific PTPase).When amino acid sequences were compared (Fig.4 B ) , these PTPases showed greaterthan 90% identitywith eachother. These PTPase sequences were then compared with thoseof six memf SI)S-polyacrylamstudies. Upper l ~ f purlel, bers of the membrane PTPase family that have tandem PTPase Fw. 5. Fusion proteinshowing the affinity-purified 59-kDa PTPase ide Coomassie-stainedgel domains (human CD45, human PTPa, human PTP6, human fusion protein (lane R 1, the affinity-purified 74-kDa extracellular doleukocyte common antigen-related PTP, Drosophila PTP, and main fusion protein (lane c),and the affinity-purified25-kDa glutathiDrosophila PTP99A), and to three members of the intracellular one S-tranferase protein(lane D )purified by glutathione affinitychromatography. Lane D is overloaded,giving two apparentbands for single domain PTP family (human T cell PTP, human PTPlp, glutathione S-transferase. The starting material (bacterial extract) for and human hematopoietic PTP) (9, 24, 26, 35-43). The single the PTPase fusion protein prior to the affinity column step is shown in domain typeI11 membrane receptor PTPases had less than 80% lane A. Upper right and middle panels,assay of phosphatase activityof identity with anyof the other PTPasesequences examined. As the purified preparations of PTPase fusion protein (open squares),the ECDfusionprotein (closedsquares), and the GST protein (closed shown in Fig. 4 B , consensus amino acid residues present in the circles) whentested for phosphataseactivityagainstp-nitrophenyl single domain type I11 PTPase receptors were similar to those phosphate substrate(top right panel), [R2P]tyrosine myelin basic protein previously described as being present in most PTPases (43). substrate (middle left panel), and [:12PltyrosineRaytide peptide substrate (middle right panel ). Note that the PTPase fusion protein exHowever, the three members of the type I11 receptor family pressed phosphatase and tyrosine phosphatase activities. shared 12 consensus amino acids, which were either not pres- The ECDenzymatic fusion protein and GST did not show phosphatase activity for ent orwere only encountered once in the15 non-type I11 recep- any of the substrates(closed circlesand closed squares overlap). Bottom tor PTPasesequences examined (bothfirst and second domains panels (immunophotomicrographs), immunizationof a guinea pig( G P ) fusion protein-producedantiserum, of tandem domain PTPases were included in the analysis).Of withtheextracellulardomain recognized epitope(s) in the same distribution in renal cortex particular note were 2 cysteine residues, a t positions 123 and which (lower left panel) a s does the mouse GLEPPl monoclonal antibody 4C3 280 according to the numbering systemused in Fig. 4B. The 2 (lower right panel). cysteines were not present in other PTPase sequences examined. In addition, at position 160, a tryptophan was present in tivity, the nucleotide sequencespanning from immediately 3'of all three typeI11 receptor PTPases. At this position, a tyrosine the putative alternate splice site to the end of the coding region, residue was present in 14 of the other 15 PTPase sequences including nucleotides 2,841-3,743 (903 nucleotides, 300 amino examined. Thus, type I11 receptor PTPases contain specific acids, molecular mass 35 kDa), was cloned into thepGEX vecamino acid residues within the PTPase domain that appear to tor toform a fusion protein with glutathione S-transferase.As define this type of receptor. This finding may reflect common a control, the region coding for the part of the GLEPPl ECD recognized by the 4C3 mAb spanning nucleotides 525-1,622 ancestor genes and/or restrictions in structure related to the (1,098 nucleotides, 366 amino acids, molecular mass 42 kDa) as a single membranePTPaseunit. enzymefunctioning These "specific" amino acid sequences will be helpful for char- was also cloned into thepGEX vector. The glutathione S-transacterizing PTPases identified by the commonly used PCR ap- ferase protein alone (approximately26 kDa) was expressed and proach for cloning PTPases. used as an additional control. The expressed fusion proteins The GLEPPl Intracellular Domain Has PTPase Activity-To were purified by glutathione-Sepharose affinity chromatograconfirm that the PTPase domain did indeed have PTPase ac- phy (shown in Fig. 5) and tested for phosphatase and tyrosine

-

n1

Y

"

GLE:PP1

19959

P8E7

4c3

NR R

NR R

BB5

200 P

-

116 t

97 P FIG.6. Immunoabsorption studies. I,# pone/, Wcstern blots (of glomerular extract) incubatedwith GLEPPl mAb 4C3 (lane A ), a control mAb of the same IgGl isotype (BR5) (lane B), the 4C3 mAb absorbed with the GLEPP1-F'TPase fusion protein (lane C ) , and the 4C3 mAb absorbed withthe GLEPP1-ECD fusion protein (lane D).Note that the ECD fusion protein,but not the PTPase fusion protein, absorbed out the reactivity of the 4C3 mAb with the 200-kDa band on Western blot. Right panel, photomicrographs developed with the 4C3 mAb absorbed with the GLEPP1-PTPase domain (A) and the GLEPP1-ECD fusion protein ( B ) by indirect immunofluorescence, showing that the ECD fusion protein completely absorbed all reactivityof the 4C3 mAb with the tissue section. Magnification, x 500.

66

45

DYE FRONT-

NR

R

FIG.7. Western blot studies. Western blot of glomerular extract

phosphatase activity. The affinity-purified PTPase fusion pro- incubated with GLEPPlmAbs 4C3 and P8E7 anda control mAb of the tein (conforming to the expected molecular mass of about 60 same IgGl isotype(BB5) under reducing conditions( R )and non-reduckDa by SDS-polyacrylamide gel electrophoresis) had p-nitro- ing conditions(NR).Both P8E7 and 4C3mAbs detect a 235-kDa band non-reducing conditions. Under reducing conditions 4C3 detects phenyl phosphatase and also tyrosine phosphatase activities under a 205-kDa band strongly, whereas P8E7 recognizes this band weakly. when testedagainst two [32P]tyrosine-phosphorylatedsub- The 4C3 mAb also recognizes an approximately 300-kDa diffuse band strates (myelin basic protein and theRaytide peptide) (Fig. 5 ) . under both reducing and non-reducingconditions. The pH optimum for the fusion PTPase for the p-nitrophenyl phosphate substrate was found to be pH 5.5, whereas the pH optimum for the tyrosine-phosphorylatedprotein substrates was pH 7.0, as has been previously demonstrated for other PTPases (data not shown) (24, 37, 43). In other experiments, the PTPase-fusion protein showed no detectable serine/ threonine phosphatase activity when tested against 32P-substrate (casein labeled with [y-32P]ATPin the presence of the catalytic subunitof cyclic AMP-dependent proteinkinase) (data not shown). The purified glutathione S-transferase alone and the purified ECD fusion protein did not demonstrate p-nitrophenyl phosphatase or tyrosine phosphatase activities(Fig. 5 ) . This resultconfirms that the PTPase domain acts asa proteintyrosine phosphatase and not as a serinelthreonine phosphatase under the conditions tested. The Nucleotide Sequence Cloned Is GLEPPl as Defined by the 4C3 d b - T h e affinity-purified GLEPP1-ECD fusion protein (coded for by nucleotides 525-1,622 and shown in Fig. 4) was Western-blotted and probed with the4C3 mAb. A band was seen corresponding to theexpected molecular mass of the ECD fusion protein at about 70 kDa (data notshown). No band was FIG.8. Northern blot. RNA used for the blot (40 pg) was isolated seen on the same Western blot with the purified glutathione from normal rabbit renalcortex. The 28 and 18 S ribosomal bands are shown by arrows. The blot was probed with 32P-labeledGLEPPl cDNA. S-tranferase protein or the purified PTPase fusionprotein (data not shown). This result confirms that the fusion protein A single major GLEPPl transcript approximately6.5 kilobases in size was found in normal renal cortex. was recognized by the 4C3 mAb used to clone the cDNAs. The affinity-purified ECD fusion protein was used to immunize a guinea pig. Polyclonal antiserum produced by the guinea epitope structure as that designated GLEPPl on the basis of pig was tested by indirect immunofluorescence against rabbit 4C3 mAb. These resultsalso confirm that thenucleotide readrenal cortex. The polyclonal antiserum showed specific staining ing frameidentified as the open reading frame is correct and in of glomerular epithelialcells in anidentical pattern of immun- the right orientation. Characterization of the GLEPPl Protein-mAbs 4C3 and ofluorescence to that seen by the original GLEPPl mAbs 4C3 and P8E7 (Fig. 5, lower panels). Absorption of the 4C3 mAb P8E7 both recognized a 235-kDa protein on Western blot of used to obtain the GLEPPl cDNA clones from the rabbit glo- SDShrea-extracted rabbit glomeruli under non-reducing conmerular cDNAlibrary with theECD domain fusion protein,but ditions (Fig. 7). Under reducing conditions, a band of about 205 not with the PTPase fusion protein abolished detection of the kDa was recognized well by 4C3 and faintly by P8E7. This 205-kDa protein by Western blot (reducing conditions) and im- finding of a protein of apparently smaller size under reducing munofluorescent staining of the glomerulus (Fig. 6). We there- conditions raises the possibility of a heterodimer with a small fore conclude thatthe GLEPP1-ECD nucleotidesequence protein of about 20 kDa being disulfide-linked to the GLEPPl cloned (bp 525-1,622) codes for a protein that has the same protein. No direct evidence of a second chain has yet been distribution in the kidney, the samesize, and contains the sameidentified.

GLEPPl

19960

filtration surface through modified tight junctions (slit diaphragms) between foot processes (44)and to sustain the charge 400and size characteristics of the glomerular basement membrane by synthesis and directional secretion into the basementmem300brane of the specialized molecules, necessary for maintenance of the extraordinary filtration characteristics of the glomerulus bases (45).In addition, foot process structure is thoughtto be main200tained in part by the negative charge conferred by heavily sialated podocalyxin molecules present on the non-basement membrane aspect of each foot process (44, 46). In considering this complex mechanism, i t seems likely that local sensing systems (receptors)will be present on the glomerular epithelial cell which monitor foot process structure, the efficiency of glomerular filtration at any single point on the filter surface, and 100A BC D E F G H I J K L M N O P Q R S that regulateprotein secretion into the basement membrane a t FIG.9. RNase protection assay.Autoradiogram of a polyacrylam- the local level. Such regulatory events would most likely be ide gel used for RNase protection assay. Lane A, RNA size markers a t controlled in partby phosphorylation of intracellular proteins. 100,200,300,400 and 500 bases. Lane B,rabbit GAPDH riboprobe (296 bases) plus yeast RNA showing the size of the intact probe. Lane C , Therefore, we speculate that GLEPPlmay participate in this rabbit GAF'DH riboprobe plus yeastRNA incubated with RNaseshow- regulatory mechanism by acting as a receptor for information ing complete hydrolysis of the probe. Lane D, rabbit GLEPPI riboprobe related to foot process structure and/or function and transmit(448 bases) plus yeast RNAshowing the size oftheprobe. intact Lane E, ting this information into the cell via a dephosphorylation rabbit GLEPPI riboprobe plus yeast RNA incubated with RNaseshowevent similar to that described for CD45. ing complete hydrolysis of the probe. Lanes F-S, both GLEPPI and The ligand for GLEPPl is not known. However, it has been GAF'DH riboprobes preincubated with RNA preparations from rabbit glomerulus ( F ) ,renal cortex ( G ) ,placenta (H), heart (I),skeletal mus- suggested that the extracellular domains of other PTPases, cle ( J ) ,large intestine ( K ) ,small intestine ( L ) ,stomach ( M ) , eye (N), which likeGLEPPlhave fibronectintype 111-like repeats, spleen ( O ) ,skin ( P I , lung (Q), brain ( R ) ,and liver ( S ) , and then submight play a role in cell-cell adhesion (9). Recent studies supjected to incubation with RNase. Lanes F (glomeruli), G (renal cortex), and R (brain) showa protected fragmentof 393 bases corresponding to porting this concept show specific membrane FTPase-dependthe size of the GLEPPIsequence in the:'*P-labeled GLEPPI riboprobe. ent clumping of cells which is independentof the intracellular All lanes from F to S show a 178-base GAPDH protected fragment PTPase domain itself, implying that these FTPases interact corresponding to the size of the GAPDH sequence in the :'?P-labeled with themselves (homophilic) via their extracellular domains GAPDH riboprobe. (10,ll). Thus the ligand for GLEPPl could be another GLEPPl molecule on an adjacent cell. In normal glomeruli, GLEPPl is Assuming that the signalpeptide is cleaved off during proc- predominantly localized to the podocyte foot processes, which essing, the mature GLEPPlprotein (minus the signalpeptide) interdigitate extensively, thereby creatinga huge filtration suris 1,158 amino acids longand hasa calculated molecular mass face area between cells. Since the major anatomic change seen of 132 kDa. The apparent discrepancy in molecular weight is in the podocyte associated with glomerular dysfunction (leakprobablyaccounted for by extensive glycosylation sinceseage of albumin across the filter) is fusion (effacement) of foot quence analysis shows 15 potentialN-linked glycosylation sites processes with loss of the complex interdigitating arrangement, in the extracellulardomain. In some glomerular extract prepa- we speculate that the GLEPPlreceptor tyrosine phosphatase rations, a variable proportion of the protein appears as a higher could play a role in regulating the structure and function of the molecular mass form (approximately 300 kDa) under both re- podocyte foot processes via a homophilic interaction. ducing and non-reducing conditions, possibly representing a Acknowledgments-We are grateful to the following: Dirk Bjornstadt more glycosylated form of GLEPPl. and Rajan Pastoriza for technical support, Robin Kunkel and Bradley Size of the GLEPPl mRNA lFanscript and Distribution of Nelson (Departmentof Pathology, University of Michigan, Ann Arbor) GLEPPl mRNA and Proteinin the Rabbit-Northern blot for their help with the transmission EM studies, Jill Baney (University analysis usingRNA from renal cortex shows a single transcript of Michigan Multipurpose Arthritis Center)for performing fusions and by NationalInstitutes of Health approximately 6.5 kilobases in length (Fig. 8). RNase protec- hybridomaproduction(supported Grant P560AR20557), Joann Knepper for preparing the figures and tion analysis of RNA preparations from glomerulus, renal cor- text, and the Clinical Research Center (University of Michigan, Ann tex, liver, brain,lung,skin, eye, skeletal muscle, placenta, Arbor) for supplying data base search facilities (supported by National Institutes of Health Grant MOlRR00042). heart, spleen, stomach, small intestine, and large intestine showed transcript for GLEPPl present in glomeruli (high level) REFERENCES and renal cortex (lower level) (Fig. 9, lanes F and G).The 1 . Hebert, S. C., and Kritz, W. (1993) in Diseases offheKidney (Schrier, R. W., and RNase protection assay also showed the presence of GLEPPl Gottschalk, C. W., eds) pp. 1-63,Little, Brown and Co., Boston, MA mRNA sequence in the brain (Fig. 9, lane R). This result was 2. Sibley, R. K., Striegel, J., and Melvin, T. (1989)in Renal fafhology(Tischer, C. C., and Brenner,B. M., eds) pp. 1291-1311,Lippincott Co., Philadelphia, PA confirmed in a second independent experiment using brain 3. Nadasdy, T., Silva, E G., and Hogg, R. J. (1989)in Renal Pathology (Tischer, C. RNA from a second rabbit. Further studies will therefore be C., and Brenner, B. M., eds) pp. 330-389, Lippincott Co., Philadelphia, PA 4. Janeway, C.A. (1992)Annu. Reu. Immunol. 10,645-674 required to define the structure, distribution, and function of 5. Koretzky, G.A. (1993)FASEB J. 7,420-426 GLEPPl protein in brain. 6. Secrist, J. P., Karnitz, L., and Abraham, R. T. (1991)J. Biol. Chem. 266, Speculation on the Function of GLEPPl-We report here the 12135-12139 7. Bemdge, M. J. (1993)Nafurr 361,315-325 identification and characterization of a new member of the 8. Pingel, J. T., and Thomas, M. L. (1989)Cell 58, 1055-1065 membrane protein-tyrosine phosphatase(FTPase) family of 9. Fischer, E.H., Charbonneau,H., and Tonks, N. K. (1991) Science 253,401-453 proteins, which is restricted in distribution to the renal visceral 10. Gebbink, M.,Zondag, G., Wubbolts, R., Beijersbergen, R., van Etten, I., and Moolenaar, W. (1993) J. Biol. Chem. 268, 16101-16104 glomerular epithelialcell (podocyte) and probably also to brain. 11. Brady-Kalnay, S., Flint, A. J., and Tonks, N. K. (1993)J . Cell Biol. 122, The podocyte is anoctopus-like cell that has multiple foot proc961-972 Nephrol. 1, 1334-1342 esses which interdigitate with the foot processes of other podo- 12. Goyal, M., and Wiggins, R. (1991)J. Am. SOC. 13. Login, G. R., Schnitt, S. J., and Dvorak, A. M. (1987)Lab. Inuest. 57,585-591 cytes to cover the surface of the glomerular basement mem- 14. Wiggins, R. C., Glatfelter, A,, Kshirsagar, B., and Beak, T. (1987)Lab. Invest. brane. The function of the podocyte is to maintain a large 56,264-272 500-

GLEPPl 15. Downer, G., Phan, S.H., and Wiggins, R. C . (1988) J. Clin. Inuest. 82, 9981006 16. Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J., and Rutter, W. J. (1979) Biochemistry 18, 5294-5299 17. M e m t t , S. E., Killen, P. D., Phan, S.H., and Wiggins,R. C . (1990) Lab. Inuest. 63,762-769 18. Younp, R. D., and Davis, R. W. (1985) Genetic Engineering, - PO. .. 2 9 4 1 , Plenum Publishing Corp., New York 19. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular CZoning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor,

NY 20. Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc. Natl. Acad. Sci. U.S. A. 74,5463-5467 21. Altschul, S. F., Gish, W., Miller, W., Myers, E. W., and Lipman, D. J. (1990) J. Mol. Biol. 215, 403410 22. Smith, D. B., and Johnson, K. S.(1988) Gene (Amst.) 6 7 , 3 1 4 0 23. Guan, K. L., and Dixon, J. E. (1991) Anal. Biochem. 192,262-267 24. Tonks, N. K., Diltz, C. D., and Fischer, E. H. (1988) J. B i d . Chem. 263, 6731-6737 S.,Dixon, J. E. (1991) Nature 350, 359-362 25. Guan, K. L., Broyles, S. and 26. Streuli, M., Krueger, N. X., Tsai, A. Y. M., and Saigo, H. (1989) Proc. Natl. Acad. Sei. U. S. A. 86, 8698-8702 27. Schraven, B., Kirschgessner, H., Graber, B., Samstag, Y., and Mener, S. (1991) Eur J. Immunol. 21,2469-2477 28. Kozak, M. (1989) J . Cell Biol. 108, 229-241 29. Austin, B. M., and Ridd, D. H. (1981) Biochem. SOC.Syrnp. 46, 235-258 30. Walter, P., and Lingappa, V. R. (1986) Annu. Reu. Cell Biol. 2, 499-516 31. von Heijne, G. (1986) Nucleic Acids Res. 14, 4683-4690

19961

32. Pearson, W. R., and Lipman, D. J. (1988) Proc. Natl. Acad. Sei. U. S. A. 8 5 , 2444-2448 33. Boyd, D., and Beckwith, J. (1990) Cell 62, 1031-1033 34. Jackson, R. L., Busch, S. J., and Cardin,A. D. (1991)Physiol. Reu. 71,481-538 35. Trowbridge, I. S . (1991) J. Biol. Chem. 266,23517-23520 36. Kreuger, N. X., Streuli, M., and Saito, H. (1990) EMBO J . 9,3241-3252 37. Wang, Y., and Pallen, C . J. (1992) J. Biol. Chem. 267,16696-16702 38. Zanke, B., Suzuki,H., Kishihara, K., Mizzen, L., Minden, M., Pawson,A,, and Mak, T.W. (1992) E u r J. Immunol. 22,235-239 39. Levy, L. B., Canoll, P. D., Silvennoinen, O., Barnea, G., Morse, B., Honegger, A. M., Huang, J. T., Cannizzaro, L. A,, Park,S.H., Druck, T.,Huebner, K., Sap, J., Ehrlich, M., Musacchio, J. M., and Schlessinger,J. (1993) J. Bid. Chem. 268,10573-10581 40. Yang, X., Seow, K. T., Bahri, S.M., Oon, S. H., and Chia, W. (1991) Cell 67, 661-674 41. Tian, S.-S., Tsoulfas, P., and Zinn, K. (1991) Cell 67, 675-685 42. Arimura, V., Hinoda, V., Itah, F., Takekawa, M., Tsujisaki, M., Adachi, M., Imai, K., and Vachi, A. (1992) %mor Biol. 13, 180-186 43. Itoh, M., Streuli, M., Kreuger, N. X., and Saito, H. (1992) J . Biol. Chem. 267, 12356-12363 44. Kurihara, H.,Anderson, J . M., Keriaschki. D., and Farquhar,M. G. (1992)Am. J. Pothol. 141, 805-816 45. Stow, J. L., Soroka, C. J., MacKay, IC, Striker, L., Striker, G., and Farquhar,M. G. (1989)Am. J. Puthol. 135, 637-646 46. Kejaschki, D., Vernillo, A. T., and Farquhar, M. G. (1985)Am.J. Pathol. 118, 343-349 47. Kornblihtt, A. R., Umezawa, K., Vibe-Pederson, K., and Baralle, F. E. (1985) EMBO J. 4, 1755-1759