Communication Vol. 267, No. 35, Issue of December 15, pp. 24921-24924,1992 THE JOURNAL OF BIOLOGICAL CHEMISTRY

0 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U. S. A.

Thrombospondin 1 and Thrombospondin 2 Are Expressed as Both Homo- and Heterotrimers”

(Received for publication, June 1, 1992, and in revised form, October 6,1992)

Karen M. O’Rourke, Carol D. Laherty, and Vishva M. Dixit3 From the Department of Pathology, University of Michigan Medical School, Ann Arbor, Michigan 48109

There exist two distinct thrombospondin molecules (designated TSPl and TSP2) which are encoded by separate genes. TSPl is a trimeric cell surface and extracellular matrix molecule. Sequence comparison reveals that the 2 cysteines involved in interchain disulfide linkage and trimer assembly in TSPl are conserved in TSPB (Laherty, C. D., O’Rourke, K., Wolf, F. W., Katz, R., Seldin, M. F., and Dixit, V. M. (1992) J. Biol. Chem. 267, 3274-3281). Swiss 3T3 fibroblasts express both TSPl and TSP2, and, there- fore, an important question is whether TSP in such cells is expressed as homotrimers or as heterotrimers. We find that Swiss 3T3 cells and epithelial cells trans- fected with TSP expression vectors express both homo- and heterotrimeric forms of TSP. In addition, homo- trimeric TSPB has a lower affinity for heparin than homotrimeric TSP1. Thus, the heparin affinity of TSP can be modulated by the expression of TSP as homo- or heterotrimers.

Thrombospondin (TSP)’ is a large (-450 kDa), trimeric, multidomain extracellular matrix molecule, which is inti- mately involved in cell proliferation, cell migration, cell at- tachment, neurite outgrowth, and blastocyst attachment and outgrowth (1-3) (reviewed in Refs. 4 and 5 ) . This myriad of activities had been attributed to a single TSP molecule. How- ever, the recent discovery of a second TSP gene product (designated TSP2) that is homologous to the previously stud- ied TSP (designated TSP1) has added an additional level of complexity (6). These two molecules are almost identical at the carboxyl terminus. In contrast, TSP2 is quite dissimilar from TSPl at its amino-terminal heparin binding domain, having only 28% homology at the amino acid level (7).

Importantly, the 2 cysteine residues (Cys-252, Cys-256) responsible for interchain disulfide linkage and trimer assem- bly are conserved in TSP2 (7). Additionally, cell lines like

* This work was supported by National Institutes of Health Grant CA58182. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ Established Investigator of the American Heart Association. To whom correspondence and reprint requests should be addressed Dept. of Pathology, University of Michigan Medical School, Box 0602, Ann Arbor, MI 48109. Tel.: 313-747-2921; Fax: 313-764-4308.

The abbreviations used are: TSP, thrombospondin; PCR, polym- erase chain reaction; bFGF, basic fibroblast growth factor; PAGE, polyacrylamide gel electrophoresis; CNBr, cyanogen bromide; CMV, cytomegalovirus.

Swiss 3T3 fibroblasts express both TSPl and TSPB (7). This raises the important question of whether the molecules are expressed as homotrimers or as heterotrimers. To address this, we have developed polyclonal antisera specific for TSPl or TSP2 and mammalian expression vectors that express TSPl or TSP2. Our results indicate that both homo- and heterotrimeric forms of TSP are expressed by Swiss 3T3 fibroblasts and transfected cells. Furthermore, despite having a heparin binding domain that is distinct from TSP1, homo- trimeric forms of TSPB are capable of binding heparin-Seph- arose, albeit with an affinity lower than that of homotrimeric TSP1.

MATERIALS AND METHODS

Preparation of Recombinant TSP Fusion Proteins and Polyclonat Antisera-To obtain proteins suitable for polyclonal antibody pro- duction, TSPl and TSPS fusion proteins were produced in the PET system (8), in which the expression of a gene 10 fusion protein is directed by the T7 promoter. Polymerase chain reaction (PCR) primers were designed to amplify a non-homologous region of the mouse TSPl and mouse TSPZ heparin binding domains, while intro- ducing BamHI restriction sites in both the 5‘ and 3’ ends of the double-stranded PCR products. Using mouse TSP cDNAs as tem- plate, the products of these PCR reactions were digested with BarnHI and ligated into BamHI-cut, phosphatased PET expression vectors. A 342-base pair fragment of TSPl corresponding to nucleotides 485- 827 of the TSPl cDNA was inserted into PET 3xb, and a 340-base pair fragment of TSPS representing nucleotides 478-818 of the TSPZ cDNA was similarly cloned into PET 3xa. The resultant plasmids were introduced into competent Escherichia coli DH5 cells, and the DNA isolated from individual colonies was sequenced in entirety to verify insert orientations, ensure fidelity of the PCR amplifications, and confirm that the correct open reading frames were maintained. Selected clones were then transformed into E. coli strain BL21(DE3)pLysS, which contains the T7 polymerase gene under control of the inducible lacUV5 promoter. Fusion protein expression and purification from the insoluble inclusion body fraction of bacte- rial cell lysates were performed as described in detail elsewhere (9, 10).

TSPl and TSP2 fusion proteins were also expressed in the PATH expression system, which produces an E. coli trpE fusion protein (11, 12). The same PCR fragments encoding TSPl and TSPZ heparin binding domain sequence were ligated into PATH10 and pATH2O plasmids, respectively. Both recombinant expression plasmids, TSP1- PATH and TSP2-PATH, were transformed into E. coli DH5 cells and subsequently sequenced to verify insert orientation and to confirm that the correct open reading frame was maintained. Transformed cells were grown and PATH fusion proteins were induced as previ- ously described (9). Polyclonal antiserum specifically recognizing TSPl or TSP2 was prepared by immunizing rabbits with purified recombinant TSP1-PET or TSP2-PET proteins as described else- where (9). Specificity of the immune sera was determined by Western blot analysis.

Western Blot Analysis-Bacterial lysates containing TSP1-PATH or TSP2-PATH fusion proteins were separated on 10% SDS-poly- acrylamide gels and transferred to nitrocellulose using a semi-dry electroblotting apparatus (LKB Multiphor, Pharmacia). The nitro- cellulose membranes were blocked in Tris-buffered saline, pH 7.4, containing 3% nonfat dry milk, incubated with immune serum diluted 1:1000, briefly washed, and then incubated with horseradish peroxi- dase-conjugated goat anti-rabbit IgG. Bound antibody was detected using the color development reagent 4-chloro-1-naphthol (Bio-Rad).

Cell Culture and Metabolic Labeling-Swiss 3T3 cells were grown in Dulbecco’s modified Eagle’s medium supplemented with 10% bo- vine calf serum. Cells were brought to quiescence by maintaining them in growth factor-deficient media for 6 days as previously de- scribed (7). Quiescent cells were then metabolically labeled and growth factor stimulated simultaneously. Briefly, quiescent cells were

24921

24922 HomolHeterotrimers of Thrombospondin rinsed twice and placed in media lacking cysteine and methionine. After 45 min, a mixture of [35S]cysteine and [35S]methionine (Tran9-label, ICN) at 200 pCi/ml and recombinant basic fibroblast growth factor, (bFGF; gift of Synergen, Boulder, CO) at 25 ng/ml was added for 6 h. Conditioned media and cell layer fractions were collected as previously described (13).

Immunoprecipitation and Immunodepletion Studies-Equivalent numbers of trichloroacetic acid-precipitable counts were immunopre- cipitated, as previously described (13), using anti-TSP1-PET or anti- TSP2-PET at a 150 dilution. Where indicated, anti-TSP1-PET and anti-TSP2-PET antisera were blocked by preincubation with purified TSPl-PET or TSPL-PET fusion protein coupled to CNBr-activated Sepharose, respectively. Immunoprecipitated products were resolved on 4% SDS-PAGE gels in the absence of reductant or on 7.5% SDS- PAGE gels in the presence of reductant. Immunodepletion studies were performed by consecutively immunoprecipitating metabolically labeled Swiss 3T3-conditioned media three times using either anti- TSP1-PET or anti-TSP2-PET serum to deplete the sample of the corresponding TSPl or TSP2 protein and then immunoprecipitating the final supernatant with antiserum specific for the other TSP.

Transfections-cDNA fragments containing the coding region for mouse TSPl or mouse TSP2 were subcloned into the expression vector pJDM (14) downstream of the CMV promoter. Exponentially growing human epithelial 293 cells were transfected essentially as previously described (X), with either vector alone, TSPl expression construct, TSPP expression construct, or the two together at an equimolar ratio. Thirty-six hours post-transfection the cells were metabolically labeled with [35S]methionine and [35S] cysteine for 1 h, and then chased for 4 h with media containing unlabeled amino acids. Cell layer and media were harvested separately.

Heparin-Sephurose Affinity Chromatography-Equivalent num- bers of trichloroacetic acid-precipitable counts of metabolically la- beled media from transfected 293 cells expressing either homotrimeric TSPl or homotrimeric TSP2 were applied to a 0.5-ml heparin- Sepharose column, washed with a buffer containing 0.02 M Tris, pH 7.6, and 0.15 M NaCl, and step-eluted in the same Tris buffer containing the indicated salt concentrations (Fig. 5). 1.5-ml fractions were collected and subjected to immunoprecipitation with either anti- TSPl or anti-TSP2 antibody at a 150 dilution.

RESULTS AND DISCUSSION

Anti-TSPl and Anti-TSP2 Polyclonal Antisera Specifically Recognize TSPl or TSP2, Respectively-Rabbit polyclonal antiserum directed against TSPl or TSP2 heparin binding domains was raised by immunizing rabbits with purified TSPl-PET or TSP2-PET fusion proteins containing heparin binding domain fragments fused to the gene 10 protein. TSPl and TSPZ fusion proteins produced in the PATH expression system which contain the same heparin binding domain frag- ments fused to the E. coli trpE protein were used to test for antibody specificity. Fig. 1 (panel A) shows an Amido Black- stained transfer of the bacterial lysates containing the TSP1- PATH and TSPZ-PATH fusion proteins used for determining antiserum specificity. The existence of antibodies specific for TSPl or TSPZ was established by Western blot analysis. As demonstrated in Fig. 1 (panel B), anti-TSP1-PET serum detected the TSP1-PATH fusion protein but not the TSPZ- PATH fusion protein. Similarly, anti-TSP2-PET serum rec- ognized the TSP2-PATH fusion protein but not the TSP1- PATH fusion protein (Fig. 1, panel C ) .

TSP2 Is Expressed by Swiss 3T3 Fibroblasts in a Trimeric Form-Since growth factor-stimulated Swiss 3T3 cells have been shown to express both mTSPl and mTSP2 mRNA (7), we used metabolically labeled, bFGF-stimulated Swiss 3T3 cells to assess whether the polyclonal antisera raised against TSP1-PET and TSPZ-PET fusion proteins could specifically immunoprecipitate TSPl and TSP2. As shown in Fig. 2 (lower panel, filled arrowheads), anti-TSPZ-PET serum specifically immunoprecipitates a protein of the appropriate molecular size (apparent mass of 180 kDa), and this immunoprecipita- tion is blocked by preincubating the antibody with TSP2, but not TSP1, fusion protein. Similarly, anti-TSP1-PET serum

5% $ 5 $ 5

m-* -'-.!I rr =m

d & n - P P

A !?!? B c !? C i !? 3:

- - 97 - a"' . 43'ln ..I

n - 2 '! Arnldo h11-TSPlpET AMI-TSp2PEl Black

FIG. 1. Specificity of anti-TSP1 and anti-TSP2 polyclonal antisera. The heparin binding domains of the two TSP molecules were expressed as recombinant fusion proteins using the PET expres- sion system. Rabbits were immunized with purified TSP1-PET or TSP2-PET. The presence of antibody specific for the TSP portion of the fusion protein was examined by generating a second recombinant fusion protein (using a PATH vector) which contained the same TSP amino acids, but a different bacterial backbone protein. Bacterial lysates containing TSPl-PATH and TSPP-PATH fusion proteins were resolved by SDS PAGE, transferred to nitrocellulose, and probed with either anti-TSP1 PET or anti-TSP2 PET immune serum. The Amido Black-stained transfer (panel A ) shows the prominent pres- ence of TSP2-PATH and TSPlpATH fusion proteins (open arrow- head) in lysates of bacteria induced to express the recombinant proteins. The anti-TSP1 antibody recognizes only TSPlpATH and the anti-TSP2 antibody recognizes only TSP2-PATH, confirming the specificity of the antibodies.

s.ph.ro5s m ! o e k l n g

Antibody WP2 TIpl WP1 Wl TSPl TJP?

.. . S a D -

c I NR

20011D- b

IR BBkD-

FIG. 2. Swiss 3T3 fibroblasts express both TSPl and TSPS. Swiss 3T3 cells were metabolically labeled with [3%]cysteine and [35S] methionine and the media subjected to immunoprecipitation with polyclonal antisera raised against the bacterial fusion proteins. As shown (lowerpanel, filled arrowhead), anti-TSP2 immunoprecipitates a protein of the appropriate molecular weight (which appears as a doublet), and this is blocked by preincubating the antibody with TSPZ fusion protein, but not TSPl fusion protein. Similarly, anti- TSPl antibody immunoprecipitates TSPl (also a doublet), and this immunoprecipitation is blocked by preincubating the antibody with TSPl fusion protein, but not with TSP2 fusion protein. This analysis, furthermore, confirms the specificity of the antibodies raised against the bacterial fusion proteins. In addition, when the immunoprecipi- tates were resolved in the absence of reductant, their molecular weight was consistent with the formation of trimers (upper panel, open arrowheads).

immunoprecipitates TSP1, but not TSPP; immunoprecipita- tion by anti-TSP1-PET serum was blocked by preincubating the antibody with TSP1, but not TSP2, fusion protein (lower panel, filled arrowheads). These data confirmed the specificity of the antibodies raised against TSPl and TSP2 recombinant fusion proteins. On close inspection, it became apparent that two closely migrating bands of slightly different apparent molecular weight were being immunoprecipitated by both antisera. These, as will be addressed later, represented either proteolytic forms of the same molecule or immunoprecipitated material composed of heterotrimers of TSPl and TSPZ, which differed in their apparent molecular size. When the immu- noprecipitation products were resolved in the absence of reductant, the molecular size (apparent mass of 450 kDa) of the complex precipitated by both anti-TSP1 and anti-TSPZ was consistent with the formation of trimers (upper panel, open arrowheads).

TSPl and TSP2 Exist as Both Homotrimers and Heterotri-

HomolHeterotrimers of Thrombospondin 24923

mers-To determine if TSPl and TSP2 can exist as homo- trimers and heterotrimers, immunodepletion studies were car- ried out on conditioned media from metabolically labeled, bFGF-induced Swiss 3T3 cells. Immunoprecipitations were performed three times consecutively, using anti-TSP1 or anti- TSPB antibodies to deplete the sample of either TSPl or TSP2. The remaining supernatant was then immunoprecipi- tated with the opposite antibody to determine if homotrimers of TSPB or TSPl were present in the sample. As shown in Fig. 3, initial immunoprecipitates were composed of two mo- lecular species, a slower migrating form (upper band) being more prominent with anti-TSP2 and a faster migrating form (lower band) being more prominent with anti-TSP1. In con- trast, following immunodepletion, subsequent immunoprecip- itation with the opposing antibody resulted in a single pre- dominant molecular species, corresponding to either the lower band for TSPl or the upper band for TSPB (Fig. 3). This result is consistent with the slower migrating (upper band) form being composed of TSPZ homotrimers and the faster migrating (lower band) being composed of TSPl homotri- mers. In addition, since both forms were precipitated during immunodepletion, heterotrimers of TSPl and TSPZ must also exist. Implicit in the interpretation of these results was the assumption that TSP2 had a slightly greater apparent molecular weight than TSP1, resulting in two migrating bands on SDS-PAGE analysis of heterotrimers. The calculated mo- lecular mass of both the TSPs is approximately 130 kDa, but on SDS-PAGE they migrate with an apparent molecular mass between 160 and 180 kDa. This anomalous migration may be due to the low isoelectric points of the TSPs and consequent reduction in the binding of SDS to the negatively charged proteins (5 ) , or due to post-translational modifications; TSP1, for instance, is a glycoprotein (5). Proteolysis is an unlikely explanation for the two forms, given that it is possible to precipitate the larger immunoreactive TSPZ (upper band) following depletion from the sample of the smaller TSPl (lower band).

Recombinant TSPl and TSP2 Can Be Expressed in Homo- and Neterotrimeric Forms-To confirm the results of the immunodepletion experiments, that TSP2 is larger (and con- sequently slower migrating on SDS-PAGE) than TSPl and that both molecules can exist as homo- and heterotrimers, the properties of recombinantly expressed TSPs were exam- ined. cDNAs encoding TSPl and TSP2 were cloned into the eukaryotic expression vector (pJDM) so that they were under

Anti Anti rAnti TSPPq TSpl rAnti TSP17 TSpp -,.e-. "V. v- ---

1 2 3 4 1 2 3 4

FIG. 3. Immunodepletion analysis demonstrates that TSPl and TSPB have different electrophoretic mobilities. Metabol- ically labeled media was subjected to three rounds of immunoprecip- itation with anti-TSP1 or anti-TSP2 immune serum to deplete the medium of the corresponding TSP protein (lanes 1-3). Depleted samples were then immunoprecipitated with the opposite antibody (lane 4 ) . Depletion immunoprecipitation with anti-TSP2 brought down a doublet, having a prominent slower migrating species (upper band); anti-TSP1 similarly immunoprecipitated a doublet, but the more rapidly migrating form (lower band) was more prominent. Immunoprecipitation of the TSPZ depleted sample with anti-TSP1 antibody resulted in a single molecular species (TSPl) corresponding to the lower of the two bands. Conversely, immunoprecipitation of the TSPl depleted sample with anti-TSP2 gives a single species (TSP2) that corresponds to the upper of the two bands.

Transfectant Vector TSPl TSPP TSPl + TSPP

Antibody TSPl. TSP;, TSPl TSPl T S R T S R T S R TSR Block B W B W B W

520kD-

Non- Reduced

Transfectant Vector TSPl TSPP TSP1 + TSPB BW Block Block Block

Antibody ~ T S P l , ~ S P ; , ~ T S P l _ T S P l ~ T S P ; , ~ T S R _ T S P 2 ~ T S P 1 , ~ T S ~ ~

2CQkD-

Reduced

FIG. 4. Transfection of 293 cells with TSP expression vec- tors results in the formation of TSP homo- and heterotrimers. 293 cells were transfected with the indicated expression constructs, metabolically labeled with [%]cysteine and methionine, and the harvested media immunoprecipitated with the indicated antibodies. Cells transfected with vector alone synthesized neither TSPl nor TSP2. The TSPl expression vector directed the synthesis of TSP1, as evidenced by immunoprecipitation by anti-TSP1, but not by blocked anti-TSP1 or anti-TSP2 antisera. Conversely, cells trans- fected with the TSPZ expression construct synthesized TSPZ which was immunoprecipitated by anti-TSP2, but not by blocked anti-TSP2 or anti-TSP1 immune sera. In addition, immunoprecipitated TSPB was slower migrating than TSP1. When the expression constructs were co-transfected, a doublet was immunoprecipitated by both anti- TSPl and anti-TSP2. The slower migrating form of the doublet corresponded to TSPB and the more rapidly migrating form to TSP1. In the absence of reductant, the immunoprecipitated material mi- grated with a molecular weight consistent with trimer formation. In aggregate, this indicates that TSPl and TSPZ form homotrimers when expressed individually, and, when co-expressed, one addi- tionally gets heterotrimer formation.

TSP 2

TSP 1

FIG. 5. TSP2 homotrimer binds to heparin-Sepharose with an affinity lower than TSPl homotrimer. Equivalent numbers of trichloroacetic acid-precipitable counts from the media of 293 cells transfected with either the TSPl or TSPZ expression constructs were applied to a heparin-Sepharose column. Following extensive washing, the column was eluted using the indicated salt step gradient and the collected fractions subjected to immunoprecipitation with anti-TSP1 or anti-TSP2 immune serum. Elution of homotrimeric TSPl began at the expected salt concentration of 0.5 M, but homotrimeric TSPB started eluting at the significantly lower salt strength of 0.35 M. Thus, TSPZ homotrimer binds to heparin with an affinity lower than that of TSPl homotrimer.

the transcriptional control of the powerful constitutive CMV promoter. 293 cells, a human epithelial cell line that was found to express neither TSPl nor TSP2,2 were transfected with the expression constructs either singly or in combination. Following metabolic labeling, media samples were subjected to immunoprecipitation with either anti-TSP1 or anti-TSP2 immune serum. Precipitates were resolved by SDS-PAGE in the absence or presence of reductant and visualized by auto- radiography. As shown in Fig. 4, neither of the TSPs was immunoprecipitated from 293 cells transfected with vector alone, confirming our previous observation that these cells do not synthesize endogenous TSP. In contrast, transfection with the TSP2 expression vector led to the synthesis of a

* K. M. O'Rourke and V. M. Dixit, unpublished data.

24924 HomolHeterotrimers of Thrombospondin

protein which was precipitable by anti-TSP2 antibody, but not by anti-TSP1 antibody (Fig. 4). In the absence of reduc- tant, TSP2 migrated with an apparent molecular weight con- sistent with its being a TSPS homotrimer. An analogous situation occurred with cells transfected with TSP1. Bxpres- sion vector in that they expressed a TSPl homotrimer. Im- portantly, when cells were transfected with both expression vectors, two molecular species were precipitable by both anti- TSPl and anti-TSP2. The larger form comigrated with TSPS and the smaller with TSP1, while in the absence of reductant, the precipitated material migrated in a trimeric form. These results confirmed that TSPS was larger than TSP1, and the difference in their apparent molecular weights was distin- guishable by SDS-PAGE. In addition, since in the co-trans- fection experiment either antibody could precipitate both forms, it follows that TSPl and TSPB when co-expressed are able to form heterotrimers. The results are not due to non- specific association of homotrimers of TSPl and TSP2, since it is possible to immunoprecipitate either without the other from a mixture of homotrimeric TSPl and TSP2 (data not shown). Depending upon the quantity of each TSP expressed, there is probably a dynamic balance between the homo- and heterotrimeric forms of the molecule. Cells such as Swiss 3T3 that express both forms can simply alter the spectrum of homo/hetero molecules by modulating the synthesis of either of the TSPs. If the two TSPs are under differential regulation by growth factors/cytokines, signals generated at the cell surface may preferentially alter the synthesis of one of the TSPs, and thereby influence the homo/heterotrimer equilib- rium.

TSP2 is a Heparin-binding Protein-The ability to synthe- size pure homotrimeric forms of TSP enabled us to determine if TSP2 was a bona fide heparin-binding protein. This is especially important since the sequence of TSP2 differs max- imally from TSPl in its NH2-terminal heparin binding do- main. The other potential heparin binding sequence having the consensus WSXW found in the Type 1 repeats is con- served between the two TSPs (7, 17). To determine heparin binding, metabolically labeled media containing either hom- otrimeric TSPl or homotrimeric TSPB was applied to a heparin-Sepharose column and step-eluted with a salt gra- dient. As shown in Fig. 5, homotrimeric TSPl bound to

heparin-Sepharose in the expected manner (16), being maxi- mally eluted by 0.5-0.6 mM NaC1. TSPB bound heparin- Sepharose in a similar manner except with a lower affinity; maximal elution began at 0.35 M NaCl instead of 0.5 M NaCl as was seen with TSP1. Taken together, these results dem- onstrate TSPB to be a heparin-binding protein with an affin- ity weaker than that of TSP1. Thus, the affinity of the expressed TSP for glycosaminoglycans may be varied by expressing the homo- or heterotrimeric form of the molecule. It remains to be determined if this will be of functional consequence. The TSPl heparin binding domain has a num- ber of important functions, including being a potent chemo- taxin for melanoma and endothelial cells, and the binding site for syndecan, a proteoglycan receptor for TSP (4). The avail- ability of the facile TSP expression system described in this paper will allow for an exploration and experimental verifi- cation of the shared and unique biological properties of TSP2 and its heparin binding domain.

Acknowledgments-We thank F. Wolf and J. Hoffman for valuable assistance and Dr. N. Perkins for helpful discussion and the gift of 293 cells.

REFERENCES 1. O’Shea, K. S., Liu, L.-H. J., and Dixit, V. M. (1991) Neuron 7, 1-7 2. Neugebauer, K. M., Emmett, C. J., Venstrom, K. A., and Reichardt, L. F.

3. O’ghea, K. S., Liu, L.-H. J., Kinnunen, L. H., and Dixit, V. M. (1990) J.

4. Frazier, W. A. (1991 Curr m. Cell Biol. 3, 792-799 5. Frazier, W. A. (19871 J. Cell%il. 106,625-632 6. Bornstein, P., ORourke, K., Wikstrom, K., Wolf, F. W., Katz, R., Li, P., 7. Laherty, C. D., O’Rourke, K., Wolf, F. W., Katz, R., Seldin, M. F., and

8. Studier, F. W., Rosenber , A H , Dunn, J. J., and Dubendorff, J. W. (1990)

9. Wolf F. W. Marks R. M. Sarma V B era M G., Katz R. W., Shows,

10. Marston, F. A. 0.’(1987) in DNA Cloning: A Practical Approach (Glover,

11. Dieckman, C. L., ak$’Tzigoloff, Z. (i985) J. Biol. Chem. 260., 1513-1520 12. Spindler, K. R., Rosser, D. S. E., and Berk, A. J. (1984) J. Vtrol. 49, 132-

13. Prochownik, E. V., ORourke, K., and Dixit, V. M. (1989) J. Cell Biol. 109 ,

1991) Neuron 6 , 345-358

Cell. Bcol. 111 , 2713-2723

and Dixit, V. M. (1991) J. Biol. Chem. 266,12821-12824

Dixit, V. M. (1992) J. Biol. Chem. 267,3274-3281

Methods Enzymol. l d , 60-89

T. B., and Dixit Q. M. (i992) J,’Bi&. &e&. 2 6 7 , 1317-1326

D. M., ed) Vol. 3 59 IRL Press Oxford

141

14.

15.

16.

17.

Cr -_ Wi ler M. Silvers 4. (i977j cell 1 I Dixlt V. M. Grant

Chkm. 269,10100-10105 Guo, N.-H., Krutzsch, H. C., Ne e, E., Vo el, T , Blake, D. A,, and Roberts,

D. D.. (1992) Proc. Natl. A c J S c i . U. .‘?A. 69,3040-3044