protein-tyrosine kinase from an Escherichia coli expression system overproducing the bacterial chaperones GroES and GroEL. (Src/signal transduction/heat ...
Proc. Natl. Acad. Sci. USA Vol. 92, pp. 1048-1052, February 1995
Purification and characterization of recombinant human p50csk protein-tyrosine kinase from an Escherichia coli expression system overproducing the bacterial chaperones GroES and GroEL (Src/signal transduction/heat shock proteins/phosphorylation)
KURT E. AMREIN*t, BELA TAKACSI, MARTIN STIEGERI, JULIETITE MoLNos*, NICHOLAS A. FLINT*, AND PAUL BURN* *Roche Research Center, Hoffmann-La Roche Inc., Nutley, NJ 07110-1199; and tF. Hoffmann-La Roche Ltd., CH-4002 Basel, Switzerland
Communicated by Christopher T Walsh, Harvard Medical School, Boston, MA, September 12, 1994
ABSTRACT An Escherichia coli expression system overproducing the bacterial chaperones GroES and GroEL was engineered and has been successfully used to produce large quantities of the recombinant human protein-tyrosine kinase pSOcsk. The co-overproduction of the two chaperones with p5Ocsk results in increased solubility of the kinase and allows purification of milligram amounts of active enzyme. Analysis of the purified protein by SDS/polyacrylamide gel electrophoresis reveals a single band with an apparent molecular mass of 50 kDa, indicating that recombinant human p5Ocsk has been purified to near homogeneity. The purified enzyme displays tyrosine kinase activity as measured by both autophosphorylation and phosphorylation of exogenous substrates. Biochemical properties, including in vitro substrate specificity and enzymatic characteristics of the enzyme, have been assessed and compared with those of members of the Src family of protein-tyrosine kinases. Results indicate that p5Ocsk and p56kk have different substrate specificities and that p59csk and p60csrc have similar kinetic parameters. The successful production and purification of an enzymatically active form of p5Ocsk will enable further characterization of this important kinase and allow clarification of its physiological role. In addition, the results suggest that the approach described may be generally applicable to improve the solubility of recombinant proteins which otherwise are produced in an insoluble form in E. coli.
Tyr527, if the two molecules are coexpressed in Saccharomyces cerevisae (5, 10), and (ii) phosphorylate and downregulate the catalytic activity of the Src family kinases p561ck, p60 c-src, p531Yn, and p59fYvn by phosphorylating their corresponding regulatory tyrosine residues in vitro (3, 11). These results are consistent with the notion that p5Ocsk is a key negative regulator of Src family kinase activity. The baculovirus system has been used for the expression and purification of small quantities of Src family kinases (12-14); however, attempts to overexpress and purify Src family kinases in large amounts have not been achieved. Moreover, this system has the disadvantage that the production of recombinant kinases often results in a nonhomogeneous population of enzyme molecules due to variable posttranslational modifications in the host cells (12). In addition, the purification of the kinases is complicated by the presence of similar endogenous PTKs. In contrast, expression of heterologous genes in bacterial cells has been successfully used for high-level production of numerous prokaryotic and eukaryotic proteins. Unfortunately, most PTKs produced in bacteria aggregate in inclusion bodies (15, 16). In vivo the cellular processes of protein folding and assembly are controlled by molecular mechanisms involving chaperones. Many chaperones are members of the heat shock protein (Hsp) family (17-20). Several Hsp subfamilies have been described, including the Hsp9O, Hsp7O, and Hsp6O families of stress proteins. The Hsp6O family member GroEL (GroEL in Escherichia coli, Hsp6O in mitochondria) together with its cofactor GroES (GroES in E. coli, HsplO in mitochondria) form high molecular weight complexes which hydrolyze ATP. GroES regulates the ATPase activity of the chaperonin complex and has been reported to be required for the full function of GroEL in the folding of some proteins. Chaperones can increase the yield of correctly folded, soluble, active proteins both in vitro and in vivo (21-26). We report here the successful overexpression and purification to near homogeneity of large amounts of an enzymatically active form of recombinant human p5Ocsk from an E. coli expression system which overproduces the bacterial chaperones GroES and GroEL. The biochemical and catalytic properties and substrate specificity of the purified enzyme are reported and compared with those of Src family PTKs p561ck and p60c-src.
Protein-tyrosine kinases (PTKs) of the Src family are important intermediates in coupling various cell surface receptors to intracellular signal transduction pathways (1). The tyrosine kinase activity of this kinase family is tightly regulated by tyrosine dephosphorylation and phosphorylation events (1, 2). The PTK p5Qcsk (C-terminal Src kinase) has been implicated in mediating some of these crucial regulatory phosphorylation events (3). p5ocsk was originally identified in neonatal rat brain (4) and subsequently cloned from human, rat, and chicken cDNA libraries (5-7). It is expressed ubiquitously as a 50-kDa protein and shows some structural similarities to p6OYrc (-45% amino acid identity). Like p6Osrc, it displays Src homology 2 and 3 domains. However, it lacks a myristoylation site at the N terminus and phosphorylation sites corresponding to the autophosphorylation site and the regulatory C-terminal tyrosine residues common to Src family kinases (3, 6). Targeted mutation (null mutation) of the p5Ocsk locus in mice results in neuronal tube defects and embryonic lethality (8). Cell lines established from these embryos show hypophosphorylation of Src family kinases at their corresponding C-terminal regulatory tyrosine residues and accordingly display an increased tyrosine kinase activity (9). Initial functional characterization demonstrated that p5Ocsk can (i) phosphorylate p6Oc-src on
MATERIALS AND METHODS Bacterial Strains and Construction of Plasmids pDS56/csk and pREP4groESL. Cloning, plasmid construction, and expression of recombinant proteins was performed in E. coli Abbreviations: GAP, GTPase-activating protein; HI-p56lck, heatinactivated p56kck; Hsp, heat shock protein; PTK, protein-tyrosine kinase. tTo whom reprint requests should be addressed.
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Biochemistry: Amrein et aL strains W3110 (ATCC 27325) and M15. Chromosomal DNA was isolated from E. coli W3110 as described (27). For the coproduction experiments bacteria were grown at 37°C in
Luria-Bertani (LB) medium supplemented with ampicillin (100 jig/ml) and kanamycin (25 ,ug/ml). To construct pDS56/ csk, primers Pl (5'-TCAGCAATACAGGCCGCCTGG-3') and P2 (5'-CCAGGCGGAAGCTTGCC-GTCACA-3') were used for PCR amplification of the coding region of csk from a plasmid encoding the full-length human p50csk (6), and the resulting PCR fragment was digested with HindIII, blunted with Klenow DNA polymerase, and cloned between the blunt-ended Nco I and HindIII sites of the expression vector pDS56/RBSII.Ncol (28). Plasmid pREP4groESL was constructed as described (22, 23). In brief, the operon containing the groES and groEL genes was amplified from E. coli W3110 genomic DNA by PCR using the specific primers P3 (5'AATATTCGTCCATTGCATG-3') and P4 (5'-AGGTGCAGGAAGCTTACATCATG-3'). The resulting PCR fragment was digested with HindIII and cloned between the filled-in Nco I and HindIII sites of the expression plasmid pDS56/RBSII.Ncol. The chaperones expression unit was then excised from this vector by digestion with Xho I and Nhe I and cloned between the Sal I and Nhe I sites of the repressor plasmid pREP4 (28), resulting in pREP4groESL. An isopropyl ,B-D-thiogalactopyranoside (IPTG)-regulatable promoter/ operator element and a synthetic ribosomal binding site (RBSII) are present in both plasmids. Both plasmids can be stably maintained in the same bacterial cell and are suited for the simultaneous overproduction of the chaperones GroES and GroEL with p5ocsk kinase. Co-Overproduction of GroES, GroEL, and p5OAk. Plasmids pREP4groESL and pDS56/csk were cotransfected into the bacterial strain W3110. The resulting strain was grown in a 100-liter fermenter, and the production of recombinant proteins was induced by the addition of 2 mM IPTG when the OD600 of the culture reached 0.9. Four hours after induction, the bacteria were harvested by centrifugation. Purification of Recombinant Human p5OCsk. One hundred grams (wet weight) of pelleted E. coli overproducing p5Ocsk together with GroES and GroEL was suspended to a final volume of 200 ml of breaking buffer [25 mM Hepes, pH 8.0/150 mM NaCl/5% (vol/vol) glycerol/5% (vol/vol) ethylene glycol/1 mM MgSO4/1 mM dithiothreitol/1 mM phenylmethanesulfonyl fluoride containing Benzonase (Merck) at 1 ,ul/ml and soybean trypsin inhibitor, antipain, pepstatin, leupeptin, and chymostatin each at 10 jig/ml]. E. coli cells were broken in a precooled French pressure cell (SLM Aminco, Urbana, IL) at 20,000 psi (1 psi = 6.89 kPa). EDTA was added (2 mM) and the lysed cell suspension was cleared by centrifugation at 23,000 X g for 30 min and then at 100,000 x g for 120 min. The cleared supernatant was adjusted with 4 M *(NH4)2SO4 to a final concentration of 1.4 M, incubated on ice for 30 min, and then centrifuged at 23,000 x g for 30 min. The soluble fraction, which contained most of the recombinant p5ocsk, was dialyzed against 2 liters of 4 M (NH4)2SO4.
Precipitated proteins, including p5Ocsk, were pelleted by centrifugation at 23,000 x g for 15 min and then dissolved in and dialyzed against buffer A (10 mM sodium phosphate, pH 8.0). The dialyzed proteins were cleared by centrifugation and by filtration through a 0.45-gm Millex-HV membrane (Millipore). The filtrate was applied onto a 50-ml Phospho-Ultrogel A4R column (Sepracor, Marlborough, MA) that was equilibrated and washed with buffer A. Bound proteins were eluted with 10 column volumes of a linear gradient of 0-0.5 M NaCl in buffer A at 1.0 ml/min. Fractions that contained p5Ocsk were pooled and dialyzed against 3.2 M (NH4)2SO4 overnight at 4°C. Precipitated proteins were pelleted by centrifugation at 10,000 x g for 15 min. The pellet was dissolved in and dialyzed against AQ buffer (20 mM Tris acetate, pH 8.0/10% glycerol/i mM EDTA) before it was filtered through a 0.45-,uM membrane as
Proc. Natl Acad ScL USA 92 (1995)
described above. The remaining proteins were then applied onto a 25-ml Q-Sepharose FF column (Pharmacia) equilibrated in AQ buffer. After the column was washed with AQ buffer, bound proteins were eluted with a linear gradient of 0-0.5 M NaCl in AQ buffer. Fractions containing p5Ocsk were
pooled and concentrated by dialysis/precipitation against 3.2 M (NH4)2SO4 as described above. Precipitated proteins were dissolved in 100 mM glycine/NaOH buffer (pH 8.5), filtered through a 0.45-,uM membrane, and applied onto a Sephacryl S-200 column (Pharmacia). Proteins were eluted with the glycine buffer, and fractions that contained pSocsk were pooled and concentrated by Diaflow filtration through a PM10 membrane (Amicon). The concentrated solution was cleared by centrifugation at 100,000 x g for 60 min and by filtration through a 0.22-,um membrane. Aliquots were kept frozen at -800C. In Vitro Kinase Reactions. In vitro kinase activity of p5Ocsk or p561ck was assayed in 20 ,ul of standard kinase reaction buffer (30 mM Hepes, pH 6.8/1 ,uM ATP/3 mM MnCl2 with a-casein at 0.3 mg/ml). In addition, each 20-,lI reaction mixture contained 0.5 ,uCi (18.5 kBq) of [y-32P]ATP and =20 ng of p5Ocsk or p561ck. Deviations from this protocol are stated in the text or figure legends. The following proteins were used as substrates: (i) dephosphorylated a-casein and 3-casein (Sigma), (ii) acid-denaturated enolase (29), (iii) heatinactivated (560C for 5 min) recombinant p56kck purified from an E. coli expression system (30), and (iv) Ras-GTPaseactivating protein (GAP) purified from a baculovirus-based expression system (31).
RESULTS Co-Overproduction of Recombinant Human p5Ocsk with the Bacterial Chaperones GroES and GroEL. With the plasmid pDS56/csk, p5Ocsk can be produced in high quantities in E. coli. However, most of the recombinant protein is found in the insoluble fraction (Fig. 1 Left). The plasmid pREP4groESL encodes the bacterial chaperones GroES and GroEL. pREP4groESL and pDS56/csk carry compatible origins of replication and different antibiotic-resistance genes (kanamycin and ampicillin, respectively) and therefore both can be stably maintained in the same bacterial cell. Co-overproduction of p5Ocsk with the two chaperones results in a dramatic E
:~ ~ ~ p0s
FIG. 1. SDS/PAGE analysis of p5Ocsk overproduced alone (Left) with GroES and GroEL (Right) in E. coli. Total cell lysates and soluble and insoluble protein fractions were prepared as described in Materials and Methods. Protein samples corresponding to 100 ,ul of culture were analyzed by SDS/PAGE and stained with Coomassie blue. Note the increased solubility of p5Ocsk when overproduced with the chaperones GroES and GroEL. Molecular weight (Mw) markers: phosphorylase b, Mr 94,000; albumin, Mr 67,000; ovalbumin, Mr 43,000; carbonic anhydrase, Mr 30,000; trypsin inhibitor, Mr 20,100; a-lactalbumin, Mr 14,400.
Biochemistry: Amrein et al.
Proc. NatL Acad ScL USA 92
increase in solubility of p5Ocsk. More than half of p5Ocsk is found in the soluble protein fraction (Fig. 1 Right). Purification of Recombinant Human p5Ocsk. Recombinant human p5ocsk was overproduced with the bacterial chaperones GroES and GroEL in E. coli and subsequently purified as described in Materials and Methods. After each purification step, pooled p50csk-containing fractions were analyzed by SDS/PAGE and visualized by Coomassie blue staining (Fig. 2). The final purification step yielded a single band at 50 kDa. Applying this purification protocol, we obtained 250 mg of highly purified, active recombinant human p5Ocsk from 100 g (wet weight) of pelleted E. coli starting material. In Vuro Substrate Specificity of Purified Recombinant Human p5Ocs. To investigate substrate specificity, purified p5ocsk (20 ng) was incubated with various substrate proteins [enolase, heat-inactivated p561ck (HI-p561ck), ci-casein, j3-casein, and Ras-GAP] (1 ,g) in 20 ,ul of kinase buffer containing [yy-32P]ATP. ci-Casein was the best substrate, followed by HI-p56lck, enolase, GAP, and 03-casein (Fig. 3A Left). To compare the substrate specificity of p50csk with that of the Src family PTK p561ck, we tested the same substrates with purified recombinant human p561Ck (30). The substrate specificity of p561ck was GAP > HI-p56lck > a-casein enolase > /3-casein (Fig. LA Right). Both kinases phosphorylated enolase and HI-p561ck with a similar efficiency. a-Casein, however, was a better substrate for p5ocsk than for p56kck. On the other hand, GAP, a very good substrate for p561ck (31), was poorly phosphorylated by p5ocsk. To investigate the substrate preferences of the kinases in more detail the two substrates HI-p56'ck and a-casein were phosphorylated by p5ocsk and by p561ck and then analyzed by two-dimensional tryptic phosphopeptide mapping (Fig. 3B). HI-p56kck was phosphorylated by p561ck mainly on one tryptic peptide (equivalent to the peptide containing Tyr394) but was phosphorylated by p5ocsk on two peptides (one peptide is equivalent to the peptide containing Tyr505, whereas the identity of the second one is not known). In contrast, a-casein was phosphorylated by both kinases on a single tryptic peptide. Two-dimensional peptide map analysis using a mixture of the labeled tryptic peptides of a-casein indicated that both kinases phosphorylated the same tyrosine residue, but with different efficiencies (data not shown). These findings indicate that p5ocsk and p561Ck have different substrate specificities.
500§ IC50 (NaCl), mM 7 15§ IC50 (CaCl2), mM 6.5t 7.0 pH optimum 4 2.4* Km (a-casein), ,uM 2 Turnover number, mol mol-1 min-1 15§ 46 2.4-4t1 Km (ATP), ILM Parameters determined for p5Ocsk are compared with the corresponding published values for p60c-srC purified from human platelets. To determine the kM of p50csk for a-casein, kinase assays were performed in standard kinase buffer with the following changes: p5ocsk, 5 ,ug/ml; ATP, 100 ,uM (10 ACi of [y-32P]ATP per 20 jl); and a-casein, 0.1-12 ,uM. To determine the Km of p50csk for ATP, kinase assays were performed in standard kinase buffer with the following changes: p50csk, 5 ,ug/ml; ATP, 1-150 ,uM (10 ,uCi of [y-32P]ATP per 20 ,l); and a-casein, 14 ,uM. *Calculated from figure 5 in ref. 32. tRef. 32. tRef. 33. §Value determined with enolase as substrate (32).
Proc. Natl. Acad. Sci USA 92 (1995)
Biochemistry: Amrein et aL
Mn25 80 lm 100
0--20 lI 00
10 15 0
6.67.0 7.4 0 10 20 300 5 10 15 mM mM pH
FIG. 5. Effect of assay conditions on p5Ocsk activity. p50cskcatalyzed phosphorylation of a-casein was assayed under standard conditions (see Materials and Methods) except that the concentration of divalent cations (Mg2+ or Mn2+), the pH or the salt concentration (NaCl or CaC12) of the kinase buffer was varied. In each sample a-casein phosphorylation was determined after SDS/PAGE separation by Phosphorlmager analysis. One hundred percent phosphorylation represents the maximal incorporation of phosphate in each series.
,uM and ATP varying from 1 to 150 ,uM. Lineweaver-Burk analysis revealed Km values of -4 ,uM for a-casein and -46 ,uM for ATP (Table 1).
DISCUSSION We report the engineering of an E. coli expression system which overproduces the chaperones GroES and GroEL and which subsequently allows the overproduction and purification of soluble, active recombinant human p5Ocsk in milligram quantities. Overproduction of recombinant human p5ocsk with the Hsp6O family members GroES and GroEL in E. coli resulted in increased yields of soluble enzyme compared with p5Ocsk expressed alone. Thus, the approach described provides a means to improve the solubility of recombinant human p5Ocsk to a level suited for purification studies. Overproduction of chaperones with other recombinant proteins in E. coli may be a generally applicable method to improve the solubility of recombinant proteins which otherwise are produced in an insoluble form in E. coli. Indeed preliminary experiments with Src family kinases (22) and trimethoprim-resistant type Si dihydrofolate reductase from Staphylococcus aureus (23) confirm this conclusion. Similar effects on the solubility of recombinant proteins were observed in experiments where p5ocsk and recombinant human Src family kinases were overproduced with bacterial chaperones of the Hsp7O family (22). At present it is not possible to draw general conclusions about the relative abilities of the two chaperone families Hsp6O and Hsp7O to promote soluble protein production, since both can increase the solubility of recombinant kinases. According to a model proposed by Hendrick and Hartl (34) nascent polypeptide chains associate with chaperones of the Hsp7O family as they emerge from the ribosome. After partial folding steps the proteins are then passed on to the chaperones of the Hsp6O family to attain their final structure. This suggests that the two chaperone families may act synergistically in the proper folding of proteins in vivo. Thus, an interesting possibility will be to examine whether co-overproduction of both chaperone families in the same cell might further increase the solubility of the respective proteins. The E. coli expression system overproducing the bacterial chaperones GroES and GroEL provided an ideal source for
sufficient amounts of soluble recombinant p5Ocsk. Initial purification studies showed that a simple, classical purification scheme including (NH4)2SO4 precipitation, cation-exchange chromatography, anion-exchange chromatography, and sizeexclusion chromatography was sufficient to obtain highly purified protein. After the first chromatographic purification step the two chaperones GroES and GroEL were already
Biochemistry: Amrein et aL
separated from p5Ocsk, suggesting that the enzyme does not form a stable complex with the chaperones under these conditions. The two additional purification steps resulted in a highly purified form of the enzyme which displayed intrinsic PTK activity in autophosphorylation and exogenous-substrate phosphorylation experiments. Since we purified recombinant human p5Ocsk to near homogeneity from E. coli, an organism from which no PTKs have been isolated (35), a contamination of our enzyme preparation with another PTK seems highly unlikely. Starting from 100 g of pelleted E. coli material, we were able to purify 250 mg of active enzyme. The substrate specificity of purified recombinant human p5Ocsk was assessed in vitro and compared with that of the purified recombinant human Src family kinase p56Ick. The two kinases have a distinct substrate specificity as demonstrated by comparing the efficiency of in vitro phosphorylation of a number of protein substrates-including enolase, HI-p561ck, a-casein, 13-casein and GAP-by both kinases. In cases where the two kinases phosphorylated the same substrate they phosphorylated either different sites within the molecule (HIp56lck) or the same site with different efficiencies (a-casein). Further, the tryptic phosphopeptide map and phospho amino acid analysis of a-casein indicate that this substrate is specifically phosphorylated on a single tyrosine residue. Since a-casein is commercially available, it seems to be an ideally suited substrate for kinetic studies of p5Ocsk. Previous studies have indicated that p50csk produced in eukaryotic cells does not undergo autophosphorylation (4, 7, 11). Our tryptic phosphopeptide mapping and phospho amino acid analysis of in vitro kinase reactions demonstrated, however, a significant phosphorylation on a single tyrosine residue of purified recombinant human p5Ocsk. Since contamination by another PTK seems rather unlikely, this phosphorylation is most likely due to autophosphorylation. Similarly it has been shown that a glutathione S-transferase-Csk fusion protein expressed in E. coli undergoes autophosphorylation (36). p5OCsk expressed in eukaryotic cells may not undergo autophosphorylation because it interacts with other cellular proteins which either prevent or reverse autophosphorylation. Alternatively, autophosphorylation of p5Ocsk may occur in eukaryotic cells only under specific conditions. The kinetics of autophosphorylation of purified recombinant human p50csk are reminiscent of the autophosphorylation reported for p6Oc-src. The reaction proceeds with a similar initial rate to approximately the same extent (Table 1). For most other kinetic parameters p5OCsk and p60c-src are similar, including the Km values and turnover rates for an exogenous substrate (Table 1). Moreover, both enzymes have a pH optimum around 7 and require the presence of a divalent cation (preferentially Mn2+) for optimal reaction conditions. In contrast the Km value of p5Ocsk for ATP (40 ,M) was found to be significantly higher than for p6oc-srC (2.2-4 ,uM) (32, 33) or for p561ck (10 ,M) (30). Since cellular ATP levels are in the millimolar range, these differences may not be relevant in vivo. In summary the development of the described E. coli expression system and the overproduction and purification of p5Ocsk will enable further characterization of this important kinase at the molecular level (37). In addition, three-dimensional structurefunction analysis should now be possible. We thank Marie-Francoise Girard for excellent technical assistance and Dr. Kari Alitalo for providing the cDNA clone of human csk. 1. Mustelin, T. & Burn, P. (1993) Trends Biochem. Sci. 18,215-220. 2. Cooper, J. A. (1990) in Peptides and Protein Phosphorylation, ed. Kemp, B. E. (CRC, Boca Raton, FL), pp. 85-113.
Proc. Natl. Acad Sci USA 92 (1995) 3. Okada, M., Nada, S., Yamanashi, Y., Yamamoto, T. & Nakagawa, H. (1991) J. Bio. Chem. 266, 24249-24252. 4. Okada, M. & Nakagawa, H. (1989) J. Biol. Chem. 264, 2088620893. 5. Nada, S., Okada, M., MacAuley, A., Cooper, J. A. & Nakagawa, H. (1991) Nature (London) 351, 69-72. 6. Partanen, J., Armstrong, E., Bergman, M., Makela, T. P., Hirvonen, H., Huebner, K. & Alitalo, K. (1991) Oncogene 6, 2013-2018. 7. Sabe, H., Knudsen, B., Okada, M., Nada, S., Nakagawa, H. & Hanafusa, H. (1992) Proc. Natl. Acad. Sci. USA 89, 2190-2194. 8. Imamoto, A. & Soriano, P. (1993) Cell 73, 1117-1124. 9. Nada, S., Yagi, T., Takeda, H., Tokunaga, T., Nakagawa, H., Ikawa, Y., Okada, M. & Aizawa, S. (1993) Cell 73, 1125-1135. 10. Superti Furga, G., Fumagalli, S., Koegl, M., Courtneidge, S. A. & Draetta, G. (1993) EMBO J. 12, 2625-2634. 11. Bergman, M., Mustelin, T., Oetken, C., Partanen, J., Flint, N. A., Amrein, K E., Autero, M., Burn, P. & Alitalo, K. (1992) EMBO J. 11, 2919-2924. 12. Ramer, S. E., Winkler, D. G., Carrera, A., Roberts, T. M. & Walsh, C. T. (1991) Proc. Natl. Acad. Sci. USA 88, 6254-6258. 13. Watts, J. D., Wilson, G. M., Ettenhadieh, E., Clark, L. I., Kubanek, C. A., Astell, C. R., Marth, J. D. & Aebersold, R. (1992) J. Biol. Chem. 267, 901-907. 14. Piwnica-Worms, H., Williams, N. G., Cheng, S. H. & Roberts, T. M. (1990) J. Virol. 64, 61-68. 15. Gilmer, T. M. & Erikson, R. L. (1981) Nature (London) 294, 771-773. 16. McGrath, J. P. & Levinson, A. D. (1982) Nature (London) 295, 423-425. 17. Craig, E. A., Gambill, B. D. & Nelson, R. J. (1993) Microbiol. Rev. 57, 402-414. 18. Ellis, J. (1987) Nature (London) 328, 378-379. 19. Hartl, F. U., Martin, J. & Neupert, W. (1992)Annu. Rev. Biophys. Biomol. Struct. 21, 293-322. 20. Gething, M.-J. & Sambrook, J. (1992) Nature (London) 355, 33-45. 21. Blum, P., Velligan, M., Lin, N. & Matin, A. (1992) BiolTechnology 10, 301-304. 22. Caspers, P., Stieger, M. & Burn, P. (1994) Cell. Mol. Biol. 40, 635-644. 23. Dale, G. E., Schonfeld, H.-J., Langen, H. & Stieger, M. (1994) Protein Eng. 7, 933-939. 24. Mendoza, J. A., Lorimer, G. H. & Horowitz, P. M. (1992) J. Biol. Chem. 267, 17631-17634. 25. Mizobata, T., Akiyama, Y., Ito, K., Yumoto, N. & Kawata, Y. (1992) J. Biol. Chem. 267, 17773-17779. 26. Schroder, H., Langer, T., Hartl, F. U. & Bukau, B. (1993) EMBO J. 12, 4137-4144. 27. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A. & Struhl, K. (1993) Current Protocols in Molecular Biology (Wiley, New York). 28. Stuber, D., Matile, H. & Garotta, G. (1990) Immunol. Methods 4, 121-152. 29. Cooper, J. A., Esch, F., Taylor, S. S. & Hunter, T. (1984) J. Biol. Chem. 259, 7835-7841. 30. Flint, N. A., Amrein, K. E., Jascur, T. & Burn, P. (1994) J. Cell. Biochem. 55, 389-397. 31. Amrein, K. E., Flint, N., Panholzer, B. & Burn, P. (1992) Proc. Natl. Acad. Sci. USA 89, 3343-3346. 32. Feder, D. & Bishop, J. M. (1990)J. Biol. Chem. 265, 8205-8211. 33. Reuter, C., Findik, D. & Presek, P. (1990) Eur. J. Biochem. 190, 343-350. 34. Hendrick, J. P. & Hartl, F.-U. (1993) Annu. Rev. Biochem. 62, 349-384. 35. Lindberg, R. A. & Pasquale, E. (1991) Methods Enzymol. 200, 557-577. 36. Bougeret, C., Rothhut, B., Jullien, P., Fischer, S. & Benarous, R. (1993) Oncogene 8, 1241-1247. 37. Cole, P. A., Burn, P., Takacs, B. & Walsh, C. T. (1994) J. Biol. Chem., in press.