Mutational Analysis of Interactions between the Gag Precursor Proteins of Murine Leukemia Viruses

VIROLOGY

216, 418–424 (1996) 0078

ARTICLE NO.

SHORT COMMUNICATION Mutational Analysis of Interactions between the Gag Precursor Proteins of Murine Leukemia Viruses KIMONA A˚LIN* and STEPHEN P. GOFF*,†,1 *Departments of Biochemistry and Molecular Biophysics and Microbiology and †Howard Hughes Medical Institute, Columbia University College of Physicians and Surgeons, New York, New York 10032 Received August 10, 1995; accepted December 12, 1995 The yeast two-hybrid system was used to test for interactions among the Gag precursor proteins of three members of the murine leukemia virus family. These Gag proteins all interact with each other in all combinations, but do not interact with the distantly related HIV-1 Gag. A series of deletion mutants of Moloney MuLV were examined to determine the minimal interaction domain. Either one of two regions of Gag was independently sufficient to mediate homodimerization, suggesting multiple points of contact in the precursor. Analysis of a set of point mutations in the CA region revealed a complex pattern of effects on Gag–Gag interactions. q 1996 Academic Press, Inc.

The gag gene of all retroviruses encodes a polyprotein responsible for the assembly of the virion particle (for review, see (1)). Expression of the Gag precursor protein in cells is sufficient, even in the absence of any other viral gene expression, to mediate the assembly and release of noninfectious particles (2–4). Interactions between the Gag precursor proteins under the plasma membrane are thought to mediate this assembly process. Mutagenesis of gag genes from many different viruses has allowed the identification and localization of assembly domains, small portions required for the successful formation of a particle in vivo (5–7). These domains may represent sites of contact between Gag proteins or simply structures that are required for proper folding of the rest of the molecule. After virion assembly, the Gag precursor protein is cleaved by the viral protease (8, 9) into products found in the infectious particle; this maturation process is associated with profound rearrangements of the Gag proteins and condensation of the virion core (10, 11). During maturation it is likely that new protein–protein contacts are formed between the Gag products. The condensed form of the virion is required for early steps of the life cycle, perhaps virion entry or uncoating. The gag genes of the various retrovirus families are not highly conserved, and the Gag proteins of the more distantly related viruses are not thought to be able to form mixed particles. The murine leukemia viruses are very similar to one another, however, and indeed there is strong genetic evidence that MuLV Gag proteins can

form mixed particles (12, 13). The MuLVs can be classified on the basis of their ability to replicate in the face of a host restriction system encoded by the Fv-1 gene: N-tropic viruses can replicate on Fv-1n/n but are restricted in Fv-1b/b hosts, while B-tropic viruses replicate on Fv-1b/b but not Fv-1n/n hosts (14, 15). The tropism of the MuLVs has been mapped to the CA region of the gag gene (16, 17). Cells coinfected by N- and B-tropic MuLVs release mixed virions sensitive to both N- and B-type restriction, as expected for mixed virions containing both Gag proteins (12). Further, the Gag protein of one MuLV can incorporate mutant proteins of another MuLV into virion particles (13). Surprisingly, coassembly of mutant Gag proteins from murine and avian retroviruses has recently been demonstrated, suggesting that some structures present on these two Gags have been conserved sufficiently well for contacts to be made (18). We have previously made use of the two-hybrid system (19–21) to demonstrate and assay the specific formation of Gag homomultimers (22, 23). In this system, two Gag proteins are coexpressed in yeast as fusions with either a transcriptional activation domain or a DNA binding domain; interactions between the Gags bring the activation domain and the DNA binding domain together, thereby activating expression of an indicator gene positioned downstream of a binding site for the DNA-binding partner. Gag proteins from several different retroviruses can homodimerize in this assay, and in general, pairs of Gags from unrelated viruses do not form heterodimers (23). To test whether Gag proteins of different MuLV strains could heterodimerize in the yeast system, we generated yeast plasmids containing gag genes from Moloney

1 To whom correspondence and reprint requests should be addressed. Fax: (212) 305-8692.

0042-6822/96 $18.00

418

Copyright q 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

/ 6a0f$$7718

01-27-96 00:31:15

viras

AP-Virology

SHORT COMMUNICATION

MuLV, N-tropic MuLV, and B-tropic MuLV. The M-MuLV constructs were as described (22). The N- and B-tropic gag genes were obtained from molecular clones of the N-Cl-35 and B-Cl-11 viruses (24), the generous gifts of P. Jolicoeur. The gag sequences were excised from the cloned DNAs (pN20-7 and pB16-5; (17)) by cleavage with AatII plus HindIII and transferred to the pGEM72f(/) vector; the coding region was then amplified by PCR with two primers (5*-CGGGATCCATATGGGACAAACCGTAACA-3*, mapping just upstream of gag, and 5*-CGGTCGACCCAAGCCGACCTGTCACTTAGG-3*, in the protease region), treated with NdeI plus SalI, and cloned into pBluescript. The genes were then transferred into yeast plasmids in two steps, first moving a 3* BamHI–SalI fragment and then a contiguous 5* BamHI fragment, into vectors pGADNOT (25) to generate Gal4 activator–Gag fusions (pACgal4 –Gag), into pMA424 (26, 27) to generate Gal4 DNA binding domain–Gag fusions (pDBGal4 –Gag), and into pSH2-1 (28) to generate lexA DNA binding domain–Gag fusions (pDBlexA –Gag). The DNA sequences at the junctions were determined to ensure that the translational reading frame was maintained in the fusion genes. The plasmids were introduced into the appropriate yeast strains in various combinations to test for activation of the lacZ indicator gene. Plasmids encoding Gal4 DNA binding domain fusions were assayed in strain GGY1::171, containing an integrated Gal1–LacZ fusion (29); plasmids encoding lexA DNA binding domain fusions were assayed in strain CTY10-5d, containing a lacZ gene downstream from a lexA operator site (gift of R. Sternglanz, Stony Brook, NY). Yeasts were made competent for transformation by treatment with lithium acetate (30). Transformants were selected for leucine and histidine prototrophy and at least 100 colonies from each transformation were scored for expression of b-galactosidase after replica plating colonies onto nitrocellulose filters. The immobilized yeasts were permeabilized by freezing at 0707 for 10 min and then soaked in buffer containing 5-bromo-4-chloro-3-indolyl-b-D-galactosidase (Xgal) for 2 to 12 hr at 307. In general, yeasts expressing particular constructs were considered to have activity if all colonies on the plate turned blue within 2–12 hr. Transformants were scored negative if all colonies were white, and //0 if all colonies were pale blue. Coexpression of any one of the three DBLexA –Gag fusion proteins with any one of the three ACGal4 –Gag fusions resulted in strong induction of b-galactosidase (Table 1). No expression of b-galactosidase was detected in yeast transformed by any one of the plasmids alone or in combination with the parental Gal4 or lexA expression plasmids or in combination with plasmids containing the HIV-1 Gag. These results suggest that each of the MuLV Gag proteins was able to interact with itself to form ho-

/ 6a0f$$7718

01-27-96 00:31:15

viras

419 TABLE 1

Interactions between Various Pairs of Gag Precursor Proteins Detected by the Two-Hybrid System Gal4-AD

LexA-DB Moloney Pr65gag N-tropic Pr65gag B-tropic Pr65gag HIV-1 Pr55gag LexA-DB Alone

Moloney Pr65gag

N-tropic Pr65gag

B-tropic Pr65gag

HIV-1 Pr55gag

Gal4-AD Alone

/

/

/

0

0

/

/

/

0

0

/

/

/

0

0

0

0

0

/

0

0

0

0

0

0

modimers and with each of the other Gags to form heterodimers. None of the MuLV Gags, in any combination, could interact with the more distantly related HIV-1 Gag. These results suggest that the interactions scored here reflect the behavior of the proteins in virion assembly in vivo. The two-hybrid system provides a rapid assay for mutational studies defining the minimal regions essential for a given interaction. To identify interaction domains, we generated a panel of deletion mutants of M-MuLV gag and tested them in various combinations for their interactions (Table 2 and Fig. 1). To generate C-terminal truncations, DNA of plasmid pBS-Mgag containing the full-length coding region was digested at various positions within gag and at the HindIII site 3* to the insert; the DNA termini were blunted by treatment with the Klenow fragment of DNA polymerase I and religated. To generate N-terminal deletions, pBS-Mgag DNA was digested with NdeI at the 5* edge of the insert and at positions within gag and processed similarly. A series of internal deletions was also introduced by cleavage with various pairs of enzymes, joined so as to retain the proper translational reading frame. The mutants are named by the nucleotide positions of the endpoints of the deletions relative to the left edge of the 5* LTR of the provirus, and for the terminal deletions, a 3* or 5* prefix to designate the portion of the gag gene affected. A BamHI–SalI fragment or a BamHI fragment of each gag deletion mutant was used to replace the wild-type sequences in plasmids pDBLexA – Mgag, pDBGal4 –Mgag, and pACGal4 –Mgag. The expression of b-galactosidase was assessed in yeast after cotransformation with pairs of DNA binding and activation domain plasmids, either with both encoding the same mutant Gag protein or with one encoding the wild-type protein and one the mutant (Table 2). In all cases, each mutant construct was also used to transform

AP-Virology

420

SHORT COMMUNICATION TABLE 2 Interactions of Various Mutant Constructs of M-MuLV Gag

b-gal activitya

ACT-mutant vs DB-WT

ACT-WT vs DB-mutant

ACT-mutant vs DBmutant

Mutant

Gal4

LexA

Gal4

LexA

Gal4

LexA

3*D2355 3*D2282 3*D2151 3*D2033 3*D2009 3*D1885 5*D1304 5*D1709 5*D2009 5*D2183 5*D2282 5*D2504 D1485–2119 D2009–2151 D2033–2190 D1709–2355 D2009–2355 D2119–2355 5*D1485–3*D1966

0 0 0 //0 //0 0 / / 0 0 0 0 0 0 / 0 0 0 nt

0 0 0 nt nt 0 / 0 0 0 0 0 / 0 / 0 0 0 0

/ sa 0 0 0 sa / / 0 0 0 0 0 //0 / 0 sa 0 sa

/ sa 0 0 0 sa / //0 0 0 0 0 //0 sa / 0 sa 0 sa

//0 sa 0 / / sa / / 0 0 0 0 0 / nt 0 sa 0 sa

/ sa 0 / / sa / //0 0 / 0 0 / sa / 0 sa 0 sa

a Interactions were scored by testing for b-galactosidase present in yeast colonies expressing the indicated pairs of constructs. ACT, activation domain; DB, DNA binding domain; WT, wild type; /, strong X-gal staining; //0, weak staining; 0, no staining. nt, not tested. sa, mutant construct was self-activating as DB fusion and could not be scored for interaction.

yeast individually, and with the control expression vectors without inserts, to identify any mutants displaying false-positive self-activation. These controls are important because several deletion mutants proved capable of self-activation, presumably by exposing an artifactual transcriptional activation function. In addition, lysates from yeast expressing all the gag mutants which failed to activate b-galactosidase expression were analyzed by Western blot to test for the appearance of a stable fusion protein of the predicted size, using antiCA antibody (NCI # 79S-804) at 1:2000 dilution. These constructs were also tested for successful interaction with a panel of Gag-interacting fusion proteins (data not shown). Those mutants from which no protein could be detected, and unable to interact with other Gag-interacting proteins, were considered to be trivial negatives and also were not included in our analysis. Mutants with deletions at the 5* end of gag showed that MA, p12, and the N-terminal part of CA were dispensable for one Gag–Gag interaction; the C-terminal half of CA and NC was sufficient for dimerization of this region in the LexA system (Fig. 1, clone 5*D2183). Interactions

/ 6a0f$$7718

01-27-96 00:31:15

viras

between the mutants and the wild-type Gag, however, required a larger region, including all of CA and NC (clone 5*D1709). This region includes the Major Homology Region (MHR), a highly conserved sequence present in all retroviral Gag proteins (31–33), and a region termed assembly domain 3 (AD3) (1, 6). Conversely, mutants with deletions at the 3* end showed that the Cterminal part of CA and NC could be deleted, and that MA, p12, and the N-terminal part of CA could mediate homodimerization in the LexA system (Fig. 1, clone 3*D2009). Here, too, interactions between the mutants and the wild-type Gag required a larger region (clone 3*D2355). These results taken together show that there were at least two separable, nonoverlapping portions of the Gag precursor protein that could independently mediate multimerization. The various internal deletions showed behaviors consistent with their utilization of either one or the other of the two interaction domains. In the deletion series, removal of increasing amounts of sequence did not always yield simple behavior. In the C-terminal deletion series, for example, interaction between mutants was first lost with removal of some sequence (3*D2282, 3*D2151) but then was restored with removal of more sequence (3*D2033). Presumably the deletions of intermediate size prevented the proper folding of the protein required for a productive interaction. This behavior indicates that the failure of a given mutant to interact cannot be taken as strong evidence that no interaction domain is present; that domain might be buried or otherwise unable to interact. Because of this, the smallest, interaction-positive clones are the best indicators of the location of functional interaction domains. On these grounds, one interaction domain lies in sequences encoded upstream of the XhoI site (nt 2009); the other is encoded downstream of the AflII site (nt 2183). Several of the mutants were able to form homomeric interactions but could not interact with the full-length Gag precursor protein (e.g., clone 5*D2183). The conformational changes induced by the deletions in these mutants must somehow prevent access of the fragment to the full-length protein but not to another copy of itself. Thus, it is possible that the interaction domain being scored with these mutants could not take place in vivo between two intact Gag precursor proteins, but may only become possible during or after cleavage of the Gag into its mature products. Essentially nothing is yet known about the different contacts that may form in the immature vs the mature virion particle. To explore the possibilities of heteromeric interactions between different deletion mutants, pairs of partially overlapping clones were tested (Fig. 2). For these experiments, the lexA protein was used as the DNA binding domain. Constructs containing most of the CA domain in common proved able to interact, while those overlapping

AP-Virology

SHORT COMMUNICATION

421

FIG. 1. Homodimeric interactions of different fragments of Moloney MuLV Gag precursor protein. Top, schematic of the M-MuLV Gag protein, with positions of the mature cleavage products indicated. The restriction sites used to generate the deletions are indicated. The scale indicates the nucleotide position relative to the left edge of the 5* LTR of the proviral DNA. The gray boxes represent the two domains found to be independently sufficient for homodimerization; point mutations were generated in the region in between these domains as indicated. The bars represent the regions retained in each construct, and the thin lines indicate regions removed. The names of the deletions are indicated on the lines. Open bars indicate constructs that scored positive for homodimer interaction, and black bars indicate constructs that did not interact. If a construct was selfactivating as a DNA-binding fusion, it was scored for interaction as an activation domain fusion against the wild-type DB-Gag. All negative constructs were tested for production of stable protein or successful interaction with other Gag-binding constructs.

elsewhere did not. The smallest region of overlap sufficient to drive heterodimerization was seen in the pair of mutants 5*D1709 1 3*D2355. These results suggest that overlapping CA regions can provide sufficient contacts to mediate dimer formation. The results of reciprocal constructs were not always the same, and thus we presume that with some noninteracting pairs of fusion proteins, although they contain sufficient portions of Gag to interact in other settings, appropriate contacts cannot be made. To explore this overlap region in more detail, we also examined a number of mutants of M-MuLV with amino

/ 6a0f$$7718

01-27-96 00:31:15

viras

acid substitutions in CA (Table 3). A detailed description of the generation and characterization of these mutations will be presented elsewhere. Briefly, mutations were targeted to CA by bisulfite treatment of heteroduplex DNAs in which the CA portion of the gag gene was placed in single-stranded loops (34, 35). After mutagenesis, the DNA was introduced into bacteria, and individual clones were picked and characterized by DNA sequence analysis; four mutants with single amino acid substitutions, and six mutants with two substitutions were identified. DNA fragments containing the mutations were isolated and used to replace the corresponding wild-type frag-

AP-Virology

422

SHORT COMMUNICATION

FIG. 2. Heterodimeric interactions between various pairs of M-MuLV Gag constructs. Top, schematic of the M-MuLV Gag protein, with positions of the mature cleavage products indicated. The restriction sites used to generate the deletions are indicated. The scale is as in Fig. 1. Pairs of constructs were tested as activation domain fusions (ACT) and DNA binding domain fusions (DB) as indicated at the left. Open bars indicate pairs that scored positive for heterodimer formation, and black bars indicate pairs that did not interact.

ment of plasmids pDBLexA and pACGal4 . Finally, yeast strain CTY10-5d was cotransformed with various pairs of plasmids expressing mutant and wild-type fusion proteins, and colonies were scored for b-galactosidase exTABLE 3 Interactions of Substitution Mutants of M-MuLV Gag

b-gal activitya

Mutant

DB-mutant vs ACT-mutant

DB-mutant vs ACT-WT

DB-WT vs ACT-mutant

L122F A137V P150L P102S P102S/A129V G111D/V139I P146L/S149F H117Y/A129V P102L/T143I R119C/P146S

// // // 0 /// 0 // // // 0

/ // 0 // / / // // / 0

0 // 0 / 0 //0 / // 0 0

a

Mutants were tested for homodimeric and heterodimeric interactions with the wild-type Gag as indicated. DB, LexA DNA binding domain fusion; ACT, Gal4 activation domain fusion. WT, wild type. ///, significantly more rapid appearance of blue color than the wild type after X-gal staining. //, strong staining comparable to wild-type. /, weak staining. //0, barely detectable staining. 0, no staining.

/ 6a0f$$7718

01-27-96 00:31:15

viras

pression by X-gal staining (Table 3). There were no problems with self-activation by any of the constructs expressed singly. Six of the ten mutants examined (L122F, A137V, P102S/ A129V, P146L/S149F, H117Y/A129V, and P102L/T143I) were able to form strong homodimeric interactions, as well as heterodimeric interactions with wild-type Gag in at least one pairwise combination. Controls with irrelevant test hybrids were always negative. The failures of some of these six to interact with wild-type in one combination (DB-wild type vs AC-mutant) is curious, but presumably reflects problems with the geometry of the interaction or other requirements for activation for that particular pair of fusion proteins. As noted above, we interpret the positive signals as indicative of the true capacity of the proteins to bind and the negative signals as not necessarily significant because trivial problems with fusion protein folding and geometry can prevent interaction. Thus, the results suggest that these six mutants all retained at least one of their interaction domains and were able to fold and bind normally. Two mutants (P102S and G111D/V139I) were unable to homodimerize, but were able to interact with wild-type Gag. These mutations presumably block mutant–mutant contacts, but since the mutant fusion proteins do function with a wild-type partner, the mutant proteins must be intact, imported into the yeast nucleus, and functional for gene activation. These

AP-Virology

SHORT COMMUNICATION

two are thus good candidates for true interaction-negative proteins. The binding to wild-type protein suggests that these mutants might be rescued by coexpression of a wild-type Gag protein in a cell. Exactly this phenotype has been observed for several substitution mutants in the HIV-1 Gag CA region: though mutant Gag–Gag multimerization was blocked, mutant Gag–pol protein could be coassembled into particles made up of wild-type Gag (32). Mutant P150L was unique among the 10 in forming strong homodimeric interactions but no heterodimeric interactions with wild-type Gag. The activity of this mutant in homodimerizing suggests that the two mutant fusion proteins are functional; their failure to interact with wild-type in all combinations is intriguing, and suggests that very specific geometric constraints are not met. It may be significant that this mutation alters a proline in the center of the MHR, a short motif almost perfectly conserved among all retroviral Gag proteins. Mutations in this region of Rous virus have been shown to exhibit similarly complex phenotypes: deletions of the whole region are tolerated for virus assembly, but smaller mutations can severely affect or even block assembly (33 ). Finally, mutant R119C/P146S was inactive in all combinations; this mutant protein may not fold or translocate into the nucleus properly, and may be trivially defective. In summary, it is clear that this CA region is critical for proper folding and binding among the Gags, and that very complex phenotypes can be expected from alterations in this region. It will be important to identify what steps in the life cycle are blocked by these mutations as a clue to determining when these interactions are needed. In conclusion, we have shown that the Gag precursor proteins of three different murine leukemia viruses can interact with each other in all combinations, but not with the HIV-1 Gag, in the yeast two-hybrid system. These results are consistent with genetic evidence for the ability of the MuLVs to form mixed virion particles that are fully infectious (12, 13 ). Thus, at least some of the interactions seen in yeast are likely to reflect the interactions that are required for virion assembly. However, the analysis of mutations of the M-MuLV gag gene suggest that not just one domain is being monitored in the assay: rather, at least two nonoverlapping regions of the protein are independently sufficient to mediate dimerization. These two distinct regions may be needed during virion assembly, during virion maturation, or during virion entry and uncoating. Examination of the phenotypes of mutants specifically blocked in one or the other domain may help determine the role of these interactions in virus replication. The CA region seems to mediate strong dimerization. Furthermore, changes of a single amino acid in CA, and especially the MHR region of CA, can have

/ 6a0f$$7718

01-27-96 00:31:15

viras

423

profound consequences on binding in the yeast assay, with very complex and selective effects on the interaction between wild-type and mutant. The results predict that the genetics of this region will be equally complex, with complementation occurring between some and not other pairs of mutants. Genetic analysis of Rous sarcoma virus mutants with alterations in the MHR region of CA are consistent with these predictions. In particular, the MHR region can be completely deleted in the Rous Gag with minimal effect on particle assembly, but several point mutations in the MHR of Rous (33 ) and Mason – Pfizer monkey virus (31 ) can block assembly and release. One proposed model consistent with all the data suggests that the MHR acts as a crucial hinge between interacting domains of CA and thus can control the ability of adjacent regions to bind. The MuLV mutants in this study will provide useful starting materials for genetic experiments that may further our understanding of the functions of CA. ACKNOWLEDGMENTS This work was supported by Public Health Service Grant CA 30488 from the National Cancer Institute. S.P.G is an Investigator of the Howard Hughes Medical Institute.

REFERENCES 1. Wills, J. W., and Craven, R. C., AIDS 5, 639–654 (1991). 2. Shields, A., Witte, O. N., Rothenberg, E., and Baltimore, D., Cell 14, 601–609 (1978). 3. Delchambre, M., Gheysen, D., Thines, D., Thiriart, C., Jacobs, E., Verdin, E., Horth, M., Burny, A., and Bex, F., EMBO J. 8, 2653– 2660 (1989). 4. Shioda, T., and Shibuta, H., Virology 175, 139–148 (1992). 5. Hansen, M., Jelinek, L., Whiting, S., and Barklis, E., J. Virol. 64, 5306–5316 (1990). 6. Weldon, R. A., Jr., and Wills, J. W., J. Virol. 67, 5550–5561 (1993). 7. Bennett, R. P., Nelle, T. D., and Wills, J. W., J. Virol. 67, 6487–6498 (1993). 8. Crawford, S., and Goff, S. P., J. Virol. 53, 899–907 (1985). 9. Katoh, I., Yoshinaka, Y., Rein, A., Shibuya, M., Odaka, T., and Oroszlan, S., Virology 145, 280–292 (1985). 10. Yoshinaka, Y., and Luftig, R. B., Proc. Natl. Acad. Sci. U.S.A. 74, 3446–3450 (1977). 11. Arcement, L. J., Karshin, W. I., Naso, R. B., and Arlinghaus, R. B., Virology 81, 284–297 (1977). 12. Rein, A., Kashmiri, S. V. S., Bassin, R. H., Gerwin, B. I., and DuranTroise, G., Cell 7, 373–379 (1976). 13. Kashmiri, S. V. S., Rein, A., Bassin, R. H., Gerwin, B. I., and Gisselbrecht, S., J. Virol. 22, 626–633 (1977). 14. Pincus, T., Hartley, J. W., and Rowe, W. P., Virology 65, 333–342 (1975). 15. Jolicoeur, P., Curr. Top. Microbiol. Immunol. 86, 67–122 (1979). 16. Boone, L. R., Myer, F. E., Yang, D. M., Ou, C.-Y., Koh, C. K., Roberson, L. E., Tennant, R. W., and Yang, W. K., J. Virol. 48, 110–119 (1983). 17. DesGroseillers, L., and Jolicoeur, P., J. Virol. 48, 685–696 (1983). 18. Bennett, R. P., and Wills, J. W., Submitted for publication. 19. Chien, C.-T., Bartel, P. L., Sternglanz, R., and Fields, S., Proc. Natl. Acad. Sci. U.S.A. 88, 9578–9582 (1991). 20. Fields, S., and Sternglanz, R., Trends Genet. 10, 286–292 (1994).

AP-Virology

424

SHORT COMMUNICATION

21. Luban, J., and Goff, S. P., Curr. Opin. Biotechnol. 6, 59–64 (1995). 22. Luban, J., Alin, K. B., Bossolt, K. L., Humaran, T., and Goff, S. P., J. Virol. 66, 5157–5160 (1992). 23. Franke, E. K., Yuan, H. E. H., Bossolt, K. L., Goff, S. P., and Luban, J., J. Virol. 68, 5300–5305 (1994). 24. Rassart, E., DesGroseillers, L., and Jolicoeur, P., J. Virol. 39, 162– 171 (1981). 25. Luban, J., Bossolt, K. A., Franke, E. K., Kalpana, G. V., and Goff, S. P., Cell 73, 1067–1078 (1993). 26. Ma, J., and Ptashne, M., Cell 51, 113–119 (1987). 27. Fields, S., and Song, O., Nature 340, 245–246 (1989). 28. Hanes, S. D., and Brent, R., Cell 57, 1275–1283 (1989).

/ 6a0f$$7718

01-27-96 00:31:15

viras

29. Gill, G., and Ptashne, M., Cell 51, 121–126 (1987). 30. Rose, M. D., Winston, F., and Hieter, P. (1990) ‘‘Methods in Yeast Genetics: A Laboratory Course Manual.’’ Cold Spring Harbor Press, Cold Spring Harbor, New York. 31. Strambio-de-Castillia, C., and Hunter, E., J. Virol. 66, 7021–7032 (1992). 32. Mammano, F., Ohagen, A., Hoglund, S., and Gottlinger, H. G., J. Virol. 68, 4927–4936 (1994). 33. Craven, R. C., Leure-DuPree, A. E., Weldon, J., R.A., and Wills, J. W., J. Virol. 69, 4213–4227 (1995). 34. Peden, K. W. C., and Nathans, D., Proc. Natl. Acad. Sci. U.S.A. 79, 7214–7217 (1982). 35. Berkowitz, R. D., and Goff, S. P., Virology 196, 748–757 (1993).

AP-Virology