Analysis of the Retrovirus Capsid Interdomain Linker Region

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Available online at www.sciencedirect.com

Virology 308 (2003) 166Ð177 www.elsevier.com/locate/yviro

Analysis of the retrovirus capsid interdomain linker region

Brian Arvidson, Joshua Seeds, Mike Webb, Liam Finlay, and Eric Barklis*

Vollum Institute and Department of Microbiology, Oregon Health and Sciences University, 3181 SW Sam Jackson Park Road, Portland, OR 97201-3098, USA Received 27 August 2002; returned to author for revision 31 October 2002; accepted 11 November 2002

Abstract In structural studies, the retrovirus capsid interdomain linker region has been shown as a ﬂexible connector between the CA N-terminal domain and its C-terminal domain. To analyze the function of the linker region, we have examined the effects of three Moloney murine leukemia virus (M-MuLV) capsid linker mutations/variations in vivo, in the context of the full-length M-MuLV structural precursor protein (PrGag). Two mutations, A1SP and A5SP, respectively, inserted three and seven additional codons within the linker region to test the effects of increased linker lengths. The third variant, HIV/Mo, represented a chimeric HIV-1/M-MuLV PrGag protein, fused at the linker region. When expressed in cells, the three variants reduced the efﬁciency of virus particle assembly, with PrGag proteins and particles accumulating at the cellular plasma membranes. Although PrGag recognition of viral RNA was not impaired, the capsid linker variant particles were abnormal, with decreased stabilities, anomalous densities, and aberrant multiple lobed and tubular morphologies. Additionally, rather than crosslinking as PrGag dimers, particle-associated A1SP, A5SP, and HIV/Mo proteins showed an increased propensity to crosslink as trimers. Our results suggest that a wild-type retrovirus capsid linker region is required for the proper alignment of capsid protein domains. © 2003 Elsevier Science (USA). All rights reserved.

Introduction involved in virus assembly (Swanstrom and Wills, 1997). Processing of PrGag proteins by the viral protease (PR) Structural studies on the capsid (CA) protein of several releases the individual MA, CA, and NC Gag proteins and retroviruses have revealed that it is composed of an N- is a necessary step in the conversion of immature to mature, terminal domain (NTD) and a C-terminal domain (CTD), infectious retrovirus particles (Swanstrom and Wills, 1997). which includes the conserved major homology region Consequently, capsid proteins and their constituent domains (MHR; Berthet-Colominas et al., 1999; Campos-Olivas et may serve different roles in immature versus mature virus al., 2000; Gamble et al., 1996; Gitti et al., 1996; Jin et al., forms. 1999; Khorasanizadeh et al., 1999; Kingston et al., 2000; The roles performed by capsid proteins of retroviruses Momany et al., 1996). The CA NTD and CTD are separated such as HIV, Moloney murine leukemia virus (M-MuLV), by an apparently ﬂexible linker region (Fig. 1) that allows and the avian Rous sarcoma virus (RSV) have been ex- the two domains to orient independently in solution, but plored largely through molecular genetic investigations which serves an unknown function in vivo (Berthet-Colo- (Alin and Goff, 1996; Borsetti et al., 2001; Bowzard et al., minas et al., 1999; Campos-Olivas et al., 2000; Gamble et 2001; Craven et al., 1995; Mammano et al., 1994; McDer- al., 1996; Gitti et al., 1996; Jin et al., 1999; Khorasanizadeh mott et al., 1996, 2000a; Schwartzberg et al., 1984; Swan- et al., 1999; Kingston et al., 2000; Momany et al., 1996). Retrovirus capsid proteins initially are synthesized as parts strom and Wills, 1997; Tang et al., 2001; von Schwedler et of the Gag precursor (PrGag) proteins, which also include al., 1998; Wang and Barklis, 1993). Mutations in the coding matrix (MA), nucleocapsid (NC), and other polypeptides regions for the CA NTD and CTD of HIV, M-MuLV, and RSV have yielded defects in virus assembly, particle sta- bility and morphogenesis, and reverse transcription. Inter- * Corresponding author. Fax: ϩ1-503-494-6862. estingly, suppressor mutations for several lethal mutations E-mail address: [email protected] (E. Barklis). in the RSV MHR have been mapped to the capsid NTD

al., 1996; Jin et al., 1999; Khorasanizadeh et al., 1999; Kingston et al., 2000; Momany et al., 1996; Petroupoulos, 1997; Tang et al., 2002), as well as secondary structure predictions of CA NTDÐCTD juncture regions (Baldi et al., 1999; Jones, 1999; McGuffin, 2000; Raghava, 2000), interdomain linker regions of 6Ð13 residues have been identified (see Fig. 1). While all of the listed observed or predicted Fig. 1. Alignment of retrovirus capsid linker sequences. Capsid linker linkers have proline residues near their N-termini, sequences located just upstream of retrovirus major homology regions M-MuLV, HIV-1, and HIV-2 also share the sequence (S/ (MHR) were aligned for Moloney murine leukemia virus (M-MuuLV), HIV-1 and -2, human T cell leukemia virus (HTLV1), mouse mammary N)PT(N/S)(I/L) in this region. This shared sequence corre- tumor virus (MMTV), Mason Pfizer monkey virus (MPMV), and Rous sponds to HIV-1 gag codons 278Ð282 and M-MuLV gag sarcoma virus (RSV). Observed and predicted (Baldi et al., 1999; Berthet- codons 347Ð351. To probe the function of the M-MuLV Colominas et al., 1999; Campos-Olivas et al., 2000; Gamble et al., 1996, capsid linker region in the context of the wild-type (WT) 1997; Gitti et al., 1996; Jin et al., 1999; Jones et al., 1999; Khorasanizadeh et al., 1999; Kingston et al., 2000; McGuffin et al., 2000; Momany et al., PrGag protein, several linker region variants were con- 1996; Petroupoulos, 1997; Raghava, 2000) capsid N-terminal domain helix structed. These included constructs A1SP and A5SP, which 7 sequences are as shown, and intervening linkers of 6Ð13 residues are were designed to lengthen the linker region by three (ASP) aligned. As shown, the M-MuLV, HIV-1, and HIV-2 sequences share the and seven (AAAAASP) residues, respectively (Fig. 2). We sequence (S/N)PT(N/S)(I/L) in this region. also constructed the HIV/Mo PrGag chimera, which fuses the HIV-1 and M-MuLV linker region SPT sequences so as (Bowzard et al., 2001), indicating a functional communication between the two domains. Despite the genetic demonstration of a functional interaction between capsid N-terminal and C-terminal domains, the nature of the interaction and the potential role of the linker region in regulation of interdomain communication are unclear. We now report the analysis of M-MuLV capsid linker region mutants and variants designed to probe the function of these interdomain sequences. To probe linker length requirements, linker regions with added residues have been tested, and to examine linker effects on heterologous capsid domains, a chimeric HIV/M-MuLV Gag protein has been characterized. Our results indicate that pertur- bation of the linker region is lethal to M-MuLV, reducing particle release and causing an accumulation PrGag proteins at plasma membranes. Although mutant particles encapsidated M-MuLV genomic RNA, they had aberrant densities and stabilities and showed lobed, elongated, and tubular morphologies, reminiscent of late assembly domain gag mutants (Demirov et al. 2002a, 2002b; Garrus et al., 2001; Fig. 2. Recombinant viral constructs. The observed and predicted HIV-1 and M-MuLV capsid (CA) linker regions between the CA N-terminal Kikonyogo et al., 2001; Patnaik et al., 2000; Strack et al., domain (NTD) and C-terminal domain (CTD) are adjacent to the CA major 2000; VerPlank et al., 2001; Yuan et al., 1999, 2000). homology region (MHR). Predicted linker sequences corresponding to Addition of extra residues in the linker region also shifted HIV-1 (HXB2) gag codons 278Ð284 and M-MuLV gag codons 347Ð356 PrGag crosslink profiles from dimers (Hansen and Barklis, are shown with their SPT sequences aligned as in Fig. 1. To examine the 1995; McDermott et al., 1996, 2000b) to trimers. Our results functions of the linker region, several Gag protein expression vectors were constructed. The WT (pXM-WT) construct expresses the wild-type M- indicate that the linker region is important in the control of MuLV Gag protein from an adenovirus late promoter in a protease-minus Gag protein interactions. (PRϪ) setting. The insertion mutation constructs A1SP (pXM- ins348A1SP) and A5SP (pXM-ins348A5SP) encode proteins with three and seven residue insertions after gag codon 348 corresponding to ASP and AAAAASP, respectively. The chimeric HIV/Mo construct (pXM-HIV/ Results Mo) encodes a fusion protein translated from HIV-1 gag codons 1Ð278 and M-MuLV gag codons 348Ð538. The predicted HIV/Mo protein is expected Linker region effects on particle release to be approximately 52 kDa and composed of the HIV-1 matrix domain and capsid NTD, a chimeric linker region, and the M-MuLV CTD and complete nucleocapsid domain. Protease-plus versions of the M-MuLV In structures available for mammalian and avian retrovi- constructs are designated WT PRϩ, A1SP PRϩ, and A5SP PRϩ and are rus capsid proteins (Berthet-Colominas et al., 1999; Cam- encoded by pXM-gpe-type constructs which express M-MuLV gag, pol, pos-Olivas et al., 2000; Gamble et al., 1996, 1997; Gitti et and env genes. 168 B. Arvidson et al. / Virology 308 (2003) 166–177

tation of all virus and cell Gag bands from several independent transfections indicated release levels (relative to WT) of 49 Ϯ 8, 38 Ϯ 7, and 50 Ϯ 18 for the A1SP, A5SP, and HIV/Mo constructs, respectively. Although these release reductions were relatively small, the mutant release levels were close to what we have observed using the same methods for M-MuLV late (p12) domain assembly mutants (Yuan et al., 1999, 2000). Thus, we believe our experimen- tal system gives a conservative estimate as to the impairment of particle release for the linker region mutants. Conceivably, reduced Gag protein release levels could have been a consequence of reduced membrane binding for Gag protein variants. In this regard, subcellular fractionation (Hansen et al., 1993) experiments demonstrated normal membrane to cytosol distributions for the A1SP, A5SP, and HIV/Mo proteins (data not shown). However, because membrane fractions can be contaminated with membrane- Fig. 3. Virus particle release. Cell lysate (C) and pelleted media (V) free PrGag oligomers, a complementary localization assay samples from 293T cells transfected with the indicated constructs were was employed. As an alternative way to evaluate Gag sub- collected, fractionated by SDSÐPAGE, and subjected to immunodetection cellular distributions for the WT, A1SP, and A5SP proteins, for Gag proteins using an anti-M-MuLV CA (Hy187) antibody. Note that while major PrGag bands migrated at the predicted 65-kDa (WT, A1SP, indirect immunofluorescence localization studies were per- A5SP) and 52-kDa (HIV/Mo) sizes, subsidiary bands were observed at 50 formed using an anti-M-MuLV p12 primary antibody kDa for the A1SP and A5SP constructs and at 60 and 45 kDa for the (Chesebro et al., 1983; Hansen et al., 1993; epitope not HIV/Mo construct. Note also that the minor bands at 30 kDa in lanes B, D, present on HIV/Mo proteins). As observed previously F, and more faintly in H correspond to a cross-reactive protein present in (Barklis et al., 1997; Hansen et al., 1993), wild-type PrGag mock-transfected cells. Gag protein release levels, relative to WT release levels (ϭ100%), were determined from ratios of total released versus proteins stained a distinct punctate pattern, with a bright cellular Gag protein levels and were quantitated from densitometrically haze of fluorescence in a perinuclear region (Figs. 4A and scanned unsaturated blots from several independent experiments. Release B). Although A1SP and A5SP proteins also often showed levels were as follows: A1SP, 49 Ϯ 8% (N ϭ 3); A5SP, 38 Ϯ 7% (N ϭ 3); perinuclear staining (Figs. 4CÐF), the WT punctate appear- Ϯ ϭ HIV/Mo, 50 18% (N 2). ance was replaced by a patchy haze of fluorescence for the two variants (Figs. 4CÐF). In most cases, Gag-staining to encode a chimera composed of the HIV-1 matrix and A1SP and A5SP cells had increased staining at cell edges, NTD, plus the M-MuLV CTD and nucleocapsid (Fig. 2). suggesting an accumulation of Gag proteins at plasma mem- Expression of WT, A1SP, A5SP, and HIV/Mo constructs branes. Tabulation of staining patterns (Fig. 4 legend) indi- in transfected 293T cells yielded major cellular PrGag cated that peripheral staining was observed on 2 to 2.5 times bands of the expected sizes (Fig. 3, lanes B, D, F, H). In more A1SP and A5SP cells than WT cells. Although such particular, the major WT, A1SP, and A5SP PrGag proteins tabulation is, at least in part, subjective, these results are migrated at approximately 65 kDa (Fig. 3B, 3D and 3F), consistent with the notion that reduced extracellular levels while minor A1S1- and A5SP-specific bands—possibly rep- of capsid linker variants were caused by the impaired re- resenting breakdown products—migrated at about 50 kDa. lease of plasma membrane-associated Gag proteins. The chimeric HIV/Mo construct showed a predominant PrGag band at 52 kDa (Fig. 3, lane H), consistent with its Characterization of virus particles predicted size, but also showed a putative breakdown prod- uct at about 45 kDa and a variable band at about 60 kDa. To Although A1SP, A5SP, and HIV/Mo proteins appeared monitor the levels of extracellular, particle-associated Gag to assemble and release particles, the nature of these parti- proteins, media samples from transfected cells were centri- cles was uncertain. Because immature M-MuLV particles fuged through sucrose cushions, and Gag proteins in pel- have densities of 1.15Ð1.17 g/ml (Jones et al., 1990, Hansen leted material (Fig. 3, lanes A, C, E and G) were detected by et al., 1990), we initially subjected pelleted (Ն165S) media immunoblotting. Since we used matched media pellet and samples to 20Ð50% sucrose density gradient fractionation. cell samples, it was possible to compare particle release With WT particles, we observed the anticipated 1.15Ð1.17 levels of our different constructs by comparison of virus to g/ml density peak, along with some higher density material cell Gag protein levels. As illustrated, while A1SP, A5SP, (data not shown). However, the three Gag variants showed and HIV/Mo Gag proteins were detected in media pellet anomalous fractionation patterns: A1SP particles were samples, the virus:cell Gag ratios were reduced somewhat slightly more dense than WT; A5SP particles gave a 1.15Ð relative to WT, suggesting an impairment in particle assem- 1.17 g/ml peak, but showed a wider spread of particle bly or release for the variants. Total densitometric quanti- densities than WT; and HIV/Mo particles appeared to be B. Arvidson et al. / Virology 308 (2003) 166–177 169

examine Gag proteinÐRNA interactions. Thus, we examined the abilities of WT and capsid linker variant proteins to encapsidate viral RNAs. For this purpose, Gag expression constructs were cotransfected into cells with the MP10 construct (Barklis et al., 1986; Mougel et al., 1996). MP10 expresses full-length (genomic) and spliced recombinant M-MuLV RNAs, which can be measured simultaneously with an antisense RNase protection probe (Barklis et al., 1986; Mougel et al., 1996). As illustrated in Fig. 6, both genomic and spliced MP10 RNAs were detected in cells transfected with MP10 alone or with MP10 in the presence of Gag protein expression constructs (Fig. 6, lanes CÐG). As expected, the RNAs were not detected in pelleted media samples from cells transfected with MP10 alone (Fig. 6, lane K), whereas WT particles preferentially packaged full- length MP10 transcripts (Fig. 6, lane H). Interestingly, A1SP, A5SP, and HIV/Mo particles also encapsidated MP10 RNAs, with an apparent preference for genomic versus spliced transcripts (Fig. 6, lanes I, J and L). Densi- tometric quantitation of particle-associated MP10 RNAs yielded roughly equivalent genomic:spliced RNA ratios (Fig. 6 legend), indicating similar selectivities for the WT and variant proteins. After normalization for Gag protein content, genomic MP10 RNA levels in A1SP, A5SP, and HIV/Mo particles were somewhat reduced (54Ð84%; Fig. 6 legend) relative to WT particles, suggesting a slight reduc-

Fig. 4. Immunofluorescent localization of Gag proteins. Three days posttransfection with the indicated constructs, Cos7 cells were fixed, permeabilized, and processed for immunolocalization of PrGag proteins using a primary anti-M-MuLV p12 antibody (Hy548) and a rhodamine-conjugated secondary antibody. Gag proteins expressed from all three constructs showed evidence of a perinuclear localization, but WT proteins also showed a punctate localization pattern (A, B), while A1SP (C, D) and A5SP (E, F) proteins showed heterogeneous, sometimes patchy, localization patterns, often with increased fluorescence at cell peripheries. Percent- ages of fluorescent cells showing perinuclear, punctate, or peripheral staining were as follows: perinuclear staining, WTϭ72%, A1SPϭ58%, A5SPϭ67%; punctate staining, WTϭ69%, A1SPϭ32%, A5SPϭ19%; increased peripheral staining, WTϭ31%, A1SPϭ60%, A5SPϭ78%. The size bar at the bottom right corresponds to 40 ␮m. considerably more dense than normal M-MuLV virions (data not shown). Capsid linker variant particles also appeared to differ from WT in their sensitivities to non-ionic detergent (Triton X-100; TX100) treatment (Fig. 5). As observed previously (Hansen et al., 1993), Triton X-100 treatment of WT particles somewhat increased the amount of soluble Gag protein. On A1SP and HIV/Mo particles, the Fig. 5. Virus particle stabilities. Extracellular particles assembled by the detergent appeared to show a slightly increased ability to indicated Gag proteins were either untreated (Ϫ) or treated (ϩ) with 0.5% solubilize Gag proteins, while treatment of A5SP particles Triton X-100 and incubated on ice 30 min. Following incubations, samples resulted in the solubilization of greater than half of the total were centrifuged at 100,000 g for 30 min at 4¡C. Pellet (black) and Gag protein (Fig. 5). supernatant (super; gray) samples were collected and fractionated by SDSÐ PAGE, after which PrGag proteins were detected by immunoblotting and While density gradient and detergent treatment experi- quantitated densitometrically. Bar graphs depict percentages of total PrGag ments implied that A1SP, A5SP, and HIV/Mo Gag proteinÐ protein levels in supernatant or pellet fractions for each sedimentation protein contacts were compromised, it was of interest to experiment. 170 B. Arvidson et al. / Virology 308 (2003) 166–177

release (Table 1), implying that WT particle assembly at the plasma membrane was relatively rapid. In contrast with WT transfections, nearly half of the viral structures observed from A1SP, A5SP, and HIV/Mo transfected cells appeared to be in the act of assembly (Table 1). These observations, exemplified in Figs. 7C, E and G, suggest that plasma membrane release of capsid linker variant particles occurred slowly and may explain the cell surface accumulation of A1SP and A5SP Gag proteins observed in immunofluorescence localization studies (Fig. 4). In addition to showing increased numbers of assembling structures, the morphologies of capsid linker variant particles were abnormal. Many (44%) of the A1SP particles were classified as irregular, with multiple lobes, elongated shapes, or crescents of electron density (Table 1). However, a high percentage (42%) were elongated tubes that were lined with tracks of electron density (Fig. 7D), reminiscent of the appearance of gag late assembly domain mutants (Demirov et al. 2002a, 2002b; Garrus et al., 2001; Kikonyogo et al., 2001; Patnaik et al., 2000; Strack et al., Fig. 6. Viral RNA encapsidation. Triplicate 10-cm plates of 293T cells 2000; VerPlank et al., 2001; Yuan et al., 1999, 2000). were transfected with a construct expressing the packagable M-MuLV Similar results were obtained with A5SP, with the distinc- vector MP10 in the absence (lanes F and K) or the presence of Gag protein tion that higher percentages of tubes were seen (Table 1), expression constructs WT (lanes C and H), A1SP (lanes D and I), A5SP and the tubes were more heterogeneous in appearance (Figs. (lanes E and J), and HIV/Mo (lanes G and L). Three days posttransfection, total cellular RNA samples were extracted from cells, and pelleted extra- 7E and F). The HIV/Mo particles differed from their WT, cellular particle (virus) samples were collected for RNA isolation and Gag A1SP, and A5SP counterparts. While very few tube forms protein detection by immunoblotting. Cell (lanes CÐG) and viral (lanes were noted, abnormal elongated and lobed particles pre- HÐL) MP10 RNAs were detected by RNase protection with a 274Ðnt dominated (Table 1). Frequent particles with incomplete antisense probe (lane A), which protects a viral genomic RNA fragment of crescents of electron density were evident, as were particles 214 nt, and a spliced RNA fragment of 205 nt. In the absence of MP10 RNA, the antisense probe is digested during the protection protocol (lane with multiple lobes (Figs. 7G and H). Thus, the general B). Quantitation of MP10 genomic and spliced RNAs in virus samples appearance of capsid linker variant particles suggested that yielded the following genomic:spliced RNA ratios: WT, 1.19; A1SP, 1.58; improper interprotein contacts were made at the cell surface. A5SP, 1.60; HIV/Mo, 1.34. Levels of particle-associated genomic MP10 Using a cell culture system, it is difficult to assess exactly RNA, normalized for particle-associated Gag protein levels, were as fol- how PrGag contacts differ between WT, A1SP, A5SP, and lows: WT, 1.0; A1SP, 0.78; A5SP, 0.54%; HIV/Mo, 0.82%. HIV/Mo proteins. However, we previously have demon- tion in encapsidation efficiencies. However, our results indicate that pertubation of the capsid linker region and re- Table 1 placement of M-MuLV N-terminal PrGag sequences with Classification of particle morphologies those of HIV-1 did not have a major effect on encapsidation. Construct Assembling Normal Irregular Tube Since capsid linker variant proteins packaged viral RNAs WT 33 71 20 9 with near normal efficiencies and selectivities, it seemed A1SP 46 14 44 42 possible that particles assembled by these proteins might A5SP 47 6 43 51 have normal morphologies. This possibility was examined HIV/Mo 49 26 70 4 through electron microscopic (EM) analysis of transfected Note. Virus-like particles were imaged by EM after processing of 293T cells, and our observations are summarized in Table 1. As cells transfected with the indicated constructs. Percentages of assembling the table indicates, viral particles and structures were ob- particles and of particles with normal, irregular, and tube morphologies served at plasma membranes and in extracellular areas from were counted from a total of 70 (WT), 166 (A1SP), 122 (A5SP), and 387 (HIV/Mo) particles. Assembling particles were defined as particles with cells transfected with each of the constructs. With WT evident plasma membrane attachments, and both assembling and released transfections, the majority (71%) of observed particles were particles were counted in the calculation of normal, irregular, and tube classified as normal in appearance, meaning that they were percentages. Normal particles were defined operationally as 120- to 180- circular in projection, with 120- to 180-nm diameters and an nm-diameter particles with a characteristic circle of electron density adja- electron-dense ring adjacent to the particle surface: exam- cent to the particle surface. The irregular particle classsification included off-sized particles; long oval or indented particles; particles with multiple ples are shown in Figs. 7A and B. A noteworthy feature lobes; and particles with incomplete crescents of electron density. Tubes with WT transfections was that only a third of the particles were defined as having an axial ratio of Ͼ2, a width of 100Ð200 nm, and appeared to have been caught in the process of assembly or electron-dense tracks adjacent to tube surfaces. B. Arvidson et al. / Virology 308 (2003) 166–177 171

Fig. 7. Virus particle morphologies. Three days posttransfection with the indicated constructs, 293T cells were processed for EM as described under Materials and Methods. Images of cells transfected with WT (A and B), A1SP (C and D), A5SP (E and F), and HIV/Mo (G and H) constructs were collected electronically at original magniﬁcations of 2810Ð11,290ϫ and are shown with 200-nm bars. Observed particle morphologies are tabulated in Table 1.

strated that Gag protein contacts in virus particles can lane C), as we observed with A1SP. These crosslink profiles be probed with membrane-permeant, cysteine-specific indicate that capsid linker sequence alterations resulted in crosslinking agents such as bis-maleimidohexane (BMH; an increase in PrGag trimer versus dimer products. Hansen and Barklis, 1995; McDermott et al., 1996, 2000a). In particular, we have shown that PrGag domains in immature M-MuLV particles are preferentially crosslinked at nucleocapsid domain cysteines to yield dimers, implicating PrGag dimers as particle building blocks (Hansen and Bark- lis, 1995; McDermott et al., 2000b). With this backround, we decided to probe capsid linker variant particles by BMH crosslinking. To do so, particles were mock- or BMH- treated, and PrGag crosslink products were detected after electrophoresis and immunoblotting (Figs. 8 and 9). Our results with the WT construct (Fig. 8, lanes A and B; Fig. 9, lanes A and B) duplicated those observed previously in that PrGag dimers were the major crosslink forms. In contrast, A1SP PrGag proteins crosslinked to yield dimers migrating at about 126 kDa and trimers at about 162 kDa (Fig. 8C), while A5SP crosslink products were skewed even more toward trimer forms (Fig. 8E). Analysis of HIV/Mo Fig. 8. Crosslinking of spacer mutant Gag proteins. Pelleted extracellular crosslink profiles was more complicated, as a consequence particles from 293T cells transfected with the indicated constructs were mock-treated (Ϫ) or treated (ϩ) with the crosslinker bis-maleimidohexane of the greater number of subsidiary crosslink products (Fig. (BMH) prior to electrophoresis and immunodetection of Gag proteins. The 9, lanes C and D). Nevertheless, crosslink products were indicated molecular weight size designations were calculated by compari- clearly evident in both the dimer and trimer regions (Fig. 9, sion of mobilities with molecular weight size markers. 172 B. Arvidson et al. / Virology 308 (2003) 166–177

whereas three separate MP10 plus WT PRϩ cotransfections yielded infectious virus titers of 3275 Ϯ 2165 virus/ml, MP10 cotransfections with either A1SP PRϩ or A5SP PRϩ produced no infectious virions. In an effort to examine the replication defect of the A1SP PRϩ and A5SP PRϩ constructs, cell lysate and media pellet (virus) samples from cells transfected with wild-type and mutant constructs were subjected to Gag protein detection as described in Fig. 3. With WT PRϩ, A1SP PRϩ, and A5SP PRϩ cell samples, the expected 65-kDa PrGag bands were observed, along with bands at 50Ð55 kDa, which presumably correspond to partial processing or breakdown products (Fig. 10, lanes DÐF). A quite different result was obtained with virus samples (Fig. 10, lanes AÐC). Whereas major PrGag and mature 30-kDa capsid bands were observed with WT PRϩ virus samples (Fig. 10, lane A), particle-associated A1SP PRϩ and A5SP PRϩ PrGag protein levels were barely detected (Fig. 10, lanes B and C). Moreover, the A1SP PRϩ and A5SP PRϩ PrGag proteins were not processed to give mature CA. In comparison with release levels for the protease-minus A1SP and A5SP constructs (Fig. 3), A1SP PRϩ and A5SP PRϩ particle release was reduced dramatically. Indeed A1SP PRϩ and A5SP PRϩ release levels were 7 Ϯ 5 and 5 Ϯ 2% those of WT PRϩ (Fig. 10 legend), about 5- to 10-fold reduced relative to Fig. 9. Crosslinking of chimeric Gag proteins. Pelleted extracellular par- the protease-minus versions. Despite this, subcellular frac- ticles from 293T cells transfected with the indicated constructs were tionation proﬁles and immunoﬂuorescence localization pat- Ϫ ϩ mock-treated ( ) or treated ( ) with the crosslinker (BMH) prior to terns for the PRϪ and PRϩ capsid linker variants were electrophoresis and immunodetection of Gag proteins. The indicated molecular weight size designations correspond to the mobilities of molecular comparable (data not shown), suggesting that the same weight marker proteins. Approximate predicted mobilities for WT and assembly or release steps were affected. It is noteworthy HIV/Mo dimers and for HIV/Mo trimers are indicated by double and triple arrows, respectively.

Linker region effects in replication-competent constructs

The above studies on WT, A1SP, A5SP, and HIV/Mo constructs in a PRϪ context revealed significant defects for the linker region variants. Nevertheless, investigations were undertaken with A1SP and A5SP to determine their effects in replication-competent constructs. One approach was to express the mutations in M-MuLV proviral constructs (pMov9-1-A1SP and pMov9-1-A5SP), stably transfected into Rat1 cells. However, while stable A1SP and A5SP Gag protein-expressing cell clones were isolated, particle release was minimal, no infectious virions were isolated, and we were unable to obtain viable, viral revertants (data not shown). A second approach was to transfer the mutations to a transient transfection vector (pXMgpe; McDermott et al., Fig. 10. Expression and release of mature Gag proteins. Cell lysate (cells) 2000a), which permits expression of M-MuLV gag, pol, and pelleted media (virus) samples from 293T cells transfected with the and env genes and mobilization of MP10 proviruses. To test indicated constructs were collected, fractionated by SDSÐPAGE, and sub- linker region effects in this system, pXMgpe variants WT jected to immunodetection for Gag proteins using anti-M-MuLV CA PRϩ, A1SP PRϩ, and A5SP PRϩ were cotransfected with (Hy187) antibody. PrGag and mature capsid (CA) bands migrated accord- MP10 plasmids into 293T cells. The viruses so generated ing to their predicted respective 65- and 30-kDa sizes, and subsidiary bands at 50 kDa also were observed. Total Gag protein release levels, relative to were used to infect NIH 3T3 cells in a single round of WT PRϩ release levels (ϭ100%), were quantitated from densitometrically infection, and infection events were scored by selection for scanned blots from three independent experiments and were 7 Ϯ 5% for MP10 provirus neomycin gene expression. Not surprisingly, A1SP PRϩ and 5 Ϯ 2% for A5SP PRϩ. B. Arvidson et al. / Virology 308 (2003) 166–177 173 that similar observations have been made concerning the particles. Problems did not seem related to Gag proteinÐ severity of assembly/release defects in PRϩ versus PRϪ RNA interactions: mutant proteins packaged genomic viral virions for other retrovirus gag mutations (Huang et al., RNAs efficiently and selectively (Fig. 6), as might be ex- 1995). pected from studies which have implicated NC as the major encapsidation determinant (Berkowitz et al., 1995; D’Souza et al., 2001; Zhang and Barklis, 1995). However, A1SP, Discussion A5SP, and HIV/Mo particles showed anomalous densities and mobilities, while A1SP PRϩ and A5SP PRϩ particles Multiple retrovirus capsid protein structures have dem- showed no evidence of PrGag processing to mature CA onstrated that the CA NTD and CTD are separated by an (Fig. 10). Particle morphologies for linker variants also apparently flexible, interdomain linker (Berthet-Colominas were aberrant. In contrast to the 120- to 180-nm spheres et al., 1999; Campos-Olivas et al., 2000; Gamble et al., assembled by WT proteins, mutant structures included in- 1996, 1997; Gitti et al., 1996; Jin et al., 1999; Khorasaniza- dented particles, particles with crescents of electron density, deh et al., 1999; Kingston et al., 2000; Momany et al., 1996; ovals, multiple lobed structures, and tubes (Table 1, Fig. 7). Tang et al., 2002). However, the alignment of CA domains Although some of these structures could have arisen as a observed by EM in structures assembled from capsid pro- consequence of perturbed associations between Gag and teins in vitro (Li et al., 2000; Mayo et al., 2002; McDermott cellular proteins as with late domain mutants (Garrus et al., et al., 2000b) suggests that the linker region does not just 2001; Kikonyogo et al., 2001; Patnaik et al., 2000; Strack et hold the NTD and CTD in close proximity. To examine its al., 2000; VerPlank et al., 2001; Yuan et al., 1999, 2000), function, we have targetted the predicted M-MuLV capsid we favor an explanation that implicates GagÐGag interac- linker region. In this regard, although secondary structure tions. Credence for this interpretation is provided by the fact predictions and alignment algorithms have been used to that crosslinking of linker variants yielded higher propor- identify the M-MuLV CA linker, we cannot exclude the tions of trimer (versus dimer) PrGag products than WT possibility of NTD or CTD involvement. For our studies, proteins (Figs. 8 and 9). Currently, we do not know the we added either three (construct A1SP) or seven (construct mechanism by which linker variant crosslink profiles are A5SP) residues to the predicted linker region. Our rationale skewed. It may be that gag NC domains have a natural was to test whether the linker might tolerate residue addi- preference for trimer formation, but are constrained into a tions or whether it serves functions which would be more dimer configuration by wild-type capsid linker region se- sensitive to disruption. We also analyzed a chimeric HIV- quences. Alternatively, increased linker flexibility may 1/M-MuLV (HIV/Mo) Gag protein, fused at the linker re- make it so that NC domains are afforded increased oppor- gion, to examine whether the linker could mediate func- tunities to adopt unnatural trimer configurations. It will be tional interactions between heterologous CA domains. of interest to investigate these and other alternatives on Analysis of the A1SP, A5SP, and HIV/Mo variants fa- capsid linker variants in vitro. vors the interpretation that the wild-type linker sequence serves a function beyond that of a simple domain connector. This conclusion is supported by results which showed that Materials and methods linker variants assembled or released virus particles inefficiently (Figs. 3 and 10). Immunofluorescence localization Recombinant DNA constructs (Fig. 4) and EM (Table 1, Fig. 7) studies indicated that the PrGag variant proteins appeared to be delivered efficiently Plasmids WT (pXM-WT), A1SP (pXM-ins348A1SP), to the plasma membrane, but that assembly at the plasma A5SP (pXM-ins348A5SP), and HIV/Mo (pXM-HIV/Mo) membrane was impaired, consistent with a posttransport, for expression of Gag proteins in the absence of the M- oligomerization defect. The accumulation of linker variant MuLV protease were based on pXM-2453, which expresses Gag proteins and viral structures at cell surfaces is rem- the M-MuLV Gag protein from the adenovirus major late inscent of results with nucleocapsid domain mutants (Wang promoter and carries an SV40 origin of replication (Hansen et al., 1994; Zhang and Barklis, 1997), and with late assem- and Barklis, 1995; McDermott et al., 2000a). The WT bly domain mutants (Garrus et al., 2001; Kikonyogo et al., plasmid differs from pXM2453 as follows: the 5Ј EcoRI 2001; Patnaik et al., 2000; Strack et al., 2000; VerPlank et cloning site has been eliminated by a fill-in reaction (creat- al., 2001; Yuan et al., 1999, 2000). Another similarity with ing the sequence GAATTAATTC); gag codon 349 has a some late domain mutants (Huang et al., 1995) is that in the conservative ACC to ACT mutation; and the 3Ј EcoRI presence of pol and env gene products (in A1SP PRϩ and cloning site immediately follows the last M-MuLV gag A5SP PRϩ), the assembly defects appeared more severe, codon with the sequence GAC TAA TGA ATT C, where perhaps because the protease activity causes the dissolution GAC is the codon for the last aspartate residue of the of nascent buds (Wills et al., 1994). protein. A1SP is identical to WT, except that the sequence In addition to assembling virus particles inefficiently, the GCT AGC CCC (encoding the three residues ASP) was A1SP, A5SP, and HIV/Mo proteins assembled abnormal inserted between codons 348 and 349. A5SP is identical to 174 B. Arvidson et al. / Virology 308 (2003) 166–177

WT, except that the sequence GCG GCC GCA GCT GCT plates were supplemented with 8 ml of media. Cells infected AGC CCC (encoding the seven residues AAAAASP) was with MP10 viruses were selected at 2Ð3 days postinfection inserted between codons 348 and 349. For HIV/Mo, the by splitting them 1:10 to 1:20 into media containing 350 pXM-2453 gag EcoRI fragment was replaced with an ␮g/ml of G418 (Gibco). G418-resistant colonies were EcoRI fragment encoding the HIV/Mo fusion protein. In counted after 7Ð10 days in selection, and viral titers were this construct, the sequence from the 5Ј EcoRI site through calculated using the formula titer of virus stock ϭ (colonies) the first HIV-1 (HXB2; Ratner et al., 1987) gag codon was (virus dilution factor)(cell split ratio)/(cell proliferation be- G AAT TCG GCT AGA AGG AGA GAG ATG, and the tween infection and selection). sequence from the last M-MuLV gag codon to the 3Ј EcoRI site was identical to the sequence in the WT construct. The Protein detection, fractionation, and crosslinking HIV/Mo fusion sequence connected the HIV-1 and M- MuLV capsid spacer regions between HIV gag codon 278 For standard detection of cell- and virus particle-associ- and M-MuLV gag codon 348 with the juncture sequence ated Gag proteins, cell pellet and media samples collected CCC/ACT. from transfections were processed for immunoblotting. Cell Plasmids for expression of the WT and insertion variants samples were processed from 20% of the cell pellet material in the presence of M-MuLV pol and env gene products were by suspension in IPB [20 mM TrisÐhydrochloride (pH 7.5), WT PRϩ (pXMgpeWT), A1SP PRϩ (pXMgpeA1SP), and 150 mM NaCl, 1 mM ethylenediaminetetraacetic acid A5SP PRϩ (pXMgpeA5SP). These differ from the previ- (EDTA), 0.1% sodium dodecyl sulfate (SDS), 0.5% sodium ously described (Hansen and Barklis, 1995; McDermott et deoxycholate, 1.0% Triton X-100, 0.02% sodium azide], 5 al., 2000b) M-MuLV gag, pol, and env expression construct min incubation on ice, resuspension, pelleting of insoluble in the capsid spacer region sequence changes described material at 13,700 g for 15 min at 4¡C, and mixing of the above. The WT and insertion variants also were tested in the soluble material with 1 volume of 2 ϫ sample buffer [12.5 context of the full-length M-MuLV provirus, pMov9-1 mM TrisÐhydrochloride (pH 6.8), 2% SDS, 20% glycerol, (Schnieke et al., 1983). For testing gag mutants in single 0.25% bromophenol blue] plus 0.1 vol of ␤-mercaptoetha- round infection assays, the packageable retroviral vector nol (␤-Me). Cell-free media samples were centrifuged MP10 (Barklis et al., 1986) was employed. The MP10 through 2-ml 20% sucrose cushions in phosphate-buffered proviral construct includes the M-MuLV 5Ј long terminal saline [PBS; 9.5 mM sodium potassium phosphate (pH 7.4), repeat (LTR) to viral nt 566; the M-MuLV splice acceptor 137 mM NaCl, 2.7 mM KCl] at 197,300 g for 45 min at (viral nt 5409Ð5768); a neomycin cassette; and the 3Ј end of 4¡C, conditions sufficient to sediment complexes of 165S or M-MuLV, from viral nt 7197 through the 3Ј LTR (Barklis larger. Pellets were suspended in 100 ␮l of PBS, and 50-␮l et al., 1986): in mammalian cells it can express a neomycin aliquots of each suspension were mixed with equal volumes transcript from either full-length or spliced RNAs. For de- of 2 ϫ xample buffer plus 0.1 vol of ␤-Me. For SDSÐ tection of MP10 transcripts, probes were prepared from the polyacrylamide gel electrophoresis (SDSÐPAGE) on 7.5Ð previously described plasmid SPMLV (Barklis et al., 1986; 10% gels (Hansen et al., 1990; Jones et al., 1990), 75-␮l Mougel et al., 1996). sample mixes, representing 13% of cell samples and 33% of virus samples, were used. After electrophoresis and elec- Cell culture, transfections, and infections troblotting onto nitrocellulose filters, Gag proteins were detected by immunoblotting using Hy548 (anti-M-MuLV Cos7 and 293T cell lines were routinely passaged at p12) or Hy187 (anti-M-MuLV CA) primary antibodies

37¡Cin5%CO2 in Dulbecco’s modified Eagle’s media (Cheesebro et al., 1983), an alkaline-phosphatase conju- (DMEM) supplemented with 10 mM HEPES, pH 7.4, pen- gated goat antimouse IgG secondary antibody (Promega), icillin, and streptomycin, plus 10% fetal calf sera (FCS). and a BCIP (5-bromo-4-chloro-3-indolyl phosphate) plus NIH 3T3 and Rat1 cells were passaged similarly, substitut- NBT (nitroblue tetrazolium)-mediated color reaction (Jones ing normal calf sera (NCS) for the FCS. For transfections, et al., 1990) for visualization. Quantitation was accom- 10-cm plates of Cos7 or 293T cells at 50% confluence were plished using NIH Image 1.61 software on scanned, digi- transfected with 16Ð24 ␮g of plasmid DNA using the cal- tized images. cium phosphate precipitation method (Barklis et al., 1986; Gag proteins also were detected by immunoblotting after Hansen and Barklis, 1995; McDermott et al., 1996, 2000a; a variety of fractionation and crosslinking procedures. Sub- Wang and Barklis, 1993). For standard protein analysis cellular fractionation into postnuclear cytosol and mem- experiments, media and washed cell pellets were collected brane fractions followed established methods (Hansen et al., at 3 days post transfection and stored at Ϫ80¡C prior to use. 1993). Fractionation of extracellular material on 20Ð50% Infections followed established protocols (Barklis et al., sucrose density gradients was as described previously (Han- 1986; Mougel et al., 1996). Briefly, 10% confluent 10-cm sen et al., 1990; Jones et al., 1990; Wang and Barklis, 1993). plates of NIH 3T3 or Rat1 cells were infected for3hat For analysis of particle stabilities, 50-␮l virus samples were 37¡C with 2 ml of diluted, filtered (0.2Ð0.45 ␮m) viral diluted to 200 ␮l in PBS, incubated 30 min at 4¡Cinthe stocks in the presence of 8 ␮g/ml of polybrene, after which absence or the presence of 0.5% Triton X-100, centrifuged B. Arvidson et al. / Virology 308 (2003) 166Ð177 175

at 100,000 g for 30 min at 4¡C, and collected as pellet and Immunofluorescence and electron microscopy supernatant fractions. BMH crosslinking of Gag proteins in particle samples was performed by treatment of samples for Immunofluorescent detection of PrGag proteins using 60 min with 1 mM BMH (Pierce; Hansen and Barklis, 1995; Hy548 (Cheesbro et al., 1983) followed the method of McDermott et al., 1996, 2000b). Molecular weight esti- Hansent et al. (1990). Transfected Cos7 cells on coverslips mates for Gag proteins and crosslink products were calcu- were fixed 20 min in 3.7% formaldehyde in PBS, washed, lated by comparison of SDSÐPAGE mobilities with the permeabilized 10 min in 0.2% Triton X-100 in PBS, incu- mobilities of known size standards (Gibco, BioÐRad). bated 1 h with primary antibody, washed, incubated 1 h with rhodamine-conjugated goat antimouse IgG secondary anti- RNase protection assays body (Biosource), washed, and mounted onto slides in PBS containing 50% glycerol. Images were collected electroni- For quantitation of cellular and particle-associated MP10 cally on a Zeiss fluorescence microscope and were evalu- RNA levels, triplicate 10-cm plates of cells were transfected ated for the appearance of a fluorescent patch adjacent to the with either MP10 alone or MP10 plus WT, A1SP, A5SP, or nucleus (perinuclear staining), the presence of fluorescent HIV/Mo constructs. Three days posttransfection, pelleted spots against an otherwise black backround (punctate stain- particle samples were resuspended in 50 mM TrisÐHCl (pH ing), and an increased fluorescent signal at cell edges (pe- 7.5), 5 mM EDTA, 0.1% SDS and divided into aliquots for ripheral staining). Gag protein quantitation by immunoblotting (see above) For EM, media were carefully removed from transfected and for viral RNA isolation. Viral RNA was isolated by 293T cells and were replaced with fixative [100 mM sodium phenolÐchloroform and chloroform extraction, followed by cacodylate (pH 7.2), 2.5% glutaraldehyde, 1.6% parafor- ethanol precipitation (Mougel et al., 1996; Zhang and Bark- maldehyde, 0.064% picric acid, 0.1% ruthenium red] for 30 lis, 1995, 1997), suspension in RNase-free water, and stor- min. After removal of fixative, cells were gently washed in age at Ϫ80¡C. Total cellular RNA samples were isolated PBS, scraped into PBS, and pelleted in Eppendorf tubes. from guanidine thiocynate-treated cells by cesium chloride Cell pellets were postfixed1hin1%osmium tetroxide plus density gradient centrifugation (Wang et al., 1993), quanti- 0.8% potassium ferricyanide in 100 mM sodium cacodylate, tated via absorbance readings at 260 nm, and stored at pH 7.2. After thorough rinsing in water, cells were Ϫ80¡C in RNase-free water. prestained in 4% uranyl acetate 1 h, thoroughly rinsed, The 274-nt antisense probe for RNase protection assays dehydrated, infiltrated overnight in 1:1 acetone:Epon 812, protects a 214-nt fragment from full-length MP10 RNAs infiltrated 1 h with 100% Epon 812 resin, and embedded in and a 205-nt fragment from spliced MP10 RNAs (Barklis et the resin. After polymerization, 60- to 80-nm thin sections al., 1986). It was prepared using [␣-32P] GTP (Amersham) were cut on a Reichert ultramicrotome, stained 5 min in lead in SP6 transcription reactions from EcoRI-linearized citrate, rinsed, poststained 30 min in uranyl acetate, rinsed, SPMLV templates and purified by denaturing polyacryl- and dried. EM was performed at 60 kV on a Philips CM120/ amide gel electrophoresis (Barklis et al., 1986; Mougel et Biotwin equipped with a 1024ϫ1024 Gatan multiscan al., 1996). Overnight 42¡C hybridization reactions em- CCD, and images were collected at original magnifications 5 ployed 1Ð1.5 ϫ 10 cpm probe plus 20 ␮g of cellular RNA, of 2810Ð11,290ϫ. For tabulation of particle morphologies, viral RNA (one third of each preparation) supplemented assembling particles were defined as particles with evident ␮ ␮ with 10 g of yeast tRNA, or just 10 g of yeast tRNA as plasma membrane attachments; normal particles were de- a control (Barklis et al., 1986; Mougel et al., 1996). Unpro- fined as 120- to 180-nm-diameter particles with a circle of tected single-stranded RNAs were digested 30 min at 30¡C electron density adjacent to the particle surface; the irregu- ␮ ␮ with 40 g/ml of RNaseA plus 2 g/ml of RNaseT1, after lar particle classification included off-sized particles, long ␮ which samples were treated with 150 g/ml of proteinase K oval particles, lobed particles, and particles with incomplete (15 min, 37¡C), cleaned by phenolÐchloroform extraction electron-dense rings; while tubes were defined as having an and ethanol precipitation, and fractionated on sequencing axial ratio of Ͼ2, a width of 100Ð200 nm, and electron- gels (Barklis et al., 1986; Mougel et al., 1996; Wang et al., dense tracks adjacent to tube surfaces. 1993; Zhang and Barklis 1995, 1997). Protected probe frag- ments were visualized by autoradiography and were quantitated using NIH Image 1.61 software after scanning and digitization. To evaluate encapsidation specificities, ratios Acknowledgments of full-length to spliced MP10 RNA levels in virus particle samples were calculated. For evaluation of encapsidation We are thankful for the help and advice of Ayna Al- efficiencies, particle-associated unspliced MP10 RNA lev- fadhli, Zac Love, Keith Mayo, Jason McDermott, and Eric els were normalized for Gag protein content, and normal- Steel. This research was supported by Grants GM52914-07 ized values were compared with those of the WT construct and GM60170-03 from the National Institute of General (ϭ1.0). Medical Sciences. 176 B. Arvidson et al. / Virology 308 (2003) 166Ð177

References Hansen, M., Barklis, E., 1995. Structural interactions betwen retroviral Gag proteins examined by cysteine crosslinking. J. Virol. 69, 1150Ð1159. Hansen, M., Jelinek, L., Jones, R.I., Stegeman-Olsen, J.J., Barklis, E., Alin, K., Goff, S., 1996. Amino acid substitutions in the CA protein of 1993. Assembly and composition of intracellular particles formed by Moloney murine leukemia virus that block early events in infection. Moloney murine leukemia virus. J. Virol. 67, 5163Ð5174. Virology 222, 3439Ð3451. Hansen, M., Jelinek, L., Whiting, S., Barklis, E., 1990. Transport and Baldi, P., Brunak, S., Frasconi, P., Pollastri, G., Soda, G., 1999. Exploiting assembly of gag proteins into Moloney murine leukemia virua. J. Virol. the past and the future in protein secondary structure prediction. Bioin- 64, 5306Ð5316. formatics 15, 937Ð946. Huang, M., Orenstein, J., Martin, M., Freed, E., 1995. p6Gag is required Barklis, E., McDermott, J., Wilkens, S., Schabtach, E., Schmid, M.F., for particle production from full-length human immunodeficiency virus Fuller, S., Karanjia, S., Love, Z., Jones, R., Rui, Y., Zhao, X., Thomp- type 1 molecular clones expressing protease. J. Virol. 69, 6810Ð6818. son, D., 1997. Structural analysis of membrane-bound retrovirus capsid Jones, D., 1999. Protein secondary structure prediction based on position- proteins. EMBO J 16, 1199Ð1213. specific scoring matrices. J. Mol. Bio. 292, 195Ð202. Barklis, E., Mulligan, R., Jaenisch, R., 1986. Chromosomal position or Jones, T.A., Blaug, G., Hansen, M., Barklis, E., 1990. Assembly of gag- virus mutation permits retrovirus expression in embryonal carcinoma b-galactosidase proteins into retrovirus particles. J. Virol. 64, 2265Ð cells. Cell 47, 391Ð399. 2279. Berkowitz, R., Ohagen, A., Hoglund, S., Goff, S., 1995. Retroviral nucleo- Jin, Z., Jin, L., Peterson, D., Lawson, C., 1999. Model for lentivirus capsid capsid domains mediate the specific recognition of genomic viral core assembly based on crystal dimers of EIAV p26. J. Mol. Biol. 286, RNAs by chimeric Gag polyproteins during RNA packaging in vivo. 83Ð93. J. Virol. 69, 6445Ð6456. Khorasanizadeh, S., Campos-Olivas, R., Summers, M., 1999. Solution Berthet-Colominas, C., Monaco, S., Novelli, A., Sibai, G., Mallet, F., structure of the capsid protein from the human T-cell leukemia virus Cusack, S., 1999. Head-to-tail dimers and interdomain flexibility re- type-I. J. Mol. Biol. 291, 491Ð505. vealed by the crystal structure of HIV-1 capsid protein (p24) com- Kikonyogo, A., Bouamr, F., Vana, M., Xiang, Y., Aiyar, A., Carter, C., plexed with a monoclonal antibody Fab. EMBO J 18, 1124Ð1136. Leis, J., 2001. Proteins related to the Nedd4 family of ubiquitin protein Borsetti, A., Ohagen, A., Gottlinger, H., 1998. The C-terminal half of the ligases interact with the L domain of Rous sarcoma virus and are human immunodeficiency virus type 1 Gag precursor is sufficient for required for Gag budding from cells. Proc. Natl. Acad. Sci. USA 98, efficient particle assembly. J. Virol. 72, 9313Ð9317. 11199Ð11204. Bowzard, J., Wills, J., Craven, R., 2001. Second-site suppressors of Rous Kingston, R., Fitzon-Ostendorp, T., Eisenmesser, E., Schatz, G., Vogt, V., sarcoma virus CA mutations: evidence for interdomain interactions. Post, C., Rossmann, M., 2000. Structure and self-association of the J. Virol. 75, 6850Ð6856. Rous sarcoma virus capsid protein. Struct. Folding Design 8, 617Ð628. Campos-Olivas, R., Newman, J., Summers, M., 2000. Solution structure Li, S., Hill, C., Sundquist, W., Finch, J., 2000. Image reconstuctions of and dynamics of the Rous sarcoma virus capsid protein and comparison helical assemblies of the HIV-1 CA protein. Nature 407, 409Ð413. with capsid proteins of other retroviruses. J. Mol. Biol. 296, 633Ð49. Mammano, F., Ohagen, A., Hoglund, S., Gottlinger, H., 1994. Role of the Chesebro, B., Britt, W., Evans, L., Wehrly, K., Nishio, J., Cloyd, M., 1983. major homology region of human immunodeficiency virus Type 1 in Characterization of monoclonal antibodies reactive with murine leuke- virons morprogenesis. J. Virol. 68, 4927Ð4936. mia viruses: use in analysis of strains of Friend MCF and Friend Mayo, K., Vana, M., McDermott, J., Huseby, D., Leis, J., Barklis, E., 2002. ecotropic murine leukemia virus. Virology 127, 134Ð148. Analysis of Rous sarcoma virus capsid protein variants assembled on Craven, R., Leure-duPree, A., Weldon, R., Wills, J., 1995. Genetic analysis lipid monolayers. J. Mol. Biol. 316, 667Ð678. of the major homology region of the Rous sarcoma virus Gag protein. McDermott, J., Farrell, L., Ross, R., Barklis, E., 1996. Structural analysis J. Virol. 69, 4213Ð27. of human immunodeficiency virus type 1 Gag protein interactions, Demirov, D., Ono, A., Orenstein, J., Freed, E., 2002a. Overexpression of using cysteine-specific reagents. J. Virol. 70, 5106Ð5114. McDermott, J., Karanjia, S., Love, Z., Barklis, E., 2000a. Crosslink anal- the N-terminal domain of TSG101 inhibits HIV-1 budding by blocking ysis of N-terminal, C-terminal, and N/B determining regions of the late domain function. Proc. Natl. Acad. Sci. USA 99, 955Ð960. Moloney murine leukemia virus capsid protein. Virology 269, 190Ð Demirov, D., Orenstein, J., Freed, E., 2002b. The late domain of human 200. immunodeficiency virus type 1 p6 promotes virus release in a cell McDermott, J., Mayo, K., Barklis, E., 2000b. Three-dimensional organi- type-dependent manner. J. Virol. 76, 105Ð117. zation of retroviral capsid proteins on a lipid monolayer. J. Mol. Biol. D’Souza, V., Malamed, J., Habib, D., Pullen, K., Wallace, K., Summers, 302, 121Ð133. M.F., 2001. Identification of a high affinity nucleocapsid protein bind- McGuffin, L., Bryson, K., Jones, D., 2000. The PSIPRED protein structure ing element within the Moloney murine leukemia virus Psi-RNA pack- prediction server. Bioinformatics 16, 404Ð405. aging signal: implications for genome recognition. J. Mol. Biol. 314, Momany, C., Kovari, L., Prongay, A., Keller, W., Gitti, R., Lee, B., 217Ð232. Gorbalenya, A., Tong, L., McClure, J., Ehrlich, L., Summers, M., Gamble, T.R., Vajdos, F.F., Yoo, S., Worthylake, D.K., Houseweart, M., Carter, C., Rossmann, M., 1996. Crystal structure of dimeric HIV-1 Sundquist, W.I., Hill, C.P., 1996. Crystal structure of human cyclophi- capsid protein. Nature Struct. Biol. 3, 763Ð770. lin A bound to the amino-terminal domain of HIV-1 capsid. Cell 87, Mougel, M., Zhang, Y., Barklis, E., 1996. cis-Active structural motifs 1285Ð1294. involved in specific encapsidation of Moloney murine leukemia virus Gamble, T.R., Yoo, S., Vajdos, F.F., von Schwedler, U.K., Worthylake, RNA. J. Virol. 70, 5043Ð5050. D.K., Wang, H., McCutcheon, J.P., Sundquist, W.I., Hill, C.P., 1997. Patnaik, A., Chau, V., Wills, J., 2000. Ubiquitin is part of the retrovirus Structure of the carboxyl-terminal dimerization domain of the HIV-1 budding machinery. Proc. Natl. Acad. Sci. USA 97, 13069Ð13074. capsid protein. Science 278, 849Ð853. Petroupoulos, C., 1997. Retroviral taxonomy, protein structure, sequences, Garrus, J., von Schwedler, U., Pornillos, O., Morham, S., Zavitz, K., Wang, and genetic maps, in: Coffin, J., Hughes, S., Varmus, H. (Eds.), Ret- H., Wettstein, D., Stray, K., Cote, M., Rich, R., Myszka, D., Sundquist, roviruses, Appendix 2. Cold Spring Harbor Laboratory Press, Cold W., 2001. Tsg101 and the vacuolar protein sorting pathway are essen- Spring Harbor, NY. tial for HIV-1 budding. Cell 107, 55Ð65. Raghava, G., 2000. Protein secondary structure prediction using nearest Gitti, R., Lee, B., Walker, J., Summers, M., Yoo, S., Sundquist, W., 1996. neighbor and neural network approach. Fourth Community Wide Ex- Structure of the amino-terminal core domain of the HIV-1 capsid periment on the Critical Assessment of Techniques for Protein Struc- protein. Science 273, 231Ð235. ture Prediction. Asilomar, CA, pp. 75Ð76. B. Arvidson et al. / Virology 308 (2003) 166Ð177 177

Ratner, L., Fisher, A., Jagodzinski, L., Mitsuya, H., Liou, R., Gallo, R., von Schwedler, U.K., Stemmler, T.L., Klishko, V.Y., Li, S., Albertine, Wong-Staal, F., 1987. Complete nucleotide sequences of functional K.H., Davis, D.R., Sundquist, W.I., 1998. Proteolytic refolding of the clones of the AIDS virus. AIDS Res. Hum. Retroviruses 3, 57Ð69. HIV-1 capsid protein amino-terminus facilitates viral core assembly. Schnieke, A., Stuhlmann, H., Harbers, K., Chumakov, I., Jaenisch, R., EMBO J 17, 1555Ð1568. 1983. Endogenous Moloney leukemia virus in nonviremic Mov sub- Wang, C-T., Barklis, E., 1993. Assembly, processing, and infectivity of strains of mice carries defects in the proviral genome. J. Virol. 45, human immunodeficiency virus type 1 gag mutants. J. Virol. 67, 4264Ð 505Ð513. 4273. Schwartzberg, P., Colicelli, J., Gordon, M.L., Goff, S.P., 1984. Mutations Wang, C-T., Stegeman-Olsen, J., Zhang, Y., Barklis, E., 1994. Assembly in the gag gene of Moloney murine leukemia virus: effects on produc- of HIV GAG-B-galactosidase fusion proteins into virus particles. Vi- tion of virions and reverse transcriptase. J. Virol. 49, 918Ð924. rology 200, 524Ð534. Strack, B., Calistri, A., Accola, M., Palu, G., Gottlinger, H., 2000. A role Wang, C-T., Zhang, Y., McDermott, J., Barklis, E., 1993. Conditional for ubiquitin ligase recruitment in retrovirus release. Proc. Natl. Acad. infectivity of a human immunodeficiency virus matrix domain deletion Sci. USA 97, 13063Ð13068. mutant. J. Virol. 67, 7067Ð7076. Swanstrom, R., Wills, J., 1997. Synthesis, assembly and processing of viral Wills, J., Cameron, C., Wilson, C., Xiang, Y., Bennett, R., Leis, J., 1994. proteins, in: Coffin, J., Hughes, S., Varmus, H., (Eds.), Retrovirues, An assembly domain of Rous sarcoma virus Gag protein required late Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. in budding. J. Virol. 68, 6605Ð6618. 263Ð334. Yuan, B., Li, X., Goff, S., 1999. Mutations altering the Moloney murine Tang, C., Ndassa, Y., Summers, M.F., 2002. Structure of the N-terminal leukemia virus p12 Gag protein affect virion production and early 283-residue fragment of the immature HIV-1 Gag polyprotein. Nature events of the virus life cycle. EMBO J 18, 4700Ð4710. Struct. Biol. 9, 537Ð543. Yuan, B., Campbell, S., Bacharach, E., Rein, A., Goff, S., 2000. Infectivity Tang, S., Murakami, T., Agresta, B., Campbell, S., Freed, E., Levin, J., of Moloney murine leukemia virus defective in late assembly events is 2001. Human immunodeficiency virus type 1 N-terminal capsid mu- restored by late assembly domains of other retroviruses. J. Virol. 74, tants that exhibit aberrant core morphology and are blocked in initiation 7250Ð7260. of reverse transcription in infected cells. J. Virol. 75, 9357Ð9366. Zhang, Y., Barklis, E., 1995. Nucleocapsid protein effects on the specificity VerPlank, L., Bouamr, F., LaGrassa, T., Agresta, B., Kikonyogo, A., Leis, of retrovirus RNA encapsidation. J. Virol. 69, 5716Ð5722. J., Carter, C., 2001. Tsg101, a homologue of ubiquitin-conjugating (E2) Zhang, Y., Barklis, E., 1997. Effects of nucleocapsid mutations on human enzymes, binds the L domain in HIV type 1 Pr55(Gag). Proc. Natl. immunodeficiency virus assembly and RNA encapsidation. J. Virol. 71, Acad. Sci. USA 98, 7724Ð7729. 6765Ð6776.