Opposing Mechanisms Involving RNA and Lipids Regulate HIV-1 Gag Membrane Binding Through the Highly Basic Region of the Matrix Domain

Opposing mechanisms involving RNA and lipids regulate HIV-1 Gag membrane binding through the highly basic region of the matrix domain

Vineela Chukkapalli, Seung J. Oh, and Akira Ono1

Department of Microbiology and Immunology, University of Michigan Medical School, Ann Arbor, MI 48109

Edited by John M. Coffin, Tufts University School of Medicine, Boston, MA, and approved December 3, 2009 (received for review July 30, 2009) Membrane binding of Gag, a crucial step in HIV-1 assembly, is understood. The HBR is also implicated in other steps of virus facilitated by bipartite signals within the matrix (MA) domain: replication such as the postentry process (for review, see ref. 20). N-terminal myristoyl moiety and the highly basic region (HBR). We Many cellular proteins that interact peripherally with the cyto- and others have shown that Gag interacts with a plasma- plasmic leaflet of membranes have basic amino acid-rich domains membrane-specific acidic phospholipid, phosphatidylinositol- (21). The subcellular localization of these proteins generally (4,5)-bisphosphate [PI(4,5)P2], via the HBR, and that this interaction depends on the specific binding of their basic domains with acidic is important for efficient membrane binding and plasma mem- lipids of target membranes. For example, the pleckstrin homology – brane targeting of Gag. Generally, in protein PI(4,5)P2 interac- domain of phospholipase C δ1(PHPLCδ1) has basic residues loca- tions, basic residues promote the interaction as docking sites for ted on the surface of a binding pocket for PI(4,5)P2, a PM-specific the acidic headgroup of the lipid. In this study, toward better acidic lipid. These basic residues specifically interact with the – understanding of the Gag PI(4,5)P2 interaction, we sought to phosphates on the inositol headgroup of PI(4,5)P2 (22). This determine the roles played by all of the basic residues in the interaction allows PHPLCδ1 to localize to the PM. Similarly, basic fi HBR. We identi ed three basic residues promoting PI(4,5)P2- residues of other phosphoinositide-binding domains play crucial dependent Gag-membrane binding. Unexpectedly, two other roles in specific interactions with target membranes. HBR residues, Lys25 and Lys26, suppress membrane binding in Previously, we showed that depleting cellular PI(4,5)P2 reduces the absence of PI(4,5)P2 and prevent promiscuous intracellular virus release of HIV-1 significantly and relocalizes Gag from the localization of Gag. This inhibition of nonspecific membrane bind- PM to cytosol or intracellular compartments (23, 24). Not only ing is likely through suppression of myristate-dependent hydro- HIV-1, but other retroviruses such as HIV-2, murine leukemia phobic interaction because mutating Lys25 and Lys26 enhances virus, and Mason-Pfizer monkey virus also require PI(4,5)P2 or binding of Gag with neutral-charged liposomes. These residues other phosphoinositides for efficient virus release (25–28). Several were reported to bind RNA. Importantly, we found that RNA also studies using various in vitro techniques, including NMR, protein negatively regulates Gag membrane binding. In the absence but foot printing, liposome binding, and monolayer membrane bind- not presence of PI(4,5)P , RNA bound to MA HBR abolishes Gag- 2 ing, suggest that MA interacts specifically with PI(4,5)P2 (7, 24, 29, liposome binding. Altogether, these data indicate that the HBR is 30). A recent lipid analysis study demonstrated that PI(4,5)P2 is unique among basic phosphoinositide-binding domains, because it enriched in the HIV envelope in an MA-dependent manner, integrates three regulatory components, PI(4,5)P , myristate, and 2 confirming the MA–PI(4,5)P2 interaction in vivo (26). These RNA, to ensure plasma membrane specificity for particle assembly. studies suggest that MA basic residues, in particular those in the HBR, promote PI(4,5)P2 interaction and membrane binding, as basic residues | PI(4,5)P2 | phosphoinositide | plasma membrane | retrovirus observed for basic residues in other phosphoinositide-binding assembly domains. However, the contribution of each basic residue in the HBR to the interaction between full-length Gag and PI(4,5)P2- IV-1 assembly is a multistep process mediated primarily by containing membrane remains to be examined. Hthe viral structural protein, Gag. This protein, synthesized as Previous reports showed that mutating lysines 29 and 31 to a polyprotein precursor, Pr55Gag, contains four major domains: threonines (29KT/31KT) within the HBR relocalizes Gag from matrix (MA), capsid (CA), nucleocapsid (NC), and p6. CA and the PM to intracellular compartments, as observed for wild-type NC contain determinants for Gag multimerization, whereas p6 (WT) Gag in PI(4,5)P2-depleted cells (31). Consistent with the facilitates release of virus particles (1, 2). In most cell types, Gag altered localization, the in vitro liposome binding assay showed fl assembles on the cytoplasmic lea et of the plasma membrane that 29KT/31KT Gag binds PI(4,5)P2 less efficiently compared to fi (PM). MA is required for this PM-speci c targeting as well as the WT Gag (24). However, the Gag–PI(4,5)P2 interaction was not lipid bilayer binding of Gag. completely abolished, suggesting a role for other basic amino MA has bipartite entities that facilitate membrane binding of acids in this interaction. i ii Gag: ( ) the N-terminal myristoyl moiety and ( ) a highly basic In this study, toward better understanding of the Gag–PI(4,5)P2 region (HBR) comprising of residues 17–31 (3–5). The N-terminal interaction, we sought to identify all of the basic residues in the myristate is normally sequestered within the globular head domain HBR important for this interaction. We confirmed that the HBR of MA (6), which also includes the HBR. A structural change, as a whole is indeed essential for PI(4,5)P2 interaction and efficient either through Gag multimerization or phosphatidylinositol- (4,5)-bisphosphate [PI(4,5)P2] binding to MA, is required to expose the myristate and enhance membrane binding (6, 7). This Author contributions: V.C., S.J.O., and A.O. designed research; V.C. and S.J.O. performed mechanism is known as the myristoyl switch (8, 9). Structural and research; V.C. and S.J.O. contributed new reagents/analytic tools; V.C., S.J.O., and A.O. biochemical studies suggest that the HBR of HIV-1 MA and analyzed data; and V.C. and A.O. wrote the paper. analogous regions of other retroviral MA are apposed to face the The authors declare no conflict of interest. cytoplasmic leaflet of membranes and facilitate membrane bind- This article is a PNAS Direct Submission. ing by interacting with acidic lipids (4, 10–14). In addition, MA, 1To whom correspondence should be addressed. E-mail: [email protected]. – especially the HBR, has been shown to interact with RNA (15 19) This article contains supporting information online at www.pnas.org/cgi/content/full/ although the role of this interaction in virus assembly is less well 0908661107/DCSupplemental.

1600–1605 | PNAS | January 26, 2010 | vol. 107 | no. 4 www.pnas.org/cgi/doi/10.1073/pnas.0908661107 Downloaded by guest on September 24, 2021 PC+PS PC+PS+PI(4,5)P membrane binding of Gag. Surprisingly, analysis of a panel of Gag A 2 mutants revealed that a part of the HBR actually suppresses M NM M NM WT membrane binding in the absence of PI(4,5)P2. This suppression is likely due to inhibition of the myristate-dependent hydrophobic 6A2T interaction. Furthermore, RNA bound to the HBR also inhibits WT Gag-membrane binding through inhibition of acidic lipid binding 25KT/26KT and to a lesser extent via suppression of myristate exposure. 120 Altogether, these results showed that the HBR regulates mem- PC+PS PC+PS+PI(4,5)P2 brane binding in both positive and negative manners. Data pre- 100 ns sented here suggest a unique, RNA-dependent mechanism by 80 which HIV-1 Gag ensures specific binding to the PI(4,5)P -con- * 2 * taining membrane, i.e., the PM. 60 * * Results 40 Efficiency (%) 20 ns *** The HBR of the MA Domain Is Essential for PI(4,5)P2-Dependent ns ns ns Membrane Binding of Gag. To examine whether the HBR of MA 0 is indeed important for Gag–PI(4,5)P interaction, we mutated Liposome Binding Relative 2 WT WT all of the positively charged residues (6A2T) within the HBR. 17KT 6A2T 17KT 6A2T This mutant was then analyzed for PI(4,5)P2-dependent mem- 17KT/29KT25KT/26KT29KT/31KT 17KT/29KT25KT/26KT29KT/31KT brane binding by previously described in vitro liposome binding WT 6A2T assay (24). In this assay, myristoylated full-length Gag is syn- B thesized by using rabbit reticulocyte lysates and incubated with liposomes. Liposome-bound and non-liposome-bound Gag proteins are separated by sucrose gradient membrane flotation centrifugation. As shown in Fig. 1A, compared to WT Gag, binding of 6A2T Gag to PI(4,5)P2-containing liposomes was significantly reduced. To examine the effect of the 6A2T change 400 on the localization of Gag in cells, we introduced the same C ** MICROBIOLOGY mutations in a Gag derivative tagged with Venus, a variant of 300 fl yellow uorescent protein (GagVenus) (24, 32). Consistent with 200 the weak liposome binding, 6A2T GagVenus showed a hazy, cytosolic signal in most cells. Additionally, perinuclear local- 100 Efficiency (%) ization of GagVenus was observed in some cells (Fig. 1B). This 0 Relative Virus Relative Release 25KT/26KT phenotype is in contrast to that of WT GagVenus that localizes WT 25KT/ predominantly on the PM. The virus release was undetectable 26KT for the 6A2T mutant. Altogether, these results suggest that the Fig. 1. Lys25 and Lys26 in the HBR inhibit membrane binding of Gag and HBR likely forms a binding site for PI(4,5)P2 and that this 35 fi suppress promiscuous localization. (A)[S]-labeled WT Gag and Gag interaction is important for ef cient Gag membrane binding and mutants with indicated amino acid substitutions were synthesized by the in virus release. vitro transcription and translation system by using reticulocyte lysates and

incubated with liposomes containing or not containing PI(4,5)P2. The reac- MA Basic Residues Lys25 and Lys26 in the HBR Suppress PI(4,5)P2- tions were subjected to membrane flotation centrifugation, and five 1-mL Independent Membrane Binding and Inhibit Promiscuous Localization fractions were collected from each tube. Representative gels are shown in Upper. M, liposome-bound; NM, non-liposome-bound. The liposome binding of Gag. To pinpoint the residues essential for Gag–PI(4,5)P2 interaction, we analyzed liposome binding of Gag mutants with efficiency was calculated as the amount of liposome-bound Gag as a fraction of total Gag. (Lower) The relative liposome binding efficiency was calculated single or double amino acid substitutions of the basic residues in fi in vitro in comparison with the binding ef ciency of WT Gag with PI(4,5)P2-con- the MA HBR by using liposome binding assay (Figs. 1, S1, taining liposomes for each experiment. Data from three independent and S2). Several of these mutants have been studied with respect to experiments are shown as means ± SD. P values were determined by using other proposed MA functions (19, 20, 33, 34). Overall, the mutants Student’s t test as a comparison with WT (ns, not significant; *, P < 0.05; **, could be categorized into three different phenotypes: (i) reduced P < 0.01; ***, P < 0.001). (B) HeLa cells expressing WT, 6A2T, and 25KT/26KT PI(4,5)P2 binding (17KT, 17KT/29KT, and 29KT/31KT), (ii) GagVenus were analyzed by using Nikon TE2000 microscope. (C) HeLa cells increased PI(4,5)P2-independent binding (25KT/26KT), and (iii) expressing HIV molecular clones encoding WT or 25KT/26KT Gag were no change compared to WT (all of the other mutants). Altogether, metabolically labeled for 2 h. Cell- and virus-associated Gag were recovered these experiments suggest that MA basic residues 17, 29, and 31 by immunoprecipitation. Virus release efficiency was calculated as the promote MA–PI(4,5)P interaction. amount of virus-associated Gag as a fraction of total Gag synthesized during 2 the labeling period and normalized to the virus release efficiency of WT. The Among the Gag mutants examined, the 25KT/26KT mutant, average virus release efficiencies for WT and 25KT/26KT are 7.22% and where lysines 25 and 26 were changed to threonines, showed no 17.85%, respectively. Data from five different experiments are shown as fi signi cant change in binding to PI(4,5)P2-containing liposomes. means ± SD. **, P < 0.01. Unexpectedly, however, this mutant had significantly elevated binding to control liposomes that lack PI(4,5)P2 (Fig. 1A). These results suggest that, in addition to facilitating membrane binding GagVenus displayed promiscuous membrane binding in cells, through PI(4,5)P2 interaction, some basic residues in the HBR localizing both at the PM and intracellular compartments (Fig. inhibit PI(4,5)P2-independent membrane binding. 1B). Altogether, these results indicate that MA basic residues 25 To characterize the 25KT/26KT mutant Gag further, we and 26 inhibit PI(4,5)P2-independent membrane binding of Gag. analyzed both virus release and localization by using cell-based assays as described (24). Consistent with the lack of requirement RNA Inhibits Liposome Binding of Gag. Previous studies showed that – for PI(4,5)P2 observed in the in vitro liposome binding assay (Fig. the residues Lys25 and Lys26 can interact with RNA (17 19). 1A), the virus release efficiency of 25KT/26KT was substantially Therefore, it is conceivable that the 25KT/26KT mutations augmented compared to WT (Fig. 1C). Notably, 25KT/26KT abolish RNA binding and allow other MA basic residues to

Chukkapalli et al. PNAS | January 26, 2010 | vol. 107 | no. 4 | 1601 Downloaded by guest on September 24, 2021 freely interact with acidic lipids, thereby enhancing promiscuous WT 25KT/26KT membrane binding. To examine whether RNA does interfere M NM M NM with Gag membrane binding, we ﬁrst tested the effect of RNase PC treatment on the liposome binding of WT Gag. Remarkably, treating WT Gag with RNase before incubation with liposomes PC+PS

greatly enhanced interaction of Gag with control liposomes PC+PS+PI(4,5)P2 containing phosphatidylcholine (PC) and phosphatidylserine 140 ns (PS) (PC+PS) (Fig. 2). A similar increase was observed with PI 120 * (4,5)P2-containing liposomes [PC+PS+PI(4,5)P2]. Notably, 100 however, as we reported previously (24), a substantial amount of * 80 Gag was bound to PI(4,5)P2-containing liposomes even without RNase treatment (Fig. 2). These results indicate that Gag- 60 40

membrane binding is suppressed almost completely by RNA Efficiency (%) 20 unless PI(4,5)P2 is present. To determine whether the phenotype of 25KT/26KT Gag can 0 Relative Liposome Binding Relative be explained by the loss of RNA binding, we examined the effect WT WT WT

of RNase treatment on liposome binding of this mutant Gag. 25KT/26KT 25KT/26KT 25KT/26KT The enhancement of membrane binding by RNase treatment was PC PC+PS PC+PS+PI(4,5)P2 smaller than that observed for WT, especially in the absence of Fig. 3. 25KT/26KT Gag binds efficiently to liposomes lacking acidic phos- PI(4,5)P2 (Fig. 2). These data suggest that the 25KT/26KT pholipids. Binding of WT and 25KT/26KT Gag to liposomes containing indi- mutation and RNase treatment enhance membrane binding by cated lipids were analyzed as in Fig. 1A. Data from three independent overlapping mechanisms. experiments are shown as means ± SD. ns, not significant; *, P < 0.05. There are no statistically significant differences between binding of 25KT/26KT to The 25KT/26KT Changes Increase the Hydrophobic Interaction with different liposomes. Membranes. Efficient membrane binding of Gag requires both the myristoyl moiety and the HBR. It is still possible that the 25KT/ fi 26KT mutant enhances membrane binding by constitutively lysines 25 and 26 suppress nonspeci c membrane binding by exposing the myristoyl moiety. To test this hypothesis, we ana- regulating hydrophobic interactions dependent on the myristate. lyzed the binding of 25KT/26KT Gag to liposomes containing PC RNA Bound to MA but Not NC Inhibits Liposome Binding of Gag. As alone. Because PC is a neutral lipid, binding of Gag to these described above, RNase treatment enhanced membrane binding liposomes would likely be mediated by hydrophobic interaction of both WT and to a lesser extent 25KT/26KT Gag (Fig. 2). As through N-terminal myristate moiety. Consistent with the pre- RNA-mediated inhibition of membrane binding may represent a vious findings that the myristate moiety is sequestered inside the unique mode of regulation in virus assembly, we sought to globular head of MA, and that PI(4,5)P binding is required for 2 understand the mechanism of this inhibition. Coimmunopreci- facilitating myristate exposure (2, 6, 7), WT Gag was unable to pitation experiments showed that both MA HBR and NC facil- bind well to PC liposomes (Fig. 3). However, the double mutant fi itate RNA binding to Gag in reticulocyte lysates (Fig. S3). To 25KT/26KT was able to bind ef ciently to liposomes containing test whether RNA bound to NC is involved in the inhibition of only PC (Fig. 3). These results suggest that the 25KT/26KT membrane binding, we analyzed NC-deleted Gag (delNC) for mutant readily exposes the myristate. Therefore, it is likely that RNase sensitivity. Liposome binding of delNC Gag was increased upon RNase treatment to a similar extent as WT Gag (Fig. 4A). These results suggest that at least in the liposome WT 25KT/26KT binding assay, NC is not required for RNA-mediated inhibition RNase MNMMNM of membrane binding of Gag. PC+PS - + RNase Treatment Increases both the Interaction of MA Basic Residues PC+PS+ - with Acidic LipidsandtheMyristate-Dependent Hydrophobic Interaction. PI(4,5)P2 + The results presented above are consistent with the possibility that 90 *** RNA bound to MA HBR blocks the interaction of the HBR basic 80 ** 70 residues with acidic lipids and interferes with Gag membrane 60 *** ns binding. However, it is also conceivable that RNA binding to the 50 HBR might reduce myristate exposure, thereby suppressing mem- 40 brane binding of Gag. Both of these possibilities are consistent with 30 the observation that membrane binding of 25KT/26KT Gag that has Efficiency (%) 20 readily exposed myristate (Fig. 3) is enhanced upon RNase treat- Liposome Binding 10 0 ment markedly but not as much as WT Gag (Fig. 2). Hence, to RNase --++- - ++ further understand the mechanism by which RNA inhibits mem- WT 25KT/26KT WT 25KT/26KT brane binding, we analyzed the myristoylation-defective Gag mutant PC+PS PC+PS+PI(4,5)P2 (1GA). Interestingly, although RNase treatment did increase Fig. 2. RNA inhibits membrane binding of Gag. WT and 25KT/26KT Gag membrane binding of 1GA Gag, the extent of increase depended on were synthesized by using rabbit reticulocyte lysates and either treated or the composition of liposomes used in the assay. 1GA Gag bound PI not treated with RNase. Subsequently, liposomes containing indicated lipids fi fi (4,5)P2-containing liposomes as ef ciently as WT Gag upon RNase were added, and the liposome binding ef ciency was calculated as described treatment (Fig. 4B). This observation strongly supports the model in Materials and Methods. M, liposome-bound; NM, non-liposome-bound. that interaction of basic residues in the HBR with acidic lipids is Data from three (PC+PS) or four [PC+PS+PI(4,5)P2] independent experiments are shown as means ± SD. ns, not significant; **, P < 0.01; ***, P < 0.001. increased upon RNase treatment.

There is a statistically signiﬁcant difference (*) in binding to PI(4,5)P2-con- In contrast, RNase treatment increased binding of 1GA Gag taining liposomes between RNase-treated WT and 25KT/26KT Gag. to PC+PS liposomes only modestly compared to that of WT

1602 | www.pnas.org/cgi/doi/10.1073/pnas.0908661107 Chukkapalli et al. Downloaded by guest on September 24, 2021 A WT delNC liposome binding upon RNase treatment, whereas 1GA Gag RNase MNMMNM showed no increase (Fig. 4C). These data suggest that removal of PC+PS - RNA has an impact on the myristoyl switch as well. Altogether, + these results indicate that RNA inhibits membrane binding both 80 ** ** by preventing the electrostatic interaction between basic residues 70 and acidic lipids and by suppressing the myristate-dependent 60 hydrophobic interaction. 50 40 Discussion 30 Phosphoinositide-binding domains found in cellular proteins

Efficiency (%) 20 contain basic residues that are essential for interaction with

Liposome Binding 10 acidic headgroups of phosphoinositides (21). Similarly, the cur- 0 rent study showed that some basic residues within the HBR RNase --++ fi WT delNC (Lys17, Lys29, and Lys31) are important for ef cient Gag membrane binding mediated by PI(4,5)P2 (Fig. 1A). However, B WT 1GA our study also revealed that the HBR contains basic residues RNase MNMMNM (Lys25 and Lys26) that suppress membrane binding of Gag (Fig. - A PC+PS 1 ). This negative regulation of membrane binding by Lys25 and + Lys26 likely occurs via myristate sequestration (Fig. 3). Con- PC+PS+ - sistent with the increased PI(4,5)P2-independent liposome PI(4,5)P2 + binding, 25KT/26KT Gag localizes promiscuously in cells and 90 releases virus particles more efficiently than WT Gag (Fig. 1 B ** 80 ** and C), a phenotype reminiscent of Gag derivatives with MA 70 ** globular head deletions (35). 60 This work also identified RNA as a negative regulator of Gag 50 membrane binding. Several in vitro studies including this study 40 (Fig. S3) showed that HIV-1 MA interacts with RNA via its MICROBIOLOGY 30 HBR (17–19). Therefore, competition between RNA and acidic Efficiency (%) * 20 lipids for binding to the HBR at least partly accounts for the Liposome Binding 10 RNA-mediated suppression of Gag membrane binding. Con- in vitro 0 sistent with this hypothesis, in other experiments per- RNase --++- - ++ formed in the absence of RNA, binding of MA to PS has been WT 1GA WT 1GA

PC+PS PC+PS+PI(4,5)P2

WT 1GA Non-specific PI(4,5)P2-containing C membrane membrane RNase MNMMNM

PC - + 20 ** X 18 Or 16 WT 14 12 10 8 6 Efficiency (%) ns 4 Liposome Binding 2 0 RNase --++ WT 1GA 25KT/26KT Fig. 4. MA-bound RNA suppresses membrane binding of Gag by inhibiting interactions between basic residues and acidic lipids and by reducing the myristate-dependent interaction. (A) Liposome binding of WT and delNC Gag was analyzed as in Fig. 2 by using PC+PS liposomes. Data from three independent experiments are shown as means ± SD. **, P < 0.01. There is no statistically signiﬁcant difference in membrane binding of RNase-treated MA Gag between WT and delNC. (B and C) Liposome binding of WT and 1GA Gag was analyzed as in Fig. 2 by using liposomes containing either PC+PS or ﬁ

PC+PS+PI(4,5)P (B) or only PC (C). Data from three to ve independent WT with RNase 2 CA experiments are shown as means ± SD. ns, not signiﬁcant; *, P < 0.05; **, P < NC 0.01. There is a statistically signiﬁcant difference (*) in binding to PC+PS p6 liposomes but not PI(4,5)P2-containing liposomes between WT and 1GA Gag RNA, treated with RNase. Key: PS, PI(4,5)P2, HBR,

exposed partially exposed sequestered myristate, myristate, myristate

Gag. These results are consistent with the possibility that RNA Fig. 5. A model for the regulation of Gag-membrane binding speciﬁcity. PI

also inhibits membrane binding partially by sequestration of (4,5)P2-dependent and -independent membrane binding of WT, 25KT/26KT, myristate. Indeed, WT Gag did show a modest increase in PC- and RNase-treated Gag are shown. See Discussion for explanation.

Chukkapalli et al. PNAS | January 26, 2010 | vol. 107 | no. 4 | 1603 Downloaded by guest on September 24, 2021 readily observed (36–38), whereas in our system, Gag binding to dependent hydrophobic interaction, and negatively by interaction the PC+PS liposomes was negligible without RNase treatment. with RNA. RNA binding to MA HBR not only interferes with Importantly, in contrast to PC+PS liposomes, a substantial interactions between MA basic amino acids and acidic lipids, but amount of Gag binds to PC+PS+PI(4,5)P2 liposomes even also appears to reduce myristate exposure. Based on this and without RNase treatment (Figs. 2 and 4). Consistent with these previous studies, we put forth the following model (Fig. 5). Both data, while this manuscript was in revision, Eric Barklis’s group RNA binding and suppression of myristate exposure mediated by reported that PI(4,5)P2-containing liposomes out-compete oli- the HBR prevent Gag from premature or nonspecific membrane fi gonucleotides for binding to beads coated with puri ed myr- binding. Once Gag reaches PI(4,5)P2-containing membranes such istylated MA domain (39). Thus, it is likely that PI(4,5)P2, but as the PM, PI(4,5)P2 binds the HBR, which then triggers myristate not other acidic lipids such as PS, can bind the HBR in the exposure and stabilizes binding to the lipid bilayer. However, if presence of RNA. In addition, exposure of myristate may also be myristate sequestration is blocked (e.g., by the 25KT/26KT more dependent on PI(4,5)P2 binding in the presence of RNA, mutations) or if RNA detaches from the HBR (e.g., by RNase because RNA binding to MA HBR also reduces myristate- treatment), Gag binds membrane in a nonspecific, PI(4,5)P2- dependent hydrophobic interaction (Fig. 4). In any case, our data independent manner (Fig. 5). Cooperation of interactions with suggest that RNA binding to MA represents an important reg- multiple membrane factors, which enhances the robustness and ulatory mechanism in Gag membrane binding. specificity of membrane binding, is prevalent in phosphoinositide- It remains to be determined whether RNA bound to the HBR binding domains (21). However, MA HBR is unique among these inhibits unregulated membrane binding in cells. In our in vitro domains for integrating opposing regulatory mechanisms to system, the RNA concentration is estimated to be lower than ensure specificity for the target membrane that contains PI(4,5)P2. that in the cytoplasm of HeLa cells. Nonetheless, the inhibition of membrane binding by RNA does not require NC, the major Materials and Methods RNA binding domain (Fig. 4A). However, in cells, it is likely that Cells and Plasmids. HeLa cells were cultured and maintained in DMEM sup- the viral genomic RNA bound to NC also binds MA because of plemented with 5% FBS. HIV-1 molecular clone pNL4-3 (61), its derivative their close proximity. This model is also consistent with a encoding YFP-tagged Gag (pNL4-3/GagVenus) and a Gag expression vector recently proposed folded-over conformation of Gag (40, 41) (not for in vitro translation reaction, pGEMNLNR, were described (24). pNL4-3/ depicted in Fig. 5). Notably, two recent reports suggest that RNA 1GA, pNL4-3/25KT/26KT, pNL4-3/17KT/19RL, pNL4-3/29KT/31KT, pNL4-3/ export from the nucleus plays a key role in MA-mediated 19RL, and pNL4-3/21RL were kind gifts from E. Freed (HIV Drug Resistance membrane binding of Gag (42, 43). These authors observed that Program, National Cancer Institute, Frederick, MD) (33, 34). Other MA HIV Gag assembly defects linked to specific Gag mRNA export changes were introduced into pNL4-3 by PCR mutagenesis. MA changes in pNL4-3 derivatives were also introduced into pGEMNLNR and pNL4-3/Gag- pathways could be rescued by increasing membrane binding Venus by using standard molecular biology techniques. pGEMNLNR/delNC ability of Gag, either by MA mutations or replacing MA with was constructed by replacing BssHII to PflMI fragment (nt 711–5297) with heterologous membrane binding domains (42, 43). In light of the that of pNL4-3/delNC (a kind gift from D. Ott) (AIDS Vaccine Program, SAIC- results presented here, it is possible that nuclear export pathways Frederick, Inc., National Cancer Institute, Frederick, MD) (62). used by viral RNAs affect MA–RNA interactions that, in turn, influence Gag membrane binding. This hypothesis is also in Liposome Binding Assay. Liposome binding assay was performed as described agreement with the observed autoinhibition of Gag membrane (24). Briefly, radiolabeled Gag is synthesized by using in vitro transcription and binding by HIV-1 MA in murine cells (44, 45). translation coupled reaction mix containing rabbit reticulocyte lysates (Prom- It has been long known that retroviral MA interacts with RNA ega). Liposomes containing corresponding lipids were added after 30 min at (15–19, 46–50). Although genomic RNA encapsidation does not 30°C and incubated further for 90 min. Then, the liposome-bound and non- fl require MA in the case of HIV-1 (51), a variety of functions have liposome-bound Gag were separated by equilibrium otation centrifugation. fi been suggested for the retroviral MA–RNA interaction in the later For RNase treatment experiments, the above protocol was modi ed as i following. Gag synthesis was performed at 30°C for 90 min. Subsequently, half of the virus life cycle. These functions include ( ) regulation of 1 μL of RNase, which is a mixture of pyrimidine-specific endoribonucleases viral mRNA translation (18, 52), (ii) encapsidation of genomic iii from bovine pancreas (Cat. No.: 11119915001, Roche Applied Science), was viral RNA (53), ( ) dimerization of encapsidated RNA (54, 55), added to 25 μL of the reaction mix and incubated for 20 min at 37°C. Lip- and (iv) enhancement of Gag assembly (56, 57). Thus, it is osomes were then added and incubated for 15 min at 30°C before equili- tempting to speculate that binding of the HBR to both PI(4,5)P2 brium flotation centrifugation was performed. and RNA may play a key role in coordinating the four-way interactions between the PM, viral RNA, MA, and NC during the late Virus Release Assay, Microscopy, and Statistical Analysis. Virus release assay phase of the retrovirus life cycle. For example, once Gag reaches was performed as described (23). Microscopy and statistical analysis were the PM, the viral RNA bound to MA might get displaced by the performed as described (24). HBR–PI(4,5)P2 interaction. This displaced RNA may then act as a scaffold for NC-dependent Gag multimerization and facilitate Coimmunoprecipitation of Gag and RNA. RNA binding to Gag synthesized in in efficient particle assembly at the PM. Such a process might also vitro transcription and translation coupled reactions was analyzed as described in SI Materials and Methods. require cellular proteins reported to play a role in Gag multimerization or RNA packaging, including ATP-binding cassette ACKNOWLEDGMENTS. We thank Drs. Mike Summers, Kalola Chancellor, Eric protein in the E subfamily (ABCE1) (58, 59) and Staufen-1 (60). Freed, and the members of our laboratory for helpful discussions and critical Clearly, further studies are necessary for addressing inter- review of the manuscript. We would also like to thank Drs. E. Freed and D. relationships between events during the late phase. Ott for providing plasmids. The following reagent was obtained through AIDS Research and Reference Reagent Program, Division of AIDS, N National In summary, our data indicate that the HBR within MA mod- Institute of Allergy and Infectious Diseases, National Institutes of Health: ulates membrane binding in three different ways: positively by HIV-Ig from NABI and National Heart, Lung, and Blood Institute. This work is interaction with PI(4,5)P2, negatively by suppression of myristate- supported by the National Institutes of Health Grant R01 AI071727 (to A.O.).

1. Freed EO (1998) HIV-1 gag proteins: Diverse functions in the virus life cycle. Virology 4. Zhou W, Parent LJ, Wills JW, Resh MD (1994) Identification of a membrane-binding 251:1–15. domain within the amino-terminal region of human immunodeficiency virus type 1 2. Adamson CS, Freed EO (2007) Human immunodeficiency virus type 1 assembly, Gag protein which interacts with acidic phospholipids. J Virol 68:2556–2569. release, and maturation. Adv Pharmacol 55:347–387. 5. Göttlinger HG, Sodroski JG, Haseltine WA (1989) Role of capsid precursor processing 3. Bryant M, Ratner L (1990) Myristoylation-dependent replication and assembly of and myristoylation in morphogenesis and infectivity of human immunodeficiency human immunodeficiency virus 1. Proc Natl Acad Sci USA 87:523–527. virus type 1. Proc Natl Acad Sci USA 86:5781–5785.

1604 | www.pnas.org/cgi/doi/10.1073/pnas.0908661107 Chukkapalli et al. Downloaded by guest on September 24, 2021 6. Tang C, et al. (2004) Entropic switch regulates myristate exposure in the HIV-1 matrix 34. Freed EO, Englund G, Martin MA (1995) Role of the basic domain of human protein. Proc Natl Acad Sci USA 101:517–522. immunodeficiency virus type 1 matrix in macrophage infection. J Virol 69:3949–3954. 7. Saad JS, et al. (2006) Structural basis for targeting HIV-1 Gag proteins to the plasma 35. Reil H, Bukovsky AA, Gelderblom HR, Göttlinger HG (1998) Efficient HIV-1 replication membrane for virus assembly. Proc Natl Acad Sci USA 103:11364–11369. can occur in the absence of the viral matrix protein. EMBO J 17:2699–2708. 8. Spearman P, Horton R, Ratner L, Kuli-Zade I (1997) Membrane binding of human 36. Dalton AK, Ako-Adjei D, Murray PS, Murray D, Vogt VM (2007) Electrostatic immunodeficiency virus type 1 matrix protein in vivo supports a conformational interactions drive membrane association of the human immunodeficiency virus type 1 myristyl switch mechanism. J Virol 71:6582–6592. Gag MA domain. J Virol 81:6434–6445. 9. Zhou W, Resh MD (1996) Differential membrane binding of the human 37. Ehrlich LS, Fong S, Scarlata S, Zybarth G, Carter C (1996) Partitioning of HIV-1 Gag and fi – immunode ciency virus type 1 matrix protein. J Virol 70:8540 8548. Gag-related proteins to membranes. Biochemistry 35:3933–3943. 10. Conte MR, Matthews S (1998) Retroviral matrix proteins: A structural perspective. 38. Alfadhli A, Huseby D, Kapit E, Colman D, Barklis E (2007) Human immunodeficiency – Virology 246:191 198. virus type 1 matrix protein assembles on membranes as a hexamer. J Virol 81: 11. Callahan EM, Wills JW (2000) Repositioning basic residues in the M domain of the 1472–1478. – Rous sarcoma virus gag protein. J Virol 74:11222 11229. 39. Alfadhli A, Still A, Barklis E (2009) Analysis of human immunodeficiency virus type 1 12. Hill CP, Worthylake D, Bancroft DP, Christensen AM, Sundquist WI (1996) Crystal matrix binding to membranes and nucleic acids. J Virol 83:12196–12203. structures of the trimeric human immunodeficiency virus type 1 matrix protein: 40. Datta SA, et al. (2007) Interactions between HIV-1 Gag molecules in solution: An Implications for membrane association and assembly. Proc Natl Acad Sci USA 93: inositol phosphate-mediated switch. J Mol Biol 365:799–811. 3099–3104. 41. Datta SA, et al. (2007) Conformation of the HIV-1 Gag protein in solution. J Mol Biol 13. Dalton AK, Murray PS, Murray D, Vogt VM (2005) Biochemical characterization of 365:812–824. rous sarcoma virus MA protein interaction with membranes. J Virol 79:6227–6238. 42. Sherer NM, Swanson CM, Papaioannou S, Malim MH (2009) Matrix mediates the 14. Murray PS, et al. (2005) Retroviral matrix domains share electrostatic homology: functional link between human immunodeficiency virus type 1 RNA nuclear export Models for membrane binding function throughout the viral life cycle. Structure 13: elements and the assembly competency of Gag in murine cells. J Virol 83:8525–8535. 1521–1531. 43. Jin J, Sturgeon T, Weisz OA, Mothes W, Montelaro RC (2009) HIV-1 matrix dependent 15. Chang CY, et al. (2008) HIV-1 matrix protein repositioning in nucleocapsid region fails membrane targeting is regulated by Gag mRNA trafficking. PLoS One 4:e6551. to confer virus-like particle assembly. Virology 378:97–104. 44. Hatziioannou T, Martin-Serrano J, Zang T, Bieniasz PD (2005) Matrix-induced 16. Lochrie MA, et al. (1997) In vitro selection of RNAs that bind to the human fi immunodeficiency virus type-1 gag polyprotein. Nucleic Acids Res 25:2902–2910. inhibition of membrane binding contributes to human immunode ciency virus type 1 – 17. Purohit P, Dupont S, Stevenson M, Green MR (2001) Sequence-specific interaction particle assembly defects in murine cells. J Virol 79:15586 15589. between HIV-1 matrix protein and viral genomic RNA revealed by in vitro genetic 45. Hübner W, Chen BK (2006) Inhibition of viral assembly in murine cells by HIV-1 matrix. – selection. RNA 7:576–584. Virology 352:27 38. 18. Cimarelli A, Luban J (1999) Translation elongation factor 1-alpha interacts specifically 46. Steeg CM, Vogt VM (1990) RNA-binding properties of the matrix protein (p19gag) of with the human immunodeficiency virus type 1 Gag polyprotein. J Virol 73: avian sarcoma and leukemia viruses. J Virol 64:847–855. 5388–5401. 47. Sen A, Todaro GJ (1977) The genome-associated, specific RNA binding proteins of 19. Hearps AC, Wagstaff KM, Piller SC, Jans DA (2008) The N-terminal basic domain of the avian and mammalian type C viruses. Cell 10:91–99. HIV-1 matrix protein does not contain a conventional nuclear localization sequence 48. Leis JP, McGinnis J, Green RW (1978) Rous sarcoma virus p19 binds to specific double-

but is required for DNA binding and protein self-association. Biochemistry 47: stranded regions of viral RNA: Effect of p19 on cleavage of viral RNA by RNase III. MICROBIOLOGY 2199–2210. Virology 84:87–98. 20. Hearps AC, Jans DA (2007) Regulating the functions of the HIV-1 matrix protein. AIDS 49. Karpel RL, Henderson LE, Oroszlan S (1987) Interactions of retroviral structural Res Hum Retroviruses 23:341–346. proteins with single-stranded nucleic acids. J Biol Chem 262:4961–4967. 21. Lemmon MA (2008) Membrane recognition by phospholipid-binding domains. Nat 50. Darlix JL, Spahr PF (1982) Binding sites of viral protein P19 onto Rous sarcoma virus Rev Mol Cell Biol 9:99–111. RNA and possible controls of viral functions. J Mol Biol 160:147–161. 22. Lemmon MA, Ferguson KM, O’Brien R, Sigler PB, Schlessinger J (1995) Specific and 51. Poon DT, Li G, Aldovini A (1998) Nucleocapsid and matrix protein contributions to high-affinity binding of inositol phosphates to an isolated pleckstrin homology selective human immunodeficiency virus type 1 genomic RNA packaging. J Virol 72: domain. Proc Natl Acad Sci USA 92:10472–10476. 1983–1993. 23. Ono A, Ablan SD, Lockett SJ, Nagashima K, Freed EO (2004) Phosphatidylinositol (4,5) 52. Anderson EC, Lever AM (2006) Human immunodeficiency virus type 1 Gag bisphosphate regulates HIV-1 Gag targeting to the plasma membrane. Proc Natl Acad polyprotein modulates its own translation. J Virol 80:10478–10486. Sci USA 101:14889–14894. 53. Wang H, Norris KM, Mansky LM (2003) Involvement of the matrix and nucleocapsid 24. Chukkapalli V, Hogue IB, Boyko V, Hu WS, Ono A (2008) Interaction between the domains of the bovine leukemia virus Gag polyprotein precursor in viral RNA fi human immunode ciency virus type 1 Gag matrix domain and phosphatidylinositol- packaging. J Virol 77:9431–9438. fi (4,5)-bisphosphate is essential for ef cient gag membrane binding. J Virol 82: 54. Parent LJ, et al. (2000) RNA dimerization defect in a Rous sarcoma virus matrix – 2405 2417. mutant. J Virol 74:164–172. fi 25. Saad JS, et al. (2008) Structure of the myristylated human immunode ciency virus 55. Garbitt RA, Albert JA, Kessler MD, Parent LJ (2001) trans-acting inhibition of genomic type 2 matrix protein and the role of phosphatidylinositol-(4,5)-bisphosphate in RNA dimerization by Rous sarcoma virus matrix mutants. J Virol 75:260–268. membrane targeting. J Mol Biol 382:434–447. 56. Ott DE, Coren LV, Gagliardi TD (2005) Redundant roles for nucleocapsid and matrix 26. Chan R, et al. (2008) Retroviruses human immunodeficiency virus and murine RNA-binding sequences in human immunodeficiency virus type 1 assembly. J Virol 79: leukemia virus are enriched in phosphoinositides. J Virol 82:11228–11238. 13839–13847. 27. Jefferies HB, et al. (2008) A selective PIKfyve inhibitor blocks PtdIns(3,5)P(2) 57. Crist RM, et al. (2009) Assembly properties of human immunodeficiency virus type 1 production and disrupts endomembrane transport and retroviral budding. EMBO Rep Gag-leucine zipper chimeras: Implications for retrovirus assembly. J Virol 83: 9:164–170. 2216–2225. 28. Stansell E, et al. (2007) Basic residues in the Mason-Pfizer monkey virus gag matrix 58. Zimmerman C, et al. (2002) Identification of a host protein essential for assembly of domain regulate intracellular trafficking and capsid-membrane interactions. J Virol immature HIV-1 capsids. Nature 415:88–92. 81:8977–8988. 59. Dooher JE, Schneider BL, Reed JC, Lingappa JR (2007) Host ABCE1 is at plasma 29. Shkriabai N, et al. (2006) Interactions of HIV-1 Gag with assembly cofactors. Biochemistry 45:4077–4083. membrane HIV assembly sites and its dissociation from Gag is linked to subsequent fi – 30. Alfadhli A, Barklis RL, Barklis E (2009) HIV-1 matrix organizes as a hexamer of trimers events of virus production. Traf c 8:195 211. on membranes containing phosphatidylinositol-(4,5)-bisphosphate. Virology 387: 60. Chatel-Chaix L, Boulay K, Mouland AJ, Desgroseillers L (2008) The host protein fi 466–472. Staufen1 interacts with the Pr55Gag zinc ngers and regulates HIV-1 assembly via its 31. Ono A, Orenstein JM, Freed EO (2000) Role of the Gag matrix domain in targeting N-terminus. Retrovirology 5:41. human immunodeficiency virus type 1 assembly. J Virol 74:2855–2866. 61. Adachi A, et al. (1986) Production of acquired immunodeficiency syndrome- 32. Nagai T, et al. (2002) A variant of yellow fluorescent protein with fast and efficient associated retrovirus in human and nonhuman cells transfected with an infectious maturation for cell-biological applications. Nat Biotechnol 20:87–90. molecular clone. J Virol 59:284–291. 33. Freed EO, Orenstein JM, Buckler-White AJ, Martin MA (1994) Single amino acid 62. Ott DE, et al. (2003) Elimination of protease activity restores efficient virion changes in the human immunodeficiency virus type 1 matrix protein block virus production to a human immunodeficiency virus type 1 nucleocapsid deletion mutant. particle production. J Virol 68:5311–5320. J Virol 77:5547–5556.

Chukkapalli et al. PNAS | January 26, 2010 | vol. 107 | no. 4 | 1605 Downloaded by guest on September 24, 2021