|Home | About | Journals | Submit | Contact Us | Français|
The HIV-1 Gag polyprotein precursor has multiple domains including nucleocapsid (NC). Although mature NC and NC embedded in Gag are nucleic acid chaperones (proteins that remodel nucleic acid structure), few studies include detailed analysis of the chaperone activity of partially processed Gag proteins and comparison with NC and Gag. Here we address this issue by using a reconstituted minus-strand transfer system. NC and NC-containing Gag proteins exhibited annealing and duplex destabilizing activities required for strand transfer. Surprisingly, unlike NC, with increasing concentrations, Gag proteins drastically inhibited the DNA elongation step. This result is consistent with “nucleic acid-driven multimerization” of Gag and the reported slow dissociation of Gag from bound nucleic acid, which prevent reverse transcriptase from traversing the template (“roadblock” mechanism). Our findings illustrate one reason why NC (and not Gag) has evolved as a critical cofactor in reverse transcription, a paradigm that might also extend to other retrovirus systems.
The human immunodeficiency type 1 (HIV-1) Gag polyprotein precursor is the only viral protein required for assembly of virus-like particles (VLPs) (reviewed in Freed, 1998; Swanstrom and Wills, 1997; Vogt, 1997). Monomeric Gag is a 55-kDa multi-domain protein, which contains (from the N- to C-terminus) matrix (MA), capsid (CA), spacer peptide 1 (SP1), nucleocapsid (NC), spacer peptide 2 (SP2), and p6 (Henderson et al., 1992) (reviewed in Adamson and Freed, 2007; Freed, 1998; Turner and Summers, 1999; Vogt, 1997) (Fig. 1A). Each of the domains contributes to the overall activity of Gag and to its unique role in HIV-1 replication. For example, the MA domain is responsible for Gag targeting to the plasma membrane during virus assembly and also binds RNA (Adamson and Freed, 2007; Chukkapalli et al., 2010; Shkriabai et al., 2006). CA is a protein interaction domain that facilitates Gag multimerization (Datta et al., 2007a,b; Gamble et al., 1997); the NC domain binds RNA, is required for genomic RNA packaging and primer placement (Darlix et al., 1995; Kleiman and Cen, 2004; Levin et al., 2005; Rein et al., 1998), has a role in viral RNA dimerization (Darlix et al., 1990; Feng et al., 1999; Liang et al., 1998), and interacts with the Bro 1 domain of the host protein Alix during virus assembly (Dussupt et al., 2009; Popov et al., 2009); and the p6 domain is needed for virus release and interaction with host factors in the ESCRT pathway (Adamson and Freed, 2007). During or shortly after budding of virions from the cell, the viral protease (PR) is activated and Gag is cleaved into the virus structural proteins. Cleavage occurs sequentially and in a highly ordered manner (Adamson and Freed, 2007; Freed, 1998; Swanstrom and Wills, 1997; Vogt, 1997).
The mature NC protein, which is released in a late cleavage reaction (Swanstrom and Wills, 1997), plays a major role in assuring the specificity and efficiency of reverse transcription (reviewed in Bampi et al., 2004; Darlix et al., 1995; Levin et al., 2005; Rein et al., 1998; Thomas and Gorelick, 2008) and is also important for other events in the virus life-cycle including maturation of the genomic RNA dimer (Feng et al., 1996; Muriaux et al., 1996) and integration of proviral DNA into the host genome (Carteau et al., 1999; Thomas et al., 2006). NC's function in virus replication is correlated with its ability to act as a nucleic acid chaperone. This activity allows NC to catalyze nucleic acid conformational changes that result in the most thermodynamically stable structures (Tsuchihashi and Brown, 1994; reviewed in Levin et al., 2005; Rajkowitsch et al., 2007; Rein et al., 1998). The NC domain in Gag also has nucleic acid chaperone activity and as mentioned above, its role in viral DNA synthesis is to mediate primer placement, i.e., annealing of the primer to the viral RNA genome, a reaction that occurs during virus assembly (Cen et al., 1999, 2000; Feng et al., 1999; Huang et al., 1997; Kleiman and Cen, 2004).
Previous studies of the chaperone activities of Gag and NC have not generally included detailed comparison with the activities of partially processed Gag proteins. To address this issue, we examined the activity of wild-type (WT) and mutant Gag and Gag-derived proteins to determine how chaperone activity is affected by removing one or more domains from Gag. Experimentally, our approach was to use our reconstituted minus-strand transfer assay system, which models the NC-dependent transfer of (–) strong-stop DNA [(–) SSDNA] to the 3′ end of the viral RNA genome, in a reaction facilitated by annealing of the complementary repeat regions in the RNA and DNA substrates (Basu et al., 2008; Levin et al., 2005; Thomas and Gorelick, 2008). This system represents a rigorous test for chaperone function (Guo et al., 1997; Heilman-Miller et al., 2004) and can also be used to examine the nucleic acid chaperone activity of proteins other than HIV-1 NC (Zúñiga et al., 2010).
Here, our results demonstrate that only Gag and partially processed Gag proteins containing the NC domain have nucleic acid chaperone activity. Although the activities of these proteins share some properties in common with mature NC, there are also fundamental differences. Thus, like NC, Gag and Gag-derived proteins exhibit annealing and duplex destabilizing activities that are required for minus-strand transfer. Surprisingly, as the protein concentration is increased, NC stimulates strand transfer, whereas Gag proteins drastically inhibit strand transfer (specifically, the DNA elongation step) by apparently blocking reverse transcriptase (RT) movement along the template. We refer to this inhibition as the “roadblock mechanism”. Moreover, endogenous RT (ERT) assays of minus-strand transfer with detergent-treated HIV-1 virions having mutations at one or more PR cleavage sites (Wyma et al., 2004) give results consistent with data obtained with the reconstituted system. Collectively, our studies provide additional insights into the question of why mature NC (and not a precursor) has evolved as a critical cofactor that facilitates efficient and specific HIV-1 reverse transcription.
In the present study, we set out to determine functional differences between the nucleic acid chaperone activities of NC embedded within larger precursors and mature NC. Our approach was to investigate the activities of unprocessed Gag and partially processed Gag proteins and to compare these activities with that of NC (Figs. 1A and B). For this work, we used our reconstituted minus-strand transfer assay system (Fig. 1C), which represents an especially sensitive readout for chaperone function (Guo et al., 1997; Heilman-Miller et al., 2004). This system measures (i) annealing of the RNA 148 acceptor to (–) SSDNA (i.e., DNA 128), which requires nucleic acid aggregation activity as well as helix destabilization of the highly structured complementary trans-activation response element (TAR) stem-loops at the 3′ ends of the repeat regions in (–) SSDNA and acceptor RNA; (ii) RT-catalyzed elongation of annealed (–) SSDNA, using the 54-nt U3 region present in the acceptor as the template, to give a 182-nt transfer product; and (iii) a competing reaction, known as self-priming (SP), in which fold-back structures at the 3′ end of (–) SSDNA (induced by the presence of the DNA TAR stem-loop) are extended by RT, forming a heterogeneous mixture of dead-end SP products (Beltz et al., 2005; Driscoll and Hughes, 2000; Guo et al., 1997, 2000; Heilman-Miller et al., 2004; Lapadat-Tapolsky et al., 1997; Levin et al., 2005 and references therein). In these experiments, “Gag” refers to the protein known as GagΔp6 (Campbell and Rein, 1999).
To evaluate the chaperone activities of HIV-1 Gag and Gag WM, a mutant with poor protein dimerization activity (Datta et al., 2007a,b), we tested the effect of increasing protein concentration on minus-strand transfer (Fig. 2). A similar set of reactions with HIV-1 NC was included as a positive control. The strand transfer and SP products were resolved by polyacrylamide gel electrophoresis (PAGE) (Fig. 2A) and were quantified by PhosphorImager analysis of the gel data (Figs. 2B and C, respectively). SP products were identified by comparison with the bands formed in a control reaction (C), which lacks acceptor and NC (Guo et al., 1997; Heilman-Miller et al., 2004).
In accord with previous results (Guo et al., 1997; Heilman-Miller et al., 2004), increasing concentrations of NC dramatically stimulated synthesis of the strand transfer product and severely inhibited SP (Figs. 2B and C, compare lanes 2 to 5 with lane 1). Gag and Gag WM also exhibited strand transfer activity in a dose-dependent manner, but significant activity occurred over a very narrow range (0.08–0.12 μM) (lanes 9 and 10 and lanes 16 and 17, respectively). At Gag and Gag WM concentrations that support strand transfer (i.e., up to 0.12 μM), inhibition of SP was more modest than that observed with NC (Fig. 2C), implying that the duplex destabilizing activity of Gag is not as effective as that of NC. At higher concentrations of Gag proteins, SP was drastically reduced, but as explained below, this is due to a general reduction in RT-catalyzed DNA extension. Gag WM facilitated strand transfer only slightly more efficiently than Gag (Fig. 2B, compare lanes 16 and 17 with lanes 9 and 10). This would suggest at first glance that the oligomeric status of these proteins might be the same under the conditions of this assay (for further discussion of this point, see below.)
Comparison of the NC, Gag, and Gag WM results showed that a higher concentration of NC was needed to achieve maximal strand transfer than was required for the Gag proteins (3.2 μM NC vs. 0.12 μM Gag or Gag WM). However, the percent strand transfer with NC was ~1.5-fold higher than the values obtained for reactions with the two Gag proteins. Curiously, in contrast to NC's behavior (Fig. 2B, lanes 4 and 5), when the concentration of Gag or Gag WM was raised, e.g., to 0.23 μM (Fig. 2B, lanes 11 and 18, respectively), there was a substantial reduction in strand transfer activity. At 0.46 μM, strand transfer activity was completely inhibited (Fig. 2B, lanes 12 and 19, respectively). In additional reactions with 0.46 μM Gag and increasing concentrations of RT from 20 to 50 nM (in our assay we use 10 nM), strand transfer was slightly increased, but not to an appreciable extent (e.g., 20 nM RT, <10%; 50 nM RT, 15%) (data not shown).
Since the minus-strand transfer assay requires both annealing and DNA elongation for a positive readout (Fig. 1C), it was important to determine which of these steps might be affected by increased concentrations of Gag and Gag WM. We therefore measured annealing alone as a function of protein concentration (Fig. 3). The data showed that for NC, Gag, and Gag WM, increasing the protein concentration resulted in a greater extent of annealing. In contrast to the strand transfer results, annealing with both Gag proteins did not lead to a reduction in the percent annealed product, even at the concentration (0.46 μM) at which strand transfer was completely abolished (Fig. 2). Moreover, heat annealing of RNA 148 to DNA 128 followed by addition of increasing concentrations of Gag plus a constant amount of RT led to results similar to those presented in Fig. 2B, i.e., inhibition of strand transfer (data not shown).
Taken together, these findings clearly demonstrate that the observed inhibition of strand transfer is caused by an inhibition of RT-catalyzed DNA extension and not annealing. Thus, when strand transfer was inhibited (concentrations of Gag and Gag WM above 0.12 μM), the reduction in SP (Fig. 2C, lanes 12 and 19) was due to a general reduction of polymerase activity and not to the helix destabilizing activity of the Gag proteins. In addition, as will be discussed in detail below, the data suggest that binding of Gag proteins to single-stranded (ss) nucleic acids creates a “roadblock”, which significantly reduces the ability of RT to traverse the template.
The results of Fig. 3 also show that with NC and the two Gag proteins, the annealing reaction reached an end point of ~60% in each case. However, the amount of Gag or Gag WM required for maximal activity (0.16 μM) was considerably lower than the amount of NC needed (between 1.6 and 3.2 μM). To account for this difference, we used FA to measure the nucleic acid binding affinity of Gag and NC to ss 20-nt DNA and RNA oligonucleotides (Table 1). In accord with the data of Fig. 3, we found that under our conditions, Gag was bound to nucleic acid ~15-fold more efficiently than NC (Table 1), e.g., the apparent dissociation constant (Kd) values for binding to the DNA oligonucleotide were 9 ± 1 nM (Gag) vs. 143 ± 12 nM (NC). The Kd values for Gag WM (with DNA, 20 ± 0.1) were similar to those of Gag, as expected from the data of Fig. 3. Note that in each case, the Kd values for binding to the RNA and DNA substrates were very similar.
Like NC, the MA domain in Gag also binds nucleic acids (Adamson and Freed, 2007; Vogt, 1997). It was therefore of interest to ascertain whether the MA domain contributes to Gag nucleic acid chaperone activity. To address this question, we initially assayed the strand transfer activity of three proteins lacking the NC domain (Fig. 4A), i.e., MA (lanes 1 to 5), CA (lanes 6 to 10), and MACA (lanes 11 to 15). MA and CA had no detectable nucleic acid chaperone activity at concentrations similar to those used for Gag and Gag WM. MACA had a very small stimulatory effect at the highest concentrations of the protein (lanes 14 and 15), but the level of activity was too low to be considered significant.
We also performed an assay to see if these proteins added in trans might affect NC-mediated strand transfer activity. NC (0.92 μM) was added to reactions containing increasing amounts of the three proteins (Fig. 4B). The first bar in each panel (lanes 1, 6, and 11) represents activity with NC alone. NC activity was unchanged in the presence of the three proteins, although in the case of MA and to a lesser extent MACA, there was a very small reduction in NC activity when the concentrations of these two proteins were raised to 0.46 μM (lanes 5 and 15).
Next we assayed the activity of two proteins that contain the NC domain: (i) GagΔ16-99, which lacks a region in MA enriched with basic residues; and (ii) CANC, which is missing all of MA (Fig. 1A). The results in Fig. 5A demonstrate that both of these proteins retained the ability to promote minus-strand transfer in a dose-dependent manner, although the most significant activity was in a narrow range of protein concentration. Thus, for GagΔ16-99, maximal activity was between 0.23 and 0.46 μM (lanes 6 and 7); this activity was somewhat lower than that achieved with the optimal concentration for Gag (0.12 μM) (Fig. 2B, lane 10). At a higher concentration of GagΔ16-99 (0.92 μM) (lane 8), the reduction in strand transfer activity was modest, unlike the complete abolition of this activity with 0.46 μM Gag (Fig. 2B, lane 12). The CANC protein behaved more like Gag and stimulated maximal strand transfer between 0.12 and 0.23 μM (lanes 13 and 14). Moreover, adding higher concentrations of CANC (0.46 and 0.92 μM, lanes 15 and 16 respectively) completely eliminated strand transfer activity. In contrast to Gag or Gag WM, neither the CANC nor the Gag Δ16-99 mutant was able to inhibit SP (Fig. 5B). Since minus-strand transfer is inhibited at 0.46 and 0.92 μM CANC, the reduction in SP products at these concentrations (lanes 15 and 16) probably reflects a general reduction in polymerase activity.
Collectively, the data in Figs. 4A and and55 demonstrate that the major determinant for nucleic acid chaperone activity is the NC domain in WT Gag and truncated Gag-derived proteins. While the MA domain appears to contribute to chaperone activity in the context of Gag, the absence of MA in an NC-containing partially processed Gag protein such as CANC has a relatively small effect on strand transfer activity in our system.
To further quantify differences in the nucleic acid chaperone activities of Gag, CANC, and GagΔ16-99, we investigated the kinetics of minus-strand transfer and annealing in a series of reactions containing three different concentrations of each protein (0.06, 0.12, and 0.23 μM) (Fig. 6). The optimal concentrations were 0.12 μM for Gag (Fig. 6A–1) and CANC (Fig. 6B–1), whereas for GagΔ16-99, it was 0.23 μM (Fig. 6C–1). Note that maximal strand transfer (about 40%) was achieved by 30 min with these concentrations and remained unchanged with further incubation for as long as 240 min (data not shown).
The rates of minus-strand transfer were calculated from the experimental data for reactions with 0.12 μM (Table 2, middle column). Of the three proteins, Gag had the highest kobs value (line 1). The rate for CANC was only slightly reduced (1.3-fold) relative to the rate for Gag (compare lines 1 and 2). In contrast, the rate for GagΔ16-99 (line 3) was 5-fold lower than the rate for Gag (compare lines 1 and 3). However, at the optimal concentration for GagΔ16-99 (0.23 μM), the rate of strand transfer was calculated as 0.125 ± 020 min−1, i.e., 1.7-fold lower than the rate for Gag and 1.4-fold lower than the rate for CANC. These results indicate that when reactions contained optimal concentrations for each protein, the rates were similar.
We also assayed the kinetics of annealing alone for each of the three proteins. In each case, a concentration of 0.23 μM gave the greatest amount of annealed product: almost 70% for Gag and CANC (Figs. 6A-2 and B-2, respectively) and over 60% for GagΔ16-99 (Fig. 6C-2). In contrast to the strand transfer results, the maximal annealing levels were reached within 1 to 5 min. At a lower concentration, 0.12 μM, the Gag reaction was significantly more efficient than the CANC and GagΔ16-99 reactions.
The rates of minus-strand annealing were determined at a concentration of 0.12 μM (Table 2, right column), since at a higher concentration, the kinetics for Gag annealing were too fast to measure (Fig. 6A-2). Interestingly, for all three proteins, the rates of annealing were significantly higher than the rates of strand transfer (Table 2, compare middle column with right column). Gag had the highest rate of annealing among the three proteins (Table 2, right column, line 1). The rates for CANC and GagΔ16-99 were 2-fold and 5.4-fold lower, respectively, than the rate for Gag (Table 2, right column, compare line 1 with lines 2 and 3).
Collectively, these results demonstrate that raising the protein concentration of Gag leads to increased stimulation of both the rate and final level of annealing, consistent with the end point data in Fig. 3 and in contrast with the strand transfer data (Figs. 2A and B). Although the rates for strand transfer were lower than the rates for annealing, it is interesting that in both assays, the ratio of the rate for Gag to the rates for either CANC and GagΔ16-99 were very similar, with activity ranked as Gag>CANC>GagΔ16-99 (Table 2).
In addition to assays of Gag and Gag-derived proteins in our reconstituted system, it was of interest to measure the activity of such proteins in a viral setting. A series of HIV-1 mutants having mutations at specific PR cleavage sites in Gag was constructed by the Aiken group (Wyma et al., 2004). These mutants include (i) MA/CA, (ii) MA/SP1, (iii) MA/NC, and (iv) MA/p6 (uncleaved Gag) (Fig. 7). The absence of an arrow at a site present in WT Gag means that cleavage is blocked at that site in the mutant (Fig. 7). Note that these mutations do not affect cleavage of the Pol moiety in Gag-Pol and RT is expressed as the mature protein.
Initial efforts to analyze synthesis of viral DNA products in cells infected with these virions were unsuccessful, since the amounts of DNA detected were very low and therefore highly variable. However, it was possible to perform ERT assays with detergent-treated virions and to measure minus-strand transfer efficiency, i.e., the amount of extended minus-strand DNA (U3-U5) divided by the amount of (–) SSDNA (R-U5), by qPCR analysis of the DNA products (Table 3). Interestingly, for those mutants able to generate mature NC, i.e., MA/CA and MA/SP1, minus-strand transfer efficiency essentially matched that observed for WT. The MA/NC virus, which contains uncleaved NC, had 55% of WT activity. MA/p6 (uncleaved Gag) exhibited ~30% of WT activity, presumably reflecting a low level of strand transfer facilitated by Gag. This is consistent with our finding in the reconstituted system that under some conditions, small amounts of the strand transfer product could be detected in reactions containing Gag and an increased concentration of RT (data not shown). In addition, some of the Gag activity could be due to the fact that in addition to uncleaved Gag, mutant virions were shown to contain some MA/NC (Wyma et al., 2004). (All of the other cleavage mutants contain only the predicted proteins (Wyma et al., 2004).) The negative control for these experiments was the viral mutant PRD25G (Ott et al., 2000) with a mutation in the PR active site.
Taken together, these results demonstrate that Gag and partially processed Gag proteins with the NC domain are able to promote minus-strand transfer in detergent-treated virus particles as well as in a reconstituted system with purified components. In virions where mature NC could be formed, activity was not affected by the presence of a partially processed Gag fragment, consistent with the results of Fig. 4B.
The goal of the present study was to determine the effect on the nucleic acid chaperone activity of HIV-1 NC when it is embedded in Gag or partially processed Gag proteins. A positive result in the assay we use (Fig. 1C) is dependent upon the characteristic activities of a nucleic acid chaperone: (i) ability to facilitate annealing and weak destabilization of nucleic acid secondary structure and (ii) rapid nucleic acid binding/dissociation kinetics (Cruceanu et al., 2006a,b; Darlix et al., 2007; Levin et al., 2005; Rein et al., 1998). Here, we show for the first time, that Gag has both annealing and duplex destabilization activities since it is able to stimulate strand transfer and reduce SP. NC reaches a higher end point value for strand transfer and suppresses SP more efficiently than Gag, suggesting that NC's duplex destabilization activity is more effective than that of the precursor protein. However, relatively high concentrations of NC are required for maximal activity (Fig. 2).
The results also demonstrate that the NC domain in Gag is crucial for its chaperone function in strand transfer, in accord with studies on primer placement with Gag (Cen et al., 2000; Feng et al., 1999; Guo et al., 2009) and Gag-derived proteins (Chan et al., 1999; Roldan et al., 2005). In vitro, mature NC also promotes primer placement (Cen et al., 2000; Feng et al., 1999; Hargittai et al., 2001; Iwatani et al., 2007; Prats et al., 1988; Rong et al., 1998; Tisné et al., 2001); for more detailed discussion and references, see Levin et al., 2005). Thus, CANC or GagΔ16-99, which have an NC domain, but are lacking all or a large portion of MA, respectively, facilitate strand transfer (Figs. 5 and and6),6), whereas proteins without an NC domain, e.g., MA, CA, and MACA do not (Fig. 4). The more effective chaperone activity of Gag in comparison to GagΔ16-99 in our assay system suggests that in the context of WT Gag, the NC domain binds and concentrates the nucleic acid, which allows additional binding by MA, thereby enhancing chaperone activity (Datta et al., 2007a). Studies of Gag WM suggest that it adopts a variety of compact conformations, allowing the MA and NC domains to be near each other (Datta et al., 2007a). More recent work provides evidence that nucleic acid interaction with highly basic residues in the MA domain of Gag, in addition to interactions with the NC domain, regulates Gag binding to lipid membranes (Chukkapalli et al., 2010). The difference in the activities of Gag and CANC might be due to the lack of MA in CANC and/or different folds of the CA N-terminal domain (NTD) in the two proteins (von Schwedler et al., 2003).
In reactions that only measure the annealing step in minus-strand transfer, Gag, Gag WM, and NC all reach the same end point, but a much higher concentration of NC is needed compared with the concentrations of the other two proteins (Fig. 3). This observation is consistent with FA analysis showing that Gag and Gag WM have significantly greater affinity for binding to ssDNA or RNA than NC (Table 1) and also with earlier data (Cruceanu et al., 2006b; Roldan et al., 2004; Roldan et al., 2005). Interestingly, both the extent and rate of Gag annealing are also higher than the corresponding values for CANC and GagΔ16-99 (Fig. 6; Table 2). This difference in annealing activity may account, at least in part, for the fact that these two proteins are unable to inhibit SP (Fig. 5), which occurs only when (–) SSDNA exists in an unannealed form (Guo et al., 1997). Note that in both the annealing and strand transfer assays the order of activity is Gag > CANC > GagΔ16-99.
A major finding of this study is the observation that when the concentrations of Gag or Gag WM are ≥0.23 μM, DNA elongation is significantly reduced or abolished completely (Fig. 2). This phenomenon was also observed with CANC and GagΔ16-99, although the effect was less dramatic with the deletion mutant (Fig. 5). Indeed, a common feature of proteins exhibiting concentration-dependent inhibition of polymerization is that they all contain the CA, SP1, and NC domains. Interestingly, the blocking effect is correlated with the behavior of Gag in single-molecule DNA stretching experiments (Cruceanu et al., 2006b), a technique that gives information on nucleic acid binding kinetics (Williams et al., 2009). In studies of Gag and NC, Cruceanu et al. (Cruceanu et al., 2006a,b) demonstrated that NC has a high on-off rate, whereas Gag, once bound to the ssDNA or RNA template, dissociates very slowly. Indeed, even with relatively low concentrations of Gag, it can act as a roadblock to DNA extension, whereas fully processed NC allows RT to continue elongating the DNA or RNA primer (Figs. 2A and B). Thus, Gag lacks a critical property of highly functional nucleic acid chaperones such as HIV-1 NC (Cruceanu et al., 2006a). Recently, it was reported that the poor chaperone activity of HTLV-1 NC is also correlated with slow dissociation kinetics (Qualley et al., 2010). In addition, Iwatani et al. (Iwatani et al., 2007) found that a roadblock mechanism could explain the deaminase-independent inhibitory effect of APOBEC3G on DNA polymerization reactions catalyzed by HIV-1 RT (Bishop et al., 2008; Iwatani et al., 2007).
These considerations led us to ask: (i) What causes Gag's slow dissociation from bound nucleic acid and the resulting roadblock to RT-catalyzed DNA extension? and (ii) Why are there negligible differences between GagΔp6 and Gag WM activity in the strand transfer assay? In addressing these questions, it is important to note that in the absence of nucleic acid, Gag multimerizes at protein concentrations in the μM range, whereas 95% of Gag WM is monomeric (Datta et al., 2007a). However, in the presence of nucleic acid (the condition that prevails in our assays), both Gag and Gag WM bind to nucleic acid with similar affinity (Table 1) and can engage in specific protein–protein interactions, mediated by the CA domain (Datta et al., 2007a; Stephen et al., 2007). We refer to this as “nucleic acid-driven multimerization”. Indeed, the fact that Gag WM can assemble into VLPs in vivo, albeit with lower efficiency and higher heterogeneity when bound to nucleic acid (Datta et al., 2007a), suggests that this protein can recapitulate basic protein–protein interactions necessary for VLP formation. Thus, both Gag and Gag WM can effectively inhibit strand transfer by the roadblock mechanism. In principle, such a mechanism could inhibit premature initiation of reverse transcription. However, at present, it is not known whether RT is released from Gag-Pol before PR cleavage of Gag or whether HIV-1 Gag-Pol can catalyze reverse transcription in the infected cell.
In earlier work, Roldan et al. (2005) showed that when a -viral RNA duplex was annealed by GagΔp6 or a mini-Gag protein, which in each case remained in the reaction mixtures throughout the incubation, elongation of the tRNA primer was either completely or almost completely inhibited, respectively. In their system, DNA extension following annealing was more efficient with a mini-Gag dimerization mutant than with WT mini-Gag (Roldan et al., 2005), a result that contrasts with our finding that Gag and Gag WM have very similar activities (Figs. 2 and and3).3). Gag and mini-Gag can each interact with nucleic acids, resulting in protein–protein interactions. However, it is likely that the mini-Gag–mini-Gag interactions are not as stable as the Gag–Gag interactions observed in our study, since the mini-Gag construct is missing the NTD of CA, which contributes to stabilization of the protein multimers. Thus, the monomeric mini-Gag mutant has greater activity than the WT protein, whereas the strength of the Gag–Gag interactions even with Gag WM is sufficient to block DNA elongation.
Another issue that we address is the behavior of Gag and partially processed Gag proteins in our reconstituted strand transfer system and the effects of such proteins in a viral setting. For this work, we took advantage of a series of HIV-1 viral mutants that are blocked at one or more Gag cleavage sites and do not have conical cores (Wyma et al., 2004). Since attempts to assay viral DNA products synthesized in infected cells were unsuccessful, we chose to measure ERT activity in detergent-treated virions (Fig. 7; Table 3). In this assay, the lack of conical cores per se does not block reverse transcription (Kaplan et al., 1994; Tang et al., 2001) (Thomas, JA. and Gorelick, R.J., unpub. obs.). The results of the ERT assays are consistent with the data from the reconstituted system. In particular, we note that transdominant effects of partially processed Gag proteins, which could potentially inhibit NC activity in our experiments, are not observed in either assay (Fig. 4B; Table 3), in contrast to the situation with virus-infected cells (Checkley et al., 2010; Lee et al., 2009; Müller et al., 2009; Rulli et al., 2006). Thus, transdominant inhibition of viral DNA synthesis by Gag cleavage proteins occurs only in a cell-based system where a block in Gag cleavage is closely linked to (i) the assembly of aberrant viral cores and/or (ii) improper uncoating of cores following virus entry into the cell.
In summary, we have demonstrated that there is a major difference between the nucleic acid chaperone activities of NC embedded in HIV-1 Gag and Gag-derived proteins and those of mature NC. Although all of these proteins have annealing and helix destabilizing activities, the activity of Gag proteins is less effective than that of mature NC. Thus, unlike NC, Gag dissociates slowly from bound nucleic acid, a consequence of nucleic acid-driven Gag multimerization that occurs even at relatively low protein concentrations. This in turn leads to drastic inhibition of DNA elongation. We refer to this inhibition as the roadblock mechanism. It will be interesting to determine whether this mechanism is used more generally by other retroviruses. In conclusion, our data help to explain why NC and not Gag is a critical cofactor for viral DNA synthesis and illustrate how the presence of NC in a multi-domain protein like Gag modulates the nature of its nucleic acid chaperone activity.
DNA oligonucleotides were obtained from Lofstrand (Gaithersburg, MD) and Integrated DNA Technologies (Coralville, IA). T4 polynucleotide kinase, proteinase K, SUPERaseIN, and Gel Loading Buffer II were obtained from Applied Biosystems (Foster City, CA). [γ-33P] ATP (3,000 Ci/mmol) was purchased from PerkinElmer (Shelton, CT). All of the columns used for protein purification were purchased from GE Healthcare Life Sciences (Piscataway, NJ). HIV-1 RT was obtained from Worthington (Lakewood, NJ). The sequence of NC and the nucleic acids used in this study were derived from the HIV-1 pNL4-3 clone (GenBank accession no. AF324493) (Adachi et al., 1986). Sequences of the purified Gag and Gag cleavage proteins were derived from HIV-1 BH10 (GenBank accession no. M15654.1) (Campbell and Rein, 1999; Ratner et al., 1985), except for GagΔ16-99, which contains sequences from HIV-1 NL4-3 and BH10 (Spel site in CA to 3′ end is BH10) (Gross et al., 2000).
Gag and Gag cleavage products (Fig. 1A) were expressed in BL21 (DE3)pLysS cells (Stratagene). Cells expressing HIV-1 matrix (MA) were induced at 37 °C for 4 h with 1 mM isopropyl-beta-d-thiogalactoside and were lysed in buffer A (20 mM Tris–HCl, pH 7.4, 10 mM 2-mercaptoethanol, and 1 mM phenylmethylsulfonyl fluoride [Sigma-Aldrich]) containing 150 mM NaCl. After centrifugation at 12,000×g for 15 min to remove cellular debris, the MA protein was purified from the total lysate by taking a 40–70% ammonium sulfate cut. The protein was dialyzed against buffer A containing 150 mM NaCl. After dialysis, ammonium sulfate was added to 40% saturation and the protein was chromatographed on a Butyl Sepharose column. Fractions containing the protein were dialyzed against buffer A with 50 mM NaCl and were run on an SP® Sepharose column. The purified protein was stored at −80 °C in buffer A with 150 mM NaCl and 10% Glycerol. HIV-1 CA and MACA were purified as described in references (von Schwedler et al., 1998; Yoo et al., 1997), respectively.
Since HIV-1 Gag expressed in E. coli is extensively degraded by bacterial proteases (Campbell and Rein, 1999), the stably expressed protein known as GagΔp6 was used instead of authentic Gag in assays for minus-strand transfer. Note that in addition to missing the p6 domain, GagΔp6 is not myristoylated at its N-terminus (Campbell and Rein, 1999). GagΔp6 proteins with additional changes were as follows: (i) GagΔ16-99, missing residues 16–99 in the MA domain (Datta and Rein, 2009; Facke et al., 1993; Gross et al., 2000) and (ii) Gag WM, GagΔp6 with Trp316 and Met317 (i.e., Trp184 and Met185 in CA, respectively) changed to Ala, which causes a defect in Gag dimerization (Datta et al., 2007a,b). GagΔp6, the two GagΔp6-derived proteins, and CANC (Campbell and Vogt, 1995) were purified by phosphocellulose affinity chromatography as described previously (Datta et al., 2007a; Datta and Rein, 2009). In the final purification step, the proteins were subjected to further chromatography on Superose 12 (GagΔp6, GagΔ16-99, Gag WM, CANC) or Superdex 75 (MA, MACA, CA) gel filtration columns to eliminate RNase and DNase contamination. Analysis of the Gag and Gag-derived protein preparations by SDS–PAGE indicated that the proteins were at least 90% pure (Fig. 1B). The residue numbers in Gag and the respective molecular mass values are given in the legend to Fig. 1B. Recombinant HIV-1 NC was prepared as described previously (Carteau et al., 1999; Wu et al., 1996).
The minus-strand transfer assay was performed as described previously (Heilman-Miller et al., 2004) with several changes. Since maximal activity was obtained at 1 mM Mg2+ in reactions with either NC (Wu et al., 2007) or Gag (data not shown), we used 1 mM Mg2+ in all of the polymerase reactions. In addition, 0.5 U of SUPERaseIN was added per 20-μl reaction. HIV-1 NC or Gag and Gag-derived proteins were added at the concentrations specified in the legends to the figures. Reaction mixtures (final volume, 20 μl) also contained DNA 128 ((–) SSDNA) labeled at its 5′ end with 33P, using the 32P labeling protocol described in Guo et al. (1995), RNA 148 (acceptor RNA), and HIV-1 RT, each at a final concentration of 10 nM. Incubation was at 37 °C for the indicated times. Termination of the reactions, electrophoresis in denaturing gels, and PhosphorImager analysis were performed as described (Wu et al., 2007). In addition to the transfer product, dead-end products caused by SP of fold-back structures at the 3′ end of (–) SSDNA (Beltz et al., 2005; Driscoll and Hughes, 2000; Guo et al., 1997, 2000; Heilman-Miller et al., 2004; Lapadat-Tapolsky et al., 1997; Levin et al., 2005) and references therein) were also formed. The percentage of minus-strand transfer product synthesized was calculated by dividing the amount of transfer product by total DNA (transfer product plus SP products and remaining DNA 128), multiplied by 100. The percentage of SP products synthesized was calculated in a similar manner. The data for strand transfer and for annealing (see below) represent the average of results obtained in at least three independent experiments. Error bars shown in the figures represent the standard deviation.
33P-labeled DNA 128 (0.2 pmol) was incubated at 37 °C with 0.2 pmol of RNA 148 in buffer containing 50 mM Tris–HCl (pH 8.0) and 75 mM KCl for the indicated times in the absence or presence of Gag proteins (final volume, 20 μl), as described (Wu et al., 2007). The final concentration of the nucleic acid substrates was 10 nM.
Equilibrium binding of HIV-1 NC and Gag to a 20-nt AlexaFluor-488-labeled ss DNA oligonucleotide (JL936, 5′-Alex488-AGCTGCTTTTTGCCTGTACT-3′) and a 20-nt fluorescein-labeled ssRNA oligonucleotide (JL943, 5′-Fl-AGCUGCUUUUUGCCUGUACU-3′) was measured using FA. The oligonucleotides contained the 20-nt ssRNA or DNA sequence from the extreme 3′ terminus of U3. JL936 (purified by high performance liquid chromatography) and JL943 (purified by PAGE) were obtained from Integrated DNA Technologies (Coralville, IA) and Dharmacon, Inc. (Lafayette, CO), respectively. FA measurements were performed in Corning® 384-well low volume black polystyrene NBS™ microplates (Corning, NY) using a SpectraMax® M5 multimode microplate reader (Molecular Devices, Sunnyvale, CA). JL936 or JL943 (each 20 nM) were incubated with increasing concentrations of NC or Gag in 50 mM Tris–HCl (pH 8.0), 75 mM KCl, 1 mM MgCl2, and 1 mM DTT. Samples were excited at 485 nm and emission intensities at both parallel and perpendicular planes were collected at 525 nm. The resulting plot of anisotropy vs. protein concentration was fit using a one-site binding model (Iwatani et al., 2007) to obtain the Kd value.
Viruses were derived by transfection of 293 T cells in 100-mm culture dishes, using the CaPO4/DNA coprecipitation method (Graraham and van der Eb, 1973), as described previously (Thomas et al., 2008). Plasmids for the production of the MA/CA, MA/SP1, MA/NC, and MA/p6 mutant viruses (Wyma et al., 2004) were a generous gift from Christopher Aiken, Vanderbilt University Medical Center, Nashville, TN. The PR mutant, PRD25G, was a generous gift from David Ott, SAIC-Frederick, Inc., NCI-Frederick, Frederick, MD (Ott et al., 2000). The WT HIV-1 plasmid pNL4-3 (Adachi et al., 1986) was obtained from Malcolm Martin (NIAID, NIH) through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH.
Viral supernatant fluids (30 ml) were treated with DNase I and were then pelleted through a 5-ml 20% (w/v) sucrose cushion (in PBS without Ca2+ or Mg2+) in a Beckman SW32 rotor at 144,000×g for 1 h at 4 °C. The pellets were resuspended and subjected to ERT as described previously (Thomas et al., 2008). Reverse transcription products, R-U5 (representative of (–) SSDNA products), and U3-U5 (representative of minus-strand transfer products) were quantified by qPCR, as described previously (Buckman et al., 2003).
We thank Christopher Aiken for the Gag cleavage mutants, David Ott for the HIV-1 PR mutant PRD25G, Malcolm Martin for the pNL4-3 clone, which was obtained from the NIH AIDS Research and Reference Reagent Program, and Hans Luecke, who generously allowed us to use his Spectromax M5 instrument for the Kd determinations. We are also indebted to Christopher Jones, Ioulia Rouzina, and Karin Musier-Forsyth for valuable discussion and for sharing unpublished data. This work was supported by the Intramural Research Program of the NIH (Eunice Kennedy Shriver National Institute of Child Health and Human Development [J.G.L] and the National Cancer Institute, Center for Cancer Research [A.R.]) and was funded in part with federal funds from the National Cancer Institute under contract numbers N01-CO-12400 and HHSN261200800001E (R.J.G.). The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. government.
Authors Judith G. Levin and Alan Rein wish to dedicate this paper to the memory of Brenda I. Gerwin, a dear friend and colleague.