ABSTRACT
The ability of human immunodeficiency virus type 1 (HIV-1) to transduce nondividing cells is key to infecting terminally differentiated macrophages, which can serve as a long-term reservoir of HIV-1 infection. The mutation N57A in the viral CA protein renders HIV-1 cell cycle dependent, allowing examination of HIV-1 infection of nondividing cells. Here, we show that the N57A mutation confers a postentry infectivity defect that significantly differs in magnitude between the common lab-adapted molecular clones HIV-1NL4-3 (>10-fold) and HIV-1LAI (2- to 5-fold) in multiple human cell lines and primary CD4+ T cells. Capsid permeabilization and reverse transcription are altered when N57A is incorporated into HIV-1NL4-3 but not HIV-1LAI. The N57A infectivity defect is significantly exacerbated in both virus strains in the presence of cyclosporine (CsA), indicating that N57A infectivity is dependent upon CA interacting with host factor cyclophilin A (CypA). Adaptation of N57A HIV-1LAI selected for a second CA mutation, G94D, which rescued the N57A infectivity defect in HIV-1LAI but not HIV-1NL4-3. The rescue of N57A by G94D in HIV-1LAI is abrogated by CsA treatment in some cell types, demonstrating that this rescue is CypA dependent. An examination of over 40,000 HIV-1 CA sequences revealed that the four amino acids that differ between HIV-1NL4-3 and HIV-1LAI CA are polymorphic, and the residues at these positions in the two strains are widely prevalent in clinical isolates. Overall, a few polymorphic amino acid differences between two closely related HIV-1 molecular clones affect the phenotype of capsid mutants in different cell types.
IMPORTANCE The specific mechanisms by which HIV-1 infects nondividing cells are unclear. A mutation in the HIV-1 capsid protein abolishes the ability of the virus to infect nondividing cells, serving as a tool to examine cell cycle dependence of HIV-1 infection. We have shown that two widely used HIV-1 molecular clones exhibit significantly different N57A infectivity phenotypes due to fewer than a handful of CA amino acid differences and that these clones are both represented in HIV-infected individuals. As such minor differences in closely related HIV-1 strains may impart significant infectivity differences, careful consideration should be given to drawing conclusions from one particular HIV-1 clone. This study highlights the potential for significant variation in results with the use of multiple strains and possible unanticipated effects of natural polymorphisms.
INTRODUCTION
The HIV-1 capsid is a closed conical structure composed of hexamers and pentamers of the viral CA protein (1) that performs an essential set of functions in the early virus life cycle, shielding the viral RNA genome from cytoplasmic innate immune sensors, regulating the completion of reverse transcription, and governing use of nuclear trafficking and import pathways (2). Capsid has been shown to interact with numerous host factors, including cyclophilin A (CypA) (3), cleavage and polyadenylation specificity factor 6 (CPSF6) (4), the karyopherin transportin-3 (TNPO3) (5), and nucleoporins Nup153 (6) and Nup358 (7). Inhibition of these interactions impairs early virus life cycle events, including completion of reverse transcription, capsid uncoating, and nuclear trafficking and entry, and attenuates virus infectivity (8–18). Overall, HIV-1 CA is poorly tolerant of mutation, with most amino acid changes resulting in a dysfunctional capsid and severely impaired virus infectivity (19, 20), limiting the ability of HIV-1 to adapt to capsid-targeting therapies. The combination of its critical function, specific host factor interactions, and mutational fragility makes the capsid a desirable target for antiretroviral intervention (21, 22).
HIV-1 is a lentivirus that is able to transduce nondividing cells, which allows infection of terminally differentiated macrophages (23–25) and microglial cells (26). While CD4+ T cells are the principal target of HIV-1 infection, macrophages and microglia express CD4 and CCR5 (27, 28) and become HIV infected in lymphoid, mucosal, and central nervous system tissues (29, 30). While not as efficiently infected as CD4+ T cells (31), macrophages are less prone to the cytopathic effects of HIV-1 infection (32). Because of this combined with the widespread distribution of macrophages throughout the body and their long life span (33), macrophages can serve as a long-term HIV-1 reservoir (34, 35). It was previously shown that CA is the major viral determinant that permits HIV-1 infection of nondividing cells (36, 37), but the specific mechanism by which this occurs remains unclear.
The CA mutation N57A together with T54A renders HIV-1 cell cycle dependent in all cell types tested, unlike other CA mutants examined, which were cell cycle dependent only in certain cell types (38, 39). N57 resides within a pocket of the N-terminal domain formed by alpha helices 3, 4, and 5 (Fig. 1A). This pocket has been identified as the CA binding site for host factors CPSF6 (40) and Nup153 (41) and the CA binding compound PF3450074 (PF74) (42). Mutating the asparagine to an alanine (N57A) renders HIV-1 infection independent of CPSF6 (12, 40), TNPO3 (11), Nup153 (41), and Nup358 (7) and prevents binding to and restriction by PF74 (40, 43). N57A was originally created in combination with the cyclosporine (CsA)-dependent CA mutation T54A (44) as part of an alanine-scanning surface mutation panel (20). We reasoned that N57A could serve as a useful tool for examining the mechanism by which HIV-1 is able to infect nondividing cells independent of any cell type-specific attributes and sought to characterize the viral infectivity defect caused by N57A.
HIV-1 CA mutation N57A exhibits an infectivity defect that differs between HIV-1LAI and HIV-1NL4-3. (A) Protein Data Bank structure 3H47 of a portion of HIV-1 CA showing the location of N57 within the pocket formed by alpha helices 3, 4, and 5. (B) The indicated cell types were infected with equal amounts of WT or N57A vesicular stomatitis virus G (VSV-G)-pseudotyped, single-cycle HIV-1NL4-3 expressing luciferase and assayed after 48 h for luciferase activity. (C) The indicated cell types were infected with viruses from panel B with and without aphidicolin treatment. Dotted lines represent the average luciferase signal of uninfected cells. (D) The indicated cell types were infected with equal amounts of WT or N57A VSV-G-pseudotyped, single-cycle HIV-1LAI expressing GFP and assayed after 48 to 72 h for GFP expression. (E) The indicated cell types were infected with viruses from panel D with and without aphidicolin treatment. Dotted lines represent the average GFP signal of uninfected cells. ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05. Error bars indicate standard error of the mean (SEM) for 2 to 4 experiments.
Here, we present the results of the early life cycle infectivity defect of N57A HIV-1 and our observation of distinct differences in infectivity and capsid permeabilization phenotypes in multiple cell types when N57A is incorporated into two widely used and closely related lab-adapted virus strains, HIV-1NL4-3 and HIV-1LAI, that differ in only four amino acids within CA. Phenotypic differences that are dependent upon CA polymorphisms can be useful for elucidating early virus life cycle mechanisms but also raise the question of the applicability of observations with a single HIV-1 molecular clone.
RESULTS
HIV-1 CA mutation N57A exhibits an infectivity defect that differs between common lab-adapted strains HIV-1NL4-3 and HIV-1LAI.Based upon the previous observation that T54A/N57A HIV-1 was cell cycle dependent in all cell types tested (38), we sought to characterize N57A HIV-1 infectivity in different cell types. N57A HIV-1NL4-3 had a significant infectivity defect in the HeLa and GHOST cell lines and in primary human CD4+ T cells (33- to 100-fold) (Fig. 1B), which was further exacerbated in nondividing cells (300- to 1,000-fold) (Fig. 1C). This defect is similar to that observed by other groups in HIV-1NL4-3 (7, 40, 41). HIV-1NL4-3 and HIV-1LAI are two of the most widely used lab-adapted HIV-1 strains (45). Unexpectedly, when incorporated into HIV-1LAI, the infectivity defect caused by N57A was considerably attenuated in the HeLa and GHOST cell lines and in primary human CD4+ T cells (2- to 5-fold), though it nonetheless was significant compared to that of wild-type (WT) virus (Fig. 1D). N57A HIV-1LAI infectivity was further reduced in the presence of aphidicolin (65- to 150-fold) (Fig. 1E).
The magnitude of reduced N57A HIV-1 infectivity is CA dependent.HIV-1NL4-3 was originally constructed in 1986 as a chimera between the 5′ half of isolate NY5 (gag through most of vpr) and the 3′ half of isolate LAV (46), which is also known as LAI (47). To determine the cause of phenotypic differences of N57A in HIV-1NL4-3 and HIV-1LAI, two chimeric viruses were created by swapping the BssHII-to-ApaI restriction fragment between the NL4-3 and LAI reporter viruses (Fig. 2A). This fragment spans the majority of the gag gene, including MA, CA, SP1, and a portion of NC. The difference in infectivity of N57 and A57 in HIV-1NL4-3 encoding LAI Gag was 2- to 5-fold in the HeLa and GHOST cell lines and in primary human CD4+ T cells (Fig. 2B), similar to what was observed with HIV-1LAI (Fig. 1C). Conversely, the difference in infectivity of WT and N57A in HIV-1LAI encoding NL4-3 Gag was approximately 30-fold in MT-4 cells (Fig. 2C), similar to that observed with HIV-1NL4-3 (Fig. 1B) and in contrast to the 8-fold difference in infectivity of WT and N57A in HIV-1LAI encoding LAI Gag (Fig. 2C). Taken together, these data reveal that the difference in the levels of N57A HIV-1NL4-3 and N57A HIV-1LAI infectivity is encoded within gag.
The magnitude of reduced N57A HIV-1 infectivity is CA dependent. (A) Schematic of the chimeric reporter viruses constructed from HIV-1NL4-3 and HIV-1LAI. (B and C) The indicated cell types were infected with equal amounts of WT or N57A HIV-1NL4-3 luciferase reporter virus with LAI Gag (B) or HIV-1LAI GFP reporter virus with LAI or NL4-3 Gag (C) and assayed after 48 h for luciferase or GFP expression in MT-4 cells. (D) Amino acid sequences of the Gag fragment in HIV-1LAI and HIV-1NL4-3, with CA shown in red. (E) The indicated cell types were infected with equal amounts of WT or N57A HIV-1NL4-3 with LAI CA and assayed after 48 h for luciferase activity. (F) GHOST cells were infected with WT or N57A HIV-1NL4-3 bearing the indicated CA mutations and assayed after 48 h for GFP expression. ***, P < 0.001; **, P < 0.01; *, P < 0.05. Error bars indicate SEM for 2 to 4 experiments.
A total of 17 amino acids differ between the BssHII/ApaI gag fragments of HIV-1NL4-3 and HIV-1LAI (Fig. 2D). Only four amino acids vary within CA between HIV-1NL4-3 and HIV-1LAI: L6I, L83V, H120N, and G208A. To determine the role of these CA amino acid differences in HIV-1NL4-3 and HIV-1LAI on the infectivity phenotype of N57A, all four LAI residues were introduced into HIV-1NL4-3, generating HIV-1NL4-3 with LAI CA. As with the gag chimeric virus, the N57A infectivity defect in HIV-1NL4-3 with LAI CA phenocopied the defect in HIV-1LAI (Fig. 2E), demonstrating that the observed N57A infectivity phenotypes are attributable to four amino acid substitutions in CA between the two virus strains. To ascertain the specific residue(s) responsible for the strain-specific N57A infectivity phenotypes, CA mutant viruses were created in HIV-1NL4-3 containing the four CA substitutions alone or in combination and with or without the N57A mutation. The L83V CA mutation alone was sufficient to confer the attenuated HIV-1LAI N57A infectivity defect in HIV-1NL4-3 (Fig. 2F).
The magnitudes of N57A HIV-1 Gag-dependent infectivity defects correspond with strain-specific defects in capsid permeabilization and reverse transcription.Alterations in capsid uncoating, caused by inhibitors or CA mutations, lead to diminished HIV-1 infectivity (48). To characterize the nature of the infectivity defect caused by N57A in HIV-1NL4-3 and HIV-1LAI, we compared early capsid uncoating kinetics of WT and N57A HIV-1 in both virus strains, using our previously published capsid permeabilization assay (49, 50). In brief, HeLa cells were infected with HIV-1 in which the viral RNA was labeled with 5-ethynyl uridine (EU). Staining of the modified HIV-1 RNA was measured at different time points, which occurs after initial dissociation of the capsid. Capsid permeabilization of N57A HIV-1NL4-3 with LAI CA was similar to that of WT HIV-1NL4-3 with LAI CA (Fig. 3A). In contrast, the capsid permeabilization kinetics of N57A HIV-1NL4-3 were significantly different from those of WT HIV-1NL4-3, with viral RNA staining peaking at an earlier time point and diminishing more rapidly over time (Fig. 3B).
Gag-dependent infectivity defects caused by N57A occur before or after reverse transcription for HIV-1NL4-3 and HIV-1LAI, respectively. (A and B) HeLa cells were infected with equal amounts of WT or N57A HIV-1NL4-3 luciferase reporter virus with LAI CA (A) or NL4-3 CA (B) and assayed for capsid permeabilization at indicated time points. (C to E) HeLa cells were infected with equal amounts of viruses from panel A (C), WT or N57A HIV-1LAI GFP reporter virus (D), or viruses from panel B (E), and reverse transcription products and 2-LTR circles were measured. ***, P < 0.001; **, P < 0.01; *, P < 0.05. Error bars indicate SEM for at least 2 experiments.
Alterations in HIV-1 capsid uncoating often affect reverse transcription (51). Thus, we measured early and late reverse transcription products and two-long-terminal-repeat (2-LTR) circles in cells infected with HIV-1NL4-3, HIV-1NL4-3 with LAI CA, and HIV-1LAI. There were no appreciable differences in early or late reverse transcripts between WT and N57A HIV-1NL4-3 with LAI CA (Fig. 3C) or between WT and N57A HIV-1LAI (Fig. 3D); however, in HIV-1NL4-3, N57A early and late reverse transcripts were significantly diminished (7- to 10-fold) compared to WT (Fig. 3E). Similar to the infectivity data, there were 2- to 3-fold-lower levels of 2-LTR circles produced in cells infected with N57A HIV-1NL4-3 with LAI CA than in those infected with WT HIV-1NL4-3 with LAI CA (Fig. 3C) and in cells infected with N57A HIV-1LAI than in those infected with WT HIV-1LAI (Fig. 3D), but there were 20-fold-lower levels of 2-LTR circles in cells infected with N57A HIV-1NL4-3 than in those infected with WT HIV-1NL4-3 (Fig. 3E). Taken together, these data further demonstrate that the N57A HIV-1NL4-3 and N57A HIV-1LAI phenotypes are CA dependent and suggest that the nature of the infectivity defect caused by N57A differs when the mutation is in the two strains, with the mutation leading to an infectivity defect prior to reverse transcription in HIV-1NL4-3 and after reverse transcription in HIV-1LAI.
N57A HIV-1 infectivity is dependent upon capsid interaction with CypA.It was previously observed that N57A relieves the CsA dependence of T54A in HIV-1 (39), suggesting that N57A might influence the interaction between CA and the host factor CypA. To determine the effect of CypA binding on N57A HIV-1 infectivity, HeLa and GHOST cells were infected with WT and N57A HIV-1NL4-3 and WT and N57A HIV-1NL4-3 with LAI CA in the presence and absence of CsA treatment. While WT HIV-1NL4-3 (Fig. 4A) and WT HIV-1NL4-3 with LAI CA (Fig. 4B) showed only a minimal reduction in infectivity with CsA, infectivity was significantly reduced for both N57A HIV-1NL4-3 (Fig. 4C) and N57A HIV-1NL4-3 with LAI CA (Fig. 4D) in the presence of CsA. This was particularly noticeable for N57A in the context of HIV-1LAI CA, in which infectivity was reduced only 2- to 5-fold compared to WT virus in the absence of CsA but was reduced 10- to 30-fold compared to WT virus in the presence of CsA.
N57A HIV-1 infectivity is dependent upon capsid interaction with CypA. (A and B) The indicated cell types were infected with equal amounts of WT HIV-1NL4-3 (A) or HIV-1NL4-3 with LAI CA (B) with or without aphidicolin treatment and in the presence or absence of CsA treatment and assayed after 48 h for luciferase activity. (C and D) The indicated cell types were infected with WT or N57A HIV-1NL4-3 (C) or WT or HIV-1NL4-3 with LAI CA (D) in the presence or absence of CsA. (E and F) The indicated cell types were infected with WT or N57A HIV-1NL4-3 (E) or HIV-1NL4-3 with LAI CA (F) treated with aphidicolin and in the presence or absence of CsA. N57A HIV-1 infectivity is shown relative to the corresponding WT virus infectivity for each treatment. Dotted lines represent average luciferase signal of uninfected cells. *, P < 0.05. Error bars indicate SEM for 3 experiments.
Given the attenuated infectivity of N57A HIV-1 in nondividing cells (Fig. 1C), the impact of CsA treatment on WT and N57A HIV-1 infectivity was examined in growth-arrested HeLa and GHOST cells. As with dividing cells, WT HIV-1NL4-3 (Fig. 4A) and WT HIV-1NL4-3 with LAI CA (Fig. 4B) showed little change in infectivity of nondividing cells with the addition of CsA. The already low infectivity of N57A HIV-1NL4-3 in nondividing cells was not further reduced by CsA treatment in either cell line (Fig. 4E). In HIV-1NL4-3 with LAI CA, N57A infectivity in nondividing cells was reduced with CsA to various degrees depending on the cell type (Fig. 4F). In HeLa cells, the reduction in infectivity was only 2-fold. However, in GHOST cells, the reduced infectivity of N57A HIV-1NL4-3 with LAI CA was 10-fold, which was similar to what was observed in dividing GHOST cells. Thus, N57A HIV-1 infectivity is dependent upon CA binding of CypA, particularly in dividing cells, and more prominently with HIV-1LAI CA than with HIV-1NL4-3 CA.
Inhibiting CypA-capsid interaction restricts N57A HIV-1 infectivity at nuclear entry.To examine the mechanism through which CypA promotes N57A HIV-1 infectivity, we measured early and late reverse transcription products and 2-LTR circles in HeLa cells infected with WT and N57A HIV-1NL4-3 and WT and N57A HIV-1NL4-3 with LAI CA in the presence and absence of CsA treatment. For WT HIV-1NL4-3, early and late reverse transcripts were reduced 1.6-fold and 2.3-fold, respectively, and 2-LTR circles were reduced 1.4-fold with CsA treatment (Fig. 5A), corresponding to a 1.6-fold reduction in infectivity of WT HIV-1NL4-3 with CsA treatment (Fig. 4A). Similarly, early and late reverse transcripts were reduced 3-fold and 2-fold, respectively, during CsA treatment of WT HIV-1NL4-3 with LAI CA infection (Fig. 5B). However, 2-LTR circles were reduced 7-fold for HIV-1NL4-3 with LAI CA (Fig. 5B), corresponding to a 5-fold reduction in infectivity of WT HIV-1NL4-3 with LAI CA upon CsA treatment (Fig. 4B). These results are consistent with previous observations (52).
Inhibiting CypA-capsid interaction restricts N57A HIV-1 infectivity at nuclear entry. HeLa cells were infected with equal amounts of WT HIV-1NL4-3 (A), WT HIV-1NL4-3 with LAI CA (B), N57A HIV-1NL4-3 (C), or N57A HIV-1NL4-3 with LAI CA (D) in the presence or absence of CsA treatment, and reverse transcription products and 2-LTR circles were measured. ND, no qPCR product detected. **, P < 0.01; *, P < 0.05. Error bars indicate SEM for 2 experiments.
CsA treatment did not alter production of early or late reverse transcripts for N57A HIV-1NL4-3 (Fig. 5C) or for N57A HIV-1NL4-3 with LAI CA (Fig. 5D) compared to WT viruses. However, no 2-LTR circles were detected during CsA treatment for either N57A HIV-1NL4-3 (Fig. 5C) or N57A HIV-1NL4-3 with LAI CA (Fig. 5D). Taken together, these data suggest that inhibiting CypA binding to capsid restricts WT HIV-1 prior to or at reverse transcription. In contrast, N57A HIV-1 infection is inhibited by CsA treatment at the step of nuclear entry.
N57A HIV-1 infectivity is rescued by CA mutation G94D in a virus strain- and host cell type-dependent manner.While it was impaired only 2- to 5-fold in infectivity compared to WT virus, we hypothesized that N57A HIV-1LAI could be amenable to adaptation to improve replication. Therefore, serial passaging of replication-competent N57A HIV-1LAI was performed in MT-4 cells until robust replication was detected. Sequencing of the virus after outgrowth revealed a second CA mutation, G94D, in addition to N57A. Addition of this mutation to N57A HIV-1LAI resulted in the partial rescue of N57A HIV-1LAI infectivity in both spreading (Fig. 6A, top panel) and single-cycle (Fig. 6A, bottom panel) infection. A similar positive effect of G94D on infectivity of N57A HIV-1LAI virus was observed for other cell types (Fig. 6B). To determine if the rescue of N57A HIV-1LAI infectivity by G94D differs between the two virus strains, the G94D mutation was combined with N57A in HIV-1NL4-3 containing either LAI Gag (Fig. 6C) or LAI CA (Fig. 6D). As with HIV-1LAI, G94D partially rescued infectivity in these viruses in HeLa, GHOST, and primary human CD4+ T cells. While the degree of rescue varied with cell type and was less apparent in single-cycle infection than in spreading infection, G94D consistently increased the infectivity of N57A HIV-1NL4-3 with LAI CA.
N57A HIV-1LAI infectivity is rescued by CA mutation G94D in a cell type-dependent manner. (A) MT-4 cells were infected with equal infectious units of WT, N57A, or N57A/G94D HIV-1LAI replication-competent (top) or single-cycle (bottom) GFP reporter virus. Virus replication was measured by GFP expression for 2 weeks (top). Single-cycle infectivity was determined after 48 to 72 h by GFP expression (bottom). (B) The indicated cell types were infected with equal amounts of WT, N57A, or N57A/G94D HIV-1LAI single-cycle GFP reporter virus and assayed after 48 to 72 h for GFP expression. (C and D) The indicated cell types were infected with equal amounts of WT, N57A, or N57A/G94D HIV-1NL4-3 single-cycle luciferase reporter virus with LAI Gag (C) or LAI CA (D). ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05; ns, not significant. Error bars indicate SEM for 2 to 4 experiments.
In contrast, the addition of G94D to N57A in HIV-1NL4-3 did not rescue the N57A infectivity defect in HeLa, GHOST, or primary human CD4+ T cells (Fig. 7A). While N57A/G94D HIV-1NL4-3 had significantly higher infectivity than N57A HIV-1NL4-3 in GHOST cells, the level of N57A/G94D HIV-1NL4-3 infectivity was still greater than 1 log lower than that of WT virus. Surprisingly, while the addition of the L83V CA mutation in HIV-1NL4-3 was sufficient to rescue the N57A infectivity defect in HIV-1NL4-3 similarly to that in HIV-1LAI (Fig. 2F), L83V in HIV-1NL4-3 did not rescue the N57A/G94D infectivity defect (Fig. 7B).
N57A HIV-1NL4-3 infectivity is not rescued by CA mutation G94D. (A) The indicated cell types were infected with equal amounts of WT, N57A, or N57A/G94D HIV-1NL4-3 luciferase reporter virus and assayed after 48 h for luciferase activity. (B) Cells were infected with L83V, N57A/L83V, or N57A/L83V/G94D HIV-1NL4-3 and assayed after 48 h for luciferase activity. ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05; ns, not significant. Error bars indicate SEM for 2 to 4 experiments.
G94D rescue of N57A HIV-1 infectivity is dependent on CA binding to CypA.G94D HIV-1 has been previously shown to infect cells independent of capsid binding to CypA, becoming CsA dependent in some cell lines (53–55). To determine whether CsA treatment would affect the rescue of N57A HIV-1 infectivity by addition of G94D, HeLa and GHOST cells were infected with WT, N57A, and N57A/G94D HIV-1NL4-3 or HIV-1NL4-3 with LAI CA in the presence or absence of CsA treatment. As expected, there was no appreciable difference in the infectivity of HIV-1NL4-3 viruses during CsA treatment and no rescue of N57A HIV-1NL4-3 infectivity by G94D (Fig. 8A). Whereas G94D rescued infectivity of N57A HIV-1NL4-3 with LAI CA in both HeLa and GHOST cells, treatment with CsA prevented this rescue (Fig. 8B). In HeLa cells, there was a nonsignificant increase in infectivity with the addition of G94D to N57A HIV-1NL4-3 with LAI CA in the presence of CsA. In GHOST cells, the infectivity of N57A/G94D HIV-1NL4-3 with LAI CA was significantly decreased (15-fold) compared to that of virus with N57A alone in the presence of CsA.
G94D rescue of N57A HIV-1LAI infectivity is dependent upon capsid binding to CypA in a cell type-specific manner. (A and B) The indicated cell types were infected with equal amounts of WT, N57A, or N57A/G94D HIV-1NL4-3 luciferase reporter virus with WT (A) or LAI (B) CA in the presence or absence of CsA treatment and assayed after 48 h for luciferase activity. (C and D) Cells treated with aphidicolin were infected with viruses from panels A and B, respectively. N57A and N57A/G94D HIV-1 infectivity is shown relative to that of the corresponding WT virus. Dotted lines represent the average luciferase signal of uninfected cells. *, P < 0.05. Error bars indicate SEM for 3 experiments.
In growth-arrested cells, CsA treatment did not change the rescue of N57A infectivity by G94D compared to N57A alone in either HIV-1NL4-3 or HIV-1NL4-3 with LAI CA. Infectivities of N57A and N57A/G94D HIV-1NL4-3 were similar with and without CsA treatment in both nondividing HeLa and GHOST cells (Fig. 8C). In HIV-1NL4-3 with LAI CA, no appreciable difference was observed between N57A and N57A/G94D viruses in nondividing HeLa cells with and without CsA treatment (Fig. 8D). However, N57A/G94D HIV-1NL4-3 with LAI CA showed a 10-fold reduction in infectivity in GHOST cells in the presence of CsA compared to N57A HIV-1NL4-3 with LAI CA (Fig. 8D). Altogether, these data demonstrate that the rescue of the N57A infectivity defect by G94D in HIV-1NL4-3 with LAI CA is both cell cycle and, in some cell types, CypA dependent. In contrast, N57A HIV-1NL4-3 infectivity is not rescued by G94D, with or without CypA interaction, in dividing or nondividing cells.
HIV-1NL4-3 and HIV-1LAI CA sequences are represented in HIV-1 clinical isolates.Given the strikingly different CA-dependent N57A infectivity phenotypes observed in lab-adapted strains HIV-1NL4-3 and HIV-1LAI, which differ by only four amino acids in CA (residues 6, 83, 120, and 208), we determined whether the CA sequences of these HIV-1 strains were representative of clinical isolates, including non-B subtypes. Over 40,000 Gag sequences from the Los Alamos National Laboratory HIV-1 sequence database were analyzed, of which approximately 14,000 were subtype B. The results showed that CA amino acids 6, 83, 120, and 208 are polymorphic within subtype B (Fig. 9A) and across multiple subtypes (Fig. 9B). These 4 amino acids in HIV-1NL4-3 and HIV-1LAI are most predominant and, with the exception of position 120 analyzed across all subtypes, collectively represent the majority of primary isolate sequences, though their relative prominence varies. An examination of a panel of 10 subtype B transmitted/founder viruses (56) similarly demonstrated that at CA residues 6, 83, 120, and 208, the HIV-1NL4-3 and HIV-1LAI amino acids together represent 70% to 100% of the amino acids found at these positions in the transmitted/founder virus panel (data not shown). Furthermore, more than 90% of CA sequences within subtype B (Fig. 9C) and across all subtypes (Fig. 9D) matched HIV-1NL4-3 and HIV-1LAI at the vast majority of CA positions, with a greater degree of diversity observed when comparing non-B sequences. Taken together, these data demonstrate that HIV-1NL4-3 and HIV-1LAI CA sequences are both representative of CA in HIV-1 primary isolates.
CA positions differing between HIV-1NL4-3 and HIV-1LAI are polymorphic and represented in most HIV-1 clinical isolates. (A and B) HIV-1 Gag sequences from the LANL HIV-1 database were compared at residues 6, 83, 120, and 208 in subtype B (A) and multiple subtypes (B), with NL4-3 (blue) and LAI (red) amino acids generally being the most predominant. Amino acids comprising at least 2% of sequences are indicated in the graphs. Gray segments represent the combination of amino acids at <2% prevalence. (C and D) HIV-1NL4-3 and HIV-1LAI amino acid prevalence at each CA position were compared to subtype B (C) or all subtype (D) HIV-1 clinical isolate sequences. The four residues at which HIV-1NL4-3 and HIV-1LAI differ are marked with asterisks.
DISCUSSION
Here, we characterize the unexpected difference in infectivity of N57A HIV-1 in two of the most widely used lab-adapted HIV-1 strains, HIV-1NL4-3 and HIV-1LAI, and offer a model to summarize our findings (Fig. 10). While N57A has been previously examined in HIV-1NL4-3 (7, 40, 41, 57) and found to have an infectivity defect similar to our results in HIV-1NL4-3, this is, to the best of our knowledge, the first study to examine N57A in HIV-1LAI and the first observation of an attenuated (2- to 7.7-fold) infectivity defect with N57A. Our results were consistent across multiple cell lines and in primary human CD4+ T cells, and while cell type-specific variation in overall infectivity was observed here as in previous studies (7, 40, 41, 57), the N57A infectivity defect was always significantly attenuated in HIV-1LAI compared to HIV-1NL4-3.
Model of N57A HIV-1NL4-3 and N57A HIV-1LAI phenotypes. The differences in capsid permeabilization, reverse transcription, nuclear import, and infection of N57A HIV-1 compared to WT virus in either HIV-1NL4-3 (top) or HIV-1LAI (bottom) are shown. The effects of CsA treatment, cell cycle arrest (via aphidicolin treatment), and the addition of the G94D CA mutation are included.
Our use of gag chimeric viruses demonstrated that the N57A infectivity phenotypes in HIV-1NL4-3 and HIV-1LAI are CA dependent, with the infectivity phenotype observed in each chimera corresponding to the phenotype observed in the gag source strain and not the background strain (i.e., HIV-1NL4-3 with LAI Gag exhibited the same N57A infectivity phenotype as HIV-1LAI, and HIV-1LAI with NL4-3 Gag phenocopied HIV-1NL4-3). We focused on CA to account for the observed differences in N57A infectivity phenotypes between HIV-1NL4-3 and HIV-1LAI because N57A was previously shown to affect interactions with CA-dependent host factors CPSF6 (12, 40), TNPO3 (11), Nup153 (41), and Nup358 (7). Indeed, mutation of CA amino acids 6, 83, 120, and 208 in HIV-1NL4-3 to those of HIV-1LAI, thereby producing HIV-1NL4-3 with LAI CA, resulted in N57A infectivity phenotypes in these strains that are CA dependent.
Examining subsets of these mutations in combination with N57A further refined this CA dependence to a single requisite amino acid substitution, L83V. Residue 83 resides within helix 4 of the N-terminal domain of CA, proximal to the CypA binding loop. It has been suggested that this residue in combination with residues 120 and 122 is involved in the modulation of restriction by TRIM5α (58). Maillard et al. (58) demonstrated that mutating these three residues in HIV-1R8.74 to their HIV-2 equivalents (V83Q, H120R, and P122Q) conferred sensitivity to restriction by human TRIM5α. However, there are no published reports characterizing a specific role for CA residue 83, particularly one modulated by the presence of leucine versus valine. This residue is located on an external surface within the N-terminal CA domain that interacts with multiple host factors and could be involved with one or more host factor interactions that are altered by the L83V mutation. Alternatively, mutation of residue 83 could be involved with proximal or distal CA conformational changes that vary based upon the amino acid.
As CA was shown to alter the N57A infectivity phenotypes in HIV-1NL4-3 and HIV-1LAI, capsid permeabilization kinetics of N57A in HIV-1NL4-3 and HIV-1NL4-3 with LAI CA were examined. N57A capsid permeabilization was accelerated in HIV-1NL4-3 but not in HIV-1NL4-3 with LAI CA, suggesting a fundamental difference in the nature of the N57A infectivity defect in these two HIV-1 strains. This was supported by the observation that production of reverse transcripts was significantly reduced for N57A HIV-1NL4-3 compared to WT HIV-1NL4-3 but not for N57A in the context of HIV-1LAI CA. N57A in all virus backgrounds led to a significant reduction in 2-LTR circles compared with WT, which is consistent with previous observations of other N57 CA mutations in HIV-1NL4-3 (57). The relative level of N57A 2-LTR circles compared to WT in HIV-1NL4-3 (20-fold) is considerably lower than the level of N57A 2-LTR circles compared with WT in HIV-NL4-3 with LAI CA and HIV-1LAI (2- to 3-fold), which might suggest, if taken on its own, that N57A HIV-1NL4-3 has a more appreciable nuclear entry defect than N57A HIV-1LAI. However, when observed within the context of the accompanying defect in N57A HIV-1NL4-3 reverse transcription products versus WT virus (7- to 10-fold), the 20-fold reduction in 2-LTR circles for N57A HIV-1NL4-3 is consistent with a combination of the same 2- to 3-fold nuclear entry defect of N57A HIV-1LAI and an earlier 7- to 10-fold reverse transcription defect. In other words, these data suggest that N57A causes a similar nuclear entry defect in both HIV-1NL4-3 and HIV-1LAI but has an additional, earlier infectivity defect that results in accelerated capsid permeabilization and reduction in reverse transcription in HIV-1NL4-3 and not in HIV-1LAI.
N57A shares several phenotypic attributes with another CA mutation on the opposite side of the same N-terminal domain binding pocket, the previously characterized N74D (4). Like N57A, N74D HIV-1 infects cells independently of CPSF6, TNPO3, and Nup153 (4, 6, 8). However, unlike for N57A, there is no cell cycle dependence caused by N74D. Thus, we examined whether N57A also is dependent on CA binding to CypA. By inhibiting interaction of CA and CypA with CsA treatment, we demonstrated that N57A HIV-1 infectivity is dependent upon CypA in both dividing (HIV-1NL4-3 and HIV-1NL4-3 with LAI CA) and nondividing (HIV-1NL4-3 with LAI CA) cells. This observation is in line with the CsA hypersensitivity of the N57S CA mutation (59), which is similar to N57A in cell cycle dependence for infection (19). The CypA dependence of two CPSF6-independent CA mutations would seem to suggest an interrelationship between CypA and CPSF6 in terms of HIV-1 infectivity, which also has been suggested in other studies (50, 60). We further showed that CsA treatment does not affect reverse transcription in N57A HIV-1 but rather that 2-LTR circle production is reduced, suggesting that the loss of N57A capsid interaction affects nuclear entry. These results support a role for CypA influencing HIV-1 nuclear entry that has been previously proposed as a result of different studies (7, 52).
The emergence of G94D as a compensatory mutation in N57A HIV-1LAI further demonstrates an interplay of CypA and other cellular factors on HIV-1 infection. Similar to the case for N57A HIV-1 infectivity, G94D rescue is dependent upon CA interaction with CypA. Inhibiting this interaction via CsA treatment reduced (HeLa cells) or abrogated (GHOST cells) the G94D infectivity rescue of N57A HIV-1LAI. In GHOST cells, CsA treatment significantly exacerbated overall infectivity of N57A/G94D HIV-1LAI, reducing it by more than 100-fold. These results are consistent with previous reports on cell type differences in HIV-1 infectivity upon CsA treatment with CA mutation G94D or the related CA mutation A92E (52–55, 61, 62). The significant reduction in infectivity of N57A/G94D HIV-1LAI in CsA-treated GHOST cells is similar to the 100-fold reduction in infectivity of CsA-dependent T54A HIV-1 with the addition of the N57A mutation (39). This suggests that the CypA dependence of N57A HIV-1 infectivity is not alleviated by the presence of a CypA-independent CA mutation (G94D) but is also dominant over the CypA-independent CA mutation, rendering the virus hypersensitive to CsA treatment.
However, the infectivity rescue of N57A HIV-1LAI by G94D does not restore the ability to efficiently infect nondividing cells. It also does not appear to restore host factor interactions that are deficient with N57A HIV-1. In the case of CPSF6, both N57A HIV-1 and N57A/G94D HIV-1 are not restricted by the truncated form of CPSF6, CPSF6-358 (4), in both HIV-1NL4-3 and HIV-1LAI (data not shown). This suggests that G94D might provide N57A HIV-1LAI access to an alternative pathway for infecting dividing cells. This idea is not unreasonable, as a similar cell cycle-dependent CA mutant, N57S, which must gain entry to the nucleus during cell division, relies on a particular set of nucleoporins for infection (59).
The inability of CA mutation L83V to induce the infectivity rescue of N57A HIV-1NL4-3 by G94D was unexpected, given the ability of the addition of the L83V CA mutation to HIV-1NL4-3 to switch the N57A HIV-1NL4-3 infectivity phenotype to that of N57A HIV-1LAI. One or more of the CA mutations L6I, H120N, and G208A appears to be required, presumably in conjunction with L83V, to facilitate G94D rescue of N57A HIV-1NL4-3 infectivity. This is a curious point for future study, the elucidation of which may reveal the mechanisms by which L83V attenuates the N57A HIV-1NL4-3 infectivity defect and by which G94D rescues N57A HIV-1LAI infectivity.
While originally derived from one or more clinical isolates, HIV-1NL4-3 (46) and HIV-1LAI (47) are lab-adapted virus strains. Adaptation by serially passaging in cell lines selects for phenotypes, such as CD4 affinity and coreceptor tropism, that differ substantially from infectivity phenotypes observed in HIV-1 clinical isolates, particularly transmitted/founder strains (45). While the primary viral determinant for these lab-adapted infectivity phenotypes was shown to map to changes to gp120 (63, 64), it is possible that lab adaptation could also result in changes to CA. HIV-1NL4-3 and HIV-1LAI CA differ at only four positions (residues 6, 83, 120, and 208), three of which (residues 6, 83, and 120) are among the most polymorphic positions within CA (19, 65). Indeed, an analysis of over 40,000 Gag sequences from primary isolates of multiple subtypes demonstrated the polymorphic nature of all four of these CA positions, including the residues present in HIV-1NL4-3 and HIV-1LAI. Within subtype B, Gag (including CA) shows low overall sequence diversity (66), and across all subtypes the N-terminal region of CA, along with IN, contains the lowest levels of amino acid diversity within the HIV-1 genome (67). The frequency of HIV-1NL4-3 and HIV-1LAI amino acids across all CA positions showed strong conservation, with over 90% of CA sequences matching HIV-1NL4-3 and HIV-1LAI at the vast majority of the positions. The high degree of similarity of CA sequences from HIV-1NL4-3, HIV-1LAI, and clinical isolates suggests that CA-dependent phenotypes, such as observed with N57A, could differ among other HIV-1 lab-adapted strains and primary isolates.
We have shown that two closely related and widely used HIV-1 molecular clones (HIV-1NL4-3 and HIV-1LAI) can exhibit significantly different infectivity phenotypes due to four CA amino acid differences and that these clones are both represented in HIV-infected individuals. Thus, careful consideration should be given to drawing conclusions from one particular HIV-1 clone used in postentry infection studies and to the potential for significant variation in results with the use of multiple strains. While phenotypic differences in HIV-1 subtypes and host cell types have been acknowledged for many aspects of the virus life cycle, differences among closely related subtype B strains, such as HIV-1NL4-3 and HIV-1LAI, or for CA-dependent host factors have garnered less attention. We hope that this study highlights the significance that minor differences in closely related HIV-1 strains may impart significant infectivity differences and that consideration should be given to examining the effects of natural polymorphisms on specific results.
MATERIALS AND METHODS
Cell lines.HEK 293T cells, GHOST cells, and HeLa cells were maintained at 37°C with 5% CO2 in Dulbecco’s modified Eagle medium (DMEM) (Gibco) supplemented with 10% fetal bovine serum (FBS) (Atlanta Biologicals), 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM l-glutamine (Thermo Fisher). GHOST cells were additionally supplemented with 500 μg/ml Geneticin G418 (Gibco), 500 μg/ml puromycin (Invitrogen), and 100 μg/ml hygromycin B (Invitrogen). MT-4 cells were cultured in Roswell Park Memorial Institute (RPMI) medium (Cellgro) supplemented with 10% FBS (Sigma), 1× penicillin-streptomycin (Cellgro), and 2 mM l-glutamine (Cellgro). Human peripheral blood mononuclear cells (PBMCs) were isolated from leukapheresis obtained from the Central Blood Bank (Pittsburgh, PA) via Ficoll-Paque Plus (GE Healthcare) density gradient centrifugation, following the manufacturer’s instructions. PBMCs were maintained at 37°C and 5% CO2 in RPMI 1640 medium (Gibco) supplemented with 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM l-glutamine, and 20 U/ml recombinant interleukin-2 (IL-2) (Thermo Fisher). PBMCs were stimulated with 50 U/ml IL-2 and 5 μg/ml phytohemagglutinin (PHA) (Gibco) for 48 to 72 h prior to infection.
Viruses.Proviral plasmid pNLdE-luc (4) was used to produce HIV-1NL4-3, and plasmids pBru3ori-ΔEnv-luc2 and pBru3ori-ΔEnv-GFP3 (38) were used to produce HIV-1LAI. CA mutations were generated via the Q5 (New England BioLabs) or QuikChange (Agilent) site-directed PCR mutagenesis kit following the manufacturers’ instructions and verified by Sanger sequencing. The pNLdE-LAI-luc chimeric plasmids (HIV-1NL4-3 with LAI Gag) were created by cloning the Gag fragment from pBru3ori-ΔEnv-GFP3 into pNLdE-luc using the restriction enzymes BssHII and ApaI. The pBru3ori-ΔEnv-NL43-GFP3 chimeric plasmids (HIV-1LAI with NL4-3 Gag) were created by cloning the Gag fragment from pNLdE-luc into pBru3ori-ΔEnv-GFP3 using BssHII and ApaI.
HEK 293T cells were plated overnight and transfected with Lipofectamine 2000 (Invitrogen) or with polyethylenimine (PolySciences) using the following plasmids: the proviral plasmids described above, pL-VSV-G or pHCMV-G, and, for imaging studies, pcDNA5-TO-Vpr-mRuby3-IN (50). Supernatants were harvested 48 h later, filtered or centrifuged to remove cells, concentrated via a Lenti-X concentrator (Clontech), and stored in aliquots at −80°C. For virus stocks used in the capsid permeabilization assay, 293T producer cells were supplemented with 1 mM 5-ethynyl uridine (EU) (Invitrogen) for 12 to 16 h, which was replaced with regular medium for the remainder of the 48 h of incubation. Virus titers were determined by infection of GHOST cells as previously described (68), and viruses were quantified for p24 with a HIV-1 p24 enzyme-linked immunosorbent assay (ELISA) kit (XpressBio). Virus yields were also quantified by measuring virion-associated reverse transcriptase (RT) activity using a SYBR green-based quantitative PCR assay (69).
Infectivity assays.HeLa cells, GHOST cells, and PBMCs were plated overnight in 24-well plates, challenged with equal p24 amounts of virus for 2 h, washed with phosphate-buffered saline (PBS), and given fresh medium. Virus infectivity was determined by luciferase production (Promega) after 48 h using a 1450 MicroBeta TriLux microplate luminescence counter (PerkinElmer). For assays including treatment with CsA and/or aphidicolin, cells were treated with CsA (2 μM) and/or aphidicolin (2 μg/ml) at the time of plating and remained in drug-containing medium throughout the assay. For green fluorescent protein (GFP) reporter virus, infection of HeLa cells and MT-4 cells was performed with equal amounts of virus input normalized by RT activity. Spinoculation (1,200 × g for 30 min) was used to enhance infection of HeLa cells. The number of GFP-expressing cells was counted using either Guava easyCyte (Millipore) or an LSRII flow cytometer (BD Biosciences) at 2 to 3 days after infection.
Measurement of reverse transcripts and 2-LTR circles.HeLa cells were plated in 6-well plates and infected with 50 ng p24 of virus treated with DNase I (Roche) for 30 min at 37°C. After 24 h, cells were washed with PBS, trypsinized, and pelleted. Control infections were performed in the presence of 150 nM efavirenz or 25 nM rilpivirine. For assays including treatment with CsA, cells were treated with CsA (2 μM) at the time of plating and remained in drug-containing medium throughout the assay. DNA was extracted using the Blood minikit (Qiagen). Early (RU5) and late (gag) HIV-1 reverse transcripts and 2-LTR circles were measured by quantitative PCR as previously described (70).
Capsid permeabilization assay and confocal imaging.The capsid permeabilization assay was performed as previously described (49, 50). Briefly, HeLa cells were plated overnight in 35-mm glass-bottom dishes (MatTek), synchronously infected with EU-labeled HIV-1 virus, and incubated for the indicated time periods at 37°C. Cells were washed with PBS, fixed with 2% paraformaldehyde, and permeabilized with 0.1% Triton X-100 (Fisher Scientific). Viral RNA was stained using the Click-iT RNA imaging kit (Invitrogen), washed with PBS, stained with Hoechst 33342 (Molecular Probes), and mounted with coverslips. Confocal imaging was performed on a Nikon A1 laser scanning confocal microscope, with 8 to 10 image Z-stacks per sample. Viral RNA staining was enumerated using Imaris image analysis software (Bitplane).
Sequence analysis.Aligned HIV-1 p24 (CA) protein sequences were retrieved from the Los Alamos National Laboratory (LANL) HIV-1 sequence database (https://www.hiv.lanl.gov) in May 2018. The vast majority (99.5%) of the 40,712 sequences were classified as group M, with 35% subtype B, 29% subtype C, 22% circulating recombinant forms, 10% subtype A, 3% subtype D, and <1% each subtypes F, G, H, J, and K. At each CA amino acid position, sequences were compared to the NL4-3 and LAI amino acids, and the prevalence of the NL4-3 and LAI amino acids (0% to 100%) was calculated. The frequencies of amino acids present at residues 6, 83, 120, and 208 were measured by Perl scripts, with amino acids representing fewer than 2% of sequences aggregated into an “other” category.
Statistics.Results were analyzed for statistical significance by the two-sided Student t test with Prism software (GraphPad). Grubbs’ extreme Studentized deviate test was used to exclude significant outliers from capsid permeabilization assay data sets. A P value of less than or equal to 0.05 was used to indicate statistical significance.
ACKNOWLEDGMENTS
We thank Callen Wallace and Samantha Teng for technical assistance.
This work was supported by National Institutes of Health P50 grant GM082251 (S.C.W., M.Y., and Z.A.), National Institutes of Health R01 grant AI100720 (M.Y.), and National Institutes of Health T32 training grant AI065380 (D.K.F.).
FOOTNOTES
- Received 7 February 2019.
- Accepted 15 February 2019.
- Accepted manuscript posted online 27 February 2019.
- Copyright © 2019 American Society for Microbiology.
