Though CypA binds mature CA, interaction was first discovered with the Gag polyprotein (37
), and the findings of most subsequent studies were consistent with the hypothesis that producer cell CypA is important for HIV-1 replication. CypA binds the Gag polyprotein more strongly than it binds mature CA (16
). CypA is incorporated into HIV-1 virions or Gag particles with a fixed stoichiometry, and disruption of CypA incorporation into virions by gag
mutants correlated with decreased infectivity (24
). The hypothesis seemed especially secure when it was found that, like the gag
mutants, competitive inhibitors that block CypA incorporation also inhibited virion infectivity (14
). These drugs had no effect on the infectivity of related viruses such as HIV-2 and SIVMAC239
), viruses that do not incorporate CypA into particles.
The experiments presented here render the original hypothesis unlikely and show that target cell CypA interaction with HIV-1 CA is important for viral infectivity. We were, in fact, unable to detect any contribution to the infectivity of HIV-1 virions by producer cell CypA. There are perhaps several technical reasons why this result was not apparent before. The simultaneous analysis of CypA function with the large number of experimental tools that have accumulated over the years, including CA mutants with altered CypA affinity or CypA dependence, three different competitive inhibitors of the CA-CypA interaction, and cell lines in which CypA expression was disrupted by two different genetic methods, permitted us to detect effects that were not apparent previously. Jurkat cells engineered to be CypA deficient by gene targeting were valuable to prove the importance of CypA for spreading infection (15
), but these cells are difficult to transfect or otherwise manipulate to produce virus for single-cycle assays. The recent development of RNAi in mammalian cells permitted us to disrupt CypA expression in transfectable cell lines and to more readily assess the role of CypA. Finally, more sensitive and accurate assays for quantitating HIV-1 infectivity in single-cycle assays are now available. This last point is critical since the effects of CypA on HIV-1 replication in human cells are modest in magnitude and dependent on the multiplicity of infection (50
These new conclusions are consistent with most reports in the literature. Aside from minimal effects on the kinetics of gag
processing or virion release (45
), disruption of CypA incorporation into virions has no effect on biochemical or ultrastructural characteristics of HIV-1 virions, including endogenous reverse transcription (13
). We had proposed that virion-associated CypA might promote virion disassembly (36
), but the stability of virion cores is not detectably altered by CypA disruption (53
), and structural models place CypA on the outside of the core (29
), where it would seem unlikely to disrupt CA-CA interactions.
The results reported here are also in agreement with recent experiments from our lab, and others, indicating that target cell CypA modulates HIV-1 sensitivity to CA-specific restriction factors (33
). The mechanism by which target cell CypA modulates HIV-1 restriction remains to be determined. CypA catalyzes the rate of cis-trans
interconversion of the peptide bond connecting HIV-1 CA residues G89 and P90 (10
): perhaps HIV-1 susceptibility to restriction factors is conformation dependent. Alternatively, as a component of a putative restriction factor complex, CypA might determine if incoming virus is recognized or somehow regulate the antiviral activity of the complex itself.
We also provided evidence that CsA inhibits infectious HIV-1 virion production and entry via independent mechanisms. The magnitude of inhibition was greater if CsA was administered during entry than if it was administered during production, and the effects were additive if the drug was present at both times (Fig. ). Consistent with the fact that the G89V mutation already abolishes CypA binding to CA, infectivity of the HIV-1NL4-3
G89V mutant was not decreased further if CsA was added during infection of target cells. However, CsA added during virion production decreased infectivity of the HIV-1NL4-3
G89V mutant to the same degree as for the wild-type virus. Others have reported similar findings (4
). These results indicate that the inhibitory effect of CsA during virion production is independent of the CA interaction with cyclophilin. An additional indication that CsA has CA-independent effects was provided by the finding that VSV-G pseudotyping suppresses the effects of CsA on virion production but not the effects of CA mutants (5
). Here these findings were extended further by showing that the inhibitory effect of CsA was of equal magnitude when virions were produced from cells with CypA knockdown by RNAi (Fig. ). Thus, the inhibitory effect of CsA on virion assembly is independent of CypA.
-CsA had the same effect as CsA did (Fig. ), indicating that inhibition of calcineurin phosphatase activity (41
) is not required for the antiviral effect. Sanglifehrin also has the same effect (Fig. ). The only property that Sanglifehrin shares with the other drugs is the ability to bind to cyclophilin family members (42
). This suggests, then, that inhibition by CsA involves targeting of one of the cyclophilin family members. Our knockdown data rule out CypA (Fig. ), but there are 14 more cyclophilins to test (15
). Consistent with the CypA independence of the drug effect, there was no correlation between the ability of the drugs to disrupt CypA binding to CA and their effectiveness at inhibiting virion infectivity. For example, Sanglifehrin was a less potent inhibitor of infectivity, but all three drugs eliminated CypA incorporation into virions with comparable efficiency (Fig. ).
How does CsA disrupt the production of infectious virions? Suppression of the drug effect by VSV-G pseudotyping (5
) suggests that CsA might block Env function. VSV-G pseudotyping also eliminates the requirement for Nef (4
), and CsA has minimal effect on nef
mutant virus (4
), suggesting that CsA inhibits Nef function, though not all investigators have reported the same result (31
). Interestingly, SIV Nef or HIV-2 Nef suppresses the inhibitory effect of CsA on HIV-1 virions (31
). Further work will be required to determine which cyclophilin family member and which viral protein are targeted by CsA during virion production.
Previous experiments suggested that the CsA dependence of the A92E mutant in H9 cells results from elevated levels of CypA in H9 cells, compared to Jurkat cells (2
). This hypothesis seems not to hold up in that CsA dependence of A92E is more pronounced in HeLa than in H9 cells (data not shown), yet the levels of CypA expression in HeLa cells are not significantly elevated compared to those in 293T cells, a cell line which exhibits the same phenotype as Jurkat cells (data not shown).
The phenotype of the A92E mutant in HeLa cells resembles a phenomenon termed Lv1 restriction. In nonhuman primate cells, HIV-1 replication is inhibited at an early postentry step of infection and can be rescued by CsA (50
). In human cells, a similar activity called Ref1 specifically restricts N-tropic MLV and HIV-1 if the CypA-CA interaction is eliminated during infection by genetic or pharmacological means (43
). In both cases, target cell CypA plays a role during infection: its interaction with HIV-1 CA protects the virus from Ref1-mediated restriction in human cells and renders it sensitive to Lv1-mediated restriction in monkey cells.
We used two different approaches, saturation by N-tropic VLPs and suppression by As2O3, to address the question whether the CsA-dependent phenotype of HIV-1 CA variants in HeLa cells results from altered Ref1 activity in these cells. Both experiments indicated that Ref1 is not restricting NL4-3/A92E (Fig. ). Additionally, we attempted to saturate the putative restriction factor by using HIV-1 VLPs bearing the A92E mutation. Evidence of a saturable restriction factor was not obtained (data not shown).
The group O viruses HIV-1CA9
are resistant to CsA in Jurkat cells or peripheral blood mononuclear cells (14
). Based on the similarity in phenotype with the A92E mutant, we hypothesized that the group O viruses would behave similarly in single-cycle assays. Vectors bearing the CypA binding loop from these viruses (Fig. ) indeed exhibited the same cell-type-specific response to CsA (Fig. and ). Thus, viruses bearing CAs that confer this phenotype occur in infected patients. Also, the dramatically different phenotypes of different human cell lines raise the possibility that such variations might occur among people and might render individuals differentially susceptible to HIV-1 infection.