Impact of Genetic Variation during Cross-Species Transmission on Lentiviral Capsid-Host Protein Interactions

Abstract

For lentiviruses such as HIV-1, the viral capsid protein (CA) plays a crucial role in replication by facilitating active transport across nuclear pore complexes (NPCs). Nucleoporin Nup358/RanBP2 – a large, multidomain protein that comprises the main component of cytoplasmic NPC filaments – was previously identified as a potential cofactor for HIV-1 nuclear entry, and its C-terminal cyclophilin-like domain (Nup358Cyp) is able to interact with the CA of both HIV-1 and HIV-2. The importance of this interaction to viral replication is unclear though as certain cell-culture experiments suggest CA interaction with Nup358Cyp is dispensable for viral replication, and the CA of several other lentiviruses like SIVmac do not appear to interact with the Nup358Cyp domain. However, we have found that CA interaction with Nup358 is widely conserved among primate lentiviruses and is maintained by natural selection. The exception, SIVmac, likely reflects an evolutionary trade-off allowing escape from rhesus macaque TRIM5Cyp. Together, our observations are strong evidence that the interaction between viral CA and the Nup358Cyp domain must be biologically relevant in vivo. Specifically, by comparing interactions between multiple SIVsm/HIV-2 lineage CAs and several primate orthologs of Nup358, we identified interspecies differences in the Nup358Cyp domain that affect the CA interaction, but only when assayed in conjunction with the preceding Ran-binding domain 4 (Nup358R4). We next found that selection preserves the interaction during cross-species transmission, resulting in adaptation to differences between the Nup358Cyp homologs of the reservoir and spillover hosts. For example, SIVsm CA does not interact with human Nup358R4-Cyp, while HIV-2 CA interacts with both the human and sooty mangabey orthologs. We confirmed these distinct interaction phenotypes in an extended set of SIVsm/HIV-2 CAs, and mapped the difference to a single position – residue 3173 – in the Nup358Cyp domain. The differing ability to interact with human Nup358R4-Cyp is due to residue 85 in the CA 4-5 loops; most SIVsm strains encode a glutamine at position 85, whereas most HIV-2 strains encode an isoleucine. Reciprocal swaps reverse the interaction phenotypes, such that the SIVsm Q85I CA mutant strongly interacts with human Nup358R4-Cyp, while HIV-2 I85Q CA mutant does not. This difference also correlates with differences in single- and multi-cycle infectivity on human cell lines and levels of nuclear import in HeLa cells. Together, these results indicate that HIV-2 adapted to human Nup358 during emergence in humans. We also examined the ability of our CA panel to interact with Cyclophilin A. While all HIV-2 CA interact with CypA, the ability to interact varied among the other SIVsm CA tested, and was absent for SIVpbj. Thus, conservation of CA interaction with Nup358Cyp does not correlate to the ability to interact with CypA, and is not simply a consequence of maintaining the CA-CypA interaction.