A long history of non-human primate infection by SIVs has resulted in a host-virus arms race and led to rapid evolution of both sides 
. Host restriction factors and their viral antagonists provide an attractive system to investigate genetic conflict between hosts and the viruses. Primate APOBEC3G, TRIM5α and Tetherin undergo positive selection and the positive selection pressures most likely come from the vial antagonists, such as Vif for APOBEC3G, viral capsid for TRIM5α, and Vpu (or SIV Nef) for Tetherin [reviewed in 
]. HIV-1 genes encoding structural proteins and the polymerase are also under positive selection 
. Thus, host restriction factors at least partially contribute to the selective pressures on vial antagonists.
SAMHD1 is a myeloid cell-specific HIV-1 restriction factor that can be counteracted by Vpx of SIVsm and HIV-2 
, while HIV-1 and its ancestor SIVcpz do not encode Vpx to overcome SAMHD1-mediated restriction. It implies that SAMHD1 from human, chimpanzee and gorilla should not be under positive selection, and the lack of vpx
gene in HIV-1 and its ancestor SIVcpz might be associated with the HIV-1 restriction function of SAMHD1. Indeed, we identified that positive selection mainly acts on SAMHD1 from orangutan, gibbon, rhesus macaque and marmoset, but not on that from human, chimpanzee and gorilla. The lack of selective pressure from Vpx to drive the evolution of SAMHD1 of human, chimpanzee and gorilla is consistent with the fact that HIV-1, SIVcpz, and SIVgor lack the vpx
gene. Although HIV-2 encodes Vpx and can infect humans, it only accounts for a small population of infected individuals and has a short history (<71 years) in humans 
. Thus, HIV-2 Vpx is unlikely to be the major driving force behind the evolution of human SAMHD1. The detection of positive selection on the SAMHD1 of other four primates is consistent with the fact that the vast majority of SIV have Vpx, which may drive the rapid evolution of SAMHD1. Our results of positive selection of SAMHD1 confirmed the findings of recent two studies 
Virus genes normally have more rapid evolutionary rates than the primate genes 
. If the primate SAMHD1 is the selective agent, we expect that Vpx that can degrade SAMHD1 should be under positive selection, while Vpx that cannot degrade SAMHD1 should not undergo positive selection. Indeed, we detected positive selection in Vpx from both the HIV-2 sub-clades and the SIV(Vpx+) sub-clades I and II (), suggesting that Vpx evolves under selective pressure from the primate SAMHD1. Of particular importance is that no positive selection was detected in Vpx of the SIV(Vpx+) sub-clade III, in which Vpx of certain SIVrcm isolates are unable to degrade SAMHD1 
. Vpx of two SIVrcm isolates from Nigeria and Gabon cannot degrade human SAMHD1 
, suggesting that the Vpx of other SIVrcm and SIVmnd are unable to degrade human SAMHD1 due to their high amino acid homology 
). Indeed, recent studies demonstrated that Vpx of SIVrcm and SIVmnd can degrade SAMHD1 from red-capped mangabeys and mandrills respectively in a species-specific manner 
Site-model (M7 vs. M8) test for vpx genes from two HIV-2 sub-clades and three different SIV(Vpx+) sub-clades.
A previous study suggested that SIVcpz is a recombinant virus between the predecessor of SIVrcm and the common ancestor of SIVgsn, SIVmus, and SIVmon that infect Greater spot-nosed, Mustached, and Mona monkeys 
. However, we found that SIVgsn, SIVmus, and SIVmon are located at the basal position in the tree of lentiviruses, far away from the clade containing SIVcpz, implying less possibility of the ancestor of SIVgsn, SIVmus, and SIVmon participating in the recombination. In addition, the inference of the ancestral host states show that the MRCA of SIVcpz, SIVrcm, SIBmnd, SIVgor and HIV-1 most likely is SIVcpz within the gag
gene regions ( and Fig. S5A
), but most likely is SIVrcm or SIVmnd in the env
gene region (Fig. S5B
), supporting an association of recombination with the formation of SIVcpz. Therefore, we presume that the loss of vpx
in SIVcpz was a result from the positive selection of the viral recombination and fitness. Furthermore, the phylogenetic position of Vpx loss have been well documented (reviewed in 
), which supports our conclusion that co-evolution of primate SAMHD1 and lentivirus Vpx leads to the loss of the vpx
gene. Of note, the application of HIV molecular clocks to long-term lentivirus evolution has its limitations because heterotachy can cause root ages to be overestimated 
. Because the sequence data did not include HIV-1 group M sequences obtained after 1996 and SIVgor sequences acquired before 2004, our estimates may not represent the most accurate ones. For example, we dated the origin of HIV-1 group M to 598 (95% HPD, 59–1059) years ago, much earlier than previous estimates based on the pol
genes (216 years ago, 95% HPD, 111–384) 
Our results suggest that HIV-1 could not inherit the lost vpx
gene from its ancestor SIVcpz (), and the lack of Vpx seems to be advantageous for HIV-1 to escape from human immune surveillance 
. Thus, the reason for the loss of vpx
might be associated with the function of chimpanzee SAMHD1. Comparison of amino acid sequences shows that there are 7 different residues between human and chimpanzee SAMHD1 (Fig. S6
), including one residue in the SAM domain and three in the HD domain. Whether these differences affect the function of the chimpanzee SAMHD1 to inhibit lentiviral infection need to be determined. Furthermore, SAMHD1 may act as a regulator of the innate immune response 
and it is unclear whether primate SAMHD1 restricts other viruses in addition to HIV-1. It is also possible that positive selection on primate SAMHD1 has been driven by other pathogens or multiple past host and pathogen conflicts.
In summary, our evolutionary analyses of primate SAMHD1 and lentiviruses provide new insights into understanding the genetic conflict between the restriction factor SAMHD1 and the viral antagonist Vpx. HIV-1 could not inherit the lost vpx gene from its ancestor SIVcpz. The lack of Vpx in HIV-1 results in restricted infection in myeloid cells that are important for antiviral immunity, which could contribute to the AIDS pandemic by escaping the immune responses.