During the process of HIV-1 reverse transcription, viral reverse transcriptase requires deoxynucleoside triphosphates (dNTPs) as a substrate for the synthesis of viral cDNA (18, 19). SAMHD1 is a cytosolic enzyme with phosphohydrolase activity that enzymatically degrades (“hydrolyzes”) dNTPs (18–20). Deoxyguanosine triphosphate in particular, binds to the allosteric site of SAMHD1 and activates SAMHD1’s hydrolytic activity (18).

SAM domain and HD domain-containing protein 1 is expressed in peripheral CD4⁺ leukocytes including myeloid cells [e.g., macrophages and dendritic cells (DCs)] and CD4⁺ T cells (19). The experiments in in vitro cell cultures demonstrate that SAMHD1 restricts HIV-1 infection in non-dividing cells such as macrophages (plus phorbol 12-myristate 13-acetate-stimulated macrophage-like THP-1 cell line), DCs, and resting CD4⁺ T cells by degrading dNTPs (18, 19).

In comparison with dividing (i.e., cycling and activated/proliferating) cells, the level of intracellular dNTP is much lower in non-dividing cells (21). Previous studies have suggested that SAMHD1 plays a crucial role in maintaining a low pool of cellular dNTPs in non-dividing cells, including resting CD4⁺ T cells, which may reduce the risk of retroviral insult without disrupting homeostasis in the non-dividing cell environment (21, 22). In dividing cells, including activated CD4⁺ T cells, SAMHD1 is post-transcriptionally inactivated: cyclin-dependent kinases 1 (CDK1) and CDK2 phosphorylate the threonine residue at position 592 of SAMHD1 (23). This phosphorylation impairs SAMHD1’s hydrolyzing activity and results in the loss of its anti-HIV-1 activity (23). CDKs, including CDK1 and CDK2, are key regulators of the cell cycle that activate cyclins during cell division (24). As dividing cells require a greater pool of dNTPs, SAMHD1’s enzymatic activity is inhibited by CDK1/2-mediated phosphorylation (25).

To overcome SAMHD1-mediated restriction, an accessory protein of human lentiviruses, viral protein X (Vpx), degrades SAMHD1 via the ubiquitin/proteasome-dependent pathway (22). The Q76A mutation in Vpx results in the loss of SAMHD1 degradation ability suggesting that the glutamine at position 76 is critical (22). Importantly, the vpx gene is not encoded by HIV-1 but HIV-2, another human lentivirus and causative agent of acquired immunodeficiency syndrome (AIDS) (26).

Human immunodeficiency virus type 1 and HIV-2 are evolutionarily and phylogenetically distinct, and more intriguingly, Etienne et al. have shown evidence indicating that the lineage of primate lentiviruses, including HIV-1, lost the vpx gene during viral evolution (27). These observations raise an incongruous insight: although RFs such as APOBEC3 and tetherin (see below) can be degraded and antagonized by HIV-1 accessory proteins, HIV-1 does not possess any counterparts to counteract SAMHD1. Additionally, HIV-1 is able to replicate in macrophages that express SAMHD1 in its anti-viral state. Moreover, HIV-1 is more pathogenic than HIV-2 in spite of the absence of anti-SAMHD1 factor(s) (26). These insights may imply that SAMHD1 is not critical for the restriction of HIV-1 replication. In this regard, a previous paper has revealed that the concentration of dNTP required for the reverse transcriptase of HIV-1 is clearly lower than that of HIV-2, and HIV-1 can efficiently reverse transcribe a single strand template with a lower level of dNTPs (21). Therefore, HIV-1 may have evolved to overcome SAMHD1-mediated anti-viral activity by decreasing the requirement for a high-dNTP concentration.

In addition to the phosphohydrolase activity, SAMHD1 possesses ribonuclease (RNase) activity. Ryoo et al. have reported that RNase activity but not phosphohydrolase is required for exhibiting an anti-HIV-1 effect (28). This is shown with the D137N mutant of SAMHD1, which possesses RNase activity but specifically loses phosphohydrolase activity and is still able to restrict HIV-1 infection. In contrast, the Q548A mutant of SAMHD1 that loses RNase activity but maintains phosphohydrolase activity is ineffective at restricting HIV-1 (28). Moreover, the phosphorylation of SAMHD1 at T592 negatively regulates its RNase activity in cells and impedes HIV-1 restriction (28), suggesting that the RNase activity of SAMHD1 is responsible for preventing HIV-1 infection by directly degrading viral RNA (28).

Apolipoprotein B mRNA editing enzyme catalytic-like 3 family proteins are cellular cytidine/cytosine deaminases and the human genome encodes seven APOBEC3 genes: APOBEC3A, B, C, D, F, G, and H. Some APOBEC3 family proteins, particularly APOBEC3D, APOBEC3F, APOBEC3G, and certain haplotypes of APOBEC3H (see below), are incorporated into released viral particles and enzymatically remove the amino group (-NH₂) of the cytosine residue in the minus-stranded viral DNA during viral reverse transcription. This deamination converts cytosine to uracil, which results in guanine (G) to adenine (A) substitution in the plus-stranded viral DNA (this step is usually referred to as “APOBEC3-mediated G-to-A mutation”). The APOBEC3-mediated G-to-A mutations can result in the insertion of premature termination mutations [e.g., if TGG codon is converted to TGA codon by APOGEC3, the codon encoding tryptophan (TGG) is converted to a stop codon (TGA)]. Also, multiple APOBEC3-mediated G-to-A mutations can lead to the accumulation of non-synonymous mutations, which may produce defective viral proteins.

To overcome APOBEC3-mediated anti-viral action, an accessory protein of HIV-1, viral infectivity factor (Vif), degrades anti-viral APOBEC3 proteins in virus-producing cells via the ubiquitin/proteasome-dependent pathway. The relationship between APOBEC3 and Vif has been well studied and reviewed previously [e.g., Ref. (29, 30)].

To elucidate the roles of endogenous APOBEC3 proteins in HIV-1 infection in vivo, hematopoietic stem cell (HSC)-transplanted “humanized” mouse models have been utilized. First, vif-deficient HIV-1 was incapable of replicating in humanized mice, indicating that Vif is a prerequisite for HIV-1 infection and replication in vivo (31). Also, some proviral DNA in infected humanized mice exhibited G-to-A hypermutations, further suggesting that endogenous APOBEC3 protein(s) potently exhibit anti-HIV-1 activity in vivo (31).

Secondly, to elucidate which endogenous APOBEC3 protein(s) crucially affect HIV-1 replication in vivo, two vif mutants have been utilized: one is designated “4A,” which is unable to antagonize APOBEC3D and APOBEC3F, while the other is designated “5A,” which is unable to antagonize APOBEC3G (32). As the replication efficacy of both 4A and 5A HIV-1 were significantly lower than that of wild-type (i.e., vif-proficient) HIV-1, endogenous APOBEC3D, APOBEC3F, and APOBEC3G are deemed to be potent intrinsic RFs in humanized mice (32). On the other hand, it is intriguing that the viral RNA in the plasma of humanized mice infected with 4A HIV-1 had greater diversity compared with 5A and wild-type HIV-1 sequences (32). This observation suggests that the G-to-A mutation caused by APOBEC3D and APOBEC3F potently contributes toward viral diversification. In this regard, APOBEC3G prefers the GG-to-AG mutation, while APOBEC3D and APOBEC3F prefer a GA-to-AA mutation (33–35). Also, an experimental-mathematical analysis has suggested that APOBEC3G-mediated substitution easily results in nonsense mutations (mainly because “TGG,” a codon encoding tryptophan, is converted to “TAG,” a stop codon), while the G-to-A mutations mediated by APOBEC3D and APOBEC3F (i.e., GA-to-AA mutation) lead only to missense mutations (33). Therefore, three endogenous APOBEC3 proteins, APOBEC3D, APOBEC3F, and APOBEC3G, possess the ability to suppress HIV-1 replication in vivo, while, at the same time, APOBEC3D and APOBEC3F may promote viral diversification.

Third, Nakano et al. have recently addressed the anti-viral effect of APOBEC3H in vivo (36). There are seven haplotypes within human APOBEC3H genes and APOBEC3H can be categorized based on the protein expression status of three phenotypes: stable (haplotypes II, V, and VII), intermediate (haplotype I), and unstable (haplotypes III, IV, and VI) (37–39). From the retrovirological point of view, only stable APOBEC3H exhibits anti-HIV-1 activity (37–39). Interestingly, although almost all of the naturally occurring HIV-1 Vif proteins can antagonize anti-viral APOBEC3 proteins including APOBEC3D, APOBEC3F, and APOBEC3G, certain Vif proteins are incapable of counteracting stable (i.e., anti-viral) APOBEC3H and are called “hypo” Vif (37). On the other hand, the Vif proteins that can antagonize stable APOBEC3H are called “hyper” Vif (37). To investigate the impact of endogenous APOBEC3H in vivo, an “in vivo competition assay” was conducted: hyper and hypo HIV-1s were co-inoculated into humanized mice encoding stable or unstable APOBEC3H and the most efficiently replicating virus was determined by RT-PCR (36). In the humanized mice encoding stable APOBEC3H, hyper HIV-1 predominantly replicated, suggesting that Vif’s ability to antagonize stable APOBEC3H is a prerequisite when the host is expressing stable APOBEC3H (36). On the other hand, since the type of virus that efficiently replicated in the humanized mice encoding unstable APOBEC3H (i.e., “hyper” or “hypo”) was stochastic, the selection pressure mediated by unstable APOBEC3H is relaxed (36). Moreover, hyper HIV-1 has emerged in the mice encoding stable APOBEC3H, originally infected with hypo HIV-1 (36). Altogether, these findings suggest that stable variants of APOBEC3H impose selective pressure on HIV-1. More importantly, the expression levels of these APOBEC3 genes are upregulated in the CD4⁺ T cells of humanized mice infected with HIV-1 (32, 36). As global transcriptome analyses have also indicated the upregulation of ISG expression levels (36), it can be said; IFN-I responses can be triggered by HIV-1 infection in humanized mice.

The human genome encodes two IFN-inducible MX dynamin-like guanosine triphosphate hydrolases (GTPases), MX1 and MX2 (also known as MXA and MXB, respectively), which are presumably created by gene duplication over the course of evolution (40, 41). In addition to MX2’s strong anti-HIV-1 activity (42–44), it has also been shown that MX1 can suppress a wide range of other pathogenic DNA and RNA viruses, not including HIV-1 (42, 44, 45).

It is well known that IFN-I stimulation strongly inhibits HIV-1 infection during the early stages of viral replication (i.e., from entry to integration process) (46). To determine the IFN-I-responsive RF(s) restricting HIV-1 replication, comparative gene expression profiling (i.e., mRNA microarray) was conducted using human cells in the presence and the absence of IFN-I, and identified MX2 as the determining factor (42, 44). Subsequent investigations revealed that MX2 participates in blocking HIV-1 infection after reverse transcription (42–44). Since MX2 overexpression reduces the levels of nuclear viral DNA (e.g., 2-LTR circles) and more efficiently suppresses HIV-1 infection in non-dividing cells when compared with dividing cells (42, 44), MX2 presumably inhibits nuclear import of the viral complex. Moreover, MX2-mediated anti-viral potency is dependent on the viral capsid protein, as N57S and G89V mutants in the HIV-1 capsid render resistance to MX2 (42). Furthermore, a previous study has suggested that MX2-mediated restriction can be overcome by the depletion of cyclophilin A (CYPA), a peptidylprolyl isomerase (officially designated PPIA), and the treatment of cyclosporin A, a compound inhibiting CYPA (43). These findings suggest that CYPA is required for MX2-mediated anti-viral activity. CYPA is a well-known interaction partner of the HIV-1 capsid protein [reviewed in Ref. (47)]. Therefore, it is plausible that MX2 is closely associated with the viral complex composed of HIV-1 capsid and lines of cellular proteins.

Guanosine triphosphate hydrolase activity is required for the anti-viral effect of MX1 (48). In contrast, the K131A and T151A mutants of MX2, which lose the ability of GTP binding and hydrolysis, respectively, still exhibit anti-HIV-1 activity that is comparable with wild-type MX2 (42), suggesting that the MX2’s enzymatic activity appears to be dispensable for its anti-viral effect. On the other hand, the deletion of the nuclear localization signal at the N-terminus of MX2 results in the loss of anti-viral activity (42). These observations suggest the importance of the nuclear localization signal for MX2 to exhibit anti-HIV-1 activity, although the functional importance of the subcellular localization of MX2 remains unclear.

Schlafen 11 is also an ISG and potently inhibits HIV-1 production (49). Since SLFN11 overexpression suppresses viral protein expression but not viral transcription, this RF restricts viral replication at a post-transcriptional stage (49). Interestingly, the protein expression of codon-optimized HIV-1 Gag as well as GFP does not affect SLFN11 overexpression or knockdown. Moreover, SLFN11 binds to tRNA and impairs protein expression based on codon usage (49). Altogether, SLFN11 restricts HIV-1-biased translation in a codon-usage-dependent manner (49). Furthermore, a subsequent study has revealed that primate SLFN11 genes are under evolutionarily positive selection pressure and commonly possess the ability to impair viral production regardless of the virus or host target (50). However, it remains unclear how SLFN11 influences HIV-1 replication in vivo.

Guanylate-binding protein 5 belongs to an IFN-inducible subfamily of GTPases with host defense activity against intracellular bacteria and parasites (51). Krapp et al. have recently demonstrated that GBP5 suppresses HIV-1 infectivity by interfering with N-linked glycosylation of the viral envelope glycoprotein (Env) (52). The cysteine residue at position 583 is critical for its anti-viral activity; however, catalytically inactive mutants still demonstrate anti-viral activity, suggesting GBP5 exhibits an anti-viral effect independent of its enzymatic activity (52).

Intriguingly, the nonsense mutations in the HIV-1 vpu gene (i.e., the deletion of initiation codon or the insertion of premature stop codons) increase Env expression and confer resistance to GBP5-mediated anti-viral activity (52). As described below, viral protein U (Vpu) is a crucial factor for the counteraction of tetherin-mediated restriction (53, 54). However, it should be noted that the initiation codons of the vpu gene in certain clinical HIV-1 isolates including HXB2 (55), BH8 (56), MAL (57), and Zr6 (58) are primarily deleted. Additionally, the expression of GBP5 gene is upregulated by HIV-1 replication in infected individuals (52). Therefore, it might be plausible to assume that conferring resistance to GBP5 is important for HIV-1 dissemination in certain tissues or organs in infected individuals, and that there is a “trade-off” relationship between anti-tetherin activity (presence of Vpu) and GBP5 resistance (absence of Vpu).

The observation that the HIV-1 accessory protein, Vpu, is required for the efficient release of HIV-1 particles depending on cell type indicated the existence of an RF counteracted by Vpu (59–64). In 2008, Neil et al. and Van Damme et al. identified tetherin (also known as bone marrow stromal antigen 2, CD317, and HM1.24) (53, 54). Tetherin is an IFN-I-inducible type II membrane protein that consists of an N-terminal cytoplasmic tail, a transmembrane domain, and an extracellular domain with a glycosylphosphatidylinositol (GPI) modification at the C-terminus (65). Due to GPI anchoring, tetherin is mainly localized in cholesterol-enriched lipid rafts (65), where HIV-1 viruses bud from, and retains budding virions on the plasma membrane of virus-producing cells (66). To antagonize the tetherin-mediated anti-viral action, Vpu downregulates tetherin from the surface of HIV-1-producing cells (54, 67). Vpu is a multifunctional type I transmembrane protein [reviewed in Ref. (68)] and sequesters tetherin molecules from the cell surface to endosomal compartments through transmembrane domain-mediated interaction (69–72). Additionally, the DSGXXS motif in the cytoplasmic tail of Vpu interacts with BTRC1 (beta-transducin repeat containing E3 ubiquitin protein ligase; also known as β-TrCP1 and Fbxw1), a subunit of E3 ubiquitin ligase. In this way, Vpu induces tetherin ubiquitination and enhances subcellular sorting of tetherin mediated by endosomal sorting complexes required for transport machinery into lysosomal compartments for degradation (72). The requirement of BTRC1 for tetherin antagonization, however, remains controversial.

To reveal the importance of Vpu in the dynamics of HIV-1 replication in vivo, Sato et al. (73) and Dave et al. (74) utilized HSC-transplanted humanized mouse models and demonstrated that Vpu strongly downregulates the expression level of tetherin on the surface of virus-producing cell in vivo. The replication kinetics of vpu-deficient HIV-1 during the early phase of infection is clearly lower than that of wild-type HIV-1 in humanized mice (73, 74), suggesting that Vpu augments HIV-1 replication during the acute phase of infection.

In addition to the tetherin’s ability to impair viral release, it can also be an inducer of NFκB activation (75, 76). The molecular mechanism of tetherin-mediated NFκB activation has been well investigated in in vitro cell cultures (75–77). However, the importance of NFκB signaling triggered by tetherin in HIV-1 replication in vivo remains unknown and needs to be addressed in future investigations.