|Home | About | Journals | Submit | Contact Us | Français|
The APOBEC3H gene is polymorphic in humans, with four major population-dependent haplotypes that encode proteins with different levels of antiviral activity. Haplotype II, present most frequently in African populations, encodes the most stable protein and is most active against human immunodeficiency virus type 1 (HIV-1). In contrast to human APOBEC3G, which can be completely counteracted by HIV-1 Vif, the protein encoded by APOBEC3H haplotype II is only partially sensitive to Vif, while the protein encoded by APOBEC3H haplotype I is completely resistant to HIV-1 Vif. We mapped a residue on APOBEC3H that determines this partial Vif sensitivity. However, it is unclear how HIV-1 can replicate in vivo without the ability to neutralize APOBEC3H antiviral activity. In order to directly address this question, we cloned vif genes from HIV-1-infected individuals with different APOBEC3H genotypes and tested them for their ability to inhibit human APOBEC3H. We found that while the APOBEC3H genotype of infected individuals significantly influences the activity of Vif encoded by their virus, none of the Vif variants tested can completely neutralize APOBEC3H as well as they neutralize APOBEC3G. Consistent with this genetic result, APOBEC3H protein expression in human peripheral blood mononuclear cells was below our limit of detection using newly developed antibodies against the endogenous protein. These results demonstrate that human APOBEC3H is not as strong of a selective force for current HIV-1 infections as human APOBEC3G.
APOBEC3 (apolipoprotein B mRNA-editing catalytic polypeptide) proteins belong to a family of cytidine deaminases that have antiviral and antiretroelement functions (16). APOBEC3 proteins have been shown to restrict various retroviruses by causing cytidine-to-uridine editing in minus-sense viral DNA and by a deaminase-independent mechanism that acts to block the completion of reverse transcription (1, 2, 10). In order to achieve productive infection in cells expressing APOBEC3, all known modern lentiviruses, except equine infectious anemia virus, encode a viral protein called Vif, which counteracts APOBEC3 antiviral activity. Vif binds to APOBEC3 and recruits the E3 ubiquitin ligase complex to APOBEC3, which leads to the polyubiquitination and subsequent degradation of APOBEC3 by the proteasome (16).
The APOBEC3 family of antiviral genes has expanded during mammalian evolution. Rodents have a single APOBEC3 gene, whereas other placental mammals encode multiple genes (4, 6, 14). In humans, chromosome 22 carries seven APOBEC3 genes: APOBEC3A, APOBEC3B, APOBEC3C, APOBEC3DE, APOBEC3F, APOBEC3G, and APOBEC3H. However, deletions of APOBEC3B are common in some human populations (12). APOBEC3G and other APOBEC3 genes have been under intense positive selection during primate evolution (19, 23), presumably to adapt to a changing landscape of viral pathogens that can evade the action of these antiviral proteins. However, it is not yet known the extent to which the evolution of different APOBEC3 family members has occurred in response to different viral pathogens.
An important characteristic of the Vif interaction with the APOBEC3 proteins is that it is often species specific. For example, the Vif protein encoded by human immunodeficiency virus type 1 (HIV-1) is active against human APOBEC3G but not APOBEC3G from African green monkey, which is the natural host of the simian immunodeficiency virus SIVagm. Similarly, APOBEC3G from African green monkey is sensitive to SIVagm Vif but not HIV-1 Vif due to one amino acid difference in APOBEC3G (3, 17, 24, 33). Compared with human APOBEC3G, human APOBEC3F inhibits HIV-1 with less potency but demonstrates greater resistance to neutralization by Vif (26, 32). Nonetheless, it was previously shown that HIV-1 Vif utilizes two distinct regions to counteract APOBEC3G and APOBEC3F, which suggests that HIV-1 has simultaneously evolved to evade at least two different APOBEC3 family members (15, 21, 26, 36).
Previous studies done by our laboratory and other groups have shown that APOBEC3H, the most diverged from other members of the APOBEC3 family in primates, is polymorphic in humans (9, 18, 19, 28). Among the four major haplotypes, haplotype II encodes a protein with the longest half-life and is the most active against vif-deficient HIV-1 and non-long-terminal-repeat retrotransposons in tissue culture (9, 18). This is a unique feature of APOBEC3H since genetic polymorphisms of other APOBEC3 genes in humans are not known to exhibit such functional dichotomy. The loss of stability of human APOBEC3H proteins can be mapped to two independent polymorphic changes (R105G and Del15N) (18). Interestingly, the protein encoded by APOBEC3H haplotype II is partially resistant to Vif from HIV-1 LAI and completely resistant to NL4-3 Vif (9, 18). It is not clear, therefore, how HIV-1 overcomes human APOBEC3H in vivo since, at least as measured by PCR, the RNA is expressed in human peripheral blood mononuclear cells (PBMCs) (19).
Here, we examined this question by looking at the sensitivity of human APOBEC3H haplotypes to different vif genes, including the vif genes cloned from HIV-1-infected people with different APOBEC3H genotypes and from other primate lentiviruses, and by looking directly at protein expression with newly developed antibodies. We identified a single polymorphic site in APOBEC3H (amino acid 121) that determines its partial sensitivity to HIV-1 Vif. Moreover, vif variants isolated from HIV-1-infected individuals with haplotype I or II demonstrate differential activity against APOBEC3H. However, no HIV-1 Vif protein is able to completely neutralize APOBEC3H, although HIV-2 Vif can do so. Consistent with these genetic results, we showed that the APOBEC3H protein is below our detection level in human PBMCs with available antibodies. Taken together, these data show that APOBEC3H has less impact on current HIV infections in human T cells than human APOBEC3G.
The hemagglutinin tag at the 5′ end of the previously described human and macaque APOBEC3H cDNAs (18, 19) was removed by PCR amplifying APOBEC3H cDNA with a primer lacking the hemagglutinin tag sequence and cloning the fragment into the EcoRI/XhoI sites of pcDNA3.1 (Invitrogen). Point mutations were introduced by site-directed mutagenesis using the QuikChange kit (Stratagene), and the entire insert was resequenced.
HEK293T cells were maintained in Dulbecco's modified Eagle's medium-1% penicillin-streptomycin-10% bovine growth serum at 37°C in a CO2 incubator. SupT1 cells were maintained similarly in RPMI medium-1% penicillin-streptomycin-10% bovine growth serum. Human PBMCs were isolated from healthy donors by using Ficoll gradient centrifugation, and whole-cell lysates were prepared from them for immunoblotting. Transfections were performed with TransIT-LT1 transfection reagent (Mirus Bio) at a reagent-to-plasmid DNA ratio of 3:1. Western blot analyses were performed as previously described (18). A 1:2,500 dilution of human APOBEC3H antibodies (P1H6-1 and P1D8-1), a 1:5,000 dilution of human APOBEC3G antibody, and a 1:10,000 dilution of cyclophilin A antibody (Biomol) were used.
Genomic DNA was isolated from human PBMCs with a QIAamp DNA blood minikit (Qiagen). Two different PCR-based approaches were developed for the genotyping of the DelN15 and R105G polymorphisms. For the Del15N polymorphism, genomic DNA was amplified with fluorescence-labeled primers, and DNA fragment analysis was performed as previously described (18). For the R105G polymorphism, two primer pairs were used in a tetraprimer ARMS-PCR to amplify the two different alleles, respectively, of the single nucleotide polymorphism at position 105 (35). The PCR products were then resolved on a 1% agarose gel to visualize the presence of a 191-bp fragment (105R/R), a 158-bp fragment (105G/G), or both fragments (105R/G).
A cloning system was developed for inserting vif genes from HIV-1 -infected people into a reporter vif-deleted viral construct. HIV-1 vif sequences were amplified in nested PCRs from genomic DNA with viral loads above 50,000 copies per ml of plasma from the University of Washington CFAR HIV Specimen Repository from HIV-1-infected people who self-identified as being African-Americans. These patients were most likely exposed to a subtype B virus, as subtype B is the predominant virus strain circulating in the United States (29). Thus, the primers for nested PCR were designed from the subtype B viral vif consensus sequence. The primers used for the first round of PCR amplified the viral genome between the end of integrase and the beginning of tat (5′-GTC TTA GGC TGA CTT CCT GGA TG and 5′-GGA ATA GAT AAG GCC CAA GAA GAA C). The primers used for the second round of PCR flanked the vif gene with MluI and Xba sites on the 5′ and 3′ ends, respectively (5′-ATA ACG CGT GGC CAC CAT GGA AAA C and 5′-GAT TCT AGA CCT AGT GTC CAT TCA TTG). The nested PCR products were ligated into a TA vector from a Promega pGEM-T Easy kit and sequenced. These vif genes were then ligated into the MluI/Xba sites of a vif-deleted HIV proviral plasmid that contains a deletion in the env gene and has the firefly luciferase gene inserted into Nef (pLai3ΔenvLuc2ΔvifLk). In this construct, all the start codons in the region from the original start of vif to the end of the integrase gene were mutated to prevent the expression of wild-type vif. The portion from the end of the integrase gene to the AvrII site in the vpr gene was deleted and replaced by new restriction sites (5′-SnaBI MluI XbaI HpaI-3′). Wild-type vif from LAI (20) was also cloned back into the vif-deleted viral construct to yield pLai3ΔenvLuc2ΔvifLk/Lai. vif genes from HIV-2 ROD9 (7) and SIVcpz TAN3.1 (27) were also amplified with specific primers and cloned into pLai3ΔenvLuc2ΔvifLk. All constructs were confirmed by sequencing.
APOBEC3H haplotype I was cloned into the bacterial expression plasmid pTrC-His (Invitrogen) under an IPTG (isopropyl-β-d-thiogalactopyranoside)-inducible promoter with a His tag at the C terminus and transformed into BL21(DE3) RIL cells (Invitrogen). Three hundred milliliters of culture was grown to an optical density at 600 nm of 0.8 and then induced with 1 mM IPTG for 3 h at 37°C. The cell pellet was resuspended in denaturing buffer, freeze-thawed, treated with 1 mg/ml lysozyme on ice for 90 min, and sonicated on ice. The soluble material was then loaded onto a Ni-Sepharose 6 Fast Flow column (GE HealthCare), washed extensively in buffer containing 30 mM imidazole, and then eluted in a buffer containing 8 M urea at pH 4.0. The peak fraction showed only one band (20 kDa) on a Coomassie blue-stained gel with a protein concentration of 5.6 mg/ml.
Polyclonal and monoclonal antibodies against human APOBEC3H were generated in RBF/DnJ outbred mice (Jackson Laboratories) as described previously (31). Two mice were injected with 50 μg of protein and then boosted eight times over a period of 1 year. After a final boost with 100 μg of protein, polyclonal sera were collected at autopsy, and fusions were generated. Monoclonal antibodies were screened for reactivity to the bacterial APOBEC3H protein by enzyme-linked immunosorbent assay and then rescreened on Western blots containing whole-cell lysates collected from 293T cells transfected with expression plasmids for the human APOBEC3H haplotype I protein, the human APOBEC3H haplotype II protein, rhesus macaque APOBEC3H, chimpanzee APOBEC3H, human APOBEC3A, human APOBEC3B, human APOBEC3C, human APOBEC3DE, human APOBEC3F, and human APOBEC3G. Three monoclonal antibodies (P1H6-1, P1D8-1, and P5H9-A11) that reacted with only the APOBEC3H proteins were identified.
Single-round HIV-1 infectivity assays were performed as previously described (19, 34). All assays were performed by the transfection of 1.25 × 105 293T cells in 24-well plates with approximately a 1:1 ratio of pcDNA3/APOBEC plasmid (200 to 250 ng) to 250 ng of pLai3ΔenvLuc2 (34) or Δvif proviral plasmid (19). For functional assays of patient vif genes, 1.25 × 105 293T cells in 24-well plates were transfected with approximately a 1:1 ratio of pcDNA3/APOBEC plasmid (250 ng) to pLai3ΔenvLuc2ΔvifLk, pLai3ΔenvLuc2ΔvifLk/Lai, or pLai3ΔenvLuc2ΔvifLk/Patient# (cloned patient vif genes). Virus equivalent to 2 ng of p24CA was used to infect 4 × 104 SupT1 cells in a 96-well plate in the presence of 20 μg/ml DEAE-dextran. After 48 h, cells from triplicate infections were lysed in 100 μl of Bright-Glo luciferase assay reagent (Promega) and read on a luminometer.
Virus stocks (4 ml per sample) were filtered and spun in an ultracentrifuge at 24,000 rpm for 1.5 h at 4°C. The viral pellets were resuspended in loading dye and analyzed by immunoblotting using antibodies specific for APOBEC3H or p24gag (mouse monoclonal antibody 24-2) (25).
Previous studies showed that human APOBEC3H is polymorphic, with four major haplotypes in the human population (called haplotypes I, II, III, and IV) (Fig. (Fig.1A).1A). Unlike APOBEC3F and APOBEC3G, the protein encoded by human APOBEC3H haplotype II is only partially neutralized by Vif of HIV-1 LAI (18) or not at all by Vif of HIV-1 NL4-3 (9). Since those previous studies were done with epitope-tagged APOBEC3H proteins (9, 18), which could potentially confound the interactions between the protein and HIV-1 Vif, we developed antibodies to APOBEC3H proteins that allowed us to detect untagged versions of this protein (see Materials and Methods). Consistent with data from previous reports using epitope-tagged proteins, the proteins encoded by both haplotypes III and IV are very poorly expressed and are completely inactive against HIV-1 (Fig. (Fig.1B).1B). The expression of the protein encoded by APOBEC3H haplotype I is more easily detectable than the proteins encoded by haplotypes III and IV and is weakly active against HIV-1 (63% infectivity compared to the no-APOBEC control) (Fig. (Fig.1B).1B). APOBEC3H haplotype II yields the highest steady-state levels of protein after transfection and is highly active against vif-deficient HIV-1 (3% infectivity compared to the no-APOBEC control) (Fig. (Fig.1B1B).
We then compared the Vif sensitivities of untagged APOBEC3H proteins. As a control, HIV-1 restriction by human APOBEC3G was completely rescued by Vif (Fig. (Fig.1B).1B). However, the protein encoded by APOBEC3H haplotype II is only partially neutralized by HIV-1 LAI Vif (about sevenfold virus rescue). On the other hand, the small amount of antiviral activity encoded by haplotype I is completely resistant to the effects of HIV-1 LAI Vif (Fig. (Fig.1B).1B). Because of the differences in the sensitivities of the proteins encoded by APOBEC3H haplotypes I and II to HIV-1 Vif, we could determine which of the polymorphic changes in this gene contributes to their resistance to Vif. Among the three polymorphic changes that differ between haplotypes I and II, the mutation at position 105 from a glycine (haplotype I) to an arginine (haplotype II) in the protein encoded by APOBEC3H haplotype I was previously shown to greatly increase expression levels (9, 18) and, in our study, the antiviral activity of this protein against HIV-1 LAI (Fig. 1A and B). However, this polymorphism did not affect the sensitivity of APOBEC3H to HIV-1 Vif (Fig. (Fig.1B),1B), which demonstrates that although position 105 is a determinant for APOBEC3H protein stability, it is not a determinant for APOBEC3H sensitivity to Vif (Fig. (Fig.1B1B).
We also analyzed the virion encapsidation of different human APOBEC3H proteins and found that the level of APOBEC3H packaging corresponds to its resistance to HIV-1 Vif. Similar levels of proteins encoded by haplotypes I and II and the haplotype I G105R mutant were packaged into the virions regardless of the presence of Vif, which suggests that HIV-1 Vif is unable to efficiently target these APOBEC3H proteins for degradation by the proteasome (Fig. (Fig.1C).1C). On the other hand, we found that mutating position 121 from a lysine (haplotype I) to an aspartic acid (haplotype II) increases APOBEC3H antiviral activity, which is consistent with a previous finding (9). This mutant demonstrates a greater inhibition of vif-deficient HIV-1 than the protein encoded by the haplotype I G105R mutant (Fig. (Fig.1B).1B). Most importantly, mutating the basic residue at position 121 to an acidic one leads to increased APOBEC3H sensitivity to HIV-1 Vif (about fivefold virus rescue) (Fig. (Fig.1B).1B). Thus, the polymorphic change at position 121 is critical for the sensitivity of APOBEC3H to HIV-1 Vif in addition to APOBEC3H antiviral activity.
We next asked why the protein encoded by human APOBEC3H haplotype II is only partially neutralized by HIV-1 Vif from LAI, while human APOBEC3G is almost completely neutralized by the same Vif (Fig. (Fig.1B).1B). If APOBEC3H poses an important block to viral replication, one would expect that HIV-1 would have to overcome it with its viral vif gene in order to replicate efficiently in vivo. We reasoned that the vif genes in the laboratory strains might not reflect the selection for Vif activity in an HIV-infected person who encodes the more active APOBEC3H haplotype II protein (9, 18). APOBEC3H haplotype II is found mostly in African and African-American populations but is uncommon among European/Caucasian and Asian populations (18). However, our study and others investigating the APOBEC3H interaction with HIV-1 Vif were done with standard HIV laboratory strains derived from individuals of unknown APOBEC3H genotypes. Therefore, we hypothesized that HIV-infected individuals who are homozygous or heterozygous for the stable APOBEC3H allele (haplotype II) might harbor viral Vif variants that have stronger antagonistic activities against the APOBEC3H haplotype II protein than the previously tested Vif proteins. In other words, if APOBEC3H is important for virus replication in patients, these viruses would encode variations of Vif that can neutralize the haplotype II protein better than the viruses infecting individuals that have one of the unstable APOBEC3H proteins.
In order to test the activity of Vif proteins isolated from HIV-infected people who encode the most stable and active form of APOBEC3H, we obtained PBMCs of HIV-1-infected African-Americans who had a viral load of greater than 50,000 copies per ml of plasma and genotyped their polymorphisms of APOBEC3H (R105G and DelN15) (see Table S1 in the supplemental material). Three individuals were chosen for further study: one individual who has two copies of haplotype II (patient 1203), one individual with one copy of haplotype II and one copy of haplotype I (patient 1440), and one individual with two copies of haplotype I (patient 1393). We then amplified vif variants from genomic DNA isolated from the PBMCs of each of these individuals. The phylogenetic relationships of the vif genes from viruses in the PBMCs of these three patients are of subtype B and were not contaminated with our standard laboratory strain LAI since the vif sequences from each patient form a separate group (Fig. (Fig.2A).2A). Consensus Vif sequences from each patient share about 85 to 91% identity with LAI Vif and 81 to 86% identity with each other (see Fig. S1 in the supplemental material).
The vif variants from these three patients were then cloned into a vif-deleted HIV-1 luciferase reporter construct in place of the original vif gene (see Materials and Methods for details). We first tested the ability of the patient Vif variants to neutralize human APOBEC3G by determining the viral infectivity measured by the luciferase activity of infected SupT1 cells in the presence of Vif. We tested five vif clones from each patient (Fig. (Fig.2A),2A), and each one was tested in triplicate in two different experiments. The infectivity data matched with the corresponding patient vif gene are shown in Table S2 in the supplemental material. We found that the proteins encoded by the vif alleles from all three patients are able to effectively neutralize APOBEC3G and rescue viral infectivity up to a level similar to that of Vif from HIV-1 LAI (Fig. (Fig.2B).2B). This indicates that the vif genes that we cloned are active and not defective.
We next tested the ability of vif genes cloned from HIV-1-infected people with different APOBEC3H haplotypes to neutralize the human APOBEC3H haplotype II protein. As shown in Fig. Fig.1B,1B, HIV-1 LAI Vif only partially neutralizes the antiviral effect of human APOBEC3H (Fig. (Fig.2B).2B). Likewise, we found that Vif variants from the patients either showed minimal to no virus rescue or showed about the same modest level of rescue as LAI (Fig. (Fig.2B).2B). Strikingly, the proteins encoded by the vif genes isolated from a homozygous haplotype II patient (patient 1203) demonstrate a wider range of activity against APOBEC3H, and some of them are able to rescue virus at up to 50 to 70% infectivity (Fig. (Fig.2B2B).
A one-way analysis of variance (ANOVA) to compare the means of the activities encoded by vif genes from the three patients against APOBEC3H found that at least one of the patient pairs is significantly different (P < 0.0001). In particular, a Student's t test showed that the infectivity of viruses carrying the vif genes isolated from the homozygous haplotype II patient (patient 1203) is significantly different (P = 0.0023) from that of viruses carrying the vif genes isolated from the homozygous haplotype I patient (patient 1393). These data suggest that the haplotype II protein is mostly resistant to inactivation by the vif genes from viruses infecting patients with one copy of haplotype II or no copies and from the standard HIV-1 laboratory strain. However, the individual carrying two copies of the stable allele of APOBEC3H does harbor viral Vif variants that can neutralize the haplotype II protein to a significantly greater extent. Carrying two copies of haplotype II might confer some advantage against HIV-1 to individuals, although a larger cohort needs to be screened. Interestingly, the activity of the Vif proteins from the homozygous haplotype II patient against APOBEC3G is also significantly different from that against APOBEC3H (P = 0.0001). In other words, none of the patient Vif variants that we tested are as potent against APOBEC3H as they are against APOBEC3G. Therefore, these data suggest that while the APOBEC3H genotype of HIV-1-infected people significantly influences the activity of the Vif protein from their own virus, the selective pressure on HIV-1 Vif due to APOBEC3H expression is not strong enough to drive the evolution of Vif to neutralize APOBEC3H as well as it neutralizes APOBEC3G.
We hypothesized that perhaps HIV-1 vif genes have not evolved to neutralize APOBEC3H as well as they neutralize APOBEC3G because human APOBEC3H is poorly expressed in target cells of HIV-1. In order to test this hypothesis, we analyzed human PBMC samples for endogenous APOBEC3H expression. Two monoclonal antibodies, P1H6-1 and P1D8-1, reacted against human protein expressed in 293T cells with minimal background (Fig. (Fig.3,3, lane 1). We probed whole-cell lysates made from PBMCs of HIV-1-infected individuals who were the same ones genotyped for the data in Fig. Fig.22 and were homozygous for haplotype I (patient 1393) (Fig. (Fig.3,3, lane 2), heterozygous haplotype I and haplotype II (patient 1440) (Fig. (Fig.3,3, lane 3), and homozygous for haplotype II (patient 1203) (Fig. (Fig.3,3, lane 4). The amount of APOBEC3H protein expression in PBMCs of infected donors was below our level of detection, as no band of 20 kDa was visible, even for the individual that is homozygous for haplotype II (patient 1203) (Fig. (Fig.3,3, lane 4). In addition, we probed whole-cell lysates made from PBMCs of healthy donors with different APOBEC3H genotypes (Fig. (Fig.3,3, lanes 5 to 8). Healthy donor 2 carries one copy of haplotype II. The amount of APOBEC3H protein expression was again below our level of detection. We also carried out threefold serial dilutions of whole-cell lysates collected from 293T cells transfected with exogenous APOBEC3H to show that our immunoblotting system could still detect APOBEC3H in a total protein level that is equivalent to a 1:720 dilution of human PBMC samples loaded. We then stripped the blot and reprobed it with human APOBEC3G antibody. A band of the expected size (about 40 kDa) was detected in the lanes loaded with lysates of patient PMBCs, which suggests that the absence of APOBEC3H in these samples is not due to the degradation of the endogenous protein by HIV-1 Vif. Interestingly, the level of APOBEC3G protein expression is higher in patient PBMCs than in healthy PBMCs, which is consistent with a previous observation of a modest induction of APOBEC3G mRNA levels in CD4+ T cells by alpha interferon treatment (13). Thus, the low level or lack of APOBEC3H protein expression in the PBMCs of HIV-1-infected individuals may explain the lack of selective pressure for Vif proteins from these individuals to evolve to completely neutralize this antiviral protein.
A previous study has shown the complete inactivation of rhesus macaque APOBEC3H by SIVmac, SIVagm, and HIV-2 Vif proteins (30). Given the differences between the abilities of rhesus and human APOBEC3H proteins to be neutralized by SIV and HIV-1 Vif proteins, we tested the proteins encoded by APOBEC3H haplotypes against Vif from other lentiviruses that infect primates to see if human APOBEC3H insensitivity to Vif is species specific. We PCR amplified and cloned the vif genes from the full-length infectious clones of SIVcpz clone TAN3.1 (27) and HIV-2 ROD9 (7) into our vif-deficient HIV-1 LAI luciferase reporter construct for testing against human APOBEC3G and APOBEC3H. Both the SIVcpz and HIV-2 Vif proteins, along with LAI Vif, are able to completely neutralize human APOBEC3G, which is shown by a virus rescue of 62- to 100-fold (Fig. (Fig.4).4). Similar to our findings with HIV-1 Vif proteins, the human APOBEC3H haplotype I and II proteins are highly resistant to SIVcpz Vif (no virus rescue) (Fig. (Fig.4).4). On the other hand, the proteins encoded by both human APOBEC3H haplotypes can be neutralized by HIV-2 Vif (5- and 131-fold virus rescues, respectively) (Fig. (Fig.4).4). These results suggest that Vif proteins from the SIVcpz/HIV-1 lineage have specifically lost the ability to neutralize APOBEC3H, while this activity has been retained in another lineage of primate lentiviruses.
We find that vif genes isolated from an HIV-1-infected individual who is homozygous for the most active form of APOBEC3H (haplotype II) are significantly better at antagonizing APOBEC3H than the vif genes from the heterozygote or the homozygote for the less active form of APOBEC3H (haplotype I). Nonetheless, in all cases, human APOBEC3H is less sensitive to neutralization by HIV-1 Vif than is human APOBEC3G. These results suggest that APOBEC3H is not as strong of a selective force on HIV-1 vif evolution as APOBEC3G and imply that the APOBEC3H protein is not well expressed in HIV-1 target cells. Indeed, using newly developed monoclonal antibodies against APOBEC3H, we find that protein is undetectable in the PBMCs of HIV-1-infected and uninfected humans. Furthermore, we show that a single polymorphic amino acid at position 121 in human APOBEC3H is the major determinant for the sensitivity of human APOBEC3H to neutralization by HIV-1 Vif. This amino acid is distinct from the polymorphism that was previously shown to be responsible for the differences in protein stabilities among human APOBEC3H haplotypes, which is determined by R105G and Del15N (18). These results demonstrate that human APOBEC3H does not play a role as important as APOBEC3G in current HIV-1 infections but suggest that this protein may have evolved to counteract other viral pathogens in the past.
We showed in this study that the amount of the APOBEC3H protein was below the detection level in PBMCs of both HIV-1-infected and uninfected humans. This result is consistent with the functional data that suggest that none of the Vif variants cloned from individuals with a stable APOBEC3H allele are able to completely rescue HIV-1 infectivity from restriction by human APOBEC3H. In other words, the endogenous expression of APOBEC3H in PBMCs might be insufficient compared to the endogenous APOBEC3G expression to completely drive the evolution of HIV-1 vif alleles from most infected individuals. On the other hand, we did see a statistically significant difference in the ability of the Vif proteins from an individual homozygous for the APOBEC3H haplotype II (the haplotype that makes the most stable protein) to neutralize APOBEC3H relative to that of the Vif proteins from people with other genotypes. This implies that there is a low level of APOBEC3H in PBMCs that is below our level of detection with our current antibodies or else in an HIV-target cell such as mucosal T cells or macrophages that we have not yet measured. Alternatively, APOBEC3H haplotype II might be linked to a haplotype of another APOBEC3 gene in the individual homozygous for haplotype II (patient 1203) that is exerting evolutionary pressure on the vif gene of his virus.
In contrast to the inability of HIV-1 Vif to encounter human APOBEC3H, SIVmac, SIVagm, and HIV-2 Vif proteins have been shown to efficiently neutralize rhesus macaque APOBEC3H (30). This suggests that these Vif proteins have likely encountered and therefore evolved to inhibit the APOBEC3H proteins from Old World monkeys. One would predict that the APOBEC3H protein is expressed at significant levels in the PBMCs of Old World monkeys even though the expression of APOBEC3H in human PBMCs is undetectable. Moreover, we showed that HIV-2 Vif can counteract human APOBEC3H more effectively than can HIV-1 Vif. Since HIV-2 has a more recent origin in Old World monkeys than does HIV-1 (8), these results imply that HIV-2 retains Vif determinants that were useful in antagonizing APOBEC3H that it encountered in sooty mangabeys. Thus, some evolutionary event might have selected for the change in the transcriptional or posttranscriptional regulation of this antiviral gene in a species-specific manner that potentially caused the differential expression of the APOBEC3H protein between humans and Old World monkeys. Whatever event this was, its modern consequence is that the antiviral repertoire in human PBMCs has been reduced compared to that in Old World monkeys. Further studies should be carried out to look at endogenous APOBEC3H expression in macaque PBMCs.
Apart from the potential species-specific differences in the protein expression level, we also observed functional differences between the various versions of human APOBEC3H. We previously showed that the stability of the APOBEC3H protein was lost twice in human evolution (18). Here, we demonstrate that haplotype I carries a basic residue at position 121 (K121) that causes it to be completely resistant to HIV-1 Vif, while haplotype II contains an acidic residue at the same position (D121) that confers partial sensitivity to HIV-1 Vif. Interestingly, the region containing the amino acid responsible for human APOBEC3H sensitivity to HIV-1 Vif can be mapped to homologous regions in the N terminus of APOBEC3G and the C terminus of APOBEC3F that are important for their interaction with Vif (11, 22). This interaction domain appears to be conserved among different APOBEC3 proteins.
It is possible that at least one of the APOBEC3H haplotypes evolved to become more resistant to a Vif-like factor of an ancient pathogen that is now extinct. Consistent with this idea, our laboratory previously demonstrated that primate APOBEC3H genes have been shown to be under positive selection, which is indicative of “genetic conflict” between a host defense gene and viral pathogens (19). Although a role for human APOBEC3H in restricting HIV-1 in vivo is yet to be demonstrated since it expression in peripheral blood is low, human APOBEC3H might play a role in inhibiting other viruses that infect and replicate in the tissue types where the APOBEC3H protein is highly expressed. The availability of antibodies against the native protein described in this paper will allow us to describe the tissue expression patterns of human APOBEC3H more fully and should provide clues to what other viruses might be targets of this potent antiviral agent. The range of viral targets of APOBEC3H and the possible consequences of the differences in the activities of the protein among human populations have yet to be discovered.
We thank the University of Washington CFAR HIV Specimen Repository for human PBMC samples; Sarah Holte of the University of Washington CFAR Biometrics Core for advice on statistical analysis; Elizabeth Wayner of the FHCRC Antibody Development Shared Resources, FHCRC Genetic Analysis core; and Jaisri Lingappa for human APOBEC3G antibody. The following reagents were obtained through the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH: HIV-1 p24 Gag monoclonal antibody (catalog no. 24-2) from Michael H. Malim and SIVcpzTAN3.1 (catalog no. 11498) from Jun Takehisa, Matthias H. Kraus, and Beatrice H. Hahn. We thank Masahiro Yamashita, Semih Tareen, Nisha Duggal, Efrem Lim, Alex Compton, and Molly OhAinle for comments on the manuscript.
This work was supported by NIH grant R37 AI30937.
Published ahead of print on 14 October 2009.
†Supplemental material for this article may be found at http://jvi.asm.org/.