Polymorphisms and Splice Variants Influence the Antiretroviral Activity of Human APOBEC3H

Abstract

Human APOBEC3H belongs to the APOBEC3 family of cytidine deaminases that potently inhibit exogenous and endogenous retroviruses. The impact of single nucleotide polymorphisms (SNP) and alternative splicing on the antiretroviral activity of human APOBEC3H is currently unknown. In this study, we show that APOBEC3H transcripts derived from human peripheral blood mononuclear cells are polymorphic in sequence and subject to alternative splicing. We found that APOBEC3H variants encoding a SNP cluster (G105R, K121D and E178D, hapII-RDD) restricted human immunodeficiency virus type 1 (HIV-1) more efficiently than wild-type APOBEC3H (hapI-GKE). All APOBEC3H variants tested were resistant to HIV-1 Vif, the viral protein that efficiently counteracts APOBEC3G/3F activity. Alternative splicing of APOBEC3H was common and resulted in variants with distinct C-terminal regions and variable antiretroviral activities. Splice variants of hapI-GKE displayed a wide range of antiviral activities, whereas similar splicing events in hapII-RDD resulted in proteins that uniformly and efficiently restricted viral infectivity (>20-fold). Site-directed mutagenesis identified G105R in hapI-GKE and D121K in hapII-RDD as critical substitutions leading to an average additional 10-fold increase in antiviral activity. APOBEC3H variants were catalytically active and, similarly to APOBEC3F, favored a GA dinucleotide context. HIV-1 mutagenesis as a mode of action for APOBEC3H is suggested by the decrease of restriction observed with a cytidine deaminase domain mutant and the inverse correlation between G-to-A mutations and infectivity. Thus, the anti-HIV activity of APOBEC3H seems to be regulated by a combination of genomic variation and alternative splicing. Since prevalence of hapII-RDD is high in populations of African descent, these findings raise the possibility that some individuals may harbor effective as well as HIV-1 Vif-resistant intracellular antiviral defense mechanisms.

APOBEC3H is a member of the APOBEC3 family of cytidine deaminases, some of which possess strong anti-human immunodeficiency virus type 1 (HIV-1) activity (e.g., APOBEC3G/3F) (3, 8, 12, 15, 21, 34). HIV-1's ability to replicate in human cells depends on the expression of the viral protein Vif, which efficiently mediates the degradation of several APOBEC3 members in the producer cell (8, 12, 21).

APOBEC3H mRNA has been detected in several human tissues (e.g., peripheral blood mononuclear cells [PBMC], liver, skin, ovary, and testis) (19, 28). APOBEC3H lacks the cytidine deaminase domain (CDA) that mediates RNA binding, homodimerization, and virion encapsidation of APOBEC3G (13, 26). In contrast to the strong Vif-independent HIV-1 restriction exerted by the rhesus macaque APOBEC3H, the human protein seems to be limited in its antiretroviral activity (9, 28). Protein expression levels of human APOBEC3H and that of the rhesus homologue differ greatly upon transfection into mammalian cells (9, 28), suggesting that the lack of potency of human APOBEC3H is a reflection of insufficient expression and/or protein stability in the producer cell, rather than a lack of enzymatic activity. Indeed, human APOBEC3H displayed cytidine deaminase activity comparable to its rhesus homologue in a bacterial mutator assay (28). Moreover, APOBEC3H has been reported to cause hypermutation in both the hepatitis B virus (19) and in human papillomavirus genomes (33), suggesting the presence of enzymatic activity in mammalian systems.

Comparison between human and rhesus sequences revealed that rhesus APOBEC3H protein is 210 amino acids long, whereas the human homologue is shorter due to a premature translation termination codon. Repairing this stop codon resulted in a human APOBEC3H protein variant which was well-expressed in mammalian cells and displayed HIV-1 Vif-independent antiretroviral activity (10). A similar activity profile was observed when expression of the short human APOBEC3H variant was optimized using a cytomegalovirus (CMV) intron A-containing expression vector (10).

In this study, we report that multiple, distinct APOBEC3H variants with antiviral activity are present in PBMC from healthy donors. Specifically, we identified a cluster of three nonsynonymous single nucleotide polymorphisms (SNP) which in conjunction with a specific splice variant confer strong, HIV-1 Vif-resistant antiretroviral activity. This restriction correlated with the introduction of G-to-A mutations in HIV-1 proviruses in a GA dinucleotide context.

MATERIALS AND METHODS

Amplification of APOBEC3H transcripts.

Human PBMC were obtained by Ficoll (GE Healthcare) density centrifugation from 12 HIV-1-negative anonymous blood donors (Mount Sinai School of Medicine Blood Bank). Cells were cryopreserved in liquid nitrogen until total cellular RNA was extracted using a Qiagen RNA extraction kit. RNA was reverse transcribed with Superscript II (Invitrogen) and random hexamers. APOBEC3H variants were amplified with PicoMax DNA polymerase (Stratagene) using primers 5′-AAC GCT CGG TTG CCG CCG GGC GTT TTT TAT TAT GGC TCT GTT AAC AG and 5′-TCT TGA GTT GCT TCT TGA TAA T. PCR products were cloned using a StrataClone kit (Stratagene) as specified by the manufacturer. Six to 14 clones per donor were sequenced using BigDye Terminator v3.1 reagents and analyzed on an ABI Prism 3730xl apparatus (Agencourt Bioscience Corp.). Sequences were manually edited and aligned using DNAStar and Bioedit software packages.

Plasmids used for HIV-1 production.

Replication-competent full-length molecular clone NL4-3 Vif mutant SLQ144AAA (NL4-3 ΔSLQ) has previously been described (24). It lacks the required motif to bind to Elongin C, which is part of the E3 ligase complex Cullin5/Elongin B/C needed for APOBEC3 degradation (12).

Plasmid HIV-1 gag-pol (pCRV1/gag-pol) (16), the packageable HIV-1 RNA genome encoding Tat, Rev, Vpu, and GFP (pV1/hrGFP), the G protein from vesicular stomatitis virus (pHCMV VSV-G), and the Vif expression plasmid pCRV1-Vif have been described previously (30).

APOBEC3 expression plasmids.

Six of the most common APOBEC3H variants (hapI-GKE and hapII-RDD; SV-182, SV-183, and SV-200) were subcloned into the mammalian expression vector pTR600 (14). We chose pTR600 because its CMV intron A improves expression of the inserted transgene (14). APOBEC3H variants were amplified from StrataClone plasmids (see “Amplification of APOBEC3H transcripts,” above) using Pfu polymerase (Stratagene) and the following primers: 5′-GAT Caa gct tCG atg GAT TAC AAG GAT GAC GAC GAT AAG atg gct ctg tta aca gcc gaa ac (FLAG tag sequence is shown in italics) and 5′-TAA TAC GAC TCA CTA TAG GG. Upon restriction enzyme digestion and ligation into pTR600, the cloned inserts were verified by sequencing.

Site-directed mutagenesis of APOBEC3H.

Plasmid pTR600-hapI-GKE and pTR600-hapII-RDD (both SV-183) were used as templates for site-directed APOBEC3H mutagenesis. We used standard overlap PCR techniques to introduce mutations at positions 56, 105, 121, and 178. Mutation E56A is located in the deaminase active site (CDA) and has been shown to abolish catalytic activity in other APOBEC3 enzymes. Introduction of the correct mutation into the cloned fragments was confirmed by sequencing.

Culture of cell lines.

HEK 293T and TZM-bl reporter cells were maintained in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum and 100 U/ml penicillin-streptomycin. TZM-bl cells were provided by J. C. Kappes and X. Wu through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, National Institutes of Health.

Transfection of HEK 293T cells.

Viral stocks were obtained by transfection of HEK 293T cells using 4 μg/ml polyethylenimine (Polysciences, Inc.). pTR600-FLAG-APOBEC3G, pTR600-FLAG-hapI-GKE, and FLAG-hapII-RDD expression vectors (range, 50 to 1,000 ng) were cotransfected with NL4-3 wild type (WT), NL4-3 Vif mutant SLQ144AAA molecular clones (500 ng), or irrelevant plasmids (500 ng) in 24-well tissue culture plates.

HIV-1 vector particles were generated by transfecting HEK 293T cells with plasmids pCRV1/gag-pol, the packageable HIV-1 RNA genome pV1/hrGFP, and pHCMV VSV-G in a 5:5:1 ratio (30). To measure APOBEC3H and Vif functions, cells were cotransfected with this plasmid mixture and additional plasmids expressing the amino-terminally FLAG-tagged APOBEC3H variants with pCRV1empty or pCRV1/Vif wild type.

For all transfections, the culture medium was replenished after 12 h. Supernatants were harvested 2 days after transfection, clarified by centrifugation, and used to infect TZM-bl reporter cells.

Assessment of viral infectivity.

TZM-bl reporter cells, which carry an HIV-1 Tat-responsive β-galactosidase indicator gene under the transcriptional control of the HIV-1 long terminal repeat, were used to assess the infectivity of viral stocks produced by transfection in the presence and absence of the different FLAG-APOBEC3H variants or FLAG-APOBEC3G. TZM-bl cells were infected in triplicate with 20 μl cell-free viral supernatant in 96-well plates. β-Galactosidase activity was quantified 48 h after infection using chemiluminescent substrate (Tropix; Perkin-Elmer), as previously described (30).

Western blotting of cell lysates.

Cells were lysed in 1% sodium dodecyl sulfate, 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM EDTA, supplemented with EDTA-free protease inhibitor cocktail (Roche) 48 h posttransfection. Proteins were separated on 10% or 4 to 12% gradient polyacrylamide-sodium dodecyl sulfate gels (Invitrogen), transferred to polyvinylidene difluoride membranes (Pierce) and probed with anti-FLAG M2 monoclonal antibody (Sigma) for FLAG-APOBEC3G and FLAG-APOBEC3H variants. Membranes were subsequently incubated with horseradish peroxidase-conjugated secondary antibodies and developed with SuperSignal West Pico (Pierce). After stripping with 0.2 M NaOH for 10 min, membranes were probed with anti-glyceraldehyde-3-phosphate dehydrogenase (anti-GAPDH; Sigma) to ensure equal protein loading.

For quantification of protein expression, Western blots were developed as described above and analyzed using the Fujifilm Intelligent light box LAS-3000 and Image Reader LAS-3000 software. Signals were detected at supersensitive settings with 10-s increments. Only the nonsaturated signals were quantified using ImageGauge 4.0 software and used to calculate protein expression levels.

APOBEC3H-driven HIV-1 mutagenesis.

Viral stocks were generated by transfecting NL4-3 WT (500 ng) and pTR600-FLAG-APOBEC3H variants (50 ng), pTR600-FLAG-APOBEC3G (50 ng), or pTR600 (50 ng) in HEK 293T cells. Culture medium was replaced the next day and supernatants were harvested 36 h later. TZM-bl cells were infected in 24-well tissue culture plates with DNase I (Invitrogen)-treated viral stocks. At 12 h postinfection, the cells were extensively washed with phosphate-buffered saline and genomic DNA was extracted using a DNeasy DNA isolation kit (Qiagen). To assess the frequency of mutations in the proviral genome, a 1,905-nucleotide-long region of pol (HXB2, nucleotides 2928 to 4833) was amplified by PCR and cloned using a StrataClone kit as previously described (24). DNA sequencing was performed by Agencourt Biosciences using BigDye Terminator v3.1 reagents. Reverse transcriptase (RT) sequences (600 bp) were manually edited and aligned using DNAStar and Bioedit software packages. The frequency of G-to-A mutations and the dinucleotide context of the mutations were analyzed with the Hypermut program (29).

Statistical analysis.

Prism software (version 4.0; GraphPad Software) was used to perform all statistical tests. P values are two-sided, and values of <0.05 were considered to be significant.

Nucleotide sequence accession numbers.

The reference accession ID for the APOBEC3H SNPs (rs numbers) are rs139292 (Δ15N), rs139293 (R18L), rs139294 (synonymous G-to-C nucleotide substitution at position 129), rs139297 (R105G), rs139298 (K121E), rs139299 (K121N), and rs139302 (E178D). Representative APOBEC3H cDNA sequences were submitted to GenBank (accession numbers FJ376611 to FJ376617).

RESULTS

The current human APOBEC3H mRNA sequence information is based on some 30 different cDNA clones submitted to GenBank/dbEST (accessed May 2008). The APOBEC3H gene comprises seven GT-AG introns, and four alternatively spliced mRNAs are predicted to encode functional proteins (63, 182, 183, and 200 amino acids) (32; see also AceView [http://www.ncbi.nlm.nih.gov/IEB/Research/Acembly/index.html?human).

APOBEC3H variant NM_181773 (Ensembl ID ENS00000100298) has served as a wild-type reference (10) and will be referred to as hapI-GKE (in agreement with the nomenclature used by OhAinele et al. [27]). For the purpose of clarity, we named the different APOBEC3H splice variants based on the length of the predicted protein (e.g., SV-182, SV-183, and SV-200).

APOBEC3H is polymorphic in sequence.

Nine APOBEC3H SNPs (one synonymous [+129C/G; T43T], two single-codon deletions [Δ14N and Δ15N], and six nonsynonymous [R18L, G37H, G105R, K121E/N, S140G, and E178D]) are listed in the Single Nucleotide Polymorphism database at NCBI (www.ncbi.nlm.nih.gov/projects/SNP).

We amplified, cloned, and sequenced APOBEC3H transcripts derived from PBMC cDNA of 12 anonymous blood donors. Six to 14 APOBEC3H clones were analyzed for each donor (total, 106 clones).

We detected six SNPs in our data set at previously published polymorphic positions (Fig. 1A displays the exon location of the mutations; the rs numbers are listed in Materials and Methods section). APOBEC3H alleles encoding GKE at position 105, 121, and 178, respectively, were amplified from 10 of 12 donors, indicating that haplotype 1 is common (Fig. 1B). APOBEC3H transcripts encoding a cluster of three substitutions, G105R, K121D, and E178D, were found in 6 of the 12 donors (Fig. 1B). Haplotype II (HapII-RDD, with RDD standing for the substitutions at position 105, 121, and 178, respectively) was seen in 3 of the 12 donors (Fig. 1B). Deletion of asparagine at position 15 (Δ15N) with or without the substitution R18L (haplotypes III and IV, respectively) (Fig. 1B) was observed in combination with cluster RDD in four independent clones derived from three different donors (donors D1, D4, and D8). The synonymous mutation +129C (residue T43) was detected in six donors, always in conjunction with hapII-RDD, hapIII, or hapIV.

FIG. 1.

Sequence diversity of APOBEC3H transcripts derived from PBMC of healthy blood donors. (A) Summary of SNP detected in APOBEC3H transcripts derived from human PBMC. Locations of the five nonsynonymous mutations (black square) and the one synonymous mutation (square with black and white stripes) are indicated. CDA denotes the cytidine deamination site of the enzyme. (B) The APOBEC3H alleles with the different mutations at position 15, 18, 105, 121, and 178 are listed together with the frequency of detection in 12 PBMC donors analyzed in this study. (C) Summary of the genetic diversity of APOBEC3H SNP detected in human populations for SNP 105R, 121E, and 178D (based on the HapMap database). The details of each genotype reveal differences in the mutation frequencies for Utah residents with ancestry from Europe (CEU), Han Chinese (HCB), Japanese in Tokyo, Japan (JPT), and Yoruba from Ibadan, Nigeria (YRI).

Two donors harbored exclusively G105R/K121D/E178D APOBEC3H transcripts (donors P6 and P7) and six donors harbored only wild-type APOBEC3H, suggesting that these individuals are homozygous for hapII-RDD or hapI-GKE, respectively. Mixtures of wild-type and mutant transcripts (hapII-RDD, D11; hapIII, D1 and D8; hapIV, D4) were recovered from the PBMC of the remaining four donors (Fig. 1B). Our data indicate that APOBEC3H is polymorphic in sequence, with hapI-GKE and hapII-RDD being commonly represented in the 12 donors studied.

The International HapMap Consortium project (www.hapmap.org) provides information on the frequency of SNPs in four populations of diverse ethnicities (11, 18) and lists distribution for some of the APOBEC3H SNPs analyzed in this study. Of note, the aspartic acid (D) at position 121 in hapII-RDD is encoded by mutations at the first and third position of the triplet (K, AAG, to D, GAC), whereas the SNP database lists the polymorphisms separately (N, AAC; E, GAG), resulting in distinct residues. Figure 1C lists the frequencies of SNPs 105R, 121E, and 178D in four different populations. For example, SNP G105R is highly prevalent in sub-Saharan Africans (e.g., 93% of Yoruba [HapMap-YRI] encode G105R) but less often observed in other groups (e.g., only 39% of Europeans [HapMap-CEU] and 31% of Asians [HapMAp-HCB and JPT] encode G105R). Data for K121E and E178D reveal a similar ethnic bias supporting our observation that G105R/K121D/E178D occur as a cluster rather than as isolated substitutions (Fig. 1C).

APOBEC3H transcripts are subject to alternative splicing.

Estimates suggest that half of all human genes are subjected to alternative splicing, thereby generating transcriptome diversity in a cell-type- or tissue-specific manner (23).

Figure 2A depicts the four alternatively spliced APOBEC3H forms (SV-183, SV-182, SV-200, and SV-154) that were found in at least two different blood donors. All transcripts with the exception of SV-154 share exons 2, 3, and 4 with SV-182 lacking only a glutamine in the second from last position in exon 5. In contrast, skipped and/or cryptic exons in SV-200 (cryptic exon 4b) and SV-154 (skipped exon 4 and cryptic exon 4b) result in 19 and 15 distinct amino acids, at the C terminus, respectively (Fig. 2B shows the predicted protein sequences of each of these transcripts). A limited number of matches are present in databases for SV-182 (nine cDNA clones) and SV-200 (two cDNA clones), while SV-154 has not been described (AceView [http://www.ncbi.nlm.nih.gov/IEB/Research/Acembly/index.html?human]).

FIG. 2.

Alternative splicing of APOBEC3H transcripts derived from PBMC of healthy blood donors. (A) The four most frequently detected APOBEC3H splice variants are shown, each with its corresponding splice pattern. SV denotes a splice variant, with the number reflecting the length of the protein (e.g., SV-182 encodes 182 residues). SV-182 and SV-183 have previously served as references. The most pronounced changes were noted for SV-200 and SV-154 due to the skipping of exon 4 and/or the usage of the cryptic exon 4b. The schematic representation has been adapted from AceView (www.ncbi.nlm.nih.gov). (B) Alternative splicing introduced variation in the 3′ end of APOBEC3H. The four alternative proteins (SV-182, SV-183, SV-200, and SV-154) are depicted with their distinct carboxyl-terminal regions underlined. (C) The assortment of APOBEC3H splice variants within each donor is shown. Between 6 and 14 independent clones were analyzed for each donor (total, 106 clones). The majority of donors harbored at least three different APOBEC3H splice variants, with SV-182 and SV-183 being detected in every donor.

An assortment of two or three APOBEC3H transcripts was present in the PBMC of all donors. SV-183 and SV-182 were detected in all donors, in contrast to some of the alternative splice forms (e.g., SV-200 in 6 of 12 donors and SV-154 in 3 of 12 donors) (Fig. 2C). In summary, alternative splicing of APOBEC3H transcripts occurred frequently in vivo and resulted in proteins with variable C-terminal regions.

The antiviral activity of APOBEC3H hapII-RDD is superior to that of hapI-GKE.

We next investigated the impact of genomic sequence variation on APOBEC3H function. We used two experimental approaches which differed in the manner by which HIV-1 Vif was provided. In approach 1, HIV-1 WT or the Vif mutant (SLQ144AAA) was provided in cis by full-length molecular clone NL4-3. In approach 2, HIV-1 Vif was supplied in trans, allowing for Vif complementation independently of the HIV-1 genome. In both systems, the infectivity of viral particles generated in the presence of human APOBEC3H variants was assessed on the TZM-bl reporter cell line.

(i) APOBEC3H activity using full-length HIV-1.

The infectivities of HIV-1 WT and mutant (SLQ144AAA) Vif viruses produced in the presence of the APOBEC3H variants were compared to the infectivities of viruses made in the presence of APOBEC3G or in the absence of any APOBEC3 (Fig. 3A). The hapI-GKE SV-183 construct inhibited HIV-1 (WT Vif) infectivity by more than 20-fold (Fig. 3A), which was comparable to the decrease obtained with APOBEC3G (hapI-GKE [NL4-3], 3.8% ± 2.9 [mean ± standard deviation], versus APOBEC3G [NL4-3], 9.3% ± 0.4; P = not significant, paired t test). hapII-RDD SV-183 showed a significantly higher antiretroviral activity than APOBEC3G (SV-183 of hapII-RDD [NL4-3], 1.5% ± 0.5, versus APOBEC3G [NL4-3], 9.3% ± 0.4; P = 0.006, paired t test).

FIG. 3.

Antiretroviral activity profiles of naturally occurring APOBEC3H variants using full-length HIV-1. (A) HIV-1 wild-type (NL4-3; 500 ng) and Vif mutant SLQ144AAA (NL4-3ΔSLQ; 500 ng) viral stocks were produced by transfection in the presence of six different APOBEC3H variants (250 ng), APOBEC3G (250 ng), or empty pTR600 plasmid (250 ng). Viral infectivity was assessed by infection of TZM-bl reporter cells. Results were normalized using the no-APOBEC3 controls as a reference and plotted as the percent relative infectivity. hapI-GKE and hapII-RDD refer to the APOBEC3H haplotype. Results represent the mean ± SD of TZM-bl infections performed in triplicates from at least two independent transfection experiments. A3H stands for APOBEC3H. (B) APOBEC3H expression levels in the absence and presence of NL4-3 HIV-1 were assessed by Western blotting of transfected HEK 293T cell lysates. APOBEC3H variants are FLAG tagged at the amino terminus. The predicted molecular mass for the normal splice form of APOBEC3H is 23.5 kDa. Detection of GAPDH served as protein loading control.

The restriction exerted by hapI-GKE and hapII-RDD APOBEC3H variants was comparable for HIV with and without functional Vif, which stands in contrast to the Vif-mediated rescue of viral infectivity observed for viruses produced in the presence of APOBEC3G (for example, compare NL4-3 WT with NL4-3 Vif mutant SLQ144AAA in Fig. 3A).

A 100-fold reduction of NL4-3 Vif mutant SLQ144AAA infectivity was observed for hapII-RDD, a level of restriction that was comparable to the one induced by APOBEC3G (SV-183 of hapII-RDD [NL4-3 SLQ144AAA], 1.07% ± 0.1, versus APOBEC3G [NL4-3 SLQ144AAA], 0.36% ± 0.6; P = 0.06, paired t test). In this system hapI-GKE and hapII-RDD APOBEC3H both function as HIV-1 Vif-resistant antiviral restriction factors, with hapII-RDD being more active.

Western blot assays of cell lysates revealed that hapI-GKE and hapII-RDD variants differed in the level of protein expression. Only hapII-RDD proteins were well expressed upon transfection (Fig. 3B, right side). However, when cotransfected with HIV-1 NL4-3, the accumulation of all APOBEC3H proteins, including hapI-GKE variants, was greatly enhanced (Fig. 3B, right side). These findings indicate that mutations within APOBEC3H (e.g., the G105R/K121D/E178D cluster) as well as HIV-1 itself can, independently, increase and/or stabilize human APOBEC3H expression.

(ii) APOBEC3H activity determined using an HIV vector system.

To confirm the HIV-1 Vif-independent nature of the APOBEC3H restriction, we generated HIV-1 vector-derived VSV-G-pseudotyped viral particles with and without Vif proteins in the presence of APOBEC3H variants and measured their infectivity (Fig. 4). Under these experimental conditions, viral infectivity was reduced by twofold in the presence of hapI-GKE and 5- to 20-fold by hapII-RDD (SV-183 and SV-200). This restriction was completely independent of the presence of HIV-1 Vif, which stands in good agreement with the findings observed with the full-length Vif SLQ144AAA mutant virus. The activity of APOBEC3G was comparable between the two systems: APOBEC3G action was HIV-1 Vif sensitive as shown by the >100-fold reduction of infectivity, which was rescued by addition of HIV-1 Vif (e.g., 10% of the level observed in the absence of APOBEC3; also, compare controls in Fig. 3A and 4A).

FIG. 4.

Antiretroviral activity profiles of naturally occurring APOBEC3H variants, as determined using HIV-1 vectors. (A) VSV-G-pseudotyped HIV-1 viral vectors were produced in the presence of APOBEC3H variants and APOBEC3G with and without HIV-1 Vif expression plasmids. Viral infectivity was quantified by TZM-bl reporter cell infections. Results were normalized using the no-APOBEC3 controls as a reference and plotted as the percent relative infectivity. Results represent the means ± SD of TZM-bl infections performed in triplicates from two independent transfection experiments. (B) Expression levels of APOBEC3H variants and APOBEC3G in the presence of viral vectors complemented with HIV-1 Vif or empty plasmid were assessed by Western blotting of transfected HEK 293T cell lysates.

Western blotting of the cell lysates revealed that the expression levels of hapII-RDD variants were comparable in the presence or absence of HIV-1 Vif, in contrast to APOBEC3G, which is readily degraded by HIV-1 Vif (Fig. 4B). HapI-GKE was, however, poorly expressed in the presence of viral vectors complemented with HIV-1 Vif or empty control plasmid (e.g., compare expression in Fig. 3B and 4B). These data suggest that the antiretroviral activity of APOBEC3H is completely independent of the presence of HIV-1 Vif, and not only in its ability to bind to the Elongin C component of the Cullin5 E3 ligase. Furthermore, it seems likely that either the absence of some of the nonstructural genes in the vector-derived viruses or a different ratio between genomic RNA and Gag-Pol in the producer cell can account for the different degree of APOBEC3H accumulation and antiviral activity seen with this approach.

Influence of alternative splicing on antiviral activity depends on genetic background of APOBEC3H.

Since alternative splice forms were frequently detected in PBMC, we tested next whether splice variants of hapI-GKE and hapII-RDD differed in their activities against HIV-1. We tested three isoforms (SV-182, SV-183, and SV-200) for hapI-GKE and hapII-RDD.

SV-182 and SV-200 in hapI-GKE were less active than SV-183 (SV-182, 14.02% ± 6.9 [P = 0.018]; SV-200, 19.46% ± 3.1 [P = 0.001, paired t test]) (Fig. 3A). In contrast, SV-182 and SV-200 in the hapII-RDD background displayed increased activity compared to SV-183 (e.g., 50- to 100-fold reductions of infectivity) (Fig. 3A and 5A). Of note, hapII-RDD/SV-200 reduces viral infectivity to the levels observed for APOBEC3G, with the notable difference that infectivity was not recued by a functional Vif (e.g., HIV-1 SLQ144AAA [Fig. 3A] and viral vectors with or without Vif [Fig. 4A]).

FIG. 5.

Alternative splicing impacts the antiviral activity of APOBEC3H. (A) Cotransfection of NL4-3 (500 ng) and serial dilutions (50, 100, and 250 ng) of the six APOBEC3H plasmids demonstrates the different impacts of alternative splicing on hapI-GKE (left panel) and hapII-RDD (right panel). Results represent the mean ± SD of TZM-bl infections performed in triplicates from a representative transfection experiment. (B) Western blot analysis of the lysates of cells used for the production of viruses in the presence of increasing APOBEC3H concentrations (50, 100, and 250 ng) as depicted in Fig. 3C. Anti-FLAG monoclonal antibody was used to probe for FLAG-APOBEC3H expression. Detection of GAPDH served as a protein loading control. (C) Transfection of serial dilutions (50 to 1,000 ng) of APOBEC3H plasmids demonstrated higher antiviral potency of SV-200 hapII-RDD compared to SV-183 and SV-200 of hapI-GKE. As a reference the relative activity of APOBEC3G against NL4-3 and NL4-3 Vif mutant SLQ144AAA is also shown in each of the plots. Error bars represent the SDs of TZM-bl infections performed in triplicates from a representative transfection experiment.

Serial dilutions of the different APOBEC3H splice variants confirmed the activity differences between wild-type and hapII-RDD splice variants (Fig. 5A). Protein expression of serially diluted APOBEC3H plasmids showed that hapII-RDD variants were expressed to higher levels than the hapI-GKE ones (Fig. 5B). Interestingly, the level of expression of hapI-GKE variants was comparable for all three splice forms despite clear differences in antiviral activity (compare Fig. 5A with B). Thus, although APOBEC3H protein expression levels might be relevant for anti-HIV-1 activity, other features (e.g., cellular localization) are likely to have an equally important role.

Figure 5C illustrates in more detail the impact of alternative splicing on antiviral activity relative to the allelic context. Extended serial dilutions of APOBEC3H expression plasmids (50 to 1,000 ng) revealed dramatically different activity profiles for hapI-GKE and hapII-RDD SV-200 (Fig. 5C, far right panels), with SV-200 of hapII-RDD achieving suppression levels comparable to those of APOBEC3G (Fig. 5C, far left panel). Taken together, these findings suggest that alternative splicing regulates the antiviral function of the two APOBEC3H haplotypes with opposite effects.

Requirements for activity in wild-type and mutant genomic APOBEC3H contexts.

To determine the amino acid substitutions within the SNP cluster (G105R/K121D/E178D) that are responsible for differences between hapI-GKE and hap-II-RDD, we constructed a panel of site-directed mutants containing the naturally occurring amino acid substitutions at positions 105, 121, and 178 in different combinations (all SV-183) (Fig. 6A). As in previous experiments, the infectivity of WT NL4-3 viruses produced in the presence of 100 ng of the APOBEC3H variants was assessed by infection of TZM-bl cells. We chose this intermediate concentration of APOBEC3H expression plasmid in order to have optimal discrimination in the lower range of the assay (see also the serial dilutions in Fig. 5C).

FIG. 6.

Identification of mutations essential for antiviral activity of APOBEC3H. (A) The impact of single mutations on antiretroviral activity was tested by cotransfecting the different APOBEC3H site-directed mutants (100 ng) with NL4-3 (500 ng). Viral infectivity was quantified by TZM-bl reporter cell infection. Results were normalized using the no-APOBEC3 controls as a reference and plotted as the percent relative infectivity. Results represent the means ± SDs of TZM-bl infections performed in triplicates from four independent transfection experiments. The cartoons on the right illustrate the panel of mutants, with the letters in the boxes symbolizing the amino acids at position 105, 121, and 178 in hapI-GKE (white boxes) and hapII-RDD (dark boxes) backgrounds. G, glycine; K, lysine; E, glutamic acid; R, arginine; D, aspartic acid. (B) Expression levels of APOBEC3H site-directed mutants were assessed by Western blotting of transfected HEK 293T cell lysates. The Fuji Intelligent light box LAS-300 system was used to quantify protein expression in a dynamic manner. The bar graph shows protein expression normalized to the signal captured for hapI-GKE (first lane). Anti-FLAG monoclonal antibody was used to probe for FLAG-APOBEC3H expression. Detection of GAPDH served as a protein loading control.

Residues in two distinct positions (105R and 121K) proved to be relevant for the antiretroviral activity of hapI and hapII APOBEC3H (Fig. 6A). Introduction of 105R into hapI-GKE resulted in a protein that was 10-fold more active than the original protein (NL4-3 infectivity in the presence of hapI-RKE, 1.2% ± 1.3%, versus hapI-GKE, 22.0% ± 7.2%). Conversely, replacement of arginine at position 105 by glycine in hapII-RDD decreased antiviral activity approximately threefold (NL4-3 in the presence of hapII-RDD, 7.5% ± 4.1%, versus hapII-GDD, 27.1% ± 9.1%), yielding levels of infectious particles comparable to hapI-GKE (Fig. 6A).

In hapII, the reversion from aspartic acid to lysine at position 121 (hapII-RKD) resulted in a fourfold increase of antiretroviral activity compared to the naturally occurring hapII-RDD (NL4-3 infectivity in the presence of hapII-RKD, 1.9% ± 1.3%, versus hapII-RDD, 7.5% ± 4.1%). Substitutions at position 178 in either haplotype did not improve the potency (Fig. 6A, compare hapI-GKD and hapII-RDE to their corresponding parental proteins).

Since expression of hapI-GKE was far inferior to hapII-RDD, we speculated that mutations with 105R may result in protein stabilization, thereby leading to enhanced activity. In agreement with our previous findings (Fig. 3B and 5B), the expression of hapII-RDD was fivefold higher than hapI-GKE (Fig. 6B). However, expression of hapI-RKE, carrying the 105R substitution from hapII, was 10 times higher than that of the natural variant hapI-GKE. In contrast, hapII protein expression was destabilized by introduction of hapI residues at positions 105 and 178 (hapII-GDD and RDE).

Taken together these findings suggest that the activities of both haplotypes I and II are suboptimal and can be improved by specific substitutions from the other haplotype. The combination of 105R and 121K (hapI-RKE and hapII-RKD) (Fig. 6A and B) resulted in stably expressed proteins with, on average, 10-fold-higher antiviral activity.

APOBEC3H variants deaminate HIV-1.

Since the ability to act as a mutator of retroviral genomes is a hallmark of APOBEC3 proteins, we next investigated whether APOBEC3H variants introduce G-to-A changes into HIV-1 upon infection of target cells. TZM-bl reporter cells infected with NL4-3 viral stocks produced in the presence of APOBEC3H variants (50 ng; hapI-GKE and hapII-RDD; SV183 and SV-200), an active site mutant (100 ng; E56A; hapII-RDD), or APOBEC3G (50 ng). Genomic DNA of these infected cells was used to amplify, clone, and sequence the HIV-1 RT region.

Parallel measurement of the infectivity of each viral stock revealed that catalytic site mutant HapII-E56A failed to restrict HIV-1 (NL4-3 infectivity, 94.2% ± 1.8%) (Fig. 7A). Under these experimental conditions, APOBEC3G reduced NL4-3 infectivity by twofold (52% ± 5.01%), while infectivity rates in the presence of APOBEC3H ranged between 21.3% (SV-200 of hapII-RDD) and 90.4% (SV-200 of hapI-GKE) (Fig. 7A).

FIG. 7.

APOBEC3H variants introduce G-to-A mutations in HIV-1. (A) Infectivity of NL4-3 viral stocks produced in the presence of hapI-GKE and hapII-RDD variants (50 ng) was assessed in TZM-bl reporter cells. hapII-RDD E56A is a catalytic site mutant (SV-183). Results represent the means ± SDs of TZM-bl infections performed in triplicates from one representative experiment. (B) TZM-bl cells were infected with viral stocks produced in the presence of APOBEC3H variants, the catalytic site mutant E56A, and APOBEC3G. Genomic DNA was extracted and a portion of RT was amplified, cloned, and sequenced (6 to 14 clones for each infection). The number of mutations was calculated relative to the total number of sequenced nucleotides for each infection. (C) Infectivity and the relative frequency of G-to-A mutations correlate inversely. hapI-GKE/SV-200, hapII-RDD-E56A (deaminase site mutant), and pTR600/no A3 control cluster closely in the upper left corner. (D) APOBEC3 mutagenesis depends on the dinucleotide context (e.g., GG versus GA). The number of G-to-A mutations per clone was plotted for SV-183 and SV-200 of hapI-GKE and hapII-RDD. APOBEC3G favors a GG dinucleotide context, whereas APOBEC3H prefers a GA dinucleotide context.

We analyzed 6 to 11 individual RT clones for each infection (total, 59 clones, 35,400 nucleotides) (Fig. 7B and D) and calculated the overall frequency of mutations and the percentage of G-to-A mutations, as well as the favored dinucleotide context in which they occurred (e.g., GG versus GA). Overall, the APOBEC3H variants introduced mostly G-to-A mutations (Fig. 7B). For example, SV-200 of hapII-GKE induced the most mutations (frequency of any mutation, 0.79%; frequency of only G-to-A mutations, 0.68%) (Fig. 7B) and restricted HIV-1 best (Fig. 7A). Similarly, SV-200 of hapI-GKE was the least active of the APOBEC3H variants and the frequency of mutations associated with this viral stock was very low. Indeed, with an overall mutation frequency of 0.09%, SV-200 of hapI-GKE was comparable to results with the virus alone (Fig. 7B, no-APOBEC3 control, 0.09%) and to the active site mutant E56A (Fig. 7B, 0.01%). Lastly, the frequency of G-to-A mutations in proviral sequences correlated inversely with the infectivity (Fig. 7C).

The majority of G-to-A mutations introduced by APOBEC3H occurred in a GA dinucleotide context, which contrasted with APOBEC3G and its clear favoring of a GG dinucleotide context (compare Fig. 7C and D). A similar preference has been reported for human APOBEC3F (20, 35) and rhesus macaque APOBEC3H (28).

DISCUSSION

Human APOBEC3H is evolutionarily distinct from the other six human APOBEC3 family members: it resembles the 3′ region of mouse APOBEC3 more closely than any APOBEC3 domain of human origin (7, 28). Interestingly, the mouse APOBEC3 is as active against HIV-1 as is the human APOBEC3G but, unlike APOBEC3G, it is fully resistant to HIV-1 Vif (22).

Human APOBEC3H is the least studied of the single-domain cytidine deaminases, which generally exert only modest anti-HIV activity (8, 12). We thought, therefore, to investigate whether sequence variation and/or splicing events may increase its antiviral activity. We started by analyzing the frequency of APOBEC3H SNP and splice variants in PBMC, a cell population known to express APOBEC3H (10, 19, 28). Here, we describe APOBEC3H to be polymorphic in sequence and subject to alternative splicing (Fig. 1 and 2).

We generated a panel of the most commonly detected haplotypes (hapI-GKE and hapII-RDD) and splice variants (SV-183, SV-182, and SV-200) and tested them for antiretroviral activity and expression (Fig. 3A and B). Four of the six APOBEC3H variants inhibited the infectivity of HIV-1 by 20- to 100-fold in a Vif-independent manner. Our findings indicate that all splice variants of hapII-RDD were well expressed and active against HIV-1 (Fig. 3A). Thus, hapII-GKE APOBEC3H variants are highly active HIV-1 Vif-resistant antiviral proteins which mimic mouse APOBEC3 anti-HIV-1 properties (4, 6).

The frequency, pattern, function, and relevance of alternative splicing of most human APOBEC3 enzymes remains unknown, but it is tempting to speculate that alternative splicing of cryptic exons could provide functional diversity and/or control. Human APOBEC3B has been reported to have two splice variants, both of them expressed in human liver, but the shorter form lacks activity against hepatitis B virus (5). In mice, two APOBEC3 splice variants display similar activities against HIV-1 (6), and in cats alternative readthrough splicing generates APOBEC3CH (A3C-H fusion protein) (25).

In this study, we found that splice variants modulate the antiviral activity of APOBEC3H. Splicing events that lead to the replacement of the carboxy-terminal region of the protein were frequent (Fig. 2B), and the majority of donors harbored combinations of two or three different variants. By using PBMC to amplify APOBEC3H transcripts, our current data do not discriminate which cell populations express the different splice variants. It is conceivable that cell-type-specific alternative splicing events lead to the accumulation of certain variants which, depending on the genotype (hapI-GKE versus hapII-RDD), could be highly active or largely defective (Fig. 5C, compare wild-type SV-200 with SNP SV-200) with respect to their ability to inhibit HIV-1 infectivity.

APOBEC3 enzymes restrict HIV-1 through editing and nonediting mechanisms (reviewed in references 8, 12, and 17). The degree of mutagenesis observed was in excellent agreement with the reduction of viral infectivity observed for the specific APOBEC3H variants. Although this is only a correlation, the catalytic mutant provides compelling evidence of the causal relationship between deamination and viral restriction (Fig. 7B). These findings for human APOBEC3H variants resemble those relative to rhesus APOBEC3H variants, which are catalytically active and display a strong preference for a GA dinucleotide context (28).

Studies of natural history cohorts have reported associations between nonsynonymous SNPs in APOBEC3G as well as in Cullin-5 proteins (1, 2). Individuals differ in their susceptibility to infection and time to AIDS disease progression, and future studies will establish whether individuals with these SNPs in the APOBEC3H gene are more resistant to HIV-1/AIDS disease.

Note added during revision.

Reports by OhAinle et al. (27) and Tan et al. (31) were published as the manuscript was under revision. We attempted to integrate the different nomenclatures to facilitate understanding. Haplotype I represents the wild-type reference sequence and is referred to as hapI-GKE. Haplotype II represents APOBEC3H alleles containing a cluster of three SNPs (G105R/K121D/E178D) and is named hapII-RDD. GKE or RDD stand for the amino acids found at positions 105, 121, and 178 of a given APOBEC3H variant.