HIV-1 Replication and APOBEC3 Antiviral Activity Are Not Regulated by P Bodies

Abstract

The APOBEC3 cytidine deaminases play a critical role in host-mediated defense against exogenous viruses, most notably, human immunodeficiency virus type-1 (HIV-1) and endogenous transposable elements. APOBEC3G and APOBEC3F interact with numerous proteins that regulate cellular RNA metabolism, including components of the RNA-induced silencing complex (RISC), and colocalize with a subset of these proteins to mRNA processing bodies (P bodies), which are sites of mRNA translational repression and decay. We sought to determine the role of P bodies and associated proteins in HIV-1 replication and APOBEC3 antiviral activity. While we established a positive correlation between APOBEC3 protein incorporation into virions and localization to P bodies, depletion of the P-body components DDX6 or Lsm1 did not affect HIV-1 replication, APOBEC3 packaging into virions or APOBEC3 protein mediated inhibition of HIV-1 infectivity. In addition, neither HIV-1 genomic RNA nor Gag colocalized with P-body proteins. However, simultaneous depletion of multiple Argonaute family members, the effector proteins of RISC, could modestly increase viral infectivity. Because some APOBEC3 proteins interact with several Argonaute proteins, we also tested whether they could modulate microRNA (miRNA) activity. We found no evidence for the specific regulation of miRNA function by the APOBEC3 proteins, though more general effects on transfected gene expression were observed. In sum, our results indicate that P bodies and certain associated proteins do not regulate HIV-1 replication or APOBEC3 protein antiviral activity. Localization to P bodies may therefore provide a means of sequestering APOBEC3 enzymatic activity away from cellular DNA or may be linked to as yet unidentified cellular functions.

INTRODUCTION

The APOBEC3 (apolipoprotein B mRNA editing enzyme catalytic polypeptide-like 3) family of cytidine deaminases provides an innate mechanism of antiviral defense against a diverse range of exogenous viruses and endogenous retroelements (20, 67). The human APOBEC3 family consists of seven proteins; APOBEC3A (A3A), APOBEC3B (A3B), APOBEC3C (A3C), APOBEC3D/E (A3D/E), APOBEC3F (A3F), APOBEC3G (A3G), and APOBEC3H (A3H) (53). At least four members (APOBEC3D/E, APOBEC3F, APOBEC3G, and APOBEC3H) have anti-HIV-1 activity and are counteracted by the viral Vif protein (11, 28, 48, 81, 90, 99). In the absence of Vif, antiviral APOBEC3 proteins are incorporated into assembling virions and, following infection of the target cell, mediate deamination of cytidine residues to uridines in (mostly) nascent minus-strand reverse transcripts. These are detected as G-to-A mutations in the plus-strand viral DNA, and excessive editing, known as hypermutation, leads to loss of sequence integrity and the production of genetically compromised virions (40, 68, 69, 106). Editing-independent effects also contribute to HIV-1 inhibition, as A3F and A3G can impede reverse transcription in target cells (10, 12, 42, 43, 51). Vif prevents these proteins from being packaged by recruiting them to a cullin5-elonginB/C-Rbx2-CBFβ E3 ubiquitin ligase complex, resulting in their polyubiquitination and subsequent proteasomal degradation (24, 52, 69, 72, 104, 107).

A3F and A3G interact, mostly via RNA bridging, with a large number of RNA binding proteins that regulate mRNA metabolism, translation, and degradation and localize to discrete, nonmembraned structures termed processing bodies (P bodies) (21, 36, 37, 57, 98). These foci are concentrated sites of translationally repressed mRNAs and mRNA decay machinery, including the Dcp1a/Dcp2 decapping complex, the decapping coactivators DDX6 and Lsm1, and the 5′-3′ exoribonuclease Xrn1 (25, 83, 91). The Argonaute proteins also localize to P bodies and interact with A3F and A3G in a partially RNase-insensitive manner, suggestive of close and potentially direct interactions (36, 37). These proteins play a fundamental role in the process of RNA interference (RNAi), as they are the effector components of the RNA-induced silencing complex (RISC), which mediates translational repression and mRNA decay via microRNAs (miRNAs) and small interfering RNAs (siRNAs) (32, 50, 94).

There is increasing evidence to suggest that P-body and RISC proteins can influence diverse viral life cycles and may act as cofactors for the replication of certain viruses (7). Hepatitis C virus requires the miR-122 miRNA as well as the DDX6, Lsm1, and PatL1 proteins for viral protein expression and replication (54, 89). DDX6 also facilitates infectious virion production of the retrovirus primate foamy virus (PFV) (103). In the yeast Saccharomyces cerevisiae, virus-like particle formation and retrotransposition of the Ty1 and Ty3 retrotransposons are dependent upon the P-body components Dhh1p/DDX6, Lsm1p, and PatL1p (8, 18, 31). Similarly, several P-body proteins promote the translation and replication of the plant RNA virus brome mosaic virus (BMV) (6, 29, 71, 80). Conversely, P-body components have also been shown to be inhibitory to some viruses and retroelements. Poliovirus infection causes disruption of P bodies and the specific degradation of Dcp1a and Xrn1 (30), and replication of PFV is restricted by the cellular miRNA miR-32 (59). In mice, inhibition of miRNA biogenesis, accomplished through knockdown of Dicer, increases the transcript abundance of intracisternal A particle (IAP) and murine endogenous retrovirus-L (MuERV-L) retrotransposons (93), and depletion of the P-body components DDX6 and eIF4E-T also increases IAP retrotransposition events (65).

A number of reports have indicated that HIV-1 has substantial interactions with miRNA and P-body-associated proteins. HIV-1 infection has been shown to modulate cellular miRNA production (41, 44, 101), and both cellular miRNAs targeting HIV-1 and putative virally derived miRNAs have been identified by some groups (2, 47, 56, 73, 82, 95), but not others (61, 84). Two groups have reported that siRNA-mediated knockdown of RISC and P-body-associated proteins results in increased virus production and/or infectivity, with both studies also suggesting that HIV-1 RNA localizes to P bodies (17, 73). In contrast, more recent reports have found that Argonaute 2 (Ago2) or DDX6 depletion decreases HIV-1 production by 50 to 80%, (13, 86), highlighting the controversial role that P-body proteins play in HIV-1 replication.

Why certain APOBEC3 proteins localize to and interact with P-body and RISC components remains unresolved. It is plausible that these complexes or specific associated proteins may promote APOBEC3 protein packaging into virions, thereby acting as cofactors for APOBEC3-mediated HIV-1 restriction. Recently, APOBEC3 antiviral activity has been positively correlated with virion incorporation (48), but it remains unclear as to exactly what cellular or viral factors are required for packaging. A3G virion incorporation is dependent upon its interaction with the nucleocapsid region of Gag (16, 74, 88), but the identity of the RNA(s) that bridges this interaction remains controversial (55). It has been proposed, for instance, that APOBEC3 proteins are packaged in association with polymerase III (Pol III) RNAs, which are themselves incorporated into virions (97). However, we previously reported that A3F does not detectably interact with these RNAs, even though it is efficiently incorporated into HIV-1 particles (36, 49). Alternatively, APOBEC3 localization to P bodies may be indicative of cellular functions that are distinct from viral inhibition, potentially concerning the control of cellular mRNA expression. In particular, the closely related family member APOBEC1 is able to bind to and stabilize several AU-rich element (ARE)-containing mRNAs (3, 4, 19), and the APOBEC3 proteins themselves have been reported to inhibit miRNA-mediated translational repression (46). Very recently, this inhibition has been attributed to competition between A3G and Ago2 for binding to the RISC and P-body component MOV10 (62). Therefore, a subset of APOBEC3 proteins may localize to P bodies and interact with the miRNA machinery to regulate cellular mRNAs.

In this study, we aimed to determine the role of P bodies and associated proteins in APOBEC3 protein activity. We established a positive correlation between localization to P bodies and packaging into viral particles and hence sought to determine the influence of these foci on APOBEC3-mediated HIV-1 inhibition and virion incorporation as well as on HIV-1 replication in general. We also identified several APOBEC3 proteins, in addition to A3F and A3G, which interact with Ago2 and hence explored the functional significance of this interaction with regard to a role of the APOBEC3 proteins as regulators of cellular mRNA translation through miRNA inhibition.

MATERIALS AND METHODS

Expression vectors and RNA silencing.

All plasmid manipulations were performed using standard PCR-based methods, and all constructs were confirmed to be correct by sequencing and restriction digest. The carboxyl-terminal hemagglutinin (HA)-tagged human APOBEC3 (A, B, C, F, and G) expression vectors have previously been described (11). cDNAs encoding APOBEC3D/E, APOBEC3H (haplotype II), or green fluorescent protein (GFP) were inserted into the same vector using the HindIII and XbaI restriction sites. cDNA encoding firefly luciferase (pLuc), with a stop codon at the end of the open reading frame to prevent expression of the HA tag, was also inserted into this vector using the Asp718 and HindIII restriction sites. A3G or yellow fluorescent protein (YFP) was cloned into the retroviral vector pNG/neo (90) by insertion using either the EcoRI and NotI or EcoRI and BglII sites to give pNG/neo/A3G and pNG/neo/YFP, respectively. Amino-terminal Myc-tagged DDX6 was constructed by inserting the DDX6 cDNA into the pcDNA3.1-myc expression vector (63) using the BamHI and XhoI restriction sites to generate pMyc-DDX6. The NL4-3 Δvif proviral plasmid (pHIV_NL4-3Δvif) was generated by introducing two stop codons into the vif reading frame of pHIV_NL4-3 (1) by overlapping PCR. The NL4-3 protease-deficient derivative (pHIV_NL4-3/Pr-) was a kind gift from E. Freed (45a). The NL4-3 proviral plasmid containing 24 binding loops from the MS2 bacteriophage (pHIV_{NL4-3/24×MS2}) was generated by insertion of the SacII and BsmB1 restriction sites at the end of the gag reading frame in the parental pNL4-3 expression plasmid by overlapping PCR to give pHIV_NL4-3/SB. The 24 MS2 binding loops (35) were then directly subcloned into pHIV_NL4-3/SB using the SacII and BsmB1 sites to give pHIV_{NL4-3/24×MS2}. Plasmids encoding 24 MS2 binding loops and MS2-YFP were kind gifts from R. Singer (35). The proviral plasmid construct that contained Gag fused to the Venus fluorescent protein (pHIV_{NL4-3/Gag-Venus}) was a kind gift from A. Ono (22). pcDNA3.1 plasmids encoding Myc-tagged Ago1, Ago2, and Ago3 were kindly provided by G. Hannon (63). The luciferase reporter constructs FF4LCS (wild type) and FF4rmLCS (mutant) were a kind gift from J. Steitz (66). The TK/Renilla plasmid (pGL4.74) was purchased from Promega. The vesicular stomatitis virus G (VSV-G) protein expression vector has previously been described (34). The short hairpin RNA (shRNA) lentiviral vectors were obtained from Open Biosystems (pGIPZ shRNAmir-GFP; Ago2/V2LHS_241069, DDX6/V2LHS_23998, Lsm1/V2LHS_213130). The Argonaute and DDX6 siRNA oligonucleotides were from Ambion Life Technologies (Silencer Select siRNA; Ago1/s25501, Ago2/s25932, Ago3/s46948, Ago4/s46949, DDX6/s4012). The ALIX siRNA oligonucleotide has previously been described (15).

Cell culture and stable cell lines.

293T, HeLa, and TZM-bl cells were cultured in Dulbecco modified Eagle medium, and HUT78 cells were cultured in RPMI supplemented with 10% fetal bovine serum plus penicillin-streptomycin. HeLa cells stably expressing A3G or YFP were produced by standard retrovirus-mediated transduction using pNG/neo/A3G or pNG/neo/YFP, respectively, and selection with G418 (0.5 mg/ml). Cell lines stably expressing either a nonsilencing control, Ago2, DDX6, or Lsm1 targeting shRNA were generated by standard lentivirus-mediated transduction and selection with puromycin (1 μg/ml). For shRNA- and DDX6 or ALIX siRNA-treated cells, protein expression and knockdown were confirmed by immunoblotting using either an anti-Ago2 (1/1,000; rat monoclonal 11A9; Sigma), anti-ALIX (1/10,000; rabbit polyclonal) (33), anti-DDX6 (1/3,000; rabbit polyclonal; Cambridge Biosciences), or anti-Lsm1 (1/500; chicken polyclonal; Sigma) antibody. For Argonaute siRNA-treated cells, RNA levels were determined by quantitative PCR (qPCR) analysis using TaqMan gene expression assays (Ago1/Hs01084653, Ago2/Hs01085579, Ago3/Hs01087121, Ago4/Hs01059731; Applied Biosystems).

Single-cycle infectivity assays.

Subconfluent monolayers of 293T cells seeded in 35-mm plates were cotransfected with 0.5 μg of pHIV_NL4-3Δvif and between 0.02 μg and 1.5 μg of pHA-APOBEC3A to APOBEC3H expression vectors using polyethylenimine (PEI). Forty-eight hours later, supernatants were harvested and virus levels were quantified by p24^Gag enzyme-linked immunosorbent assay (ELISA; Perkin-Elmer). Virus equivalent to 5 ng p24^Gag was used to challenge 10⁵ TZM-bl indicator cells. At approximately 30 h postinfection, whole-cell lysates were assayed for the induction of β-galactosidase expression using a Galacto-Star system (Applied Biosystems).

Virion incorporation assays.

Virus equivalent to 20 ng p24^Gag was purified through a 20% (wt/vol) sucrose cushion at 21,000 × g for 2 h at 4°C. Viral pellets were resuspended in loading buffer (60 mM Tris-HCl [pH 6.8], 0.1 M dl-dithiothreitol [DDT], 10% glycerol, 2% sodium dodecyl sulfate [SDS], 0.1% bromophenol blue) and analyzed by immunoblotting. Whole-cell lysates from the corresponding virus producer cells were analyzed in parallel. A3G virion incorporation was also assessed by immunoblot analysis using an anti-A3G antibody (1/1,000; rabbit polyclonal) (76).

Virus infections and replication.

Subconfluent monolayers of HeLa cells seeded in 35-mm cultures were infected with 293T cell-generated VSV-G-pseudotyped HIV_NL4-3 or HIV_NL4-3Δvif equivalent to 50 ng p24^Gag. Four hours later, input virus was removed and cells were thoroughly washed before the addition of fresh medium. At 48 h postinfection, supernatants were harvested and viral infectivity was measured by a single-cycle infectivity assay.

Approximately 3 × 10⁶ HUT78 cells were infected with HIV_NL4-3 or HIV_NL4-3Δvif equivalent to 5 ng, 25 ng, or 125 ng p24^Gag. Two hours later, input virus was removed and cells were thoroughly washed before the addition of fresh medium. Every 2 days postinfection, 5 ml of culture medium was collected (and replaced with fresh medium) and centrifuged at 500 × g for 5 min. The supernatant was harvested for quantification by p24^Gag ELISA, and cell pellets were resuspended in loading buffer for determination of protein expression by immunoblot analysis.

HIV-1 genome incorporation assays.

Subconfluent layers of 293T cells were cotransfected with equivalent amounts of pHIV_{NL4-3/24×MS2} or pHIV_NL4-3/Pr- and either the Gag-Pol plasmid p8.91 or the luciferase-encoding expression plasmid pLuc. Forty-eight hours later, viral supernatants were harvested and treated with 20 U/ml RQ1 DNase (Promega) in 10 mM MgCl₂ for 3 h at 37°C, prior to sucrose cushion purification. Viral pellets were subject to RNA extraction using an miRNeasy kit (Qiagen). cDNA was synthesized using a high-capacity cDNA archive kit (Applied Biosystems) with 50 ng of extracted RNA. qPCRs were performed as previously described (39), using primers oNS172f (5′-GGCCAGGGAATTTTCTTCAGA) and oNS173r (5′-TTGTCTCTTCCCCAAACCTGA) and probe oNS174p (5′–6-carboxyfluorescein [FAM]–ACCAGAGCCAACAGCCCCACCAGA–6-carboxytetramethylrhodamine [TAMRA]).

Coimmunoprecipitation.

Subconfluent monolayers of 293T cells, in 35-mm cultures, were cotransfected with 2 μg of HA-tagged APOBEC3 expression vectors (pHA-APOBEC3A to pHA-APOBEC3H and 2 μg of pMyc-Argonaute 1 to Argonaute 3 expression plasmids. Forty-eight hours later, cells were harvested in phosphate-buffered saline (PBS) and pelleted at 500 × g for 5 min. Cell pellets were resuspended in 0.05% HCHO and incubated at 37°C for 10 min. Cells were washed in PBS and resuspended in 0.25 M glycine for 5 min at room temperature. Cells were washed in PBS and resuspended in 1× BB lysis buffer (0.5% NP-40, 150 mM KCl, 10 mM HEPES [pH 7.0], 6 mM MgCl₂, 2 mM DDT, 10% glycerol, and complete protease inhibitor cocktail [Roche]) for 10 min on ice. Lysates were sonicated (3 times for 10 s) before being cleared by centrifugation for 10 min at 1,000 × g. Protein G agarose (Invitrogen) was preincubated with an anti-Myc (1/1,000; mouse monoclonal 9E10) antibody for 1 h 30 min at 4°C and then mixed with lysate for 2 h at 4°C on a rotational tumbler. Samples were washed 5 times in 1× radioimmunoprecipitation assay buffer (10 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% Triton X-100, 1% sodium deoxycholate) and resuspended in reverse cross-linking buffer (10 mM EDTA, 5 mM DDT, 0.5% SDS) for 45 min at 65°C before eluting in loading buffer. Samples were analyzed by SDS-PAGE using 10% gels for APOBEC3 proteins and 7% gels for Argonaute proteins. Resolved proteins were detected by immunoblotting using anti-HA (1/1,000; rabbit polyclonal; Rocklands) or anti-Myc (1/1,000) antibodies. Bound antibodies were visualized either by appropriate horseradish peroxidase-conjugated secondary antibodies (1/3,000; Pierce) and enhanced chemiluminescence or by infrared IRDye-conjugated secondary antibodies (1/5,000; LI-COR Biosciences) and LI-COR infrared imaging technology (LI-COR UK Ltd.).

Immunofluorescence.

HeLa or 293T cell monolayers (10⁵ cells) were directly seeded onto 13-mm glass coverslips. For nonadherent HUT78 cells, coverslips were first treated with poly-l-lysine (Sigma) as per the manufacturer's instructions and dried overnight. Twenty-four hours later, cells were washed three times in PBS and fixed with 4% paraformaldehyde (Electron Microscopy Sciences) for 15 min. Cells were washed again, permeabilized with 0.2% Triton X-100 for 12 min at room temperature, and then simultaneously blocked and quenched with NGB buffer (50 mM NH₄Cl, 1% goat serum, 1% bovine serum albumin) for 1 h. For P-body detection, cells were first incubated with either an anti-DDX6 (1/500) or an anti-Ge1 (1/500; mouse monoclonal; Santa Cruz) antibody for 2 h. Cells were washed and incubated with the appropriate species-specific Alexa Fluor 488- or 594-conjugated secondary antibody (1/500; Molecular Probes) for 1 h. To visualize nuclei, cells were additionally incubated with 4′,6′-diamidino-2-phenylindole (DAPI) dilactate (1/25,000; Molecular Probes) for 1 min. Coverslips were mounted onto slides using Mowiol mounting medium (Calbiochem). Laser scanning confocal imaging was performed on a DM IRE2 microscope (Leica), and the images were processed using LCS (Leica), Openlab (Improvision), and Photoshop (Adobe) software packages.

For APOBEC3 and P-body localization studies, 10⁵ HeLa cell monolayers seeded on 13-mm coverslips were cotransfected with 0.2 μg of pHA-APOBEC3(A-H) expression plasmids and 0.2 μg of pMyc-DDX6 or pMyc-Ago2 expression plasmids using Fugene 6 (Roche). At approximately 30 h posttransfection, cells were fixed and stained with anti-HA (1/500) and anti-Myc (1/500) primary antibodies and appropriate species-specific Alexa Fluor 488- or 594-conjugated secondary antibodies and DAPI. Coverslips were treated and analyzed by confocal microscopy, as described above.

For HIV-1 genomic RNA (gRNA) and P-body localization studies, 0.5 μg of the MS2-YFP expression vector was transfected alone or cotransfected with between 0.018 and 0.5 μg of the pHIV_{NL4-3/24×MS2} proviral expression plasmid. For pHIV_{NL4-3/24×MS2}, expression of Pol is disrupted, and thus, some experiments were also performed with complementation using wild-type virus. For analysis of HIV-1 Gag and P-body localization, 0.5 μg of the pHIV_{NL4-3/Gag-Venus} proviral plasmid was transfected into HeLa cells. pLuc was also included to keep total DNA concentrations equivalent for each transfection. At approximately 30 h posttransfection, cells were fixed and stained with an anti-GFP antibody (1/1,000; mouse monoclonal; Roche) to enhance the YFP signal (for MS2-YFP-containing samples) and either anti-DDX6 or anti-Ge1 antibodies for detection of P bodies and then appropriate species-specific Alexa Fluor 594-conjugated secondary antibodies and DAPI. Coverslips were treated and analyzed by confocal microscopy, as described above.

Luciferase reporter assays.

HeLa cells plated at 10⁵ cells in 24-well plates were cotransfected with 0.1 μg of firefly luciferase expression plasmid, 0.08 μg of Renilla luciferase expression plasmid, and 0.9 μg of pHA-APOBEC3(A-H) expression plasmids using Lipofectamine 2000 (Invitrogen). At 30 h posttransfection, cells were harvested and assayed using a dual-luciferase assay kit (Promega) as per the manufacturer's instructions.

RESULTS

Antiviral activities of the APOBEC3 proteins.

The seven members of the APOBEC3 subfamily have various levels of antiviral activity against different viral substrates (20, 26, 67). We directly compared the anti-HIV-1 phenotypes of all the APOBEC3 proteins in a single-cycle infectivity assay, using Δvif viruses produced with the pHIV_NL4-3Δvif provirus in the presence of HA-tagged APOBEC3 expression vectors [pHA-APOBEC3(A-H)]. Consistent with previous reports (11, 48, 60, 99, 105), A3G was the most potent member of this family, even at very low concentrations (see Fig. S1 in the supplemental material). The other APOBEC3 proteins showed a range of inhibitory phenotypes, from relatively strong (A3B, A3F, and A3H) to weak (A3A, A3C, and A3D/E) antiviral activity.

APOBEC3 virion incorporation correlates with P-body localization.

Although not all of the APOBEC3 proteins are antiviral (see Fig. S1 in the supplemental material), they are all reportedly packaged into virions when ectopically expressed (27, 28, 38, 99, 102). In order to establish which APOBEC3 proteins can be detected in virions, 293T cells were cotransfected with pHA-APOBEC3(A-H) and pHIV_NL4-3Δvif expression plasmids. Forty-eight hours later, supernatants were harvested and virions were purified through a 20% sucrose cushion (see Fig. S2A in the supplemental material). At a 1:3 ratio of APOBEC3 plasmid DNA to proviral plasmid DNA, A3G was clearly packaged with the highest efficiency, followed by A3F, A3H, and A3D/E.

To determine if APOBEC3 localization to P bodies (37, 77, 98) correlated with virion incorporation, HeLa cells were cotransfected with pHA-APOBEC3(A-H) and an expression vector for Myc-tagged DDX6 or Ago2, cellular proteins known to accumulate in P bodies. These cells were then analyzed by indirect immunofluorescence, and colocalization of APOBEC3 proteins with DDX6-marked (see Fig. S2B and C in the supplemental material) or Ago2-marked (see Fig. S3 in the supplemental material) foci was scored for 100 cells in each sample. A3DE, A3F, A3G, and A3H were found to localize with DDX6- and Ago2-marked P bodies in ∼90% of cells. Thus, there appears to be a positive correlation between the localization of APOBEC3 proteins at P bodies and their incorporation into viral particles. It is important to note, however, that the proteins that most strongly associate with P bodies are also the ones that are predominantly cytoplasmic.

Role of P bodies in APOBEC3 antiviral activity and HIV-1 replication.

As APOBEC3 protein localization to P bodies positively correlated with virion incorporation (see Fig. S2 and S3 in the supplemental material), we sought to investigate whether microscopically visible P bodies are necessary for infectious HIV-1 particle production and APOBEC3 antiviral activity. We first depleted P bodies from HeLa cells that stably express either A3G or YFP by transfecting an siRNA targeting the DDX6 mRNA. This led to the almost complete loss of P bodies from these cells, as observed by confocal microscopy (Fig. 1A and B) and a consistent reduction of DDX6 protein levels of between 80 and 90% compared to that for a cell line transfected with a nonsilencing control siRNA (Fig. 1E). Importantly, expression of another P-body protein, Lsm1, was unaffected by knockdown of DDX6, confirming the specificity of the silencing (data not shown).

Fig 1.

Knockdown of DDX6 does not affect HIV-1 infectious virion production or A3G antiviral activity. (A) HeLa cells stably expressing either A3G or YFP were twice transfected with siRNAs targeting the DDX6 mRNA (DDX6) or a nonsilencing control (NSC). Cells were plated onto coverslips and 24 h later were fixed, permeabilized, and stained with a mouse anti-Ge1 primary antibody and an anti-mouse Alexa Fluor 594-conjugated secondary antibody. Coverslips were mounted onto slides, dried overnight, and imaged using a Leica DM IRE2 confocal microscope. Shown are representative images of immunofluorescence analyses to assess P-body depletion of cells. Bars = 10 μm. (B) Cells were also quantified for the number of cells containing Ge1-marked foci (top) and the average number of these foci per cell (bottom). n = 100. (C) Cells, as described for panel A, were infected with VSV-G-pseudotyped HIV_NL4-3 (wild type [wt]) or HIV_NL4-3Δvif (Δvif) equivalent to 50 ng p24^Gag. Forty-eight hours later, supernatants were harvested and quantified by p24^Gag ELISA. (D) The infectivity of the virions produced for panel C was assessed by infecting TZM-bl reporter cells with virus equivalent to 5 ng p24^Gag. For panels C and D, values are normalized to the value for the control siRNA transfected in the YFP-expressing cell line infected with wild-type virus, which is set at 100%. Results are presented as the average of three independent experiments, with error bars denoting the standard deviation. (E) A3G virion incorporation was assessed by purifying virus equivalent to 20 ng p24^Gag through a 20% sucrose cushion. Purified virions were lysed and subjected to immunoblotting (right). Lysates from the corresponding producer cells were analyzed in parallel (left).

All cell lines were then infected with wild-type or Δvif HIV-1, corresponding to 50 ng p24^Gag, and 48 h later the abundance and infectivity of the virions in the media were analyzed by p24^Gag ELISA and a single-cycle infectivity assay, respectively. Contrary to published findings (17, 73), we were unable to detect any increase in virus production (Fig. 1C) or HIV-1 infectivity (Fig. 1D) upon knockdown of DDX6 and depletion of visible P bodies. Further, DDX6 depletion did not affect the antiviral activity of A3G (Fig. 1D). Consistent with these findings, immunoblot analysis of purified virions demonstrated that there was no detectable difference in A3G packaging between the control and DDX6-knockdown cell lines (Fig. 1E). Equivalent results were obtained when HeLa cells were stably transduced with an shRNA-encoding lentiviral vector targeting DDX6 and a similar analysis was undertaken (data not shown). To verify these results further, we also depleted the P-body protein Lsm1 using lentiviral shRNA vectors. Analogous to knockdown of DDX6, depletion of Lsm1 led to a significant loss of visible P bodies and Lsm1 protein (∼85%) (Fig. 2A, B, and E), but when the knockdown cell lines were infected with wild-type or Δvif HIV-1, similar amounts of virus were produced, with no alteration in infectivity (Fig. 2C and D).

Fig 2.

Knockdown of Lsm1 does not affect HIV-1 infectious virion production or A3G antiviral activity. (A) HeLa cells stably expressing either A3G or YFP were transduced with shRNA-encoding vectors targeting the Lsm1 mRNA (Lsm1) or a nonsilencing control (NSC). P-body knockdown was assessed by confocal microscopy as described for Fig. 1A. (B) The number of cells containing Ge1-marked P bodies and the average number of these bodies per cell were quantified as described for Fig. 1B. (C) Cells were infected with wild-type or Δvif virus and analyzed as described for Fig. 1C. (D) The infectivity of the virions produced for panel C was assessed by infecting TZM-bl reporter cells as described for Fig. 1D. (E) A3G virion incorporation was assessed as described for Fig. 1E.

To determine if an effect of P-body depletion could be discerned over multiple rounds of viral replication, we then stably transduced HUT78 cells, a CD4⁺ T cell line that naturally expresses anti-HIV-1 APOBEC3 proteins, including A3G, with either a nonsilencing control shRNA or an shRNA targeting Lsm1 mRNA. Transduction of HUT78 cells with the Lsm1 shRNA led to a substantial reduction in protein levels (∼80%) and P bodies compared to control cells (Fig. 3A and B). Cells were then infected with wild-type or Δvif virus, and spreading replication was monitored over several days (Fig. 3C). Knockdown of Lsm1 had no effect on HIV-1 replication, as there was no detectable difference between the two cell lines in terms of wild-type virus production. Furthermore, APOBEC3 protein-mediated antiviral activity was unaffected by loss of Lsm1, as observed by equivalent levels of suppression of the Δvif virus in relation to the wild-type virus for both the control and knockdown cell lines. Knockdown of Lsm1 was maintained throughout the replication cycle (see Fig. S4A in the supplemental material), and similar results were observed, with virus inputs corresponding to 5 ng or 125 ng p24^Gag (see Fig. S4B and C in the supplemental material). Therefore, both in single-cycle assays and in multiple rounds of replication, we found no effect of depletion of P-body proteins on HIV-1 virion production, infectivity, or A3G antiviral function.

Fig 3.

Knockdown of Lsm1 does not affect HIV-1 replication or A3G antiviral activity. (A) Immunoblot analysis of Lsm1 protein levels in HUT78 cells stably transduced with a nonsilencing control (NSC) or Lsm1-targeting (Lsm1) shRNA lentiviral vector. HSP90 is included as a loading control. (B) HUT78 cells, as described for panel A, were plated onto poly-l-lysine-coated coverslips, fixed, and stained with anti-Ge1 and anti-DDX6 primary antibodies and appropriate species-specific Alexa Fluor 594-conjugated secondary antibodies. Samples were imaged by confocal microscopy. Bars = 10 μm. (C) Cells, as described for panel A, were infected with HIV_NL4-3 (wt) or HIV_NL4-3Δvif (Δvif) equivalent to 25 ng p24^Gag. Every 2 days postinfection, supernatants were harvested and analyzed by p24^Gag ELISA. A representative result from one of two independent experiments is shown.

Two groups have recently reported that knockdown of several P-body-associated proteins results in an increase in HIV-1 production and infectivity, utilizing either infection- or transfection-based experimental approaches (17, 73). In order to reconcile our results with published reports, we next tested whether depletion of DDX6 or Lsm1 affected virion production or infectivity when HeLa cells were transiently transfected with a proviral expression plasmid instead of being directly challenged with infectious virus. HeLa cells stably transduced with nonsilencing or DDX6- or Lsm1-targeting lentiviral shRNAs were cotransfected with pHA-APOBEC3(A-H) or pHA-GFP and pHIV_NL4-3Δvif expression plasmids at a 1:1 ratio. Forty-eight hours later, viral supernatants were harvested, quantified by p24^Gag ELISA, and used to infect the TZM-bl reporter cell line. Transfection of DDX6-depleted HeLa cells consistently resulted in a modest 2- to 4-fold increase in HIV-1 production (see Fig. S5C in the supplemental material). However, depletion of Lsm1 had no effect on virus production (see Fig. S6C in the supplemental material), even though knockdown of both proteins resulted in similar extents of P-body loss (see Fig. S5A and B and S6A and B in the supplemental material). In both cases, P-body depletion had no effect on HIV-1 infectivity (with all assays normalized for viral input), APOBEC3 protein antiviral activity, or virion incorporation (see Fig. S5D and E and S6D in the supplemental material), even when the amount of A3G-encoding plasmid DNA was reduced from 1 μg to 0.1 μg (see Fig. S7 in the supplemental material). Based on these observations, we propose that variations in experimental configuration can account for the different conclusions drawn regarding the role of P-body components in HIV-1 production. However, we consider the infection-based methods used for Fig. 1 to 3 to be more reflective of natural HIV-1 replication.

HIV-1 gRNA and Gag do not colocalize with P-body proteins.

It has been reported that HIV-1 RNA localizes to P bodies (17, 73). Since we were unable to detect any effect of P-body depletion on HIV-1 replication, we asked whether we could detect viral structural components in P bodies. We constructed an HIV_NL4-3 provirus containing 24 MS2 binding loops (pHIV_{NL4-3/24×MS2}) and cotransfected it into HeLa cells with pMS2-YFP (Fig. 4A). Importantly, the modified viral RNA is efficiently packaged into virus particles and is able to generate infectious virions comparable to those for an unmodified wild-type control (see Fig. S8 in the supplemental material). MS2-YFP was nuclear in the absence of the HIV-1 proviral construct (35) (Fig. 4A1) but showed a diffuse cytoplasmic distribution in its presence. Unlike a previous report (17), we were unable to observe any localization of gRNA with DDX6-positive P bodies (Fig. 4A2), even when pHIV_{NL4-3/24×MS2} was titrated down to lower expression levels (see Fig. S9 in the supplemental material). We then determined whether the viral structural protein Gag localized to these foci, which may be anticipated if these are important subcellular packaging domains. To do this, we used a second proviral construct that contained Gag fused to the Venus fluorescent protein (pHIV_{NL4-3/Gag-Venus}) (22). There was no apparent colocalization of Gag with P bodies, using either Ge1 or DDX6 as a P-body marker (Fig. 4B). Therefore, we were unable to detect any accumulation of either gRNA or Gag at P bodies.

Fig 4.

HIV-1 gRNA and Gag do not colocalize with the P-body proteins DDX6 and Ge1. (A) Schematic diagram of the proviral construct containing 24 MS2 binding sites (pHIV_{NL4-3/24×MS2}) used to visualize HIV-1 genomic RNA. A DNA fragment encoding 24 binding loops from the MS2 bacteriophage RNA was introduced 3′ to the terminus of the gag reading frame in an NL4-3 proviral plasmid. Coexpression of an MS2-YFP fusion protein, which binds with high affinity to the MS2 binding loops, allows visualization of the viral RNA. HeLa cells were transfected with either MS2-YFP alone (panel 1) or in combination with pHIV_{NL4-3/24×MS2} (panel 2). Cells were fixed at 24 h after transfection and stained with anti-GFP and anti-DDX6 primary antibodies and appropriate species-specific secondary antibodies conjugated to Alexa Fluor 488 or 594. LTR, long terminal repeat. (B) A second NL4-3 proviral construct containing the Venus fluorescent protein fused in frame with Gag (pHIV_{NL4-3/Gag-Venus}) was transfected into HeLa cells. Twenty-four hours later, the cells were fixed and stained with anti-DDX6 (panel 1) or anti-Ge1 (panel 2) primary antibodies and appropriate species-specific secondary antibodies conjugated to Alexa Fluor 594. For panels A and B, all samples were imaged by confocal microscopy and images are compilations of between 4 and 6 z stacks, with merged images presented on the right. Bars = 10 μm.

Role of the Argonaute proteins in HIV-1 infectious virion production and APOBEC3 antiviral activity.

There are conflicting reports concerning the effect of depletion of the RISC and P-body component Ago2 on HIV-1 virion production (13, 73). To address the role of the Argonaute protein family in regulating HIV-1 particle production and infectivity, we first depleted each Argonaute protein individually in HeLa cells using siRNAs, which led to a substantial reduction (at least 80%) of each corresponding RNA (Fig. 5A). We then infected these cells with wild-type HIV-1 corresponding to 50 ng p24^Gag and measured infectious virion production. Individual depletion of the Argonaute proteins did not significantly affect HIV-1 virion production or infectivity (Fig. 5B and C). As all of the human Argonaute proteins are able to mediate miRNA repression (64, 85, 100) and are thus functionally redundant in this respect, we then attempted to knock down all four proteins simultaneously. We transfected HeLa cells stably transduced with a nonsilencing or Ago4-targeting shRNA with either control siRNA or siRNAs targeting Ago proteins 1, 2, and 3. In the Ago4 shRNA-expressing cell line, Ago1 mRNA expression was reduced by 80%, Ago2 by 60%, Ago3 by 20%, and Ago4 by 50% (Fig. 5D). Importantly, knockdown of the Argonaute proteins either individually or in combination did not affect endogenous P-body formation (Fig. 5E and data not shown). When these cells were challenged with HIV-1, as before, there was a 2-fold decrease in subsequent virion production and a 4-fold increase in the infectivity of the virions produced (Fig. 5F and G). Of note, at this level of Argonaute depletion, there was no overt cell toxicity, but more efficient depletions of the total Argonaute mRNA pool resulted in substantial cell death (data not shown). Therefore, simultaneous depletion of multiple Argonaute proteins can modestly increase HIV-1 infectivity.

Fig 5.

Simultaneous knockdown of Argonaute family members increases HIV-1 infectivity. (A) HeLa cells were twice transfected with siRNAs targeting the Ago proteins 1 to 4 individually, and RNA levels of the Ago transcripts were measured by qPCR. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; con, control. (B) Cells, as described for panel A, were infected with wild-type virus equivalent to 50 ng p24^Gag. Forty-eight hours later, supernatants were harvested and virus was quantified by p24^Gag ELISA. (C) Infectivity of the virions produced for panel B was measured by a TZM-bl reporter assay. (D) HeLa cells stably expressing either a nonsilencing control (NSC) or Ago4-targeting shRNA were twice transfected with equivalent amounts of control siRNA (con) or siRNAs targeting Ago proteins 1, 2, and 3. Cells were harvested, and the RNA levels of the Ago transcripts were measured by qPCR. (E) Cells, as described for panel D, were plated onto coverslips, stained with an anti-DDX6 antibody, and imaged by confocal microscopy. Bars = 10 μm. (F) Cells, as described for panel D, were infected with wild-type virus equivalent to 50 ng p24^Gag, and virus levels were measured as described for panel B. (G) Infectivity of the virions produced for panel F, measured as described for panel C. For panels B, C, F, and G, values are normalized to the value for the nonsilencing control cell line transfected with control siRNA, which is set at 100%. Results are presented as the average of three independent experiments, with error bars denoting the standard deviation. *, P < 0.05 (unpaired two-tailed t test).

To determine if Ago2 is necessary for APOBEC3 antiviral activity, we derived a 293T cell line stably expressing an Ago2 shRNA. Ago2 protein expression was reduced by 85 to 95%, as determined by immunoblotting (see Fig. S10A in the supplemental material), and functional impairment of Ago2 activity was confirmed by using an siRNA targeting the ESCRT (endosomal sorting complex required for transport) component ALIX. RNAi-mediated suppression of endogenous ALIX expression was less efficient in Ago2-depleted 293T cells (see Fig. S10A in the supplemental material). We then cotransfected these cells with pHIV_NL4-3Δvif and pHA-APOBEC3A, -C, -F, -G, and -H expression plasmids and analyzed virion production and infectivity as before (see Fig. S10B and C in the supplemental material). APOBEC3 protein antiviral activity was unperturbed by Ago2 knockdown, even when A3G was expressed at lower levels (see Fig. S10D to F in the supplemental material).

APOBEC3 proteins do not specifically regulate miRNA-mediated translational repression.

We next wished to determine the possible role of APOBEC3 proteins in Argonaute activity, specifically, in the regulation of mRNAs through miRNAs. It has been proposed that several APOBEC3 proteins can stimulate cellular gene expression by inhibiting miRNA-mediated repression (46). The Argonaute proteins colocalize with A3G and A3F in P bodies and coimmunoprecipitate in a partially RNase-insensitive manner (36, 37). Therefore, we tested the entire APOBEC3 family for the ability to coimmunoprecipitate with Ago2 using formaldehyde cross-linking. This procedure freezes the RNP complex to prevent dissociation and allows stringent washing to remove proteins that associate nonspecifically (79, 96). Forty-eight hours after 293T cells were cotransfected with pHA-APOBEC3(A-H) and pMyc-Ago2 plasmids, the cells were fixed with 0.05% HCHO, lysed, and sonicated to disrupt insoluble complexes. Ago2 was immunoprecipitated using an anti-Myc antibody and washed, the cross-links were reversed, and proteins were analyzed by immunoblotting (Fig. 6A). In agreement with previously published reports (36, 37, 98), Ago2 efficiently immunoprecipitated A3G and also interacted with A3F and A3H but to lesser extents. Small amounts of A3A and A3C were consistently found to be associated with Ago2, while no detectable interactions were observed for A3B and A3D/E. A3F, A3G, and A3H also coimmunoprecipitated with Ago1 and Ago3 (data not shown). Therefore, unlike localization to P bodies, the ability to interact with Ago2 does not appear to correlate with the packaging efficiencies of the APOBEC3 proteins (see Fig. S2A in the supplemental material).

Fig 6.

APOBEC3 proteins do not inhibit let-7 miRNA-mediated translational repression. (A) 293T cells cotransfected with equivalent amounts of pHA-GFP or pHA-APOBEC3(A-H) and pMyc-Ago2 expression vectors were incubated with 0.05% formaldehyde, lysed, sonicated, immunoprecipitated (Ip) with an anti-Myc antibody, and then analyzed by immunoblotting using anti-Myc and anti-HA antibodies. A representative result from one of four independent experiments is shown. (B) Schematic diagram of the firefly luciferase reporter constructs used in this study. The wild-type construct contains four binding sites for the let-7 miRNA, while the mutant construct contains mutated binding sites. The sequence of the let-7 binding site is highlighted in bold. CMV, cytomegalovirus; UTR, untranslated region. (C) HeLa cells were cotransfected with pHA-GPF or pHA-APOBEC3(A-H) expression vectors and either the wild-type or mutant firefly luciferase reporter construct as well as a Renilla luciferase construct as a transfection control. At 30 h posttransfection, the cells were lysed and a dual-luciferase assay was performed. Raw values of firefly luciferase expression levels are shown. (D) Corresponding raw values of Renilla luciferase expression levels from cells as described for panel B. For both panels C and D, values are normalized to the amount of GFP expressed with the mutant reporter, which is set at 100%. Results are presented as the average of four independent experiments, with error bars denoting the standard deviation.

We then sought to replicate the inhibitory activity of the APOBEC3 proteins on miRNA-mediated repression using established luciferase reporter constructs containing four binding sites for the let-7 miRNA (66). In the absence of any regulatory proteins, expression of luciferase from a construct containing wild-type let-7 binding sites is suppressed by the action of endogenous let-7 compared to that for the mutated control in which all four binding sites are disrupted (Fig. 6B). We cotransfected pHA-GFP and the luciferase reporter constructs and observed a 4- to 5-fold decrease in expression of the wild-type construct, as previously reported (66) (Fig. 6C). Importantly, coexpression of the APOBEC3 proteins generally led to similar changes for the expression of both reporter constructs, implying that the changes observed for the let-7 binding site-containing construct are not due to altered activity of the cognate miRNA. A3C had the most profound effect, enhancing firefly luciferase levels ∼10-fold, whereas A3D/E dramatically decreased luciferase expression. Analogous effects were also observed for Renilla luciferase, which is commonly used as a transfection control and contains no known miRNA binding sites (Fig. 6D). Our results therefore argue against a role for the APOBEC3 proteins as modulators of miRNA-mediated translational repression but suggest that they could act in other ways to modify gene expression.

DISCUSSION

The expansion of the APOBEC3 family in higher primates, coupled with the fact that the genes themselves are subject to positive selection (87), strongly suggests that APOBEC3 proteins play necessary and fundamental roles in their hosts. The APOBEC3 proteins, along with Trim5α, tetherin, and most recently, SAMHD1, have been identified as host restriction factors that have the intrinsic ability to defend against invading exogenous viruses and endogenous retroelements (45, 58, 75, 90, 92). Although the antiviral activities of the APOBEC3 proteins have been well documented, many questions remain regarding the scope of their functions, potential cellular cofactors, and mechanisms for regulating their protein activity.

Several APOBEC3 proteins localize to P bodies (see Fig. S2 and S3 in the supplemental material), which are RNA-rich microdomains that have previously been reported to influence HIV-1 infectious virion production, positively or negatively (13, 17, 73). However, contrary to these reports, we did not find any substantial effects of P-body depletion or perturbation on HIV-1 infectivity, even when several different experimental approaches were employed (Fig. 1 to 3 and 5). This leads us to conclude that P bodies and at least some P-body components do not appear to regulate HIV-1 replication. It is important to note, however, that although in each experiment protein abundance and, hence, P-body numbers were substantially depleted, without complete elimination of protein expression (e.g., via gene knockout), a role in HIV-1 regulation cannot be comprehensively ruled out. Despite this, with levels of protein depletion comparable to our own, several groups have reported that P-body loss affects HIV-1 production and infectivity (17, 73). The reasons for these disparities are unclear at present, though our own results are consistent across different cell lines, with different experimental configurations, and with altered expression of multiple P-body constituents. In agreement with two reports (17, 73), we did observe a slight increase in virus production upon DDX6 knockdown when employing proviral transfections (see Fig. S5C in the supplemental material). However, this was not observed upon depletion of Lsm1 (see Fig. S6 in the supplemental material), which indicates that this modest effect is not due to the general loss of P bodies but is more specifically a consequence of reduced levels of DDX6, a known translational repressor (23). The fact that this was not reproduced using the more physiologically relevant method of infection (Fig. 1 and 2) also questions the possible in vivo relevance of this phenomenon.

The role of the Argonaute proteins in modulating HIV-1 virion production is controversial, as depletion of Ago2 has been reported to increase or decrease progeny virion production (13, 73). We individually knocked down each Argonaute protein and did not observe any significant effects on HIV-1 particle production or infectivity (Fig. 5A to C). In order to address potential functional redundancy, we also performed a combined (partial) knockdown of all four Argonaute proteins (Fig. 5D) and found that even though virion production was slightly reduced (Fig. 5F), viral infectivity was modestly increased (Fig. 5G). Because Argonaute depletion did not alter P-body abundance (Fig. 5E), we hypothesize that the Argonaute proteins regulate HIV-1 infectivity by a mechanism which is independent of these foci, perhaps similar to Ago2-mediated regulation of hepatitis C virus replication (9).

Close association of HIV-1 RNA or proteins with P bodies might be expected if the virus was modulated in some manner by these foci. Several reports have shown that HIV-1 RNA and Gag protein localize to P bodies (17, 70, 73). Using an MS2-based tethering system and fluorescently tagged Gag in the context of HIV-1 proviral constructs, we were unable to detect colocalization of HIV-1 gRNA and Gag with DDX6 (Fig. 4), findings consistent with one previous report (14). Therefore, we conclude that HIV-1 gRNA and Gag do not appear to localize to P bodies and the production of infectious virions is not affected when these bodies are disrupted.

It is evident from the work of several groups that certain APOBEC3 proteins interact and colocalize with a diverse array of cellular proteins involved in RNA regulation and metabolism (21, 37, 57, 98). Whether these interactions are necessary for antiviral activity or a different cellular function (or functions) remains to be determined. As the most efficiently packaged APOBEC3 proteins colocalize with DDX6 and Ago2 at P bodies (see Fig. S2 and S3 in the supplemental material), we also sought to determine their influence on APOBEC3 antiviral activity. Using siRNAs and shRNAs targeting DDX6 and Lsm1, respectively, we found that P-body depletion had no effect on A3G antiviral activity or virion incorporation, for both the ectopically expressed and endogenous protein (Fig. 1 to 3). This was also true for other family members (see Fig. S5 and S6 in the supplemental material) and implies that P bodies are not sites required for the packaging of APOBEC3 proteins into HIV-1 virions.

Although several APOBEC3 proteins interact with Ago2 (Fig. 6A), this protein was not necessary for APOBEC3-mediated viral inhibition (see Fig. S10 in the supplemental material). We also addressed the possibility that the APOBEC3 proteins may be involved in the regulation of Argonaute activity. It has previously been reported that several APOBEC3 proteins inhibit miRNA-mediated repression (46) and A3G specifically inhibits this process through disruption of the interaction between Ago2 and MOV10 (62). However, we found that the APOBEC3 proteins had nonspecific effects on transfected gene expression, as they were able to affect the expression of reporter genes irrespective of miRNA binding sites, indicating that they do not regulate miRNA-mediated translational repression (Fig. 6C and D). We have also recently reported that MOV10 is not required for this cellular process (5). While the basis for this discrepancy is not yet clear, we note that several APOBEC3 family members, particularly A3C, exhibited more general effects on reporter gene expression (Fig. 6C and D). It will be of future interest to identify cellular RNA targets/ligands of the APOBEC3 proteins, as this will focus efforts aimed at understanding the contributions that these proteins may make to RNA regulation and metabolism.

Taken together, our data suggest that P bodies and certain P-body components are not involved in the regulation of APOBEC3 protein antiviral activity or, more generally, HIV-1 replication. However, the Argonaute protein family and miRNAs may contribute to viral regulation. The functional relevance of APOBEC3 colocalization and interaction with numerous P-body proteins therefore remains enigmatic. Provocatively, recent work has implicated the APOBEC3 proteins in causing somatic mutations leading to breast cancer (78). Therefore, it is possible that incorporation into high-molecular-weight ribonucleoprotein complexes and/or P bodies may be mechanisms for regulating APOBEC3 cytidine deaminase activity (21) by sequestering these proteins away from chromosomal DNA.