TRIM5α Degradation via Autophagy Is Not Required for Retroviral Restriction

ABSTRACT

TRIM5α is an interferon-inducible retroviral restriction factor that prevents infection by inducing the abortive disassembly of capsid cores recognized by its C-terminal PRY/SPRY domain. The mechanism by which TRIM5α mediates the disassembly of viral cores is poorly understood. Previous studies demonstrated that proteasome inhibitors abrogate the ability of TRIM5α to induce premature core disassembly and prevent reverse transcription; however, viral infection is still inhibited, indicating that the proteasome is partially involved in the restriction process. Alternatively, we and others have observed that TRIM5α associates with proteins involved in autophagic degradation pathways, and one recent study found that autophagic degradation is required for the restriction of retroviruses by TRIM5α. Here, we show that TRIM5α is basally degraded via autophagy in the absence of restriction-sensitive virus. We observe that the autophagy markers LC3b and lysosome-associated membrane protein 2A (LAMP2A) localize to a subset of TRIM5α cytoplasmic bodies, and inhibition of lysosomal degradation with bafilomycin A1 increases this association. To test the requirement for macroautophagy in restriction, we examined the ability of TRIM5α to restrict retroviral infection in cells depleted of the autophagic mediators ATG5, Beclin1, and p62. In all cases, restriction of retroviruses by human TRIM5α, rhesus macaque TRIM5α, and owl monkey TRIM-Cyp remained potent in cells depleted of these autophagic effectors by small interfering RNA (siRNA) knockdown or clustered regularly interspaced short palindromic repeat (CRISPR)-Cas9 genome editing. Collectively, these results are consistent with observations that the turnover of TRIM5α proteins is sensitive to autophagy inhibition; however, the data presented here do not support observations that the inhibition of autophagy abrogates retroviral restriction by TRIM5 proteins.

IMPORTANCE Restriction factors are a class of proteins that inhibit viral replication. Following fusion of a retrovirus with a host cell membrane, the retroviral capsid is released into the cytoplasm of the target cell. TRIM5α inhibits retroviral infection by promoting the abortive disassembly of incoming retroviral capsid cores; as a result, the retroviral genome is unable to traffic to the nucleus, and the viral life cycle is extinguished. In the process of restriction, TRIM5α itself is degraded by the proteasome. However, in the present study, we have shown that in the absence of a restriction-sensitive virus, TRIM5α is degraded by both proteasomal and autophagic degradation pathways. Notably, we observed that restriction of retroviruses by TRIM5α does not require autophagic machinery. These data indicate that the effector functions of TRIM5α can be separated from its degradation and may have further implications for understanding the mechanisms of other TRIM family members.

INTRODUCTION

Tripartite motif-containing proteins (TRIMs) are a large family of proteins that participate in diverse cellular activities, including cell cycle regulation, embryonic development, regulation or direct activation of cellular signaling pathways, and intrinsic immunity to viral infection (1–4). Expression of many TRIM family proteins is induced by interferon treatment (5, 6), and many TRIM family proteins have been shown to activate cellular signaling pathways through the generation of K-63-linked ubiquitin chains (7, 8).

The tripartite motif present in all TRIM proteins includes an N-terminal RING domain, one or two B-box domains, and a coiled-coil (CC) domain. In most cases, the RING domain of TRIM family proteins functions as an E3 ligase (2, 9), while the B-box and CC domains promote the self-association of TRIM proteins (10–13), leading many TRIM family members to assemble into cytoplasmic or nuclear bodies (14). Variability between TRIM proteins is found mostly at the C terminus, where numerous domains are thought to confer distinct cellular activities to TRIM family proteins (2, 4).

Primate TRIM5α proteins are distinguished from other TRIM family members by their expression of a C-terminal PRY/SPRY (SPRY) domain, which allows TRIM5α to bind to retroviral capsids and inhibit viral replication. The C-terminal SPRY domain itself has been subjected to intense selective pressure (15), such that the SPRY domains of different primate species have evolved to inhibit different viruses (16, 17). For example, the TRIM5α protein expressed in rhesus macaques (rhTRIM5α) restricts human immunodeficiency virus type 1 (HIV-1) and N-tropic murine leukemia virus (N-MLV) (18, 19), while the human variant of TRIM5α (huTRIM5α) inhibits N-MLV but has a limited ability to restrict HIV-1 (18, 19). Furthermore, in certain primates, including owl monkeys, the C-terminal PRY/SPRY domain has been functionally replaced by the retrotranspositional insertion of cyclophilin A, creating a TRIM-Cyp fusion that potently inhibits HIV-1 infection in these monkeys (20).

Numerous studies have found that retroviral restriction by TRIM5α proteins occurs by a two-step mechanism (21–25). In the first step, which is sufficient to prevent infection, TRIM5α recognizes the viral capsid via its C-terminal PRY/SPRY domain (or CypA in the case of TRIM-Cyp). In the second step, TRIM5α induces the abortive disassembly of the viral capsid core and prevents the accumulation of reverse transcription (RT) products. The latter step requires the E3 ligase activity of the TRIM5α RING domain and is sensitive to proteasome inhibitors, although restriction of infection remains potent in both cases (9, 21–24). Additionally, the presence of restriction-sensitive virus triggers the degradation of TRIM5α, and this degradation is also sensitive to proteasome inhibitors (25). Furthermore, rhTRIM5α cytoplasmic bodies have been shown to associate with proteasomal subunits and viral capsids (26).

Previous studies have also interrogated autophagy, another cellular degradative pathway, and its role in TRIM5α-mediated restriction. We previously observed that TRIM5α associates with the autophagic adaptor protein p62/sequestosome1, and depletion of p62 by small interfering RNA (siRNA) caused a reduction in retroviral restriction in cells expressing huTRIM5α or rhTRIM5α (27); however, because the depletion of p62 also reduced the expression level of TRIM5α, we could not conclude that p62 is directly required for the restriction of retroviruses by TRIM5α. Nevertheless, one recent study observed that the depletion of autophagic mediators abrogated rhTRIM5α-mediated restriction of HIV-1 (28). Although this observation appears to disagree with the results of numerous studies suggesting a proteasome-dependent step in restriction, this apparent discordance might be explained by cross talk between autophagic and proteasomal pathways, which is known to occur in many contexts (29, 30), or by gross perturbation of ubiquitin homeostasis caused by pharmacological inhibition of the proteasome.

We therefore sought to determine how the inhibition of autophagy affected the turnover of TRIM5α and its ability to inhibit retroviral infection. In agreement with data from a previous study by Mandell and colleagues (28), we observe that the turnover of yellow fluorescent protein (YFP)-tagged rhTRIM5α (YFP-rhTRIM5α) is sensitive to bafilomycin A1 (BafA1), a drug that perturbs the final steps of autophagic degradation, but not to the proteasome inhibitor MG132. Additionally, we observe that rhTRIM5α colocalizes with the autophagy markers LC3b and lysosome-associated membrane protein 2A (LAMP2A), and this colocalization is increased when autophagic degradation is prevented by BafA1. However, when critical mediators of macroautophagy, including ATG5, Beclin1, and p62, were depleted in cells by siRNA or clustered regularly interspaced short palindromic repeat (CRISPR)-Cas9 genome editing, the restriction of retroviral infection and reverse transcription by huTRIM5α, rhTRIM5α, or TRIM-Cyp remained intact. Collectively, these results demonstrate that the degradation of TRIM5α by autophagy is not required for its ability to restrict retroviral infection.

MATERIALS AND METHODS

Cells and pharmaceuticals.

Expression plasmids for YFP-rhTRIM5α and hemagglutinin (HA)-tagged rhTRIM5α or TRIM-Cyp were described previously (31, 32). To quantify TRIM5α accumulation, a lentiviral plasmid (pLVX; Clontech) expressing human TRIM5α containing a C-terminal firefly luciferase reporter gene was created. The HeLa and TE671 cell lines were obtained from the American Type Culture Collection. THP-1 cells were obtained from the AIDS Reagent Repository. Wild-type (wt) and ATG5^−/− mouse embryo fibroblasts (MEFs) were generously provided by Noboru Mizushima (University of Tokyo). HeLa cells, TE671 cells, and wt and ATG5^−/− MEFs were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (HyClone, Logan, UT, USA), 100 U/ml penicillin, 100 μg/ml streptomycin, and 10 μg/ml ciprofloxacin. THP-1 cells were cultured in RPMI medium with FBS and antibiotics identical to those described above. Cells were maintained in the presence of 5% CO₂ at 37°C. Bafilomycin A1 and MG132 (Cayman Chemical Company, Ann Arbor, MI, USA) were used at 100 nM and 1 μg/ml, respectively. Cycloheximide (CHX) was used at 20 μg/ml. Cyclosporine (CsA; Sigma-Aldrich) was used at a final concentration of 2.5 μM.

Steady-state protein analysis.

Cell lines stably expressing the indicated TRIM5α-tagged proteins were treated with BafA1 or MG132, and cells were harvested at the indicated time points. Whole-cell lysates were prepared by lysing 2 × 10⁵ cells in lysis buffer (100 mM Tris [pH 8.0], 1% NP-40, 150 mM NaCl) containing a protease inhibitor cocktail (Roche Applied Science, Indianapolis, IN, USA). After lysis, Laemmli 2× SDS sample buffer was added, and samples were boiled for 5 min. Equal amounts of protein were loaded onto a 10% polyacrylamide gel for SDS-PAGE. After separation of proteins via SDS-PAGE, proteins were transferred onto nitrocellulose membranes and detected by incubation with the following antibodies: anti-β-actin (Sigma-Aldrich, St. Louis, MO, USA), anti-HA (clone 3F10) conjugated to horseradish peroxidase (HRP) (Roche Applied Science, Indianapolis, IN, USA), anti-green fluorescent protein (GFP) (Clontech Laboratories, Inc., Mountain View, CA, USA), and monoclonal antibodies to TRIM5α (clone 3F1-1-9) (AIDS Reagent Program). Secondary antibodies conjugated to HRP (Thermo Fisher Scientific, Waltham, MA, USA) were used where necessary, and antibody complexes were detected by using the SuperSignal West Femto chemiluminescent substrate (Thermo Fisher Scientific, Waltham, MA, USA). Chemiluminescence was detected by using a UVP EC3 imaging system (UVP LLC, Upland, CA, USA).

Protein turnover assay.

Cell lines stably expressing the indicated TRIM5α-tagged proteins were treated with cycloheximide alone or in the presence of BafA1 or MG132. Cells were harvested 6 h following the addition of cycloheximide. Equivalent amounts of protein from individual samples were subjected to SDS-PAGE, and TRIM5α was detected by Western blotting.

Flow cytometry.

For steady-state protein analysis by flow cytometry, equivalent numbers of cells stably expressing YFP-rhTRIM5α were plated in a 12-well plate and treated with cycloheximide, BafA1, or MG132 for 6 h or 18 h, after which the cells were fixed in a 1% formaldehyde–phosphate-buffered saline (PBS) solution. Protein levels were determined by measuring the mean fluorescence intensity (MFI) in the fluorescein isothiocyanate (FITC) channel for 10,000 events per sample, using a FACSCanto II flow cytometer (Becton Dickinson, San Jose, CA, USA). For viral infectivity assessment by flow cytometry, equivalent numbers of cells of the indicated cell lines were plated in 24-well plates. Dilutions of the viral supernatant were applied to the cells, after which the cells were subjected to spinoculation (1,200 × g for 2 h at 13°C). For experiments involving cells expressing TRIM-Cyp, cyclosporine (final concentration of 2.5 μM) or dimethyl sulfoxide (DMSO) was added to the cells concurrently with the viral supernatant. Following spinoculation, the medium was subsequently changed, and after 48 h, the cells were harvested and fixed in a 1% formaldehyde–PBS solution for flow cytometric analysis. Percent infectivity was determined by measuring the proportion of GFP-positive cells in the FITC channel for 10,000 events per sample, using a FACSCanto II flow cytometer (Becton Dickinson, San Jose, CA, USA).

Luciferase measurement.

THP-1 cells stably expressing human TRIM5α-luciferase were plated into a 6-well plate. The cells were differentiated by treating the cells with 100 ng phorbol myristate acetate (PMA) for 48 h. Subsequently, the cells were treated with BafA1 or MG132 for 18 h, after which cells were harvested for luciferase measurement (catalog no. E4530; Promega).

Immunofluorescence microscopy.

Cells were allowed to adhere to fibronectin-treated glass coverslips and fixed with 3.7% formaldehyde (Polysciences) in 0.1 M PIPES [piperazine-N,N′-bis(2-ethanesulfonic acid)] (pH 6.8). Cells were permeabilized with 0.1% saponin, 10% normal donkey serum, and 0.01% sodium azide in PBS. We used the following primary antibodies: rabbit anti-LC3b (Sigma-Aldrich, St. Louis, MO, USA) and mouse anti-LAMP2A (BD Pharmingen, San Diego, CA, USA). Primary antibodies were labeled with fluorophore-conjugated donkey anti-mouse or anti-rabbit antibody (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA). Images were collected with a DeltaVision microscope (Applied Precision, Issaquah, WA, USA) equipped with a digital camera (CoolSNAP HQ; Photometrics, Tucson, AZ, USA), using a 1.4-numerical-aperture (NA) 100× objective lens, and were deconvolved with SoftWoRx software (Applied Precision, Issaquah, WA, USA).

Image analysis.

Twenty z-stack images were acquired by using identical acquisition parameters. Deconvolved images were analyzed by using Imaris software (Bitplane). Surfaces were generated around YFP-rhTRIM5α, and the maximum fluorescence intensities of LC3b and LAMP2A on each surface were quantified. Background fluorescence intensities were calculated and used to set LC3b and LAMP2A intensity thresholds. Data were graphed with Prism (GraphPad Software, Inc.).

siRNA transfections.

Transcripts for several macroautophagy factors were targeted by using siRNAs against ATG5 (catalog no. sc-41445; Santa Cruz), Beclin1 (catalog no. sc-29797; Santa Cruz), and p62/SQSTM1 (catalog no. sc-29679; Santa Cruz) and control siRNA (catalog no. sc-37007; Santa Cruz). A total of 300,000 TE671 cells were plated in 6-well plates and transfected with 30 nM the indicated siRNAs twice over a 48-h period. The siRNAs were transfected by using Lipofectamine 2000 (catalog no. 11668027; Life Technologies, Grand Island, NY, USA), according to the manufacturer's instructions. Whole-cell lysates were prepared 72 h following the second transfection, as described above. Proteins were separated via SDS-PAGE and transferred onto nitrocellulose membranes. Membranes were probed with anti-Atg5 (catalog no. NB110-53818; Novus), anti-Beclin1 (catalog no. 3738; Cell Signaling), anti-p62/SQSTM1 (catalog no. 7695S; Cell Signaling), anti-β-actin, and anti-β-tubulin antibodies. Secondary antibodies conjugated to HRP (Thermo Fisher Scientific, Waltham, MA, USA) were used where necessary, and antibody complexes were detected by using the SuperSignal West Femto chemiluminescent substrate (Thermo Fisher Scientific, Waltham, MA, USA). Chemiluminescence was detected by using a UVP EC3 imaging system (UVP LLC, Upland, CA, USA).

Generation of knockout cells using CRISPR-Cas9 genome editing.

The indicated knockout TE671 and HeLa cell lines were generated by using LentiCRISPRv2 (Addgene plasmid 52961), a gift from Feng Zhang (33). Guide sequences were generated by using the CRISPR design tool (see http://www.crispr.mit.edu) or were taken from available guide sequences from the Genome-Scale CRISPR Knockout (GeCKO2) library (33). An oligonucleotide targeting ATG5, 5′-CACCGGATGGACAGTTGCACACACT-3′, and an oligonucleotide targeting Beclin1, 5′-CACCGATCTGCGAGAGACACCATCC-3′, were annealed and cloned into LentiCRISPRv2. Lentivirus was prepared by transfecting equal amounts of vesicular stomatitis virus G (VSV-G), psPAX2 (catalog no. 11348; Didier Trono, NIH AIDS Reagent Program) (34), and LentiCRISPRv2 (containing the guide RNA of interest) into HEK293T cells. The viral supernatant was harvested at 48 h posttransfection, filtered through 0.45-μm filters (Millipore), and applied to TE671 or HeLa cells. Forty-eight hours after transduction, 5 μg/ml of puromycin was added to the cells, and following selection, cells were collected for knockout assessment by Western blotting and phenotypic analysis.

Virus generation and quantitative real-time PCR for viral RT products.

HIV and MLV reporter viruses were prepared as described previously (27). Briefly, HIV-1 reporter virus was produced by polyethylenimine (PEI) transfection of HEK293T cells with 10 μg of VSV-G and 15 μg of the proviral construct R7ΔEnvGFP, in which the Nef gene was replaced with GFP. MLV reporter virus was produced by PEI transfection of HEK293T cells with equal amounts of VSV-G, the pCigN or pCigB packaging plasmid (for the generation of N-MLV and B-tropic MLV [B-MLV], respectively), and the GFP reporter vector. Virus was harvested as previously described (35). MLV was titered onto CRFK cells to normalize virus input in infectivity studies, as described previously (21). Quantitation of viral RT products was performed as previously described (21, 23). Briefly, equivalent numbers of the indicated cells were seeded into 12-well plates. Cells were infected with the indicated virus, and they were subsequently incubated for 18 h at 37°C. Genomic DNA was harvested by using a DNeasy tissue kit (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions and digested with 1 U/μl DpnI (New England BioLabs, Ipswich, MA, USA) for 4 h at 37°C to remove residual plasmid DNA. Real-time PCR was performed with SYBR green PCR reagent (Applied Biosystems, Carlsbad, CA, USA), using primers for late RT, GFP, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Dilutions of the proviral plasmid and GAPDH (10-fold) were used to generate standard curves. Samples were normalized to 10 ng of total cellular DNA or GAPDH standards.

RESULTS

Inhibition of autophagy decreases rhTRIM5α turnover.

To understand the role of proteasomal and autophagic degradation pathways in the turnover of rhTRIM5α, we examined the accumulation of YFP-rhTRIM5α in cells subject to the inhibition of these pathways. We utilized cells expressing YFP-rhTRIM5α from a stably integrated retroviral vector (19, 31). In these cells, TRIM5α localizes to small, highly dynamic cytoplasmic bodies (31) that associate with HIV-1 virions during restriction (22, 26) and potently restrict HIV-1 infection (19, 31). Conversely, the protein localizes to large aggregates and fails to restrict viral infection following transient transfection with YFP-rhTRIM5α (36). To block autophagic degradation, we used bafilomycin A1 (BafA1), which prevents hydrogen flux through the ATPase present on lysosomal and endosomal membranes (37). This prevents the autophagic degradation of substrates within the autophagolysosome and additionally is thought to perturb autophagy by preventing the fusion of autophagosomes with lysosomes (38, 39). To inhibit proteasome-mediated turnover, we utilized the proteasome inhibitor MG132. HeLa cells stably expressing YFP-rhTRIM5α were treated with 100 nM BafA1 or 1 μg/ml MG132 for 18 h, and changes in YFP-rhTRIM5α expression were measured by Western blotting (Fig. 1A). YFP-rhTRIM5α accumulated in HeLa cells following treatment with BafA1, compared to untreated cells. Consistent with data from previous studies, MG132 had little effect on steady-state protein levels (23). Similar results were observed when the total amount of cellular YFP-rhTRIM5α was measured by flow cytometry following similar treatments (Fig. 1B). These experiments show that inhibition of autophagy by BafA1 increases the steady-state expression level of TRIM5α. To ensure that the steady-state accumulation observed in the presence of BafA1 was due to altered TRIM5α turnover, we measured the turnover of YFP-rhTRIM5α following the treatment of cells with cycloheximide (CHX), which inhibits protein translation. Most of the YFP-rhTRIM5α was turned over following 6 h of CHX treatment (Fig. 1C and D), consistent with the short half-life of TRIM5α reported in other studies (23, 40). However, rhTRIM5α protein turnover was reduced following CHX treatment in the presence of BafA1 (Fig. 1C and D). Notably, this rescue was not observed in cells treated with CHX and MG132 (Fig. 1C and D), indicating that YFP-rhTRIM5α protein turnover is insensitive to proteasome inhibition. Similar results were obtained when YFP-rhTRIM5α accumulation was measured by flow cytometry under the same conditions (Fig. 1E). These data suggest that in the absence of restriction-sensitive virus, YFP-rhTRIM5α is basally turned over by autophagy.

FIG 1.

rhTRIM5α protein turnover is sensitive to the autophagy inhibitor BafA1. (A) HeLa cells stably expressing YFP-rhTRIM5α were treated with BafA1 or MG132 or left untreated for 18 h, after which equivalent amounts of total protein were analyzed by SDS-PAGE and Western blotting. YFP-rhTRIM5α levels determined by densitometry, normalized to cellular β-actin levels, are plotted in the bottom panel. Data are representative of results from three independent experiments. (B) YFP-rhTRIM5α expression in HeLa cells stably expressing YFP-rhTRIM5α following treatment with BafA1 or MG132 or no treatment for 18 h was measured by flow cytometry. The mean fluorescence intensity (MFI) of YFP is reported for each treatment group. Data are representative of results from three independent experiments. (C) HeLa cells stably expressing YFP-rhTRIM5α were plated in triplicate and left untreated or treated with cycloheximide alone (CHX), CHX and BafA1 (CHX + BafA1), or CHX and MG132 (CHX + MG132) for 6 h. Equivalent amounts of total protein were analyzed by SDS-PAGE and Western blotting. Data are representative of results from three independent experiments. (D) YFP-rhTRIM5α levels as determined by densitometry, normalized to cellular β-actin levels, are plotted based on the treatments described above for panel C. Data are representative of results from three independent experiments (P < 0.05, compared to CHX alone). (E) YFP-rhTRIM5α expression was measured by flow cytometry in HeLa cells stably expressing YFP-rhTRIM5α following treatments as described above for panel C. The MFI of YFP is reported for each treatment group. Data are representative of results from three independent experiments (P < 0.05, compared to CHX alone).

Degradation of human TRIM5α is sensitive to both autophagy and proteasome inhibition.

To determine the cellular degradative pathway responsible for the degradation of human TRIM5α (huTRIM5α), we measured the accumulation of endogenous huTRIM5α following treatment with autophagy or proteasome inhibitors. Human TE671 cells were treated with 1 μg/ml MG132 or 100 nM BafA1, and endogenous TRIM5α levels were measured by Western blotting. In contrast to what was observed in the case of HeLa cells expressing YFP-rhTRIM5α, we observed that treatment with both MG132 and BafA1 induced the accumulation of endogenous huTRIM5α relative to the untreated control, with MG132 treatment inducing a more pronounced accumulation of huTRIM5α (Fig. 2A). To confirm these results, we generated a lentiviral vector expressing huTRIM5α containing a C-terminal luciferase reporter. We used this vector to generate a THP-1 cell line expressing this protein. Following differentiation with PMA, these cells were treated with MG132 or BafA1 as described above, and luciferase activity was measured to determine the degree to which either drug induced the accumulation of huTRIM5α-Luc. Similar to our observations of endogenous huTRIM5α induction in TE671 cells, treatment with both drugs produced an increase in the amount of huTRIM5α-Luc present in these cells, with MG132 inducing a more pronounced accumulation of huTRIM5α-Luc than BafA1 (Fig. 2B). These results demonstrate that turnover of huTRIM5α is sensitive to both autophagy and proteasome inhibitors.

FIG 2.

huTRIM5α turnover is sensitive to autophagic and proteasomal inhibitors. (A) TE671 cells were treated with BafA1 or MG132 for 18 h, and endogenous huTRIM5α expression was measured by Western blotting. (B) THP-1 cells transduced with a retroviral vector expressing huTRIM5α-Luc were differentiated with 100 ng PMA for 48 h. Differentiated cells were treated with BafA1 or MG132 for 18 h, after which luciferase activity (signified by relative luminescence units [RLU]) was quantified. Error bars represent the standard deviations of data from three replicates. Results are representative of data from at least three experiments.

Inhibition of autophagy alters the cellular localization of rhTRIM5α.

A novel characteristic of the TRIM family of proteins is their intrinsic ability to form higher-order assemblies, and in the case of TRIM5α, this activity is essential for the ability of the protein to act as a retroviral restriction factor (32). Normally, rhTRIM5α localizes to both small, discrete cytoplasmic puncta, termed cytoplasmic bodies (19), and a diffuse pool of cytoplasmic protein that is capable of forming cytoplasmic bodies de novo around individual virions (22, 31). Given our observations that the turnover of YFP-rhTRIM5α is sensitive to the inhibition of autophagy, we further asked if changes in the cellular localization of YFP-rhTRIM5α are similarly sensitive to autophagy inhibition. Although we observed that MG132 did not dramatically alter the turnover of the cellular pool of rhTRIM5α (Fig. 1), previous studies observed that treatment of cells stably expressing rhTRIM5α with MG132 drives the protein to form cytoplasmic bodies that are larger than normal bodies (23). To study the effects of autophagic or proteasomal inhibition on the cellular localization of YFP-rhTRIM5α, HeLa cells stably expressing YFP-rhTRIM5α were treated with BafA1 or MG132 for 18 h, after which the abundance of YFP-rhTRIM5α in cells was quantified. As shown in Fig. 3, inhibition of autophagy by BafA1 altered the localization of TRIM5α, resulting in the accumulation of more cytoplasmic bodies than observed in untreated cells (Fig. 3A and C), while treatment with MG132 recapitulated previously reported findings (Fig. 3B). These observations were validated by quantitative image analysis to characterize YFP-TRIM5α localization in data sets obtained from the individual treatment groups. BafA1 treatment produced a substantial increase in the number of cytoplasmic bodies per cell (Fig. 3D), compared to both untreated and MG132-treated cells. Therefore, as BafA1 treatment inhibits the turnover of rhTRIM5α, it also alters the subcellular localization of rhTRIM5α, resulting in more numerous cytoplasmic bodies, consistent with these bodies being autophagosomal structures destined for clearance via lysosomal degradation pathways.

FIG 3.

The subcellular localization of YFP-rhTRIM5α changes in the presence of BafA1 and MG132. (A to C) HeLa cells stably expressing YFP-rhTRIM5α were seeded onto fibronectin-treated coverslips for 18 h. Cells were left untreated or treated with BafA1 or MG132 during this time. Cells were subsequently fixed and stained with DAPI. z-stack images were collected with a DeltaVision microscope equipped with a digital camera using a 1.4-NA 100× objective lens and were deconvolved with SoftWoRx deconvolution software. Individual channel images were superimposed to create the merged panels. Images of cells left untreated (A), treated with MG132 (B), or treated with BafA1 (C) are presented. Images are representative of results from at least three experiments. (D) To quantify the number of rhTRIM5α cytoplasmic bodies in each treatment group, 20 images were taken per treatment under identical acquisition parameters. Each image was analyzed by using Imaris imaging software. The means and standard errors of the means are highlighted in red. *, P < 0.0001.

rhTRIM5α colocalizes with LC3b and LAMP2A following BafA1 treatment.

Our observations that treatment with BafA1 reduces the turnover and increases the accumulation of YFP-rhTRIM5α suggest that TRIM5α is degraded by an autophagic pathway. Accordingly, if YFP-rhTRIM5α is degraded by autophagy, then we would expect the cytoplasmic bodies of YFP-rhTRIM5α, which accumulate upon BafA1 treatment, to colocalize with markers of autophagy. To test this hypothesis, we utilized immunofluorescence microscopy to quantify the degree of colocalization between YFP-rhTRIM5α and LC3b, a common marker of autophagosomes, and lysosome-associated membrane protein 2A (LAMP2A). In untreated cells, a subset of YFP-rhTRIM5α puncta was observed to colocalize with LC3b and LAMP2A (Fig. 4A and B), with ∼30% of the YFP-rhTRIM5α cytoplasmic puncta being positive for at least one of these markers (Fig. 4C). However, after 6 h of treatment with BafA1, the colocalization of YFP-rhTRIM5α with both markers substantially increased (Fig. 4A and B), such that 32% of puncta were positive for one of the two markers and ∼35% were positive for both LC3b and LAMP2A (Fig. 4C). These data suggest that YFP-rhTRIM5α is rapidly turned over by autophagic degradation. When autophagy is inhibited by BafA1, YFP-rhTRIM5α that has been targeted for degradation accumulates in compartments containing LC3b and LAMP2A.

FIG 4.

rhTRIM5α colocalizes with the autophagy markers LC3b and LAMP2. (A and B) HeLa cells stably expressing YFP-rhTRIM5α were seeded onto fibronectin-treated coverslips. Cells were left untreated or treated with BafA1 for 6 h. Cells were fixed, permeabilized, and costained with rabbit anti-LC3b (A), mouse anti-LAMP2A (B), and DAPI. Representative images of cells left untreated or treated with BafA1 are presented. (C) To quantify the number of rhTRIM5α cytoplasmic bodies that were positive for LC3b or LAMP2A following each treatment, 20 z-stack images per treatment were taken under identical acquisition parameters. Imaris imaging software was used to identify YFP-rhTRIM5α puncta, and the maximum LC3b and LAMP2A staining intensity on each surface was calculated and plotted. Percentages indicate the numbers of rhTRIM5α cytoplasmic bodies that are positive for LAMP2A, LC3b, both, or neither. Images are representative of results from at least three independent experiments.

Depletion of autophagic effectors does not relieve N-MLV restriction by human TRIM5α.

The above-described studies provide evidence that YFP-rhTRIM5α is degraded by an autophagic pathway. We next asked if the depletion of key macroautophagy effector proteins was able to perturb TRIM5α-mediated retroviral restriction. To this end, we assessed retroviral restriction in human TE671 cells, which endogenously express human TRIM5α and therefore potently restrict N-tropic murine leukemia virus (N-MLV) but are permissive to infection by B-tropic MLV (B-MLV) (41). TE671 cells were transfected with siRNAs targeting ATG5, Beclin1, or p62, and the infectivity of N-MLV and B-MLV was assessed. As expected, N-MLV infection was potently inhibited compared to B-MLV infection in TE671 cells subjected to control siRNA transfection (Fig. 5B). Notably, knockdown of ATG5, Beclin1, or p62 did not relieve the restriction of N-MLV infection (Fig. 5B), suggesting that these effectors of macroautophagy are not required for the restriction of N-MLV by huTRIM5α. To confirm and extend this observation, we generated TE671 cells in which the ATG5 gene or the Beclin1 gene was disrupted using CRISPR-Cas9 genome editing (Fig. 5C and E). Similar to our findings for cells depleted of ATG5 or Beclin1 by siRNA, we observed no relief of TRIM5α-mediated restriction of N-MLV in TE671 cells in which ATG5 or Beclin1 was knocked out (Fig. 5D and F). To determine if ATG5- or Beclin1-dependent macroautophagy is required for the second step of TRIM5α-mediated restriction, in which the retroviral capsid is destabilized and viral reverse transcription is inhibited, we also measured reverse transcription products generated by N-MLV and B-MLV in these cells. As we and others have previously observed, reverse transcription by N-MLV was reduced, relative to reverse transcription by B-MLV, in unmodified TE671 cells (Fig. 5G and H). Importantly, the restriction of N-MLV reverse transcription, relative to that of B-MLV, was preserved in cells depleted of ATG5 or Beclin1 (Fig. 5G and H). These data demonstrate that perturbation of macroautophagy does not abrogate the restriction of retroviral infection or reverse transcription by endogenously expressed huTRIM5α.

FIG 5.

Depletion of autophagic mediators does not affect N-MLV restriction by huTRIM5α. (A) TE671 cells were transfected with siRNAs targeting ATG5, Beclin1, or p62 or a control siRNA. Expression of the indicated proteins was detected by Western blotting at 72 h posttransfection. (B) TE671 cells transfected with siRNAs targeting ATG5, Beclin1, or p62 or a control siRNA were collected at 72 h posttransfection. Equal numbers of siRNA-transfected cells were plated and infected with equivalent titers of VSV-G-pseudotyped N-MLV or B-MLV. Cells were harvested 48 h after infection, and infectivity, signified by the percentage of GFP-positive cells, was measured by flow cytometry. The data shown here are representative of results from three independent experiments. (C) TE671 cells were depleted of ATG5 by using CRISPR-Cas9 genome editing. Protein expression of ATG5 in wild-type and knockout cells was confirmed by Western blotting. (D) Infectivity of VSV-G-pseudotyped N-MLV or B-MLV in wild-type or ATG5 knockout TE671 cells was assayed as described above for panel B. The data shown here are representative of results from three independent experiments. (E) TE671 cells were depleted of Beclin1 by using CRISPR-Cas9 genome editing. Protein expression of Beclin1 in wild-type and knockout cells was confirmed by Western blotting. (F) Infectivity of VSV-G-pseudotyped N-MLV or B-MLV in wild-type or Beclin1 knockout TE671 cells was assayed as described above for panel B. The data shown here are representative of results from three independent experiments. (G) Wild-type or ATG5 knockout TE671 cells were infected with equivalent titers of VSV-G-pseudotyped N-MLV or B-MLV. Eighteen hours following infection, cells were harvested, and the abundance of viral DNA products was measured by quantitative PCR. For each sample, the amount of viral DNA, as measured by the number of GFP reporter copies detected, was normalized to the amount of GAPDH observed in parallel samples. Three independent experiments were conducted, and the amount of viral DNA detected in each experiment was normalized to the amount detected in the wild-type untransduced sample infected with B-MLV in that experiment. Error bars represent the standard deviations of the relative numbers of viral DNA products detected across three independent experiments. (H) Measurement of viral reverse transcription products was conducted on wild-type or Beclin1 knockout TE671 cells subjected to infection by VSV-G-pseudotyped N-MLV or B-MLV, as described above for panel G. Error bars represent the standard deviations of data from three independent experiments.

Depletion of autophagic effectors does not relieve HIV-1 restriction by rhesus macaque TRIM5α or owl monkey TRIM-Cyp.

We next assessed if autophagic adaptors are also required for the restriction of HIV-1 by rhTRIM5α and owl monkey TRIM-Cyp. We generated HeLa cell lines in which ATG5 or Beclin1 was disrupted by CRISPR-Cas9 genome editing (Fig. 6A and C and 7A). In the challenge with HIV-1, we observed extensive infection in wild-type HeLa cells and in HeLa cells depleted of ATG5 or Beclin1 (Fig. 6B and D). In contrast, when wild-type and ATG5- or Beclin1-depleted HeLa cells were transduced to stably express rhTRIM5α, these cells potently restricted HIV-1 infection relative to their untransduced counterparts (Fig. 6B and D). Analogous results were obtained with ATG5 knockout mouse embryonic fibroblasts (not shown). Furthermore, we observed similar restriction in wild-type and ATG5 or Beclin1 knockout HeLa cells stably expressing owl monkey TRIM-Cyp, compared to untransduced cells (Fig. 7B). Notably, relief of restriction by TRIM-Cyp was observed only when infection was carried out in the presence of cyclosporine (CsA), which is known to inhibit the interaction of TRIM-Cyp with the capsid of HIV-1 (Fig. 7B) (20, 42). Collectively, these data reinforce that the restriction of HIV-1 infection by both rhTRIM5α and owl monkey TRIM-Cyp is independent of the macroautophagy adaptors ATG5 and Beclin1.

FIG 6.

Depletion of autophagic mediators does not affect the restriction of HIV-1 by rhTRIM5α. (A) HeLa cells depleted of ATG5 by CRISPR-Cas9 genome editing were transduced to stably express HA-tagged rhTRIM5α (or left untransduced). Protein expression of ATG5 and rhTRIM5α in both untransduced and transduced wild-type and knockout cells was confirmed by Western blotting. (B) Wild-type or ATG5 knockout HeLa cells, either with or without exogenous rhTRIM5α expression (as depicted in panel A), were plated and infected with VSV-G-pseudotyped HIV-1. Cells were harvested 48 h after infection, and infectivity, signified by the percentage of GFP-positive cells, was measured by flow cytometry. The data shown here are representative of results from three independent experiments. (C) HeLa cells depleted of Beclin1 by CRISPR-Cas9 genome editing were transduced to stably express HA-tagged rhTRIM5α (or left untransduced). Protein expression of Beclin1 and rhTRIM5α in both untransduced and transduced wild-type and knockout cells was confirmed by Western blotting. (D) Infectivity of VSV-G-pseudotyped HIV-1 in wild-type or Beclin1 knockout HeLa cells, either with or without exogenous rhTRIM5α expression, was assayed as described above for panel B. The data shown here are representative of results from three independent experiments. (E) Wild-type or ATG5 knockout HeLa cells, either with or without exogenous rhTRIM5α expression, were infected with equal titers of VSV-G-pseudotyped HIV-1. Eighteen hours following infection, cells were harvested, and the abundance of viral DNA products was measured by quantitative PCR. For each sample, the amount of viral DNA, as measured by the number of GFP reporter copies detected, was normalized to the amount of GAPDH observed in parallel samples. Three independent experiments were conducted, and the amount of viral DNA detected in each experiment was normalized to the amount in the wild-type untransduced sample from that experiment. Error bars represent the standard deviations of the relative numbers of viral DNA products detected across three independent experiments. (F) Viral reverse transcription products were measured in wild-type or Beclin1 knockout HeLa cells, either with or without rhTRIM5α expression, following infection by VSV-G-pseudotyped HIV-1, as described above for panel E. Error bars represent the standard deviations of the relative numbers of viral DNA products detected across three independent experiments.

FIG 7.

Depletion of ATG5 does not affect the restriction of HIV-1 by owl monkey TRIM-Cyp. (A) HeLa cells depleted of ATG5 by CRISPR-Cas9 genome editing were transduced to stably express HA-tagged owl monkey TRIM-Cyp (or left untransduced). Protein expression of ATG5 and TRIM-Cyp in both untransduced and transduced wild-type and knockout cells was confirmed by Western blotting. (B) Wild-type or ATG5 knockout HeLa cells, either with or without exogenous owl monkey TRIM-Cyp expression (as depicted in panel A), were plated and infected with VSV-G-pseudotyped HIV-1 in either the absence or the presence of cyclosporine (− CsA and + CsA, respectively). Cells were harvested 48 h after infection, and infectivity, signified by the percentage of GFP-positive cells, was measured by flow cytometry. The data shown here are representative of results from three independent experiments.

We next examined the ability of rhTRIM5α to inhibit the formation of reverse transcription products in these cells. In untransduced ATG5 and Beclin1 knockout HeLa cells, reverse transcription was reduced relative to that in unmodified HeLa cells, consistent with the reduction in infectivity observed in these cells (Fig. 6E and F). In each case, however, potent restriction of reverse transcription was observed in cells expressing rhTRIM5α, compared to their untransduced counterparts. Taken together, these data indicate that rhTRIM5α does not require ATG5 or Beclin1 to complete restriction of HIV-1 infection and reverse transcription.

DISCUSSION

In this study, we observed that the degradation of rhTRIM5α and huTRIM5α is mediated, at least in part, by autophagic degradation pathways. We observed that YFP-rhTRIM5α degradation, in the absence of restriction-sensitive virus, was mediated almost exclusively by autophagic degradation, as MG132 had little effect on the steady-state expression of this protein (Fig. 1). This finding is consistent with data from previous reports (23, 25). We also observed that the inhibition of autophagy increased the expression of huTRIM5α, although in this case, we noted that protein accumulation was increased in the presence of both autophagic and proteasomal inhibitors (Fig. 2). This observation that TRIM5α is degraded by autophagy is consistent with results of a previous study by Mandell and colleagues (28).

Although our experiments support a role for autophagic degradation in TRIM5α turnover, we observed a substantial difference between rhTRIM5α and huTRIM5α, with rhTRIM5α turnover being insensitive to proteasome inhibitors (Fig. 1) and huTRIM5α being more sensitive to proteasome inhibitors than to inhibitors of autophagic degradation (Fig. 2). This difference in turnover between rhTRIM5α and huTRIM5α may reflect species-specific differences between the two proteins. However, it is important to note that our studies of huTRIM5α turnover were performed with proteins that did not include an N-terminal epitope tag. One possibility is that N-terminal tags alter the turnover of TRIM5α, potentially by perturbing the N-terminal acetylation observed in other studies (9, 43). Alternatively, N-terminal tags may alter the ubiquitination of TRIM5α by perturbing the dimerization of the RING domains (44). Either of these possibilities might shift the degradative fate toward autophagic degradation pathways. However, our studies of endogenously expressed huTRIM5α support the idea that endogenous huTRIM5α is degraded, at least in part, by autophagy (Fig. 2), and data from high-content image analyses similarly support the observation that human TRIM5α is degraded via autophagy (28).

We also examined the requirement of autophagic mediators in retroviral restriction by TRIM5α. We observed that retroviral restriction was not impacted by the depletion of autophagic mediators by siRNA. This was true in the case of endogenously expressed huTRIM5α, which still mediated potent inhibition of N-MLV following ATG5, Beclin1, and p62 knockdown. However, these studies do not exclude the possibility that small amounts of these mediators remaining after siRNA knockdown are sufficient to preserve TRIM5α-mediated restriction. We therefore used the CRISPR-Cas9 genome-editing system to deplete cells of ATG5 and Beclin1 and assess N-MLV restriction by huTRIM5α (Fig. 5). Similarly to our knockdown studies (Fig. 5B), no relief in retroviral restriction was observed (Fig. 5D and F) in cells depleted of ATG5 and Beclin1. In addition, restriction of viral reverse transcription was intact following ATG5 and Beclin1 knockout (Fig. 5G and H). Although we cannot discount the possibility that cells depleted of ATG5 or Beclin1 may possess alternative mechanisms of substrate degradation via autophagy, these results collectively demonstrate that restriction of retroviral infection and reverse transcription by TRIM5α is independent of ATG5 and Beclin1.

We obtained similar results when the restriction of HIV-1 by rhTRIM5α or TRIM-Cyp was examined. Although ATG5 depletion caused a decrease in HIV-1 infection (Fig. 6B), restriction of HIV-1 was not impacted by ATG5 or Beclin1 depletion (Fig. 6B and D). These data are in apparent contrast to the observations of Mandell and colleagues, who observed that the depletion of autophagic adaptor proteins abrogated rhTRIM5α-mediated restriction of HIV-1. One noticeable difference between these studies was that the study by Mandell et al. utilized primary rhesus macaque fibroblasts to examine the role of autophagic adaptors in the restriction mechanism of rhTRIM5α; they observed modest restriction of HIV-1 by primary rhesus macaque fibroblasts, with minimal relief of restriction being observed following the depletion of rhTRIM5α by siRNA (28). Our study dissected the role of autophagy in retroviral restriction in the context of much more potent restriction, as observed for the restriction of N-MLV by huTRIM5α and of HIV-1 by rhTRIM5α or TRIM-Cyp. Thus, we suspect that the differences between our results and those in the study by Mandell et al. stem from the more pronounced degree of restriction observed in our studies. However, we cannot exclude the possibility that cell type or species-specific differences explain the apparent discordance between these observations.