HIV-1 Gag co-localizes with euchromatin histone marks at the nuclear periphery

Plasmid, cells, and inductions

HIV-1 Gag-mCerulean rtTA HeLa cells contain a doxycycline-inducible HIV-1 viral DNA stably integrated into a HeLa cell line containing rtTA, as previously described (13). Briefly, the original HIV-1 plasmid was constructed from the HIV-1 NL4.3 strain containing 24 copies of the MS2-MBL stem loops (47) (a kind gift from Dr. Nathan Sherer, University of Wisconsin-Madison). The plasmid was modified by deletion of 21 MS2-MBL loops, substitution of the HIV-1 LTR with a dox-inducible promoter at both the 5′ and 3′ ends of the provirus (48), excision of a 1.9-kb region of the pol gene using MscI sites, and replacement of the nef gene with rtTA. The resulting DNA was cloned into the Piggybac vector (System Biosciences, Palo Alto, CA, USA) containing a selective marker for puromycin drug resistance. A stable cell line was created using selection with puromycin (Cytiva HyClone, Malborough, MA, USA) and the correct sequence of the insertion confirmed by DNA sequencing of the cellular genomic DNA (data not shown). The HeLa cell line was maintained in Dulbecco’s modified Eagle medium containing 10% fetal bovine serum. Expression of the HIV-1 construct was induced by treating cells with doxycycline (GoldBio, St. Louis, MO, USA) at various concentrations, as explained in the Results. J-Lat T cells (clone 10.6, obtained through the NIH AIDS Reagent Program) (49), derived from Jurkat T cells, which express low levels of the CD4 receptor (50), were maintained in RPMI medium supplemented with 10% fetal bovine serum. Both cell lines were grown in the presence of penicillin, streptomycin, and amphotericin B at 38°C and 5% CO₂. Reversal of latent HIV-1 gene expression in J-Lat 10.6 cells was performed by treatment with prostratin (Sigma, St. Louis, MO, USA) or tumor necrosis factor alpha (TNF-α) (Abcam, Cambridge, MA, USA), as noted in the Results.

Subcellular fractionation

Fractionation of HIV-1 Gag-mCerulean rtTA HeLa cells was performed as described with minor modifications (13). Briefly, cells were grown in 10-cm² cell culture dishes (Genesee Scientific GenClone, San Diego, CA, USA), doxycycline induced, trypsinized, washed with cold 1× phosphate-buffered saline (PBS), and centrifuged at 16,000 × g at 4°C for 30 s to remove residual culture medium. Cells were resuspended in 400-µL cold cytoplasmic lysis buffer [10-mM HEPES, pH 7.9, 10-mM KCl, 0.1-mM EDTA, 0.4% (vol/vol) Nonidet P40, 1-mM dithiothreitol (DTT), Roche Protease Inhibitor cOmplete tablets EDTA free (MilliporeSigma, St. Louis, MO, USA) to 1×] and incubated on ice for 5 min, followed by centrifugation at 16,000 × g at 4°C for 5 min. Supernatants were collected as the cytoplasmic fraction, and pellets were resuspended in 200-µL cold nuclear lysis buffer (20-mM HEPES, pH 7.9, 100-mM NaCl, 1-mM DTT, Roche Protease Inhibitor cOmplete tablets EDTA free) in the presence of OmniCleave (Biosearch Technologies, Novato, CA, USA). Pellets were incubated in a 37°C water bath for 10 min, and cell suspensions were brought up to 400-mM NaCl and 1-mM EDTA. Pellets were vortexed for 15 s and rotated end-over-end at 4°C for 20 min. Samples were spun at 16,000 × g at 4°C for 10 min, and the supernatant was collected as the nuclear fraction.

Fractionation of J-Lat 10.6 cells was performed similarly as described above with minor modifications for the cytoplasmic lysis. Prior to lysis, J-Lat 10.6 cells were grown in T75 flasks (Corning Inc., Corning, NY, USA) and induced by prostratin or TNF-α treatment. Cells were resuspended in 1,000-µL cold cytoplasmic lysis buffer (10-mM HEPES, pH 7.9, 10-mM KCl, 2-mM Mg acetate, 3-mM CaCl₂, 340-mM sucrose, 1-mM DTT, Roche Protease Inhibitor cOmplete tablets EDTA to 1× and incubated on ice for 5 min. Nonidet P40 (0.05%) was added to samples and immediately vortexed for 15 s followed by centrifugation at 3,500 × g at 4°C for 10 min. Supernatants were collected as the cytoplasmic fraction. Nuclear fractions were extracted as described above.

Chromatin fractionation of HIV-1 Gag-mCerulean rtTA HeLa and J-Lat 10.6 cells were performed as described by Chase et al. (51). Following induction and harvest, cells were resuspended in 1,000-µL cold sucrose buffer containing 10-mM HEPES, pH 7.9, 10-mM KCl, 2-mM Mg acetate, 3-mM CaCl₂, 340-mM sucrose, 1-mM DTT, and Roche Protease Inhibitor cOmplete tablets EDTA to 1× and incubated on ice for 5 min, followed by centrifugation at 3,500 × g at 4°C for 10 min. The supernatant was collected as the cytoplasmic fraction. Cell pellets were then resuspended in 1,000-µL cold nucleoplasm extraction buffer [50-mM HEPES, pH 7.9, 150-mM K acetate, 1.5-mM MgCl₂, 0.1% (vol/vol) Nonidet P40, 1-mM DTT, Roche Protease Inhibitor cOmplete tablets EDTA free] and homogenized using Dounce homogenizers on ice. Samples were transferred into fresh tubes and rotated end-over-end at 4°C for 20 min. The supernatant was collected as the nucleoplasm fraction after centrifuging samples at 16,000 × g at 4°C for 10 min. Remaining cell pellets were resuspended in 500-µL cold nuclease incubation buffer [50-mM HEPES, pH 7.9, 10-mM NaCl, 1.5-mM MgCl₂, 1-mM DTT, 100-U/mL Omnicleave (Biosearch Technologies), Roche Protease Inhibitor cOmplete tablets EDTA free] and incubated in a 37°C water bath for 10 min. The buffer in the cell suspension was adjusted to 150-mM NaCl and incubated on ice for 20 min. Samples were centrifuged at 16,000 × g at 4°C for 10 min, and the supernatant was collected as the euchromatin fraction. Cell pellets were then resuspended in 500-µL cold chromatin extraction buffer (50-mM HEPES 7.9, 500-mM NaCl, 1.5-mM MgCl₂, 0.1% Triton X-100, 1-mM DTT, Roche Protease Inhibitor cOmplete tablets EDTA free) and incubated on ice for 20 min. Samples were centrifuged at 16,000 × g at 4°C for 10 min, and the supernatant was collected as the heterochromatin fraction. Chromatin fractionations in experiments involving treatment with 18.5-nM romidepsin (Sigma) were performed as described above with drug treatment prior to fractionations.

Western blot analysis

After cell fractionation, the protein concentration of each sample was measured using the Bradford Protein Assay protein with Coomassie protein assay reagent (Thermo Fisher Scientific-Invitrogen, Waltham, MA, USA). Next, 5 µg of each protein fraction was added to 1× loading buffer (250-mM Tris, pH 6.8, 40% glycerol, 8% β-mercapto-ethanol, 0.4% bromophenol blue, 8% SDS) and subjected to immunoblot analysis using the Stain-Free Imaging Technology (BioRad, Hercules, CA, USA). After SDS-PAGE, gels were transferred to polyvinylidene fluoride membrane and blocked in 5% milk 1× Tris-buffered saline with 0.1% Tween-20 for 30 min. Gag was detected using purified mouse anti-p24 supernatant from anti-HIV-1 Gag hybridoma 183 [National Institutes of Health (NIH) AIDS Reagent Program, Division of AIDS, NIAID, NIH, Germantown, MD, USA] at 1:2,000 dilution. Antibodies targeting calnexin (1:3,000; Enzo, cat #ADI-SPA-865-F), mediator complex subunit 4 (Med4) (1:5,000; Abcam, cat #Ab129170), H3K4me3 (1:5,000; Abcam, cat #Ab8580), H3K27ac (1:2,500; Abcam, cat #Ab4729), H3K9me3 (1:1,000; Active Motif, cat #39065), and total H3 (1:10,000; ProteinTech, cat #68345-1-Ig) were used to assess fraction purities. Mouse (1:5,000) and rabbit (1:10,000) secondary horseradish peroxidase-conjugated antibodies (Jackson ImmunoResearch, West Grove, PA, USA; mouse cat #711-036-152, rabbit cat #715-036-150) were used with Pierce ECL Western blotting substrate (Thermo Scientific, cat #32106) and visualized by a ChemiDox MP Imager (BioRad). Densitometry analysis was performed using ImageLab (BioRad) with adjusted total band volume data for each lane extracted for analysis. To account for the differences in the volume of nuclear versus cytoplasmic compartments, the adjusted total band volume derived from the signal representing 5 µg of protein loaded in each lane of the Western blot was normalized to represent 1% of the total protein of each fraction. The adjusted cytoplasmic and nuclear band volumes were used to calculate the percentage of cytoplasmic or nuclear signal by dividing each fraction by the sum of the total band volume from both fractions. Three independent biological replicates were performed for each experiment. The mean percentage ± standard error of the mean (SEM) of nuclear Gag and the mean percentage of chromatin marks ± SEM in the nuclear fractions in the three replicates were statistically analyzed using two- and one-way analyses of variance (ANOVAs), respectively, with statistical significance defined as P < 0.5.

Single-molecule fluorescence in situ hybridization

HIV-1 Gag-mCerulean rtTA HeLa cells were induced with doxycycline (2 µg/mL) for 24 hours, washed with PBS and fixed with 3.7% formaldehyde in 1× PBS for 10 min. Single-molecule fluorescence in situ hybridization (smFISH) was performed using Stellaris RNA FISH protocol for adherent cells (Biosearch Technologies). The 48 RNA FISH probes targeting the gag gene were labeled with Quasar570 to specifically visualize USvRNA at the level of single-molecule detection, as described previously (13, 52). After fixation, cells were washed twice with 1× PBS and permeabilized with 70% EtOH overnight at 4°C, then rehydrated in wash buffer (10% formamide, 2× saline-sodium phosphate-EDTA [SSPE], diethyl pyrocarbonate [DEPC] water) for 20 min at room temperature. Coverslips with cells were inverted onto RNA smFISH probes diluted 1:50 in hybridization buffer (10% dextran, 10% formamide, 2× subsp., DEPC water) in a humid chamber overnight at 37°C. Coverslips were then washed with wash buffer for 30 min at 37°C, followed by 4′,6-diamidino-2-phenylindole (DAPI) staining (1:1,000 in wash buffer) for 30 min at 37°C. Coverslips were mounted onto microscope slides using ProLong Diamond Antifade Mountant (Thermo Fisher Scientific-Invitrogen) and cured for 72 hours.

Immunofluorescence of histone marks

HIV-1 Gag-mCerulean rtTA HeLa cells were briefly washed with PBS and fixed using 3.7% formaldehyde for 10 min at room temperature. Following fixation, cells were washed three times for 5 min with PBS and permeabilized with permeabilization buffer (0.12-M KCl, 0.02-M NaCl, 0.01-M Tris-HCl, pH 7.7, 0.1% Triton X-100) for 10 min. Cells were washed again three times with the same solution and blocked with 3% bovine serum albumin in PBS for 10 min. Primary antibodies against H3K4me3 (1:600; Abcam, cat #Ab8580), H3K27ac (1:600; Abcam, cat #Ab4729), and H3K9me3 (1:600; Active Motif, cat #39766) were diluted in blocking buffer and applied to an inverted coverslip containing cells for 45 min at room temperature. Cells were then washed three times using blocking buffer for 5 min each, incubated with anti-rabbit AlexaFluor 647 secondary antibody (1:1,000; Life Technologies, cat #A31573) for 45 min at room temperature, washed three times with PBS, and mounted onto microscope slides using ProLong Diamond Antifade Mountant (Thermo Fisher Scientific-Invitrogen).

Confocal microscopy

Cells on coverslips were imaged using a Leica SP8 laser scanning confocal microscope (Leica Microsystems, Buffalo Grove, IL, USA). Imaging was performed using a white light laser and a ×63/NA1.4 oil immersion objective lens with sequential scanning of each channel with a frame average of 4 at 400-Hz imaging speed using hybrid detectors. Gag-mCerulean was excited at 458-nm laser line, and fluorescent emission was detected from 463 to 520 nm. USvRNA detected by smFISH probes labeled by Quasar570 was excited using 548-nm laser, and emission was captured from 553 to 635 nm. All histone marks were excited by 647-nm laser and captured from 652 to 751 nm. Single fluorophore controls were imaged separately to ensure no cross talk occurred. Images for the HIV-1 Gag time course experiments were processed by Gaussian filters, and histograms were adjusted to best visualize nuclear foci. Images of Gag and histone mark co-localization were processed by deconvolution using Huygens Image Processing Software (Scientific Volume Imaging). The 3D surface renderings of Z-stacks, cross-sections, and co-localization channels were generated in Imaris v.10.0 image analysis software (Bitplane).

For co-localization experiments presented in Movies S1 to S3, surface renderings were rotated 360° along the Y-plane followed by a scan through of the Z-stack using the orthogonal clipping plane function to allow visualization of co-localized complexes within the nucleus. Movies created for experiments involving the visualization of histone marks (Movies S3 to S5) were made by first rotating 3D surface renderings 360° along the Y-plane and then rotating 90° along the X-plane. Cells were zoomed in to visualize the co-localized Gag:histone spots (white).

Quantitative image analysis

To measure the intensity of Gag localization in cytoplasmic and nuclear compartments for the time course experiments, fluorescent intensity analysis was performed using ImageJ (53). Images were selected in Leica LAS X by isolating a single Z plane through the center of the cell nucleus. Nuclei were traced in ImageJ and the fluorescent intensity was measured. Background subtraction was performed by sampling a small region outside the cell and averaged with all cell replicates to obtain the corrected total cell fluorescence.

To determine the percent of Gag and USvRNA co-localization, Imaris image analysis software (Bitplane) was used to calculate Mander’s coefficient based on the HIV-1 Gag and USvRNA signal from images. The nuclear signal for each channel was isolated by creating a mask of the nucleus generated from surface rendering of the DAPI stain. Co-localization analysis between Gag and USvRNA was performed using the co-localization function to generate a co-localization channel and to obtain Mander’s coefficient. The same methods were employed to determine the co-localization of HIV-1 Gag with various histone marks. To measure the distances of co-localized spots from the chromatin periphery, a surface rendering of the histone mark fluorescence was made, and the spot function was used to generate spots from the Gag:histone co-localization channel. The distance from Gag:histone co-localized spots to the periphery of the surface was measured. Statistical tests were used to determine significance, defined as P < 0.05, as indicated in the figure legends.

HIV-1 Gag nuclear trafficking and co-localization with USvRNA occurred at early time points after induction of HIV-1 gene expression

To examine the temporospatial dynamics of HIV-1 Gag nuclear localization, we constructed a HeLa cell line containing an integrated, HIV-1 genome expressing HIV-1 Gag-mCerulean controlled by a doxycycline promoter, which allowed synchronization of gene expression (Fig. 1A). In this construct, the insertion of the mCerulean gene in-frame at the 3′ end of the Gag sequence resulted in the elimination of viral protease activity, allowing the study of full-length Gag trafficking without the presence of free CA, which would have been a confounding factor in these experiments. We performed a time course using the HIV-1 Gag-mCerulean rtTA HeLa cell line by inducing HIV-1 gene expression with doxycycline (2 µg/mL). Subcellular fractionation was performed every 4 hours for 24 hours to measure the amount of Gag protein present in cytoplasmic and nuclear fractions (Fig. 1B, n = 3). Gag expression was detected by Western blot using an anti-p24 antibody targeting the CA domain of Gag, and quantification of band intensities was measured by densitometry analysis (Fig. 1C). There was no detectable Gag expression in untreated cells (Fig. 1B), so no further analysis of these samples was performed. Purity of the nuclear and cytoplasmic fractions was assessed using antibodies against calnexin (a cytoplasmic protein) and Med4 (a nuclear-localized protein), and the immunoblots consistently did not detect contamination of the nuclear or cytoplasmic fractions (Fig. 1B).

Our results indicated that HIV-1 Gag was present in cytoplasmic and nuclear fractions as early as 4 hours post-induction (h.p.i.; Fig. 1B and C), and expression in each fraction increased through 20 hours, reaching a plateau at 24 hours. At each time point, Gag levels were statistically significantly higher in the cytoplasmic fractions compared to the nuclear fractions (Fig. 1D), with a range of approximately 30%–37% of Gag in the nuclear fraction. The percentage of Gag in the nucleus compared to that in the cytoplasm remained relatively constant throughout the 24-hour period (Fig. 1D), indicating that the nuclear population of Gag was either stable or replenished after it was depleted. The presence of full-length Gag in the nucleus at early time points (4–8 hours) and low levels of protein expression indicated that Gag nuclear trafficking occurred shortly after Gag production began and suggested that nuclear entry was not driven primarily by a concentration gradient.

To examine HIV-1 Gag-mCerulean subcellular localization using an imaging approach, cells at several time points (8, 12, 16, 20, and 24 h.p.i.) were imaged through the center of the nuclear plane using confocal microscopy. Beginning at 8 h.p.i, both focal and diffuse Gag-mCerulean fluorescence (green) were observed within the nucleus (nuclear DNA stained blue by DAPI; Fig. 2A), in the cytoplasm, and at the plasma membrane. To detect HIV-1 USvRNA, FISH was performed using a set of oligonucleotide probes targeting the gag gene, as we previously described (13). At each time point, Gag and USvRNA co-localization was assessed using Imaris, with distinct co-localized foci appearing as yellow in the nucleus, cytoplasm, and at the plasma membrane (Merged image panels in Fig. 2A). The 3D cross-sections of Z-stacks confirmed Gag:USvRNA co-localization in the nucleus in the X, Y, and Z planes, indicated by arrows (white, co-localization channel). Additionally, a surface rendering of the nucleus and Gag-USvRNA spots based on the white co-localization channel was created to confirm co-localization was occurring inside the nucleus as defined by the DAPI signal (8 and 24 h.p.i., shown in Fig. 2B). Analysis of 3D renderings of Z-stacks from individual cells imaged at the 8- and 24-h.p.i. time points demonstrated Gag-USvRNA co-localization in the nucleus at numerous individual foci (see also Video S1 and Video S2, respectively).

To quantitate nuclear Gag levels obtained by confocal imaging, nuclear Gag fluorescence was measured from a single Z-plane through the center of the nucleus (n = 30–41 cells per data point across three independent biological replicates). Comparison of Gag levels in each subcellular location yielded similar trends as observed in Fig. 1, with more Gag in the cytoplasm compared to the nucleus. In contrast to the Western blot analysis, the percentage of nuclear Gag was observed highest at 8 h.p.i. (16.5% ± 0.9%) and decreased to 11.6% ± 0.4% by 12 h.p.i., where it remained relatively constant throughout the remaining time points (Fig. 2C). It should be noted that there was a lower nuclear-to-cytoplasmic ratio of Gag detected using fluorescence imaging compared to Western blotting. This discrepancy may be attributed to several factors. First, the number of cells analyzed was different for each method; the fractionation and Western blot analysis were performed using a large population of cells (n = 9 × 10⁶), whereas imaging was performed on a single-cell basis and represented a smaller number (n = 30–41 cells). Second, because the HIV-1 Gag-mCerulean rtTA HeLa is a polyclonal cell line with varying numbers of integrated proviral constructs, viral transcription and Gag expression levels varied from cell to cell even though all cells were treated with equal concentrations of doxycycline. Another factor is the difference in sensitivity of immunoblotting, with the ability to detect Gag by chemiluminescence being more efficient than visualization of individual molecules of Gag-mCerulean in single cells, where single fluorescence molecules or faint Gag foci below the threshold of detection would not be measured. Finally, it is important to note that the imaging analysis was performed by analyzing a single Z-slice through the center of the nucleus, which did not include Gag molecules above or below the plane that were nonetheless located within the 3D volume of the nucleus. Bearing in mind the numerous explanations for discrepancies in the absolute amounts, the overall trend of nuclear Gag localization was similar for both methods.

Additionally, single-cell images obtained at each time point containing HIV-1 Gag and USvRNA (n = 9–26 cells) were analyzed in 3D using the Imaris signal-based co-localization function to measure the overall fluorescence intensity of the Gag and USvRNA signals and to obtain Mander’s coefficient for each channel. At 8 h.p.i., we observed that nuclear Gag co-localized with USvRNA (11.9% ± 2.3%, Fig. 2D), and nuclear USvRNA co-localized with HIV-1 Gag (13.5% ± 3.9%, Fig. 2E) to similar degrees. At 12 h.p.i., the percentage of nuclear USvRNA that was co-localized with Gag increased (26.5% ± 2.1%) and then plateaued, followed by a slight decrease in co-localization at the 24-hour time point (22.6% ± 7.0%, Fig. 2E). The percentage of Gag co-localized with USvRNA increased after 8 h.p.i. and remained relatively constant through 16 h.p.i. (25.5% ± 2.8%, Fig. 2D). The overall percentage of HIV-1 Gag co-localized with USvRNA was similar to that previously reported by Tuffy et al. (13), although a different analytical method (object-based co-localization) was used in that study.

HIV-1 Gag preferentially localized to the euchromatin-associated protein fraction

Our previous results suggest that a population of full-length Gag localizes to the proviral transcription site in HeLa and T cells (13), indicating that Gag may be associated with transcriptionally active euchromatin. To test this idea, we examined whether Gag localized to specific chromatin-associated regions using a differential salt fractionation method (51). To begin, we examined whether the level of HIV-1 Gag expression affected its localization to specific subsets of chromatin. HIV-1 Gag-mCerulean rtTA HeLa cells were treated with varying concentrations of doxycycline (2–2,000 ng/mL) to regulate the level of Gag expression. After isolation of nuclei, extraction of euchromatin and heterochromatin-associated proteins was achieved using buffers containing either 150-nM NaCl or 500-nM NaCl, respectively, and Western blotting was performed to measure the level of Gag present in the euchromatin (Ch150) and heterochromatin (Ch500) protein fractions (Fig. 3A; n = 3). Each fraction was also analyzed using antibodies against calnexin (cytoplasm), Med4 (nuclear), tri-methylated H3K4 (H3K4me3, euchromatin), H3K9 (H3K9me3, heterochromatin), and total H3 histone (chromatin). H3K4me3 marks are associated with active promoters, and H3K9me3 marks are located at constitutive heterochromatin regions (54–57). As reported by Chase et al., chromatin fractions extracted using low salt buffer contained proteins involved in active transcription that were distinct from the high salt chromatin fraction, despite the presence of H3K4me3 in both chromatin fractions (51). Our results, consistent with Chase et al., indicated that the H3 histone signal was primarily in the Ch500 fraction (51). Interestingly, we found that HIV-1 Gag signal was barely detectable at 2 ng/mL and then increased in a dose-dependent fashion starting at 20 ng/mL with maximum reached at 200 mg/mL with no further increase at 2,000 ng/mL, suggesting saturation of the doxycycline-induction system (Fig. 3A and S1A), consistent with previous reports that this promoter reached maximal HIV-1 p24 release into cell supernatants at a dose of 500-ng/mL doxycycline (48). Densitometry analysis of Western blots revealed that HIV-1 Gag preferentially localized with euchromatin (69.1% ± 6.3%) compared to heterochromatin (30.9% ± 6.3%, P = 0.032) at lower levels of Gag expression (20-ng/mL doxycycline;) (Fig. 3B). Similar results were obtained using 200-ng/mL doxycycline induction, with 58.1% ± 0.9% of Gag in the euchromatin and 41.9% ± 0.9% in heterochromatin fraction (P = 0.001). At 2,000 ng/mL, the highest doxycycline induction level used, the ratio of Gag in euchromatin versus heterochromatin fractions remained significantly higher but moved closer to being equivalent in each fraction (53.4% ± 0.03% and 46.6% ± 0.03%, respectively). Three different concentrations of the chromatin lysates were loaded onto the gel, and the Gag signal was detected using immunoblotting, indicating that the Gag proteins were within the linear range of detection (Fig. S1E). To determine whether treatment with doxycycline altered the distribution of chromatin proteins, H3K4me3, H3K9me3, and total H3 histone ratios for euchromatin and heterochromatin were assessed and found to be unchanged across all drug concentrations (Fig. 3C; Fig. S1).

We next asked whether opening the chromatin structure to adopt a more transcriptionally-active state would affect HIV-1 Gag localization. To promote euchromatin formation, cells were treated with 18.5 nM (10 ng/mL) of histone deacetylase inhibitor (HDACi) romidepsin (RMD), which effectively reactivates latently infected CD4+ T cells, resulting in a 10-fold increase in HIV-1 virion production (58). Based on the prior experiment, a dose of 20-ng/mL doxycycline was initially used to induce Gag expression for 24 hours (Fig. 3D). Treatment with doxycycline alone resulted in higher levels of Gag in euchromatin (65.1% ± 4.8%) compared to heterochromatin (34.9% ± 4.8%, P = 0.013), similar to that shown in Fig. 3B (Fig. 3E). Of note, cells treated with RMD alone (without doxycycline) expressed slightly higher levels of Gag than the dimethyl sulfoxide control, with more Gag in the euchromatin (68.8% ± 6.2%) compared to the heterochromatin fractions (31.2% ± 6.2%, P = 0.015; Fig. 3E). This result suggests that changing the global chromatin environment with RMD treatment promoted HIV-1 gene expression and enhanced Gag localization to the transcriptionally active chromatin fraction, independently of the doxycycline-sensitive promoter. Cells treated with the combination of doxycycline and RMD displayed increased Gag expression by 4.2-fold compared to doxycycline treatment alone, with equivalent levels of Gag in the chromatin fractions (Fig. 3D and E), suggesting that the movement of Gag into the euchromatin fraction was saturable. To assess the effectiveness of the HDACi drug treatment on facilitating euchromatin formation, chromatin fractions were assessed by Western blotting using antibodies against H3K4me3 (associated with active promoters), acetylated H3K27 (H3K27ac, associated with active enhancer regions) (59), H3K9me3 (heterochromatin), and total H3 histone (Fig. S2). Cells treated with RMD exhibited increased H3K4me3 and H3K27ac histone marks, signifying a shift toward euchromatin formation, as expected (Fig. 3F; Fig. S2B). H3K9me3 histone marks were found primarily in the heterochromatin fractions (Fig. S2C). The total H3 histone antibody detected a higher proportion of H3-associated marks in the euchromatin in the presence of RMD, indicating that a subset of H3 was modified and moved into the transcriptionally active chromatin region, consistent with the efficacy of the HDACi to alter the global chromatin landscape (Fig. S2D). Calnexin antibody was used to demonstrate that chromatin fractions did not contain cytoplasmic contamination. Taken together, these results indicate that Gag preferentially localized to euchromatin, most prominently at low to moderate levels of Gag expression, suggesting that the factor that allows Gag to accumulate in euchromatin may be limiting in the cell.

HIV-1 Gag nuclear trafficking in infected T cells reactivated from latency

Based on the results observed in HeLa cells, we examined the timing and subcellular distribution of Gag in infected T cells reactivated from latency using a model system more representative of natural infection. We previously showed that HIV-1 Gag traffics to the nucleus in J-Lat 10.6 cells treated with the latency-reversal agent prostratin (13), a protein kinase C (PKC) agonist that acts by stimulating PKC to activate the NF-κB signaling pathway (60). To determine the timing of Gag expression and nuclear trafficking in T cells, we performed a 24-hour time course following latency reversal. Cells were treated with prostratin (1 µg/mL), and every 4 hours, subcellular fractionation and Western blotting (n = 3) were performed. Gag was detected in both cytoplasmic and nuclear fractions in J-Lat 10.6 cells starting at 4 h.p.i. (Fig. 4A and B). Although not statistically significant, the amount of Gag in the nucleus was higher (58.5% ± 17.9%) compared to that in the cytoplasm (41.5% ± 17.9%), whereas all other time points exhibited slightly higher cytoplasmic levels of Gag (Fig. 4C). The nuclear Gag concentrations decreased over the 24-hour period, similar to the HeLa cell model, with the exception of the last time point. These subtle differences in the timing of Gag localization between the HeLa and J-Lat 10.6 cell line may be attributed to the cell type or the presence of Gag-Pol, which is absent in the HIV-1 Gag-mCerulean rtTA HeLa cells. Interestingly, Gag-Pol was noted to be present in the nuclear fraction beginning at 12 h.p.i. (Fig. 4A), which has not been previously reported and warrants additional study.

HIV-1 Gag preferentially localized to euchromatin in latently infected T cells treated with latency-reversal agents

Previously, we reported that Gag and USvRNA co-localize at proviral transcriptional burst sites in reactivated J-Lat 10.6 cells, suggesting that Gag traffics to sites of USvRNA transcription upon reactivation of viral RNA expression after latency reversal (13). To determine whether Gag was associated with the euchromatin-associated protein fraction under these conditions, we assessed the chromatin localization of Gag after reactivation of HIV-1 gene expression in J-Lat 10.6 cells using varying concentrations of prostratin (Fig. 5A, n = 3) and TNF-α (Fig. 5D, n = 3). In chronically infected T cells, HIV-1 relies on activation of the tumor necrosis factor receptor by TNF-α to promote NK-κB recruitment to the HIV-1 LTR promoter (61–63). Thus, the use of TNF-α mimics natural reactivation of latently infected T cells. Densitometry analysis of Western blots revealed that Gag expression increases with increasing prostratin and TNF-α concentrations (Fig. S3A and F, respectively) and preferentially localized with euchromatin, particularly at lower expression levels (Fig. 5B and E, respectively). In chromatin fractions of J-Lat 10.6 cells reactivated by 0.25 µg/mL of prostratin, we found that 57.0% ± 1.3% of Gag localized to euchromatin and only 43.0% ± 1.3% localized to heterochromatin (Fig. 5B, P = 0.011). The euchromatin:heterochromatin ratio of Gag was equivalent at higher levels of Gag expression. Interestingly, J-Lat 10.6 cells reactivated with 2.50-ng/mL TNF-α exhibited much higher Gag localization to the euchromatin (61.3% ± 4.3%) compared to heterochromatin (38.7% ± 4.3%, P = 0.002; Fig. 5E). The preferential localization of Gag to euchromatin over heterochromatin was observed in cells treated at all concentrations of TNF-α, ranging from 2.5 to 10 ng/mL. Treatment of cells with prostratin or TNF-α did not alter euchromatin mark expression, as shown by the stable levels of H3K4me3 in the Ch150 fractions (Fig. 5C and F, respectively), indicating that the change in Gag localization was specific. Furthermore, measurement of H3K4me3, H3K9me3, and total H3 histone levels in chromatin fractions did not exhibit significant changes from prostratin or TNF-α treatment (Fig. S3). Quantitative assessment of the chromatin lysates immunoblots was performed across three concentrations, affirming that the Gag bands measured were in the linear range (Fig. S3E and J).

The upregulation of HIV-1 gene expression by HDACis in latently infected T cells has been extensively characterized (27, 35, 58, 64, 65); however, we are not aware of any studies that have examined whether HDACi affects Gag subnuclear localization upon latency reversal. To determine how HDACi impacts Gag localization in reactivated T cells, we treated J-Lat 10.6 cells with 18.5-nM RMD (10 ng/mL) combined with either prostratin or TNF-α at concentrations of 0.25 or 2.5 ng/mL, respectively. Chromatin fractions were prepared as previously described, and immunoblotting using anti-p24 antibody was performed to measure Gag levels in each fraction (Fig. 5G, n = 4). Purity of each fraction was assessed using anti-calnexin (cytoplasm), anti-Med4 (nucleus), anti-H3K4me3, and anti-H3K27ac (euchromatin), anti-H3K9me3 (heterochromatin), and anti-H3 (chromatin) antibodies. Similar to the findings shown in Fig. 5A and D, Lat cells treated with prostratin or TNF-α exhibited higher levels of Gag in euchromatin (Fig. 5H; Fig. S4A; P = 0.01 and 0.0002, respectively). Cells treated with RMD also showed a preference for Gag euchromatin localization, although the increase was not statistically significant (54.1% ± 3.4% euchromatin and 46.0% ± 3.4% heterochromatin, P = 0.33). H3K27ac levels were slightly increased in the presence of RMD, verifying the increased formation of euchromatin with drug treatment (Fig. 5I and S4B). The H3K9me3 heterochromatin mark and total H3 expression also increased in the presence of RMD (Fig. S4C and D), suggesting that RMD treatment changes the global chromatin landscape to promote euchromatin formation. However, latency reversal using prostratin or TNF-α plus RMD treatment resulted in an equivalent ratio of Gag in the euchromatin and heterochromatin fractions, consistent with results seen at high levels of Gag induction in HeLa cells (Fig. 3). Altogether, these results indicate that HIV-1 Gag preferentially localizes to the euchromatin under lower expression levels following reactivation, suggesting that localization to transcriptionally active chromatin is a property of HIV-1 Gag in T cells as well as HeLa cells, and in both cases, it is a saturable process.

HIV-1 Gag co-localization with transcriptionally active versus repressive histone marks

To determine whether we could visualize HIV-1 Gag co-localization with specific chromatin modifications, we performed immunofluorescence to detect H3K4me3, H3K27ac, and H3K9me3 histone marks (red, Fig. 6). Following 24 hours of doxycycline induction of the HIV-1 Gag-mCerulean rtTA HeLa cells, we observed Gag co-localization with each of the three histone marks independently, as indicated by the co-localization channel shown in white (Fig. 6A; n = 32–34 cells over three independent replicates). Imaris co-localization analysis revealed that a higher percentage of nuclear Gag co-localized with the euchromatin marks H3K4me3 and H3K27ac (73.5% ± 0.03% and 87.3% ± 0.02%, respectively) compared to the heterochromatin mark H3K9me3 (25.9% ± 0.02% with P < 0.0001; Fig. 6B), supporting our previous observation from the biochemical chromatin fractionation assays (Fig. 3). Interestingly, Gag was found to co-localize significantly more with H3K27ac histone marks compared to H3K4me3 histone marks (Fig. 6), suggesting Gag may have a specific preference for active enhancer regions.

Next, we examined the localization of HIV-1 Gag in 3D chromatin space. To address this question, we created a surface rendering of the chromatin space (red) and used the Imaris spot function to determine the position of co-localization of Gag with the chromatin marks. Over 800 co-localized spots consisting of Gag and chromatin marks spots were generated and used for analysis. As shown in zoomed-in panels in Fig. 6A, Gag co-localization with H3K4me3 (n = 807) and H3K27ac (n = 1463) preferentially occurred near the periphery of the nucleus, in some cases extending beyond the outer limits of the chromatin space (Video S3 and Video S4, respectively). Conversely, Gag and H3K9me3 (n = 882) co-localized foci were more centrally located and appeared largely buried in the 3D rendering, with far fewer spots protruding from the chromatin mask (Video S5).

Additionally, we measured the distance between Gag-histone co-localized spots and the periphery of the chromatin space, defined by the fluorescence signal of the histone channel in 3D space. Interestingly, we found that Gag co-localized with the euchromatin marks H3K4me3 and H3K27ac more toward with the nuclear periphery when compared to the heterochromatin mark. Gag co-localized with H3K27ac marks at a mean distance of 0.12 µm ± 0.002 µm from the nuclear periphery and co-localized with H3K4me3 at an average of 0.10 µm ± 0.002 µm from the outer edge of the chromatin border (P < 0.0001, Fig. 6C). Strikingly, co-localization of Gag with the heterochromatin histone mark H3K9me3 occurred significantly farther away from the outer border of the chromatin (at a mean distance of 0.23 µm ± 0.002 µm, P < 0.0001; Fig. 6C) compared to the euchromatin marks, indicating that HIV-1 Gag more closely associated with the transcriptionally active marks associated with promoters and enhancers near the periphery of the chromatin space compared to the heterochromatin region.