Human Galectin-9 Is a Potent Mediator of HIV Transcription and Reactivation

Abstract

Identifying host immune determinants governing HIV transcription, latency and infectivity in vivo is critical to developing an HIV cure. Based on our recent finding that the host factor p21 regulates HIV transcription during antiretroviral therapy (ART), and published data demonstrating that the human carbohydrate-binding immunomodulatory protein galectin-9 regulates p21, we hypothesized that galectin-9 modulates HIV transcription. We report that the administration of a recombinant, stable form of galectin-9 (rGal-9) potently reverses HIV latency in vitro in the J-Lat HIV latency model. Furthermore, rGal-9 reverses HIV latency ex vivo in primary CD4+ T cells from HIV-infected, ART-suppressed individuals (p = 0.002), more potently than vorinostat (p = 0.02). rGal-9 co-administration with the latency reversal agent "JQ1", a bromodomain inhibitor, exhibits synergistic activity (p<0.05). rGal-9 signals through N-linked oligosaccharides and O-linked hexasaccharides on the T cell surface, modulating the gene expression levels of key transcription initiation, promoter proximal-pausing, and chromatin remodeling factors that regulate HIV latency. Beyond latent viral reactivation, rGal-9 induces robust expression of the host antiviral deaminase APOBEC3G in vitro and ex vivo (FDR<0.006) and significantly reduces infectivity of progeny virus, decreasing the probability that the HIV reservoir will be replenished when latency is reversed therapeutically. Lastly, endogenous levels of soluble galectin-9 in the plasma of 72 HIV-infected ART-suppressed individuals were associated with levels of HIV RNA in CD4+ T cells (p<0.02) and with the quantity and binding avidity of circulating anti-HIV antibodies (p<0.009), suggesting a role of galectin-9 in regulating HIV transcription and viral production in vivo during therapy. Our data suggest that galectin-9 and the host glycosylation machinery should be explored as foundations for novel HIV cure strategies.

Fig 1. rGal-9 is a potent mediator of HIV transcription in vitro.

in vitro HIV reactivation in the J-Lat latency model (A) 5A8 clone, (B) 6.3 clone, and (C) 11.1 clone by varying doses of rGal-9 and other galectins (-1, -3, -4, -7, -8, and -9) after 24 hours of stimulation. αCD3 /αCD28 antibodies conjugated to beads, PMA/ionomycin (16 nM/500 nM), and TNFα (10 ng/ml) were used as positive controls. J-Lat cells were analyzed by flow cytometry to assess HIV-encoded GFP expression. Mean ± SEM is displayed, and statistical comparisons were performed using two-tailed unpaired t tests. * = p<0.05; ** = p<0.01, *** = p<0.001, and **** = p<0.0001.

Fig 2. rGal-9 is a potent mediator of HIV transcription ex vivo and synergizes with JQ1 in reactivating latent HIV.

(A) Treatment of CD4+ T cells isolated from ART-suppressed HIV-infected individuals with DMSO 0.5% (negative control), PMA/ionomycin (2 nM / 500 nM), vorinostat (1μM), or varying concentrations of rGal-9 (500 nM and 1000 nM) for 24 hours. Fold increase in cell-associated HIV RNA was determined relative to the corresponding DMSO-treated control for each individual time point. Mean ± SEM is displayed, and statistical comparisons between rGal-9 and other treatments were performed using two-tailed paired Wilcoxon signed-rank tests. (B-C) CD4+ T cells were isolated from PBMCs of three HIV-infected ART-suppressed individuals using negative selection. Resting CD4+ T cells were further enriched through depletion of cells expressing CD69, CD25, or HLA-DR surface markers from half of the isolated CD4+ T cells. The remaining half was processed through the exact enrichment procedure, except PBS was added instead of the depleting antibodies. Both cell populations were treated with 0.5% DMSO (negative control), 500 nM rGal-9, 1000 nM rGal-9 or αCD3/αCD28-conjugated beads. Induction of cell-associated HIV RNA was measured 24 hours post-treatment using RT-qPCR. Each individual is represented with a different symbol. Mean ± SEM is displayed, and statistical comparisons were performed using two-tailed paired t tests. Percentages reported reflect average values measured in the CD69- / CD25- / HLA-DR- CD4+ T cells with respect to values observed in total CD4+ T cells. (D) CD4+ T cells from HIV-infected ART-suppressed individuals were treated with 500 nM of rGal-9, 1 μM vorinostat, 40 nM romidepsin, 10 nM bryostatin, 300 nM prostratin, 1 μM JQ1, or 30 nM panobinostat alone or in combination with 500 nM of rGal-9 for 24 hours, and fold induction of cell-associated HIV RNA was determined using quantitative real-time PCR. * = p<0.05 compared with rGal-9 500 nM treatment alone. (E) The Bliss independence model was utilized for calculation of synergy for drug combinations. Δf_axy = 0 signifies a pure additive effect. Δf_axy>0 signifies synergy, while Δf_axy<0 signifies antagonism. Statistical significance was calculated using a two-tailed paired t-test comparing predicted and observed drug combination effects. * = p < 0.05.

Fig 3. rGal-9 induces HIV transcription and reactivation in a glycan-dependent manner.

(A) Effects of anti-Tim-3 antibody, anti-CD44 antibody, or anti-PDI antibody administration on rGal-9-mediated reactivation of HIV in J-Lat 5A8 cells. Antibodies were added 30 minutes prior to administration of 200 nM rGal-9. α-lactose (30 mM) was used as a positive control. (B, C) Treatment of J-Lat 5A8 cells with either 1 μg/ml tunicamycin, or with an enzymatic deglycosylation mix for 24 hours prior to rGal-9 stimulation. J-Lat cells were analyzed by flow cytometry to assess HIV-encoded GFP expression. Statistical comparisons were performed using two-tailed Mann-Whitney tests. (D) Effects of deglycosylation enzyme combinations on rGal-9-mediated HIV latency reversal in J-Lat 5A8 cells. N = PNGase F (Elizabethkingia miricola); O = O-Glycosidase (recombinant from Streptococcus pneumonia); S = α-(2→3,6,8,9)-Neuraminidase (recombinant from Arthrobacter ureafaciens); B = β(1→4)-Galactosidase (recombinant from Streptococcus pneumonia) + β-N-Acetylglucosaminidase (recombinant from Streptococcus pneumonia). Mean ± SEM is displayed, and statistical comparisons were performed using two-tailed unpaired t tests. * = p<0.05; ** = p<0.01, *** = p<0.001, and **** = p<0.0001.

Fig 4. rGal-9 modulates the expression of genes involved in several signaling pathways associated with HIV latency.

(A) Venn diagram showing the number of genes modulated by >2 fold with FDR<0.05 in sorted GFP-positive and GFP-negative cells containing reactivated (transcriptionally active) HIV proviruses, and latent (transcriptionally inactive) proviruses, respectively, after either rGal-9 treatment or αCD3/αCD28 stimulation. (B) Heat maps describing effects of rGal-9 treatment on host gene expression, organized by signaling pathways. All statistical comparisons were performed using t tests, and p values were adjusted for multiple comparisons using false discovery rate. Asterisks indicate >2-fold, statistically significant differences in gene expression between r-Gal9-treated, GFP+ cells and unstimulated control, as follows: * = FDR<0.05; ** = FDR<0.01, and *** = FDR<0.001.

Fig 5. rGal-9 partially activates primary CD4+ T cells and induces proliferation primarily in naïve CD4+ T cells.

(A, B) Effects of rGal-9 stimulation on the cell surface expression of CD69 and CD25 activation markers on CD4+ T cells isolated from six ART-suppressed individuals. Mean ± SEM is displayed. Asterisks represent statistically significant differences as compared to DMSO control (p < 0.05, two-tailed Wilcoxon signed-rank test). (C) Effects of rGal-9 stimulation on the proliferation of CD4+ T cells isolated from three ART-suppressed individuals. Primary CD4+ T cells were stained with CFSE and cultured for 5 days, stained with CD4 and CD45RA monoclonal antibodies, and proliferation was quantified as the percentage of CFSE_low cells on CD4+ CD45RA+ (Naïve, Tn) or CD4+ CD45RA- (Memory, Tm) T cells. Mean ± SEM is displayed. (D) Example of the flow cytometry gating strategy. (E, F) Effects of rGal-9 on proliferation of (E), memory CD4+ T cells, and (F), naïve CD4+ T cells.

Fig 6. rGal-9 induces the expression of the APOBEC3G anti-HIV host restriction factor in vitro.

(A–C) Digital droplet PCR gene expression profiling quantifying HIV gag, host APOBEC3G, p21, and RNAseP (housekeeping control) mRNA in J-Lat 5A8 cells sorted into GFP-positive and GFP-negative populations after either rGal-9 treatment, αCD3/αCD28 stimulation, or a combination of both. Mean ± SEM is displayed, and statistical comparisons against the unstimulated control were performed using two-tailed unpaired t tests. * = p<0.05; ** = p<0.01, *** = p<0.001, and **** = p<0.0001. (D-E) APOBEC3G protein expression in J-Lat 5A8 cells treated with varying concentrations of rGal-9 (100 nM, 200n M, and 500 nM) or interferon-α (5000 units/ml), as determined by western blot. Immunoblotting bands were quantified with ImageJ software. The quantified APOBEC3G protein expression levels were normalized to corresponding Tubulin protein levels to account for variation in loading.

Fig 7. rGal-9 induces the expression of several anti-HIV host restriction factors including APOBEC3G ex vivo.

(A) Heat map representing expression levels of host restriction factors in CD4+ T cells isolated from ART-suppressed individuals, after treatment with either 0.5% DMSO as negative control, 500 nM rGal-9, 1000 nM rGal-9, 1μM vorinostat, or a combination of PMA (2 nM) and Ionomycin (0.5 μM). Heat colors indicate fold modulation compared to the DMSO control. Red indicates induction of expression, and blue indicates reduction of expression. Statistical comparisons were performed using t tests, and p values were adjusted for multiple comparisons using false discovery rate. Asterisks indicate >3-fold, statistically significant modulation of gene expression as compared to DMSO control, as follows: * = FDR<0.05; ** = FDR<0.01, and *** = FDR<0.001. (B) APOBEC3G expression in isolated CD4+ T cells from HIV-infected ART-suppressed individuals, treated as described in panel A. Statistical comparisons were performed using two-tailed Wilcoxon signed-rank tests compared to the DMSO-treated control. (C-D) APOBEC3G protein expression in CD4+ T cells treated with either 500 nM rGal-9 or interferon-α (5000 units/ml), as determined by western blot. Immunoblotting bands were quantified with ImageJ software. The quantified APOBEC3G protein expression levels were normalized to corresponding Tubulin protein levels to account for variation in loading.

Fig 8. rGal-9 treatment reduces viral infectivity.

(A) Illustrative schematic of the viral infectivity experiment. (B-D) Effects of rGal-9 treatment of producer cells on HIV infectivity. The MOLT4-CCR5 cell line was infected for 6 hours, cells were washed and treated with either PBS, rGal-9 200nM, or interferon-α (5000 U/ml) for 24 hours, and cultures were incubated for 3 days. (B) HIV p24 levels produced by MOLT4-CCR5 cells were quantified after concentrating the culture supernatants. Concentrated culture supernatants were used to infect Jurkat cells by spinoculation. (C) Levels of integrated HIV DNA measured at days 3, 6, 9, and 12 post-infection of Jurkat cells. (D) Levels of integrated HIV DNA at days 3, 6, 9, and 12 post-infection of Jurkat cells, normalized to producer cell p24 supernatant levels. Mean ± SEM is displayed, and statistical comparisons were performed using two-tailed unpaired t test. * = p<0.05; ** = p<0.01, *** = p<0.001, and **** = p<0.0001.

Fig 9. sGal-9 levels correlate with measures of HIV transcription and viral production in vivo.

Correlations between levels of soluble Gal-9 and (A) levels of HIV cell-associated RNA, and (B) anti-HIV-1 antibodies in the plasma of 72 HIV-infected ART-suppressed individuals. Correlations were evaluated using Spearman's rank correlation coefficient tests.