The Effect of JAK1/2 Inhibitors on HIV Reservoir Using Primary Lymphoid Cell Model of HIV Latency

Abstract

HIV eradication is hindered by the existence of latent HIV reservoirs in CD4⁺ T cells. Therapeutic strategies targeting latent cells are required to achieve a functional cure, however the study of latently infected cells from HIV infected persons is extremely challenging due to the lack of biomarkers that uniquely characterize them. In this study, the dual reporter virus HIV_GKO was used to investigate latency establishment and maintenance in lymphoid-derived CD4⁺ T cells. Single cell technologies to evaluate protein expression, host gene expression, and HIV transcript expression were integrated to identify and analyze latently infected cells. FDA-approved, JAK1/2 inhibitors were tested in this system as a potential therapeutic strategy to target the latent reservoir. Latent and productively infected tonsillar CD4⁺ T cells displayed similar activation profiles as measured by expression of CD69, CD25, and HLADR, however latent cells showed higher CXCR5 expression 3 days post-infection. Single cell analysis revealed a small set of genes, including HIST1-related genes and the inflammatory cytokine, IL32, that were upregulated in latent compared to uninfected and productively infected cells suggesting a role for these molecular pathways in persistent HIV infection. In vitro treatment of HIV-infected CD4⁺ T cells with physiological concentrations of JAK1/2 inhibitors, ruxolitinib and baricitinib, used in clinical settings to target inflammation, reduced latent and productive infection events when added 24 hr after infection and blocked HIV reactivation from latent cells. Our methods using an established model of HIV latency and lymphoid-derived cells shed light on the biology of latency in a crucial anatomical site for HIV persistence and provides key insights about repurposing baricitinib or ruxolitinib to target the HIV reservoir.

Keywords: HIV; JAK-STAT signaling pathway; LRA (latency reversing agent); latency; scRNAseq; tonsil.

Figure 1

HIV_GKO infection in tonsil derived CD4⁺ T cells. (A) HIV_GKO viral construct showing mutated env gene, nef gene replaced by an eGFP reporter, and insertion of EF1a promoter directly upstream of mKO2 reporter. (B) Schematic showing experimental design for activation and infection with HIV_GKO of purified tonsillar CD4⁺ T cells. (C) Representative flow plot showing expression of dual fluorescent reporters from HIV_GKO on day 3 post-infection. (D) Summary data from 6 tonsil donors with frequency of productive and latent infected cells based on gating shown in panel (C). (E) Mean fluorescent intensity of cell surface protein expression for CD4 on cells infected with HIV_GKO on day 3 post-infection. Red data points in panels D and E indicate the individual donor used for single cell analyses in single cell RNA Seq experiments. Paired t test was performed to compare Uninfected, Latent, and Productive cell populations, *p < 0.05, **p < 0.01.

Figure 2

Cellular activation and differentiation marker expression on productive and latent infected CD4⁺ T cells. (A) Measurement of memory marker CD45RO expression on productive (P, GFP+, green), Latent (L, GFP-mKO+, red), and total (T, light gray) CD4⁺ T cells 3 days post-infection with HIV_GKO. (B) Representative plots showing expression of activation markers (CD69, HLADR, CD25) and CD45RO expression on productive, latent, and total cells. (C) Representative plot showing expression of TFH markers, CXCR5 and PD-1 on productive, latent, and total cells. CXCR5^hiPD-1^hi cells define TFH population. Box and whisker plots show data from 3-5 independent experiments (individual tonsil donors). Paired t-test analysis was performed to determine differences between the groups, *p < 0.05, ns, not significant.

Figure 3

Single cell RNA Seq in HIV_GKO model of latency. (A) HIV transcript MI counts per cell in each sorted cell type. Dotted line is cutoff for background levels of transcript expression based on blank control wells in which no cells were sorted. Solid lines indicate significant difference between productive cells and all other groups using ANOVA one-way test with multiple comparisons (p < 0.05). (B) Schematic showing cell characterization based on fluorescent reporter and HIV transcript expression. (C) Venn diagram showing number of differentially expressed genes between the cell populations defined in (B). (D) Heatmap showing shared enriched pathways between Latent cell populations and productive cells. Colored boxes indicate significant pathways p < 0.01, * indicates pathways with p < 0.05 and > 0.01. (E) Bar graph showing the difference in average expression of each DEG between Latent and Uninfected cells with p values from ANOVA analysis shown as a super-imposed line graph (blue).

Figure 4

Unique gene signature in latent cells. (A) Individual graphs showing mean (error bars indicate 95% Confidence Interval) expression of each of the genes identified in Differentially Expressed Gene analysis between Latent cells and other infected populations (from Figures 3B, C). EEF1A1 expression (gray box, top left) was used as a control for EF1α promoter activity in cells. (B) Correlation matrix showing the co-expression of each of the genes from (A) on a single cell level. Color intensity is associated with the spearman correlation coefficient and the size of circle is related to the p value (larger circle, smaller p value). (C) Validation of IL32 expression in bulk sort-purified cells from 2 donors by RT-PCR.

Figure 5

Effect of JAK1/2 inhibitors on productive and latent infection. (A) Schematic showing experimental design for activation and infection with HIV_GKO of purified tonsillar CD4⁺ T cells. Bar graphs showing average frequency of (B) GFP⁺ (productive) cells and (C) mKO+GFP- (latent) cells on day 4 post-infection in presence of JAK1/2 inhibitors at different concentrations. Error bars represent standard error of the mean (SEM), students t test was used to compare all conditions against the no drug control (black bar), *p < 0.05, **p < 0.01. (D) Bar graphs showing average frequency of cells expressing activation markers on day 4 post-infection in presence of JAK1/2 inhibitors at different concentrations. Results shown represent 3 independent experiments.

Figure 6

The EC₅₀ for blocking seeding of productive and latent infection with baricitinib and ruxolitinib. The approved dose for chronic long-term use of baricitinib is 2 mg (USA), or 2 and 4 mg (Japan, other non-USA jurisdictions), and 4 mg is approved for hospitalized COVID-19 patients. The plasma concentrations for both doses of baricitinib fall within the range to effectively block both productive and latent infection in our single cycle model (A). Ruxolitinib is not approved for chronic long-term use, but approved doses range within 10-25 mg (B). Only the higher dose of ruxolitinib fell within plasma concentration ranges to block both productive and latent infection in our single cycle model (B). Baricitinib demonstrates ~ half a log greater potency for both assays versus ruxolitinib (summarized in C).

Figure 7

Baricitinib inhibits viral reactivation in latent HIV_GKO infected CD4⁺ T cells. (A) Schematic showing experimental design for sorting and re-activation of HIV_GKO in purified tonsillar CD4⁺ T cells. (B) Representative flow plot showing expression of dual fluorescent reporters from HIV_GKO on day 3 post-infection and the sorting gate for reactivation experiments. (C) Dot plots showing expression of dual fluorescent reporters from HIV_GKO on day 2 following in vitro reactivation. (D) Summary bar graph showing average frequency of GFP⁺ (productive) cells on day 2 post-LRA treatment in presence of Baricitinib at different concentrations. Error bars represent standard error of the mean (SEM), students t test was used to compare all conditions against the no drug control (white bar), *p < 0.05. Results shown represent 3 independent experiments.