Inhibition of autophagy by ATG5 disruption or bafilomycin A1 leads to increased IFN-I signaling in THP-1 cells. Monocytes, macrophages, and DCs are major IFN-I–producing cells in the course of viral infection and play essential roles in driving immunopathogenesis during chronic viral infection (20, 47). Thus, using the monocytic cell line THP-1, we first investigated how autophagy regulates IFN-I responses. We generated autophagy-impaired THP-1 cells by using CRISPR/Cas9 to disrupt expression of autophagy-related gene 5 (ATG5), an indispensable component of autophagic vesicle formation. As shown in Figure 1A, ATG5 disruption led to reduced protein expression of ATG5 in the presence or absence of autophagy and the lysosomal inhibitor chloroquine in ATG5-sg52 cell lines. Importantly, we observed decreased microtubule-associated protein 1A/1B-light chain 3 (LC3)–II expression and a lower ratio of LC3-II/actin level in ATG5-sg52 cells by immunoblotting (Figure 1A) with or without chloroquine treatment, indicating impaired autophagosome formation in ATG5-sg52 cells (48, 49). Intriguingly, we observed increased production of IFN-β1 in ATG5-sg52 cells without stimulation, and this was further elevated with stimulation by 2′3′–cyclic GMP-AMP (c-GAMP), which triggers stimulator of IFN genes (STING)–dependent IFN-I responses by binding to DNA-sensor cGAMP synthase (Figure 1B) (50, 51).

CRISPR/CAS9–mediated ATG5 disruption led to increased IFN-I signaling in THP cells. THP-1 cells were transduced with lentiviral particles containing sgRNA52 targeting ATG5 or scrambled sgRNA and subsequently incubated with puromycin to isolate stable cell lines. (A) THP-1-sg52 and scrambled control cell lines were incubated for 6 hours in the absence or presence of 10 μM chloroquine (CQ) to block lysosomal degradation. Whole-cell lysates were collected and ATG5 and LC3 expression was analyzed by Western blotting. β-Actin expression was assessed as a protein-loading control for ATG5 (top 2 blots from same gel) and LCI-II (bottom 2 blots from same gel). The LC3-II/actin ratio was analyzed by ImageJ. (B) IFN-β expression was measured in cell-culture supernatant of scrambled control or THP-1 ATG5-sg52 cells with or without cGAMP stimulation. (C) Relative RNA expression level of MX1, IRF7, and OAS in scrambled or THP-1 ATG5-sg52 cells with or without cGAMP stimulation. Data show the mean values of 3–5 independent experiments ± SEM (represented by error bars) *P < 0.05; **P < 0.01, ****P < 0.0001, Kruskal-Wallis with Dunn’s test.

To further investigate whether impaired autophagy affects expression of IFN-I–stimulating genes (ISGs), we measured the RNA expression of ISGs MX1, IRF7, and OAS1 and internal control HPRT1 by real-time PCR in scrambled control and THP-1 ATG5-sg52 cells in the presence and absence of 2′3′-cGAMP activation. As shown in Figure 1C, compared with the scrambled control, THP-1 ATG5-sg52 cells have spontaneous, elevated ISG expression in the absence of stimulation, which is further increased upon stimulation.

Bafilomycin A1 (BafA1) is a known inhibitor of the later stage of autophagy, inhibiting fusion between autophagosomes and lysosomes (52). To examine if inhibition of late-stage autophagy also affects IFN-I signaling, THP-1 cells were treated with BafA1 (50 nM) for 2 days, and we measured expression levels of ISGs MX1, IRF7, OAS1, and internal control HPRT1 by real-time PCR. As shown in Supplemental Figure 1 (supplemental material available online with this article; https://doi.org/10.1172/jci.insight.159136DS1), without additional stimulation, BafA1 treatment alone led to significant elevation of ISGs as compared with the control. Overall, these results indicate that impairment of autophagy by either ATG5 gene disruption or inhibitor BafA1 could lead to elevated IFN-I signaling, suggesting a crucial role of autophagy in regulating IFN-I production.

Induction of autophagy by rapamycin or spermidine decreases IFN-I responses in monocytes and macrophages and is dependent on ATG5. Next, we examined if induction of autophagy can reduce IFN-I signaling using 2 autophagy inducers, rapamycin and spermidine, that target distinct components of the autophagy pathway. Rapamycin, an inhibitor of mTOR, is a potent inducer of autophagy and has been tested in diverse cells and animal models for autophagy induction (53). Spermidine, in contrast, activates autophagy via mTOR-independent pathways (54–56). To examine if autophagy induction affects IFN-I signaling in primary macrophages, we treated primary monocyte-derived macrophages with rapamycin or spermidine for 2 days, followed by stimulation with either LPS or 2′3′-cGAMP for 6 hours. Prior to stimulation, we confirmed increased autophagy flux by rapamycin or spermidine in macrophages as measured by LC3-II/actin ratio (Figure 2A). After stimulation, we observed that elevation of ISG MX1 level was significantly reduced by rapamycin or spermidine treatment for both LPS-stimulated (Figure 2B) and cGAMP-stimulated (Figure 2C) macrophages, suggesting that autophagy induction via distinct pathways can effectively lower LPS or 2′3-cGAMP–stimulated ISG expression in primary macrophages.

Autophagy induction by rapamycin (Rapa) reduces IFN-I signaling in activated macrophages and THP1 cells and is dependent on ATG5. CD14+ monocytes were sorted from healthy primary PBMCs with CD14 microbeads and differentiated into macrophages with macrophage colony-stimulating factor at 10 ng/mL for 3 days. Afterward, cells were treated with the autophagy inducers rapamycin (50 pM and 500 pM) or spermidine (Spe; 10 nM and 100 nM) for 2 days. (A) Autophagy flux was measured by western blotting for LC3 and actin. The ratio of LC3-II/actin was calculated by ImageJ. After rapamycin or spermidine treatment, cells were stimulated by (B) LPS or (C) cGAMP for 6 hours; the ISG MX1 and internal control HPRT1 were measured by real-time PCR. (D and E) THP-1 scram (D) or THP-1 ATG5-sg52 (E) cells were treated with 50 pM rapamycin or 100 nM spermidine for 2 days and followed by cGAMP stimulation for 6 hours. The ISGs MX1 and IRF7 and the internal control HPRT1 were measured by real-time PCR. Data are reported as the mean values of 3–5 independent experiments ± SEM. *P < 0.05, Kruskal-Wallis analysis with Dunn’s test.

To further investigate whether rapamycin or spermidine suppression of IFN-I signaling is dependent on autophagy induction, we treated either scrambled control or autophagy-deficient ATG5-sg52 THP1 cells with either mock, 2′3′-cGAMP only, 2′3′-cGAMP with rapamycin, or 2′3′-cGAMP with spermidine. Similar to what we observed in primary macrophages, rapamycin and spermidine reduced activation of ISG MX1 by 2′3′-cGAMP stimulation in control THP1 cells (Figure 2D). However, rapamycin and spermidine suppression of 2′3′-cGAMP activation was blunted in ATG5-sg52 cells (Figure 2E), further demonstrating that rapamycin and spermidine downregulation of IFN-I is dependent on ATG5.

Combining rapamycin with ART decreases chronic IFN-I–mediated inflammation and reduces viral RNA in HIV-1–infected humanized mice. Anti-HIV T cell functions are critical for controlling HIV replication. However, during chronic infection, HIV immune invasion and persistent inflammation lead to dysfunctional HIV-specific T cells that are defective in eliminating HIV-infected cells (57). Consequently, T cell exhaustion remains one of the major barriers for achieving sustained immune surveillance for HIV infection (58). This leads to the idea that alleviating chronic activation could halt disease progression and restore immunological defects. We (20) and others (19) showed that persistent IFN-I signaling is one of the culprits that drives T cell exhaustion and blocking IFN-I receptor could restore anti-HIV T cell responses and lower the viral load or reservoir in vivo. However, because IFN-Is are key regulators in both adaptive and innate responses, safer approaches to curb persistent activation are needed.

To observe if the autophagy inducer rapamycin can treat persistent inflammation during HIV infection in vivo, we examined the effect of rapamycin treatment in humanized mice infected with HIV-1. Rapamycin is an FDA-approved drug for the prevention of transplant rejection. It is a well-characterized autophagy inducer and has an excellent safety profile, including in patients with HIV (59, 60). Humanized bone marrow/liver/thymus (BLT) mice were constructed and infected with HIV-1 for 8 weeks. Afterward, mice were given ART with either rapamycin or DMSO control for 4 weeks, as outlined in Figure 3A. After HIV infection, and prior to rapamycin treatment and ART, we observed gradual elevation of exhaustion marker PD-1 (Supplemental Figure 2A, 0–8 weeks; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.159136DS1) and ISG MX1 (Supplemental Figure 2B, 0–8 weeks) in the blood of infected animals as expected. We did not observe any differences in PD1, MX1 expression, or viral load (Supplemental Figure 2C, 0–8 weeks) between rapamycin or DMSO treatment groups prior to treatment initiation. In contrast, after ART with rapamycin or DMSO control, as shown in Figure 3B, we observed reduced expression levels of exhaustion markers PD-1 and Tim-3 and activation markers CD38 and HLA-DR in CD8+ T cells from peripheral blood in animals that received both rapamycin and ART. As shown in Figure 3C, we also observed increased ATG5 expression and LC3-II/actin expression in mice that received rapamycin treatment, suggesting increased autophagy flux by rapamycin in vivo.

Combination of ART and autophagy inducer rapamycin (Rapa) treatment effectively decreases inflammation, ISG expression, and viral replication in HIV-infected humanized BLT mice. (A) Eight weeks after immune reconstitution, BLT humanized mice were infected with HIVNFNSXL9 for 8 weeks. Afterward, mice were treated with ART and rapamycin or DMSO control for 4 weeks before necropsy. (B) PD-1, TIM-3, CD38, and HLA-DR expression was measured by flow cytometry (quantitatively by gating of percentages positive ± SEM) on peripheral blood CD8+ T cells from rapamycin or control mice (n = 6–7 per group). (C) Autophagy flux was detected by Western blotting of LC3-I, LC-II, ATG5, and actin using pooled splenocytes from DMSO or rapamycin-treated groups at necropsy (n = 5 per group). The ratio of LC3-II/actin was calculated by ImageJ. (D) Expression levels of human ISGs MX1, OAS1, and IRF7 in multiple lymphoid tissues from humanized BLT mice after treatment were measured by real-time PCR (n = 6–7 per group). (E) Splenocytes from HIV-1–infected mice treated with ART and rapamycin or DMSO were isolated and stained with intracellular Abs against human IRF7 and MX1. MFIs of the ISGs IRF7 and MX1 on human monocytes were measured by flow cytometry (n = 5–7 per group). (F) Relative HIV cellular RNA/HPRT1 expression from multiple lymphoid tissues after the indicated treatment, as compared with control blood. (n = 5–7 per group). Each dot represents an individual mouse; horizontal bars indicate median values. *P < 0.05, **P < 0.01, ***P < 0.001, Mann-Whitney U test.

To investigate the impact of rapamycin treatment on IFN-I signaling, ISGs were measured at the terminal time point by both real-time PCR (Figure 3D) and flow cytometry (Figure 3E) to measure the changes in RNA and protein expression level. Consistent with the previous in vitro data, ART combined with rapamycin led to significantly decreased expression of ISGs such as MX1, IRF7, and OAS1 in multiple lymphoid tissues (blood, spleen, and bone marrow) (Figure 3D). There was also reduction of ISG IRF7 and MX1 protein levels in human monocytes from spleen of mice receiving combined ART and rapamycin treatment (Figure 3E).

Plasma viremia for all mice was undetectable at the time of necropsy 4 weeks after ART initiation (Supplemental Figure 2C). However, we observed a trend of lower cellular viral RNA level in blood, splenocytes, and bone marrow in the ART and rapamycin-treated group (Figure 3F) as compared with ART and DMSO control group.

To investigate whether rapamycin leads to changes in the composition of immune cell types with different abilities to express ISGs (61, 62), we investigated alterations in the percentages of T cells (CD45+CD3+CD20–), B cells (CD45+CD20+CD3–), and monocytes (CD45+CD14+CD3–CD20–) among total human CD45+ lymphocytes in spleens of humanized mice after treatment. Results show that rapamycin treatment did not affect major lymphocyte composition in humanized mice (Supplemental Figure 3A). In addition, we performed T cell subset analysis to investigate whether rapamycin treatment has a differential impact on T cell subsets during infection (63, 64) or inhibition of T cell proliferation (65–67). We found rapamycin treatment did not significantly change CD4 or CD8 percentages in the peripheral blood in HIV-infected mice (Supplemental Figure 3B). Importantly, we observed downregulation of PD-1 expression in peripheral blood CD8 T cells across all non-naive subsets, including central memory (CD45RA–CCR7+), effector memory (CCR7–CD45RA–), and terminally differentiated effector memory T cells (CD45RA+CCR7–) T cell subsets, whereas the naive subset (CD45RA+CCR7+) had low PD-1 expression levels with or without rapamycin treatment (Supplemental Figure 3C). Last, we observed an increase of naive CD4 T cells in bone marrow of mice treated with rapamycin (Supplemental Figure 3D), which may have contributed to reduced viral RNA in bone marrow (Figure 3F). These observations led us to further evaluate the impact of rapamycin on viral load before ART and viral rebound after ART withdrawal.

Rapamycin treatment alone reduces IFN-I inflammation and viral load; combined with ART, it improves antiviral T cell function and reduces viral rebound after ART discontinuation. To further investigate the effects of rapamycin treatment on inflammation and antiviral T cell function in the presence and absence of ART, we first treated chronically HIV-infected humanized BLT mice with rapamycin (or DMSO control) for 2 weeks prior to ART, followed by ART with continued rapamycin (or DMSO) for 3.5 weeks. After plasma viremia became undetectable, ART was interrupted to evaluate viral rebound (Figure 4A). Prior to ART, rapamycin treatment significantly reduced both activation-marker HLA-DR and exhaustion-marker PD-1 expression on CD8+ T cells (Figure 4B), suggesting that rapamycin treatment alone in HIV-1–infected humanized mice could lead to reductions in activation- and exhaustion-marker expression among CD8+ T cells.

Autophagy inducer rapamycin (Rapa) effectively decreases inflammation and reduces IFN-I signaling. (A) At 5.5 weeks after HIV infection, BLT humanized mice were treated with rapamycin or DMSO control for 2 weeks. Afterward, while continuing rapamycin or DMSO treatment, mice were treated with ART for 3.5 weeks, followed by ART interruption for 10 days. (B) HLA-DR and PD-1 expression was measured by flow cytometry (quantitatively by gating of percentages positive) on peripheral blood CD8+ T cells before and after rapamycin or control treatment (n = 5–8 per group) prior to ART. (C) Expression levels of the ISGs MX1, OAS1, and IRF7 in human PBMCs from humanized BLT mice after treatment were measured by real-time PCR throughout HIV-1 infection and from rapamycin- or DMSO control–treated mice in comparison with uninfected animals (n = 5–8 per group). Each dot represents an individual mouse; horizontal bars indicate median values. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, LMM.

Next, we investigated longitudinal expression level of ISGs MX1, IRF7, and OAS in HIV-1–infected mice treated with rapamycin (or DMSO control), followed by ART and discontinuation of ART. As shown in Figure 4C, HIV infection led to chronic elevation of IFN-I signaling, even in the presence of ART (Figure 4C, comparing the HIV-infected, DMSO-treated group versus uninfected controls). Rapamycin treatment alone significantly reduced IFN-I signaling (Figure 4C, week 7), which was further reduced when combined with ART to levels similar to those in uninfected controls (Figure 4C, weeks 9–11). After ART discontinuation and viral rebound, we observed significant elevation of ISGs in the DMSO control–treated mice as compared with uninfected control mice, whereas the ART and rapamycin combined-treatment group maintained low levels of ISGs (Figure 4C, week 13). Overall, ISG expression was significantly reduced in the rapamycin group as compared with the DMSO group (MX1: P < 0.0001; IRF7: P = 0.0167; OAS1: P = 0.0007, by linear mixed model [LMM]).

In addition, we investigated rapamycin’s effects on different immune cell types and T cell subsets. Rapamycin treatment did not alter the overall percentages of T cell, B cell, and monocytes subsets among total CD45+ lymphocytes (Supplemental Figure 4A). Rapamycin treatment also did not affect the percentages of CD4 and CD8 cells among T cells in HIV-infected mice (Supplemental Figure 4B). Last, apart from a slight increase of bone marrow–naive CD8 T cells, we did not observe a significant impact of rapamycin treatment on T cell subsets across different tissues (Supplemental Figure 4, C–E).

We next examined longitudinal changes in plasma viremia in treated mice. As shown in Figure 5A, prior to ART, 10 days of rapamycin treatment alone reduced plasma viremia as compared with the DMSO control (Figure 5A, week 7). Combination of rapamycin and ART led to faster viral suppression as compared with ART and DMSO control treatments (Figure 5A, weeks 9–11). Intriguingly, after ART withdrawal, the rapamycin-treated group had significantly lower plasma viremia after viral rebound (~1 log lower) (Figure 5A, week 13). Overall, we observed significantly lower viral load in the rapamycin group than in the DMSO group (P = 0.0002, by LMM). In addition, we also observed significantly lower levels of viral DNA (Figure 5B) and HIV RNA (Figure 5C) in both blood and spleen at necropsy after ART withdrawal, suggesting a reduction in overall viral replication in the rapamycin-treated group.

ART and autophagy inducer rapamycin cotreatment effectively reduces viral replication. (A) Longitudinal HIV viral load in plasma from humanized BLT mice after treatment were measured by real-time PCR. *P < 0.05, **P < 0.01, LMM. (B) HIV DNA copies per 103 cells (measured by huRRP30 expression) from blood PBMCs or splenocytes as measured by real-time PCR (n = 6–8 per group). (C) Relative HIV cellular HIV/HPRT1 compared with control blood from at end point (n = 5–8 per group). Each dot represents an individual mouse; horizontal bars indicate median values. *P < 0.05, **P < 0.01, Mann-Whitney U test.

To investigate whether HIV-specific CD8+ T cell responses were improved in the ART and rapamycin combined-treatment group, we stimulated splenocytes from uninfected, rapamycin-treated, or DMSO control mice with either the mitogens PMA or ionomycin or an HIV clade B peptide pool (Gag, Env, Nef, and Pol). To control for potential enrichment effects of rapamycin on naive T cells (63, 64, 68), we closely examined non-naive CD8+ T cells, as shown in Supplemental Figure 5 (representative gating plots), Figure 6A (representative flow), and Figure 6B (flow summary). Compared with the uninfected control, non-naive CD8+ T cells from DMSO-treated infected mice produced significantly lower levels of pro-inflammatory IFN-γ and IL-2 cytokines after PMA or ionomycin stimulation, suggesting functional exhaustion of T cells. In contrast, non-naive CD8+ T cells from infected mice treated with rapamycin produced increased levels of pro-inflammatory cytokines in response to both PMA or ionomycin and HIV-1–specific peptide pool stimulation. As a whole, these data suggest that combination of rapamycin and ART improved HIV-specific and mitogen stimulation–induced T cell responses and is correlated with increased control of viral replication after ART withdrawal.

ART and autophagy inducer rapamycin cotreatment improve anti-HIV immune responses. Splenocytes from HIV-1–infected, DMSO-treated, or rapamycin-treated mice were stimulated with PMA or ionomycin or an HIV-1 clade B peptide pool (Pol, Gag, Env, and Nef), and production of IFN-γ by CD8 cells was measured by flow cytometry (representative of n = 4–6 per group). (A and B) Representative flow (A) and cytokine assay summary (B) showing percentage of IFN-γ+ and IL-2+ among CD3+CD8+ non-naive T cells from HIV-1–infected, DMSO-treated, or rapamycin-treated mice. (n = 4–6 per group). Each dot represents an individual mouse; horizontal bars indicate median values. *P < 0.05, **P < 0.01 Kruskal-Wallis analysis with Dunn’s test.