Previous reports have implicated caspase-3 activation and apoptosis in most instances of cell death caused by HIV-1^3,7,8. To explore the role of caspase-1 in dying HIV-infected CD4 T cells, HLACs formed with freshly dissected human tonsillar tissues were infected with a GFP reporter virus (NLENG1), prepared from the X4-tropic NL4-3 strain of HIV-1. This reporter produces fully replication-competent viruses. An IRES upstream of the nef gene preserves Nef expression and supports LTR-driven GFP expression¹⁶, allowing simultaneous quantification of HIV-1 infection and caspase activation in CD4 T cells. NL4-3 was selected because tonsillar tissue contains a high percentage of CD4 T cells that express CXCR4 (90–100%). Consistent with our previous report¹¹, infection with HIV-1 produced extensive depletion of “bystander” non-productively infected CD4 T cells. No more than 4% of the CD4 T cells were productively infected with HIV-1, but most of the remaining CD4 T cells underwent abortive infection and ultimately died after four days in culture (Fig. 1a).

To determine the distribution of active caspase-1 and caspase-3 in the dying CD4 T cells, we used fluorescently labeled inhibitor of caspases (FLICA) probes with sequences targeted by specific activated caspases¹⁷. Interestingly, the majority of non-productively infected CD4 T cells exhibited activation of caspase-1. Conversely, essentially no caspase-1 activity was detected in the productively infected cells (Fig. 1b). Caspase-3 activity was markedly less abundant, and mainly confined to the productively infected subset of cells (Fig. 1c). Treatment, with efavirenz (a non-nucleoside reverse transcriptase inhibitor, NNRTI) or AMD3100 (an inhibitor of CXCR4-dependent HIV entry) prevented activation of both caspases. Infection with the primary, dual-tropic 89.6 HIV isolate¹⁸ produced similar results (Extended Data Fig. 1). The two FLICA probes appeared to bind their respective caspases with reasonable specificity based on exclusive caspase-3 staining in cells treated with staurosporine, a protein kinase inhibitor known to induce apoptosis versus robust caspase-1 staining in cells treated with the cationic ionophore nigericin that promotes NLRP3 inflammasome assembly, caspase-1 activation, and pyroptosis¹⁹.

IL-1β activity is controlled at multiple levels including pro-IL-1β expression, processing, and secretion. Proinflammatory stimuli induce expression of pro-IL-1β while processing and release are regulated by caspase-1 activation in inflammasomes²⁰. The signals required for caspase-1 activation and release of IL-1β differ between immune cells. In circulating human blood monocytes, caspase-1 is constitutively active²¹. Stimulation of these cells with LPS promotes pro-IL-1β expression leading to the rapid release of bioactive IL-1β. In contrast, macrophages and dendritic cells require a second signal to activate caspase-1²². Nigericin can function as this second signal activating caspase-1 in LPS-primed macrophages^19,23. Surprisingly, nigericin proved sufficient alone to activate caspase-1 in uninfected lymphoid CD4 T cells (Fig. 1b) and to promote the release of the 17-kDa bioactive form of IL-1β (Fig. 2a). Treatment with monensin, a different monovalent cationic ionophore, or A23187, a calcium ionophore, did not promote mature IL-1β release^23,24. Maturation and secretion of the bioactive form of IL-1β was inhibited by Z-VAD-FMK (a pan-caspase inhibitor), Z-WEHD-FMK, or Z-YVAD-FMK (two independent caspase-1 inhibitors, which also block other inflammatory caspases i.e. caspase-4 and -5), but not by Z-FA-FMK (negative control for caspase inhibitors) suggesting that caspase-1 activation was required.

Pro-IL-1β expression in human tonsil and spleen HLACs was next examined. Western blotting analysis surprisingly revealed large amounts of intracellular pro-IL-1β in both untreated tonsil and spleen HLACs (Fig. 2b). Removal of dead cells by ficoll-hypaque density centrifugation resulted in an even higher intracellular pro-IL-1β signal, suggesting that these normal lymphoid tissues constitutively express high levels of pro-IL-1β. The presence of pro-IL-1β in spleen argued that expression in tonsil is not solely caused by infection (tonsillitis). Fractionation of the lymphocytes present in these HLACs revealed high levels of intracellular pro-IL-1β in isolated CD4 T cells, but not in CD8 T or B-cell populations.

Most tonsillar CD4 T cells express CXCR4, but only ∼5% of these cells also express CCR5^12,25. Interestingly, when CCR5-positive and -negative lymphoid CD4 T-cell subsets were isolated and studied, the CCR5-expressing cells displayed much higher levels of intracellular pro-IL-1β (Fig. 2b). The CCR5-expressing CD4 T cells also released significantly more 17 kDa IL-1β into the supernatant after infection with HIV-1 (Fig. 2c). These results suggest that most of the mature form of IL-1β is released by the small population of CCR5-expressing CD4 T cells. The resident CCR5-expressing cells in lymphoid tissues are primarily memory CD4 T cells, which might be more permissive for productive HIV infection²⁶. However, the activation status of these cells varied (Fig. 2d). Two thirds exhibited a memory phenotype as determined by surface expression of CD45RO but only a small fraction of these cells were permissive to productive infection with either X4-tropic or R5-tropic HIV-1 strains (Extended Data Fig. 2). Notably, lymphoid CCR5-expressing CD4 T cells also express CXCR4 and thus can be targeted by either X4 or R5-tropic HIV-1 strains^12,27,28. Memory T cells continually recirculate within lymphoid tissues scanning for presentation of their cognate antigen^29-31. It seems likely that many of these cells have returned to a sufficient state of quiescence that they are susceptible to abortive HIV infection and thus could contribute importantly to chronic inflammation through the release of bioactive IL-1β.

Caspase-1 is a proinflammatory caspase whose catalytic activity is tightly regulated by signal-dependent autoactivation within inflammasomes²⁰. Inflammasome-dependent caspase-1 activity results in a highly inflammatory form of cell death known as pyroptosis, primarily described in myeloid cells infected with intracellular bacterial pathogens^9,15,32. Pyroptosis is caspase-1-dependent by definition and occurs independently of other proapoptotic caspases^9,32. Based on our finding that caspase-1 is activated in lymphoid CD4 T cells following abortive HIV infection, we investigated whether pyroptosis is triggered within these cells.

Fresh HLACs were infected with HIV-1 and cultured for 12 hours to initiate viral spread and then treated with various caspase inhibitors or controls. Extensive and selective depletion of CD4 T cells occurred in untreated, HIV-infected cultures after 3 days. However, treatment with either pan-caspase or caspase-1 inhibitors prevented the depletion of CD4 T cells as efficiently as the viral inhibitors efavirenz and AMD3100 (Fig. 3a). Inhibitors of caspase-3 or caspase-6 and the control compound did not prevent CD4 T-cell depletion. Necrostatin-1, a RIP1 inhibitor, did not inhibit CD4 T-cell depletion (Extended Data Fig. 3a, b), suggesting that cell death does not reflect necroptosis. Analysis of spleen cells yielded similar results (Extended Data Fig. 3c). Inhibiting type-I interferon signaling with neutralizing antibodies directed against IFNα/β receptor did not prevent CD4 T-cell death (Extended Data Fig. 4), indicating that this antiviral response is not critical for the innate immune-mediated onset of programmed cell death. Distinct from apoptosis, pyroptosis features cellular swelling, plasma membrane rupture, and release of intracellular content into the extracellular milieu¹⁵, including cytosolic enzymes like lactate dehydrogenase (LDH)³³. LDH release was readily detected after HIV infection (Fig. 3b), but completely blocked by an inhibitor of caspase-1, efavirenz and AMD3100, not by a caspase-3 inhibitor. Thus, the form of cell death associated with abortive HIV infection appears to involve caspase-1 activation and the release of cytoplasmic contents. Caspase-1 inhibitors also prevented death of CCR5-expressing CD4 T cells in HLACs infected with a CCR5-dependent strain of HIV-1 (Fig. 3c). Inhibition of cell death by the caspase-1 inhibitor was as effective as the CCR5 receptor antagonist TAK779, suggesting that most CCR5-expressing CD4 T cells die by caspase-1-mediated pyroptosis. These findings are consistent with the large amounts of bioactive IL-1β released by these cells after HIV-1 infection.

Because caspase inhibitors are not exquisitely specific. we designed shRNA vectors to silence the expression of caspase-1, the ASC (PYCARD) adaptor, which recruits pro-caspase-1 to inflammasome complexes²⁰, caspase-3, and NLRP3 (Extended Data Fig. 5). For these experiments, a third generation shRNA-encoding lentiviral vector (shRNA LV) pSico³⁴, bearing an EF1α:mCherry reporter expression cassette was used. To relieve the resistance of lymphoid CD4 T cells to shRNA LV infection, target cells were initially challenged with Vpx-harboring lentiviral particles (Vpx-VLPs), which induce proteasomal degradation of SAMHD1 in nonpermissive human resting CD4 T cells³⁵. Infections with Vpx-VLPs did not lead to activation of resting CD4 T cells, as measured by surface expression of the CD69 and CD25 activation markers (not shown). The shRNA LV particles and Vpx-VLPs were pseudotyped with a CXCR4-tropic Env of HIV-1, which supports efficient fusion to quiescent CD4 T lymphocytes³⁶. Under these conditions, infection with shRNA LVs markedly suppressed expression of a variety of targeted genes while the scrambled shRNA LV control did not (Fig. 3d). We next investigated whether any of these shRNA LVs inhibited pyroptosis induced by nigericin. Nigericin induced massive pyroptosis in mCherry positive CD4 T cells infected with scramble and caspase-3 shRNA LV particles, but this response was blocked by the caspase-1, ASC, or NLRP3 shRNA LV particles (Fig. 3e). Next, the effect of these shRNAs on CD4 T-cell death elicited by HIV-1 was examined. HIV-1 infection caused extensive death of mCherry-positive CD4 T cells expressing shRNAs against scramble, caspase-3 and NLRP3, but not caspase-1 and ASC. Thus, cell death occurring during abortive HIV infection appears to be mediated through caspase-1 dependent pyroptosis involving an inflammasome that contains ASC but lacks NLRP3.

To independently confirm that abortive HIV-1 infection leads to the activation of caspase-1, we investigated appearance of the active p10 subunit of caspase-1. As controls for pyroptosis and apoptosis, uninfected cells were treated with nigericin or staurosporine, respectively. An active 10kDa subunit of caspase-1 (p10) was detected in the lysates of HIV-infected cultures as well as in nigericin-treated cells, and in blood monocytes where caspase-1 is constitutively active²¹. Treatments with viral or caspase-1, but not caspase-3, inhibitors prevented caspase-1 cleavage (Fig. 3f). These findings confirm the induction of caspase-1 in quiescent CD4 T following abortive infection with HIV-1. Caspase-3 activation in these infected cultures was markedly less abundant (Extended Data Fig. 6). To test whether caspase-1 activation leads to proteolytic maturation of pro-IL-1β, we used various caspase inhibitors and analyzed the culture media for the mature 17-kDa form of IL-1β. Interestingly, release of mature IL-1β was completely inhibited by a pan-caspase inhibitor, and by two different caspase-1 inhibitors (Fig. 3g). Inhibitors of apoptotic caspases, caspase-3, -6, or -8, or necrostatin did not interrupt this release. Similar findings were observed using a quantitative IL-1β enzyme-linked immunosorbent assay (ELISA) (Extended Data Fig. 7a). Thus, caspase-1 activation is specifically required for the release of bioactive IL-1β in lymphoid CD4 T cells infected with HIV-1. In accord with shRNA analyses, treatment with four separate NLRP3 inhibitors did not prevent release of bioactive IL-1β by HIV-1 (Fig. 3g), nor CD4 T-cell death by HIV-1 (Extended Data Fig. 7b, c).

To extend our ex vivo HLAC findings, we next examined fresh lymph node tissue obtained from a consenting untreated subject infected with R5-tropic HIV and displaying a high viral load and a low CD4 T cell count. In-situ immunostaining revealed a distinct zone of HIV p24^gag expression between the mantle zone and germinal centers, where activated CD4 T and B cells proliferate (Ki67) and interact in the follicles (Fig. 4). Conversely, staining for caspase-1 revealed abundant activity in the surrounding paracortical zone (CD3) comprised primarily of resting CD4 T cells. Staining of uninfected tonsil or spleen (not shown) tissues revealed no such positive signals (Extended Data Fig. 8). Because this antibody reacts with both the active p20 component of caspase-1 and pro-caspase-1, we cannot completely exclude the possibility that abortive HIV-1 infection produced localized increase in pro-caspase-1 expression. However, large amounts of IL-1β were also detected in the paracortical zone, particularly in the extracellular space between the T cells, as well as the cell death marker annexin V. In sharp contrast, active caspase-3 staining was limited to the areas in the germinal center where HIV-1 p24^gag expression was detected. These findings strongly agree with the HLAC results (Fig. 1b) suggesting that caspase-3 activity occurs in a set of productively infected cells, anatomically separated from the majority of resting CD4 T cells undergoing abortive infection, caspase-1 activation, IL-1β processing, and pyroptosis.

Identifying pyroptosis as the predominant mechanism mediating CD4 T-cell depletion during HIV infection provides novel targets, such as caspase-1, for potential therapeutic intervention. The role of caspase-1 in the chronic inflammatory response has attracted therapeutic interest³⁷. VX-765 is caspase-1 inhibitor that has been tested in chronic epilepsy and psoriasis (Extended Data Fig. 9a)^38-41, and found in a phase IIa trial to be safe and well tolerated⁴². In our studies, VX-765 inhibited IL-1β secretion by nigericin-induced lymphoid CD4 T cells (Extended Data Fig. 7b), indicating it efficiently blocks caspase-1 activity in these cells. VX-765 also blocked caspase-1 cleavage (Fig. 3f), IL-1β secretion (Fig. 3g), and CD4 T-cell death in HIV-infected tonsillar and splenic HLACs (Fig. 3c, 5a, 5b, Extended Data Fig. 9b). Cell death was not markedly inhibited by VRT-043198 (the active form of the VX765 prodrug), likely because of reduced cellular permeability³⁸. HIV-1 infection was not restored to productive infection when caspase-1 was blocked (Extended Data Fig. 10). These findings demonstrate that a small-molecule inhibitor of caspase-1 shown to be safe in humans suppresses CD4 T-cell death and inflammation elicited in lymphoid tissues by HIV-1.