IFI16 DNA Sensor Is Required for Death of Lymphoid CD4 T Cells Abortively Infected with HIV

See allHide authors and affiliations

Science  24 Jan 2014:
Vol. 343, Issue 6169, pp. 428-432
DOI: 10.1126/science.1243640


The progressive depletion of quiescent “bystander” CD4 T cells, which are nonpermissive to HIV infection, is a principal driver of the acquired immunodeficiency syndrome (AIDS). These cells undergo abortive infection characterized by the cytosolic accumulation of incomplete HIV reverse transcripts. These viral DNAs are sensed by an unidentified host sensor that triggers an innate immune response, leading to caspase-1 activation and pyroptosis. Using unbiased proteomic and targeted biochemical approaches, as well as two independent methods of lentiviral short hairpin RNA–mediated gene knockdown in primary CD4 T cells, we identify interferon-γ–inducible protein 16 (IFI16) as a host DNA sensor required for CD4 T cell death due to abortive HIV infection. These findings provide insights into a key host pathway that plays a central role in CD4 T cell depletion during disease progression to AIDS.

Sensing HIV

The depletion of quiescent CD4+ T cells from lymphoid organs is a major event contributing to the development of AIDS. The accumulation of incomplete HIV DNA transcripts in the cytoplasm of these cells, which do not themselves become productively infected, is somehow sensed, which triggers cell death. Monroe et al. (p. 428, published online 19 December; see the Perspective by Gaiha and Brass) now identify the host DNA sensor as interferon-γ–inducible protein 16, which senses viral DNA and activates pyroptosis, an inflammatory cell death pathway.

HIV-AIDS is a devastating global epidemic with over 70 million infections and 35 million deaths (according to the World Health Organization). AIDS is primarily caused by loss of the quiescent “bystander” CD4 T cells that populate lymphoid organs. These cells are not permissive for viral replication, which results in abortive infection and the accumulation of incomplete DNA transcripts (1). These cytosolic viral DNAs trigger an innate immune response that activates cell death. Here, we sought to identify the host DNA sensor that initiates cell death in abortively infected CD4 T cells.

An unbiased proteomic approach involving DNA affinity chromatography and mass spectrometry was used to identify potential viral DNA sensor candidates. Cytosolic fractions of tonsillar CD4 T cell lysates were incubated with a biotinylated 500–base pair (bp) HIV-1 Nef DNA fragment and subjected to streptavidin immunoprecipitation, SDS–polyacrylamide gel electrophoresis (SDS-PAGE), and silver staining (Fig. 1A). The Nef region is reverse-transcribed early; thus, this DNA reverse-transcription product is likely present during abortive HIV infection. Streptavidin immunoprecipitation samples incubated with biotinylated HIV DNA showed numerous bands (Fig. 1A). Nonspecific background binding was very low: Protein was not detected when nonbiotinylated DNA was tested. The cytosolic lysates appeared free of nuclear contamination, as immunoblotting showed no histone H3 (Fig. 1B). Mass spectrometry was used to identify cytosolic proteins from the tonsillar CD4 T cells that bound to HIV DNA. The top six hits, based on protein discriminant scores (27), correspond to Ku80, PARP-1, Ku70, RPA-1, interferon-γ–inducible protein 16 (IFI16), and interferon-inducible protein X (IFIX) (Fig. 1C) (see data file S1 for the complete list).

Fig. 1 Biochemical analysis of cytosolic DNA binding proteins in CD4 T cells.

(A) Tonsillar CD4 T cell lysates were incubated with a 500-bp biotinylated HIV Nef DNA probe or control nonbiotinylated DNA and immunoprecipitated with streptavidin-coated beads. Samples were separated by SDS-PAGE and silver stained. (B) Western blot analysis of nuclear histone H3 and β-actin in whole CD4 T cells or digitonin lysis buffer–prepared CD4 T cell lysates. (C) Top-ranked hits (rank based on protein discriminant scores described in materials and methods) from mass spectrometry samples prepared as in (A). (D) Western blot analysis of candidate DNA sensors. (E) SDS-PAGE and silver stain analysis of biotinylated dsDNA or ssDNA samples prepared as in (A) and competed with a 10-fold excess of ssDNA or dsDNA. (F) Western blot analysis of IFI16 and RIG-I–binding samples in (E). (G) Western blots with high levels of protein input showing IFI16 binding to biotinylated ssDNA and dsDNA and RIG-I–RNA controls.

A rational approach to investigating biologically relevant DNA sensor candidates was pursued in parallel. Expression of various known innate immune sensors was assessed by immunoblotting cytosolic lysates from resting tonsillar CD4 T cells, which confirmed the presence of IFI16 (2, 3); AIM2 (47); DAI (8); STING (911); DNPK-1 (12); NLRP3 (1315); and IFIX (PYHIN-1) (16) (Fig. 1D). Cyclic guanosine monophosphate–adenosine monophosphate synthase (cGAS) (17, 18) was detected neither at the protein level in tonsillar CD4 T cells (fig. S1D) nor in the affinity chromatography–mass spectrometry experiments (data file S1). We were intrigued with IFI16 because it was identified in both approaches and shown to form an inflammasome (3, 16). Of the known inflammasome DNA sensors, IFI16, but not AIM2, bound HIV-1 DNA (Fig. 1D). Because AIM2 binds DNA in a non–sequence-specific manner, we had expected that AIM2 would be a top candidate, but it was not identified by mass spectrometry (data file S1). IFI16 mRNA levels are five times those of AIM2 mRNA in resting tonsillar CD4 T cells (fig. S1A). Of note, all three IFI16 isoforms were detected in the cytosol and nucleus of primary tonsillar CD4 T cells (fig. S1B).

Reverse transcription of the HIV RNA genome initially generates single-stranded DNA (ssDNA) and then double-stranded DNA (dsDNA); either might be sensed during abortive infection. A biotinylated dsDNA probe was incubated with cytosolic extracts from tonsillar CD4 T cells with 10-fold excess of unlabeled ssDNA as a competitor (Fig. 1E). IFI16 effectively bound dsDNA (Fig. 1F) as described (2, 19) and was in competition with “cold” ssDNA. Biotinylated ssDNA was subjected to binding and competition with cold dsDNA, but IFI16 was not initially detected by immunoblotting. However, further analysis with higher protein input confirmed that IFI16 binds to ssDNA, albeit more weakly than to dsDNA (Fig. 1G). Retinoic acid–inducible gene I (RIG-I) selectively bound dsRNA as a control (Fig. 1, F and G).

Standard methods, including liposome-mediated delivery of small interfering RNAs (siRNAs) or infection with vesicular stomatitis virus glycoprotein (VSVG)–pseudotyped lentiviruses encoding short hairpin RNAs (shRNAs) are ineffective for targeted gene knockdown in resting CD4 T cells (20, 21). siRNA nucleofection is highly variable, often toxic, and associated with extensive cell death in tonsillar cultures. To overcome these challenges and to test whether specific DNA sensor candidates are required for cell death in primary lymphoid CD4 T cells undergoing abortive HIV infection, we used an “activation-rest” strategy. Splenic CD4 T cells were activated with phytohemagglutinin (PHA) and cultured in 100 U/ml of interleukin-2 (IL-2), which rendered cells permissive for infection with VSVG-pseudotyped lentiviruses encoding shRNA and mCherry. mCherry-positive cells were isolated by cell sorting (fig. S2), expanded by two rounds of activation with beads conjugated with antibodies against CD3 and CD28, and then rested by reducing IL-2 levels to 10 U/ml for 3 to 4 days (22). IFI16 protein expression was markedly decreased in the mCherry-positive splenic CD4 T cells receiving the lentivirus encoding shIFI16-A compared with cells receiving the lentivirus encoding the control scramble shRNA (Fig. 2A). Next, the rested mCherry-positive CD4+ T cells were cocultured with tonsil or spleen CD4 T cells infected with a green fluorescent protein–HIV (HIV-GFP) reporter virus (NLENG1). In cells expressing the scramble shRNA, marked depletion of CD4 T cells occurred (Fig. 2B); this death was rescued by adding a nonnucleoside reverse-transcriptase inhibitor, efavirenz (EFV), which implicates abortive HIV infection as previously described (1). In sharp contrast, introduction of shIFI16-A resulted in survival of the mCherry-positive CD4+ T cells. In the same experiments, mCherry-negative CD4+ T cells were markedly depleted, which suggested that they had returned to a sufficient state of rest to undergo abortive infection.

Fig. 2 IFI16 shRNA knockdown rescues activated and rested splenic CD4 T cells from depletion after abortive HIV infection.

(A) Western blot analysis of IFI16 and β-actin expression in shRNA-expressing mCherry-positive CD4+ T cells after activation and rest in reduced IL-2. (B) Flow cytometric analysis of mCherry+ CD4+ T cell survival after knockdown with shSCR (scramble shRNA) or shIFI16-A and coculture with either donor-matched mCherry CD4+ T cells or tonsillar HLACs spinoculated with an HIV-1–GFP reporter virus. Cells were cocultured in the presence or absence of 100 nM efavirenz or with uninfected cells. Data represent the average of three independent experiments from three different donors. Error bars, SEM; *P < 0.05 (Student’s t test); n.s., not significant, P > 0.05. (C) Flow cytometric analysis of CD25 and CD69 expression after IL-2 reduction. (D) Flow cytometric analysis of mCherry+ GFP+ populations in shRNA-expressing spleen cells post coculture.

To exclude more formally the possibility that the “activated and rested” CD4 T cells were dying as a result of productive infection, we assessed the activation status of these cells. Flow cytometric analysis revealed that CD4 T cells cultured in reduced concentrations of IL-2 had lower levels of CD25 and CD69 than cells activated with 100 U/ml IL-2 and 10 μg/ml PHA (Fig. 2C). However, CD25 levels were higher than found in unactivated cells, which indicated that these cells had not fully returned to a resting state. This finding likely relates in part to the up-regulation of CD25 expression by IL-2 (23). To directly test how permissive these cells were to productive HIV infection, we utilized an HIV-1–GFP reporter virus. In cells expressing shScramble or shIFI16-A, only ~1 to 2% of the mCherry-positive cells and ~1 to 2% of mCherry-negative cells were productively infected, as indicated by GFP expression (Fig. 2D). Thus, the 60 to 70% depletion of CD4 T cells observed was not due to high levels of productive viral infection.

To confirm IFI16 as an HIV-1 DNA sensor and to test a broader array of potential candidates, we used a second, more rapid shRNA knockdown strategy. Virus-like particles (VLPs) were packaged with the simian immunodeficiency virus (SIV) accessory protein Vpx that degrades the SAMHD1 restriction factor and renders cells susceptible to lentiviral infection (24, 25). This method was adapted for use in resting CD4 T cells on the basis of prior success in cells of myeloid origin (26, 27). Twenty-four hours after VLP-Vpx spinoculation, complete tonsillar HLACs were spinoculated with shRNA-mCherry lentiviral vectors pseudotyped with HIV glycoprotein 160 (gp160) envelope protein (Env) (fig. S3) (28). Cells were cocultured 3 days later with human embryonic kidney (HEK) 293T cells producing or not producing HIV-1 virions. CD4 T cell death was assessed 1 or 2 days later in mCherry-positive CD4+ T cells expressing the shRNA and mCherry-negative CD4+ T cells lacking the shRNA. In parallel, EFV was added to select wells.

Three independent shRNAs targeting IFI16 reduced IFI16 protein expression in mCherry-positive CD4+ T cells compared to the shScramble control (Fig. 3, A and C). All three IFI16 shRNAs prevented depletion of mCherry-positive CD4+ T cells, whereas shScramble did not (Fig. 3, B and D). EFV rescued depletion of scramble-shRNA–expressing cells, which supported the notion that the CD4 T cell depletion resulted from abortive infection (Fig. 3B). Moreover, mCherry-negative CD4+ T cells were depleted regardless of the shRNA, which demonstrated that experimental conditions were sufficient for abortive infection in all infected samples (Fig. 3, B and D). Thus, using an independent method for shRNA knockdown, we confirmed that IFI16 is required for lymphoid CD4 T cell depletion by HIV after abortive HIV infection.

Fig. 3 shRNA knockdown of IFI16 rescues HIV-induced tonsillar CD4 T cell depletion.

(A) Western blot analysis of IFI16 and β-actin expression in mCherry+ tonsillar CD4 T cells receiving shScramble (shSCR), shIFI16-A, or shIFI16-B. (B) Quantification of flow cytometry of HLACs infected with VLP-Vpx, followed by shSCR, shIFI16-A, or shIFI16-B lentiviruses pseudotyped with HIV gp160 Env then cocultured with HEK293T cells producing HIV-1. **P > 0.01 (Student’s t test); n.s., not significant, P > 0.05. (C) Western blot analysis of shIFI16-C knockdown. (D) Quantification of flow cytometry results as in (B). ***P < 0.001. (E) Quantification of mCherry+ gate of HLACs treated as in (B) with single-round HIV-1ΔEnv pseudotyped with gp160 envelope or HIV-1 D116N integrase mutant. **P < 0.01, *P < 0.05. (F) Flow cytometric analysis of FLICA-660 caspase-1 and IFN-β intracellular staining in mCherry+ cells. Histograms are representative of results obtained from two donors.

To confirm that shScramble mCherry-positive CD4 T cells die via abortive infection, which requires reverse transcription but not integration (1), cells were cocultured with HEK293T cells to produce single-round HIV-1 (ΔEnv with gp160 coexpressed) or HIV containing a disabling integrase mutation, D116N, in which aspartic acid at codon 116 is replaced by asparagine (Fig. 3E). These replication-defective, nonspreading viruses induced depletion of mCherry-positive CD4 T cells expressing shScramble. In contrast, introduction of shIFI16-A rescued cells from HIV-1–mediated depletion. Thus, neither productive infection nor HIV integration is required for cell death. Knockdown of IFI16 decreased caspase-1 activation in the mCherry-positive cells, whereas interferon-β (IFN-β) was induced in HIV-infected cells with the shScramble control but not in cells expressing shIFI16-A (Fig. 3F). These findings suggest that IFI16 is required to sense incomplete DNA reverse transcripts that accumulate in abortively infected cells; their accumulation leads to caspase-1 activation, which results in the subsequent death of these cells via pyroptosis (29). IFI16 sensing also leads to IFN-β induction.

Although IFI16 shRNAs effectively rescued death of lymphoid CD4 T cells during abortive infection, other DNA sensor candidates were also evaluated. The VLP-Vpx method was used to render resting lymphoid CD4 T cells permissive to infection with lentiviruses encoding shRNAs directed against AIM2 and STING. Although effective in inhibiting expression of AIM2 and STING protein in human (THP-1 cells) (Fig. 4A), neither of these shRNAs rescued the mCherry-positive cells from depletion (Fig. 4B). Validated shRNAs targeting IFIX (Fig. 4C) or DNPK-1 (Fig. 4E) also did not rescue mCherry-positive CD4 T cell depletion (Fig. 4, D and F). Moreover, small-molecule inhibitors of DNPK-1, Nu7026, and Nu7441 (12) did not rescue cells from abortive infection and pyroptosis (Fig. 4G). These findings, and a recent publication, suggest that DNPK-1 may play a role in DNA sensing only within the small fraction of cells (5% in tonsil) that are permissive for productive HIV infection and trigger noninflammatory apoptosis (12). In contrast, IFI16 appears to be required to detect abortive infection and induction of highly inflammatory pyroptosis in nonpermissive CD4 T cells (Fig. 4H). These cells form the majority of HIV-1 cellular targets in most lymphoid tissues (95% in tonsillar cultures). Both mechanisms likely contribute to HIV-1–induced AIDS, but at different frequencies determined by the number of permissive versus nonpermissive cellular targets residing within various lymphoid tissues.

Fig. 4 VLP-Vpx–facilitated shRNA knockdown of other candidate DNA sensors does not rescue cells from depletion after abortive HIV infection.

(A) Western blot analysis of AIM2, STING, and heat shock protein 90 (HSP90) in shRNA expressing mCherry+ THP-1 cells. (B) Quantification of flow cytometry results for HLACs infected with shScramble (shSCR) shAIM2, or shSTING. n.s., not significant, P > 0.1 (Student’s t test). (C) Western blot analysis of IFIX in mCherry+ SupT1 cells. (D) Quantification of flow cytometric analysis as in (B) of shSCR and shIFIX. (E) Western blot analysis of DNPK-1 in shRNA expressing mCherry+ Jurkat T cells. (F) Flow cytometric analysis as in (B) with shDNPK-1. (G) Carboxyfluorescein succinimidyl ester (CFSE)–labeled HLACs were pretreated with dimethyl sulfoxide (DMSO) alone in uninfected and no drug conditions, 10 or 20 μM Nu7026, or 1 or 2 μM Nu7441, or 250 nM AMD3100. CFSE+ cells were cocultured with donor-matched HLACs productively infected with HIV and analyzed 3 days post coculture. Quantified data represent the average of three independent experiments from three different donors. Error bars, SEM. (H) Summary model.

IFI16 evolved as an antiviral DNA sensor (2, 3). In addition, IFI16 exerts novel antiviral activity, including restriction of herpesvirus replication by inhibiting viral gene expression (30). That IFI16 is targeted for degradation by herpesviruses (31) further highlights an evolutionary pressure to counteract its activity. Our studies reveal that IFI16 initiates an innate immune response that, rather than protecting the host, drives the debilitating CD4 T cell depletion that underlies progression to AIDS in untreated HIV-infected individuals. The cycle of abortive infection, inflammatory death, and recruitment of new cells likely explains how this innate host response is undermined and, in fact, centrally contributes to HIV pathogenesis. Our findings now identify IFI16 as a critical DNA sensor required for cell death during abortive HIV-1 infection. Therapies directed against this host death pathway might preserve CD4 T cells and reduce chronic inflammation––two signature pathologies in HIV infection.

Supplementary Materials

Materials and Methods

Figs. S1 to S3

Table S1

Data File S1

References (3234)

References and Notes

  1. Materials and methods are available as supplementary materials in Science Online.
  2. Acknowledgments: We are grateful for technical assistance from the Gladstone Flow Core, including M. Cavrois, M. Gesner, and J. Tawney. We thank M. Spindler for the generous gift of the pSicoR-mCherry vector, and D. N. Levy for the NLENG1 and D116N plasmids. Efavirenz and azidothymidine were obtained from the AIDS Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, NIH. We thank L. Chavez, I. Munoz-Arias, and N. Byers for technical advice; T. Johnson for technical assistance; G. Howard and A. L. Lucido for editorial assistance; J.C.W. Carroll for graphic arts; and R. Givens and S. Cammack for administrative assistance. The data presented in this manuscript are tabulated in the main paper and the supplementary materials. A patent application has been filed (U.S. provisional serial number 61/511,023) regarding the identification of abortive HIV infection as a driver of CD4 T cell depletion involving the activation by IFI16 of caspase-1 in inflammasomes leading to pyroptosis, an intensely inflammatory form of programmed cell death. This work was made possible with support from the University of California San Francisco–Gladstone Center for AIDS Research (CFAR), an NIH-funded program (P30 AI027763), with additional funding from the Gladstone Institutes. This work was also supported by NIH R21 AI102782, U19 AI096113 (Martin Delaney CARE Collaboratory), and 1DP1036502 to W.C.G. and NIH P50 GM082250, P01 AI090935, and P50 GM081879 to N.J.K. N.J.K. is a Keck Young Investigator. K.M.M. is an A. P. Giannini Fellow supported by the A. P. Giannini Foundation.
View Abstract

Stay Connected to Science

Navigate This Article