Broad CTL response is required to clear latent HIV-1 due to dominance of escape mutations

Abstract

Despite antiretroviral therapy (ART), human immunodeficiency virus (HIV)-1 persists in a stable latent reservoir, primarily in resting memory CD4(+) T cells. This reservoir presents a major barrier to the cure of HIV-1 infection. To purge the reservoir, pharmacological reactivation of latent HIV-1 has been proposed and tested both in vitro and in vivo. A key remaining question is whether virus-specific immune mechanisms, including cytotoxic T lymphocytes (CTLs), can clear infected cells in ART-treated patients after latency is reversed. Here we show that there is a striking all or none pattern for CTL escape mutations in HIV-1 Gag epitopes. Unless ART is started early, the vast majority (>98%) of latent viruses carry CTL escape mutations that render infected cells insensitive to CTLs directed at common epitopes. To solve this problem, we identified CTLs that could recognize epitopes from latent HIV-1 that were unmutated in every chronically infected patient tested. Upon stimulation, these CTLs eliminated target cells infected with autologous virus derived from the latent reservoir, both in vitro and in patient-derived humanized mice. The predominance of CTL-resistant viruses in the latent reservoir poses a major challenge to viral eradication. Our results demonstrate that chronically infected patients retain a broad-spectrum viral-specific CTL response and that appropriate boosting of this response may be required for the elimination of the latent reservoir.

Figures

Extended Data Figure 1. CTL escape variants dominate the latent reservoir of CP-treated HIV-1-positive individuals, but not AP-treated individuals

Frequency of variants in Gag CTL epitopes in proviruses from resting CD4+ T cells. Results of all 25 patients tested are shown. Only optimal CTL epitopes relevant to each patient’s HLA type are listed in linear positional order on the X axis. Results from both Pacbio (left bar) and MiSeq (right bar) sequencing platforms are shown for each epitope. The absence of bars above a listed epitope indicates that only wild type sequences were detected. For each mutation in a CTL epitope, information regarding the effect of the mutation on CTL recognition from the Los Alamos National Laboratory (LANL) HIV Molecular Immunology Database or from ELISpot assays described herein was used to assign the mutation to one of the categories indicated at the bottom. See Methods for definitions of these categories.

Extended Data Figure 2. Characterization of CTL responses against HIV-1 Gag epitopes by interferon-γ ELISpot

Results of 7 patients tested are shown. The peptides tested are listed for each patient in each graph. Error bars represent s.e.m., n = 3.

Extended Data Figure 3. Partial Gag sequences from proviral DNA and outgrowth virus from resting CD4+ T cells of 8 CP-treated patients (following Fig. 1e)

CTL epitopes with no observed variation are highlighted in blue. Documented escape mutations (red shading), inferred escape mutations (yellow shading), diminished response (pink shading), susceptible form (green shading) or undetermined variations (gray shading) in relevant optimal epitopes are indicated. See Methods for definitions of these types of mutations.

Extended Data Figure 4. CD8+ T cells pre-stimulated with a mixture of consensus B Gag peptides eliminate autologous CD4+ T cells infected with autologous HIV-1 from resting CD4+ T cells

a, HIV-1 isolated from ART-treated individuals replicates as well as lab strain virus BaL. p24 values represent mean of 3 replicates. Error bars represent s.e.m., n = 3. b, CD8+ T cells are not stimulated after co-cultured with PHA-activated CD4+ T cells. c, A representative flow cytometric analysis of CTL-mediated killing after co-culture of infected CD4+ T cells with autologous CD8+ T cells. CTL activity is measured by the percentage of Gag-positive, CD8-negative cells after 3 days of co-culture relative to cultures without CD8+ T cells. d, Pre-stimulated CD8+ T cells eliminate autologous infected CD4+ T cells more efficiently than non-stimulated CD8+ T cells. All results were normalized to the CD4 only control group. Error bars represent s.e.m., n = 3. *: p<0.05; **: p<0.01; ***: p<0.001; ns: p>0.05, paired t test.

Extended Data Figure 5. The elimination of infected CD4+ T cells is mediated by direct killing by autologous CD8+ T cells

a, Killing of infected CD4+ T cells is enhanced by increased E:T ratios for both pre-stimulated and non-stimulated CD8+ T cells. b, Killing of the infected CD4+ T cells depends on direct cell-cell contact between CD4+ T cells and CTLs. All results were normalized to the CD4 only control group. Error bars represent s.e.m., n = 3. *: p<0.05; **: p<0.01; ***: p<0.001; ns: p>0.05, paired t test.

Extended Data Figure 6. Viral dynamics and depletion of CD4+ T cells in humanized mice

a, Viral dynamics in CP18-infected MIS(KI)TRG mice. CP18-derived MIS(KI)TRG mice were infected with autologous HIV-1. Plasma HIV-1 RNA levels were measured from day 0 to day 56. b, Depletion of CD4+ T cells in peripheral blood of HIV-1 BaL-infected mice. MIS(KI)TRG mice engrafted with fetal liver CD34+ cells were infected with HIV-1 BaL. CD4 to CD8 ratio in peripheral blood was measured by FACS from day 0 to day 29 after infection. c, Depletion of CD4+ T cells in spleen of HIV-1 BaL-infected mice. MIS(KI)TRG mice engrafted with fetal liver CD34+ cells were infected with HIV-1 BaL. CD4 to CD8 ratio in spleen was measured by FACS 20 days after infection. Median and P value from Mann-Whitney test are shown. d, Detection of cell-associated HIV-1 RNA in T cells and macrophages/monocytes. CD3+ and CD14+ human cells from HIV-1-infected MIS(KI)TRG mice from spleen and lung were purified by FACS. CD3−CD14− cells were also collected as controls. Cell associated HIV-1 RNA was quantified by gag-specific qPCR. *: p<0.05, unpaired t test.

Extended Data Figure 7. HIV-1 infection occurs in peripheral blood and tissues in humanized mice

a, Engraftment levels of MIS(KI)TRG mice with fetal liver or patient CD34+ cells. b, Memory CD4+ T cells are detected in MIS(KI)TRG mice after infection. MIS(KI)TRG mice were infected with HIV-1 BaL. Peripheral blood and indicated tissues from infected mice were collected at 20 dpi. Memory CD4+ T cells were determined by CD45RO staining. c, Total number of cell associated HIV-1 DNA in blood and tissues. DNA Leukocytes from peripheral blood or indicated tissues were isolated for the measurement of total amount of cell-associated HIV-1 DNA by real-time PCR. For b and c, median and P value from Mann-Whitney test are shown.

Extended Data Figure 8. Broad-spectrum cytotoxic T lymphocytes suppress in vivo infection of patient-derived humanized mice with autologous latent HIV-1

Generation of patient CP36-derived humanized mice was described in Fig. 4. Mice were infected with autologous viruses at 6 weeks old. CD8+ T cells from CP36 were pre-stimulated with the mixture of Gag peptides or left untreated for 6 days in vitro and were injected into mice by i.v. 9 days after infection. Plasma HIV-1 RNA and HIV-1 DNA in peripheral blood were measured by real-time PCR.

Figure 1. CTL escape variants dominate the latent reservoir of CP-treated but not AP-treated patients

a, Frequency of variants in Gag CTL epitopes in proviruses from resting CD4+ T cells. Representative results of 6 patients are shown. Only optimal CTL epitopes relevant to each patient’s HLA type are listed. Results from both Pacbio (left bar) and MiSeq (right bar) sequencing are shown. The effect on CTL recognition (denoted by color) is determined from information in the Los Alamos National Laboratory (LANL) HIV Molecular Immunology Database. b, CTL escape variants identified by sequencing are specific to HLA type. Frequencies of documented escape-associated variants in four well-characterized epitopes are shown for all 15 CP-treated patients. Median and P value from Mann-Whitney test are shown. c, Comparison of CTL escape variant frequency in proviruses between CP- and AP-treated patients. Only well-characterized epitopes are shown. Median and P value from Mann-Whitney test are shown. d, Characterization of CTL responses against HIV-1 Gag epitopes by interferon-γ ELISpot. The peptides tested are listed below the x-axis (black type, epitopes in which sequence variation was detected; blue type, no variation). The observed mutation is underlined in red, and CTL escape (defined by the lost of positive response) is denoted by * above the bar. The peptide concentration was 10 μg/ml. Error bars represent s.e.m., n = 3. e, Sequences in Gag CTL epitopes for proviral DNA and outgrowth virus from resting CD4+ T cells in CP39. CTL epitopes with no observed variation are highlighted in blue. Epitopes with documented escape mutations are shaded in red.

Figure 2. CD8+ T cells pre-stimulated with a mixture of Gag peptides eliminate autologous CD4+ T cells infected with autologous HIV-1 from resting CD4+ T cells

a, Pre-stimulated CD8+ T cells (sCD8) eliminate autologous infected CD4+ T cells more efficiently than unstimulated CD8+ T cells (uCD8). Each symbol represents the mean of 3 replicates. Median and P value from Mann-Whitney test are shown. b, sCD8 inhibit viral growth in autologous infected CD4+ T cells with higher efficacy than uCD8. c, sCD8 pre-stimulated by different viral peptides eliminate autologous CD4+ T cells infected with viruses derived from resting CD4+ T cells. For b and c, results are compared with CD4 only using paired t tests. Error bars represent s.e.m., n = 3. *: p<0.05; **: p<0.01; ***: p<0.001; ns: p>0.05.

Figure 3. CD8+ T cells targeting unmutated epitopes, not epitopes with identified escaped mutations, eliminate CTL escape variants

a, Frequency of variants in Gag CTL epitopes in proviruses from resting CD4+ T cells of CP36 and CP39. Epitopes tested with single peptide stimulation herein are denoted in colors (red or pink, epitopes with escape observed; blue, no escape observed). b, Epitope-specific CD8+ T cells proliferate significantly after single peptide stimulation. Only CD8+ cells are shown. Percentages of CFSElow, pentamer-positive cells are indicated for unstimulated cultures (uCD8) with or without IL2, for cultures stimulated with Gag peptide mixture (sCD8) and IL2, and for cultures stimulated with the indicated single peptides and IL2. Wild type versions of peptides were used for all single peptide stimulations. c, CD8+ T cell proliferative responses after single peptide stimulation. Only CD8+ cells are shown. Percentages of CFSElow cells are indicated. d, CD8+ T cells targeting unmutated epitopes, not epitopes with identified escaped mutations, eliminate autologous CD4+ T cells infected with CTL escape variants. uCD8: unstimulated CD8+ T cells; sCD8: Gag peptide mixture stimulated CD8+ T cells. Error bars represent s.e.m., n = 3. *: p<0.05; **: p<0.01; ***: p<0.001; ns: p>0.05, paired t test.

Figure 4. Broad-spectrum cytotoxic T lymphocytes suppress in vivo replication of HIV-1 from the latent reservoir of the same patients in patient-derived humanized mice

a, Experimental design. b and c, Efficient engraftment of patient CP18-derived hematopoietic cells in MIS(KI)TRG mice at week 6. Representative flow cytometry analysis (b) and summary (c) of human CD45+ cells, human T-lymphocyte and monocyte subsets. 11 out of 15 mice (enclosed in the rectangle) were used for HIV-1 infection. d, Correlation between frequency of peripheral human CD45+ cells (6 weeks after engraftment) and plasma HIV-1 RNA levels (14 days after infection). e, Depletion of peripheral CD4+ T cells after HIV-1 infection. *: p<0.05, paired t-test, n=11. f and g, Reduction of levels of plasma HIV-1 RNA and copies of peripheral blood HIV-1 DNA after injection of viral-specific CTLs. Filled: above detection limit; open: below detection limit. *: p<0.05, unpaired t test. h, Effect of CTL on the level of viral replication in vivo. The area under the curve of the viremia vs. time plot for each mouse in f and g before (AUC1) or after (AUC2) injection of CD8+ T cells was calculated to quantitatively represent viral replication over time. *: p<0.05, unpaired t test.