Broad CTL Response Is Required to Clear Latent HIV-1 Due to Dominance of Escape Mutations

LETTER doi:10.1038/nature14053

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

Kai Deng1, Mihaela Pertea1,2, Anthony Rongvaux3, Leyao Wang4, Christine M. Durand1, Gabriel Ghiaur5, Jun Lai1, Holly L. McHugh1, Haiping Hao6, Hao Zhang7, Joseph B. Margolick7, Cagan Gurer8, Andrew J. Murphy8, David M. Valenzuela8, George D. Yancopoulos8, Steven G. Deeks9, Till Strowig3, Priti Kumar10, Janet D. Siliciano1, Steven L. Salzberg2,11, Richard A. Flavell3,12, Liang Shan3 & Robert F. Siliciano1,13

Despite antiretroviral therapy (ART), human immunodeficiency virus To investigate CTL escape variants in the latent reservoir, we deep (HIV)-1 persists in a stable latent reservoir1,2, primarily in resting sequenced the proviral HIV-1 DNA in resting CD41 T cells from 25 memory CD41 T cells3,4. This reservoir presents a major barrier to patients (Extended Data Table 1). Among them, 10 initiated ART during the cure of HIV-1 infection. To purge the reservoir, pharmacolog- the acute phase (AP; within 3 months of infection) while the other 15 ical reactivation of latent HIV-1 has been proposed5 and tested both initiated ART during the chronic phase (CP) of infection. The sequenc- in vitro and in vivo6–8. A key remaining question is whether virus- ing was focused on Gag because it is an important target of the CTL re- specific immune mechanisms, including cytotoxic T lymphocytes sponse23 and is highlyconserved, which facilitates thedetection ofescape (CTLs), can clear infected cells in ART-treated patients after latency variants. Our data show that previously documented CTL escape var- is reversed. Here we show that there is a striking all or none pattern iants completely dominate the viral reservoirs of nearly all CP-treated for CTL escape mutations in HIV-1 Gag epitopes. Unless ART is patients (Extended Data Fig. 1 and Supplementary Table 1). This trend is started early, the vast majority (.98%) of latent viruses carry CTL especially obvious for several well characterized CTL epitopes: the human escape mutations that render infected cells insensitive to CTLs directed leukocyte antigen (HLA)-A2-restricted epitope SLYNTVATL(SL9), the at common epitopes. To solve this problem, we identified CTLs that HLA-A3-restricted epitope RLRPGGKKK (RK9) and the HLA-B57/58- could recognize epitopes from latent HIV-1 that were unmutated in restricted epitope TSTLQEQIGW (TW10) (Fig. 1a and Extended Data every chronically infected patient tested. Upon stimulation, these Fig. 1, highlighted in coloured boxes.). In these epitopes, close to 100% CTLs eliminated target cells infected with autologous virus derived of the sequences harboured escape mutations. Comparison of muta- from the latentreservoir, both in vitro andin patient-derived human- tion frequencies between HLA allele-relevant and -irrelevant epitopes ized mice. The predominance of CTL-resistant viruses in the latent in CP-treated patients suggests that the CTL escape mutations identified reservoir poses a major challenge to viral eradication. Our results are specific to each patient’s HLA type (Fig. 1b). By contrast, except for demonstrate that chronically infected patients retain a broad-spectrum SL9 from patient AP01 and RK9 from patient AP08, few if any CTL viral-specific CTL response and that appropriate boosting of this escape mutations were archived in AP-treated patients (Fig. 1c and Ex- response may be required for the elimination of the latent reservoir. tended Data Fig. 1). The striking difference between AP- and CP-treated HIV-1 establishes latent infection in resting CD41 T cells3,4. Recent patients (Fig. 1c) indicates that, unless treatment is initiated within the efforts to eradicate HIV-1 infection have focused on reversing latency first several months of infection, the latent reservoir becomes almost without global T-cell activation5. However, inducing HIV-1 gene expres- completely dominated by variants resistant to dominant CTL responses. sion in latently infected cells is not sufficient to cause the death of these To confirm that variants detected at high frequency in the latent res- cells if they remain in a resting state9. Boosting HIV-1-specific immune ervoir represent functional CTL escape mutants, cells from seven CP- responses, including CTL responses, may be required for clearance of treated subjects were tested for reactivity to synthetic peptides representing the latent reservoir9. CTLs have a significant role in suppressing HIV-1 wild-type and mutant versions of the relevant epitopes. As expected, replication in acute infection10–14. Because of this strong selective pres- there were only minimal responses to previously documented CTL escape sure, HIV-1 quickly acquires mutations to evade CTL recognition12,13,15–18. mutants by patient CD81 T cells, and no de novo response was detected CTL escape has been studied primarily through the analysis of plasma (Fig. 1d and Extended Data Fig. 2). In contrast, all tested subjects retained virus12,13,16,18–20, and CTL-based vaccines have been designed based on a strong response to peptides representing the wild-type epitopes, sug- conserved epitopes21,22. A systematic investigation of CTL escape in the gesting that the wild-type virus was initially transmitted, with subsequent latent reservoir will be of great importance to the ongoing CTL-based evolution of CTL escape variants. Most HIV-1 proviruses detected in virus eradication efforts, because latent HIV-1 probably represents the patients are defective24. Therefore, to determine whether these CTL major source of viral rebound after treatment interruption. Earlier escape variants can be reactivated and lead to viral rebound if therapy studies have suggested the presence of CTL escape mutations in pro- is stopped, we isolated replication-competent viruses from the latent viral DNA15,17, but it remains unclear to what extent the latent reservoir reservoirs of nine CP-treated patients. We found that all the dominant in resting CD41 T cells is affected by CTL escape, whether mutations CTL escape mutations that had been identified in proviruses in resting detected in proviral DNA are representative of the very small fraction CD41 T cells were alsopresent in the replication-competent viruses that of proviruses that are replication competent, and, most importantly, grew out after T-cell activation (Fig. 1e and Extended Data Fig. 3), indicat- whether the CTL response can recognize and clear infected cells after ing that these CTL escape variants not only dominate the population of latency is reversed. proviruses, but can also be released and replicate once latency is reversed.

1Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, USA. 2Center for Computational Biology, McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, USA. 3Department of Immunobiology, Yale University School of Medicine, New Haven, Connecticut 06510, USA. 4Department of Chronic Disease Epidemiology, Yale School of Public Health, New Haven, Connecticut 06510, USA. 5Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, USA. 6Deep Sequencing and Microarray Core, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, USA. 7Department of Molecular Microbiology and Immunology, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland 21205, USA. 8Regeneron Pharmaceuticals Inc., Tarrytown, New York 10591, USA. 9Department of Medicine, University of California, San Francisco, San Francisco, California 94110, USA. 10Department of Medicine, Yale University School of Medicine, New Haven, Connecticut 06510, USA. 11Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, USA. 12Howard Hughes Medical Institute, New Haven, Connecticut 06510, USA. 13Howard Hughes Medical Institute, Baltimore, Maryland 21205, USA.

P P P P a 100 b = 0.001 = 0.157 = 0.032 = 0.121 100 50 AP10 A*02:01 0 GK9 EV9SL9 TV9 EI8GLY9 DL9 FK10 80 100 HLA allele- 60 irrelevant patients 50 CP18 A*02:01 0 WF9 SL9 TV9 TL9 HA9 PY9 VI9 FK10 40 HLA allele- 100 relevant patients 50 AP12 20 A*03:01 Frequency of variants (%) 0 KK9 RK9 SV9 TL9 HA9 GL9 0 100 TW10(T242N) IW9(I147L) SL9(Y79F/T84V) RK9(K28Q) 50 CP39 Variation in well characterized epitopes A*03:01 0 c KK9 RK9 LY9 SV9 TL9 HA9 GL9 AP-treated patients P < 0.001 100 SL9 100 Frequency of variants (%) 50 AP07 B*58:01 80 0 AP01 AP04 AP05 AP08 AP10 AP11 RY11 VL8 TW10 YL9 QW9 60 RK9 TW10 KK10 100 40 CP12 50 B*57:01 20 AP08 AP01 AP09 AP12 AP07 AP03

0 Frequency of CTL LY9 IW9 KF11 TW10 QW9 escape variants (%) CP-treated patients 0 AP CP SL9 IW9 CTL epitopes in HIV-1 Gag Wild type Documented escape Diminished response Mutation type not determined CP18 CP36 CP40 CP42 CP47 CP49 CP05 CP12 RK9 TW10 CTL escape variant d CP12 CP18 CP39 CP38 CP39 CP42 CP50 CP05 CP12 CP19 CP31 250 350 300 300 250 200 250 200 e Consensus B 150 200 150 CP39 provirus 150 100 100 CP39 outgrowth * 100 * 50 * * * 50 * 50 RK9KK9 LY9 SV9 0 0 0 Consensus B

SFC per million PBMCs CP39 provirus CP39 outgrowth

ISPRTLNAW SLYNTVATLSLFNTVAVL LYNTVATLYLFNTIAALF TL9 HA9 GL9 LSPRTLNAW QATQEVKNW WASRELERFTLNAWVKVV RLRPGGKKKRLRPGGKKS TPQDLNTMLGPGHKARVL TSTLQEQIGWTSNLQEQIGWQASQEVKNWKAFSPEVIPMF Unmutated CTL epitope Documented escape HIV-1 Gag CTL epitopes Figure 1 | CTL escape variants dominate the latent reservoir of CP-treated Medians and P values from Mann–Whitney test are shown. d, Characterization but not AP-treated patients. a, Frequency of variants in Gag CTL epitopes of CTL responses against HIV-1 Gag epitopes by interferon-c ELISpot. The in proviruses from resting CD41 T cells. Representative results from six peptides tested are listed below the x-axis (black type, epitopes in which patients are shown. Only optimal CTL epitopes relevant to each patient’s HLA sequence variation was detected; blue type, no variation). The observed type are listed. Results from both PacBio (left bar) and MiSeq (right bar) mutation is underlined in red, and CTL escape (defined by the absence of sequencing are shown. The effect on CTL recognition (denoted by colour) is positive response after mutation) is denoted by an asterisk above the bar. The determined from information in the Los Alamos National Laboratory (LANL) peptide concentration was 10 mgml21. PBMCs, peripheral blood mononuclear HIV Molecular Immunology Database. b, CTL escape variants identified by cells; SFC, spot-forming cell. Error bars represent standard error of the mean sequencing are specific to HLA type. Frequencies of documented escape- (s.e.m.), n 5 3. e, Sequences in Gag CTL epitopes for proviral DNA and associated variants in four well characterized epitopes are shown for all 15 outgrowth virus from resting CD41 T cells in patient CP39. CTL epitopes with CP-treated patients. Medians and P values from Mann–Whitney test are no observed variation are highlighted in blue. Epitopes with documented shown. c, Comparison of CTL escape variant frequency in proviruses between escape mutations are shaded in red. CP- and AP-treated patients. Only well characterized epitopes are shown.

We next asked whether the host CTL response could recognize and by Gag peptides generally had thehighest activity.Together, theseresults eliminate the cells infected with these escape variants. We infected acti- demonstrate that chronically infected patients retain CTL clones that vated CD41 T cells from these patients with autologous, replication- can recognize and eliminate autologous infected CD41 T cells, despite competent virus derived from the latent reservoir (Extended Data Fig. 4a). the presence of CTL escape mutations in dominant epitopes. However, The infected cells were then co-cultured with autologous CD81 T cells, these clones require stimulation with antigen for optimal activity. either unstimulated or pre-stimulated, to assess HIV-1-specific cyto- To characterize further which CTL population contributed to the lytic activity. Non-specific activation of CD81 T cells was not observed elimination of cells infected by CTL escape variants, we compared the after co-culture with phytohaemagglutinin (PHA)-activated CD41 T killing activity of two specific CTL populations: the population that targets cells (Extended Data Fig. 4b). From all 13 CP-treated subjects tested, epitopes in which escape has been identified and the one that targets CD81 T cells pre-stimulated by a Gag peptide mixture efficiently killed unmutated epitopes (Fig. 3a). CD81 T cells from patients CP36 and autologous infected CD41 T cells (median 61% elimination), while CP39 were pre-stimulated with interleukin (IL)-2 and different synthetic unstimulated CD81 T cells from most subjects had significantly less peptides representing the wild-type forms of the relevant epitopes. After effect (median 23% elimination) (Fig. 2a and Extended Data Fig. 4c, d). incubation for 6 days, each CTL population exhibited significant pro- CD81 T cells from 7/7 healthy donors completely failed to eliminate liferation compared with no treatment or IL-2 alone (Fig. 3b, c). Pentamer autologous infected cells (Fig. 2a), confirming that the observed killing staining for three available epitopes revealed that the number of epitope- was HIV-1 specific. The killing effect was enhanced by increasing the specific CD81 T cells increased dramatically after stimulation with wild- effector to target ratio (Extended Data Fig. 5a), and was cell–cell contact type peptides (Fig. 3b). After co-culture with autologous target cells dependent (Extended Data Fig. 5b). When the co-culture was main- infected with latent reservoir-derived viruses, CTLs targeting unmu- tained over timeinthe absence of ART, viral replication was significantly tated epitopes clearly showed stronger cytolytic activity than the IL-2 reduced, but not completely inhibited by pre-stimulated CD81 T cells only controls, while CTLs targeting epitopes with identified escaped (Fig. 2b). We found that peptide mixtures from other HIV-1 proteins mutations showed no significant killing (Fig. 3d). CTLs pre-stimulated (Nef, Tat, Rev and Env) could also boost CTL responses and facilitate by the Gag peptide mixture exhibited stronger killing than all single- the elimination of infected cells (Fig. 2c), and that CTLs pre-stimulated peptide-stimulated populations (Fig. 3d).

1 a NS c CP32 Figure 2 | CD8 T cells pre-stimulated with a mixture of Gag peptides NS NS 1 *** NS 100 ** ** * * * eliminate autologous CD4 T cells infected with autologous HIV-1 from *** * 60 resting CD41 T cells. a, Pre-stimulated CD81 T cells (sCD81) eliminate 1 1 20 T cells (%) 100 autologous infected CD4 T cells more efficiently than unstimulated CD8 + 0 1 CP36 T cells (uCD8 ). Each symbol represents the mean of three replicates. Medians

CD4 1 + 100 NS and P values from Mann–Whitney test are shown. b, sCD8 cells inhibit viral * * 10 CD4+ only 60 * * growth in autologous infected CD41 T cells with higher efficacy than uCD81 + + CD4 + uCD8 T cells (%) ** 1 + 20 + + cells. p24, HIV-1 capsid (core) protein. c, sCD8 cells pre-stimulated by CD4 + sCD8 0 1 Residual Gag CD4

1 + CP37 different viral peptides eliminate autologous CD4 T cells infected with viruses 100 1 1 HIV-1+ patients Healthy donors *** derived from resting CD4 T cells. b, c, Results were compared with CD4 only 60 b *** using paired t-tests. Error bars represent s.e.m., n 5 3. *P , 0.05, **P , 0.01, CP26 CP27 CP28 *** *** 20 *** *** 800 CD4+ only ***P , 0.001, NS, not significant (P . 0.05). CD4+ + uCD8+ 800 800 Residual Gag 0 600 + + CD4 + sCD8 600 600 CP39 ) 400 ** 100 *** –1 400 400 ** 25 *** ** ** reported mouse system named MISTRG . Whereas the previously 200 * ** 60 *** 200 200 *** 0 0 0 20 reported MISTRG mice bear a bacterial artificial chromosome (BAC) 036 036 6 03 0 (KI) CP31 CP36 CP39 + + + + + + transgene encoding human SIRP-a, the newly generated MIS TRG 500 250 1200 + only uCD8 400 200 Tat CD8 mice harbour a knock-in replacement of the endogenous mouse Sirpa

HIV-1 p24 (ng ml Nef CD8 800 CD4 Gag CD8 Rev CD8 Env CD8 300 150 * gene with a humanized version. With humanization by knock-in replace- 200 ***100 ** *** 400 2/2 2/2 100 50 ** ment of the Csf1, Csf2, Il3, Tpo and Sirpa genes in the Rag2 Il2rg 0 0 0 (KI) 036 036 036 genetic background, MIS TRG mice are highly permissive for human Days post-infection haematopoiesis and support the reconstitution of robust human lymphoid and myelomonocytic systems. With the demonstrated development To test whether CTLs that recognize unmutated viral epitopes can offunctional T lymphocytes and monocytes/macrophages, MIS(KI)TRG inhibit HIV-1 replication and clear infected cells in vivo, we generated mice provide a useful humanized mouse host for HIV-1 infection stud- patient-derived humanized mice using an improved version of a recently ies. Bone marrow biopsies were obtained from study participants and

1 100 Figure 3 | CD8 T cells targeting unmutated a 50 CP36 epitopes, not epitopes with identified escape 0 mutations, eliminate CTL escape variants.

TV9 a KK9 RK9 WF9 SL9 HA9 QW9 , Frequency of variants in Gag CTL epitopes in 100 EW10 proviruses from resting CD41 T cells from patients 50 CP39 CP36 and CP39. Epitopes tested with single- 0 peptide stimulation are denoted in colours (red KK9 RK9 LY9 SV9 TL9 HA9 GL9 Frequency of variants (%) or pink, epitopes with escape observed; blue, no CTL epitopes in HIV-1 Gag escape observed). b, Epitope-specific CD81 b CP36 SL9 CP39 RK9 CP39 GL9 c CP36 CP39 T cells proliferate markedly after single-peptide stimulation. Only CD81 T cells are shown. uCD8+ + IL-2 uCD8+ + IL-2 CFSElow 1.4% CFSE 0.9% CFSElow 2.3% low CFSE CFSE Percentages of carboxyfluorescein succinimidyl 0.01% low low 0.01% 0.02% uCD8+ ester (CFSE)low, pentamer-positive cells are indicated for unstimulated cultures (uCD81) with 4.5% 5.9% or without IL-2, for cultures stimulated with Gag 1 TV9 LY9 peptide mixture (sCD8 ) and IL-2, and for cultures CFSElow 6.3% CFSElow 5.7% CFSElow 6.2% CFSElow CFSElow + stimulated with the indicated single peptides and 0.01% 0% 0.01% uCD8 +IL-2 IL-2. Wild-type versions of peptides were used for 1 9.1% 20.6% all single-peptide stimulations. c,CD8 T-cell proliferative responses after single-peptide WF9 TL9 1 CFSElow 17.6% CFSElow 14.9% CFSElow 15.5% CFSElow CFSElow stimulation. Only CD8 T cells are shown. 0.06% 1 0.08% 0.07% sCD8+ Percentages of CFSElow cells are indicated. d, CD8 T cells targeting unmutated epitopes, not epitopes 14.2% 16.5% with identified escaped mutations, eliminate 1 EW10 autologous CD4 T cells infected with CTL escape CFSElow 14.9% CFSE 16.7% CFSElow 16.6% low Single- CFSElow + 0.22% variants. Error bars represent s.e.m., n 5 3. 0.41% 0.46% peptide- *P , 0.05, **P , 0.01, ***P , 0.001, NS, not stimulated + CD8+ significant (P . 0.05), paired t-test.

CD8 12.7% Pentamer CFSE CFSE d CP36 ** CP39 * NS** * 120 120 NS NS NS 100 *** 100 ** T cells (%) T cells (%) + 80 + 80 CD4

CD4 60 60 + +

40 40

20 20 Residual Gag Residual Gag 0 0 + + + +

+ only -CD8 -CD8 + only -CD8 -CD8 uCD8+ + IL-2 sCD8 uCD8+ + IL-2 sCD8 SL9 TV9-CD8WF9-CD8 RK9 LY9 TL9-CD8GL9-CD8 CD4 EW10 CD4 uCD8 uCD8

1 (KI) purified CD34 cells were used to reconstitute the MIS TRG mice. CD8+ T cell a Peripheral Weinfected these patient-derived humanized mice with primary HIV-1 blood In vitro antigen-specific stimulation isolates grown from resting CD41 T cells from the same patient and 1 then evaluated the antiviral effect of autologous CD8 T cells (Fig. 4a). In vitro (KI) 1 latent HIV-1 outgrowth Adoptive MIS TRG mice engrafted with bone marrow CD34 cells from patient transfer Infection CP18 successfully developed human T-lymphocyte and monocyte/ Resting CD4+ T cell macrophage subsets (Fig. 4b, c), which were sufficient to support HIV-1 Patient-derived 14 dpi humanized mice infection (Fig. 4d). Plasma HIV-1 RNA levels peaked 20–30 days after Newborn 6–8-week-old 1 Bone Bone marrow mice mice infection (Extended Data Fig. 6a). Depletion of CD4 T cells was Patient marrow CD34+ cell clearly evident 12 days after infection in peripheral blood and spleen Blood sample collection b d 107 (Fig. 4e and Extended Data Fig. 6b, c). Cell-associated HIV-1 RNA was Leukocytes hCD45+ hCD3+ 5 detected in both T cells and macrophages/monocytes (Extended Data 34.7 20.6 44.5 10 Fig. 6d). Viral infection was also observed in various tissues in which a 103 1 44.7 101 large number of memory CD4 T cells were detected (Extended Data P = 0.002 1 64.3 (copies per ml) 51.5 Plasma HIV-1 RNA 10–1 hCD8 hCD3 Fig. 7). In control mice or mice that received autologous patient CD8 mCD45 hCD45 hCD14 hCD4 020406080 T cells pre-stimulated with a peptide representing the unmutated dom- hCD45+ (week 6) (%) inant SL9 epitope, levels of plasma HIV-1 RNA and proviral DNA in c Week 6 Week 6 e 100 60 *

80 peripheral blood continued to increase from day 14 to day 29 after + 80 * cells 1 cells 60 + infection (Fig. 4f). In sharp contrast, mice that received CD8 T cells 60 40 + cells (%)

+ 40 pre-stimulated with unmutated epitopes (Gag mix or WF9) had a sig- 40 20

Percentage of 20 20 nificantly lower level of viral replication (Fig. 4g, h). Dramatic decreases total CD45 hCD45 cells in hCD3 in plasma HIV-1 RNA of 100- to 1,000-fold were observed in all three 0 0 Percentage hCD4 0 1 hCD45 CD14 CD3 CD8 CD4 –14 –7 0 7 14 mice that received CD8 T cells pre-stimulated with the mixture of Days post-infection Gag peptides including dominant and subdominant epitopes. Two of f CD8 injection ghCD8 injection 8 × 106 three mice had undetectable levels of plasma HIV-1 RNA and proviral * 108 DNA in peripheral blood measured at three time points (Fig. 4g). We 6 × 106 NS 106 * 6 performed the same experiments using patient CP36-derived human- * Mix-1 4 × 10 104 Mix-2 ized mice and a reductionof peripheral HIV-1RNA and DNA levels was SL9-1 Mix-3 2 × 106 2 SL9-2 also observed in mice that received CP36 CD81 T cells pre-stimulated 10 WF9-1 AUC (HIV-1 RNA) No CD8-1 WF9-2 0 (copies per ml) 100 No CD8-2 WF9-3 with the mixture of Gag peptides (Extended Data Fig. 8). Since the post- Plasma HIV-1 RNA 4 3 × 10 * engraftment lifespan of MIS(KI)TRG mice is only 10–12 weeks25,wewere 01020300102030 4 only able to investigate the acute phase of HIV-1 infection and dem- 10 2 × 104 1 NS onstrate the in vivo functionality of patient CD8 T cells. Future devel- Mix-1 (KI) 102 * Mix-2 1 × 104 * opments of the MIS TRG model will prolong the post-engraftment Mix-3 SL9-1

WF9-1 AUC (HIV-1 DNA) lifespan of these mice and allow studies of the establishment and clear- SL9-2 0

HIV-1 DNA 0 WF9-2 10 No CD8-1 AUC1 AUC2 ance of the HIV-1 latent reservoir in vivo. Together, our in vitro and in No CD8-2 WF9-3 vivo experiments demonstrate that only CTL clones targeting unmu- (copies per 100 μ l blood) 01020300102030 Mix and WF9 SL9 and no CD8 tated epitopes are effective against cells infected with the viral variants Days post-infection that are likely to represent the major source of rebound HIV-1 after Figure 4 | Broad-spectrum CTLs suppress in vivo replication of HIV-1 from reversal of latency. the latent reservoir of the same patients in patient-derived humanized mice. The seeding of the HIV-1 latent reservoir starts just a few days after a, Experimental design. dpi, days post-infection. b, c, Efficient engraftment infection26, before the development of a robust CTL response14. This is of patient CP18-derived haematopoietic cells in MIS(KI)TRG mice at week 6. consistent with our finding that patients who initiated treatment early, Representative flow cytometry analysis (b) and summary (c) of human CD451 in the acute infection stage, have few if any CTL escape variants archived cells, human T-lymphocyte and monocyte subsets. Eleven out of fifteen mice (enclosed in the rectangle) were used for HIV-1 infection. h, human; in the latent reservoir. However, if treatment was initiated in chronic 1 infection, CTL escape variants became dominant in the latent reservoir, m, mouse. d, Correlation between frequency of peripheral human CD45 cells (6 weeks after engraftment) and plasma HIV-1 RNA levels (14 days after indicating a complete replacement of the initially established ‘wild-type’ infection). e, Depletion of human CD41 T cells in peripheral blood after HIV-1 reservoir. The mechanism behind this replacement warrants further infection. *P , 0.05, paired t-test, n 5 11. f, g, Reduction of levels of plasma investigation, but probably reflects the dynamic nature of the reservoir HIV-1 RNA and copies of peripheral blood HIV-1 DNA after injection of in untreated infection. In any event, the overwhelming presence of viral-specific CTLs. Filled symbols, above detection limit; open symbols, below escape variants in the latent reservoir of chronic patients certainly pres- detection limit. *P , 0.05, unpaired t-test. h, Effect of CTLs on the level of viral ents an additional barrier to eradication efforts. The striking difference replication in vivo. The area under the curve (AUC) of the viraemia versus time plot for each mouse in f (n 5 3) and g (n 5 6) before (AUC1) or after between AP- and CP-treated patients presents another argument for 1 early treatment of HIV-1 infection; early treatment not only reduces (AUC2) injection of CD8 T cells was calculated to represent quantitatively the size of the latent reservoir27, but also alters the composition of the viral replication over time. Mouse SL9-2 was excluded from AUC analysis owing to early death. Error bars represent s.e.m. *P , 0.05, unpaired t-test. reservoir, as shown here, in a way that may enhance the efficacy of poten- tial CTL-based eradication therapies. The hierarchy of HIV-1-specific CTL response in acute infection Therefore, directingCTLresponses to unmutated viral epitopes isessen- appears to have an important role in initial viral suppression, as dem- tial to clear latent HIV-1. Owing to bias in antigen presentation or onstrated by the fact that certain immunodominant CTL populations recognition29, common vaccinationstrategies will probably re-stimulate are frequently linked to lower set point viraemia later in infection17,28. immunodominant CTL clones that do not kill infected cells after the These immunodominant responses in acute infection have been iden- reversal of latency. Stimulation of CTL responses with viral peptides tified asthemajor selectionforce drivingthe development of CTLescape circumvents antigen processing and is able to elicit broad-spectrum mutations13,20. Here we show that these immunodominant response- CTL responses against unmutated regions of viral proteins. Our study driven mutations are not only archived in the latent reservoir, but also suggests that latent HIV-1 can be eliminated in chronically infected in fact dominate the latent provirus population in CP-treated patients. patients despite the overwhelming presence of CTL escape variants.

Future directions in therapeutic vaccine design need to focus on boost- 19. Henn, M. R. et al. Whole genome deep sequencing of HIV-1 reveals the impact of 30 early minor variants upon immune recognition during acute infection. PLoS ing broad CTL responses, as also reported elsewhere , and/or manip- Pathog. 8, e1002529 (2012). ulating immunodominance. 20. Liu, M. K. et al. Vertical T cell immunodominance and epitope entropy determine HIV-1 escape. J. Clin. Invest. 123, 380–393 (2013). Online Content Methods, along with any additional Extended Data display items 21. Rolland, M., Nickle, D. C. & Mullins, J. I. HIV-1 group M conserved elements vaccine. and Source Data, are available in the online version of the paper; references unique PLoS Pathog. 3, e157 (2007). to these sections appear only in the online paper. 22. Le´tourneau, S. et al. Design and pre-clinical evaluation of a universal HIV-1 vaccine. PLoS ONE 2, e984 (2007). Received 16 December 2013; accepted 11 November 2014. 23. Yu, X. G. et al. Consistent patterns in the development and immunodominance 1 Published online 7 January 2015. of human immunodeficiency virus type 1 (HIV-1)-specific CD8 T-cell responses following acute HIV-1 infection. J. Virol. 76, 8690–8701 1. Siliciano, J. D. et al. Long-term follow-up studies confirm the stability of the latent (2002). reservoir for HIV-1 in resting CD41 T cells. Nature Med. 9, 727–728 (2003). 24. Ho, Y. C. et al. Replication-competent noninduced proviruses in the latent reservoir 2. Strain, M. C. et al. Heterogeneous clearance rates of long-lived lymphocytes increase barrier to HIV-1 cure. Cell 155, 540–551 (2013). infected with HIV: intrinsic stability predicts lifelong persistence. Proc. Natl Acad. 25. Rongvaux, A. et al. Development and function of human innate immune cells in a Sci. USA 100, 4819–4824 (2003). humanized mouse model. Nature Biotechnol. 32, 364–372 (2014). 3. Finzi, D. et al. Identification of a reservoir for HIV-1 in patients on highly active 26. Whitney, J. B. et al. Rapid seeding of the viral reservoir prior to SIV viraemia in antiretroviral therapy. Science 278, 1295–1300 (1997). rhesus monkeys. Nature 512, 74–77 (2014). 4. Chun, T. W. et al. Quantification of latent tissue reservoirs and total body viral load 27. Ananworanich, J. et al. Impact of multi-targeted antiretroviral treatment on gut in HIV-1 infection. Nature 387, 183–188 (1997). T cell depletion and HIV reservoir seeding during acute HIV infection. PLoS ONE 7, 5. Richman, D. D. et al. The challenge of finding a cure for HIV infection. Science 323, e33948 (2012). 1304–1307 (2009). 28. Goulder, P. J. et al. Evolution and transmission of stable CTL escape mutations in 6. Archin, N. M. et al. Expression of latent HIV induced by the potent HDAC inhibitor HIV infection. Nature 412, 334–338 (2001). suberoylanilide hydroxamic acid. AIDS Res. Hum. Retroviruses 25, 207–212 29. Le Gall, S., Stamegna, P. & Walker, B. D. Portable flanking sequences modulate CTL (2009). epitope processing. J. Clin. Invest. 117, 3563–3575 (2007). 7. Contreras, X. et al. Suberoylanilide hydroxamic acid reactivates HIV from latently 30. Hansen, S. G. et al. Cytomegalovirus vectors violate CD81 T cell epitope recognition infected cells. J. Biol. Chem. 284, 6782–6789 (2009). paradigms. Science 340, 1237874 (2013). 8. Archin, N. M. et al. Administration of vorinostat disrupts HIV-1 latency in patients on antiretroviral therapy. Nature 487, 482–485 (2012). Supplementary Information is available in the online version of the paper. 9. Shan, L. et al. Stimulation of HIV-1-specific cytolytic T lymphocytes facilitates Acknowledgements We thank all study participants. We thank J. Blankson for critical elimination of latent viral reservoir after virus reactivation. Immunity 36, 491–501 advice to the project; L. Alston and R. Hoh for coordinating patient recruitment; (2012). J. Alderman, C. Weibel and E. Henchey for technical assistance in the animal study. We 10. Walker, B. D. et al. HIV-specific cytotoxic T lymphocytes in seropositive individuals. thank the National Institutes of Health (NIH) AIDS Reagent Program for providing HIV-1 Nature 328, 345–348 (1987). consensus B peptides. R.F.S. is supported by the Howard Hughes Medical Institute, by 11. Koup, R. A. et al. Temporal association of cellular immune responses with the initial the Martin Delaney CARE and DARE Collaboratories (NIH grants AI096113 and control of viremia in primary human immunodeficiency virus type 1 syndrome. 1U19AI096109), by an ARCHE Collaborative Research Grant from the Foundation for J. Virol. 68, 4650–4655 (1994). AIDS Research (amFAR 108165-50-RGRL), by the Johns Hopkins Center for AIDS 12. Borrow, P. et al. Antiviral pressure exerted by HIV-1-specific cytotoxic T Research (P30AI094189), and by NIH grant 43222. L.S. is supported by NIH grant T32 lymphocytes (CTLs) during primary infection demonstrated by rapid selection of AI07019. R.A.F. is supported by the Bill and Melinda Gates Foundation and the Howard CTL escape virus. Nature Med. 3, 205–211 (1997). Hughes Medical Institute. 13. Goonetilleke, N. et al. The first T cell response to transmitted/founder virus contributes to the control of acute viremia in HIV-1 infection. J. Exp. Med. 206, Author Contributions K.D., L.S., R.A.F. and R.F.S. conceived and designed the research 1253–1272 (2009). studies; K.D., J.L., H.Z., J.B.M. and L.S. performed the in vitro experiments; K.D., A.R., L.W., 14. McMichael, A. J., Borrow, P., Tomaras, G. D., Goonetilleke, N. & Haynes, B. F. The C.G., A.J.M., D.M.V., G.D.Y., T.S., P. K. and L.S. performed animal experiments; C.M.D., immune response during acute HIV-1 infection: clues for vaccine development. G.G., H.L.M. and S.G.D. provided patient samples; K.D., M.P., L.W., H.H., J.D.S., S.L.S., L.S. Nature Rev. Immunol. 10, 11–23 (2010). and R.F.S. analysed data; K.D., L.S. and R.F.S. wrote the manuscript. 15. Phillips, R. E. et al. Human immunodeficiency virus genetic variation that can escape cytotoxic T cell recognition. Nature 354, 453–459 (1991). Author Information Reprints and permissions information is available at 16. Koup, R. A. Virus escape from CTL recognition. J. Exp. Med. 180, 779–782 (1994). www.nature.com/reprints. The authors declare competing financial interests: details 17. Goulder, P. J. et al. Late escape from an immunodominant cytotoxic T-lymphocyte are available in the online version of the paper. Readers are welcome to comment on response associated with progression to AIDS. Nature Med. 3, 212–217 (1997). the online version of the paper. Correspondence and requests for materials should 18. Goulder, P. J. & Watkins, D. I. HIV and SIV CTL escape: implications for vaccine be addressed to R.F.S. ([email protected]), L.S. ([email protected]) or design. Nature Rev. Immunol. 4, 630–640 (2004). R.A.F. ([email protected]).

METHODS PCR amplicons were then purified by gel extraction and sequenced by regular Human subjects. Peripheral blood or bone marrow for the isolation of primary Sanger sequencing. 1 1 1 In vitro HIV-1 infection. PBMCs from HIV-1-infected patients and healthy donors CD4 , CD8 T cells or CD34 cells was obtained from 30 HIV-1-infected patients 21 21 (Extended Data Table 1) and 7 healthy adult volunteers. All patients had been on were stimulated by adding 0.5 mgml PHA and IL-2 (100 U ml ) to basal media ART for at least 2 years and had maintained undetectable plasma HIV-1 RNA levels (RPMI with 10% heat-inactivated fetal bovine serum and antibiotics) for 3 days before isolation of CD41 T cells. Each patient’s activated CD41 T cells were infected (,50 copies per ml) for at least 1 year before study. Ten AP-treated patients were 1 recruited from the OPTIONS cohort at the University of California San Francisco withviruses recovered from the same patient’s resting CD4 T cells. Healthy donors’ CD41 T cells were infected with a laboratory strain virus, BaL. The virus concen- (UCSF). This study was approved by the Johns Hopkins Internal Review Board 21 and by the UCSF Committee on Human Research. Written informed consent was tration used in infection was equivalent to the p24 concentration of 200 ng ml . provided by all study participants. HLA typing for each patient was performed by All infections were performed by centrifugation of target cells with virus at 1,200g the Johns Hopkins University Immunogenetics Laboratory. for 2 h. 1 Sample preparation for deep sequencing. Peripheral blood mononuculear cells Stimulation of CD8 T cells. PBMCs from CP-treated patients were cultured in 21 (PBMCs) were isolated from whole blood by Ficoll gradient separation. CD41 T the presence of IL-2 (100 U ml ) with a mixture of consensus B Gag (or Nef, Rev, 21 cells were purified from PBMCs by negative selection using CD41 Isolation Kit II Tat, Env) peptides (800 ng ml for each) (NIH AIDS Reagent Program), or with 21 1 (Miltenyi). Resting CD41 T cells were then purified from CD41 T cells by negative individual synthetic peptide (0.5 mgml ) (Genemed Synthesis). CD8 T cells were selection using CD25, CD69 and HLA-DR microbeads (Miltenyi). Genomic DNA purified after 6 days of incubation by positive selection using human CD8 microbe- was extracted from 5 million resting CD41 T cells from each patient using QIAamp ads (Miltenyi). To monitor CTL proliferation, PBMCs were stained with CFSE (Life DNA Mini Kit (Qiagen). The gag gene was amplified from genomic DNA by a two- Technologies) before incubation and with the relevant pentamer (Proimmune) round nested PCR using these primers: 59 outer primer (59-TTGACTAGCGGAG after incubation. PBMCs were then stained with CD8-APC (Becton Dickson (BD)) GCTAGAAGG-39); 39 outer primer (59-GATAAAACCTCCAATTCCCCCTAT and analysed by flow cytometry using FACS Canto II (BD). 1 1 C-39); 59innerprimer(59-GAGAGATGGGTGCGAGAGCGTC-39); 39inner primer Co-culture of autologous CD4 and CD8 T cells. Three hours after infection, 1 1 (59-CTGCTCCTGTATCTAATAGAGC-39). For each patient, the entire genomic CD4 T cells were mixed with autologous unstimulated or stimulated CD8 T DNA from 5 million resting CD41 T cells was evenly distributed as a template into cells at a 1:1 ratio in basal media at 5 million cells per ml. Two days after co-culture, 80 PCR reactions. The reactions were performed by using High Fidelity Plantinum enfuvirtide (T-20, Roche) was added into the culture at 10 mM to prevent further Taq Polymerase (Life Technologies) following the manufacturer’s instructions. infection events except if the measurement was p24 ELISA. Three days after co- PCR amplicons were purified by gel extraction after gel electrophoresis. culture, cells were stained with CD8-APC (BD) first, fixed and permeabilized with Deep sequencing. For PacBio RS single-molecule sequencing, amplicons were bar- Cytoperm/Cytofix (BD pharmingen), then stained for intracellular p24 Gag (PE, coded with a group of 10 bp indexes and then multiple samples were pooled together Coulter). Cells were analysed by flow cytometry using FACS Canto II (BD). For to generate a SMRTbell sequencing library following the Pacific Biosciences tem- measurement of viral growth, 5 ml of supernatant was taken from the co-culture at plate preparation and sequencing-C2 user guide for 2 kb insert size and using the days 0, 3 and 6, and subjected to p24 ELISA. For analysis of cell contact dependence, 1 1 Pacific Biosciences DNA template preparation kit. For MiSeq sequencing, the pooled CD4 and CD8 T cells were placed in separate chambers of trans-well plates amplicon DNA was end repaired, adenylated, and ligated to Illumina TruSeq adap- (0.4 mm, Costar). tors and PCR enriched for 10 cycles. The resulting library was then run on a bio- Generation and infection of patient-derived humanized mice. The previously 2 2 2 2 analyser high-sensitivity DNA chip for size and concentration determination. The reported MISTRG mouse in the Rag2 / Il2rg / 129 3 Balb/c (N2) genetic back- library was then sequenced on MiSeq for paired-end 250 bp reads. The sequence ground harbours knock-in replacements of the endogenous mouse Csf1, Csf2, Il3 reads from PacBio and MiSeq were demultiplexed using Fastx-Toolkit. and Tpo genes with humanized versions and a BAC transgene encoding human Data analysis for deep-sequencing results. For the paired MiSeq reads, the two SIRP-a25. We generated the Sirpa(KI) mouse, which harbours a knock-in replace- reads were first merged using FLASH31. MiSeq and PacBio reads from each indi- ment of the endogenous mouse Sirpa gene with a humanized version. The Sirpa(KI) vidual were then aligned to the reference HIV-1 consensus B Gag sequence using mouse will be thoroughly described elsewhere (manuscript in preparation). The Bowtie2 (ref. 32). A custom program was written using Perl scripts to identify and improved MIS(KI)TRG mouse was generated by breeding Sirpa(KI) mice to MITRG compute the frequency of all sequence variants that caused non-synonymous amino mice. All animal experimentations were performed in compliance with Yale Insti- acid changes in each individual’s relevant optimal Gag epitopes (based on reported tutional Animal Care and Use Committee protocols. MIS(KI)TRG mice were main- information in the HIV Molecular Immunology Database, Los Alamos National tained with continuous treatment with enrofloxacin in the drinking water (Baytril, Laboratory (http://www.hiv.lanl.gov/content/immunology/index.html)) according 0.27 mg ml21). Patient bone marrow or fetal liver CD341 cells were isolated by to their HLA type. For each individual, variants that occurred at a frequency .3% CD34 microbeads selection (miltenyi). Newborn mice (within their first 3 days of were retained. Additionally, for PacBio reads, sequences with identified premature life) were sublethally irradiated (X-ray irradiation with X-RAD 320 irradiator, PXi; stop codons were eliminated from the analysed results. For each identified vari- 1 3 150 cGy) and 100,000 fetal liver or 250,000 patient CD341 cells in 20 mlofPBS ation, the mutation type regarding CTL recognition was determined by matching were injected into the liver with a 22-gauge needle (Hamilton Company). Both with the information in the before-mentioned database. The five mutation types male and female mice with comparable engraftment levels (percentage of hCD451, adopted in this paper are: documented escape (no CTL response when patient cells hCD31 and hCD141 cells in the blood) were separated randomly into the experi- are challenged with the variant peptide); inferred escape (variant is predicted to be mental groups 6–8 weeks after engraftment. Mice engrafted with patient CD341 an escape mutant by longitudinal study or transmission study, but the reactivity of cells were infected by retro-orbital injection with HIV-1 (100 ng p24), which was the variant is not tested experimentally); diminished response (experimental data recovered and expanded from the resting CD41 T cells of the same patient (CD341 suggest partial escape as evidenced by decreased CTL response); susceptible form cell donor), as mentioned earlier. Mice engrafted with fetal liver CD341 cells were (CTL response is elicited when patient cells are challenged with the variant pep- infected by intravenous injection with HIV-1 BaL (100 ng p24). Mice with less than tide); and mutation type not determined (no experimental data on CTL recog- 5% human CD451 cells in the peripheral blood were excluded from the infection nition of this variant). study. Mice with more than 70% human CD451 cells in the peripheral blood were ELISpot assays. The ELISpot assays were performed using Human IFN-c ELISpot also excluded because they were unhealthy due to human macrophage/monocyte- PLUS kit (Mabtech) according to methods previously described33 and the manu- caused anaemia25. Twenty million autologous CD81 T cells with or without pre- facturer’s instructions. PBMCs were added at 200,000 cells per well and synthetic stimulation were injected intravenously 9 or 14 days after infection. Group allocation peptides were added in a final concentration of 0.1, 1 or 10 mgml21. A response was blinded. Peripheral blood samples were collected by retro-orbital bleeding at was considered positive if it was threefold higher than the mean background (cell different time points before and after injection of CD81 T cells. Engraftment of only control) and greater than 55 SFC per million cells. The number of specific T human CD451 cells as well as lymphoid and myeloid subsets was determined by cells was calculated by subtracting the mean background values. flow cytometry. Plasma HIV-1 RNA in peripheral blood was measured by one-step Recovery and sequencing of patient viruses from resting CD41 T cells. Co- reverse transcriptase (Invitrogen) real-time PCR using the following primers and culture assays were performed to recover and amplify replication-competent viruses probe, described previously8: forward (59 R 39) ACATCAAGCAGCCATGCAAAT, as previously described34. The viruses were recovered from 5–10 million resting reverse (59 R 39) TCTGGCCTGGTGCAATAGG, and probe (59 R 39) VIC-CTA CD41 T cells. The concentration of outgrowth viruses was determined by p24 TCCCATTCTGCAGCTTCCTCATTGATG-TAMRA. Assay sensitivity is 200 RNA ELISA (PerkinElmer). Total RNA of outgrowth viruses was extracted using TRIzol copies per ml of plasma. HIV-1 DNA in peripheral blood was also measured by real- LS reagent (Life Technologies). Residual DNA was then removed by TURBO DNase time PCR using the same primers and probe mentioned earlier, with assay sens- (Life Technologies) treatment. First-strand complementary DNA was synthesized itivity at 5 copies per 100 ml of blood. Total viral DNA in PBMCs was determined using SuperScript III Reverse Transcriptase (Life Technologies) and the gag gene by measuring copies of viral DNA per 100 ml blood and blood volume per mouse was amplified from cDNA using the gag outer primer pair mentioned above. The (80 ml blood per 1 g body weight). To quantitate total viral DNA in tissues, spleens,

©2015 Macmillan Publishers Limited. All rights reserved LETTER RESEARCH livers and lungs of infected mice were collected. For the spleen, single-cell suspen- GraphPad Prism 6 software, and conducted as two-tailed tests with a type I error sions were treated with ACK lysis buffer. Liver and lung leukocytes were isolated by rate of 5%. No statistical method was used to predetermine sample size. digesting tissues with 100 U ml21 collagenase IVand 0.02 mg ml21 DNase I (Sigma), 31. Magocˇ, T. & Salzberg, S. L. FLASH: fast length adjustment of short reads to followed by density gradient centrifugation. improve genome assemblies. Bioinformatics 27, 2957–2963 (2011). Statistical analysis. For comparison of HIV-1 variant frequency (Fig. 1b, c) and 32. Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nature viral infection in HIV-1 BaL-infected mice (Extended Data Figs 6 and 7), we Methods 9, 357–359 (2012). applied Mann–Whitney tests. For comparison of the inhibitory effect of autolo- 33. Streeck, H., Frahm, N. & Walker, B. D. The role of IFN-c Elispot assay in HIV vaccine research. Nature Protocols 4, 461–469 (2009). gous CTLs (Fig. 2a), we applied a Wilcoxon matched pairs test. For comparison of 34. Siliciano, J. D. & Siliciano, R. F. Enhanced culture assay for detection and viral replication in humanized mice (Fig. 4f–h), we applied an unpaired t-test. For quantitation of latently infected, resting CD41 T-cells carrying replication- all other comparisons, paired t-tests were applied. All tests were calculated by the competent virus in HIV-1-infected individuals. Methods Mol. Biol. 304, 3–15 (2005).

100 AP04 100 A*02:01 AP09 50 A*24:02 50 A*03:01 B*14:01 A*11:01 B*35:01 B*35:01 0 Cw*04:01 0 B*44:02 Cw*15:05 Cw*04:01 Cw*05:01

100 AP05 100 A*02:01 AP12 50 B*07:02 50 B*51:01 A*01:01 Cw*01:02 A*03:01 0 Cw*07:02 0 B*07:02 KK9 RK9 SV9 TL9 HA9 GL9 Cw*07

100 AP10 100 CP38 A*0101 A*03:01 A*02:01 A*24:02 50 B*08:01 50 B*44:02 B*15:01 B*81:01 0 Cw*03 0 Cw*05:01 GK9 EV9 SL9 TV9 EI8 GLY9 DL9 FK10 Cw*07 KK9 RK9 KW9 TL9 RL11 AW11 Cw*18:01 100 AP11 100 CP39 A*02:01 A*03:01 B*35:01 A*29:02 50 B*51:01 50 B*07:02 Cw*02:02 B*15:16 0 Cw*04:01 0 Cw*07:02 RL10 WF9 SL9 TV9 HA9 PY9 FK10 KK9 RK9 LY9 SV9 TL9 HA9 GL9 Cw*14:02 100 CP18 100 CP50 A*02:01 A*03:01 A*33:03 A*30:04 50 B*35:01 50 B*35:01 B*81:01 B*49:01 0 Cw*04:01 0 Cw*04:01 WF9 SL9 TV9 TL9 HA9 PY9 VI9 FK10 Cw*18:01 KK9 RK9 WF9 RY11 HA9 PY9 Cw*07:01

AP03 100 CP36 100 A*24:02 A*02:02 A*24:07 50 A*03:01 50 B*15:02 B*35:01 B*27:05 0 B*53:01 0 Cw*01:02 Cw*04:01 IK9 KW9 VL8 KK10 YL9 Cw*03:03

AP07 100 CP40 100 A*02:07 A*02:02 A*33:03 50 A*03:01 50 B*46:01 B*15:16 B*58:01 B*51:01 0 Cw*01:02 0 Cw*14:02 RY11 VL8 TW10 YL9 QW9 Cw*03:02

CP05 100 100 CP47 A*29:02 A*02:01 A*30:01 50 A*25:01 50 B*42:01 B*18:01 B*57:01

Frequency of variants (%) 0 Cw*07:01 0 Cw*06:02 Cw*12:03 Cw*17:01 SL9 QW11 TV9 EW10 FRK10 FK10 RY11 LY9 IW9 KF11 TL9 TW10 QW9 100 100 CP12 CP49 A*01:01 A*01:01 A*29:02 50 A*02:01 50 B*44:03 B*08:01 B*57:01 0 B*13:02 0 Cw*06:02 Cw*06:02 Cw*16:01 Cw*07:01 LY9 IW9 KF11 TW10 QW9

CP19 AP01 100 100 A*23:01 A*02:01 A*68:02 50 A*03:01 50 B*15:03 B*35:01 B*58:01 B*44:02 Cw*02:10 0 Cw*04:01 0 Cw*07:18 KK9 RK9 WF9 SL9 TV9 HA9 PY9 RL11 AW11 FK10 Cw*05:01 RY11 VF9 GHL9 TW10 QW9 CP31 100 CP42 100 A*23:01 A*02:01 A*66:02 50 A*03:01 50 B*41:01 B*07:02 B*58:01 B*38:01 0 Cw*07:18 0 Cw*07:02 Cw*08:02 Cw*12:03 RY11 KF11 TL9 EW10 TW10 QW9

CP32 100 AP08 100 A*34:02 A*02:01 A*68:01 A*03:01 50 50 B*44:03 B*07:02 B*53:01 Cw*07 0 0 Cw*04:01 LY9 TL9 QW9 100 CP48 A*33:03 50 A*37:01 B*44:03 Cw*06:02 0 Cw*07:06 LY9 RL11 AW11

CTL epitopes in HIV-1 Gag

Documented Escape Inferred Escape Diminished Response Susceptible Form Mutation Type Not Determined

Extended Data Figure 1 | CTL escape variants dominate the latent reservoir indicates that only wild-type sequences were detected. For each mutation in a of CP-treated HIV-1-positive individuals, but not AP-treated individuals. CTL epitope, information regarding the effect of the mutation on CTL Frequency of variants in Gag CTL epitopes in proviruses from resting CD41 recognition from the Los Alamos National Laboratory (LANL) HIV Molecular T cells. Results of all 25 patients tested are shown. Only optimal CTL epitopes Immunology Database or from ELISpot assays described in Methods was used relevant to each patient’s HLA type are listed in linear positional order on to assign the mutation to one of the categories indicated at the bottom. See the x-axis. Results from both PacBio (left bar) and MiSeq (right bar) sequencing Methods for definitions of these categories. platforms are shown for each epitope. The absence of bars above a listed epitope

©2015 Macmillan Publishers Limited. All rights reserved xeddDt iue2 Figure Data Extended h etdstse r itdfrec ain nec rp.Errbr ersn s.e.m., represent bars Error graph. each in patient each for listed are tested peptides The |

hrceiaino T epne gis I- a ptpsb interferon- by epitopes Gag HIV-1 against responses CTL of Characterization SFC/million PBMCs 120 160 1000 120 160 120 160 200 100 200 300 400 100 150 200 250 40 80 40 80 40 80 200 400 600 800 100 150 200 250 300 50 0 0 0 0 50 0 0 0 0.1 0.1 0.1 0.1 0.1 0.1 0.1 - - - - A- - - - K - RSLYNTVATLY(RY11) TSTLQEQIGW(TW10) ------L - ISPRTLNAW(IW9) ------N- QASQEVKNW(QW9) T-- - - -T ---- - RLRPGGKKK(RK9) ------S SLYNTVATL --F ----V - SLYNTVATL -----V---- - 1 1 1 1 1 1 1 ©2015 10 10 10 10 10 10 CP32 CP19 CP18 10 Macmillan Publishers Limited. All rights reserved 100 150 200 250 120 100 200 300 400 500 100 200 300 400 200 400 600 50 120 160 200 40 80 40 80 0 40 80 0 0 0 0 0 0 Peptide concentration (µg/ml) 0.1 0.1 0.1 0.1 0.1 0.1 0.1 TSTLQEQIGW(TW10) GHQAAMQML(GHL9) TLNAWVKVV(TV9) WASRELERF(WF9) TSTLQEQIGW(TW10) ------N------N- TPQDLNTML(TL9) TLNAWVKVV - -- -I ------F -I - -A-F LYNTVATLY CP39 1 1 1 1 1 1 1 10 10 10 10 10 10 10 CP12 CP36 CP31 1000 2000 3000 4000 1000 1200 120 40 80 200 400 600 800 120 160 0 40 80 0 0 0 n 0.1 0.1 0.1 QASQEVKNW(QW9) 0.1 5 --T ------GPGHKARVL(GL9) TPQDLNTML(TL9) 3. ------D - ETINEEAAEW(EW10) QASQEVKNW(QW9) --T ------1 1 1 1 10 10 10 10 c ELISpot. 10 20 30 40 50 60 70 200 400 600 800 100 200 300 400 500 0 0 0 0.1 0.1 0.1 KAFSPEVIPMF(KF11) ETINEEAAEW(EW10) WASRELERF(WF9) KAFSPEVIPMF(KF11) TPQDLNTML(TL9) eut fsvnptet etdaeshown. are tested patients seven of Results 1 1 1 10 10 10 LETTER RESEARCH RESEARCH LETTER

Consensus B RSLYNTVATLY ISPRTLNAW KAFSPEVIPMF TPQDLNTML TSTLQEQIGW QASQEVKNW CP05 provirus ...... L...... A...... N...... CP05 outgrowth ...... L...... A...... N...... RY11 LY9 IW9 KF11 TL9 TW10 QW9

PPIPVGEIY VRMYSPTSI Consensus B WASRELERF SLYNTVATL TLNAWVKVV TPQDLNTML HPVHAGPIA FLGKIWPSHK ...... CP18 provirus ...... F....V...... G...... V...... CP18 outgrowth ...... F....V...... G...... WF9 SL9 TV9 TL9 HA9 PY9 VI9 FK10

Consensus B RSLYNTVATLY VKVVEEKAF GHQAAMQML TSTLQEQIGW QASQEVKNW CP19 provirus ...... I.V...... N.....T. ....D.... CP19 outgrowth ...... I.V...... N.....T. ....D.... RY11 VF9 GHL9 TW10 QW9

Consensus B RSLYNTVATLY KAFSPEVIPMF TPQDLNTML ETINEEAAEW TSTLQEQIGW QASQEVKNW CP31 provirus K....A...... D...... N.....A. ....D.... CP31 outgrowth K....A...... D...... N.....A. ....D.... RY11 KF11 TL9 EW10 TW10 QW9

Consensus B KIRLRPGGKKK WASRELERF SLYNTVATL TLNAWVKVV ETINEEAAEW HPVHAGPIA QASQEVKNW CP36 provirus ...... S ...... V...... I ...... T...... CP36 outgrowth ...... S ...... V...... I ...... T ...... KK9 RK9 WF9 SL9 TV9 EW10 HA9 QW9

Consensus B KIRLRPGGKKKYKLKHIVW TPQDLNTML RDYVDRFYKTL AEQASQEVKNW CP38 provirus ...... Q.R...L.. ..G...L...... T...... CP38 outgrowth ...... Q.R...L.. ..G...L...... T...... KK9 RK9 KW9 TL9 RL11 AW11

Consensus B KIRLRPGGKKKYKL SLYNTVATL TLNAWVKVV CP40 provirus R...... T... ..F.AI.V...... I CP40 outgrowth R...... T... ..F.AI.V...... I KK9 RK9 RL10 SL9 TV9

Consensus B KIRLRPGGKKK SLYNTVATL SPRTLNAWVKVV TPQDLNTML HPVHAGPIA GPGHKARVL FLGKIWPSHK CP42 provirus ...... Q ..F..I...... CP42 outgrowth ...... Q ..F..I...... KK9 RK9 SL9 SV9 TV9 TL9 HA9 GL9 FK10

Unmutated CTL epitope Documented Escape Mutation Type Not Determined

Inferred Escape Diminished Response Susceptible Form

Extended Data Figure 3 | Partial Gag sequences from proviral DNA and diminished response (pink shading), susceptible form (green shading) or outgrowth virus from resting CD41 T cells from eight CP-treated patients. undetermined variations (grey shading) in relevant optimal epitopes are CTL epitopes with no observed variation are highlighted in blue. Documented indicated. See Methods for definitions of these types of mutations. This figure escape mutations (red shading), inferred escape mutations (yellow shading), supplements Fig. 1e, as a total of nine CP-treated patients were tested.

1000 abBAL CP05 100 CP12 + ++ ++ CP18 CD4 only CD4 + uCD8 CD4 + sCD8 CP26 CP27 0.4% 2.4% 5.7% 10 CP28 CP31 CP32 CP35 1 CP36

HIV-1 p24 (ng/ml) HIV-1 CP37 CP38 CP39 0.1 CD25 036 0.4% 3.2% 7.4% Days post-inoculation

- Gated on CD8 cells + No virus CD4 only c 0% 7.5% HLA-DR CD8

+ ++ CD4+ + uCD8 CD4 + sCD8 7.6% 2.7% Gag

CD8

d + 140 NS CD4 only ++ NS CD4 + uCD8 120 + + CD4 + sCD8 100 NS T cells (%) NS + NS NS * NS *** 80 ** CD4

+ ** NS 60 ** * *** ** 40 * *** ** ** ** ** ** Residual Gag Residual 20 ** *** 0

Extended Data Figure 4 | CD81 T cells pre-stimulated with a mixture of after co-culture of infected CD41 T cells with autologous CD81 T cells. consensus B Gag peptides eliminate autologous CD41 T cells infected with CTL activity is measured by the percentage of Gag-positive, CD8-negative cells autologous HIV-1 from resting CD41 T cells. a, HIV-1 isolated from after 3 days of co-culture relative to cultures without CD81 T cells. d,Pre- ART-treated individuals replicates as well as the laboratory strain virus BaL. stimulated CD81 T cells eliminate autologous infected CD41 T cells more p24 values represent mean of three replicates. Error bars represent s.e.m., n 5 3. efficiently than non-stimulated CD81 T cells. All results were normalized to the b,CD81 T cells are not stimulated after co-culture with PHA-activated CD41 CD4 only control group. Error bars represent s.e.m., n 5 3. *P , 0.05, T cells. c, A representative flow cytometric analysis of CTL-mediated killing **P , 0.01, ***P , 0.001, NS, not significant (P . 0.05), paired t-test.

a 120 CP05 120 CP12 120 CP28

+ 100 +uCD8 100 100 +

T cells (%) +sCD8

+ 80 80 80

CD4 60 60 60 +

40 40 40

20 20 20

Residual Gag Residual 0 0 0 0124 0124 0124

120 CP35 120 CP36 120 CP39

100 100 100 T cells (%)

+ 80 80 80

CD4 60 60 60 +

40 40 40

20 20 20

Residual Gag Residual 0 0 0 0124 0124 0124 Effector to target ratio Effector to target ratio Effector to target ratio

b + 160 CD4 only NS + NS sCD8 coculture + 140 sCD8 coculture in transwell NS NS 120 NS NS T cells (%)

+ 100 CD4 + 80 * * 60 * * 40 Residual Gag Residual ** ** 20

0 CP05 CP12 CP28 CP32 CP36 CP39

Extended Data Figure 5 | The elimination of infected CD41 T cells is CD41 T cells depends on direct cell–cell contact between CD41 T cells and mediated by direct killing by autologous CD81 T cells. a, Killing of infected CTLs. All results were normalized to the CD41 only control group. Error bars CD41 T cells is enhanced by increased effector to target ratios for both represent s.e.m., n 5 3. *P , 0.05, **P , 0.01, ***P , 0.001, NS, not pre-stimulated and non-stimulated CD81 T cells. b, Killing of the infected significant (P . 0.05), paired t-test.

a b 1 10 7 0.8 10 5 0.6 10 3 0.4

1 10 0.2 CD4 to CD8 ratio CD8 to CD4 Plasma HIV-1 RNA (copies/ml) 10-1 0 0 20 40 60 dpi 0 5 10 15 20 25 30 dpi

c P=0.004 d 0.03 * 4

3 0.02 2 0.01 1 CD4 to CD8 ratio hCD3 0 hCD14 Uninfected Infected of Copies cell associated HIV-1 RNA per cell 0.00 CD3- CD3+ CD14+ CD14- Extended Data Figure 6 | Viral dynamics and depletion of CD41 T cells in mice. MIS(KI)TRG mice engrafted with fetal liver CD341 cells were infected humanized mice. a, Viral dynamics in CP18-infected MIS(KI)TRG mice. with HIV-1 BaL. The CD4 to CD8 ratio in spleen was measured by FACS CP18-derived MIS(KI)TRG mice were infected with autologous HIV-1. Plasma 20 days after infection. Medians and P values from Mann–Whitney test are HIV-1 RNA levels were measured from day 0 to day 56. b, Depletion of CD41 shown. d, Detection of cell-associated HIV-1 RNA in T cells and macrophages/ T cells in peripheral blood of HIV-1 BaL-infected mice. MIS(KI)TRG mice monocytes. CD31 and CD141 human cells from HIV-1-infected MIS(KI)TRG engrafted with fetal liver CD341 cells were infected with HIV-1 BaL. The mice from spleen and lung were purified by FACS. CD32CD142 cells were CD4 to CD8 ratio in peripheral blood was measured by fluorescence-activated also collected as controls. Cell-associated HIV-1 RNA was quantified by cell sorting (FACS) from day 0 to day 29 after infection. Error bars represent gag-specific quantitative polymerase chain reaction (qPCR). Error bars s.e.m., n 5 5. c, Depletion of CD41 T cells in spleen of HIV-1 BaL-infected represent s.e.m., n 5 3. *P , 0.05, unpaired t-test.

a 100 week 6

80 + 60

20 % of total CD45 cells 0 Fetal liver CP18

b PBMCs Spleen 100 P< 0.0001 80 + + 60 26.2% 65.4% 40

20 % hCD45RO cells in hCD4 cells 0 hCD4 hCD45RO PBMCs Spleen

c P=0.07 P<0.01 6 10 P=0.02

105

104

103

Copies of cell associated HIV-1DNA 10

Liver Lung Spleen PBMCs Extended Data Figure 7 | HIV-1 infection occurs in peripheral blood collected at 20 days post-infection. Memory CD41 T cells were determined by and tissues in humanized mice. a, Engraftment levels of MIS(KI)TRG mice CD45RO staining. c, Total number of cell-associated HIV-1 DNA in blood with fetal liver or patient CD341 cells. b, Memory CD41 T cells are detected and tissues. DNA from peripheral blood or indicated tissues was isolated for in MIS(KI)TRG mice after infection. MIS(KI)TRG mice were infected with the measurement of total amount of cell-associated HIV-1 DNA by real-time HIV-1 BaL. Peripheral blood and indicated tissues from infected mice were PCR. b, c, Medians and P values from Mann–Whitney test are shown.

a b 10 5 10 7 CD8 injection 10 4 CD8 injection 10 5 10 3 10 2 10 3 10 1 1 10 unstimulated CD8 0 unstimulated CD8

HIV-1DNA (copies/100ul blood) 10 Plasma HIV-1 RNA (copies/ml) pre-stimulated CD8 pre-stimulated CD8 10-1 10-1 0102030 0102030dpi

Extended Data Figure 8 | Broad-spectrum CTLs suppress in vivo infection CP36 were pre-stimulated with the mixture of Gag peptides or left untreated of patient-derived humanized mice with autologous latent HIV-1. The for 6 days in vitro, and were injected into mice intravenously 9 days after generation of patient CP36-derived humanized mice is described in Fig. 4. Mice infection. Plasma HIV-1 RNA (a) and HIV-1 DNA (b) in peripheral blood were were infected with autologous viruses at 6 weeks old. CD81 T cells from patient measured by real-time PCR.

Extended Data Table 1 | Characteristics of study subjects

* Patient CD4 count was measured during this study. { Plasma HIV-1 RNA levels for all patients were ,50 copies per ml for at least 1 year before this study.