HIV reservoir size and persistence are driven by T cell survival and homeostatic proliferation

Transcription

1 HIV reservoir size nd persistence re driven by T cell survivl nd homeosttic prolifertion Nicols Chomont, Mohmed El-Fr, Petronel Ancut, Lydie Trutmnn, Frncesco A Procopio, Bder Yssine-Dib, Geneviève Boucher, Mohmed-Rchid Boulssel, Georges Ghtts 5, Json M Brenchley 6, Timothy W Schcker 7, Brenn J Hill 8, Dniel C Douek 8, Jen-Pierre Routy,9, Elis K Hddd,9 & Rfick-Pierre Sékly,9 9 Nture Americ, Inc. All rights reserved. HIV persists in reservoir of ltently infected in individuls treted with highly ctive ntiretrovirl therpy (HAART). Here we identify centrl memory ( ) nd trnsitionl memory ( ) s the mjor cellulr reservoirs for HIV nd find tht virl persistence is ensured by two different mechnisms. HIV primrily persists in cells in subjects showing reconstitution of the CD + comprtment upon HAART. This reservoir is mintined through T cell survivl nd low-level ntigen-driven prolifertion nd is slowly depleted with time. In contrst, provirl DNA is preferentilly detected in cells from viremic individuls with low CD + counts nd higher mounts of interleukin-7 medited homeosttic prolifertion, mechnism tht ensures the persistence of these cells. Our results suggest tht virl erdiction might be chieved through the combined use of strtegic interventions trgeting virl repliction nd, s in cncer, drugs tht interfere with the self renewl nd persistence of proliferting memory T cells. Tretment of HIV infection hs mrkedly reduced the deth rte from AIDS nd improved the qulity of life of HIV-infected individuls. However, complete erdiction of HIV with ntiretrovirl drugs seems impossible, s the virus persists in cellulr reservoirs,.themjor HIV reservoir is smll pool of ltently infected resting memory CD + T cells crrying n integrted form of the virl genome,5 tht lcks the bility to produce virl proteins 6.CD + T cell depletion is ssocited with lrger virl reservoir size 7, wheres erly initition of HAART is often ssocited with reduced size of the HIV reservoir 8 nd the normliztion of the CD/CD8 rtio 9. These observtions suggest tht CD + T cell depletion, which is directly ssocited with incresed levels of CD + Tcellprolifertion,, my drive the size of the HIV reservoir by s yet unidentified mechnisms. Two distinct mechnisms could contribute to the persistence of this reservoir. First, low levels of virl repliction could led to de novo infection of memory, ensuring the continuous replenishment of the HIV reservoir,. However, the bsence of genetic evolution in virl reservoirs does not support this possibility 7. Moreover, plsm virl RNA sequences under HAART diverge from cell-ssocited virl DNA, providing further evidence tht de novo infection of cells is unlikely to be implicted in the persistence of the reservoir 8. The intrinsic stbility of ltently infected constitutes n lterntive explntion for the stbility of the reservoir, s cells survive for yers 9. Indeed, we hve shown tht prosurvivl pthwys re specificlly triggered in cells,,indicting tht cells my encompss the long-lived HIV reservoir. The differentition of cells into T EM cells is observed fter T cell receptor (TCR) triggering, nd, to lesser extent, in response to homeosttic cytokines such s interleukin-7 (IL-7) nd IL-5 (ref. 5). The contribution of ntigen-induced or homeosttic prolifertion to the size nd mintennce of the reservoir hs yet to be defined. This is prticulrly relevnt in the context of HAART, where high levels of immune ctivtion, incresed plsm concentrtions of IL-7 nd n incresed percentge of cycling hve been observed in individuls with low bsolute CD + T cell counts 6.Hereweshowthe presence of two distinct HIV reservoirs: one in cells, regulted by ntigen-driven prolifertion, nd one in cells, regulted by homeosttic prolifertion. RESULTS HIV provirl DNA is detected in nd cells To identify the mjor moleculr forms of HIV ssocited with its persistence, we mesured levels of totl DNA, integrted DNA nd -LTR circles, which re lbile intermedites in the virus life cycle, in Lbortoire d Immunologie, Centre de Recherche du Centre Hospitlier de l Université de Montrél (CR-CHUM) Sint-Luc, Montrél, Québec, Cnd. Lbortoire d Immunologie, Déprtement de Microbiologie et d Immunologie, Université de Montrél, Québec, Cnd. Institute Ntionl de l Snté et de l Recherche Médicle U7, CR-CHUM, Université de Montrél, Montrél, Québec, Cnd. Immunodeficiency Service nd Division of Hemtology, Royl Victori Hospitl, McGill University Helth Centre (MUHC), McGill University, Montrél, Québec, Cnd. 5 Deprtment of Gstroenterology, MUHC, Montrél, Québec, Cnd. 6 Lbortory of Moleculr Microbiology, Ntionl Institute of Allergy nd Infectious Diseses, US Ntionl Institutes of Helth, Bethesd, Mrylnd, USA. 7 Deprtment of Medicine, University of Minnesot, Minnepolis, Minnesot, USA. 8 Humn Immunology Section, Vccine Reserch Center, Ntionl Institute of Allergy nd Infectious Diseses, US Ntionl Institutes of Helth, Bethesd, Mrylnd, USA. 9 Deprtment of Microbiology nd Immunology, McGill University, Montrél, Québec, Cnd. Deprtment of Microbiology nd Immunology, Oregon Helth nd Science University, Portlnd, Oregon, USA. Vccine nd Gene Therpy Institute, Port-Ste Lucy, Florid, USA. Correspondence should be ddressed to R.-P.S. (rfick-pierre.sekly@umontrel.c). Received 7 November 8; ccepted 9 April 9; published online June 9; doi:.8/nm.97 NATURE MEDICINE ADVANCE ONLINE PUBLICATION

2 HIV DNA in 6,,, HAART Off On On Virl lod 7 <5 <5 Weeks Subject A Subject D Subject K Subject L Off On On 5 <5 <5 + + Off On On <5 < Off On On 5 5 <5 < Totl HIV DNA Integrted HIV DNA -LTR circle DNA c Percentge of cells mong T N T EM T TD 9 Nture Americ, Inc. All rights reserved. bhiv DNA in 6, Totl HIV DNA Integrted HIV DNA -LTR circle DNA ND Subject Subject 5 CD + count = 67 cells per µl from successfully treted individuls, including four subjects from whom we obtined peripherl blood mononucler cells (PBMCs) before nd fter initition of HAART. The results indicted rpid nd profound drop in levels of -LTR circles upon initition of HAART (Fig. ), wheres HIV provirl DNA ws more stble thn nd present in similr mounts to totl HIV DNA fter prolonged HAART (Fig. b), confirming tht the totl HIV DNA found in CD + T cells fter prolonged virus suppression lrgely represents integrted HIV genomes, s recently reported 7. These results identify integrted virl DNA s the mjor nd most stble moleculr form of HIV in from successfully treted subjects. To determine whether the stbility of integrted virl DNA cn be ttributed to the persistence of specific memory CD + T cell comprtments, we sorted CD + T cell subsets from 7 viremic subjects (Supplementry Tble ) on the bsis of surfce expression of CD5RA, CC chemokine receptor-7 (CCR7) nd CD7 nd performed highly sensitive quntifiction of integrted HIV provirl DNA on the sorted subsets. On the bsis of CD5RA nd CCR7 expression, we identified vrious CD + subsets, including nive (T N ), centrl memory ( ) nd terminlly differentited (T TD ) cells (Supplementry Fig. ). CD7 expression llowed us to distinguish effector memory (T EM ) cells from the trnsitionl memory CD + T cell subset ( cells); the ltter cells show functionl nd trnscriptionl chrcteristics tht re intermedite between those of cells nd T EM cells. d Integrted HIV DNA copies per 6 cells 6 Figure Integrted DNA is the mjor moleculr form of HIV during HAART nd is hrbored by nd cells in vivo. () Frequencies of totl, integrted nd -LTR circle DNA molecules in before nd fter HAART initition. Moleculr forms of HIV were quntified by rel-time PCR in longitudinl smples from four individuls who initited HAART during the follow-up period. Off nd On refer to the therpy sttus. The time before ( ) nd fter (+) HAART initition (in months) nd the virl lod t ech time point (in RNA copies per ml of plsm) re indicted. (b) Frequencies of totl, integrted nd -LTR circle HIV DNA in 6 from ten subjects receiving HAART for more thn months. (c) Percentge of CD + T cell subsets mong totl in cohort of individuls with undetectble virl lod. Percentges of cells were determined by flow cytometry. (d) Frequencies of cells hrboring integrted e Subject 5 CD + count = cells per µl, 8 Contribution of CD + T cell subsets to the pool of HIV-infected cells (%) 8 6 Subject 6 CD + count = 56 cells per µl T N T EM T TD T N T EM T TD T N T EM T TD T N +EM T EM T TD CD5RA + + CCR7 + + CD HIV DNA in CD + T cell subsets from three representtive viremic subjects. Results re expressed s the HIV copy number in 6 cells of given subset. (e) Contribution of T N,,+EM nd T TD cell subsets to the pool of HIV-infected cells clculted in 7 viremic HIV-infected subjects. HIV integrted DNA copy number ws determined in sorted subsets by highly sensitive rel-time PCR. Ech symbol represents different individul. In ten subjects, CD5RA CCR7 were further sorted ccording to the expression of CD7 to distinguish between cells nd T EM cells. Horizontl lines indicte men vlues. We determined the percentge of ech cellulr subset within the pool of from viremic individuls (Fig. c). The frequency of cells within ech subset hrboring HIV provirl DNA rnged from to. copies of HIV provirus per 6 cells of given subset (Fig. d nd Supplementry Fig. b). Our results clerly indicte tht the HIV reservoir constitutes cells hrboring memory phenotype (men contributions of cells nd nd T EM cells of 5.7% nd 6.6%, respectively), wheres T N cells nd T TD cells mrginlly contribute to the pool of cells hrboring HIV provirl DNA (men contributions of.9% nd.%, respectively, Fig. e). Overll, the contributions of cells nd cells to the HIV reservoir were higher thn tht of T EM cells (men contributions of 5.7%,.% nd.9%, respectively). To confirm tht hrbor repliction-competent virus, we sorted cells from ech subset of four viremic subjects nd mesured virl production fter co-culture nd stimultion with llogeneic dendritic cells nd phytohemgglutininctivted from HIV-negtive donors (Supplementry Fig. ). These results showed tht subsets hrboring HIV provirl DNA re ble to produce infectious virus fter stimultion. We lso mesured the frequency of cells hrboring HIV provirl DNA in lymph nodes nd gut biopsies from successfully treted individuls (Supplementry Fig. f). We found tht lymph nodes were enriched (fold enrichment of.) for memory s compred to blood, but the frequencies of memory hrboring provirl DNA were comprble in both comprtments ADVANCE ONLINE PUBLICATION NATURE MEDICINE

3 9 Nture Americ, Inc. All rights reserved. per 6,, P <. b c d ρ =.57 P =.5 e 5 f g / rtio ρ =. P =. < CD/CD8 rtio 7, per 6 /+EM infection frequency rtio, ρ =.88 P <. (mens of 56 copies nd 6 copies of HIV integrted DNA in 6 memory from PBMCs nd lymph node mononucler cells, respectively). We did the sme quntifictions in PBMCs nd mtched gut biopsies from eight subjects on HAART; the results showed tht the frequencies of cells hrboring HIV provirl DNA re similr nd highly correlted between these two comprtments (Supplementry Fig. f). Together, these experiments show tht, nd T EM cells contribute to vrious degrees to the HIV reservoir nd cn produce infectious virl prticles upon ctivtion. Moreover, experiments with PBMCs of individuls on HAART, imed t quntifying provirl DNA, did indeed reflect the contribution of the lymphoid tissue environment to the HIV reservoir. Absolute CD + count identifies the HIV reservoir Our results lso showed tht the reltive contributions of nd cells to the pool of HIV-infected cells re highly vrible from one subject to nother (Fig. e). We observed tht CD/CD8 rtios, high ndir CD + count nd initition of HAART within the first yer of HIV infection re strongly ssocited with n HIV reservoir of limited size (P o., P ¼.5 nd P o., respectively; Fig. c). As CD + T cell depletion ws lso ssocited with n incresed HIV reservoir size (P ¼., Fig. d), we exmined whether bsolute CD + counts ffect the reservoir locliztion in nd cells. We found tht cells were underrepresented in subjects with low CD + counts, suggesting tht cells re selectively depleted or differentited into cells in these subjects (P ¼., Fig. e). Moreover, the frequency of infected cells ws higher in subjects with low CD + counts, wheres cells were preferentilly infected, 7, per 6 Percentge of cells mong HIV + cells,, 8 6 P <. > < yer > yer Ndir Durtion of exposure to HIV ρ =.66 P =. 7, per 6 Percentge of +EM cells mong HIV + cells,, ρ =.8 P =. ρ =.6 P =.6 in subjects with high bsolute CD + counts (P o., Fig. f). As cells were overrepresented nd preferentilly infected in individuls with high CD + counts, they constituted the mjor reservoir for HIV in those subjects (P ¼., Fig. g). Conversely, the HIV reservoir ws minly loclized in cells in individuls with low CD + counts (P ¼.6). Immune ctivtion identifies the HIV reservoir Depletion of is ssocited with high levels of immune ctivtion nd incresed prolifertion of,,6.aswe found tht the bsolute CD + count determines the size nd the locliztion of the HIV reservoir, we exmined the impct of the residul immune ctivtion on these prmeters. Absolute CD + counts were negtively correlted with the expression of the immune ctivtion nd prolifertion mrker Ki67 in (P ¼.7, Fig. ). Notbly, high frequencies of Ki67 + were ssocited with lrger size of the virl reservoir (P ¼., Fig. b), suggesting tht T cell prolifertion might provide mechnism for the mintennce of the HIV reservoir. Moreover, we found tht the levels of cellulr prolifertion could predict not only the size but lso the locliztion of the HIV reservoir, s high levels of Ki67 protein expression were correlted with preferentil infection of cells (P ¼.8, Fig. c) nd with higher contribution of this subset to the HIV reservoir (P ¼.6, Fig. d). Conversely, cells were the min reservoir in individuls with limited levels of cellulr prolifertion (P ¼.; Fig. d). Together, these results indicte tht low bsolute CD + counts re ssocited with high levels of cellulr prolifertion nd n incresed 8 6 7, 7, Figure CD + T cell depletion drives the size nd the locliztion of the HIV reservoir. ( d) Integrted HIV DNA copy number, s determined in the CD + T cells from virlly suppressed HIV-infected subjects by Alu rel-time PCR. () Comprison of the size of the HIV reservoir in subjects with CD/CD8 rtio o or. (b) Correltion between the size of the HIV reservoir nd the ndir CD + count. (c) Comprison of the size of the HIV reservoir in subjects who hve strted HAART before or fter the first yer of infection. (d) Correltion between the size of the HIV reservoir nd bsolute CD + counts. (e) Correltion between the bsolute CD + count nd the / rtio in subjects receiving HAART. The frequencies of nd cells within totl were mesured by flow cytometry fter stining for the CD, CD5RA, CCR7 nd CD7 mrkers. (f) Correltion between the /+EM infection frequency rtio nd bsolute CD + counts in the 7 individuls on whom HIV quntifictions in CD + T cell subsets were performed. HIV integrted DNA copy number ws determined in sorted subsets by highly sensitive rel-time PCR. The rtio ws clculted by dividing the number of copies of integrted HIV DNA in 6 by the number of copies of integrted HIV DNA in 6 +EM cells. (g) Correltions between the contribution of cells nd +EM cells to the HIV reservoir nd the bsolute CD + count in the sme individuls. To clculte the contribution of ech CD + T cell subset to the globl pool of HIV-infected cells, the percentge of ech subset within the CD + comprtment s well s the frequency of the cells in the subset hrboring HIV integrted DNA were tken into ccount. Horizontl lines indicte men vlues. NATURE MEDICINE ADVANCE ONLINE PUBLICATION

4 9 Nture Americ, Inc. All rights reserved. Percentge of KI67 + e Percentge of Ki67 + cells ρ =.7 P =.7 7, T N b per 6 T EM,, ρ =. P =. Percentge of Ki67 + Percentge of PD- + cells size of n HIV reservoir minly hrbored by cells. Conversely, we observed virl reservoir of limited size minly hrbored by cells in individuls with high bsolute CD + counts nd limited prolifertion of. nd define distinct HIV reservoirs We next chrcterized the phenotypes of cells nd cells to determine whether their contribution to the HIV reservoir is function of their prolifertive sttus. cells included higher frequencies of Ki67 + nd PD- + cells thn cells but lower frequencies thn T EM cells (Fig. e). Moreover, nd irrespective of the phenotype of the CD + subset, the HIV reservoir ws minly composed of cells endowed with limited prolifertion indexes (Ki67, P o. nd PD-, P ¼.), confirming the quiescent nonctivted stte of cells tht constitute the HIV reservoir in HAART-treted subjects (Fig. f). Our results indicte tht highly proliferting T cells such s T EM cells (rnge of Ki67 + cells,..9%) do not constitute stble reservoir for HIV, which is consistent with their limited contribution to the pool of ltently HIV-infected cells. In contrst, cells nd cells re chrcterized by low to intermedite levels of prolifertion nd constitute the min virl reservoir. These results suggest model in which HIV cn persist in nd cells by continuous low-level prolifertion, ensuring the persistence of integrted virl DNA through mitosis. We thus sorted cells nd cells from four HAART-treted subjects ccording to their PD- expression, mrker of homeosttic 8,9 nd c /+EM infection frequency rtio T TD T N T EM T TD f ρ =.6 P =.8 Percentge of Ki67 + Figure Prolifertion of drives the size nd the locliztion of the HIV reservoir. () Correltion between the bsolute CD + count nd the level of cellulr prolifertion, s mesured by intrcellulr Ki67 expression in from individuls receiving potent HAART. (b) Correltion between the size of the HIV reservoir nd the frequencies of Ki67 + cells within totl. The frequency of cells hrboring HIV integrted DNA in totl from individuls ws determined by Alu reltime PCR. (c) Correltion between the /+EM infection frequency rtio nd the expression of Ki67 on totl in the 7 subjects on whom HIV quntifictions in CD + T cell subsets were performed. (d) Correltions between contributions of nd +EM cells to the HIV Contribution of CD + T cell subsets to the pool of HIV-infected cells (%) g d 8 6 P =. ρ =.7 P =. Percentge of Ki67 + ntigen-induced prolifertion, s indicted by the strong correltion between PD- nd Ki67 expression in from HAARTtreted individuls (P ¼., Fig. g). Quntifiction of HIV provirl DNA in ech frction showed tht PD- hi cells were enriched for HIV provirl DNA when compred to the corresponding PD- lo frctions (P ¼.6), confirming tht proliferting nd cells constitute privileged cellulr reservoir for HIV (Fig. h). These results indicte tht HIV preferentilly persists in nd cells tht proliferte t low levels. The cell reservoir is fueled by proliferting cells We mplified Env sequences from provirl DNA obtined from cells nd cells of ten HAART-treted subjects nd performed phylogenetic nlyses to ssess the interdependence of ech cellulr reservoir by compring HIV sequences obtined from both cellulr reservoirs (Fig. ). With the exception of three clones obtined from one subject, ll viruses (n ¼ 7) were predicted to use CCR5; hence, co-receptor usge could not llow us to evidence the interdependence of the nd reservoirs. However, sequences obtined from cells nd cells of individuls tht were good responders to HAART (see subject D s n exmple, CD + Ki67 + ¼.%) did not cluster together, indicting tht in those subjects these reservoirs re geneticlly independent (Fig. ). In contrst, in viremic subjects chrcterized by higher levels of CD + T cell prolifertion (for exmple, subject A, CD + Ki67 + ¼.5%), virl qusispecies were shred by cells nd cells, s shown by the unique cluster h per 6,, PD- hi ρ =.75 P =.6 ρ =.7 ρ =.6 P <. P =. 8 8 T EM Percentge of Ki67 + cells Percentge of PD- + cells Percentge of PD- + Percentge of cells mong HIV + cells Percentge of Ki Percentge of Ki67 + P =.6 reservoir nd the expression of Ki67 in the sme individuls s in c. (e) Percentge of from ech cellulr subset expressing Ki67 nd PD-. (f) Correltions between the contribution of given CD + T cell subset to the HIV reservoir nd the expression of Ki67 nd PD-. (g) Correltion between PD- nd Ki67 expression in totl CD + Tcells.(h) cells nd cells from three nd four virlly suppressed subjects were sorted ccording to their reltive expression of PD-. The frequency of cells hrboring HIV provirl DNA ws determined in both frctions by Alu rel-time PCR. Percentge of +EM cells mong HIV + cells PD- lo ADVANCE ONLINE PUBLICATION NATURE MEDICINE

5 9 Nture Americ, Inc. All rights reserved. b Men genetic distnce of HIV between nd cells (%) Subject A CD + Ki67 + =.5% HXB c ρ =.9 P =.5 ρ =.86 P =. 5 Ki67 + cells (%) Subject B CD + Ki67 + =.9% HXB HXB HXB HIV diversity in cells (%) Subject C CD + Ki67 + =.6% Ki67 + cells (%) Ki67 + cells (%) of vrints found in the neighbor-joining tree (Fig. ). When performed in ten subjects, the phylogenetic nlysis confirmed tht incresed turnover of cells is ssocited with reduced HIV genetic distnce mesured between the two comprtments (P ¼.5, Fig. b), consequence of the residul immune ctivtion nd cell prolifertion leding to the differentition of cells into cells,. Cell prolifertion ensures HIV reservoir stbility We determined the impct of cellulr prolifertion on the genetic diversity of provirl DNA within both cellulr reservoirs. We did not find ny ssocition between the genetic diversity of the reservoir nd levels of cell prolifertion (Fig. c), indicting tht cells hrboring rchived provirl DNA constitute stble reservoir with miniml prolifertion levels, in greement with their enhnced survivl. In contrst, prolifertion of cells ws ssocited with HIV diversity in cells (%) Subject D CD + Ki67 + =.% , 8 987,. Figure nd cells define distinct HIV reservoir. () Neighborjoining trees derived from HIV sequences obtined from cells nd cells of four representtive HAART-treted HIV-infected subjects. nd cells from ten viremic subjects were sorted, nd minimum of clones deriving from t lest ten independent positive PCRs of the Env gene were sequenced in ech cse. The numbers ner nodes indicte the percentge of bootstrp replictes (, resmpling). The scle refers to the distnce between sequences. Verticl lines re for clrity only. (b) Correltion of the men genetic distnces of HIV Env sequences between nd cells to the percentge of Ki67 + cells in ten viremic subjects. Ech dimond represents the men genetic distnce existing between the clones obtined from cells nd cells. (c) Correltions between the HIV genetic diversity in cells nd cells nd the expression of Ki67 in these subsets. HIV diversity ws estimted by clculting the men genetic distnce between the clones within given subset. reduced HIV diversity in this subset (P ¼., Fig. c). This result suggested mechnism by which HIV-infected cells mintin the size nd the lck of HIV genetic diversity of the virl reservoir by continuous self-renewl through the prolifertion of smll number of cells. To verify this hypothesis, we performed longitudinl nlysis in five viremic subjects from whom we obtined virl sequences t two time points seprted by t lest months (Supplementry Tble nd Supplementry Fig. ). We ligned sequences obtined from cells nd cells t these two time points nd drew phylogenetic trees reflecting the evolution of virl sequences (Fig. 5 nd Supplementry Fig. 5). To mesure the genetic evolution of the HIV reservoir, we quntified the divergence s the verge genetic distnce between sequences from the two time points within ech T cell subset (Fig. 5b). Our results indicted tht Ki67 + cells =.% Subject A Ki67 + cells =.5%, ,, Subject E Ki67 + cells =.6% Ki67 + cells =.% Time point, Time point 955 T TM Time point 999,,, Time point Figure 5 Prolifertion of cells is ssocited with genetic stbility of the HIV reservoir over time. () Neighbor-joining trees derived from HIV sequences obtined from cells nd cells of two representtive viremic individuls t first nd second time points (closed nd filled symbols, respectively). Percentges of cells nd cells expressing the Ki67 prolifertion mrker re indicted. (b) Genetic evolution of the HIV reservoir in cells nd cells from five viremic individuls followed longitudinlly. Results re expressed s the men genetic distnce existing between HIV sequences from both time points, corresponding to the estimted genetic divergences of the Env region (left). Percentge of Ki67 + cells (middle) nd plsm IL-7 concentrtions (right) re lso shown. (c) Evolution of the HIV reservoir size in eight viremic individuls. The integrted HIV DNA copy number ws mesured by Alu rel-time PCR in purified t two time points. For ech subject, the line between the two time points reflects the slope of the decrese in the frequency of hrboring HIV provirl DNA. (d) Correltion between the slope decrese in the number of HIV-infected nd the percentge of Ki67 + cells. The decrese in the HIV reservoir size is expressed s decrese of the percentge of cells hrboring HIV provirl DNA per dy. (e) Correltion between the slope decrese in the number of HIV-infected nd the plsm IL-7 level mesured by ELISA. b A B C D HXB c d e per 6,, HXB 8, Time (d) Slope decrese (% decrese per dy). A B C D HXB E E E Genetic divergence (%) Ki67 + cells (%) IL-7 (pg ml ) ρ =.88 P =.7 Slope decrese (% decrese per dy) A B C D ρ =.76 P = Ki67 + cells (%) IL-7 (pg ml ) 96. HXB NATURE MEDICINE ADVANCE ONLINE PUBLICATION 5

6 9 Nture Americ, Inc. All rights reserved. Figure 6 IL-7 induces homeosttic prolifertion of nd ensures HIV reservoir stbility through cellulr prolifertion t low levels. () Phenotype of fter TCRinduced (CD/CD8) or IL-7 prolifertion. Cells were hrvested t dy 8, nd their phenotype ws determined by flow cytometry. (b) Percentge of expressing the prolifertion mrker Ki67 fter 8 d of CD/CD8 or IL-7 stimultion. (c) HIV genetic diversities t bseline nd fter 8 d of tretment. Ech dot represents the men genetic distnce between one given clone nd the entire popultion. The horizontl br represents the men of these vlues nd is reflecting the genetic diversity of the virl popultion in the vrious conditions. 8 6 high prolifertion levels in cells re ssocited with limited genetic evolution of the reservoir over time, in greement NS P =. with mechnism by which persistence of this reservoir is ensured by smll number of HIV-infected proliferting (subject A, Fig. 5,b). To our surprise, in subjects with low frequencies of Ki67 + cells, sequences obtined from the two time points did not cluster together (subject E, Fig. 5,b). The Bseline pprent evolution of HIV sequences in individuls with high CD + T cell counts probbly reflects the smpling of different rchived sequences rther thn genetic evolution of the provirl popultions. These sequences my be present t vrible frequencies s consequence of the prolifertion nd deth of distinct T cell clones in subjects who re still endowed with functionl CD + comprtment. HIV reservoir is mintined by IL-7 medited prolifertion CD + T cell depletion is directly correlted with high levels of plsm IL-7 (refs. 6, 6), cytokine responsible for the survivl nd homeosttic prolifertion of 5,7,8. We observed higher mounts of IL-7 in the plsm of individuls with high frequency of proliferting cells nd limited evolution of the virl reservoir over time (Fig. 5b). We then ssessed the role of this cytokine in the mintennce of the size of the reservoir. We determined the frequency of hrboring provirl DNA in eight viremic HAART-treted subjects t two different time points with t lest months between independent mesurements (Fig. 5c). The size of the virl reservoir decresed very slowly with time in the eight viremic subjects tested. We nlyzed our dt by using previously described regression model 9, nd we found men hlf-life of this reservoir of 9.5 months, indicting tht n verge of 65.7 yers of tretment would be necessry to erdicte 6 infected cells. Of note, we observed significnt negtive correltion between this decrese nd the percentge of proliferting cells (P ¼.7, Fig. 5d). Moreover, we observed tht plsm IL-7 concentrtions inversely correlted with the decrese of the reservoir size over time (P ¼.7, Fig. 5e). Altogether, these results indicte tht IL-7 is responsible for the persistence of ltently HIV-infected by promoting homeosttic prolifertion of memory, resulting in the quntittive nd qulittive stbility of the HIV reservoir. Subject Percentge of CD + T cell subset b Subject Percentge of Ki67 + c Subject Genetic diversity (%) 8 6 Mock Mock NS CD/ CD8 CD/ CD8 CD/ CD8 ng ml ng ml ng ml ng ml Subject Percentge of CD + T cell subset 8 6 Subject Percentge of Ki Subject Genetic diversity (%) Mock CD/ CD8 Mock Cellulr prolifertion influences the HIV reservoir We performed in vitro experiments to evlute the reltive impct of ntigen-induced nd homeosttic prolifertion on the mintennce of the HIV reservoir. We stimulted from two virlly suppressed subjects through TCR triggering (CD nd CD8) or incubte them with IL-7 in the presence of zidovudine nd ritonvir to void possible de novo infection of cells with HIV. We observed tht TCR triggering induced modifictions in the distribution of CD + T cell subset nd ws ccompnied by high levels of prolifertion (Fig. 6,b). In contrst, IL-7 t ng ml nd ng ml ws ble to induce prolifertion of with limited impct on the distribution of the vrious nive nd memory T cell subsets, confirming the homeosttic nture of the prolifertion induced by this cytokine. Anlysis of HIV sequences fter IL-7 tretment indicted tht this cytokine induced the prolifertion of without modifying the diversity of the virl reservoir, s indicted by the bsence of significnt difference in the genetic diversity of the provirl popultions t bseline nd fter 8 d of culture (Fig. 6c). In contrst, treted with CD nd CD8 showed significnt reduction in the diversity of the provirl popultion (subjects nd, P ¼. nd P o., respectively). Of note, we used the sme number of positive PCR rections for cloning of provirl qusispecies under ech experimentl condition (n ¼ 7), indicting tht the decrese in virl genetic diversity is unlikely to be ttributble exclusively to cell deth. Thus, the decrese in the virl genetic diversity observed fter TCR triggering is most probbly ttributble to the specific expnsion of smll number of HIVinfected memory CD + T cell clones, which re undergoing substntil prolifertion in our cultures t dy 8, s indicted by incresed frequencies of Ki67 + cells (Fig. 6b). This observtion supports our findings in vivo showing tht TCR-induced prolifertion is responsible NS NS P <. CD/ CD8 ng ml ng ml Bseline CD/ CD8 IL-7 ng ml IL-7 ng ml Dy Dy 8 Dy Dy T N T EM T TD Others ng ml ng ml ng ml ng ml 6 ADVANCE ONLINE PUBLICATION NATURE MEDICINE

7 9 Nture Americ, Inc. All rights reserved. for the specific expnsion of restricted number of CD + T cell clones hrboring limited set of divergent HIV sequences (Fig. 5). Thus, these results indicte tht IL-7 cn medite the survivl of HIVinfected cells through homeosttic prolifertion, thereby ensuring the persistence of geneticlly stble HIV reservoir, wheres ntigeninduced prolifertion leds to the genetic evolution of the provirl popultion through the preferentil prolifertion nd survivl of restricted number of CD + Tcellclones. DISCUSSION We hve identified two virl reservoirs within memory CD + T cell subsets of virlly suppressed subjects. The reservoir is the mjor long-lsting reservoir in immune responders to HAART. As cells re chrcterized by their extremely low degree of cellulr prolifertion, nd, becuse of their intrinsic cpcity to survive for decdes,, these cells provide long-lsting cellulr reservoir for HIV. In immune responders to HAART with norml CD + counts, the size of the HIV reservoir decreses very slowly with time, indicting tht this cellulr reservoir could be prtilly depleted through the use of intensive ntivirl strtegies for prolonged periods of time. The second reservoir is hrbored by cells nd is the min reservoir in individuls with low CD + counts, the mjority of whom re chrcterized by persistent immune ctivtion. Our results clerly show tht this reservoir persists by homeosttic prolifertion of infected cells, ensuring the stbility of this virl reservoir in its size nd its genetic vribility. As the size of this reservoir is reduced in individuls who hve been treted erly in infection, our findings confirm the importnce of erly therpeutic intervention, s it limits the size of this proliferting HIV reservoir 8. Severl studies hve clerly shown the continuous production of virions during HAART,, ; recent nlysis of longitudinl plsm smples suggests tht this low-level, persistent viremi seems to rise from t lest two cellulr comprtments, one in which virl production decys over time nd second tht remins stble for t lest 7 yers. Our identifiction of cells nd cells s two mjor reservoirs tht re chrcterized by differing decy rtes in HAARTtreted individuls supports these findings. Our observtions lso provide the first evidence, to our knowledge, for the vlidity of the mthemticl model proposed recently, in which bystnder prolifertion of HIV-infected cn ensure HIV reservoir persistence without ny demonstrble evidence for virl production. The impct of virl production on the replenishment of the HIV reservoir is still controversil 7,8, 6. A recent study concluded tht persistence of the HIV reservoir is unlikely to result from ongoing virl repliction, s sequences from the predominnt plsm clones re different from those found in the provirl reservoir 8.Moreover,genetic nlysis of rebounding viruses fter tretment interruptions rgues ginst persistence of ongoing low-level repliction in individuls on suppressive HAART 7. Accordingly, our findings suggest limited role for virl repliction in the persistence of the reservoir. First, the nonstochstic distribution of HIV-infected cells mong T cell subsets tht re permissive to HIV infection suggests tht de novo infection of does not occur. Second, HAART-treted individuls with high levels of immune ctivtion, in whom ongoing repliction is most likely to occur, do not show ny evidence of genetic evolution in their provirl popultions. Altogether, these observtions strongly suggest tht ongoing virl repliction is unlikely to be responsible for the persistence of HIV-infected cells nd strengthen the role of cell prolifertion s mjor mechnism to ensure HIV persistence. IL-7 hs been shown to induce the prolifertion nd survivl of memory 5. We found tht CD + T cell depletion is ccompnied by incresed mounts of IL-7, which provides potentil mechnism by which ltently infected could proliferte in response to this cytokine, s mesured by Ki67 expression. Our results re in line with recent study indicting tht CD + T cell prolifertion in HIV-infected individuls (s ssessed by BrdU incorportion) is driven by IL-7 s homeosttic response to CD + T cell depletion 8. However, our Ki67 mesurements my hve overestimted the frequency of dividing cells, s Ki67 + cells my not represent true in vivo prolifertion in HIV-infected individuls, prticulrly in untreted subjects 9,5. These results nd our observtions suggest tht quntittively stble pool of memory tht include cells with integrted HIV DNA is mintined through continuous prolifertion nd poptosis t low levels in HAARTtreted individuls with low CD + counts. Our longitudinl nlysis shows tht the stbility of the HIV reservoir size nd the conservtion of virl sequences over time re ssocited with IL-7 concentrtions in plsm. We observed here tht IL-7 is ble to induce homeosttic prolifertion of in vitro, thereby ensuring the persistence of HIV provirl sequences. These observtions show role for IL-7 in the mintennce of ltently infected through cytokineinduced homeosttic prolifertion nd survivl. Our results indicte tht HIV persists in two reservoirs tht re mintined by distinct mechnisms. Individuls who hve strted tretment erly in infection crry virl reservoir of limited size tht is hrbored minly by cells, which hve the cpcity to survive for long periods of time,. Although these cells tht hrbor rchived provirl DNA re highly stble, the reservoir my decrese t very slow rte s consequence of the bility of these cells to mount ntigen-induced responses, leding to the elimintion of frction of infected cells through CTL killing, cytopthic effect nd poptosis during the contrction phse. The HIV reservoir is minly retrieved in cells in subjects with incresed frequencies of proliferting CD + T cells becuse HAART tretment ws initited t lter stge of the disese. This subset of continuously prolifertes t low levels, leding to the persistence of geneticlly stble HIV reservoir through IL-7 induced mitosis of cells. These results suggest tht therpeutic strtegies relying solely upon ntiretrovirl molecules will never rech the objective of virl erdiction. Finlly, our results indicte tht HIV provirl DNA is hrbored by cells tht express immune ctivtion nd prolifertion mrkers such s PD- nd Ki67. New therpies should trget pthwys downstrem of homeosttic prolifertion including inhibitors of the IL-7 pthwy or pthwys ssocited with self-renewl nd stem cell ness, such s those developed for the tretment of leukemis nd cncers 5.Indeed, by limiting immune ctivtion nd ffecting long-lived infected CD + T cells by trgeting IL-7 dependent prolifertion nd the self-renewl of memory T cells in ssocition with HAART, erdiction of virus in viremic individuls could become more relistic endevor. METHODS Methods nd ny ssocited references re vilble in the online version of the pper t Accession codes. Sequences for Env genes hve been deposited in GenBnk with ccession numbers EU75 to EU79. Note: Supplementry informtion is vilble on the Nture Medicine website. ACKNOWLEDGMENTS We thnk the study subjects for their prticiption in this study. We lso thnk M. Legult nd C. Grignon for clinicl ssistnce with subjects, N. Kettf, NATURE MEDICINE ADVANCE ONLINE PUBLICATION 7

8 9 Nture Americ, Inc. All rights reserved. M. Linesse, V. Lfontine nd Y. Chouikh for technicl ssistnce, S. Gimmig nd L. Lejeune for flow cytometric cell sorting nd T. Sing for ssistnce with the co-receptor prediction softwre. We re grteful to E. Hunter for his expertise in phylogenetic nlyses. N.C. is supported by the Americn Foundtion for AIDS Reserch (mfar, fellowship number 66-8-RFRL). M.E. nd L.T. re funded by the Cndin Institutes of Helth Reserch. J.P.R., is clinicin-scientist supported by Fonds de l Recherche en Snté du Québec (FRSQ). R.-P.S. is the Cnd Reserch Chir in Humn Immunology. This study ws supported by funds from mfar (grnts 6-7--RGRL, RGRL nd RGRL), CANFAR (grnt 88), the US Ntionl Institutes of Helth, the Cndin Institutes of Helth Reserch, the Cndin HIV Trils Network, Vccines nd Immunotherpeutics core nd the Réseu FRSQ-SIDA/mldies infectieuses. AUTHOR CONTRIBUTIONS N.C. plnned nd performed the experiments, nlyzed the dt nd wrote the mnuscript. M.E. nd F.A.P. ssisted with the cell sorting nd quntifiction experiments. P.A., L.T. nd B.Y.-D. helped with the flow cytometry experiments nd with writing the mnuscript. M.-R.B., G.G., J.M.B., T.W.S. nd J.-P.R. provided smples from reserch subjects. J.-P.R. helped with writing the mnuscript. M.-R.B. perfomed interleukin-7 mesurements. G.B. did the co-receptor prediction usge nlysis. B.J.H. nd D.C.D. performed lipopolyscchride mesurements. E.K.H. nd R.-P.S. plnned nd supervised ll experiments nd wrote the mnuscript. Published online t Reprints nd permissions informtion is vilble online t reprintsndpermissions/.. Plell, F.J. Jr. et l. Declining morbidity nd mortlity mong ptients with dvnced humn immunodeficiency virus infection. HIV Outptient Study Investigtors. N. Engl. J. Med. 8, (998).. Finzi, D. et l. Identifiction of reservoir for HIV- in ptients on highly ctive ntiretrovirl therpy. Science 78, 95 (997).. Chun, T.W. et l. Presence of n inducible HIV- ltent reservoir during highly ctive ntiretrovirl therpy. Proc. Ntl. Acd. Sci. USA 9, 9 97 (997).. Chun, T.W. et l. Quntifiction of ltent tissue reservoirs nd totl body virl lod in HIV- infection. Nture 87, 8 88 (997). 5. Finzi, D. et l. Ltent infection of provides mechnism for lifelong persistence of HIV-, even in ptients on effective combintion therpy. Nt. Med. 5, 5 57 (999). 6. Hermnkov, M. et l. Anlysis of humn immunodeficiency virus type gene expression in ltently infected resting CD + T lymphocytes in vivo. J. Virol. 77, (). 7. Chun, T.W. et l. Reltionship between the size of the humn immunodeficiency virus type (HIV-) reservoir in peripherl blood nd CD + :CD8 + T cell rtios in viremic HIV- infected individuls receiving long-term highly ctive ntiretrovirl therpy. J. Infect. Dis. 85, (). 8. Strin, M.C. et l. Effect of tretment, during primry infection, on estblishment nd clernce of cellulr reservoirs of HIV-. J. Infect. Dis. 9, 8 (5). 9. Pln, M. et l. Immunologicl benefits of ntiretrovirl therpy in very erly stges of symptomtic chronic HIV- infection. AIDS, 9 9 ().. Anthony, K.B. et l. Incomplete CD T cell recovery in HIV- infection fter months of highly ctive ntiretrovirl therpy is ssocited with ongoing incresed CD T cell ctivtion nd turnover. J. Acquir. Immune Defic. Syndr., 5 ().. Sous, A.E., Crneiro, J., Meier-Schellersheim, M., Grossmn, Z. & Victorino, R.M. CD T cell depletion is linked directly to immune ctivtion in the pthogenesis of HIV- nd HIV- but only indirectly to the virl lod. J. Immunol. 69, 6 ().. Chun, T.W. et l. Gene expression nd virl production in ltently infected, resting in viremic versus viremic HIV-infected individuls. Proc. Ntl. Acd. Sci. USA, 98 9 ().. Persud, D. et l. Continued production of drug-sensitive humn immunodeficiency virus type in children on combintion ntiretrovirl therpy who hve undetectble virl lods. J. Virol. 78, ().. Mens, H. et l. Investigting signs of recent evolution in the pool of provirl HIV type DNA during yers of successful HAART. AIDS Res. Hum. Retroviruses, 7 5 (7). 5. Persud, D. et l. Slow humn immunodeficiency virus type evolution in virl reservoirs in infnts treted with effective ntiretrovirl therpy. AIDS Res. Hum. Retroviruses, 8 9 (7). 6. Ruff, C.T. et l. Persistence of wild-type virus nd lck of temporl structure in the ltent reservoir for humn immunodeficiency virus type in peditric ptients with extensive ntiretrovirl exposure. J. Virol. 76, (). 7. Kieffer, T.L. et l. Genotypic nlysis of HIV- drug resistnce t the limit of detection: virus production without evolution in treted dults with undetectble HIV lods. J. Infect. Dis. 89, 5 65 (). 8. Sedght, A.R., Silicino, J.D., Brennn, T.P., Wilke, C.O. & Silicino, R.F. Limits on replenishment of the resting CD + T cell reservoir for HIV in ptients on HAART. PLoS Pthog., e(7). 9. Hmmrlund, E. et l. Durtion of ntivirl immunity fter smllpox vccintion. Nt. Med. 9, 7 ().. Combdiere, B. et l. Distinct time effects of vccintion on long-term prolifertive nd IFN-g producing T cell memory to smllpox in humns. J. Exp. Med. 99, ().. Riou, C. et l. Convergence of TCR nd cytokine signling leds to FOXO phosphoryltion nd drives the survivl of CD + centrl memory T cells. J. Exp. Med., 79 9 (7).. vn Grevenynghe, J. et l. Trnscription fctor FOXO controls the persistence of memory during HIV infection. Nt. Med., 66 7 (8).. Sllusto, F., Lenig, D., Forster, R., Lipp, M. & Lnzvecchi, A. Two subsets of memory T lymphocytes with distinct homing potentils nd effector functions. Nture, 78 7 (999).. Sllusto, F., Gegint, J. & Lnzvecchi, A. Centrl memory nd effector memory T cell subsets: function, genertion, nd mintennce. Annu. Rev. Immunol., (). 5. Gegint, J., Sllusto, F. & Lnzvecchi, A. Cytokine-driven prolifertion nd differentition of humn nive, centrl memory nd effector memory CD + Tcells.Pthol. Biol. (Pris) 5, 6 66 (). 6. Npolitno, L.A. et l. Incresed production of IL-7 ccompnies HIV- medited T-cell depletion: implictions for T-cell homeostsis. Nt. Med. 7, 7 79 (). 7. Koelsch, K.K. et l. Dynmics of totl, liner nonintegrted nd integrted HIV- DNA in vivo nd in vitro. J. Infect. Dis. 97, 9 (8). 8. Kinter, A.L. et l. The common g-chin cytokines IL-, IL-7, IL-5 nd IL- induce the expression of progrmmed deth- nd its lignds. J. Immunol. 8, (8). 9. Lin, S.J., Pecock, C.D., Bhl, K. & Welsh, R.M. Progrmmed deth- (PD-) defines trnsient nd dysfunctionl oligoclonl T cell popultion in cute homeosttic prolifertion. J. Exp. Med., (7)..Hokey,D.A.et l. Activtion drives PD- expression during vccine-specific prolifertion nd following lentivirl infection in mcques. Eur. J. Immunol. 8, 5 5 (8).. Siddique, M.A. et l. Low CD + T cell ndir is n independent predictor of lower HIVspecific immune responses in chroniclly HIV- infected subjects receiving highly ctive ntiretrovirl therpy. J. Infect. Dis. 9, (6).. Younes, S.A. et l. The durtion of exposure to HIV modultes the bredth nd the mgnitude of HIV-specific memory CD + Tcells.J. Immunol. 78, (7).. Younes, S.A. et l. HIV- viremi prevents the estblishment of interleukin producing HIV-specific memory endowed with prolifertive cpcity. J. Exp. Med. 98, 99 9 ().. Fry, T.J. et l. A potentil role for interleukin-7 in T-cell homeostsis. Blood 97, (). 5. Llno, A. et l. Interleukin-7 in plsm correltes with CD T-cell depletion nd my be ssocited with emergence of syncytium-inducing vrints in humn immunodeficiency virus type positive individuls. J. Virol. 75, 9 5 (). 6. Mercier, F. et l. Persistent humn immunodeficiency virus- ntigenemi ffects the expression of interleukin-7r on centrl nd effector memory CD + nd CD8 + Tcell subsets. Clin.Exp.Immunol.5, 7 8 (8). 7. Seddon, B., Tomlinson, P. & Zmoysk, R. Interleukin 7 nd T cell receptor signls regulte homeostsis of CD memory cells. Nt. Immunol., (). 8. Kondrck, R.M. et l. Interleukin 7 regultes the survivl nd genertion of memory CD cells. J. Exp. Med. 98, (). 9. Silicino, J.D. et l. Long-term follow-up studies confirm the stbility of the ltent reservoir for HIV- in resting CD + Tcells.Nt. Med. 9, ().. Zhng, L. et l. Quntifying residul HIV- repliction in ptients receiving combintion ntiretrovirl therpy. N. Engl. J. Med., 65 6 (999).. Dorndul, G. et l. Residul HIV- RNA in blood plsm of ptients tking suppressive highly ctive ntiretrovirl therpy. J. Am. Med. Assoc. 8, 67 6 (999).. Plmer, S. et l. Low-level viremi persists for t lest 7 yers in ptients on suppressive ntiretrovirl therpy. Proc. Ntl. Acd. Sci. USA 5, (8).. Kim, H. & Perelson, A.S. Virl nd ltent reservoir persistence in HIV--infected ptients on therpy. PLOS Comput. Biol., e5 (6).. Chun, T.W. et l. HIV-infected individuls receiving effective ntivirl therpy for extended periods of time continully replenish their virl reservoir. J. Clin. Invest. 5, 5 55 (5). 5. Tobin, N.H. et l. Evidence tht low-level viremis during effective highly ctive ntiretrovirl therpy result from two processes: expression of rchivl virus nd repliction of virus. J. Virol. 79, (5). 6. Rmrtnm, B. et l. The decy of the ltent reservoir of repliction-competent HIV- is inversely correlted with the extent of residul virl repliction during prolonged ntiretrovirl therpy. Nt. Med. 6, 8 85 (). 7. Joos, B. et l. HIV rebounds from ltently infected cells, rther thn from continuing low-level repliction. Proc. Ntl. Acd. Sci. USA 5, (8). 8. Ctlfmo, M. et l. HIV infection-ssocited immune ctivtion occurs by two distinct pthwys tht differentilly ffect CD nd CD8 T cells. Proc. Ntl. Acd. Sci. USA 5, (8). 9. Combdière, B. et l. CD + Ki67 + lymphocytes in HIV-infected ptients re effector T cells ccumulted in the G phse of the cell cycle. Eur. J. Immunol., (). 5. Sieg, S.F., Bzdr, D.A. & Ledermn, M.M. S-phse entry leds to cell deth in circultingtcellsfromhiv-infectedpersons.j. Leukoc. Biol. 8, 8 87 (8). 5. Dick, J.E. Stem cell concepts renew cncer reserch. Blood, (8). 8 ADVANCE ONLINE PUBLICATION NATURE MEDICINE

9 9 Nture Americ, Inc. All rights reserved. ONLINE METHODS Ptient popultion. Thirty-four HIV-seropositive subjects on suppressive HAART for n verge of 5 months (Supplementry Tble ) enrolled in this study nd signed informed consent pproved by the Royl Victori Hospitl nd the CR-CHUM hospitl review bord. All individuls were receiving vrious ntivirl regimens contining protese inhibitor, nonnucleoside reverse trnscriptse inhibitor or three nucleoside reverse trnscriptse inhibitors. None of these subjects showed ny detectble plsm viremi fter HAART initition, s ssessed by the Amplicor HIV- monitor ultrsensitive Method (Roche). All subjects underwent leukpheresis to collect lrge numbers of PBMCs. Isoltion of cellulr subsets nd flow cytometry. We purified totl CD + T cells by cell sorting fter stining with FITC-lbeled ntibody to CD (5559) nd llophycocynin (APC)-lbeled ntibody to CD (). To isolte CD + subsets, we lbeled PBMCs with FITC-lbeled ntibody to CD (5559), APC-lbeled ntibody to CD (), APC-Cy7 lbeled ntibody to CD5RA (67), phycoerythrin (PE)-Cy7 lbeled ntibody to CCR7 (55768) nd PE-lbeled ntibody to CD7 (555). Alterntively, we used PerCP-Cy5.5 lbeled ntibody to CD (65) nd APC-lbeled ntibody to PD- ( ) to isolte nd cells expressing or not expressing the PD- mrker. We seprted T cell subsets by FACSAri (BD Biosciences) to very high purity (98.%). For phenotypic nlysis, we stined PBMCs with Pcific Blue lbeled ntibody to CD (5587), Alex 7 lbeled ntibody to CD (5579), APC-Cy7 lbeled ntibody to CD5RA (67), PE-Cy7 lbeled ntibody to CCR7 (55768), PE-lbeled ntibody to CD7 (555), APC-lbeled ntibody to CD5 (555) nd FITC-lbeled ntibody to CD7(75). Alterntively, we used APC-lbeled ntibody to PD- ( ) to the surfce cocktil nd FITC-lbeled ntibody to Ki67 (5566) for intrcellulr stining. We purchsed ll ntibodies from BD Biosciences except for APC-lbeled ntibody to PD- (ebioscience). We performed nlyses on LSRII (BD Bioscience) s previously described 5. HIV DNA quntifiction. We digested sorted cells s described previously 5 nd used the lystes directly for mplifiction. We used modified nested Alu PCR to quntify both integrted HIV DNA nd CD gene copy numbers. As stndrd curve for both quntifictions, we mplified seril dilutions of ACH cells (US Ntionl Institute of Allergy nd Infectious Diseses Regent Progrm) rnging from 5 to cells t the sme time s the experimentl smples. We mplified integrted HIV sequences 5 nd the CD gene 5 for cycles in triplicte wells. We diluted the PCR products nd determined HIV nd CD copy numbers in seprte second mplifiction rections on the Light Cycler instrument (Roche Dignostics). We determined HIV nd CD copy numbers by using pproprite primers nd probes 5,5. We crried out mplifiction rections in Jumpstrt Mix (Sigm) with U of Tq Polymerse (Invitrogen),.5 mm primers nd. mm fluorescent probes. We used similr methods to quntify totl HIV DNA nd -LTR circles, with the pproprite primers nd probes 5. For -LTR circle quntifictions, we used seril dilutions of plsmid hrboring single copies of the CD gene nd -LTR junction s stndrd curve. Coculture ssys. We cultured sorted CD + T cell subsets from viremic subjects with ctivted nd dendritic cells from uninfected donors for their cpcity to mplify virl production. We differentited dendritic cells from monocytes in the presence of,5 U ml grnulocyte-mcrophge colony stimulting fctor nd ng ml IL- (R&D Systems) for d. We stimulted purified from uninfected donors with mg ml phytohemgglutinin (Sigm) nd ng ml IL- (R&D Systems) for d. We cultured.5 6 cells from sorted CD + T cell subsets with. 6 dendritic cells nd.5 6 uninfected. We collected hlf of the medium every d nd replced it with fresh medium supplemented with IL-. After d of culture, we monitored HIV repliction by p ELISA in superntnts. Env gene mplifiction. We performed Env gene mplifictions, cloning, sequencing nd phylogenetic nlyses s described previously 55.Weobtined 5 independent PCRs from ech smple nd sequenced 6 clones. To exclude the possibility of PCR resmpling, we performed single genome mplifiction ssy in some smples nd found similr results. In ddition, the clcultion of genetic divergence fter excluding geneticlly close sequences obtined from the sme cloning rection gve similr results. All sequences hve been submitted to GenBnk. Co-receptor usge prediction. We ligned trnslted gp sequences nd did co-receptor usge predictions with the support vector mchine nlysis tool genopheno[co-receptor] version. (ref. 56) with flse positive rte of.5 for CXCR detection. Interleukin-7 determintions. We determined plsm IL-7 concentrtions by the ultrsensitive Quntikine HS IL-7 immunossy Kit (R&D Systems), ccording to the mnufcturer s instructions. CD nd CD8 nd interleukin-7 stimultions of. We isolted by negtive selection (Miltenyi Biotec) nd cultured them t.5 6 cells per well. We precoted the wells with CD-specific ntibody ( mg ml ) (produced by the OKT hybridom clone), nd dded CD8-specific ntibody ( mg ml ) (BD Biosciences, 8) to the culture medium. We used IL-7 (R&D Systems) t finl concentrtions of ng ml or ng ml. We cultured cells in the presence of % FBS, mm AZT nd 5 nm ritonvir. We dded fresh medium t dy. After 8 d, we used the cells for the determintion of cell surfce phenotypes nd Env gene mplifiction. For ech condition, we used seven independent PCR products for cloning. We verified the bsence of HIV production by p ELISA of the superntnts. Sttisticl nlyses. We performed Spermn s rnk correltion nd Mnn- Whitney U tests with Prism. softwre. We considered P vlues of less thn.5 significnt. 5. Trutmnn, L. et l. Upregultion of PD- expression on HIV-specific CD8 + Tcells leds to reversible immune dysfunction. Nt. Med., 98 (6). 5. Poulin, J.F. et l. Evidence for dequte thymic function but impired nive T cell survivl following llogeneic hemtopoietic stem cell trnsplnttion in the bsence of chronic grft-versus-host disese. Blood, 6 67 (). 5. Brussel, A. & Sonigo, P. Anlysis of erly humn immunodeficiency virus type DNA synthesis by use of new sensitive ssy for quntifying integrted provirus. J. Virol. 77, 9 (). 55. Chomont, N. et l. Erly rchives of geneticlly-restricted provirl DNA in the femle genitl trct fter heterosexul trnsmission of HIV-. AIDS, 5 6 (7). 56. Skrbl, K. et l. Determining humn immunodeficiency virus coreceptor use in clinicl setting: degree of correltion between two phenotypic ssys nd bioinformtic model. J. Clin. Microbiol. 5, 79 8 (7). NATURE MEDICINE doi:.8/nm.97