ARTICLE https://doi.org/10.1038/s41467-019-12732-2 OPEN Resident memory T cells are a cellular reservoir for HIV in the cervical mucosa Jon Cantero-Pérez 1,10, Judith Grau-Expósito 1,10, Carla Serra-Peinado 1, Daniela A. Rosero 1, Laura Luque-Ballesteros 1, Antonio Astorga-Gamaza1, Josep Castellví 2, Tamara Sanhueza3, Gustavo Tapia3, Belen Lloveras4, Marco A. Fernández 5, Julia G. Prado 6, Josep M. Solé-Sedeno 7, Antoni Tarrats 8, Carla Lecumberri 8, Laura Mañalich-Barrachina9, Cristina Centeno-Mediavilla9, Vicenç Falcó 1, Maria J. Buzon 1* & Meritxell Genescà 1* 1234567890():,; HIV viral reservoirs are established very early during infection. Resident memory T cells (TRM) are present in tissues such as the lower female genital tract, but the contribution of this subset of cells to the pathogenesis and persistence of HIV remains unclear. Here, we show that cervical CD4+TRM display a unique repertoire of clusters of differentiation, with enrichment of several molecules associated with HIV infection susceptibility, longevity and self-renewing capacities. These protein profiles are enriched in a fraction of CD4+TRM expressing CD32. Cervical explant models show that CD4+TRM preferentially support HIV infection and harbor more viral DNA and protein than non-TRM. Importantly, cervical tissue from ART-suppressed HIV+ women contain high levels of viral DNA and RNA, being the TRM fraction the principal contributor. These results recognize the lower female genital tract as an HIV sanctuary and identify CD4+TRM as primary targets of HIV infection and viral persis- tence. Thus, strategies towards an HIV cure will need to consider TRM phenotypes, which are widely distributed in tissues. 1 InfectiousDiseases Department, Hospital Universitari Vall d’Hebron, Institut de Recerca (VHIR), Universitat Autònoma de Barcelona, Barcelona, Spain. 2 Pathology Department, Hospital Universitari Vall d’Hebron, UAB, Barcelona, Spain. 3 Pathology Department, Hospital Universitari Germans Trias i Pujol, Badalona, Barcelona, Spain. 4 Pathology Department, Hospital del Mar, Parc de Salut Mar, Universitat Autònoma de Barcelona, Barcelona, Spain. 5 Flow Cytometry Facility, Institut d’Investigació en Ciències de la Salut Germans Trias i Pujol, Badalona, Spain. 6 AIDS Research Institute IrsiCaixa, Institut d’Investigació en Ciències de la Salut Germans Trias i Pujol, Universitat Autònoma de Barcelona, Badalona, Spain. 7 Obstetrics and Gynecology Department, Hospital del Mar, Parc de Salut Mar, Universitat Autònoma de Barcelona, Barcelona, Spain. 8 Department of Obstetrics and Gynecology, Hospital Universitari Germans Trias i Pujol, Badalona, Barcelona, Spain. 9 Department of Obstetrics and Gynecology, Hospital Universitari Vall d’Hebron, Universitat Autònoma de Barcelona, Barcelona, Spain. 10These authors contributed equally: Jon Cantero-Pérez, Judith Grau-Expósito. *email: mariajose.buzon@vhir.org; meritxell. genesca@vhir.org NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 T he major hurdle to HIV-1 eradication is the establishment the exit of lymphocytes to tissues25,26. Further, in TRM from lung of reservoirs harboring latent HIV and low levels of viral or spleen, CD69 expression is not associated with expression of replication while on antiretroviral therapy (ART)1. The activation markers, namely CD25, CD38 and HLA-DR. This characteristics of this reservoir and the viruses that emerge during phenotype, together with a reduced expression of the prolifera- ART interruption suggest that it is established in different tissues, tion marker Ki67 and maintenance of expression of CD127 and including the mucosa2, a few days after becoming infected. Early CD28, suggests that TRM may have a low turnover and a more treatment with ART can prevent viremia and limit the size of the quiescent state, which could prevent inadequate activation in viral reservoir, but it does not prevent its establishment1. While tissues with high antigenic exposure while promoting the long- the main HIV cellular target, activated CD4+ T cells, die mas- evity of these cells25. sively during the initial infection, it is assumed that a very low In the macaque model of SIV infection, gut CD69+CD4+ proportion of the initially infected cells live long enough to be T cells, which are predominant in this tissue, are the main viral able to enter into a resting state (quiescence), establishing a target that is rapidly depleted by direct infection27. Furthermore, memory reservoir. The fact that memory CD4+ T cells generate it has been shown that HIV entry in cytobrush-derived cervical long-term protection (even during the whole life of the indivi- CD4+ T cells of HIV− women is preferential in activated cells dual), without needing a second exposure to the pathogen and expressing CD69, α4β7, and α4β1, and its enhanced susceptibility considering that its physiology includes the homeostatic pro- strongly correlates with increased CCR5 expression28. While liferation of these cells, supports theories that only the long life of these studies highlight tissue CD69+CD4+ T cells as preferential this population is sufficient for the maintenance of these reser- targets of acute HIV infection, it remains to be elucidated if these voirs2. In addition, the presence of pharmacological and immu- cells belong to the TRM phenotype and if they encompass a long- nological sanctuary sites has been associated to residual viral term reservoir during ART. Thus, although CD69 expression on replication in tissues, thus contributing to viral persistence3,4. TRM may have precluded their study in relation to HIV-1 latency, It has been determined that in blood the virus persists mostly several of their features suggest that these cells could contribute to in the central memory and, to a lesser extent, in the transitional HIV persistence: HIV-1 susceptibility, wide tissue distribution memory (CD27+) subsets5. Within the memory compartment, (including mucosal portals of entry for HIV) and local pro- CD4+ T follicular helper cells recently emerged as an important liferation and longevity of different antigen specificities25,29. Here, population that sustains HIV reservoirs6. Likewise, the con- to resolve if CD4+ TRM embody a cellular HIV-1 reservoir, we tribution to the total HIV reservoir of the stem cell-like memory focus on cervical tissue, which we also postulate may represent an T cells (TSCM), defined mainly by the expression of CD95 in a unrecognized HIV sanctuary. After detailed analysis of the CD4+ population phenotyped as naive, increases significantly after long- TRM protein profile in these mucosal tissues, which predicted term therapy, compared to other memory subpopulations7. Fur- increased susceptibility and long-term persistence, we show pre- ther, CD4+ T cells with high expression of CD2 have been pos- ferential infection of these cells using a cervical ex vivo model of tulated as potentially associated with latently infected cells8. More HIV infection. To address if CD4+ TRM are indeed a source of recently, CD32 expression in a subset of CD4+ T cells was also HIV persistence, we studied cervical tissues from aviremic ART- associated to HIV-latently infected cells9, but soon after it was treated HIV-1 infected women. Critically, we detect higher vDNA demonstrated to be linked to transcriptionally active infected content in cervical tissues compared to contemporaneous blood cells10–13. Similarly, the effector memory phenotype and the samples from these patients, and show that CD4+ TRM harboring expression of CD30 have also been associated to transcriptionally vDNA and vRNA are the main contributors to this reservoir. active infected cells14,15. Nevertheless, most of these populations have been determined in blood, while in tissues is where the main reservoir resides. In this sense, assessment of viral DNA (vDNA) Results and viral RNA (vRNA) positive cells in effectively ART-treated Evaluation of TRM cell markers in cervical CD69+ cells. In macaques (plasma vRNA < 50 copies/mL) has shown virus recent extensive TRM analyses of human tissues25,26, the female remaining in every tissue examined3. Further, in ART-suppressed genital tract has not been included and, thus, the precise char- patients proviral DNA determined in cells obtained from ileum acteristics of cervical TRM remain largely undefined. Still, in the biopsies16 or from bronchoalveolar lavage17 is enriched compared female reproductive tract of mice, a recent publication demon- to blood, and even CD4+ T cells from adipose tissue represent a strates that >90% of the CD4+ T cells are essentially TRM30. In niche of viral persistence18. In macaque models these reservoirs order to confirm that CD69 expression in these human mucosal are established within three days after infection19 and therefore tissues defines cells compatible with a TRM profile reported by target cells present in the initially infected tissues may have great others25,31,32, we first analyzed the expression of several tran- relevance in the generation and persistence of these reservoirs. scriptional factors and surface proteins previously associated to Thus, tissues from the female genital mucosa, the portal of entry TRM in a small group of samples. The gating strategy used for of HIV, may represent an unrecognized reservoir, as these analyses and examples of the expression of these markers in suggested20,21. In this sense, the most abundant hematopoietic the CD69− or the CD69+ fraction of CD4+ T cells derived from cells in the cervix are CD14+/CD68+ cells, compatible with dif- cervical tissues are shown in Fig. 1a, b, respectively. As reported ferent myeloid phenotypes, and CD4+ T cells22,23 likely belong- for human and mouse TRM32, Eomes and T-bet were mostly ing to the resident memory T cell (TRM) phenotype, according to absent from cervical CD69+ CD4+ T cells, while a median of their CD69 expression24. 1.51% and of 3.5% CD69−CD4+ T cells expressed Eomes and CD4+ TRM reside in various tissues in mice and humans and T-bet, respectively (Fig. 1c, p = 0.0625 for T-bet). As recently some of their phenotypic characteristics have recently been reported for CD69+ CD8+ T cells derived from other human established25,26. Since CD69 is a hallmark of these cells and, in tissues25, Hobit was minimally expressed in the CD69− fraction general, the majority of cervical CD4+ T cells express this acti- (median of 1.71%), but absent from the CD69+ CD4+ T cells vation marker24, this would question the role for these cells in the (p = 0.0625). S1PR1 and CCR7, molecules expected to be down- establishment of viral latency, as presumably they do not repre- regulated in TRM25,31,32, were only detected in 3.18% and 3.05% of sent quiescent cells. However, the expression of this molecule in the CD69+ CD4+ T cells, respectively (Fig. 1c, p = 0.0156 for TRM may have a role in tissue retention, since it regulates the CCR7). In addition, CD49a and PD-1, the expression of which sphingosine-1 phosphate receptor, which in turn participates in has also been associated to the core signature of TRM25, were 2 NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 ARTICLE a b 250 K 250 K 250 K CD69– CD69+ 200 K CD45 200 K 200 K 10 5 10 5 5.87% Single cells Single cells2 77.7% FSC-H 98.8% SSC-H SSC-A 150 K 150 K 150 K 104 16.9% 104 0.076% 100 K 100 K 100 K T-bet BV421 3 3 10 10 50 K 50 K 50 K 0 0 3 6.12% 3 2.18% 0 0 0 –10 –10 3 3 4 5 –10 0 10 10 10 0 50 K 100 K 150 K 200 K 250 K 0 50 K 100 K 150 K 200 K 250 K 3 3 4 5 3 3 4 5 –10 0 10 10 10 –10 0 10 10 10 CD45 AlexaFluor 700 FSC-A SSC-A Eomes PE-Cy7 250 K 10 5 105 CD4 105 105 2.72% 39.1% 200 K 46.7% CD69 PE-CF594 4 4 CD69+ CD4 BV605 10 10 104 104 80.6% SSC-A 150 K 103 103 103 100 K 103 CD69– 0 Lymphos 0 50 K 0 10.1% 71.6% 0 3.11% 3 2.43% 3 –103 –10 –10 0 3 3 4 5 3 3 4 5 3 3 4 5 3 3 4 5 4 5 –10 0 10 10 10 –10 0 10 10 10 –10 0 10 10 10 –10 0 10 10 10 0 10 10 S1PR1 eFluor 660 CD19 V500 / viability CD3 eVolve 655 CD103 FITC 105 105 4.45% 39.9% c p = 0.0156 104 104 30 100 p = 0.0156 CD49a PE p = 0.0625 CD69– 25 103 103 20 80 CD69+ 0 0 4.45% 3 0.046% –103 –10 Frequency (%) 15 –103 0 103 104 105 –103 0 103 104 105 60 Hobit AF647 p = 0.0313 105 38.5% 105 76.8% 10 p = 0.0625 40 104 104 5 20 PD-1 BV421 103 103 0 0 0 0 –103 7.55% –103 0.86% –103 103 104 105 –103 103 104 105 a es -1 0 0 1 t it 7 be 49 ob PR R PD m CCR7 PE C T- D H Eo S1 C C Fig. 1 CD4+ TRM identification in cervix. a General gating strategy for phenotyping of CD4+ T cells obtained from cervicovaginal tissue of healthy donors. Gating strategy consisted of selecting hematopoietic CD45+ cells, followed by a double doublet exclusion, dead and CD19+ cells exclusion and finally a CD3+ CD4+ T cell gate from where CD69+/− cells were identified. b Representative flow cytometry plots of the expression of different cell-surface proteins and transcriptional factors in the CD4+CD69+/− T cell subsets from the cervical tissue of healthy donors (CD69− on the left column, CD69+ on the right column). c Frequency of different cell-surface proteins and transcriptional factors shown in b for CD4+CD69− T cells (empty circles) and CD4+CD69+ T cells (full circles; n = 5 for S1PR1; n = 6 for Eomes, T-bet, Hobit, CD49a; n = 7 for CCR7, PD-1). Lines and error bars represent median and interquartile ranges. Statistical comparisons were performed using Wilcoxon matched-pairs signed rank test to compare the two groups. Source data are provided as a Source Data file remarkably higher in CD69+ CD4+ T cells with values of around that a fraction of cervical CD4+ TRM consistently expressed low 40.4% for CD49a and of 76.45% for PD-1 compared to the levels of CD32 in uninfected tissues, a molecule which expression CD69− fraction (median values of 4.26% for CD49a and of 36.3% has been associated with the identification of transcriptionally for PD-1, p = 0.0156 for PD-1 and p = 0.0313 for CD49a). All active infected cells during HIV infection10,11,13, we analyzed this together, these analyses confirmed that >95% of the CD4+ T cells fraction separately to determine phenotypic characteristics expressing CD69 in cervix are phenotypically associated to bona that could render these cells more vulnerable to HIV infection. fide TRM25,31,32. Figure 2a shows the general gating strategy established for these analyses (which follows the gating already established in Fig. 1a). Since several recent papers focused on demonstrating that CD32 CCR5+ CD4+ TRM protein expression signature in cervix. In expression does not enrich for HIV-latently infected cells raised order to determine if CD4+ T cells expressing CD69, from now concerns on the purity of these cells33–35 and, in order to elim- on called TRM, represent a reservoir for HIV, we first further inate potential contamination by B cells doublets (CD32 high characterized the expression profile of TRM as potential HIV-1 subset), we only analyzed the fraction of CD4+ T cells with dim targets. Of note, for these analyses, we considered endocervix and expression of CD32. We used the levels of CD32 expression in B ectocervix separately, since differences exist between these cells (indicated with a dotted black line, Fig. 2a) and of the cor- mucosal tissues in terms of structure, cellular components and responding fluorescence minus one control (Supplementary ex vivo HIV replicative capacity23. Considering that HIV-1 iso- Fig. 1a) to define a narrow dim CD32+ gate. Further, the exis- lates from newly infected individuals predominantly use CCR5 as tence of these cells in tissue was confirmed by Amnis imaging a co-receptor, we performed a detailed analysis of the phenotype flow cytometry, which allowed visualization of individual CD4+ of cervical CD4+ CCR5+ CD69+ (TRM) and CD69− (non-TRM) CD69+ (TRM) cells expressing dim levels of CD32 (Supplemen- cells from HIV uninfected tissues. Moreover, since we observed tary Fig. 1b and 1c). These data together with recent publications NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 a 250 K b c + + + 200 K CD4 cells CD4 CCR5 cells 100 35 p = 0.031 SSC-A 150 K 100 K Lymphos 30 p = 0.031 105 71.6% 80 CD4 46.7% 50 K 20 Frequency (%) B cells Frequency (%) 104 CD4 BV605 6.19% 0 3 3 4 5 60 p = 0.004 –10 0 10 10 10 15 103 CD19 V500/viability 40 0 10 –103 0 103 104 105 20 5 CD3 eVolve 655 TRM CD32 5.70% TRM 0 0 105 DR+ 105 105 CD69+ + + + + + 8.82% M M R R 32 32 R R T /T 5 79.5% CD69 PE-CF594 + R HLA-DR BV421 D D HLA-DR BV421 D D C 104 104 104 5 C C R C M M R R C T T M M n- C R R T T no n- 103 103 103 no Ectocervix 0 0 0 CCR5 Excluded 36.5% DR+ CD32+ B cells Endocervix –103 –103 –103 –104 0 104 105 0 104 105 0 104 105 CCR5 APC-Cy7 CD32 PE-Cy7 CD32 PE-Cy7 d p = 0.008 e p = 0.031 p = 0.016 f 50 100 15.0 p = 0.031 p = 0.031 p = 0.031 p = 0.031 p = 0.016 40 80 12.5 p = 0.031 p = 0.008 10.0 p = 0.031 % CD103 % α4β7 30 60 % α1 7.5 20 40 5.0 10 20 2.5 0 0 0.0 CD69 – + + – + + CD69 – + + – + + CD69 – + + – + + CD32 ± – + ± – + CD32 ± – + ± – + CD32 ± – + ± – + Ectocervix Endocervix Ectocervix Endocervix Ectocervix Endocervix g p = 0.008 h i 25 p = 0.031 30 p = 0.031 25 p = 0.008 20 25 20 p = 0.008 p = 0.008 20 % CXCR6 % CXCR4 % CCR6 15 15 p = 0.008 p = 0.031 15 p = 0.016 10 10 10 5 5 5 0 0 0 CD69 – + + – + + CD69 – + + – + + CD69 – + + – + + CD32 ± – + ± – + CD32 ± – + ± – + CD32 ± – + ± – + Ectocervix Endocervix Ectocervix Endocervix Ectocervix Endocervix j k p = 0.008 l p = 0.031 p = 0.016 20 30 80 p = 0.016 p = 0.031 p = 0.031 p = 0.008 70 25 p = 0.031 15 60 20 % CD132 % CD127 p = 0.016 % CD122 50 p = 0.008 p = 0.031 10 15 p = 0.031 40 30 10 5 20 5 10 0 0 0 CD69 – + + – + + CD69 – + + – + + CD69 – + + – + + CD32 ± – + ± – + CD32 ± – + ± – + CD32 ± – + ± – + Ectocervix Endocervix Ectocervix Endocervix Ectocervix Endocervix demonstrating that CD32–RNA correlates with HIV transcrip- within total CD4+ T cells or within CD4+ CCR5+ T cells was tional activity in T cells from the gut13, and from lymph nodes10, around 80% in both tissues (Fig. 2b). Also, a small subset of CD4+ not only confirms their existence in human tissues, but also raises CCR5+ TRM expressed HLA-DR or CD32, which in the case of interest on defining their nature. CD32+ were significantly more frequent in ectocervix than in Out of total CD4+ T cells, a median of 45.8% expressed CCR5 endocervix (p = 0.031, Fig. 2c). Importantly, while HLA-DR in ectocervix and a median of 41.9% expressed CCR5 in expression was similar in CD4+ CCR5+ regardless of CD69 endocervix (Fig. 2b). Further, the median frequency of TRM expression, with a median of 5.9% and 3.7% in TRM and of 9.9% 4 NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 ARTICLE Fig. 2 CCR5+ CD4+ TRM protein expression signature in cervix. a Gating strategy for phenotyping of CD4+ T cells obtained from cervicovaginal tissue of healthy donors. Based on the gating strategy shown in Fig. 1a, a lymphocyte gate (marked in red) was selected in a plot where dead and CD19+ were excluded. From one hand, CD3+ CD4+ T cells and then CCR5 positive cells were selected to further analyze three different subsets based on CD69 and CD32 expression: non-TRM (yellow gate), TRM CD32− (orange gate) and TRM CD32+ (blue gate). Of note, to identify CD4+ T cells expressing low levels of CD32 (CD32dim, blue gate), we previously defined a gate of exclusion in putative B cells (CD19+, depicted in dotted black lines) as these cells express high levels of CD32. b Frequency of diverse subsets within CD4+ T cells from human ectocervix (n = 6) and endocervix (n = 9). c Frequency of diverse subsets within CD4+ CCR5+ T cells from human ectocervix (n = 6) and endocervix (n = 9). d–l Frequency of expression of diverse cell-surface proteins within three different CD4+CCR5+ T cell subsets; namely CD69− non-TRM (yellow), CD32− TRM (orange) and CD32+ TRM (blue), in human ectocervix (n = 6) and endocervix (n = 7 for 2e, 2f; n = 8 for 2d, 2h, 2i, 2j, 2k, 2l; n = 9 for 2b, 2c, 2g). Lines and error bars represent median and interquartile ranges. All statistical comparisons were performed using Wilcoxon matched-pairs signed rank test. Source data are provided as a Source Data file and 5.3% in non-TRM for ectocervix and endocervix respectively; we also analyzed CCR7, which was minimally expressed in all CD32 expression was strongly associated to TRM compared to subsets (<5% in general) and, as expected, more associated to non-TRM (median of 14.3% and 8.1% in TRM vs. 5.2% and 3.7% in non-TRM that will recirculate (p = 0.031, Supplementary Fig. 2d). non-TRM for ectocervix and endocervix respectively; p = 0.031 for CXCR3 expression was frequent in CD32− TRM and even more ectocervix and p = 0.004 for endocervix, Fig. 2c). so in CD32+ TRM of both tissues (p < 0.032 in all cases); CD161 Then, we analyzed the expression of several surface markers was highly expressed in both CD4+ CCR5+ TRM cell fractions related to HIV-susceptibility and persistence in a total of three (p < 0.032 for each comparison), as shown before for other TRM different CD4+ CCR5+ T cell subsets: non-TRM, CD32− TRM and phenotypes in tissue26; while CCR2 and γδ T cells were hardly CD32+ TRM (Fig. 2a). Examples of individual markers are shown detected in any of the subsets studied, except for the CD32+ TRM in Supplementary Fig. 2a. We observed that TRM, including CD32+ fraction resident in the endocervix, which showed higher TRM, expressed significantly higher levels of CD103 (integrin αE) expression of CCR2 compared to non-TRM (p = 0.031, Supple- than non-TRM in ectocervix (p = 0.031 in both cases) and mainly mentary Fig. 2e-h). in endocervix (p = 0.008 in both cases, Fig. 2d). This was expected Together, these data not only confirm expression of some of since CD103 is associated to intraepithelial retention of CD8+ TRM the markers associated to the TRM phenotype in other tissues25,26 and, to less extent, of CD4+ TRM25,26. Other integrins have also in cervical CCR5+ TRM, but also strongly suggest two concepts been associated with retention of TRM in tissues, and we observed related to our hypothesis: (1) TRM express several molecules that that in cervical tissue, integrin α1 (CD49a) is highly associated to may render them highly susceptible to HIV infection (namely CD4+ CCR5+ T cells displaying a TRM phenotype (p < 0.035 for all α4β7, CXCR4, CXCR6 and CCR6), and (2) these cells are comparisons, Fig. 2e). Additionally, we detected an enrichment of potentially self-renewed and maintained long-term in tissues (i.e., α4β7 expression on CD32− and CD32+ TRM in the endocervix and CD122, CD132, CD127). Both concepts apply to TRM in general on CD32+ TRM in the ectocervix (p = 0.031 for all comparisons, and, in particular, to a small fraction of TRM co-expressing low Fig. 2f), indicating that some TRM could represent preferential levels of CD32, which is even more enriched in most of these targets for HIV-128,36. Other HIV co-receptors, such as CXCR4 surface proteins. Thus, although CD32 expression in blood is and CXCR637, were also more frequent in the TRM fractions, where associated to concomitant expression of CD69 indicating early CD32+ TRM presented even higher frequencies compared to TRM activation10,11, in cervical tissue CD4+ T cells expressing CD32 not expressing CD32 (Fig. 2g, h). Expression of CCR6, a are enriched in markers associated to truthfully TRM phenotypes. chemokine receptor associated to T helper 17 (Th17) cells, was Overall, considering that the TRM phenotype is the predominant also strongly associated to TRM and, mostly, to CD32+ TRM from CD4+ T cell subset in cervical mucosal tissues and their quiescent the endocervix, which again showed even higher expression state, exemplified by low HLA-DR here or, as reported before, low compared to CD32− TRM (p = 0.008, Fig. 2i). Ki6725, these cells are strong candidates to embody viral We also studied the presence of several interleukin receptors reservoirs in different peripheral tissues. (IL), namely CD122, CD132 and CD127, as they may confer characteristics of longevity and homeostatic proliferation, with CD4+ TRM are preferentially infected during ex vivo infection. strong implication for maintenance of HIV reservoirs5. CD122 In order to demonstrate that CD4+ TRM from cervix represent a (β-chain) and CD132 (γ-chain) together form the IL-2 and IL-15 significant cellular reservoir, we first aimed to address if this receptors, while CD127 (α-chain) and CD132 form the IL-7 subset supports infection ex vivo and which are the dynamics of receptor. We detected higher frequency of all three IL receptors in CD69 after HIV infection in these tissues. For this, cervicovaginal CD4+ CCR5+ TRM compared to non-TRM in both tissues (p < tissues from healthy donors were infected with 7200 TCID50 of an 0.032 for all comparisons, Fig. 2j–l). However, CD122 and CD132 HIV-1BaL viral stock38 for 10–12 days, and infection was mea- expression was associated to a small fraction of CD4+ CCR5+ sured by flow cytometry analysis of intracellular p24 staining and TRM, which was higher in the CD32+ TRM fraction compared to by cell-associated HIV-1 DNA in purified subsets. Since CD4 and the CD32− fraction (Fig. 2j, k). In contrast, CD127 was expressed CCR5 molecules are down-regulated by the accessory protein Nef in a median of 34.8% of non-TRM and in > 55% of CD32+/− TRM, during HIV-1 infection39, we analyzed p24 viral antigen expres- confirming that, as occurs with other tissues25, there is also sion in a gate including CD4+ and CD4−(CD8−) T cells, as retention of this receptor in memory cells from the cervical shown in the gating strategy (Fig. 3a). For each individual cervical mucosa (Fig. 2l). Surprisingly, while the frequency of each IL- explant infected, the frequency of p24+ cells ranged from 1.92 to chain individually was remarkably higher in TRM compared to 13.4% of CD4+/− T cells (median of 7.3%), in all cases evidencing non-TRM, there were few cells co-expressing CD132 and CD122, CD4 down-regulation (Fig. 3b). When analyzed by subsets, or CD127 and CD132 within the CD32− TRM fraction expression of p24 viral antigen was significantly higher in TRM, (Supplementary Fig. 2b and 2c). In contrast, a small but CD103+ TRM and CD32+ TRM compared to non-TRM (p = 0.004, significant fraction of cells co-expressing these receptors was p = 0.020 and p = 0.008 respectively; Fig. 3b). This increase cor- found in non-TRM compared to TRM, while CD32+ TRM showed responded to a median of 5.02 times more infection in TRM even higher frequencies (Supplementary Fig. 2b and 2c). Further, compared to non-TRM, and of 7.09 times more in the case of NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 a 250K 250K 105 CD45 105 105 Non-conjugates 71.2% LIVE CD3+ CD19– 99.0% CD19 V500/Viability HLA-DR BV421 200K 200K HLA-DR BV421 10 4 86.5% 104 104 Single Cells SSC-A FSC-H 150K 150K 103 96.0% 100K 100K 103 103 0 0 0 50K 50K Lymphocytes 6.71% –103 –10 3 –103 0 0 0 50K 100K 150K 200K 250K 0 50K 100K 150K 200K 250K –103 0 103 104 105 –103 0 103 104 105 –103 0 103 104 105 FSC-A FSC-A CD45 Alexa700 CD3 PerCP CD32 PE-Cy7 105 105 TRM 82.8% 105 HLA-DR 105 105 7.07% BaL (HIV+) Control (HIV–) CD69 PE-CF594 HLA-DR BV421 CD4 BV605 104 104 CD103+ 104 104 104 CD8 APC 17.0% p24+ p24+ CD4 11.8% 0.29% CD4 BV605 103 59.9% 103 103 103 103 0 0 0 0 0 CD32 –103 –10 3 Non-TRM 17.2% –10 3 9.09% –103 –10 3 –103 0 103 104 105 –103 0 103 104 105 –103 0 103 104 105 –103 0 103 104 105 –103 0 103 104 105 CD4 BV605 CD103 FiTC CD32 PE-Cy7 p24 PE b p = 0.008 c p = 0.016 d p = 0.004 50 100 101 p = 0.020 % of cells within p24+ % of intracellular p24 40 HIV DNA molec/cell 80 p = 0.004 0 10 60 30 20 40 10–1 10 20 0 10–2 0 – Non-TRM TRM Non-TRM TRM M M +/ M M R R R R -T T T T 4 + + D on 3 32 C 10 N D D C C e f g p = 0.008 50 100 40 Frequency of HLA-DR (%) 45 p24+ HIIV DNA molec/TRM cell CD32 frequency (%) p24– 10 30 30 p = 0.027 1 20 20 10 0.1 10 0 0.01 0 CD69 + + CD69 + + + + M M 32 3 R R -T T 10 CD32 + – p24 + – D on D C C N M M R T R T Fig. 3 TRM are preferentially infected in the cervical explant model of HIV infection. a General gating strategy for phenotype analyses and intracellular p24 detection in CD4+/− TRM and non-TRM obtained from cervicovaginal tissue infected ex vivo with HIV-1BaL strain. Gating strategy consisted of a lymphocyte gate based on FSC vs. SSC, followed by doublet exclusion, hematopoietic CD45+ cells, live CD3+ T cells with CD19+ cells exclusion, further B cell exclusion and a CD4+/− T cell gate from where the different gates shown were quantified: CD4+/− non-TRM (purple) or CD4+/− TRM (orange) or HLA-DR/ CD32dim expression. An example of p24 expression from an infected and a non-infected sample is shown. b Frequency of p24 viral antigen in total CD4+/− cells (green), CD4+/− non-TRM (purple), CD4+/− TRM (orange), CD4+/− CD103+ TRM and CD4+/− CD32+ TRM detected after 10–12 days of ex vivo infection and measured by flow cytometry (n = 9, except for CD4+/− CD32+ TRM where n = 8). c Number of HIV-DNA molecules per cell in non- TRM and TRM detected after 10–12 days of ex vivo infection and measured by qPCR (n = 7). d Frequency of non-TRM and TRM from the total p24 positive cells measured by flow cytometry (n = 9). e Frequency of HLA-DR expression in different subsets based on their expression of p24 viral antigen after 10–12 days of ex vivo infection (n = 9, except for CD4+/− CD32+ TRM where n = 8). f Number of HIV-DNA molecules per cell in TRM CD32+ and TRM CD32− detected after 10–12 days of ex vivo infection and measured by qPCR (n = 4). g CD32 frequency in p24+ TRM compared to p24− TRM, measured by flow cytometry (n = 8). For b and e the median is shown as a solid line, the box indicates the 25–75 percentile range, while the whiskers show the range. Statistical analysis consisted of a Wilcoxon matched-pairs signed rank test. Source data are provided as a Source Data file 6 NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 ARTICLE CD32+ TRM. Moreover, regardless of the frequency of infection, indicating preferential infection of these cells, as opposed to cell-associated HIV-1 DNA measured by qPCR was more HIV-induced changes in receptor expression41. It remains to be abundant in CD4+/− TRM with a median of 4.4 vDNA molecules/ elucidated if CD32 expression on CD4+ TRM remains stable in cell compared to 0.43 in non-TRM (p = 0.016, Fig. 3c). This the HIV-1 infected cells over the virus life cycle or is modulated finding was also observed for productively infected cells, namely by infection, and mostly associated to a fraction of activated cells CD4+/− p24+ cells, which contained higher proportion of TRM as proposed10. (>50% in all cases) than of non-TRM (p = 0.004, Fig. 3d). Unexpectedly, HLA-DR expression in all subsets from these HIV infection impacts on cervical CD4+ TRM in women. Next tissues remained very low, regardless of productive infection we determined if CD4+ TRM remain in tissue from HIV infected (Fig. 3e). As shown, the percentages of this marker were most patients, since they may be impaired as a consequence of the often <20% in all subsets. Productive infection (p24+) of total retroviral infection, as suggested42. Thus we compared cervical TRM was associated with a significant increase of HLA-DR CD4+ T cells from two groups of patients: cervicovaginal tissue of expression compared to their uninfected counterparts (p = 0.027; previously analyzed normal healthy donors (ND) and ART- Fig. 3e). Yet, these values were disproportionally lower than the suppressed HIV+ patients (Table 1). As expected, the frequency levels observed on HIV infected cells from blood where, for of CD4+ cells out of total CD45+ lymphoid-cells decreased from example, CD32+ CD4+ infected cells have been shown to have a median of 32.2% in ND to about 20.2% in HIV+ patients (p = frequencies of HLA-DR expression >80%10. These data suggest 0.002, Fig. 4a). Importantly, the frequency of TRM within the that in cervical tissue activation markers and their dynamics may same fraction was also lower in HIV+ patients compared to ND not be comparable to blood. In fact, since HIV infection per se (p = 0.006, Fig. 4a). Further, the median of CD32+ TRM out of the could induce up-regulation of CD69 on infected cells from CD45+ lymphoid-cells was of 0.52% in HIV+ patients compared peripheral blood, we determined the dynamics of CD69 to 1.87% in ND (p = 0.006, Fig. 4a). Additionally, cervical CD4+ expression and HLA-DR over 10 days of infection in cervical HLA-DR+ T cells were proportionally increased in HIV+ women tissue (Supplementary Fig. 3a). Surprisingly, the frequency of with a median of 3.15% compared to 1.29% in ND (Fig. 4a), CD69 expression decreased over time, with no significant changes similar to what has been described20. in HLA-DR expression, similarly to what we observed in the When we determined the frequency of HLA-DR in different concomitant non-infected control. fractions, we observed that in virally suppressed HIV+ patients, In addition, we separated CD69+ and CD69− CD4+ T cells HLA-DR expression was significantly enhanced compared to ND from fresh cervical suspensions, which we immediately infected to in total CD4+ TRM (p < 0.0001), TRM CD103+ (p = 0.01), and in evaluate infection (p24) and CD69 expression. From a total of four non-TRM (p = 0.001), suggesting a higher degree of activation of individual tissues, a median of 3.23% CD69+ were p24+ 3 days all these cells in infected compared to uninfected tissues (Fig. 4b). after infection, while only in one out of four tissues we detected few In contrast, the frequency of CD32 in the different subsets was p24 positive cells (0.21%) in the CD69− fraction (Supplementary similar between the two groups, except for a trend to lower Fig. 3b). Moreover, in these experiments we detected minimal frequencies of CD32+ CD103+ TRM cells in virally suppressed enhancement of CD69 expression in the CD69− fraction HIV+ patients compared to ND (p = 0.0756; Fig. 4c). However, (Supplementary Fig. 3b). To further confirm the residency nature as mentioned earlier, it was clear that CD32 was strongly of most of the cervical cells supporting infection, we stimulated 10 associated to the CD4+ TRM phenotype, since higher percentages day-infected tissue blocks with CCL19, CCL21, and S1P overnight of CD32 expression were detected in total CD4+ TRM and CD4+ to attract non-TRM out of the tissue in a transwell migration assay. CD103+ TRM of both groups compared to their corresponding CCL19 and CCL21 are chemokine-ligands attracting CCR7 non-TRM fractions (p < 0.05 for all comparisons, Fig. 4c). expressing cells, while S1P promote egress of cells expressing Together, these data suggest a direct impact of HIV infection S1PR140. Next day, we determined the level of infection in tissue on the CD4+ TRM fraction of the lower genital tract, as evidenced blocks, as well as in the supernatant (Supplementary Fig. 3c). This by its depletion and activation, which remains after >10 years of experiment demonstrated higher frequency of p24+ cells retained effective ART suppression in HIV+ women. within the tissue compared to the supernatant (Supplementary Fig. 3c). In addition, CD69 expression in total CD8− T cells was higher within the tissue (~60–81%) than in the supernatant TRM are the main cellular reservoir in the cervical mucosa. (~35–52%). Interestingly, while productive infection was again Tissue body reservoirs of HIV-1 or SIV have mainly focused on strongly associated to the TRM phenotype in the tissue (with >72% the analyses of virus burden in blood, lymphoid tissues and of the p24+ cells expressing CD69), most of these infected cells did gastrointestinal tract3,16, while other mucosal tissues such as the not express the α-chain of the IL-7 receptor, CD127, also associated female genital tract have not been accurately evaluated for HIV-1 to the TRM phenotype in healthy cervical tissue (Supplementary persistence. To address if the cervical mucosa is a site of viral Fig. 3c). persistence and if mucosal TRM represent a cellular reservoir, we Lastly, in four of these cervicovaginal explants infected ex vivo, compared contemporaneous blood and cervical samples from 8 in which, after tissue processing, a high number of T cells were HIV-infected women who had been ART-suppressed for at least a obtained, we further purified CD4+/− TRM expressing CD32 to year, with a median of 4 years of suppression (Table 1). First, we determine their vDNA content. Although limited by the small compared the total amount of vDNA in total CD4+ and CD4− number of experiments, there was a trend towards higher content (CD8−) T cells from tissue and blood, using the same gating of vDNA per cell in CD32+ vs. CD32− CD4+/− TRM (Fig. 3f). In strategy shown for the ex vivo model to include infected CD4+ addition, although most of the p24+ cells were not CD32+, CD32 T cells that down-regulate this receptor (Fig. 3a). Importantly, expression was significantly more frequent in p24+ TRM cervical tissue contained significantly more HIV-DNA copies per compared to p24− TRM (p = 0.008, Fig. 3g). Taken together, 106 CD4+/− T cell than blood (p = 0.015; Fig. 5a). While CD4+/− these results indicate that CD4+ TRM preferentially support HIV T cells from blood harbored between 28 and 4812 vDNA copies/ infection ex vivo, as demonstrated by higher vDNA content per 106 CD4+/− T cells, cervical cells contained between 1366 and cell and by high expression of intracellular p24 viral antigen. 23,270 copies per million cells, which corresponded to fold These data concur with recently reported high CD69 expression changes of up to >200 times more provirus per cell in cervix than on infected tonsillar T cells and predicted precursor cells, in blood (Fig. 5a). Of note, the number of sorted cells highly NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 Table 1 Clinical data of HIV-infected patients included in the study # Patient ID Analyses Age (yr) Time since HIV Viral load CD4 cell count Time on ART- ART regimen diagnosis (yr) (copies/ml) (cells/µl) suppressed (yr) M00 DNA 55 >12 <50 566 >10 TDF + FTC + ETR M01 Flow 51 3.41 <50 420 3 3TC + ABC + DTG M02 DNA/Flow 52 27 <50 940 13.58 3TC + ABC + RAL M03 DNA/Flow 54 16.58 <50 1260 16 TAF + FTC + EVG/c M04 DNA/Flow 26 4 <50 520 3.67 3TC + ABC + DTG M05 DNA/Flow 57 27 <50 1400 15.33 TDF + FTC + EVF M11 Flow 44 13.16 <50 440 11.66 3TC + DRV/c M15 Flow 48 29.50 <50 380 16.50 3TC + ABC + DTG M16 Flow 49 30.50 <50 490 2.64 DRV/c + DTG M17 Flow 47 26.58 <50 820 20.67 3TC + ABC + DTG M18 Flow 63 19.54 <50 160 18.45 3TC + DTG M19a Flow 51 31.50 <50 200 6.82 DRV/c + DTG M20a DNA/Flow 53 29 <50 1257 1.46 ATV/r + 3TC + ddI M21 Flow 49 22.79 <50 850 14.66 3TC + ABC + NVP M22 DNA/Flow 48 26.50 <50 850 2.55 3TC + ABC + DTG M23a DNA/Flow 46 23.46 <50 2850 2.09 3TC + DTG M25 DNA/Flow 48 17.62 <50 1040 13.87 EVG/c + FTC + TAF M26 Flow 44 26 <50 511 >15 3TC + ABC + DTG M27a Flow 54 34.39 <50 1380 0.71 DRV/c + RPV M28 DNA 54 20 <50 750 4.5 3TC + ABC + DTG P1 RNA 34 9.5 5100–10,400 760 − UNTREATED P2 RNA 30 3.42 <50 380 3.16 3TC + ABC + DRV + RTV P3 RNA 49 4.25 <50 910 4 3TC + ABC + DTG P4 RNA 49 0.5 <50 280–350 0.25 TAF + FTC + EVG/c P5a RNA 47 24.16 <25 811 4 EFV + FTC + TDF P6 RNA 51 11.25 <25 215 2 EFV + FTC + TDF + EFV P7 RNA 46 20.16 <25 828 5.5 DRV aAll infections correspond to diagnosed heterosexual HIV-1 transmission except M19, M20, M23, M27, and P5, which used to be injecting drug-users. TDF tenofovir disoproxil fumarate, FTC emtricitabine, ETR etravirine, 3TC lamivudine, ABC abacavir, DTG dolutegravir, RAL raltegravir, TAF tenofovir alafenamide fumarate, EVG/c elvitegravir boosted with cobicistat, EVF efavirenz, NVP nevirapine, DRV/c darunavir boosted with cobicistat, DRV darunavir, RTV ritonavir, RPV rilpivirine, ATV atazanavir, ddI didanosine varied based on the sample size and the patient, and probably In contemporary blood samples from the same patients, we limited detection of vDNA in tissue from other patients (Sup- used the same flow cytometry panel to isolate four different plementary Table 1). In fact, 15 more samples were also pro- fractions from total CD4+/− T cells: CD69+ (in blood identifying cessed, but the low yield obtained after tissue processing limited recently activated cells), CD103+ (a potential surrogate marker of performing any of these analyses. Nonetheless, patients with tissue-derived T cells43,44), CD32+ (here not CD69 or CD103) detectable vDNA in cervix had undetectable viral load in blood and CD4+/− T cells not expressing any of these three markers for at least 1 year and up to >11 years (Table 1). In cervical tissue (Fig. 5d). We detected similar median levels of vDNA content in of seven of these patients, by sorting CD69+ TRM and CD69− all subsets and no statistical differences were observed among non-TRM, we demonstrated that the vDNA detected in these them (Fig. 5d), regardless of different sorted cell numbers patients was associated to the TRM fraction in all but one patient, (Supplementary Table 1). However, when we analyzed the who had detectable vDNA in both TRM (5073 copies/106) and contribution of the different subsets to the blood reservoir, most non-TRM cells (1606 copies/106). Of note, although in most tis- of the contribution to the total pool of HIV in CD4+/− T cells sues TRM were purified in higher numbers than non-TRM, in was due to CD69−CD103− CD32− CD4+/− T cells, since they patient #M04 higher numbers of CD4+/− non-TRM were sorted were the most frequent subset, representing >96% of the CD4+/− (Supplementary Table 1), yet only the CD4+/− TRM fraction was T cells (p = 0.0002) (Fig. 5e). It is noteworthy that when we positive. Thus, we observed that the TRM fraction contained analyzed the phenotype of these PBMC from HIV+ patients, we significantly more vDNA than the whole population of CD4+/− detected that in blood CD32 was also strongly associated to T cells present in cervical tissue (p = 0.015; Fig. 5b and Supple- the CD69 fraction, in which a median of 38.2% of the cells mentary Fig. 4a). Next, we calculated the global contribution of expressed CD32 (p = 0.004 compared to total CD4+/− T cells, TRM and non-TRM to the total pool of infected CD4+/− T cells in Supplementary Fig. 4b). On the other hand, HLA-DR protein cervix of HIV-infected women. For the fractions that were expression was enriched in the CD69+ and the CD103+ fractions undetectable for vDNA, we used the limit of detection of the assay (p = 0.008 for CD69+, p = 0.004 for CD103+ compared to total based on the individual cellular inputs and the resulting number CD4+/− T cells, Supplementary Fig. 4b). This implies that we was plotted as the maximum possible contribution. We found partially decreased the total frequency of CD32+ CD4+/− T cells that the TRM subpopulation significantly contributed to the pool by gating on CD69+ CD4+/− T cells (which represented a of infected cells by representing a median of 95.65% compared to median of 8.1% of the total CD32), thus underestimating their 4.34% of non-TRM (p = 0.015; Fig. 5c), which was partially due to contribution to the total reservoir in blood. Simultaneously, the their different frequencies (a median of 83.20% and 16.79%, contribution of the CD69 fraction to the total pool of cells is respectively). strongly sustained by its high expression of CD32. Further 8 NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 ARTICLE a p = 0.002 45 p = 0.006 ND % of lymphoid CD45 cells 40 HIV patients 35 + 30 25 20 p = 0.006 15 p = 0.006 10 5 0 + + + + M R 4 3 32 R T + D 10 -D D C 4 D LA C D C C H M + M R T + R 4 T + D 4 C 4 D D C C b CD4 T cells + c + CD4 T cells p < 0.05 p < 0.0001 100 p = 0.030 50 p < 0.05 p = 0.010 p < 0.001 80 p < 0.0001 40 % of HLA-DR % of CD32 60 30 p = 0.001 40 20 20 10 0 0 + + M M M M R R R R 3 3 T T T T 10 10 n- n- no no D D C C M M R R T T Fig. 4 Impact of HIV infection on CD4+ TRM cell subsets from cervix. a Frequency of cervical CD4+, TRM, TRM CD103+, TRM CD32+, and HLA-DR+ CD4+ T cells from the total cervical CD45+ lymphoid cells was determined following the gating strategy described in Fig. 3 in normal donors (ND, blue circles, n = 9) and HIV infected patients (red circles, n = 18, corresponding to M01-M05, M11, M15-M23, and M25-M27). b HLA-DR expression in cervical TRM, CD103+ TRM and non-TRM in normal donors and HIV infected patients. c CD32 expression in cervical TRM, CD103+ TRM and non-TRM in normal donors and HIV infected patients. The median is shown as a solid line, the box indicates the 25–75 percentile range, while the whiskers show the range. Statistical comparisons were performed using two-tailed Mann–Whitney rank test to compare the two groups and using Friedman test with post hoc Dunn’s correction for multiple comparisons for intra-group comparisons. Source data are provided as a Source Data file analyses discriminating these two fractions should determine co-express CD69. We performed this by combining ISH for HIV- their separate real contribution to the viral reservoir in blood. 1 viral RNA with immunohistochemistry of CD69, since CD69 Finally, by including CD4− (CD8−) T cells in these analyses we mRNA expression was hardly detectable (data now shown). enriched the fraction of activated cells, since CD4− T cells from Cervical preparations from one untreated viremic controller and blood expressed significantly more CD69, HLA-DR, CD103 and six ART-suppressed women were examined (Table 1, #P1-P7). In CD32 than CD4+ T cells (p = 0.004 for CD69 and HLA-DR, p = the cervical mucosa of the viremic controller patient we detected 0.012 for CD103, p = 0.019 for CD32, Supplementary Fig. 4c), while vRNA+ cells co-expressing CD69 in both the epithelium and the in cervix only CD32 was associated to the CD4− fraction of T cells, lamina propria (Fig. 6a). As shown in the associated pie charts, yet HLA-DR had a strong opposite trend (p = 0.004 for both, from the total of 15 HIV-vRNA+ cells detected in this sample, Supplementary Fig. 4d). Of note, in terms of general expression, close to 73% were located in the submucosa and the rest in the both CD32 and HLA-DR frequencies were higher in cervicovaginal epithelium (Fig. 6a). Although a lower proportion of CD4+ TRM cells than in blood of HIV+ patients, indicating that a fraction of are found in the intraepithelial compartment compared to CD8+ tissue-resident cells are potentially more activated than circulating TRM, CD4+ TRM are retained in both compartments47. Interest- cells. In any case, altogether, these results provide evidence that ingly, some of the vRNA+ cells in the submucosa were associated CD4+ TRM are infected in vivo and represent an HIV target, at least to CD69 enriched zones (45% of them), which could concur with in cervical tissue. Importantly, regardless of effective ART- previously described semi-organized lymphoid structures con- suppression for more than 10 years, high levels of proviral DNA taining resident memory lymphocyte clusters48,49. Of note, the are still detected in CD4+ TRM from the cervical mucosa of HIV+ strong staining of the HIV-1 RNA probe in some positive cells women, consistently suggesting that this cell type could represent an precluded distinctive determination of CD69 positivity and thus important cellular reservoir elsewhere. we did not count double positive cells in each compartment. Importantly, when we examined cervical preparations of ART- suppressed HIV+ women, we also detected vRNA+ cells expres- HIV−RNA+ CD69+ cells are detected in cervical tissue. We sing CD69 in the epithelium and the submucosa (Fig. 6b). In these used in situ hybridization (ISH) to determine if cervical tissue patients the overall mean frequency of HIV-vRNA+ cells was of harbors vRNA+ cells, as performed before10,45,46, and if these cells 0.114 cells/mm2, which was similar to what was observed in the NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 a p = 0.015 b p = 0.015 105 HIV-1 DNA molec/106 CD4 HIV-1 DNA molec/106 CD4 30,000 + Total CD4 T 20,000 104 M00 M22 + 10,000 CD4 TRM M02 M23 103 M04 M25 7500 M20 M28 5000 102 2500 101 0 Cervix Blood M T R + T + 4 4 D D c lC C Contribution to the total viral reservoir in cervix ta To p = 0.015 TRM 100 Contribution to the total HIV-reservoir in cervix Non-TRM 80 60 M02 M04 M20 40 20 0 M22 M23 M25 M28 TRM Non-TRM d 105 HIV-1 DNA molec/106 cell 104 M02 M20 103 M03 M22 M04 M23 102 M05 M25 1 10 M28 CD69 + – – – CD103 – + – – CD32 +/– +/– – + e Contribution to the total viral reservoir in blood p < 0.05 p < 0.0001 p < 0.05 CD69 100 Contribution to the total HIV-reservoir in blood CD103 – – 80 CD69 CD103 CD32 60 M02 M03 M04 M05 40 20 0 CD69 + – – – CD103 – + – – M20 M22 M23 M25 M28 CD32 +/– +/– – + Fig. 5 Cellular reservoirs in CD4+/− T cells from cervix and blood of ART-suppressed HIV-1 infected women. a Comparison of HIV-1 DNA molecules per million CD4+/− T cells in cervix and PBMCs from the same ART-treated women (n = 8). b Comparison of HIV-1 DNA molecules per million CD4+/− T cells and per million CD4+/− CD69+ (TRM) in cervix from seven ART-suppressed women. c Individual contribution of CD4+/− TRM and non-TRM to the total HIV reservoir in cervicovaginal tissue (left) and summary data from all patients (right). d HIV-1 DNA molecules per million cells in four different subsets based on the expression of CD69, CD103 and CD32 in CD4+/− T cells from blood in HIV-infected women (n = 9). Empty circles represent values under the limit of detection. e Contribution of analyzed subsets to the total HIV reservoir in blood (left) and summary data from all patients (right). Boxes and error bars represent median and range. Statistical comparisons in Fig. 5a–c were performed using Wilcoxon matched-pairs signed rank test and in Fig. 5d and e using Friedman test with post hoc Dunn’s correction for multiple comparisons. Source data are provided as a Source Data file cervical tissue from the viremic controller woman, where we Discussion detected 0.106 cells/mm2. In five ART-suppressed HIV infected Here we assert the existence of an overlooked HIV-1 reservoir in women where the epithelium was preserved, >70% of the vRNA+ cervical tissue from women, since we detected substantial levels of cells were found in the submucosa of the cervix, in general less HIV-1 DNA and RNA in the mucosa of ART-treated patients frequently associated to high density CD69+ staining, as shown in with undetectable plasma viral load for up to 10 years. Moreover, the pie charts for #P2 (Fig. 6b). Individual quantification and we identify CD4+ TRM as a critical cellular reservoir that sustains distribution of vRNA+ cells in different tissue areas is shown in persistence in peripheral tissues. This subset, recently character- Fig. 6c, d for all patients. In any case, these analyses demonstrate ized in multiple tissues, includes highly HIV-susceptible pheno- substantial residual viral transcription in the cervical mucosa of types in the female genital tract, which are simultaneously shaped ART-suppressed women often associated to CD69+ and thus for long-term permanence in tissues. Accordingly, and con- to TRM. sidering the proportion of CD4+ T cells estimated to belong to 10 NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 ARTICLE a Viremic CD69 HIV-1 Distribution of vRNA+ cells #P1 4 (26.7%) 5 6 (45.4%) 11 (54.6%) (73.3%) Epithelium High CD69 Submucosa Low CD69 b ART-treated Distribution of vRNA+ cells CD69 HIV-1 #P2 3 2 (27.3%) (25%) 8 6 (72.7%) (75%) Epithelium High CD69 Submucosa Low CD69 Viremic ART-suppressed c 0.062 0.014 d Distribution of vRNA+ cells 2.5 0.25 14.5% HIV-RNA+ cells/mm2 HIV-RNA+ cells/mm2 2.0 0.20 27.7% 1.5 0.15 72.2% 85.5% 1.0 0.10 0.5 0.05 0.0 0.00 Epithelium High CD69 Submucosa Low CD69 Epithelium Submucosa High CD69 Low CD69 the TRM phenotype in tissues all over the human body25,26, the TRM longevity in diverse sites may reflect the accumulation of contribution of this subset to HIV-1 persisting reservoirs may be antigenic experiences over the human lifespan25, which may be of great importance. In fact, among these tissues with large particularly relevant in peripheral tissues that represent a mucosal proportions of TRM we find lymphoid and mucosal sanctuaries barrier against a broad range of microbes, such as the female such as the gut, estimated to be the main reservoirs during genital tract. However, TRM of the cervicovaginal mucosa of treatment3. Thus, dissecting CD4+ TRM heterogeneity in relation women have not been characterized in depth, regardless of being to HIV pathogenesis and determining the relative weight of a site of exposure to many viral and bacterial infections with very specific CD4+ TRM subsets as cellular reservoirs should be a high global burden. Recent efforts to define transcriptional and priority for the HIV field. functional signatures of CD69+ TRM have revealed consistent NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 Fig. 6 Detection of HIV-RNA+ CD69+ cells in the cervical mucosa of virologically suppressed women. Paraffin-embedded cervical tissue from one viremic and six ART-suppressed HIV-infected women were stained for HIV-1 RNA (rose) using the RNAscope 2.5 HD Duplex Reagent Kit in combination with immunohistochemistry of CD69 protein (brown). a Two representative images of a viremic HIV-infected patient (#P1) with a cell co-expressing HIV-1 RNA and CD69 in the epithelium (left) and in the submucosa (right) are shown inside a white square. Lower panels show high magnification of each of these squares. Scale bar is 100 µm and 20 µm in top and bottom panels, respectively. Left pie chart shows the proportion of HIV-RNA+ cells detected in the epithelium (light blue) or in the submucosa (dark blue) from the total of 15 HIV-RNA+ cells detected in the cervix of #P1. Right pie chart shows the proportion of HIV-RNA+ cells detected in the submucosa associated to CD69 enriched clusters (dark brown) or independent from CD69 clusters (light brown). b Two representative images of an ART-suppressed HIV-infected patient (#P2) with a cell co-expressing HIV-1 RNA and CD69 in the epithelium (left) and in the submucosa (right) are shown inside a white square. Lower panels show high magnification of each of these squares. Scale bar is 100 µm and 20 µm in top and bottom panels, respectively. Pie charts represent the same proportions as in a for a total of 11 HIV-RNA+ cells detected in the cervix (left) or of 8 HIV-RNA+ cells in the submucosa (right) of #P2 patient. c Comparison of positive cells harboring vRNA per square millimeter in the epithelium and the submucosa of n = 7 HIV-infected patients. Triangle symbols correspond to a viremic patient and circle symbols correspond to data from ART-suppressed patients. Statistical comparisons were performed using Wilcoxon matched-pairs signed rank test. d Distribution of HIV-RNA+ cells in different localizations of the cervical tissue in n = 7 HIV-infected patients. Source data are provided as a Source Data file patterns across human tissues, but also increased heterogeneity of that emigrated in the supernatant, confirming its association with the CD4+ TRM fraction compared to CD8+ TRM25,26. Here we mucosal retention. Importantly, after 10 days of infection, most of confirm that expression of CD69 in human cervical mucosa the productively infected cells remained associated to the TRM associates to a protein signature identifying bona fide TRM in phenotype within the tissue. Thus, in principle, CD4+ CCR5+ >95% of these cells25,26,31,32, which suggests residency associated TRM from a wide range of peripheral mucosal and lymphoid to most CD4+ T cells from the human female genital tract, as tissues may represent preferential targets of HIV infection. recently reported in the mouse model30. Importantly, we report Interestingly, we detected a small subset of cervical CD4+ that cervical CD4+ TRM from both ectocervix and endocervix CCR5+ TRM that readily expressed CD32, a marker originally frequently express several surface proteins associated to increased associated to HIV-latently infected cells9, which was later linked HIV susceptibility28,36,50,51. In fact, CD4+ T cells derived from to transcriptionally active infected cells10–13. We observed that the cervix of HIV− women expressing α4β7, α4β1, or the same the fraction of TRM expressing CD32 was enriched for most of the CD69, were already shown to represent preferential targets of molecules investigated, many of them associated to HIV sus- HIV entry28. Our study extends these findings, revealing that the ceptibility. The fact that CD32 expression was strongly associated majority of CCR5+ CD4+ T cells indeed belong to the TRM to cells expressing the highest levels of CD69 may explain why phenotype, which is highly enriched in the expression of α4β7 these cells are enriched for most of these molecules, since many and α4β1 integrins, as suggested50, but also in HIV-1 co-receptors are connected to the TRM phenotype definition per se25. We also such as CXCR4 or CXCR637. Further, CCR6 expression, which observed an association between CD32 and CD69 expression in may indicate a Th17 response, was also more frequent in CD4+ peripheral blood of these ART-suppressed patients, and CD32+ CCR5+ TRM from the endocervix. Importantly, Th17 cells not CD69+ CD4+ T cells have already been shown to correlate with only represent preferential targets after SIV vaginal transmis- proviral DNA in blood10. Supporting their higher susceptibility to sion51, but also are enriched for replication-competent HIV-1 HIV infection in cervix, we detected remarkably high copies of DNA in colon and blood from ART-treated patients52. HIV-1 DNA per cell associated to this fraction. Moreover, the Critically, we confirmed that TRM are indeed preferentially overall level of tissue infection appeared to determine the level of infected in human cervical tissue explants infected with HIVBaL, infection of the CD32− TRM fraction, where less infection mag- where this phenotype harbored more vDNA and was more pro- nified the preference towards CD32 expressing TRM. Further, ductively infected than their counterparts, CD69− non-TRM. As productively infected cells expressed more CD32 than uninfected mentioned, by using high-dimensional CyTOF analysis, Cavrois TRM and these cells were associated to higher expression of HLA- et al. recently demonstrated preferential infection of tonsillar DR, as recently reported10. Therefore, CD32 expression in CD4+ CD69 positive cells, where predicted precursors of the HIV- TRM could identify a subset remarkably susceptible to HIV-1 infected cells had similarly high expression levels as the infected infection, potentially highly activated within an overall more cells41. Although we cannot discard that non-TRM present in the quiescent TRM compartment, which may or may not continue to cervical tissue may upregulate CD69 upon HIV infection, the express this marker throughout the different stages of the retro- preference to support HIV-infection of sorted CD69+ cervical viral infection. Future studies will confirm if CD32 is indeed CD4+ T cells, the lack of CD69 up-regulation after HIV-exposure predicted in the precursors of HIV-1 infected cells and if no in the sorted CD69− fraction, the dynamics of the expression of down-regulation occurs in between HIV-1 infection stages. Our CD69 and HLA-DR during infection overtime, the infection level results, including the increased coexpression of IL-receptors in observed in CD103+ TRM, and the fact that most of the pro- this subset, together with the fact that a recent paper demon- ductive infection is retained within the tissue even after stimu- strates higher proportion of HIV-RNA+ cells co-expressing lation with S1P and CCR7-ligands, strongly suggest an important CD32-RNA in ART-suppressed than in viremic individuals13, role for TRM supporting HIV infection, as implicit on the work highlight the need to better understand the dynamics of CD32 in from others18,27,28. The down-regulation of CD69 detected in the context of tissue reservoir establishment and persistence. cervical tissue culture has already been observed in the tonsil We found that HIV infection has long-term consequences for explant model53 as well as in adipose tissue18, and could be the CD4+ TRM population of the cervical mucosa. In cervix from related to disruptions on cell trafficking and retaining stimulus. ART-suppressed women, the frequencies of CD4+ and CD4+ However, in the tonsil explant model, there was an increase in TRM cells out of the total CD45+ lymphocytes decreased com- activation measured by HLA-DR and CD25 expression over pared to tissues from uninfected women, and these cells expressed time53 that was not observed in our results. Moreover, although higher percentages of HLA-DR. Concomitant cervical viral culturing tissue may affect retention signals, CD69 expression infections such as herpes simplex virus type 2 (HSV-2) or human remained higher within the tissue blocks compared to the cells papilloma virus (HPV) have been associated with systemic T cell 12 NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 ARTICLE activation54,55, which could be conditioning our results. While that the virus takes advantage of this population at the moment of Spain has low prevalence of HSV-2 infection compared to other contraction of the immune response and the transition to a countries and regions56,57, six out of the 19 patients included in resting state in order to establish latency in these cells. Further, these analyses tested positive for HPV and presented cervical while both CD127 and CD57-expressing cells in tonsils are per- intraepithelial neoplasia grade II (CIN II). However, HPV and missive to HIV infection, only CD57+ cells lead to productive high-grade CIN on HIV+ women are associated with high cer- infection41. According to detailed phenotyping of TRM in differ- vical CD4+ cell density in both stroma and epithelium54, and we ent tissues across the human body, maintenance of CD127 and did not detect any correlation between the studied frequencies low expression of CD57 (the latter at least in CD8+ TRM) are and these conditions. On the other hand, it has been observed features of TRM25. Additional lack of recent activation markers that menstrual blood of HIV-infected patients has lower expres- and reduced proliferative turnover shape TRM as quiescent long- sion of the integrin CD103 compared to uninfected women42; lived cells designed for local immunity25. Here we have confirmed however, we did not detect a significant impact of HIV infection high expression of CD127 in CD4+ TRM from the cervical on the cervical CD4+ CD103+ TRM fraction in our patients. mucosa, as well as increased expression of CD122 and CD132, Although CD103 has been established as one of the core surface with low expression of HLA-DR, suggesting that they indeed markers for the TRM lineage, CD4+ TRM express lower levels of could constitute a long-term stable reservoir. Further, we also this integrin compared to the CD8+ fraction25, which may be concur the lack of expression of CD127 among productively related to higher intraepithelial retention of CD8+ vs. CD4+ infected cells, thus it remains to be determined if this receptor can T cells47. Still, detection of HIV-1 RNA in cervical tissue from identify latently infected subpopulations of TRM in HIV patients. women demonstrated intraepithelial and submucosal positive Consistent with this idea, it has been shown that stimulation with cells, highlighting the presence of these cells in close proximity to IL-7 induces reverse transcription and virus integration63, while the mucosal surface, which has obvious implications for HIV addition of this cytokine to the cervical explant model enhances acquisition. In this sense, Hladik et al. demonstrated that HIV replication by promoting not only proliferation but also intraepithelial vaginal CD4+ T cells, which highly expressed the survival of infected cells64. The fact that cervical TRM express high major HIV-1 coreceptors (CD4 and CCR5), were remarkably levels of these receptors reinforces their potential contribution to permissive to HIV-1 infection58. It remains to be elucidated how HIV persistence in these tissues. long these cells remain within these mucosal tissues and if it they Importantly, regardless of long ART-mediated viral suppres- can establish a latent cellular reservoir. sion, we detected very high levels of proviral DNA in the TRM Generation of CD4+ TRM has been associated to protection fraction of cervical tissue from HIV+ treated patients. The values against herpesvirus in the female genital tract of mice48, and it is of HIV-1 DNA copies detected here concur with recent data also believed to limit recurrence in humans47. In mice, protective reporting higher provirus copy numbers found in the lamina CD4+ TRM were found in macrophage-organized memory lym- propria of vaginal tissue of a single ART-treated woman com- phocyte clusters that developed after vaginal but not systemic pared to PBMC21. In addition, our results on proviral DNA are attenuated infection48. Protection against other sexually trans- remarkably similar to the ones determined in ileum biopsies16 mitted infections such as Chlamydia is also associated to presence and in bronchoalveolar lavage17 from ART-suppressed patients, of effective CD4+ TRM in the genital mucosa47,49. Female in which effector memory phenotypes were reported to sustain reproductive tract conditions, such as bacterial vaginosis, induce these reservoirs. Since most tissue TRM express phenotypic mar- an increase in the expression of α4β1 and CCR5 of circulating kers that would classify them as circulating memory effector cells, effector CD4+ lymphocytes59, which may correspond to recruited based on the expression of CD45RA and CCR726, it seems cells that contribute to the establishment of a memory pool. plausible that the aforementioned reservoirs are largely composed Remarkably, it is now clear that pre-existing TRM populations are of TRM phenotypes. Still, in blood samples from ART-treated not displaced after subsequent infections, which enables multiple patients, most of the HIV-1 DNA detected contains fatal muta- TRM specificities to be stably maintained within the tissue, as tions and only 2–10% of proviruses are estimated to be intact, recently shown in the skin29. Thus, in principle, infections over indicating that only a minor fraction is more likely to produce the lifetime of an individual will contribute to increase the TRM replication competent virus,65. Moreover, we showed that after pool. Further, novel vaccine designs combining a general prime long-term ART, TRM contained transcriptionally active HIV. strategy with recruitment of activated T cells to the genital tract Although we performed a quantitative viral growth assay in (by means of topical chemokine application) have been proposed sorted TRM obtained from cervix of one hysterectomy, the small to establish local memory and prevent reinfection48. However, in number of cells recovered limited this assay. Future studies the context of HIV infection, establishment of these clusters of involving techniques with lower cell number requirement may CD4+ TRM may also provide increased number of HIV-1 targets better clarify the portion of intact virus within TRM from the for the nascent infection, which can ultimately contribute to a cervix of ART-treated women66. However, the fact that a recent larger reservoir after treatment. In this sense, we detected CD69+ study reports that proviral copy numbers correlate with levels of enriched areas compatible with TRM clusters in the cervical cell-associated HIV RNA and predict control after treatment submucosa of ART-suppressed patients that contained tran- interruption55 denotes the importance of the remaining levels of scriptionally active HIV cells. Therefore, past and concomitant vDNA and vRNA encountered in this study. These findings infections of the genital tract may contribute to lowering the recognize this subpopulation of cells, not only as a long-term threshold for HIV acquisition60 by increasing preferential TRM cellular reservoir for HIV, but also as the possible source of targets and, in addition, augment long-lived cellular reservoirs. residual viremia observed in HIV+ patients despite ART. In this The effect of cytokines on the maintenance and homeostatic sense, a high degree of variability in penetration of the different proliferation of certain populations of memory lymphocytes may antiretroviral drugs, both between and within drug classes, has be determinant in the establishment and durability of reservoirs, been reported in the female genital tract67, potentially con- as occurs for IL-7 driven homeostatic proliferation of latently tributing to this phenomenon, as observed in other tissues3,4. infected cells5 or for IL-15 with the TSCM subset61. It seems that In summary, in this study we have identified, CD4+ TRM as a the maintenance of effector memory cells after the contraction critical cellular reservoir of the cervical tissue, and most likely of phase of the immune response corresponds to 5–10% of the other tissues not yet identified. These cells are preferentially effectors that express high levels of CD12762. Thus, it is possible infected during ex vivo HIV infection and, after years of NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications 13 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 suppressive treatment, still harbor HIV-1 nucleic acids in the Explant infection and subset purification. Eight to twelve blocks of tissue per cervical mucosa of infected women. With current clinical guide- condition were placed in RPMI1640 media supplemented with 20% FBS (R20) in a 24-well plate and infected with 7200 TCID50 of the viral stock HIV-1BaL or medium in lines recommending early ART initiation, the relevance of control wells. For each experimental condition at least duplicates were performed. mucosal tissues where the virus first replicates as sources of viral After 2 h of incubation at 37 °C, tissue blocks were washed three times with 3 mL of persistence may increase in years to come. Overall, the con- PBS in 6-well plates and placed back into a 12-well plate at 8 blocks/well in 1 mL of tribution of CD4+ TRM to viral persistence in lymphoid and R20. Infected cervical tissue blocks and controls were cultured for additional mucosal tissues as well as their capacity to establish latent or 10–12 days, with a change of medium every 3 days. Tissue digestion with collagenase IV (Invitrogen) and manual dissociation were immediately executed, as described active viral reservoir in these tissues requires major attention in above68. Supernatants and cell suspensions were incubated with Aqua vital dye and order to reach a functional cure in HIV infected patients. stained with CD69-Horizon-PE-CF594 (FN50, dilution 1/20, cat. no. 562645), CD4 or CD127-BV605 (RPA-T4, dilution 1/12.5, cat. no. 562658 and A019D5, dilution 1/100, cat. no. 351333 respectively), CD19-V500 (HIB19, dilution 1/50, cat. no. 561121), Methods CD3-PerCP (SK7, dilution 1/10, cat. no. 345766), CD14-APC-H7 (MφP9, dilution 1/ Study subjects and patient samples. Cervical tissue was obtained from non- 66.7, cat. no. 560180), CD8-APC (RPA-T8, dilution 1/40, cat. no. 555369), HLA-DR- neoplastic hysterectomies ranging 26–74 years performed in participating hospi- BV421 (G46-6, dilution 1/200, cat. no. 562804) (all from BD Biosciences), CD49a- tals. Additionally, blood and cervical tissue from ART-treated HIV+ women were APC-Vio770 (REA1106, dilution 1/50, cat. no. 130-119-410, Miltenyi Biotec), CD45- obtained from women undergoing a hysterectomy or a cone biopsy in the same Alexa Fluor 700 (Hl30, dilution 1/100, cat. no. 304024), CD32-PE-Cy7 (FUN-2, hospitals. Study protocols were approved by the corresponding Ethical Committees dilution 1/50, cat. no. 303214) and CD103-FiTC (BER-ACT8, dilution 1/50, cat. no. (Institutional Review Board numbers PR (IR)294/2017 for the HUVH, PI-17–159 350203) (BioLegend). After surface staining cells were intracellularly stained with for the HUGTP and 2018/8017/I for the Parc de Salut Mar), and written informed KC57-PE antibody (FH190-1-1, dilution 1/40, cat. no. 6604667, Beckman Coulter, consent was provided by all patients recruited to this study. Patient information on Brea, CA) using Fixation & Cell Permeabilization Kit (Invitrogen) according to plasma viral loads, CD4+ T cell counts, time of ART-suppression and treatment is manufacturer’s instructions to determine p24 expression10. 1% PFA fixed samples summarized in Table 1. were acquired in a BD FACSAriaTM II Cell Sorter for purification of Aqua−CD19− CD45+ CD3+ CD8−CD4± T cells: 1) TRM as CD69+; 2) non-TRM as CD69− and 3) CD32+ TRM as CD69+CD32+. These samples were immediately lysed and stored for Cervical tissue digestion and flow cytometry phenotyping. A piece of ectocervix posterior vDNA analyses by qPCR as follows. Further, samples were acquired for and endocervix from healthy donors was delivered to the laboratory once healthy phenotype analysis using FlowJo vX.0.7 software. tissue status was confirmed by the Pathology Service. Tissue was delivered in antibiotics-containing RPMI 1640 medium and processed within the 24 h following Cervical TRM purification and infection. Cervical tissue from healthy donors was surgery68. The mucosal epithelium and the underlying stroma were separated from obtained and digested, and the resulting cell suspension was stained for viability with the muscular tissue and dissected into ~8-mm3 blocks. Tissue blocks were then Aqua vital dye (Invitrogen) and the following antibodies: CD69-FiTC (FN50 dilution enzymatically digested by incubating them with 5 mg/ml collagenase IV in RPMI 1/20, cat. no. 557049), CD45-PE (HI30, dilution 1/100, cat. no. 560975), CD3-PerCP 1640 5% FBS for 30 min at 37 °C and 400 rpm, followed by a mechanical dis- (SK7, dilution 1/10, cat. no. 345766), CD19-V500 (HIB19, dilution 1/50, cat. no. sociation of the tissue with a disposable pellet pestle68. Once digested, the cellular 561121), PD1-BV421 (EH12.1, dilution 1/20, cat. no. 562516) (all from BD Bios- suspension was washed twice with PBS and stained with Live/Dead Aqua (Invi- ciences), CD4-APC (OKT4 BioLegend, dilution 1/80, cat. no. 300514). Aqua− trogen) to identify dead cells. After washing, cells were equally divided into 5 CD19−CD45+ CD3+ CD4+ T cells were acquired in a BD FACSAriaTM II Cell Sorter different tubes and stained with the following antibodies: HLA-DR-PerCP-Cy5.5 and separated according to their CD69 expression into CD69+ (TRM) and CD69− (G46-6 dilution 1/100, cat. no. 560652), CD14-V450 (MØP9, dilution 1/40, cat. no. (non-TRM). Sorted cells were then infected with 3655 TCID50 of the viral stock HIV- 560349), CD69−Horizon-PE-CF594 (FN50, dilution 1/20, cat. no. 562645), CCR5- 1BaL for 4 h at 37 °C or with medium in control conditions, washed with PBS and APC-Cy7 (2D7, dilution 1/50, cat. no. 557755), CD163-BV786 (GHI/61, dilution 1/ cultured in a 96-wells plate. After three days, cells were stained for viability with Live/ 20, cat. no. 741003), CD4-BV605 (RPA-T4, dilution 1/12.5, cat. no. 562658), Dead Fixable Far Red (Invitrogen), CD69-FiTC (FN50 dilution 1/20, cat. no. 557049) CD19-V500 (HIB19, dilution 1/50, cat. no. 561121), CCR7-PE (3D12, dilution 1/ and HLA-DR-PerCP-Cy5.5 (G46-6, dilution 1/100, cat. no. 560652), before staining 10, cat. no. 552176), CD49d-FiTC (9F10, dilution 1/25, cat. no. 560840), β7-APC for p24 as described above. Fixed cells were finally acquired in a BD FACS Calibur (FIB504, dilution 1/50, cat. no. 551082), CD184-APC (12G5, dilution 1/40, cat. no. flow cytometer and analyzed with FlowJo vX.0.7 software (TreeStar). 555976) (all from BD Biosciences), CD3-SuperBright 645 (OKT3, dilution 1/20, cat. no. 64-0037-42), S1PR1 (CD363)-eFluor660 (SW4GYPP, dilution 1/72, cat. no. 50-3639-42) (eBiosciences), CD45-Alexa Fluor 700 (Hl30, dilution 1/100, cat. Cervical TRM migration assay. Nine blocks of tissue per experimental condition no. 304024), CD32-PE-Cy7 (FUN-2, dilution 1/50, cat. no. 303214), CD103-FiTC were infected and cultured for 10 days as described above. On day 10, tissue was (BER-ACT8, dilution 1/50, cat. no. 350203), CD49a-PE (TS2/7, dilution 1/200, cat. transferred to a transwell insert containing CCL19 (100 ng/ml, Biolegend), CCL21 no. 328303), CXCR3-FiTC (G025H7, dilution 1/33.3, cat. no. 353703), CD127- (100 ng/ml, R&D) and sphingosine-1-phosphate (S1P, 10 µM, Sigma) in the lower FiTC (A019D5, dilution 1/100, cat. no. 351312), CD132-PE (VI C-89, dilution 1/ chamber medium (R20). Triplicates were established for each experimental condition, 100, cat. no. 338606), CD122-APC (TU27, dilution 1/100, cat. no. 339008), CCR6- as well as the corresponding controls. Tissue was left overnight at 37 °C and then APC (G034E3, dilution 1/100, cat. no. 353416) (all from BioLegend), CXCR6-APC digested as described above. Migrating and tissue resident cells were stained separately (56811, dilution 1/25, cat. no. FAB699A), CCR2-PE (48607, dilution 1/33.3, cat. no. for viability, surface markers listed above and p24, fixed, acquired in a BD FAC- FAB151P) (R&D Systems), TCR-γδ-FiTC (11F2, dilution 1/40, cat. no. 130-096- SAriaTM II Cell Sorter and analyzed with FlowJo vX.0.7 software (TreeStar). 884) and CD161-PE (191B8, dilution 1/50, cat. no. 130-092-677) (Miltenyi Biotec). Staining of T-bet (T-bet-BV421, 4B10, dilution 1/40, cat. no. 644815, Biolegend), Sample subset purification from ART-treated HIV+ patients. Samples from Eomes (Eomes-PE-Cy7, WD1928, dilution 1/33.3, cat. no. 25-4877-42, eBioscience) blood and cervical tissue from HIV-infected women under suppressive ART were and Hobit (Hobit-AF647, Sanquin-Hobit/1, dilution 1/20, cat. no. 566250, BD processed to obtain PBMC by Ficoll–Paque density gradient centrifugation or a cell Biosciences) was performed using FoxP3 transcription factor staining buffer set suspension from digested tissue, as described above. Cells were surface stained with (eBioscience), according to manufacturer’s protocols. Preliminary controls on the the same panel described for the ex vivo infection, without p24 intracellular staining. effect of collagenase IV treatment on the stability of the cell surface markers Fixed cells were isolated by cell sorting using a BD FACSAriaTM II Cell Sorter. In both analyzed were performed in PBMC68. Fluorescence Minus One (FMO) controls tissues sorted cells were Aqua−CD19−CD45+ CD3+ CD4±CD8−T cells and, were performed to delineate gates of continuous markers in at least one tissue depending on the tissue, different subsets were purified: for cervix (1) TRM were sample. Samples were acquired in a BD LSRFortessa flow cytometer and analyzed CD69+ and non-TRM were CD69−; for blood (1) CD69+ (CD103− CD32+/−), (2) with FlowJo vX.0.7 software (TreeStar). CD103+ (CD69− CD32+/−), (3) CD32+(CD69− CD103−) and (4) CD4+/− T cells did not express any of these three markers (CD69− CD103− CD32−). These samples were immediately lysed and stored for posterior vDNA analyses by qPCR as follows. Flow cytometry imaging of CD32 expressing cervical TRM. Specific analysis of Further, samples were acquired for phenotype analysis using FlowJo vX.0.7 software. CD32 expression on cervical CD4+ cells was performed by using Amnis® tech- nology. To do so, digested tissue cells were stained with CD32-PE (FUN-2 dilution 1/20, cat. no. 303205), CD45-Alexa Fluor 700 (Hl30, dilution 1/100, cat. no. Quantification of cell-associated HIV-1 DNA by qPCR. All sorted subsets 304024, both from BioLegend), CD69-FITC (FN50, dilution 1/20, cat. no. 557049), obtained from the ex vivo cervical model or from blood and cervix of HIV+ CD4-BV605 (RPA-T4, dilution 1/12.5, cat. no. 562658 both from BD Biosciences) patients were immediately lysed with a proteinase K-containing lysis buffer (55 °C and a viability dye (LIVE/DEAD Fixable Violet Dead Cell Stain Kit, dilution 1/250, over-night and afterwards 5 min at 95 °C) and cell lysates were subjected to total cat. no. L34966, Invitrogen), and then acquired with an AMNIS ImageStremX HIV-1 DNA quantification by qPCR with primers against HIV long terminal imaging flow cytometer (Merck). Data were analyzed using IDEAS v6.1 software. repeat (LTR forward 5′-TTAAGCCTCAATAAAGCTTGCC-3′ and LTR reverse Gradient RMS value > 40 was established as a threshold for best focus, and only 5′-GTTCGGGCGCCACTGCTAG-3′; LTR probe 5′-CCAGAGTCACACACCA focused cells were considered for analysis. A CD32-FMO control was performed to GACGGGCA-3′). For quantification of total HIV-1 DNA, a standard curve was delineate this gate in CD69+ and CD69− cervical CD4+ cells. prepared and CCR5 gene was used for cell input normalization (CCR5 forward 5′- 14 NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 ARTICLE GCTGTGTTTGCGTCTCTCCCAGGA-3′ and CCR5 reverse 5′-CTCACAGCCC 7. Buzon, M. J. et al. HIV-1 persistence in CD4+ T cells with stem cell-like TGTGCCTCTTCTTC-3′; CCR5 probe 5′-AGCAGCGGCAGGACCAGCCCCA properties. Nat. Med. 20, 139–142 (2014). AG-3′). The cycling parameters were 50 °C for 2 min, 95 °C for 10 min and then 8. Iglesias-Ussel, M., Vandergeeten, C., Marchionni, L., Chomont, N. & Romerio, 95 °C for 15 s and 60 °C for 1 min for 55 cycles of amplification. Contribution of F. High levels of CD2 expression identify HIV-1 latently infected resting each subset to the HIV reservoir was calculated by considering the frequency of memory CD4+ T cells in virally suppressed subjects. J. Virol. 87, 9148–9158 each subset (analyzed by flow cytometry) within the total CD4 compartment and (2013). their relative infection frequency. 9. Descours, B. et al. CD32a is a marker of a CD4 T-cell HIV reservoir harbouring replication-competent proviruses. Nature 543, 564–567 (2017). RNA in situ hybridization and immunohistochemistry. Paraffin-embedded cer- 10. Abdel-Mohsen M., et al. CD32 is expressed on cells with transcriptionally vical tissue samples from one viremic and six aviremic ART-treated HIV-infected active HIV but does not enrich for HIV DNA in resting T cells. Sci. Transl. patients were obtained from the Pathology Department of the HUVH. For detection Med. 10, 6759 (2018). of RNA, we used ultrasensitive RNA detection assay RNAscope 2.5 HD Duplex 11. Badia, R. et al. CD32 expression is associated to T-cell activation and is not a Reagent Kit (Anacrome). A high sensitivity target-specific probes to GagPol HIV marker of the HIV-1 reservoir. Nat. Commun. 9, 2739 (2018). mRNA sequence were used (bases 507–4601; reference #31769146). Cervical tissue 12. Noto A., et al. CD32(+) and PD-1(+) Lymph Node CD4 T cells support sections of 5 µm were mounted on Superfrost Plus microscope slides (Fisher Scien- persistent HIV-1 transcription in treated aviremic individuals. J. Virol. 92, tific). The assay was performed according to manufacture’s instructions. Briefly, e00901-18 (2018). samples were deparaffinized with xylene and dehydrated in 100% ethanol. Heat- 13. Vasquez, J. J. et al. CD32-RNA Co-localizes with HIV-RNA in CD3+ Cells induced epitope retrieval and protease digestion were used as sample pre-treatment. Found within Gut Tissues from Viremic and ART-Suppressed Individuals. Next, probes were incubated for 2 h at 40 °C and samples were stored overnight in 5× Pathog. Immun. 4, 147–160 (2019). saline sodium citrate buffer. Next day, amplification and signal development of RNA 14. Grau-Exposito J., et al. A novel single-cell FISH-flow assay identifies effector probes were performed by sequential incubation of indicated reagents. Afterwards, an memory CD4(+) T cells as a major niche for HIV-1 transcription in HIV- immunohistochemisty protocol was carried out. Samples were first blocked by infected patients. mBio 8, e00876-17 (2017). incubation with 1× Tris-buffered saline buffer containing 1% bovine serum albumin 15. Hogan, L. E. et al. Increased HIV-1 transcriptional activity and infectious and 10% normal donkey serum for 30 min. Then, anti-CD69 antibody (1/100, burden in peripheral blood and gut-associated CD4+ T cells expressing CD30. Abcam) was incubated overnight at 4 °C. Antibody detection was performed with PLoS Pathog. 14, e1006856 (2018). DAB detection kit (Abcam), according to manufacture’s instructions. Images were 16. Yukl, S. A. et al. The distribution of HIV DNA and RNA in cell subsets differs captured using a DP71 digital camera (Olympus, Center Valley, PA, USA) attached to in gut and blood of HIV-positive patients on ART: implications for viral a BX41 microscope (Olympus). Specificity of HIV-RNA positive cells was determined persistence. J. Infect. Dis. 208, 1212–1220 (2013). by quantifying the relative false detection rate in cervical tissues from two HIV 17. Costiniuk, C. T. et al. HIV persistence in mucosal CD4+ T cells within the negative patients, which corresponded to 0.03 positive cells/mm2, with no positive cell lungs of adults receiving long-term suppressive antiretroviral therapy. Aids 32, detected in their epitheliums or in CD69 enriched areas (Supplementary Fig. 5). 2279–2289 (2018). Additional negative controls consisted of a matched sample without target-specific 18. Couturier, J. et al. Human adipose tissue as a reservoir for memory CD4+ probes and a control without primary antibody (CD69) for immunohistochemistry T cells and HIV. Aids 29, 667–674 (2015). (Supplementary Fig. 5). A cell was considered HIV RNA positive based on the 19. Whitney, J. B. et al. Rapid seeding of the viral reservoir prior to SIV viraemia diameter of the staining, while small punctate signals were not considered in the in rhesus monkeys. Nature 512, 74–77 (2014). quantification46. Positive cells within the whole tissue section were counted by visual inspection and the total tissue area was measured by ImageJ software. 20. Gibbs, A. et al. Presence of CD8+ T cells in the ectocervical mucosa correlates with genital viral shedding in HIV-infected women despite a low prevalence of HIV RNA-expressing cells in the tissue. J. Immunol. 192, 3947–3957 (2014). Statistical analyses. All the data were analyzed using the software GraphPad 21. Pena-Cruz V., et al. HIV-1 replicates and persists in vaginal epithelial Prism 7.0 (GraphPad Software, La Jolla, CA, USA). All values in the graphs are dendritic cells. J. Clin. Investig. 92, e00901-18 (2018). expressed as the median and the range or the interquartile range. Non-parametric 22. Hirbod, T., Kaldensjo, T. & Broliden, K. In situ distribution of HIV-binding two-tailed Mann–Whitney rank tests, Wilcoxon matched-pairs signed rank test CCR5 and C-type lectin receptors in the human endocervical mucosa. PLoS and Friedman test with Dunn’s correction for multiple comparisons were used. A ONE 6, e25551 (2011). P value < 0.05 was considered significant. 23. Trifonova, R. T., Lieberman, J. & van Baarle, D. Distribution of immune cells in the human cervix and implications for HIV transmission. Am. J. Reprod. Reporting summary. Further information on research design is available in Immunol. 71, 252–264 (2014). the Nature Research Reporting Summary linked to this article. 24. Saba, E. et al. HIV-1 sexual transmission: early events of HIV-1 infection of human cervico-vaginal tissue in an optimized ex vivo model. Mucosal Immunol. 3, 280–290 (2010). Data availability 25. Kumar, B. V. et al. Human tissue-resident memory T cells are defined by core The source data underlying both the Main and Supplementary Figures are provided as a transcriptional and functional signatures in lymphoid and mucosal sites. Cell Source Data file. All other data are available from the corresponding authors upon Rep. 20, 2921–2934 (2017). reasonable requests. 26. Wong, M. T. et al. A high-dimensional atlas of human T cell diversity reveals tissue-specific trafficking and cytokine signatures. Immunity 45, 442–456 (2016). Received: 5 February 2019; Accepted: 30 September 2019; 27. Wang, X. et al. Distinct expression patterns of CD69 in mucosal and systemic lymphoid tissues in primary SIV infection of rhesus macaques. PLoS ONE 6, e27207 (2011). 28. Joag, V. R. et al. Identification of preferential CD4+ T-cell targets for HIV infection in the cervix. Mucosal Immunol. 9, 1–12 (2016). 29. Park, S. L. et al. Local proliferation maintains a stable pool of tissue-resident memory T cells after antiviral recall responses. Nat. Immunol. 19, 183–191 References (2018). 1. Chun, T. W. et al. Early establishment of a pool of latently infected, resting 30. Beura, L. K. et al. CD4(+) resident memory T cells dominate CD4(+) T cells during primary HIV-1 infection. Proc. Natl Acad. Sci. USA 95, immunosurveillance and orchestrate local recall responses. J. Exp. Med. 216, 8869–8873 (1998). 1214–1229 (2019). 2. Murray, A. J., Kwon, K. J., Farber, D. L. & Siliciano, R. F. The Latent Reservoir 31. Beura, L. K. et al. T cells in nonlymphoid tissues give rise to lymph-node- for HIV-1: how immunologic memory and clonal expansion contribute to resident memory T cells. Immunity 48, 327–338 e325 (2018). HIV-1 persistence. J. Immunol. 197, 407–417 (2016). 32. Topham, D. J. & Reilly, E. C. Tissue-resident memory CD8(+) T cells: from 3. Estes, J. D. et al. Defining total-body AIDS-virus burden with implications for phenotype to function. Front. Immunol. 9, 515 (2018). curative strategies. Nat. Med. 23, 1271–1276 (2017). 33. Bertagnolli, L. N. et al. The role of CD32 during HIV-1 infection. Nature 561, 4. Fletcher, C. V. et al. Persistent HIV-1 replication is associated with lower E17–E19 (2018). antiretroviral drug concentrations in lymphatic tissues. Proc. Natl Acad. Sci. 34. Osuna, C. E. et al. Publisher Correction: Evidence that CD32a does not mark USA 111, 2307–2312 (2014). the HIV-1 latent reservoir. Nature 563, E33 (2018). 5. Chomont, N. et al. HIV reservoir size and persistence are driven by T cell 35. Perez, L. et al. Conflicting evidence for HIV enrichment in CD32(+) CD4 survival and homeostatic proliferation. Nat. Med. 15, 893–900 (2009). T cells. Nature 561, E9–E16 (2018). 6. Pallikkuth, S. et al. Peripheral T follicular helper cells are the major HIV 36. Kader, M. et al. Alpha4(+)beta7(hi)CD4(+) memory T cells harbor most Th- reservoir within central memory CD4 T cells in peripheral blood from 17 cells and are preferentially infected during acute SIV infection. Mucosal chronically HIV-infected individuals on combination antiretroviral therapy. J. Immunol. 2, 439–449 (2009). Virol. 90, 2718–2728 (2015). NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications 15 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12732-2 37. Gorry, P. R. & Ancuta, P. Coreceptors and HIV-1 pathogenesis. Curr. HIV/ 64. Introini, A., Vanpouille, C., Lisco, A., Grivel, J. C. & Margolis, L. Interleukin-7 AIDS Rep. 8, 45–53 (2011). facilitates HIV-1 transmission to cervico-vaginal tissue ex vivo. PLoS Pathog. 38. Richardson-Harman, N. et al. Analytical advances in the ex vivo challenge 9, e1003148 (2013). efficacy assay. AIDS Res. Hum. Retroviruses 33, 395–403 (2017). 65. Ho, Y. C. et al. Replication-competent noninduced proviruses in the latent 39. Michel, N., Allespach, I., Venzke, S., Fackler, O. T. & Keppler, O. T. The Nef reservoir increase barrier to HIV-1 cure. Cell 155, 540–551 (2013). protein of human immunodeficiency virus establishes superinfection 66. Bruner, K. M. et al. A quantitative approach for measuring the reservoir of immunity by a dual strategy to downregulate cell-surface CCR5 and CD4. latent HIV-1 proviruses. Nature 566, 120–125 (2019). Curr. Biol. 15, 714–723 (2005). 67. Thompson, C. G., Cohen, M. S. & Kashuba, A. D. Antiretroviral 40. Griffith, J. W., Sokol, C. L. & Luster, A. D. Chemokines and chemokine pharmacology in mucosal tissues. J. Acquired Immune Defic. Syndromes 63 receptors: positioning cells for host defense and immunity. Annu. Rev. (Suppl 2), S240–S247 (2013). Immunol. 32, 659–702 (2014). 68. Cantero J., Genesca M. Maximizing the immunological output of the 41. Cavrois, M. et al. Mass cytometric analysis of HIV entry, replication, and cervicovaginal explant model. J. Immunol. Methods 460, 26–35 (2018). remodeling in tissue CD4+ T cells. Cell Rep. 20, 984–998 (2017). 42. Moylan, D. C. et al. Diminished CD103 (alphaEbeta7) expression on resident T cells from the female genital tract of HIV-positive women. Pathog. Immun. Acknowledgements 1, 371–387 (2016). We would like to thank all the patients who participated in the study and their providers. 43. Bromley, S. K., Yan, S., Tomura, M., Kanagawa, O. & Luster, A. D. We thank José L. Poza, Mª Elena Suárez, Mª Assumpció Pérez-Benavente and Cristina Recirculating memory T cells are a unique subset of CD4+ T cells with a Carrato for referral of patients and sample collection, Irian Lorencés and Laia Pérez-Roca distinct phenotype and migratory pattern. J. Immunol. 190, 970–976 (2013). from the Tumor Bank of the IGTP-HUGTiP for sample management, Alba Ruiz and 44. Kiravu, A. et al. Evaluation of CD103 (alphaEbeta7) integrin expression by Ruth Peña for generating the viral stock R5-Bal and Gerard Requena from the Flow CD8 T cells in blood as a surrogate marker to predict cervical T cell responses Cytometry Facility at the IGTP for excellent technical assistance, as well as Isabel Crespo in the female genital tract during HIV infection. Clin. Immunol. 141, 143–151 from the Flow Cytometry Platform at the IDIBAPS for her excellent technical assistance (2011). on Amnis technology. This work was primarily supported by grants from the Spanish 45. Deleage, C. et al. Defining HIV and SIV reservoirs in lymphoid tissues. “Ministerio de Economía y Competitividad, Instituto de Salud Carlos III” (ISCIII, PI14/ Pathog. Immun. 1, 68–106 (2016). 01235 and PI17/01470) and a fellowship award from the Dexeus foundation for women’s 46. Vasquez, J. J. et al. Elucidating the burden of HIV in tissues using multiplexed health research to M.G., grants R21AI118411 and SAF2015-67334-R (from the Spanish immunofluorescence and in situ hybridization: methods for the single-cell Secretariat of Science and Innovation and FEDER funds) to M.J.B and grants from the phenotypic characterization of cells harboring HIV in situ. J. Histochem. ISCIII (PI14/01058 and PI17/00164) to J.G.P. M.G., M.J.B., and. J.G.P. are supported by Cytochem. 66, 427–446 (2018). the Spanish AIDS network Red Temática Cooperativa de Investigación en SIDA (RD16/ 47. Turner, D. L. & Farber, D. L. Mucosal resident memory CD4 T cells in 0025/0007) M.G. is currently supported by the “Pla estratègic de recerca i innovació en protection and immunopathology. Front. Immunol. 5, 331 (2014). salut” (PERIS, SLT002/16/00353), from the Catalan government, while the Miguel Servet 48. Iijima, N. & Iwasaki, A. T cell memory. A local macrophage chemokine program from the ISCIII supports M.J.B. (CP17/00179) and J.G.P. (CP15/00014). network sustains protective tissue-resident memory CD4 T cells. Science 346, 93–98 (2014). Author contributions 49. Johnson, R. M. & Brunham, R. C. Tissue-resident T cells as the central J.C.-P., J.G.-E., C.S.-P., D.A.R., L.L.-B., A.A.-G., M.A.F., J.G.P., M.J.B., and M.G. designed paradigm of chlamydia immunity. Infect. Immun. 84, 868–873 (2016). and carried out experiments. J.C.-P., J.G.-E., C.S.-P., D.A.R., M.J.B., and M.G. analyzed 50. Perciani, C. T. et al. alphaEbeta7, alpha4beta7 and alpha4beta1 integrin and interpreted data. J.C., T.S., G.T., B.LL., J.M.S.-S., A.T., C.L., L.M.-B., C.C.-M. and contributions to T cell distribution in blood, cervix and rectal tissues: potential V.F. selected study subjects and provided samples. J.C.-P., M.J.B. and M.G. drafted the implications for HIV transmission. PLoS ONE 13, e0192482 (2018). manuscript and all authors edited the final version of the manuscript. 51. Stieh, D. J. et al. Th17 cells are preferentially infected very early after vaginal transmission of SIV in Macaques. Cell Host Microbe 19, 529–540 (2016). 52. Gosselin, A. et al. HIV persists in CCR6+CD4+ T cells from colon and blood Competing interests during antiretroviral therapy. Aids 31, 35–48 (2017). The authors declare no competing interests. 53. Biancotto, A. et al. HIV-1 induced activation of CD4+ T cells creates new targets for HIV-1 infection in human lymphoid tissue ex vivo. Blood 111, 699–704 (2008). Additional information Supplementary information is available for this paper at https://doi.org/10.1038/s41467- 54. Papasavvas, E. et al. High-risk oncogenic HPV genotype infection associates 019-12732-2. with increased immune activation and T cell exhaustion in ART-suppressed HIV-1-infected women. Oncoimmunology 5, e1128612 (2016). Correspondence and requests for materials should be addressed to M.J.B. or M.Gà. 55. Sheth, P. M. et al. Coinfection with herpes simplex virus type 2 is associated with reduced HIV-specific T cell responses and systemic immune activation. J. Peer review information Nature Communications thanks the anonymous reviewers for Infect. Dis. 197, 1394–1401 (2008). their contribution to the peer review of this work. Peer reviewer reports are available. 56. Garcia-Corbeira, P., Dal-Re, R., Aguilar, L., Granizo, J. J. & Garcia-de-Lomas, J. Is sexual transmission an important pattern for herpes simplex type 2 virus Reprints and permission information is available at http://www.nature.com/reprints seroconversion in the Spanish general population? J. Med Virol. 59, 194–197 (1999). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in 57. Looker, K. J. et al. Global estimates of prevalent and incident herpes simplex published maps and institutional affiliations. virus type 2 infections in 2012. PLoS ONE 10, e114989 (2015). 58. Hladik, F. et al. Initial events in establishing vaginal entry and infection by human immunodeficiency virus type-1. Immunity 26, 257–270 (2007). 59. Qualai, J. et al. Adhesion molecules associated with female genital tract Open Access This article is licensed under a Creative Commons infection. PLoS ONE 11, e0156605 (2016). Attribution 4.0 International License, which permits use, sharing, 60. Fleming, D. T. & Wasserheit, J. N. From epidemiological synergy to public adaptation, distribution and reproduction in any medium or format, as long as you give health policy and practice: the contribution of other sexually transmitted appropriate credit to the original author(s) and the source, provide a link to the Creative diseases to sexual transmission of HIV infection. Sexually Transmitted Infect. Commons license, and indicate if changes were made. The images or other third party 75, 3–17 (1999). material in this article are included in the article’s Creative Commons license, unless 61. Gattinoni, L. et al. A human memory T cell subset with stem cell-like indicated otherwise in a credit line to the material. If material is not included in the properties. Nat. Med. 17, 1290–1297 (2011). article’s Creative Commons license and your intended use is not permitted by statutory 62. Kaech, S. M. et al. Selective expression of the interleukin 7 receptor identifies regulation or exceeds the permitted use, you will need to obtain permission directly from effector CD8 T cells that give rise to long-lived memory cells. Nat. Immunol. 4, the copyright holder. To view a copy of this license, visit http://creativecommons.org/ 1191–1198 (2003). licenses/by/4.0/. 63. Coiras, M. et al. IL-7 induces SAMHD1 phosphorylation in CD4+ T lymphocytes, improving early steps of HIV-1 life cycle. Cell Rep. 14, 2100–2107 (2016). © The Author(s) 2019 16 NATURE COMMUNICATIONS | (2019)10:4739 | https://doi.org/10.1038/s41467-019-12732-2 | www.nature.com/naturecommunications