ARTICLE https://doi.org/10.1038/s41467-019-11556-4 OPEN Expression of CD20 after viral reactivation renders HIV-reservoir cells susceptible to Rituximab Carla Serra-Peinado 1, Judith Grau-Expósito 1, Laura Luque-Ballesteros 1, Antonio Astorga-Gamaza1, Jordi Navarro 1, Jenny Gallego-Rodriguez1, Mario Martin 1, Adrià Curran 1, Joaquin Burgos 1, Esteban Ribera 1, Berta Raventós 1, Rein Willekens 1, Ariadna Torrella 1, Bibiana Planas 1, Rosa Badía 1, Felipe Garcia 2, Josep Castellví 3, Meritxell Genescà1, Vicenç Falcó 1 & Maria J. Buzon 1 1234567890():,; The identification of exclusive markers to target HIV-reservoir cells will represent a sig- nificant advance in the search for therapies to cure HIV. Here, we identify the B lymphocyte antigen CD20 as a marker for HIV-infected cells in vitro and in vivo. The CD20 molecule is dimly expressed in a subpopulation of CD4-positive (CD4+) T lymphocytes from blood, with high levels of cell activation and heterogeneous memory phenotypes. In lymph node samples from infected patients, CD20 is present in productively HIV-infected cells, and ex vivo viral infection selectively upregulates the expression of CD20 during early infection. In samples from patients on antiretroviral therapy (ART) this subpopulation is significantly enriched in HIV transcripts, and the anti-CD20 monoclonal antibody Rituximab induces cell killing, which reduces the pool of HIV-expressing cells when combined with latency reversal agents. We provide a tool for targeting this active HIV-reservoir after viral reactivation in patients while on ART. 1 Infectious Disease Department, Hospital Universitari Vall d’Hebrón, Institut de Recerca (VHIR), Universitat Autònoma de Barcelona, Barcelona, Spain. 2 Infectious Disease Department, Hospital Clínic de Barcelona, Universitat de Barcelona, Barcelona, Spain. 3 Department of Pathology, Hospital Vall d’Hebrón, Universitat Autònoma de Barcelona, Barcelona, Spain. Correspondence and requests for materials should be addressed to M.J.B. (email: mariajose.buzon@vhir.org) NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 H IV infection represents an incurable disease to date. The load, VL; median time on suppressive ART 48 months), 20 HIV- persistence of the HIV-reservoir that is not susceptible to viremic patients (median VL 119,000 copies/ml) and 6 uninfected current antiretroviral therapy (ART) represents one of the control donors. The characteristics of individual patients are main barriers to cure HIV1,2. shown in Supplementary Table 1. The population of CD4+ T cells During the last two decades, intense efforts have focused on the expressing CD20 was identified according to the fluorescence search for new markers to successfully identify latently HIV- minus one (FMO) control for CD20 expression (Fig. 1a). As infected cells; cells that contain integrated HIV genetically intact shown in Fig. 1a and in agreement with previous studies12, we but able to remain quiescent for prolonged periods in vivo. found that CD4+ T cells expressed dim levels of the CD20 Recently, the paradigm that most latent HIV is transcriptionally receptor (CD20dim). Importantly, the inclusion of the CD19 inactive has been questioned by a new investigation that showed marker to exclude B cells did not alter the frequency of CD20dim that most HIV-infected CD4-positive (CD4+) T cells were able to CD4+ T cells (Supplementary Fig. 1a). To define the lymphocyte successfully initiate HIV transcription3. During ART, reservoir lineage of CD20dim CD4+ T cells, we performed fluorescent flow cells expressing HIV RNA are ~10 times more frequent than the cytometry imaging (Amnis®) on samples from HIV-infected proportion of cells containing replication-competent HIV and are patients. When cells were gated based on the expression of CD3, found in diverse populations of infected cells, including physio- CD4 and dim levels of CD20, we observed the unequivocal logically activated cells that produce viral particles upon cell expression of CD20 uniformly distributed through the plasma stimulation4,5. HIV-infected cells might also become pharmaco- membrane of CD4+ T cells. Lack of expression of CD19 on these logically activated to produce HIV transcripts by latency reversal CD20dim cells corroborated that they did not belong to a cano- agents (LRAs), which are compounds that are currently being nical B-cell phenotype (Fig. 1b). B cells were highly positive assessed in clinical trials for their ability to reactivate HIV from for CD19 and CD20 expression (Fig. 1c). Of note, we also found its dormant state. LRAs have been shown to be effective in CD4+ T cells with high expression of CD20, however, the flow inducing HIV transcription, but importantly, viral-reactivated cytometry imaging analysis showed that they corresponded to cell cells are not completely cleared by the immune system or killed aggregates consisting of T-B-cell doublets (Supplementary by viral cytopathic effects in vivo6–8. Thus, the identification of Fig. 1b). In addition, conventional quantitative PCR showed the novel markers to identify transcriptionally active HIV-infected dim expression of CD20-mRNA in previously isolated CD20dim cells, including physiologically and/or pharmacologically HIV- CD4+ T cells (Supplementary Fig. 1c). Next, we quantified the reactivated cells, could considerably advance the development of percentage of CD20 in CD4+ T cells of HIV+ patients and novel therapies to target and deplete the HIV-reservoir. uninfected controls. We observed that viremic HIV+ patients CD20 is a cell membrane receptor preferentially expressed on B presented statistically significant higher proportions of CD4+ cells. Early reports have also demonstrated the existence of a T cells expressing CD20 compared with uninfected donors proportion of CD3-positive (CD3+) T cells with dim expression (Fig. 1d). Moreover, CD4+ T cells from HIV+ viremic individuals levels of CD20 (between 1 and 5% in blood)9–12. In healthy had slightly higher expression of CD20 than virologically sup- individuals, CD20-positive (CD20+) T cells are phenotypically pressed HIV+ patients. The antiretroviral treatment normalized different than their counterpart CD20-negative (CD20−) T cells; the expression of CD20 since no differences were found between approximately half of them are CD4+ cells, they are more likely ART-suppressed patients and uninfected controls (Fig. 1d). When to express the gamma/delta T-cell receptor, and they express we compared the activation levels, defined by the expression of more CD45RO and fewer CD38 molecules on their surface12. In HLA-DR, CD69, CD25 or any combination of these cell markers, patients suffering from rheumatoid arthritis13,14 CD20+ T cells we observed that in HIV+ patients CD20dim CD4+ T cells were have been found to have a low proliferative capacity, a high more activated than CD4+ T cells not expressing CD20 (Fig. 1e). activation state, and an enhanced susceptibility to apoptosis9. In Next, we studied the memory phenotype of these cells in HIV patients with multiple sclerosis, in vivo administration of the anti- infection. In terms of the distribution of CD20dim among the CD20 monoclonal antibody Rituximab effectively induces long- different CD4+ T cell subsets, most of the CD20dim cells had a term depletion of T cells expressing CD2011. Moreover, CD20+ central memory (TCM) phenotype, followed by the effector T cells have been postulated as early T-cell precursors in the memory (TEM), transitional memory (TTM), and naive (TNA) replenishment of diverse populations of T cells11,12. Therefore, phenotypes. CD20dim cells were less distributed among memory the intrinsic characteristics of CD20+ T cells and their suscept- stem cells (TSCM) and terminally differentiated (TTD) cells ibility to Rituximab administration in vivo make the CD20 (Fig. 1f). Thus, all the memory subsets accounted for almost molecule a good candidate to explore in the setting of HIV 80% of the total expression of CD20dim CD4+ T cells. Simulta- reservoirs. Here, we investigate CD20 as a putative marker for neously, we observed that memory subsets, including CD4+ TCM, HIV infection and during in vivo viral persistence. We find that CD4+ TEM, and CD4+ TTM cells, contained significantly higher CD20 expression in CD4+ T cells is expanded during untreated percentages of CD20dim-expressing cells, as compared with the HIV infection, and during ART, CD20dim T cells contain high less differentiated CD4+ TNA and CD4+ TSCM, or with the CD4+ levels of HIV transcripts. Importantly, Rituximab induces cell TTD cells (Supplementary Fig. 2a). In this regard, small differ- death and decreases the size of the transcriptionally active HIV- ences in CD4+ T cell subsets were observed between reservoir when combined with LRAs ex vivo. Thus, we identify ART-suppressed individuals, HIV-viremic individuals, and CD20 as a new marker for HIV-infected cells in patients and uninfected controls (Supplementary Fig. 2b). When we analyzed provide evidence for the combined use of LRAs and anti-CD20 all CD4+ T memory cells together, defined by the expression of monoclonal antibody therapy for the depletion of HIV-reservoir CD45RO, we found the same trend observed in the total CD4+ T- cells after viral reactivation. cell population (shown in Fig. 1f), suggesting that these differ- ences are mainly supported by the memory pool (Supplementary Fig. 2c). We also studied if CD20dim CD4+ T cells expressed Results cell markers characteristics of follicular T helper (TFH) cells, A subset of CD4+ T cells expresses CD20 in HIV infection. defined by the expression of PD1 and CXCR5. We observed that First, we aimed to determine the proportion of CD4+ T cells CD20dim cells expressed significantly more double positive expressing the CD20 molecule during HIV infection in samples CXCR5+ and PD1+ than CD20− CD4+ T cells (Supplementary from 21 ART-suppressed patients (<50 copies/ml of plasma viral Fig. 1d). 2 NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 ARTICLE a From live cells 100 105 CD4+ T cells 33.60 250 K 105 NA+SCM CM EM TM 47.50 52.50 CCR7-PE-CF594 200 K 8.26 45.00 80 104 104 60 CD3-PE-Cy7 150 K 103 100 K 103 40 99.39 0.61 SSC-A Count 0 50 K 0 20 TD –103 20.10 0 –103 2.64 TM+EM 44.0 0 3 –10 0 3 10 104 5 10 –103 0 103 104 105 –10 3 0 103 10 4 5 10 –10 3 0 103 104 105 CD4-AF700 CD20-BV785 CD45RO-BV605 CD27-FITC FMO 25 250 K 250 K 105 200 K 200 K 20 SCM 15.80 CD95-PE-Cy5 104 15 150 K 150 K B cells 100 K 10 103 100 K SSC-A 29.00 99.96 0.04 SSC-A Count 98.10 5 0 50 K 50 K –103 NA 84.20 0 0 0 –103 0 10 3 104 105 –103 0 103 104 105 –103 0 103 104 105 –103 0 103 104 105 CD20-BV785 CD20-BV785 CD27-FITC CCR7-PE-CF594 b CD20 CD19 CD3 Live/dead CD4 5452 26,084 67,507 88,135 14,297 33,965 c 1249 1471 2190 d e <0.0001 0.019 f 0.002 <0.0001 % CD25+ or CD69+ or HLA-DR+ cells 0.037 100 0.001 4 0.013 0.001 0.015 100 % CD20dim in CD4+ T cells <0.0001 % Distribution of CD20dim cells 80 <0.0001 3 80 in CD4 T subsets 60 2 60 40 + 40 1 20 20 0 0 0 – – m m ic T r d di di 20 20 m no te AR 20 20 TNA TSCM TCM TEM TTM TTD re D do fec D C C Vi D D n C C ni U ART Viremic Viremic g Spearman r = – 0.450 Spearman r = – 0.324 Spearman r = 0.342 ART 4 p = 0.040 4 p = 0.046 3 p = 0.139 % CD20dim in CD4+ T cells in CD4+ T cells in CD4+ T cells 2 2.0 2 2.0 2 1.5 1.5 1.0 dim dim 1.0 1 % CD20 % CD20 0.5 0.5 0.0 0.0 0 0 50 100 150 0 500 1000 1500 2000 2500 3 4 5 6 7 8 + Time on suppressive ART CD4 T cells/µl Log viral load (RNA copies/ml plasma) (month) Finally, we evaluated the relationship between the expression of CD4+ T cells compared with uninfected controls that inversely CD20 in T cells and markers of HIV progression. A significant correlated with CD4+ T cell counts and time on suppressive ART. inverse correlation was observed between the percentage of Moreover, the CD20 marker was associated with activated cells CD20dim CD4+ T cells and the number of months that patients and mainly expressed and distributed in memory CD4+ T-cell were under suppressive treatment (VL < 50 copies/ml), and with subsets. their CD4+ T-cell count (Fig. 1g). A tendency toward a positive correlation was also observed with the plasma viral load (Fig. 1g). CD20dim CD4+ T cells are enriched in intracellular HIV RNA. In summary, HIV+ individuals had higher expression of CD20 on Since CD4+ T cells expressing the marker CD20 seem to be more NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 Fig. 1 CD20dim CD4+ T cells involve memory phenotypes with more activation than CD20− cells. PBMC from uninfected donors, ART-suppressed patients and viremic patients were stained with CD20, activation markers (HLA-DR, CD69, CD25), and T-cell differentiation markers (CCR7, CD45RO, CD27, CD95). a Gating strategy used to identify CD20dim CD4+ T cells. Previous sequential gates are represented in Supplementary Fig. 9. Viable CD4+ T cells (identified by CD3 and CD4 expression) were first selected for the analysis of CD20 expression, defined by its Fluorescence Minus One (FMO) control. T- cell memory subsets were selected as follows: CD4+ TNA (CCR7+, CD45RO−, CD27+, CD95−), CD4+ TSCM (CCR7+, CD45RO−, CD27+, CD95+); CD4+ TCM (CCR7+, CD45RO+), CD4+ TTM (CCR7−, CD45RO+, CD27+); CD4+ TEM (CCR7−, CD45RO+, CD27−), CD4+ TTD (CCR7−, CD45RO−). b, c Representative bright-field and pseudo-color fluorescence images of CD20dim CD4+ T cells (b) and B cells (c) from two ART-suppressed patients (#9 and #22) using the Amnis imaging flow cytometer technology. Scale bar 10 µm. d Percentage of CD20dim expression within CD4+ T cells in uninfected controls and the two patient cohorts. Mann–Whitney comparison was used to compare n = 6 uninfected controls, n = 21 ART-suppressed patients, n = 20 viremic patients. e Expression of the activation markers CD25, CD69 or HLA-DR in CD20dim and CD20− CD4+ T cells in different cohorts of HIV+ patients. Comparisons were performed using the two-tailed Wilcoxon test (n = 21 ART-suppressed patients and n = 20 viremic patients). f Distribution of CD20dim in CD4+ T-cell subsets in ART-suppressed (orange) and viremic (red) HIV-infected patients. NA, naive; SCM, memory stem cell; CM, central memory; TM, transitional memory; TD, terminal differentiated. ANOVA and Dunn’s multiple comparison test were performed, n = 10 ART-suppressed and n = 9 viremic patients. Median values and min and max ranks are represented in panels d–f. g Correlation of CD20dim CD4+ T cells with time on suppressive ART (left), CD4+ T-cell counts (middle) and plasma viral load (right) are shown. ART-suppressed patients in orange and viremic patients in red. Spearman’s nonparametric correlation coefficients and associated P values are shown. Panels (d) and (e) included patients #2–10, 15–26, 60–65, 67–74, and 76–81. Panel (f) patients #1–10, 60–65, and 67–69. Panel (g) patients #2–10, 15–26, 60–65, 67–74, and 76–81. Data underlying this Figure are provided as Source Data file activated than the general population of CD4+ T cells, and cell CD20+ T cells are infected in human lymph nodes. We next activation has been related to higher levels of HIV-1 infection and visualized and quantified the CD20 receptor in anatomically transcription, we sought to investigate if CD20dim CD4+ T cells intact lymph node tissue preparations from two ART-suppressed from HIV+ patients might also contain more transcriptionally patients and four viremic HIV-infected patients (Supplementary active HIV. In general, no correlation was found between the Table 1). First, we focused on the quantification of T cells co- expression of CD20 in CD4+ T cells and viral nucleic acids expressing the marker CD20. We detected CD20+ T cells in both during ART (Supplementary Fig. 2d). Using the novel RNA the extrafollicular zone and within the B-cell follicle in all samples FISH-flow assay4, we measured the frequency of CD20dim CD4+ analyzed. Representative micrographs are shown in Fig. 4a. In the T cells that were co-expressing HIV RNA. We included samples sample from one HIV-viremic patient (#82), we observed sig- from 12 ART-suppressed patients (median time on suppressive nificantly higher absolute number of CD20+ T cells within the B- ART 23 months), 11 viremic patients (median VL 74,500 copies/ml) cell follicle compared with the extrafollicular region, while the and 4 uninfected controls. The characteristics of the included two ART-suppressed patients had lower quantities of CD20+ patients are shown in Supplementary Table 1. The gating strategy T cells in B-cell follicles compared with the viremic patients used to identify CD20dim CD4+ T cells and HIV RNA expression (Fig. 4b, upper panel). Also, we quantified the percentage of is shown in Fig. 2a. Both cohorts of HIV-infected patients, vire- follicular CD3+ cells that co-expressed CD20. We found that, in mic and ART-suppressed patients, presented a significantly general, the median percentage of CD3+ cells positive for CD20 higher proportion of CD20dim CD4+ T expressing viral tran- expression ranged from 2 to 4% in individual patients (Fig. 4b, scripts compared with their counterpart CD20− CD4+ T cells lower panel). Next, we investigated whether HIV-infected cells (Fig. 2b). These infected CD20dim CD4+ T cells contributed to a were also expressing CD20 in lymph node tissue specimens from median of 18.55% and 25.0% to the total pool of HIV-expressing three viremic and one ART-suppressed HIV-infected patients. cells in ART-suppressed and viremic patients, respectively Since productively HIV-infected cells have been previously (Fig. 2c). Overall, we observed that CD20dim CD4+ T are enri- identified as being CD45RO-positive memory CD4+ T cells15, we ched in HIV RNA and contribute significantly to the total pool of did not perform phenotypic identification of such cells. Instead, HIV-expressing cells. we directly performed immunostaining of the viral protein p24 and the CD20 antigen. P24 showed the typical two-staining pattern observed in the B-cell follicles16; a network-like staining Proviral HIV DNA levels of CD20dim CD4+ T cells. We next probably corresponding to virions captured by follicular dendritic asked if the increased viral transcription observed within the cells (FDCs), and cells with densely spherical signal indicating CD20dim CD4+ T population was associated with enrichment in productively HIV-infected cells. Notably, within the B-cell folli- viral DNA. We used fluorescence activated cell sorting to isolate cles of all patients, we observed productively HIV-infected cells total CD4+ T cells, activated CD4+ T cells (CD20−), resting CD4+ co-expressing the CD20 molecule (Fig. 4c), but cell quantification T cells (CD20−), activated CD20dim CD4+ T cells, and resting was difficult due to the high density of B cells expressing CD20. CD20dim CD4+ T cells from nine HIV-infected ART-suppressed To finally demonstrate the expression of CD20 in productively individuals (median time on suppressive ART of 156 months) HIV-infected cells, we used the in situ hybridization ultrasensitive (Supplementary Table 1). Resting CD20dim CD4+ T cells had RNA detection assay (eBiosciences) to visualize HIV RNA and significantly higher levels of HIV DNA compared with the total CD20 RNA in a lymph node section of one HIV+ viremic patient. CD4+ T cells (Fig. 3a, left panel). This enrichment represented a In this particular patient, productively HIV-infected cells were 2.1-fold change in the amount of HIV DNA per million cells often associated with the expression of CD20, inside and outside between both subsets (Fig. 3a, right panel). Moreover, we calcu- of the B-cell follicle (Fig. 4d). Overall, we confirm the existence of lated the global contribution of these subsets to the total pool of HIV-infected cells expressing CD20 in lymph nodes from ART- HIV-infected cells, and we found that resting CD20dim con- suppressed and viremic HIV-infected patients. tributed to a median of 13.26% compared with 25.44% and 57.79% of activated CD4+ T cells and resting CD4+ T cells, HIV upregulates CD20 in T cells during early viral infection. respectively (Fig. 3b). To investigate the role of HIV-1 replication in the expression of 4 NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 ARTICLE a Viremic ART Uninfected donor 250 K 250 K 250 K 200 K 200 K 200 K 150 K 150 K 150 K 100 K 100 K 100 K 0.013 0.007 0.001 50 K 50 K 50 K 0 0 0 –103 0 103 104 105 –103 0 103 104 105 –103 0 103 104 105 250 K 250 K 250 K 250 K From live cells 200 K 200 K 200 K 200 K 105 19.0 150 K 150 K 150 K 150 K 104 97.30 2.70 CD3-PE-Cy5 100 K 100 K 100 K 100 K SSC-A SSC-A 0.27 0.08 0 103 50 K 50 K 50 K 50 K 0 0 0 0 0 –103 0 103 104 105 –103 0 103 104 105 –103 0 103 104 105 –103 0 103 104 105 –103 CD20-BV785 HIV-RNA-AF647 0 102 103 104 CD4-BV510 250 K 250 K 250 K 200 K 200 K 200 K 150 K 150 K 150 K 64 36 80 20 SSC-A 100 K 100 K 100 K SSC-A 0.02 50 K 50 K 50 K 0 0 0 –103 0 103 104 105 –103 0 103 104 105 –103 0 103 104 105 HIV-RNA-AF647 CD20-BV785 b c Distribution of HIV-RNA+ cells dim + 0.5 0.002 0.004 CD20 CD4 T cells CD20dim CD4+ T cells – + CD20 CD4 T cells CD20– CD4+ T cells % HIV-1 RNA cells 0.4 0.3 0.2 18.55% 25% 0.1 81.45% 75% 0.0 CD20dim CD20– CD20dim CD20– ART Viremic CD4+ T CD4+ T CD4+ T CD4+ T ART Viremic Fig. 2 CD20dim CD4+ T cells are enriched in HIV-1 RNA. Unstimulated PBMCs were subjected to the RNA FISH-flow protocol to detect HIV RNA expression in CD20dim CD4+ T cells. a Gating strategy used to determine the HIV RNA expression on CD20− and CD20dim CD4+ T cells and expression of CD20 in HIV RNA+ cells. Previous sequential gates are represented in Supplementary Fig. 9. b Frequency comparison of HIV RNA+ cells in CD20dim CD4+ T cells and CD20− CD4+ T cells of ART-suppressed and viremic patients. Wilcoxon test was used. c Mean distribution of HIV RNA+ cells in the CD20+ and CD20− phenotypes. Median values and min and max ranks are represented in panels (b) and (c). In all panels, n = 12 ART-suppressed and n = 11 viremic patients are shown. Data from patients #15–26, 60, 63, and 73–81 are included. Data underlying this Figure are provided as Source Data file CD20, we performed ex vivo infection experiments. Unstimulated upregulation of CD20 in infected cells impacted the overall PBMCs from three uninfected donors were infected with the viral expression of CD20 in the cell culture. Notably, a significant strain NL4.3 for 5 days. The gating strategy for productively HIV- positive correlation was found between the percentage of pro- infected cells, identified as HIV RNA-positive (HIV RNA+) and ductive HIV-infected cells and the percentage of CD20+ infected p24-positive (p24+) T cells, and expression of CD20 are shown in cells (Fig. 5d left), or with the mean fluorescence intensity of Fig. 5a. Ex vivo infection was successfully detected using this CD20 expression of the infected fraction (Fig. 5d right), sug- novel assay; we obtained increasing percentages of HIV infection gesting a selective upregulation of CD20 in HIV-infected cells. In over time (Fig. 5b left). Interestingly, when we focused on addition, in order to show the unequivocal upregulation of CD20 CD20dim T cells, we observed that this subpopulation supported after HIV infection, we performed cell sorting of the CD4+ CD20− HIV infection more efficiently than CD20− T cells (Fig. 5b right). population before cell infection. Again, a clear induction of CD20 Moreover, the contribution of infected CD20dim T cells to the expression after 3 days of cell infection was observed (Supple- total pool of HIV-infected cells significantly increased over time, mentary Fig. 3a, b). This result was confirmed by quantification of reaching 30% at day 5 post-infection (Fig. 5c left). Of note, we CD20-mRNA before and after cell infection using qPCR (Supple- also observed that CD20 expression was increased in the general mentary Fig. 3c). Since we have recently described that CD32 is population of T cells after infection (Fig. 5c right), indicating that also upregulated in HIV-infected cells17, we next investigated the NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 a 0.017 FC = 2.1 b 0.017 Activated CD20– CD4+ T Cells 10 4 4 Resting CD20– CD4+ T Cells 10 dim + Activated CD20 CD4 T Cells HIV-DNA copies/10*6 CD4+ T cells HIV-DNA copies/10*6 CD4+ T cells Resting CD20dim CD4+ T Cells 103 103 102 102 1 10 100 101 Total Activated resting Activated resting 4+ l D a T T C ot CD4+ T 4+ C g T im tin CD20– CD4+ T CD20dim CD4+ T D 20 d es D R C Fig. 3 HIV-1 DNA quantification in CD20dim CD4+ T cells. HIV DNA was measured in sorted CD4+ T cells (black), activated CD20− CD4+ T cells (dark violet), resting CD20− CD4+ T cells (light violet), activated CD20dim CD4+ T cells (dark orange), and resting CD20dim CD4+ T cells (light orange) in PBMCs from nine ART-suppressed patients. The activation state was defined by the expression of HLA-DR, CD69 or CD25 markers. a Copies of viral DNA molecules per 106 CD4+ T cells are represented for all sorted populations (left panel). Medians and min–max rank are represented. Comparison of HIV DNA copies per 106 cells between CD4+ T cells and resting CD20dim CD4+ T cells with the fold-change (FC, right panel). Comparisons were performed using the Wilcoxon test. b Contribution of each cell population to the total pool of HIV DNA. In panels (a) and (b), data from patients #11–14, 19, and 27–30 are shown. Data underlying this Figure are provided as Source Data file relationship between these two markers. We observed that in HIV- significantly reduced the proportion of CD20dim CD4+ T cells infected cells, CD20 was upregulated to a greater extent than CD32 (Fig. 6c). Gating strategy and the individual data are shown in (Supplementary Fig. 4). Finally, we observed that CD20dim HIV- Supplementary Fig. 6. infected cells expressed more frequently HLA-DR than CD20− Since we have observed that CD20 was upregulated upon HIV HIV-infected cells (ANOVA p = 0.001), and also the immune infection and Rituximab efficiently targeted CD20dim CD4+ checkpoint marker PD-1 (ANOVA p = 0.01) (Fig. 5e). To ascertain T cells, we next asked if Rituximab had any impact on ex vivo the dynamics of CD20 expression in HIV-infected cells, we took infection cultures of unstimulated PBMCs. Addition of Rituximab advantage of the CD4 downregulation induced by HIV in highly to the cell culture completely abrogated productive HIV infection productive HIV-infected cells. During early infection, defined by of CD20dim CD4+ T cells (Fig. 6d, left panel). It is noteworthy infected cells expressing CD4, levels of CD20 were notably higher that the entire population of CD20dim CD4+ T cells was not than during late infection, as defined by cells with marked down- depleted by Rituximab (as observed in Fig. 6b); however, cells that regulation of the CD4 receptor (Fig. 5f). The opposite trend was were efficiently eliminated were those supporting productive HIV observed for the activation marker HLA-DR (Supplementary infection. Moreover, Rituximab impacted the total pool of Fig. 5a). These results suggest that this early upregulation of CD20 infected CD4+ T cells, reducing HIV infection by 38.6% at day upon HIV infection might be a virus-specific effect, since the 3 (Fig. 6d, right panel). general activation of T cells with anti-CD3 and CD28 antibodies only increased the expression of CD20 over time (96 h) slightly. In In vivo administration of Rituximab targets HIV-RNA+ cells. contrast, CD69 and HLA-DR were strongly upregulated after CD3/ To assess the in vivo effect of Rituximab on the HIV-reservoir, we CD28 activation (Supplementary Fig. 5b). took advantage of available samples obtained from three ART- suppressed patients receiving Rituximab for non-malignant rea- Rituximab targets HIV-infected CD20dim CD4+ T cells sons. Patient #52 received Rituximab at 1000 mg/dose on days 0 in vitro. Rituximab is a chimeric monoclonal antibody directed and 14 due to an IgG4-related systemic disease. Patient #53 only against the B-cell-specific antigen CD20 used to deplete B cells in received one dose of 1000 mg due to fibrillar glomerulonephritis. certain types of lymphoma18. We tested if Rituximab could also Patient #59 received three doses of 375 mg/m2 of Rituximab at target and deplete T cells expressing low levels of CD20. PBMCs days 0, 7, and 33 due to a mucocutaneous ulcer caused by from uninfected donors and three ART-suppressed individuals Epstein–Bar virus. Consistent with our in vitro data, we observed (median time on suppressive ART of 336 months) (Supplemen- that Rituximab efficiently depleted more than 80% of CD20dim tary Table 1) were treated with 10 µg/ml of Rituximab for 48 h. As CD4+ T cells in the three patients, and this depletion was con- expected, Rituximab significantly reduced the number of B cells tinually sustained throughout the study (Fig. 7a, b). Moreover, we (Fig. 6a). Moreover, a marked decreased of CD20dim CD4+ measured intracellular HIV RNA in isolated CD4+ T cells in T cells in culture was also observed (Fig. 6b). We also studied if these patients. Although no LRAs were administrated to the Rituximab induced antibody-dependent cell-mediated cytotoxi- patients, we observed a transient 2.38 fold-decrease of HIV RNA city (ADCC) of CD20dim CD4+ T cells. To achieve this goal, in one patient (#52), a sustained 2.4-fold reduction of HIV RNA Rituximab-treated CD4+ T cells, stained with the eFluor670 dye in the second patient (#53), and no significant changes in patient as previously shown19, were used as target cells in combination #59 (Fig. 7c). The low basal HIV transcription level observed in with Natural killer (NK) effector cells. We found that NK cells the third patient might explain the lack of rituximab effect on were responsible for inducing a potent ADCC response that HIV+ cells. As body distribution of different cell subpopulations 6 NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 ARTICLE a b 0.060 Intrafollicular 0.004 Extrafollicular 150 CD20 CD3 CD20 CD3 Merge CD20+ CD3+ cell/mm2 100 15 µm 15 µm 15 µm 50 0 15 µm 15 µm 15 µm Viremic 1 Viremic 2 ART 1 ART 2 % CD20+ in intrafollicular T cells 25 20 15 µm 15 15 µm 15 µm 10 5 100 µm 0 ic 1 ic 2 T1 T2 AR AR em em Vir Vir c d Viremic 2 Viremic 3 CD20-RNA HIV-RNA Merge CD20 p24 CD20-RNA HIV-RNA DAPI 10 µm 10 µm 10 µm 15 µm 15 µm Viremic 4 ART1 10 µm 10 µm 10 µm 15 µm 15 µm 10 µm 10 µm 10 µm 100 µm 50 µm Fig. 4 CD20 expression in T cells of intact human lymph nodes. Paraffin-embedded lymph node sections from HIV-infected patients were stained with anti-CD20 combined with anti-CD3 (a) or anti-p24 antibodies (b), alternatively sections were proceed for in situ RNA-hybridization of CD20-mRNA or mRNA and HIV RNA (c). a Representative micrograph of a lymph node section from one viremic HIV+ patient stained with anti-CD20 (green) and anti- CD3 (violet) antibodies. The discontinuous line indicates a B-cell follicle and the white arrows double positive cells. Right panels correspond to zoomed views from different patients. b Upper panel shows the number of cells per mm2 expressing CD20+ and CD3+ in lymph nodes (circles, each dot corresponds to one follicle) and in the extrafollicular region (triangles, each dot corresponds to one optical field). Two HIV-viremic (#84 and 85, in red) and two ART-suppressed patients (#31 and 32, in orange) are shown. Comparison between values inside B-cell follicles from viremic and ART-suppressed patients was performed using a Mann–Whitney test, and comparison within viremic 1 was performed with the Wilcoxon test. Lower panel shows the percentage of CD3+ cells within the B-cell follicle expressing CD20. Each dot corresponds to one follicle. Medians and min–max rank are represented. c Representative micrograph from a lymph node section from one HIV-viremic patient stained with anti-CD20 (green) and anti-p24 (violet) antibodies. High magnification images correspond to B-cell regions from different ART-suppressed and viremic patients, white arrows indicate double positive cells and the discontinuous line indicates B-cell follicle. Patients #31 and 82–84 are shown. d Representative image of a B-cell follicle section from one viremic patient (#82) subjected to the in situ hybridization technique. CD20-mRNA is represented in green and HIV RNA in violet. At the middle, zoomed images of double positive cells (white arrows) or HIV RNA single positive cells (orange arrows). Data underlying this Figure are provided as Source Data file might impact detection of HIV-RNA+ cells in blood, we quan- PMA/ionomycin were added to cells and the expression of HIV tified the proportion of major CD4+ T-cell subsets after Ritux- RNA was assessed by the RNA FISH-flow assay (Fig. 8a). The imab administration. No consistent changes in cell proportions addition of romidepsin and ingenol significantly increased the were observed (Supplementary Fig. 7). proportion of CD20dim cells expressing viral transcripts, with higher frequencies observed in CD20dim cells compared with Rituximab impacts HIV+ cells after viral reactivation. As the CD20-negative cells in all conditions (Fig. 8b), and with a con- frequency of CD20 expression in CD4+ T cells is relatively low, comitant increase in the proportion of infected cells expressing and HIV infection upregulates CD20, we wondered if reactivation CD20dim in the whole population of CD4+ T cells (Fig. 8c). Under of the latent reservoir might also upregulate CD20 in HIV- all conditions tested, cells expressing HIV RNA presented higher infected cells and render cells susceptible to Rituximab. Samples levels of CD20dim compared with uninfected cells. In addition, from ART-suppressed patients were included in these studies ingenol and PMA/ionomycin induced a significant upregulation of (median time on suppressive ART of 23 months) (Supplementary CD20dim in the whole population of infected CD4+ T cells after Table 1). First, we performed in vitro reactivation assays using viral reactivation. A very low upregulation of CD20 in uninfected fresh CD4+ T cells from eight ART-suppressed HIV-infected cells and CD8 T cells was also observed for ingenol and PMA/ patients. The LRAs romidepsin, ingenol or the positive control ionomycin (Fig. 8d). NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 a From live T cells 250 K 104 200 K 103 CD4-BV510 150 K 27.30 100 K From live cells 102 250 K 5 50 K 10 104 85.30 0 200 K 0 104 103 –103 0 103 104 105 –103 0 103 104 105 0.15 HLA-DR-PE-CF594 CD3-PE-Cy7 150 K 2.42 102 250 K 100 K 103 105 P24-PE FSC-A 0 200 K 50 K 0 104 99.85 –103 150 K CD4-BV510 0 2.36 –103 0 103 104 105 –103 0 103 104 105 –103 0 103 104 105 100 K 103 FSC-A CD20-BV785 CD20-BV785 HIV-RNA-AF647 0 50 K –103 0 –103 0 103 104 105 –103 0 103 104 105 CD20-BV785 PD-1-BV605 b + CD3 T cells CD20dim c Infected CD3+ T cells CD20 – Uninfected 0.4 8 40 5 % HIV-1 RNA+ p24+ % HIV-1 RNA+ p24+ 4 0.3 6 30 % CD20dim % CD20dim 3 0.2 4 20 2 0.1 2 10 1 0.0 0 0 0 3 4 5 3 4 5 3 4 5 3 4 5 Days post infection Days post infection Days post infection Days post infection d f + Infected CD4 T cells 0.4 0.4 40 p = 0.004 p = 0.001 Infected CD4– T cells Spearman r = 0.857 Spearman r = 0.941 % HIV RNA+ p24+ % HIV RNA+ p24+ 0.3 0.3 30 in CD3+ T cells in CD3+ T cells % CD20dim 0.2 0.2 20 0.1 0.1 10 0.0 0.0 0 0 10 20 30 40 200 300 400 500 600 1 2 3 % CD20dim in HIV RNA+ p24+ MFI of CD20 in HIV RNA+ p24+ Days post infection e 100 100 dim Infected CD20 – 80 80 Infected CD20 Uninfected CD20dim % HLA-DR+ % PD-1+ 60 60 – Uninfected CD20 40 40 20 20 0 0 3 4 5 3 4 5 Days post infection Days post infection Next, we investigated whether Rituximab was able to impact Thereafter, CD4+ T cells were isolated and HIV RNA quantified cells actively transcribing HIV RNA after pharmacological viral by conventional qPCR. We observed successful viral reactivation reactivation in 12 ART-suppressed patients. PBMCs from these in 7 out of 12 samples tested (median fold-change reactivation of patients were incubated for 24 h with Rituximab and romidepsin 3.7). Importantly, the potency to induce viral reactivation by or panobinostat, and for an additional 24 h with Rituximab. LRAs was directly associated with the fold-change reduction of 8 NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 ARTICLE Fig. 5 Ex vivo infection upregulates CD20 expression. Unstimulated PBMCs from three uninfected donors were infected with HIV strain NL4.3. Infection was monitored by simultaneous staining of HIV RNA using the RNA FISH-flow assay and the viral protein p24 at days 3, 4, and 5. a Gating strategy used to monitor HIV infection and expression of CD20dim. Previous sequential gates are represented in Supplementary Fig. 9. b Percentage of productively HIV- infected cells in CD3+ T cells (left panel) and in CD20dim or CD20− CD3+ T cells (right panel). c Expression of CD20dim in infected and uninfected CD3+ T cells (left panel) and in the total CD3+ T cell population (right panel). d Correlation between the proportion of infected cells within the CD3+ T cell population and CD20 expression (left panel) and the mean fluorescence intensity (MFI) of CD20 (right panel). e Percentage of HLA-DR and PD-1 in infected and uninfected cells expressing or not expressing CD20dim. f Proportion of CD20dim in infected cells expressing the CD4 cell receptor versus infected cells with marked downregulation of the CD4 receptor. In all panels, the mean and SEM value of three independent experiments is represented. In panel (d), Spearman’s nonparametric correlation coefficients and associated p values are shown. Data underlying this Figure are provided as Source Data file a 0.031 b 0.031 c 0.031 d +RTX % dead CD3+p24+ cells by RTX %p24+ in CD3+CD20dim cells % CD20dim in CD4+ T cells 15 6 25 0.5 80 –RTX 20 0.4 10 4 60 % B cells 0.3 % ADCC 15 0.2 40 5 2 10 0.1 20 5 0 0 0.0 0 0 – RTX + RTX – RTX + RTX – RTX + RTX 3 4 5 3 4 5 Days post infection Days post infection Fig. 6 Rituximab treatment eliminates infected CD20dim CD4+ T cells and reduces HIV-1 infection in vitro. a, b Isolated PBMCs from uninfected donors (green) and ART-suppressed HIV+ patients (orange) were treated with Rituximab (RTX) or antibody control for 48 h. Cell death was assessed by flow cytometry using the gating strategy shown in Fig. 1a. Panels represent percentages of B cells (a) or CD20dim in CD4+ T cells (b) for n = 3 uninfected donors and n = 3 ART-suppressed patients (#56–58). Median values and min–max ranks are represented. c Isolated CD4+ T cells incubated with Rituximab or antibody control were used as a target in combination with isolated autologous NK cells. The percentage of ADCC was measured by the reduction of eFluor670 staining by flow cytometry using the gating strategy shown in Fig. S6. Summary results for six uninfected donors are represented. d PBMCs from n = 3 uninfected donors were infected with the HIV strain NL4.3 and treated with Rituximab for different periods of time. Percentages of infected cells (p24+) in CD20dim CD3+ T cells with (dark blue) or without (light blue) Rituximab are represented in the left panel (mean ± SEM). The percentage of dead infected cells induced by Rituximab is shown in the right panel, which represents the difference in the proportion of infected cells between Rituximab-treated cells versus Rituximab non-treated cells (mean ± SD). Flow gating strategy used for this analysis is shown in Fig. 5a. All comparisons were performed using the Wilcoxon test. Data underlying this Figure are provided as Source Data file RNA-expressing cells by Rituximab (Fig. 8e). Moreover, in the shown for T cells expressing high levels of CD3222. We found that seven samples responding to LRA viral reactivation, Rituximab a small proportion of CD4+ T cells expressed dim levels of CD20, induced significant depletion of HIV RNA-expressing cells and this population was significantly expanded by ~2-fold in compared with cells non-responding to LRAs (Fig. 8f), with a viremic patients. These findings, together with the enrichment of median reduction of HIV RNA of 64% [18–87]. For samples HIV RNA in this subpopulation, the upregulation of CD20 after responding to LRA treatment, individual patterns of viral RNA in vitro infection, and the normalization of CD20 levels in ART- expression in the presence of LRAs with or without Rituximab are suppressed individuals, strongly suggests a direct link between shown in Supplementary Fig. 8. Overall, Rituximab was capable HIV infection and the expression of CD20. Moreover, the fre- of targeting and deplete cells expressing HIV RNA after viral quency of CD4+ T cells expressing CD20 inversely correlated reactivation. However, the ability of Rituximab to impact this with CD4+ T-cell counts and the time on suppressive ART. This transcriptionally active viral reservoir directly depended on the phenomenon has been extensively reported for conventional ability of LRAs to induce viral transcription. activation markers, such as CD38 and HLA-DR23,24. Thus, it is tempting to speculate that CD20 might behave as a non- Discussion conventional activation marker during HIV infection for CD4+ Understanding the mechanisms that contribute to HIV-1 per- T cells. We found that in general, ~1% of CD4+ T cells expressed sistence despite ART, and the development of therapeutic stra- CD20, and most of those cells had a memory phenotype. These tegies to target persistent virus, represent current priorities in percentages of expression have been previously reported in HIV-1 research20,21. Here we report the existence of a significant patients with multiple sclerosis and rheumatoid arthritis9,25. In fraction of circulating CD4+ T cells that express CD20, a cell agreement with our results, Schuh and colleagues studied features membrane receptor preferentially expressed on B cells, which of human CD3+ CD20dim cells during multiple sclerosis and associates with HIV infection and cells enriched for viral RNA in found that these T cells were enriched in CD45RO+ memory ART-suppressed patients. Importantly, treatment with Rituximab cells25. We also found that CD4+ T cells expressing the CD20 eliminates this fraction while impacting on the HIV-viral reser- marker were in general more activated than their CD20− coun- voir when combined with LRAs. terparts. Accordingly, increased activation of T cells expressing Using different methodologies we unequivocally demonstrated CD20 has been widely reported26. that CD20dim CD4+ T cells are a truly single cell fraction, and The identification of specific cell subpopulations harboring they are not the result of flow cytometry artifacts, as previously HIV is important for the design of novel therapies directed to NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 a 250 K 250 K 250 K 250 K Day 0 Day 2 Day 7 Day 30 200 K 200 K 200 K 200 K 150 K 150 K 150 K 150 K 100 K 100 K 100 K 100 K SSC-A 0.12 0.01 0 0.005 50 K 50 K 50 K 50 K 0 0 0 0 3 –103 0 103 104 105 3 –10 0 103 104 105 –10 0 103 104 105 3 –10 0 103 104 105 CD20-BV785 b Patient #52 Patient #53 Patient #59 0.25 0.25 0.25 % CD20dim in CD4+ T cell % CD20dim in CD4+ T cell % CD20dim in CD4+ T cell 0.20 0.20 0.20 0.15 0.15 0.15 0.10 0.10 0.10 0.05 0.05 0.05 0.00 0.00 0.00 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 Days Days Days C Patient #52 Patient #53 Patient #59 HIV-RNA copies/10*6 CD4+ T cells HIV-RNA copies/10*6 CD4+ T cells HIV-RNA copies/10*6 CD4+ T cells 4000 4000 4000 3000 3000 3000 2000 2000 2000 1000 1000 1000 0 0 0 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 Days Days Days Fig. 7 Effects of Rituximab treatment in CD20dim CD4+ T cells and intracellular HIV RNA. Three ART-suppressed patients were treated with Rituximab at indicated times (arrows) and blood samples were longitudinally collected. a Flow plots, following the same gating strategy as shown in Fig. 1a, of CD20 expression in CD4+ T cells in patient #53 are shown. b Frequencies of CD20dim CD4+ T cells in longitudinal samples from three patients are shown. c Copies of HIV RNA per million CD4+ T cells after Rituximab treatment. Individuals #52, #53, #59 are represented. Data underlying this Figure are provided as Source Data file deplete the latent viral reservoir. We observed that CD20dim CD4+ or its cellular origin might be different. To date, the mechanism T cells were not particularly enriched in HIV DNA, compared and function of these two markers in transcriptionally active with levels reported in other memory subsets27,28. However, we HIV-infected cells remain unknown and merit further observed a significantly higher proportion of these cells expres- investigation. sing HIV RNA transcripts with a significant contribution to the In our in vitro model of HIV infection we showed that HIV total pool of HIV RNA-expressing cells. This result concur with replication increased the expression of CD20. Although only 30% HIV-RNA expression by other memory pools4 and with the of HIV-infected cells expressed CD20, a clear association was higher activation status of CD20dim cells. Other markers for observed between infection and proportion of CD20. For instance, transcriptionally active HIV-infected cells have been described we observed that CD20 upregulation was stronger during the early recently: CD3217,29,30 and CD3031. A common link between three events of HIV infection. Importantly, this result might explain the markers, CD20, CD30 and CD32, is their relationship with lack of CD20 expression in all HIV-infected cells analyzed. Based increased levels of cell activation. However, we found that a on these results, we conclude that CD20 expression is specially minority of HIV-infected cells co-expressed CD20 and CD32 in upregulated during HIV infection, but also it might slightly blood. Thus, although CD20 and CD32 were both upregulated in increase upon cell activation. We present several evidences sup- HIV-infected cells, we speculate that the expression kinetics and/ porting this argument: (i) we demonstrated that stimulation of 10 NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 ARTICLE a 250 k 250 k 250 k 5 10 105 99.8 200 k 200 k 200 k 99.9 HLA-DR-BV711 HLA-DR-BV711 4 10 104 150 k 150 k 94.6 150 k 100 k 100 k 100 k 103 103 FSC-H SSC-A SSC-A 0 0 50 k 50 k 50 k 99.7 98.2 –10 3 –10 3 0 0 0 0 50 k 100 k 150 k 200 k 250 k 0 50 k 100 k 150 k 200 k 250 k 0 50 k 100 k 150 k 200 k 250 k –103 0 103 104 105 0 103 104 FSC-A FSC-A SSC-H CD20-BV785 CD32-PE-Cy7 250 k 250 k 105 200 k 200 k 104 92.7 150 k 10.0 150 k CD3-AF700 103 100 k 100 k 0.032 SSC-A SSC-A 0 50 k 50 k –103 0 0 3 –10 0 103 104 105 –10 3 0 103 104 105 0 102 103 104 105 CD20-BV785 HIV-RNA-AF647 Viability b 0.015 c f 0.015 0.039 0.015 0.008 0.048 0.078 0.054 0.039 10 % HIV RNA+CD20dim in CD4+ T cells depletion by RTX in CD4 T cells 0.20 0.031 0.008 0.008 0.006 8 + 0.023 0.015 % HIV RNA+ cells 0.15 Fold change 0.004 6 0.10 4 0.002 0.05 2 0.00 0.000 0 CD20 + – + – + – + – Responders Non-responders m no G D iu M IN Io ed R A/ Medium RMD ING PMA/Iono M PM d 0.008 e 0.008 0.008 0.046 p = 0.037 25 0.008 0.008 8 Spearman r = 0.615 reactivation in CD4+ T cells 20 6 Fold change % CD20dim 15 0.008 0.008 4 10 2 5 0 0 HIV + – CD8+ + – CD8+ + – CD8+ + – CD8+ 0 2 4 6 8 CD4+ CD4+ CD4+ CD4+ Fold change depletion by RTX in CD4+ T cells Medium RMD ING PMA/Iono Fig. 8 Rituximab impacts pharmacologically HIV-reactivated cells. a–c Freshly isolated CD4+ cells from (#33–40) and CD8+ cells (#17–19) from n = 8 and n = 3 ART-suppressed patients, respectively, were treated with romidepsin (RMD), ingenol (ING), or PMA/ionomycin (PMA/Iono) for 22 h and subjected to the RNA FISH-flow assay for the detection of HIV RNA transcripts. a Representative flow gating strategy used for the analysis. Proportions of cells expressing HIV RNA within CD20dim and CD20-negative CD4+ T cells (b) or cells expressing HIV RNA and CD20dim in CD4+ T cells (c) are shown. Medians and min–max rank are represented. The median percentages of CD20dim in infected and uninfected CD4+ T cells, and CD8+ T cells after treatment with different LRAs are shown in (d). Comparisons were performed using the Wilcoxon test. e, f Unfractionated PBMCs from n = 12 ART- suppressed patients (#18, 19, 21, 23, 41–48) were treated with the latent reversal agents (LRAs) romidepsin or panobinostat for 24 h, and with Rituximab for 48 h. HIV RNA expression were measured by qPCR. Panel (e) shows the correlation between the fold-change reactivation by the LRA and the fold- change cell depletion by Rituximab. Spearman’s nonparametric correlation coefficient and associated p value are shown. Panel (f) shows the fold-change in cell depletion by Rituximab in patients, stratified by responders (fold-change reactivation by LRA > 1) and non-responders (fold-change reactivation by LRA < 1). The Mann–Whitney test was used for statistical analyses in all panels. Median and min–max ranks are represented in all panels. Data underlying this Figure are provided as Source Data file NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 uninfected donors with anti-CD3 and anti-CD28 antibodies slightly regimen, which might help to explain the sustained reduction of promotes CD20 expression in uninfected CD4+ T cells (up to 4%) HIV RNA observed in this patient. (Supplementary Fig. 5b); (ii) LRA treatment slightly increases the Furthermore, our ex vivo data indicate that viral reactivation of expression of CD20 in HIV-negative cells (up to 5%), but the latent cell reservoirs increased the levels of cells co-expressing upregulation is stronger in HIV+ cells (up to 20%) (Fig. 8d); and HIV RNA and CD20, and consequently, the addition of Ritux- (iii) ex vivo HIV infection of FACS-sorted CD4+CD20− negative imab induced depletion of viral-reactivated cells. Based on this cells strongly upregulates CD20 (~30%) (Supplementary Fig. 3). data, we speculate that a large in vivo reactivation of the latent Thus, it is tempting to speculate that viral proteins, i.e., Nef, which viral reservoir, i.e., using LRAs, may be an indispensable pre- has been shown to modulate T-cell activation32, may be involved in requisite for Rituximab to significantly impact long-term HIV this process. More mechanistic studies are needed to finally eluci- persistence in vivo. Although it is important to consider that a date the real contribution of the virus to CD20 upregulation. In possible complication of current anti-CD20 therapy could be concordance with this idea, we found higher expression of CD20 in immunodeficiency and/or immunopathologies, we provide evi- HIV-infected cells in lymph nodes, a key sanctuary viral reser- dence for the combined use of LRAs and anti-CD20 therapies for voir33–36. Thus, it is tempting to speculate that the infected cells the depletion of viral-reactivated cells in patients on ART. The expressing CD20 that we observed in lymph nodes are TFH, a future development of new anti-CD20 therapies with less toxicity subpopulation of cells found within B-cell follicles with elevated and specifically directed to CD4+ T cells might help to implement expression levels of PD-1 and CXCR5, which are highly permissive this new therapeutic approach. to HIV infection and significant contributors to HIV persistence during treatment37. In agreement with that, we observed higher Methods proportions of CD20dim CD4+ T cells expressing PD-1 and Ethics statement. PBMCs from HIV-1-infected patients were obtained from the CXCR5 in blood compared with cells lacking CD20 expression, and HIV unit of the Hospital Universitari Vall d’Hebron and from the Clinic Hospital follicular T cells expressing CD20 in lymph nodes. This may sug- in Barcelona, Spain. Study protocols were approved by the Comitè d’Ètica d’In- vestigació Clínica (Institutional Review Board numbers 39–2016, 196–2015, and gest that the origin of this population may be the lymph nodes. If 476–2018) of the Hospital Universitari Vall d’Hebron and the Hospital Clinic in this hypothesis is fully confirmed in additional studies, targeting Barcelona, Spain. Samples were obtained from adults, all of whom provided written this subpopulation of HIV-infected cells will be fundamental to informed consent, and were prospectively collected and cryopreserved in the impact HIV persistence. Biobanc (register number C.0003590). All samples received were completely Critically, we found that Rituximab was able to target CD4+ anonymous and untraceable. T cells expressing dim levels of CD20 and induced ADCC Study samples. Samples from HIV-1-infected patients with CD4+ T-cell counts mediated by NK cells. A number of studies have also demon- higher than 100 cells/mm3 recruited in the HIV unit of the Hospital Universitari strated the effects of anti-CD20 therapy on the T-cell Vall d’Hebron and the Clinic Hospital in Barcelona, Spain, were included in this compartment9,38–40. Importantly, using our in vitro HIV infec- study. Information on plasma viral loads, CD4+ T-cell counts, and time on ART tion method, we observed that HIV-infected cells expressing from suppressed patients is summarized in Supplementary Table 1. CD20 were successfully eliminated by Rituximab. This finding suggests that Rituximab treatment during HIV infection might Cells and virus. PBMCs were obtained from HIV-infected patients by Ficoll-Paque have a positive effect on controlling viral load. In this regard, in density gradient centrifugation and cryopreserved in liquid nitrogen. PBMCs from uninfected donors were obtained anonymously from the BST (Banc de Sang i untreated SIV-infected rhesus monkeys, Rituximab administra- Teixits, Barcelona, Spain) and isolated as described above. PBMCs were cultured in tion prior to and during SIV infection was associated with RPMI medium (Gibco) supplemented with 10% Fetal Bovine Serum (Gibco), decreased viral load. Interestingly, during chronic follow-up 100 µg/ml streptomycin (Fisher Scientific), and 100 U/ml penicillin (Fisher Sci- infection, the viral load was undetectable in six out of seven entific) (R10 medium). The plasmid encoding HIV-1 strain NL4.3 (pNL4.3) was obtained through the animals treated with Rituximab41. However, in another study NIH AIDS Reagent Program from Malcom Martin. Viral stocks were generated by with African green monkeys, depletion of CD20+ T cells had no transfection of 293T cells (ATCC, CRL-3216) with pNL4.3, and the resulting viral significant impact on viral load42. In humans, a case report particles were titrated in TZMbl cells (NIH AIDS Reagent Program) using an described the impact of Rituximab alone on HIV viremia; HIV enzyme luminescence assay (britelite plus kit; PerkinElmer) as described plasma viral load was maintained while on ART and Rituximab previously47. treatment. After therapy discontinuation, however, the HIV-viral load was significantly increased, most likely due to a loss of Phenotyping of CD4+ T cells. To identify CD4+ T-cell subsets and the expression of CD20dim cells, PBMCs were stained for cell surface markers in a final volume of antibody production43. 100 µl with CD3 (PE-Cy7, 2 µl, BD 557851), CD4 (AF700, 1.25 µl, BD 557922), The use of anti-CD20 therapy to reduce viral reservoirs in B- CD20 (BV785, 2.5 µl, Biolegend 302356), HLA-DR (PE, 2.5 µl, Biolegend 63-9956- cell follicles had been previously considered44; the destruction of 41), CD25 (PE, 1 µl, Biolegend 356103), CD69 (PE, 2.5 µl, Biolegend 310905), B-cell follicles using anti-CD20 monoclonal antibodies has been CD27 (FITC, 4 µl, BD 555440), CD45RO (BV605, 1.2 µl, Biolegend 304238), CCR7 (BV711, 1 µl, BD 566602 or PE-CF594, 1 µl, BD 150503) and CD95 (PE-Cy5, 10 µl, proposed as a novel strategy to expose HIV-1 reservoirs for BD 559773) antibodies. Different CD4+ T-cell subset phenotypes were identified as elimination. For instance, in a recent study, HIV-infected patients follows: Naive (NA): CD3+, CD4+, CCR7+, CD45RO−, CD27+, CD95−; stem cell with lymphoma or Kaposi Sarcoma received multiple doses of memory (SCM): CD3+, CD4+, CCR7+, CD45RO−, CD27+, CD95+; central Rituximab, but reductions in cell-associated HIV RNA or HIV memory (CM): CD3+, CD4+, CCR7+, CD45RO+; transitional memory (TM): DNA levels were not observed45. Of note, these patients were CD3+, CD4+, CCR7−, CD45RO+, CD27+; effector memory (EM): CD3+, CD4+, CCR7−, CD45RO+, CD27-, and terminal differentiated (TD): CD3+, CD4+, treated with a combination of different antineoplastic drugs, CCR7−, CD45RO−. To identify CXCR5 and PD-1 expression in CD4+ CD20dim which included Rituximab. In our study, in two patients treated T cells, PBMCs were surface stained in 100 µl with CD3 (PE-Cy5, 0.6 µl, Biolegend only with Rituximab for non-malignant reasons we observed 300410), CD4 (AF700, 1.25 µl, BD 557922), CD20 (BV785, 2.5 µl, Biolegend more than twofold reductions in HIV RNA. Note that, in spite of 302356), CD19 (FITC, 2.5 µl, BD 560994), CXCR5 (PE, 2 µl, Biolegend 356903), and PD-1 (PE-Cy7, 0.6 µl, BD 561272) antibodies. Cell viability was determined cellular associated HIV RNA might naturally fluctuate during with a violet viability dye for flow cytometry (LIVE/DEAD Fixable Violet Dead Cell time in ART-treated patients, only 5% of patients are likely to Stain Kit; Invitrogen L34963). All samples were acquired with a LSRFortessa flow present a random twofold variation in HIV-RNA levels46. cytometer (Becton Dickinson), and the results were analyzed with FlowJo Ongoing viral replication in particular tissues, i.e., lymph nodes, v10 software (Tree Star). For fluorescent flow cytometry imaging analyses, PBMCs were stained in 100 µl potentially impacted by Rituximab might help to explain the with CD3 (PE-Cy7, 2 µl, BD 557851), CD4 (APC, 5 µl, ThermoFisher 17-0048-42), reduction in HIV RNA. Patient #53 was under a dual anti- CD20 (FITC, 2 µl, Biolegend 302303), CD19 (PE, 0.6 µl, Biolegend 302207), and retroviral therapy, instead of the recommended triple therapy viability dye (LIVE/DEAD Fixable Violet Dead Cell Stain Kit, Invitrogen L34963). 12 NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 ARTICLE Cells were acquired with an AMNIS ImageStremX imaging flow cytometer secondary antibody: Alexa Fluor 546 donkey anti-goat (Invitrogen, 712-586-150) (Merck), and data was analyzed using IDEAS v6.1 software. Only focused cells were and Alexa Fluor 647 donkey anti-mouse (Invitrogen, A-31571), counterstained considered for analysis, gradient RMS value > 40 was established as a threshold for with DAPI (4′,6-diamidino-2-phenylindole-dilactate, ThermoFisher), and moun- best focus. ted with Fluoromount G (eBioscience). Quantification of HIV DNA and HIV RNA by quantitative PCR. In accordance Detection of RNA in lymph nodes by in situ hybridization. We used the with the manufacturer’s instructions, CD4+ T lymphocytes were enriched from ultrasensitive ViewRNA ISH Tissue 2-Plex Assay Kit (eBioscience) to detect HIV total PBMCs by negative selection (MagniSort Human CD4+ T Cell Enrichment, RNA and CD20-mRNA in lymph node tissue samples. Briefly, tissue was fixed and eBioscience) when appropiate. CD4+ T cells were subjected to RNA extraction embedded in paraffin. Sections were mounted on Superfrost Plus microscope slides using the mirVana™ miRNA Isolation Kit following the protocol instructions (Fisher Scientifics). Before the assay was performed, samples were dewaxed in (Ambion). Reverse transcription of RNA to cDNA was performed with Super- Xylene and dehydrated with ethanol. Sample pretreatment was performed by ScriptIII Reverse Transcriptase (Invitrogen) with random primers, and cDNA was boiling sections in pretreatment solution for 10 min and protease digestion for quantified by qPCR with the following primers and probe: LTR R: 5′ 20 min at 40 °C. Hybridization was carried out by slide incubation with target GTTCGGGCGCCACTGCTAG 3′; LTR F: 5′ TTAAGCCTCAATAAAGCTTGCC probes for 2 h at 40 °C. Next day, signal amplification was performed by sequential 3′ and LTR Probe: 5′ /56-FAM/ CCAGAGTCA/ZEN/CACAACAGACGGGCA/ incubation of Pre-Amplifiers, Amplifiers and label probes. Finally, samples were 31ABkFQ/ 3′. Quantification of RNA copies was performed using a standard curve, counterstained with DAPI. and values were normalized to 1 million CD4+ T cells. In addition, CD4+ T cells were used to quantify total HIV DNA with the same primers and probe detailed Confocal microscopy and quantification. Samples were imaged on an Olympus above. CD4+ T cells were immediately lysed with proteinase K-containing lysis Spectral Confocal Microscope FV1000 using a ×20 and ×40 phase objective and buffer, and HIV DNA was quantified. Contribution of each subset to the total HIV- sequential mode to separately capture the fluorescence from the different fluor- reservoir was calculated by considering the frequency of each subset (analyzed by ochromes at an image resolution of 800 × 800 pixels. ImageJ software was used to flow cytometry) within the total CD4 compartment and their relative infection perform image processing and analysis. For CD20/CD3 colocalization analysis, a frequency. binary mask was generated for each channel, and the “analyze particles” tool was used to count single and double-stained cells. In all experiments, at least 10 B-cell Detection of HIV RNA+ cells by the RNA FISH-flow assay. The RNA FISH-flow follicles were analyzed per patient. assay was performed according to manufacturer’s instructions (Human Primer- Flow RNA Assay, eBioscience) with some modifications. After antibody staining Ex vivo infection of unstimulated PBMCs. PBMCs from uninfected donors were and cell fixation and permeabilization, cells were ready for hybridization with a set thawed and incubated overnight in R10 with 40 U/ml interleukin-2 (IL-2). The of 50 probes spanning the whole Gag-Pol HIV mRNA sequence (bases 1165 to next day, PBMCs were incubated for 4 h at 37 °C with 350,000 TCID50 (50% tissue 4402 of the HXB2 consensus genome). Next, the cells were subjected to amplifi- culture infectious dose) of NL4.3 viral strain. Cells were then thoroughly washed cation signal steps, and HIV RNA was detected using Alexa Fluor 647-labeled and cultured in 96-well plates with R10 containing 100 U/ml IL-2 for the next probes. As a negative control, non-HIV-infected donors were included in each 6 days. To follow HIV RNA expression, on days 3, 4, and 5, at least 5 million cells experiment; to normalize the data, the percentage of HIV RNA+ cells detected in were collected and subjected to the RNA FISH-flow assay. Target-specific Alexa donor samples was subtracted to the real values obtained in the HIV-infected Fluor 647-labeled probes were used to detect HIV-1-RNA; the expression of Gag samples. In these experiments, the following antibody panel was used in 100 µl: p24 viral protein was determined using a p24-PE antibody (Beckman Coulter). CD3 (PE-Cy5, 0.6 µl, BD 300410), CD4 (BV510, 1.25 µl, BD 344633), CD20 Surface staining was performed in 100 µl with CD4 (BV510, 1.25 µl, Biolegend (BV785, 2.5 µl, Biolegend 302356), HLA-DR (PE, 2.5 µl, Biolegend 307605), CD25 344633), CD3 (PE-Cy5, 0.6 µl, Biolegend 300410), CD20 (BV785, 2.5 µl, Biolegend (PE, 1 µl, Biolegend 356103), and CD69 (PE, 2.5 µl, Biolegend 310905). Cell via- 302356), HLA-DR (Pe-Dazzle 594, 2.5 µl, BioLegend 307653), and PD-1 (BV605, bility was analyzed with a violet viability dye (Invitrogen, L34963). Samples were 2.5 µl, BD 563245). analyzed with the LSRFortessa flow cytometer, and the results analyzed with In the experiments with Rituximab treatment during ex vivo infection, at 4 h FlowJo v10 software. after infection cells were cultured with R10 supplemented with 100 U/ml IL-2 and 10 µg/ml Rituximab, or 10 µg/ml control antibody. At day 3, Rituximab (10 µg/ml) Fluorescence activated cell sorting of CD4+ T cells. For the sorting experi- and the control antibody (10 µg/ml) were replaced. Infection was analyzed by p24- ments, 100 million PBMCs surface stained in 100 µl with the following: CD3 PE intracellular staining at days 3, 4, and 5. Surface staining was performed in (PerCP, 5 µl, BD 340663), CD4 (AF700, 1.25 µl, BD 557922), CD20 (FITC, 2 µl, 100 µl with CD3 (PE-Cy7, 2.5 µl, BD 557851), CD4 (AF700, 1.25 µl, BD 557922), Biolegend 302303), HLA-DR (PE, 2.5 µl, Biolegened 307605), CD69 (PE, 2.5 µl, CD20 (BV785, 2.5 µl, Biolegend 302356), and CD19 (FITC, 2.5 µl, BD 560994). All Biolegend 310905), and CD25 (PE, 1 µl, Biolegend 356103) or CD3 (PE-Cy7, 2 µl, samples were acquired on a LSRFortessa and analyzed with FlowJo v10 software. BD 557851), CD4 (AF700, 1.25 µl, BD 557922), CD20 (FITC, 2 µl, Biolegend 302303), and CD19 (PE, 0.6 µl, Biolegend 302207) antibodies, and both stained Ex vivo infection of CD20− CD4+ T cells. Seventy million PBMCs were used for with a violet viability dye (Invitrogen, L34963). Cells were immediately sorted using cell sorting experiments. Briefly, cells were stained in 100 µl with antibodies a BD FACSAria Cell Sorter (Flow Cytometry Platform, Institut d’Investigació en recognizing CD3 (PE-Cy7, 2.5 µl, BD 557851), CD4 (AF700, 1.25 µl, BD 557922), Ciències de la Salut Germans Trias i Pujol, IGTP). For HIV DNA measurements, and CD20 (FITC, 2 µl, Biolegend 302303). A violet dye was used as a viability different sorted population were defined as follows: total CD4+: CD3+, CD4+, marker (Invitrogen, L34963). The population CD20− CD4+ CD3+ was imme- CD20−; CD4+ activated: CD3+, CD4+, CD20−, HLA-DR+, CD69+, CD25+; CD4+ diately sorted using a BD FACSAria Cell Sorter. Cells were incubated overnight in resting: CD3+, CD4+, CD20−, HLA-DR−, CD69−, CD25−; CD20 activated: R10 supplemented with 100 U/ml of IL-2, and the day after cells were infected CD3+, CD4+, CD20dim, HLA-DR+, CD69+, CD25+; CD20 resting: CD3+, CD4+, with 350,000 TCID50 of the NL4.3 viral strain by spinoculation (2 h, 1200 g at CD20dim, HLA-DR−, CD69−, CD25−. After cell isolation, cells were subjected to 37 °C). Cells were thoroughly washed and cultured in 96-well plates with R10 HIV DNA quantification by qPCR as described above. For CD20-mRNA mea- containing 40 U/ml IL-2 for the next 3 days. On days 0 and 3 cells were collected surement, the following populations were isolated: CD4+ CD20− T cells: CD3+, and subjected to flow cytometry and total RNA extraction. Upregulation of CD20 CD4+, CD20−; CD4+ CD20dim T cells: CD3+, CD4+, CD20dim; B cells: CD3−, expression was identified by flow cytometry using the antibodies in 100 µl CD3 CD4−; CD20+, CD19+. Total RNA was extracted and cDNA was generated as (PE-Cy7, 2.5 µl, BD 557851), CD4 (AF700, 1.25 µl, BD 557922), and CD20 described above. Levels of CD20-mRNA were relatively quantified using the β-actin (BV785, 2.5 µl, Biolegend 302356). Levels of CD20-mRNA were relatively reference gene (PerfeCTa® qPCR FastMix II®, ROX, Quantabio) and applying the quantified using the β-actin reference gene (PerfeCTa® qPCR FastMix II®, ROX, 2−ΔΔCT method. Quantabio) and applying the 2−ΔΔCT method. Immunohistochemistry of human lymph nodes. For immunodetection of CD20, Quantification of cell death after Rituximab treatment. PBMCs from uninfected CD3, and p24, formalin-fixed and paraffin-embedded (FFPE) sections from human donors and ART-suppressed HIV-infected individuals were cultured with R10 in lymph nodes of HIV-infected patients were dewaxed and placed in decreasing the presence of 10 µg/ml Rituximab or the control antibody Palivizumab for 48 h. ethanol concentrations. Heat-induced epitope retrieval was performed by auto- Subsequently, cells were processed for flow cytometry and stained with violet claving at 120 °C for 15 min in a citrate buffer pH 6 (Abcam). Then, the slides were viability dye (Invitrogen, L34963), Annexin V (PE, 2 µl in 100 µl, Biolegend permeabilized in 1X Tris-buffered saline (TBS) (Fisher scientific) with 0.1% Triton 640907), CD3 (PE-Cy7, 2.5 µl in 100 µl, BD 557851), CD4 (AF700, 1.25 µl in 100 µl, X-100 (Sigma-Aldrich) and 1% BSA (Sigma-Aldrich) for 10 min. Subsequently, BD 557922), and CD20, for which we used the combination of Rituximab and anti- blocking was added for 2 h with 1X TBS supplemented with 10% normal donkey human AlexaFluor488 (ThermoFisher, 31531). The analysis was performed using a serum (Jackson Immunoresearch) and 1% BSA. The slides were incubated with LSRFortessa cytometer. Antibody-dependent cell-mediated cytotoxicity (ADCC) primary antibodies overnight at 4 °C: anti-CD20 (goat-polyclonal anti-CD20 was measured as follows. Target CD4+ T cells were isolated from uninfected donor antibody, 1/100, Abcam ab194970), anti-CD3 (mouse-monoclonal anti-CD3 PBMCs as described above. CD4+ T cells were stained with the cell proliferation antibody, 1/100, Leica Biosystems NCL-L-CD3-565), or anti-p24 (mouse-mono- dye eF670 following manufacturer’s instructions (Labclinics) and then incubated clonal anti-p24 antibody, 1/10, Dako-Agilent M0857), diluted in TBS 1 × −1% with 10 µg/ml of Rituximab for 30 min; cells treated with with the control antibody BSA. Next, samples were washed and incubated for 1 h with the appropriate Palivizumab were used as a negative control. Target cells were plated in U-bottom NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications 13 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 96-well plates (5.000 cells/well, 20 wells per condition). In parallel, NK cells were 5. Pasternak, A. O., Lukashov, V. V. & Berkhout, B. Cell-associated HIV RNA: a isolated from PBMCs from uninfected donors using a commercial kit (MagniSort™ dynamic biomarker of viral persistence. Retrovirology 10, 41 (2013). Human NK cell Enrichment Kit, eBiosciences) and plated at 100.000 cells/well 6. Rasmussen, T. A. et al. Panobinostat, a histone deacetylase inhibitor, for (ratio target:effector of 1:20). After incubation, cells were stained with 10 µg/ml latent-virus reactivation in HIV-infected patients on suppressive antiretroviral Rituximab plus anti-human Alexa Fluor 488 and then analyzed using a LSRFor- therapy: a phase 1/2, single group, clinical trial. Lancet HIV 1, e13–e21 (2014). tessa cytometer. 7. Leth, S. et al. Combined effect of Vacc-4x, recombinant human granulocyte macrophage colony-stimulating factor vaccination, and romidepsin on the Ex vivo viral reactivation of CD4+ T cells. CD4+ T lymphocytes were isolated as HIV-1 reservoir (REDUC): a single-arm, phase 1B/2A trial. Lancet HIV 3, described above and cultured in R10 alone or stimulated with romidepsin (40 nM), e463–e472 (2016). ingenol (100 nM), or PMA/ionomycin (PMA, 81 nM; ionomycin 1 µM) for 22 h in 8. Archin, N. M. et al. Administration of vorinostat disrupts HIV-1 latency in the presence of raltegravir (1 µM). Then, cells were washed and subjected to the patients on antiretroviral therapy. Nature 487, 482–485 (2012). RNA FISH-flow assay, as described above. For HIV RNA detection, we used target- 9. Wilk, E. et al. Depletion of functionally active CD20+ T cells by rituximab specific Alexa Fluor 647-labeled probes. Surface staining was performed in 100 µl treatment. Arthritis Rheum. 60, 3563–3571 (2009). with CD3 (AF700, 1 µl, Biolegend 300423), CD95 (PE-Cy5, 10 µl, BD 559773), 10. Forster, F., Singla, A., Arora, S. K., Schmidt, R. E. & Jacobs, R. CD20+ T cell HLA-DR (BV711, 0.6 µl, Biolegend 307644), CD20 (BV785, 2.5 µl, Biolegend numbers are decreased in untreated HIV-1 patients and recover after HAART. 302356), CCR7 (PE-CF594, 1 µl, BD 150503), p24 (PE, 5 µl, Beckman Coulter), Immunol. Lett. 146, 74–78 (2012). CD45RO (BV605, 1.25 µl, Biolegend 304238), and CD27 (FITC, 4 µl, BD 555440). 11. Palanichamy, A. et al. Rituximab efficiently depletes increased CD20- Samples were analyzed using a LSRFortessa flow cytometer. expressing T cells in multiple sclerosis patients. J. Immunol. 193, 580–586 (2014). Quantification of HIV RNA after viral reactivation. Total PBMCs from ART- 12. Hultin, L. E., Hausner, M. A., Hultin, P. M. & Giorgi, J. V. CD20 (pan-B cell) suppressed patients were incubated for 21 h with R10 containing raltegravir antigen is expressed at a low level on a subpopulation of human T (1 µM), nevirapine (100 nM), 10 µg/ml Rituximab or the control antibody Palivi- lymphocytes. Cytometry 14, 196–204 (1993). zumab, and the LRA romidepsin (40 nM) or panobinostat (30 nM). The LRA was 13. Edwards, J. C. & Cambridge, G. Sustained improvement in rheumatoid then extensively washed, and cells were cultured for an additional 24 h in R10 with arthritis following a protocol designed to deplete B lymphocytes. 10 µg/ml Rituximab or the control antibody. Controls consisting of a cell culture Rheumatology 40, 205–211 (2001). without the LRA and a culture with the control antibody Palivizumab were 14. Edwards, J. C. et al. Efficacy of B-cell-targeted therapy with rituximab in included for each sample tested. After incubation, CD4+ T cells were isolated by patients with rheumatoid arthritis. N. Engl. J. Med. 350, 2572–2581 (2004). negative selection as described above, and intracellular RNA was extracted and 15. Tenner-Racz, K. et al. The unenlarged lymph nodes of HIV-1-infected, quantified as described above. The fold-change reactivation was calculated by asymptomatic patients with high CD4 T cell counts are sites for virus dividing the quantification of HIV RNA (molecules of HIV RNA per million CD4+ replication and CD4 T cell proliferation. The impact of highly active cells) produced in the presence of romidepsin or panobinostat by the levels (or antiretroviral therapy. J. Exp. Med. 187, 949–959 (1998). quantification) of viral RNA without the addition of the LRA. The fold-change 16. Deleage, C. et al. Defining HIV and SIV reservoirs in lymphoid tissues. depletion of HIV RNA induced by Rituximab was calculated by dividing the levels Pathog. Immun. 1, 68–106 (2016). of viral RNA induced by romidepsin or panobinostat by the HIV RNA levels after 17. Abdel-Mohsen, M. et al. CD32 is expressed on cells with transcriptionally LRA and Rituximab incubation. active HIV but does not enrich for HIV DNA in resting T cells. Sci. Trans. Med. 10, eaar6759 (2018). Cell activation. PBMCs from uninfected donors were cultured in R10 plus 18. Maloney, D. G. et al. IDEC-C2B8 (Rituximab) anti-CD20 monoclonal 20 U/ml IL-2 with or without the activating antibodies anti-CD3 (1 µg/ml) and antibody therapy in patients with relapsed low-grade non-Hodgkin’s anti-CD28 (2 µg/ml). Expression levels of CD20, CD69, and HLA-DR were ana- lymphoma. Blood 90, 2188–2195 (1997). lyzed at 0, 24, 48, 72, and 96 h. At each time point, cells were stained with 100 µl 19. Gomez-Roman, V. R. et al. A simplified method for the rapid fluorometric CD4 (AF700, 1.25 µl, BD 557922), CD3 (PE-Cy7, 2.5 µl, BD 557851), CD20 (FITC, assessment of antibody-dependent cell-mediated cytotoxicity. J. Immunol. 2 µl, Biolegend 302303), HLA-DR (Pe-Dazzle 594, 2.5 µl, BioLegend 307653), Methods 308, 53–67 (2006). CD69 (PE, 2.5 µl, Biolegend 310905), and CD14 (APC, 5 µl, BD 561708), and 20. Richman, D. D. et al. The challenge of finding a cure for HIV infection. Science samples were analyzed using a LSRFortessa flow cytometer. 323, 1304–1307 (2009). 21. Trono, D. et al. HIV persistence and the prospect of long-term drug-free remissions for HIV-infected individuals. Science 329, 174–180 (2010). Statistical analyses. Statistical analyses were performed with Prism software, version 6.0 (GraphPad). Nonparametric Mann–Whitney U tests, Wilcoxon rank 22. Osuna, C. E. et al. Evidence that CD32a does not mark the HIV-1 latent and ANOVA Dunn’s multiple comparison tests were used as appropriate. For reservoir. Nature 561, E20–E28 (2018). correlations, Spearman’s correlation coefficient was calculated. A p value < 0.05 was 23. Orendi, J. M. et al. Activation and cell cycle antigens in CD4+ and CD8+ considered significant. T cells correlate with plasma human immunodeficiency virus (HIV-1) RNA level in HIV-1 infection. J. Infect. Dis. 178, 1279–1287 (1998). 24. Massanella, M., Fromentin, R. & Chomont, N. Residual inflammation and Reporting summary. Further information on research design is available in viral reservoirs: alliance against an HIV cure. Curr. Opin. HIV AIDS 11, the Nature Research Reporting Summary linked to this article. 234–241 (2016). 25. Schuh, E. et al. Features of human CD3+CD20+ T cells. J. Immunol. 197, Data availability 1111–1117 (2016). Data underlying figures are provided as Source Data files. All other data are available 26. de Bruyn, M. et al. CD20(+) T cells have a predominantly Tc1 effector from the corresponding author upon reasonable requests. memory phenotype and are expanded in the ascites of patients with ovarian cancer. Oncoimmunology 4, e999536 (2015). 27. Buzon, M. J. et al. HIV-1 persistence in CD4+ T cells with stem cell-like Received: 16 July 2018 Accepted: 19 July 2019 properties. Nat. Med. 20, 139–142 (2014). 28. Chomont, N. et al. HIV reservoir size and persistence are driven by T cell survival and homeostatic proliferation. Nat. Med. 15, 893–900 (2009). 29. Melanie Wittner, G. A. D. et al. CD32 expression of different memory T cell subpopulations in the blood and lymph nodal tissue of HIV patients and healthy controls correlates with immune activation. J. Acquir. Immune. Defic. References Syndr. 77, 345–349 (2018). 1. Palmer, S., Josefsson, L. & Coffin, J. M. HIV reservoirs and the possibility of a 30. Badia, R. et al. CD32 expression is associated to T-cell activation and is not a cure for HIV infection. J. Intern. Med. Suppl. 270, 550–560 (2011). marker of the HIV-1 reservoir. Nat. Commun. 9, 2739 (2018). 2. Coiras, M., Lopez-Huertas, M. R., Perez-Olmeda, M. & Alcami, J. 31. Hogan, L. E. et al. Increased HIV-1 transcriptional activity and infectious Understanding HIV-1 latency provides clues for the eradication of long-term burden in peripheral blood and gut-associated CD4+ T cells expressing CD30. reservoirs. Nat. Rev. Microbiol. 7, 798–812 (2009). PLoS. Pathog. 14, e1006856 (2018). 3. Yukl, S. A. et al. HIV latency in isolated patient CD4(+) T cells may be due to 32. Markle, T. J., Philip, M. & Brockman, M. A. HIV-1 Nef and T-cell activation: a blocks in HIV transcriptional elongation, completion, and splicing. Sci. Transl. history of contradictions. Future Virol. 8, 391–404 (2013). Med. 10, eaap9927 (2018). 33. Lorenzo-Redondo, R. et al. Persistent HIV-1 replication maintains the tissue 4. Grau-Expósito, J. S.-P. C. et al. A novel single-cell FISH-flow assay identifies reservoir during therapy. Nature 530, 51–56 (2016). effector memory CD4 T cells as a major niche for HIV-1 transcription in HIV- 34. Connick, E. et al. CTL fail to accumulate at sites of HIV-1 replication in infected patients. mBIO 8, e00876-17 (2017). lymphoid tissue. J. Immunol. 178, 6975–6983 (2007). 14 NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11556-4 ARTICLE 35. Folkvord, J. M., Armon, C. & Connick, E. Lymphoid follicles are sites of de Investigación en SIDA (RD16/0025/0007). M.B. is supported by the Miguel Servet heightened human immunodeficiency virus type 1 (HIV-1) replication and program funded by the Spanish Health Institute Carlos III (CP17/00179). M.G. is sup- reduced antiretroviral effector mechanisms. AIDS Res. Hum. Retrovir. 21, ported by the “Pla estrategic de recerca i innovacio en salut” (PERIS), from the Catalan 363–370 (2005). Government. The funders had no role in study design, data collection and analysis, the 36. Fukazawa, Y. et al. B cell follicle sanctuary permits persistent productive decision to publish, or preparation of the paper. simian immunodeficiency virus infection in elite controllers. Nat. Med. 21, 132–139 (2015). Author contributions 37. Perreau, M. et al. Follicular helper T cells serve as the major CD4 T cell M.B. designed, directed, and interpreted experiments. C.S.-P., J.G.-E., L.L.-B., A.A.-G., compartment for HIV-1 infection, replication, and production. J. Exp. Med. J.G.-R. and M.G. performed experiments, analyzed the data, and interpreted experiments. 210, 143–156 (2013). R.W., J.N., A.C., J.B., E.R., M.M., B.R., A.T., B.P., F.G., R.B., J.C. and V.F. were 38. Stroopinsky, D., Katz, T., Rowe, J. M., Melamed, D. & Avivi, I. Rituximab- responsible for recruitment, specimen handling and storage, and related clinical data induced direct inhibition of T-cell activation. Cancer Immunol., Immunother. collection. M.B. and C.S.-P. wrote the initial paper, and all of us contributed to editing of 61, 1233–1241 (2012). the paper. 39. Cross, A. H., Stark, J. L., Lauber, J., Ramsbottom, M. J. & Lyons, J. A. Rituximab reduces B cells and T cells in cerebrospinal fluid of multiple sclerosis patients. J. Neuroimmunol. 180, 63–70 (2006). Additional information 40. Piccio, L. et al. Changes in B- and T-lymphocyte and chemokine levels with Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- rituximab treatment in multiple sclerosis. Arch. Neurol. 67, 707–714 (2010). 019-11556-4. 41. Gaufin, T. et al. Limited ability of humoral immune responses in control of viremia during infection with SIVsmmD215 strain. Blood 113, 4250–4261 Competing interests: The authors declare no competing interests. This work was (2009). produced solely by the authors and that no other individuals or entities influenced any 42. Gaufin, T. et al. Effect of B-cell depletion on viral replication and clinical aspects of the work. outcome of simian immunodeficiency virus infection in a natural host. J. Virol. 83, 10347–10357 (2009). Reprints and permission information is available online at http://npg.nature.com/ 43. Huang, K. H. et al. B-cell depletion reveals a role for antibodies in the control reprintsandpermissions/ of chronic HIV-1 infection. Nat. Commun. 1, 102 (2010). 44. Bronnimann, M. P., Skinner, P. J. & Connick, E. The B-cell follicle in HIV Peer review information: Nature Communications thanks the anonymous reviewers for infection: barrier to a Cure. Front. Immunol. 9, 20 (2018). their contribution to the peer review of this work. Peer reviewer reports are available. 45. Henrich, T. J. et al. Human immunodeficiency virus type 1 persistence following systemic chemotherapy for malignancy. J. Infect. Dis. 216, 254–262 Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in (2017). published maps and institutional affiliations. 46. Leth, S. et al. HIV-1 transcriptional activity during frequent longitudinal sampling in aviremic patients on antiretroviral therapy. AIDS 30, 713–721 (2016). Open Access This article is licensed under a Creative Commons 47. Li, M. et al. Genetic and neutralization properties of subtype C human Attribution 4.0 International License, which permits use, sharing, immunodeficiency virus type 1 molecular env clones from acute and early adaptation, distribution and reproduction in any medium or format, as long as you give heterosexually acquired infections in Southern Africa. J. Virol. 80, appropriate credit to the original author(s) and the source, provide a link to the Creative 11776–11790 (2006). Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the Acknowledgements article’s Creative Commons license and your intended use is not permitted by statutory We are indebted to the Cytomics core facility of the IDIBAPS for their technical help regulation or exceeds the permitted use, you will need to obtain permission directly from with the Amnis® Flow cytometer. This study was supported by the American National the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Institutes of Health (grant R21AI118411 to M.B.), the Spanish Secretariat of Science and licenses/by/4.0/. Innovation and FEDER funds (grant SAF2015-67334-R [MINECO/FEDER]), the Spanish “Ministerio de Economia y Competitividad, Instituto de Salud Carlos III” (ISCIII, PI17/01470), GeSIDA and the Spanish AIDS network Red Temática Cooperativa © The Author(s) 2019 NATURE COMMUNICATIONS | (2019)10:3705 | https://doi.org/10.1038/s41467-019-11556-4 | www.nature.com/naturecommunications 15