Tonsils were obtained from children at low risk for HIV infection who had undergone routine tonsillectomy. Use of tonsil specimens for these studies was reviewed by the University of Arizona Institutional Review Board and determined to not constitute human subjects research, in accordance with guidelines issued by the Office of Human Research Protections and, consequently, informed consent was not required.

Archived inguinal lymph nodes from untreated people living with HIV without AIDS, obtained as previously described (33, 34), were used with permission by the University of Arizona’s IRB.

293T cells were cultured in DMEM (Gibco) supplemented with 10% FBS and Primocin (Invivogen) and were transfected with X4-tropic pNLENG1-IRES (32) or R5 tropic pNLYUV3-GFP (35) virus plasmid using Lipofectamine 2000 (ThermoFisher). Supernatant was collected 2 days post transfection, centrifuged at 800 xg, and stored in 500 µl aliquots at -80°C prior to use.

Tonsil CD4⁺ T cells were enriched from disaggregated tonsil tissue using negative selection (STEMCELL Technologies). CD4⁺ T cells were labeled with CD3-AF700 (clone SP34-2, BD), CD8-BV510 (Clone RPA-T8, BioLegend), CXCR5-PE (Clone MU5UBEE, eBioscience), CD25-PECy7 (Clone BC96, Tonbo), and 7-AAD (Tonbo) and sorted using a FACSAriaIII cell sorter equipped with an 85 μm nozzle into TFR (CD3⁺CD8^-CXCR5⁺CD25⁺7-AAD^-) and TFH (CD3⁺CD8^-CXCR5⁺CD25^-7-AAD^-) populations. A minimum of 4 x 10⁶ TFH were spinoculated with X4-tropic or R5-tropic GFP reporter virus and labeled with violet proliferation dye 450 (VPD) (Becton Dickinson). Spinoculated VPD⁺TFH were cultured 1:1 with uninfected, unlabeled TFH or TFR at a final concentration of 2x10⁶ cells/ml for five days in Advanced RPMI (Gibco) supplemented with 10% heat inactivated FBS, 1X nonessential amino acids (Gibco), 1X Glutamax (Gibco), 1X Primocin (In vivogen), (Advanced R-10) in the presence of 5 μM saquinavir (Sigma). IL-2 (NIH AIDS reagent program) (36), IL-6 (PeproTech), and 10 μg/ml blocking antibodies to CD39 (Clone A1, Bio-Rad), CTLA-4 (Clone BE0099, BioXcell), IL-10 (Clone JES3-9D7, Biolengend), or TGF-β (Clone 19D8, Biolegend) were added where indicated.

For studies evaluating the impact of media on sex differences, cells were cultured in Advanced R-10 as indicated above with the following alterations: 10% charcoal-stripped FBS (Sigma), 1 μM estrogen receptor antagonist ICI 182,780 (Sigma), or 300 pg/ml β-estradiol (Sigma) where indicated.

Live VPD⁺TFH (VPD⁺7-AAD^-) were sorted after 5-day coculture with unlabeled, uninfected TFH or TFR. DNA was isolated using a Puregene kit (Qiagen). Cellular DNA was quantified by QPCR for cell equivalents of DNA per volume as described previously (37). Total and integrated DNA was quantified as described previously (38) using equal quantities of cellular DNA.

Cells were washed in PBS and resuspended in PBS and Ghost Dye Red 780 (Tonbo). After 2 minutes, antibodies in PBS, and where appropriate, Brilliant Violet Stain Buffer (BD) were added and incubated for 30 minutes in the dark at room temperature. Cells were washed in PBS and resuspended in 2% paraformaldehyde in PBS. A minimum of 2.55 x 10⁴ CountBright absolute counting beads (Molecular Probes) were added and used for cell count determination. For phenotyping, the following antibodies were used: CD3-AF700 (Clone SP34-2, BD), CD8-BV510 (Clone RPA-T8, BioLegend), PD-1-APC or BV785 (Clone EH12.1H7, BioLegend), CXCR5-PE (Clone MU5UBEE, eBioscience), CD25-PECy7 (Clone BC96, Tonbo), CTLA-4-PerCP-eFluor710 (Clone 14D3, ThermoFisher), GITR-PE-eFluor610 (Clone eBioAITR, eBioscience), LAG-3-Superbright 702 (Clone 3DS223H), and CD39-BV711 (Clone TU66, BD). Cells were analyzed on a Fortessa flow cytometer, and data analyzed using FloJo v10.

For intracellular staining, in place of 2% PFA, cells were fixed and permeabilized with FoxP3/Transcription factor staining buffer kit (Tonbo) according to manufacturer’s instructions. Labeling of intracellular FoxP3 and CTLA-4 was performed using the following antibodies: Foxp3-APC (Clone PHC101, eBioscience) and CTLA-4-PerCP-eFluor710 (Clone 14D3, ThermoFisher) for 30 minutes. Cells were washed, suspended in PBS and analyzed on a Fortessa flow cytometer and data analyzed using FloJo v10.

Cryopreserved tonsil cells were thawed and cultured in Advanced R-10 overnight. 2ml PBS was added and samples were centrifuged at 300xg for 10 minutes. Cell pellets were resuspended in Advanced R-10 supplemented with 50 ng/ml PMA (Sigma), 1 μg/ml ionomycin (Sigma), and monensin (eBioscience) and cultured 5 hours. Samples were washed and stained as indicated for extracellular proteins, then fixed and permeabilized with Cytofix/Cytoperm (BD) according to manufacturer’s instructions. Samples were stained for 30 minutes with either IL-10-AF647 (Clone JES3-9D7, Biolegend) or TGF-β (Clone TW4-2F8, Biolegend), and analyzed on a Fortessa flow cytometer. Gates for IL-10 and TGF- β were determined using unstimulated, monensin treated samples as controls.

On day 5 of infection, cell cultures were centrifuged at 800 g for 10 minutes and supernatant was collected and stored at -80°C. IL-2 ELISA (Sigma) was performed according to the manufacturer’s instructions using diluted supernatant. Samples were analyzed on a plate reader at 450 nm and IL-2 concentrations were calculated using IL-2 standards provided.

Six micron sections of frozen inguinal lymph nodes were thaw mounted onto glass slides and fixed in 4% paraformaldehyde for 30 min. After blocking with hydrogen peroxide, sections were stained for IL-2 and CD4 RNA using multiplexed fluorescent in situ hybridization (ACDbio). Following in situ hybridization, sections were incubated overnight with antibodies to FoxP3 (clone 236A/E7, Invitrogen) and either CD20 (Rabbit pAb, Abcam) or Ki67 (Rat clone OT1567, Abcam) on adjacent sections. Appropriate fluorescent secondary antibodies (Invitrogen) were added and all sections were counterstained with DAPI. Whole sections were imaged on an Aperio slide scanning system (Leica) at 40X.

Using Aperio Imagescope (v 12.4.0.5043, Leica), total, follicular (based on CD20 staining) and germinal center (based on Ki67 staining) areas were defined. IL-2 RNA positive cells were counted and categorized as extrafollicular, follicular, or germinal center IL-2 positive cells and their frequency determined by quantitative image analysis. Percentages of IL-2 positive cells that co-expressed CD4 were also determined. In follicles or germinal centers where there were IL-2+ cells, FoxP3+CD4+ cells were counted and the ratio of IL-2+ cells to FoxP3+CD4+ cells was determined. A final ratio was obtained by averaging data from each IL-2 containing follicle or germinal center for each lymph node.

To determine the impact of TFR on HIV replication in TFH, both populations were sorted from tonsils, TFH were spinoculated (39) with the CXCR4-tropic GFP reporter virus (19), labeled with violet proliferation dye (VPD), and cultured for five days with unlabeled TFR or unlabeled, uninfected TFH as a control. Cultures were performed in the presence of saquinavir, to prevent spreading infection, and low concentrations of IL-2 (10 IU/ml). A representative example of flow cytometry gating used for sorting TFH (CD3⁺CD8^-CXCR5⁺CD25^lo/-) and TFR (CD3⁺CD8^-CXCR5⁺CD25^hi) is shown in Figure 1A . Median purity of sorted populations was 98% for TFH (range 91-100%) and 94% for TFR (85-100%). Using this gating strategy, the median ratio of TFR : TFH was 1:7 (range 1:4 to 1:13; n=32). Approximately half of all TFR expressed FoxP3 (median, 51% FoxP3⁺) while very few TFH expressed FoxP3 (median, 1% FoxP3⁺) ( Figure 1B ), as reported previously (19, 30). Furthermore, CXCR5⁺FoxP3⁺ cells were predominantly classified as TFR (median, 74%) compared to TFH (median, 14%) ( Figures 1C, D ) thus validating this sorting strategy. After five days in culture, GFP expression was measured by flow cytometry in VPD⁺TFH as depicted in a representative flow plot ( Figure 1E ). Less than 0.1% of VPD- cells were GFP⁺, demonstrating the efficacy of saquinavir in preventing spreading infection. When compared to controls, TFR cultures reduced percentages of GFP⁺ VPD⁺TFH (median reduction, 36%) ( Figure 1F ) as well as median fluorescence intensity (MFI) of GFP (median reduction, 23%) ( Figure 1G ).

Additional experiments to evaluate the impact of TFR on CCR5-tropic HIV GFP reporter virus expression revealed a similar magnitude of TFR-mediated suppression of HIV replication as seen in CXCR4-tropic HIV GFP reporter virus infection ( Supplementary Figure 1 ). Thus, virus tropism did not appear to impact TFR-mediated suppression of HIV replication. Because of the more robust infection seen with CXCR4-tropic virus, subsequent experiments were performed using the CXCR4-tropic GFP reporter virus except when noted.

The defined TFH population consisted of both CXCR5^mid and CXCR5^hi cells. To determine if the impact of TFR on HIV replication in TFH was different between CXCR5^mid and CXCR5^hi populations, these subsets were sorted, spinoculated with CXCR4-tropic GFP reporter virus, labeled with VPD, and cultured five days with unlabeled control or TFR cells ( Supplementary Figure 2 ). Although percentages of GFP+ cells were elevated in CXCR5^hi TFH compared to CXCR5^mid TFH (p<0.0001), TFR reduced percentages of GFP+ cells similarly for both subsets (p=0.226). The TFH (CD3⁺CD8^-CXCR5⁺CD25^lo/-) population was thus used for all subsequent experiments.

TFR exert immunosuppressive effects through the expression of multiple regulatory molecules including CD39 and CTLA-4 (40), and cytokines including IL-10 and TGF-β (41). To determine whether TFR utilize these regulatory mechanisms to inhibit HIV replication in TFH, TFR were first analyzed for expression of these markers. At baseline, a greater percentage of TFR were CD39⁺ (median, 43%) compared to TFH (median, 9%) ( Figure 2A ). Nearly all TFR (median, 87%) and approximately half of TFH (median, 46%) expressed CTLA-4 intracellularly ( Figure 2B ). Some TFR expressed IL-10 (median 24%) upon stimulation with PMA/ionomycin in the presence of monensin, while most TFH did not (median, 1%) ( Figure 2C ). Little TGF-β was expressed by either TFR (median, 0.9%) or TFH (median, 0.2%) ( Figure 2D ). Despite expression of these regulatory molecules by TFR, blocking antibodies to CD39, CTLA-4, IL-10, or TGF- β did not reverse TFR-mediated reductions in percentages of GFP⁺ TFH ( Figure 2E ).

To determine whether TFR-mediated reductions in HIV GFP reporter virus expression were related to reductions in numbers of infected cells rather than inhibition of HIV transcription, total HIV DNA was quantified from six additional tonsils that were spinoculated and cultured for five days in the same way as described above. Live VPD⁺TFH were then isolated on a cell sorter, DNA was isolated, and HIV DNA measured using quantitative PCR (qPCR). As shown previously in Figure 1 , percentages of GFP⁺VPD⁺TFH ( Figure 3A ) and GFP-MFI ( Figure 3B ) were again consistently reduced in TFR cocultures compared to controls, with median reductions of 36% and 23%, respectively. Total HIV DNA per VPD⁺TFH was markedly decreased in TFR cocultures compared to control cultures with a median reduction of 61% ( Figure 3C ). In two additional experiments, reductions in integrated HIV DNA were observed in the presence of TFR and paralleled reductions seen in total HIV DNA ( Supplementary Figure 3 ). Absolute VPD+ cell counts were quantified in three experiments and revealed that viable VPD⁺TFH cells were consistently reduced in TFR cocultures compared to controls with reductions in viable cells ranging from 33-49% ( Figure 3D ). To determine if TFR suppressed TFH proliferation, VPD dye dilution was assessed. With the exception of one subject, few TFH proliferated and TFR did not inhibit TFH proliferation. Indeed, the percentages of cells that proliferated was statistically elevated in wells with TFR compared to controls, although the absolute differences were trivial (median, 0.5% compared to 0.3%) ( Figure 3E ), indicating that TFR-mediated reductions in HIV DNA and TFH numbers could not be accounted for by alterations in TFH proliferation. Collectively, these data indicate that TFR reduce viability of TFH in the context of ex vivo HIV infection and that HIV expressing cells and HIV DNA+ cells are preferentially targeted for depletion.

Because prior ex vivo studies indicated HIV infection activates TFR (30), TFR were assessed for their ability to reduce TFH viability in the presence or absence of HIV infection ( Figure 3F ). Fewer viable TFH were found in HIV-infected cultures compared to uninfected cultures. TFR reduced TFH viability in both HIV infected (median reduction, 26%; range, 24-30%) and uninfected cocultures (median reduction, 16%; range, 8-27%). These results indicated that TFR mediated reductions in TFH viability were not driven by HIV infection.

Females with HIV infection have lower levels of plasma HIV RNA than males (42–44) and we previously demonstrated that women have lower frequencies of HIV RNA+ cells in lymph nodes as well as produce lower amounts of plasma HIV RNA per infected lymph node cell compared to men (45). To evaluate whether sex differences in viral load could be mediated by sex differences in TFR activity, we sought to test whether there were sex differences in the magnitude of TFR suppression of HIV replication in the tonsil model. Sex differences were not observed in percentages of TFR (median, males 5.9%; females 5.9%; p=0.43) ( Supplementary Figure 4A ). To determine whether media supplemented with conventional bovine serum confounded results, TFH and TFR coculture experiments were performed in the following serum media conditions: conventional FBS, FBS supplemented with an estrogen receptor 1 inhibitor ICI (1μm), charcoal stripped FBS, or charcoal stripped FBS supplemented with β-estradiol (300pg/ml). No sex differences were detected in the TFR mediated reduction of GFP expression when conventional FBS was used (p=0.31). The addition of ICI (male, p>0.99; female, p>0.99), use of charcoal stripped media (male, p=0.13; female, p=0.79), or the addition of β-estradiol (male, p=0.08; female, p>0.99) did not affect the TFR mediated reduction in GFP expression when compared with conventional FBS ( Supplementary Figure 4B ). Next, we combined results from experiments in Figs 1-3 with additional experiments and analyzed them by sex. There were no significant sex differences in percentages of GFP+ TFH in the absence (p=0.89) or presence of TFR (p=0.98) ( Supplementary Figure 4C ), and the magnitude of reduction in percentages of GFP+ cells (median, males 35%, females 26%; p=0.26) did not differ significantly between the two sexes ( Supplementary Figure 4D ). Thus, there was no evidence for sex differences in TFR-mediated elimination of HIV-expressing cells.

T regulatory cells have been shown to reduce T cell viability through sequestration of IL-2 (46). To investigate whether TFR also possess this capability, we first determined the impact of IL-2 on GFP expression and TFH viability in the tonsil cell model. VPD-labeled TFH were spinoculated with CXCR4-tropic HIV GFP reporter virus and then cultured with uninfected, unlabeled TFH in the presence of varying concentrations of IL-2. GFP expression and TFH viability were measured after 5 days in culture ( Supplementary Figure 5 ) and presented as fold differences for each parameter compared to no IL-2 ( Figure 4 ). Approximately two-fold increases in percentages of GFP⁺ TFH and GFP-MFI were observed at IL-2 concentrations equal to or greater than 10 IU/ml ( Figures 4A, B ). Absolute numbers of TFH were also increased at these concentrations by approximately two-fold ( Figure 4C ). GFP+ cell viability, however, was more sensitive to exogenous IL-2 and numbers significantly increased at IL-2 concentrations equal to and greater than 3 IU/ml and were more than five-fold greater in number at the highest IL-2 concentration ( Figure 4D ). Similar patterns were observed in CCR5-tropic HIV GFP reporter virus infected TFH as in CXCR4-tropic GFP reporter virus infected TFH ( Supplementary Figure 6 ). Thus, IL-2 enhanced the magnitude of HIV replication in TFH, TFH survival, and particularly GFP⁺ TFH survival in both CXCR4-tropic and CCR5-tropic virus infection.

In a mouse model of influenza infection, germinal center TFH were found to be unresponsive to IL-2 through an IL-6 dependent mechanism (47). As the experiments reported here demonstrated that TFH are responsive to IL-2, but were performed in the absence of FDC, which are the major IL-6 producing cells (48), we sought to determine whether IL-6 would enhance HIV replication and TFH viability in IL-2 treated cultures, and negate TFR-mediated suppression of TFH viability and HIV replication. The addition of IL-6 to cultures did not alter the percentage of GFP+ cells ( Figure 5A ) in control wells. However, in control wells treated with IL-2 and IL-6 both GFP MFI ( Figure 5B ) and TFH cell count ( Figure 5C ) were elevated compared to IL-2 alone, suggesting a positive effect of IL-6 on these parameters. In cocultures of TFH and TFR supplemented with 10 IU/ml IL-2 in the presence or absence of 5 ng/ml IL-6, TFR-mediated reductions in percentages of GFP⁺ VPD+TFH were similar in cultures treated with IL-2 alone compared to those with IL-2 and IL-6 ( Figure 5A ). Similarly, TFR-induced reductions in GFP MFI, TFH cell count and GFP+ cell count were unaffected by the presence of IL-6 ( Figures 5B–D ). These results suggest that human TFH maintain responsiveness to IL-2 in the presence of IL-6, and human TFR exert their suppressive functions on TFH viability in the presence of IL-6.

To evaluate whether TFR-mediated reductions in HIV-producing TFH are related to IL-2, TFR and TFH co-culture experiments were repeated in the presence of varying concentrations of IL-2 ranging from 0 to 100 IU/ml IL-2 ( Figure 6 ). At concentrations of IL-2 of 10 IU/mL and 30 IU/mL, significant reductions in percentages of GFP+VPD+ TFH, MFI of GFP, and numbers of either VPD+TFH cells or VPD+GFP+ cells occurred in the presence of TFR ( Figures 6A–D ). In contrast, no alterations in these parameters were observed in the presence of TFR at concentrations of IL-2 of 100 IU/mL or when no IL-2 was added. Similar observations were seen in CCR5-tropic infection as in CXCR4-tropic infection ( Supplementary Figure 7 ).

TFR cell counts from these experiments were increased in conditions with exogenous IL-2 compared to conditions without IL-2 ( Figure 6E ). However, varying concentrations of IL-2 did not affect TFR cell counts, suggesting no dose effect on TFR. In separate experiments prepared as in Figure 6 , TFR expression of CD39, GITR, and LAG-3 was measured after five days of culture with varying concentrations of IL-2 ( Supplementary Figure 8 ). TFR expression of these effector molecules was lower in cells cultured without IL-2, but again there was no dose effect and no statistically significant differences in expression of these effector molecules among the varying concentrations of IL-2.

IL-2 concentrations were measured in supernatants from control and TFR-treated 5-day cocultures ( Figure 6F ). In the absence of exogenous IL-2, IL-2 concentrations were low in 3 and undetectable in 9 of 12 cocultures. IL-2 supernatant concentrations were significantly reduced in TFR cocultures compared to TFH control cocultures supplemented with 10 IU/ml IL-2 (median, TFR: 2.31 ng/ml; Control: 5.40 ng/ml) and 30 IU/ml IL-2 (median, TFR: 6.14 ng/ml; Control: 13.85 ng/ml). At high concentrations of IL-2, i.e., 100 IU/mL, there was no significant difference between TFR and control co-cultures in terms of IL-2 concentrations. Taken together, these results indicate that TFR reduce TFH viability and HIV replication by reducing extracellular IL-2 concentrations, and that this occurs in the context of low or moderate IL-2 concentrations, but not at high IL-2 concentrations.

To evaluate whether IL-2 expressing cells and TFR are found in the follicles and germinal centers of lymph nodes where the majority of TFH reside, inguinal lymph node sections of individuals with untreated chronic HIV infection without AIDS were evaluated for IL-2 and CD4 expression by in situ hybridization for mRNA, and FoxP3 by antibody staining ( Figure 7 ). All lymph node sections contained IL-2 expressing cells, and the majority (92%; range 79-97%) of IL-2+ cells were CD4+. No IL-2+FoxP3+ cells were detected. The majority (68%; range, 48-92%) of IL-2+ cells were found outside follicles and the fewest IL-2 positive cells were found in the germinal centers ( Figure 7A ). Frequencies of IL-2+ cells/mm² were slightly elevated in germinal centers compared to extrafollicular regions (p=0.0286); however, there was no statistically significant difference in frequencies of IL-2 positive cells between follicles and germinal centers, and follicles and extrafollicular areas ( Figure 7B ). We previously reported that FoxP3+CD4+ cells are most abundant in extrafollicular regions, intermediate in frequency in follicular regions, and least frequent in germinal centers of lymph nodes of people living with HIV (30). We confirmed again that FoxP3+CD4+ cell frequencies were higher in follicles compared to germinal centers (p=0.0039; Figure 7C ). The median ratio of FoxP3+CD4+ cells to IL-2+ cells was 28:1 (range, 11:1 to 36:1) in follicles and 10:1 (range, 5:1 to 13:1) in germinal centers ( Figure 7D ). FoxP3+CD4+ cells found within follicles and germinal centers were detected adjacent to IL-2+ cells in some instances ( Figure 7E ). Thus, IL-2 expressing cells exist within the follicular and germinal center milieu of lymph nodes in HIV infection with the potential to influence surrounding TFH, and even larger numbers of TFR compared to IL-2 producing cells are present with the capacity to reduce local concentrations of IL-2.