Aging impacts CD103+ CD8+ T cell presence and induction by dendritic cells in the genital tract

Summary As women age, susceptibility to systemic and genital infections increases. Tissue‐resident memory T cells (TRMs) are CD103+ CD8+ long‐lived lymphocytes that provide critical mucosal immune protection. Mucosal dendritic cells (DCs) are known to induce CD103 expression on CD8+ T cells. While CD103+ CD8+ T cells are found throughout the female reproductive tract (FRT), the extent to which aging impacts their presence and induction by DCs remains unknown. Using hysterectomy tissues, we found that endometrial CD103+ CD8+ T cells were increased in postmenopausal compared to premenopausal women. Endometrial DCs from postmenopausal women were significantly more effective at inducing CD103 expression on allogeneic naïve CD8+ T cells than DCs from premenopausal women; CD103 upregulation was mediated through membrane‐bound TGFβ signaling. In contrast, cervical CD103+ T cells and DC numbers declined in postmenopausal women with age. Decreases in DCs correlated with decreased CD103+ T cells in endocervix, but not ectocervix. Our findings demonstrate a previously unrecognized compartmentalization of TRMs in the FRT of postmenopausal women, with loss of TRMs and DCs in the cervix with aging, and increased TRMs and DC induction capacity in the endometrium. These findings are relevant to understanding immune protection in the FRT and to the design of vaccines for women of all ages.

implantation and pregnancy (Wira et al., 2015). In particular, T cells are tightly regulated in a compartmentalized manner to protect from infections without interfering with implantation (Erlebacher, 2013;Wira et al., 2015). However, after menopause, when pregnancy is no longer a primary function, endometrial T cells undergo phenotypical and functional changes, such as increased Th17 cell frequency or increased CD8 + T cell cytotoxic activity (Rodriguez-Garcia, Barr, Crist, Fahey & Wira, 2014;White et al., 1997).
T cells are critical components of the adaptive immune system that provide specific long-lasting protection against pathogens (Sallusto, Geginat & Lanzavecchia, 2004). Na€ ıve T cells generated in the thymus migrate to lymphoid tissues throughout the body. Following antigen recognition, na€ ıve T cells differentiate into effector and memory T cells. Tissue-resident memory T cells (TRMs) are a subset of memory cells that remain in tissues and do not recirculate (Gebhardt et al., 2009;Mueller & Mackay, 2016). CD8 + TRMs can be identified by the expression of CD69 and CD103, the aE portion of the aEb7 integrin that allows interactions with E-cadherin expressed on epithelial cells (Rosato, Beura & Masopust, 2017). CD103 + CD8 + T cells provide critical protection against viral infections due to privilege anatomical location within the epithelium, rapid local cytotoxic function, and induction of tissue antiviral state through the production of IFNc and other pro-inflammatory cytokines (Rosato et al., 2017;Schenkel et al., 2014).
In humans, memory T cells with TRM characteristics have been identified at multiple mucosal surfaces including the lung, the intestines, and the FRT (Thome & Farber, 2015;Thome et al., 2014). In the FRT, more than 95% of T cells are effector memory (Saba et al., 2010;Yeaman et al., 1997), with a high proportion of T cells expressing CD103 in the vagina, endocervix (CX), ectocervix (ECX), and endometrium (EM) (Duluc et al., 2013;Moylan et al., 2016;Rodriguez-Garcia et al., 2017). Along with resident T cells, mucosal surfaces are populated by multiple subsets of resident DCs critical for the induction and maintenance of T-cell responses (Schlitzer, McGovern & Ginhoux, 2015). DCs are known to be strongly influenced by the tissue environment, regarding their numbers, phenotype, and function (Schlitzer et al., 2015). We recently described a compartmentalized distribution of DCs in the FRT, with specific DC subsets restricted to the EM, while other subsets were distributed throughout the upper and lower FRT (Rodriguez-Garcia et al., 2017).
Studies with human DCs from the lung and vaginal mucosa demonstrate that certain DC subsets can induce CD103 + T cells with tissue-resident characteristics (Duluc et al., 2013;Yu et al., 2013), suggesting that tissue DCs have the ability to control mucosal accumulation of TRMs.
While it is known that TRMs are present in the FRT, the extent to which aging influences their distribution and regulation in different anatomical locations within the FRT is unknown. Moreover, it is also unknown whether DCs from FRT sites other than the vagina have the ability to induce TRMs, and how aging and menopause impact their function. Given the epidemiological evidence for increased STIs in older women and the recognition that the FRT is a mucosal site with restricted access to T cells (Nakanishi, Lu, Gerard & Iwasaki, 2009), understanding the distribution, dynamics, and mechanisms for induction of TRMs is critical to unravel the underlying mechanisms that regulate immune protection in women of all ages.
Here, we report that menopausal status and aging control CD103 + CD8 + T cell presence in a site-specific manner in the FRT, through the modulation of DCs in a TGFb signaling-dependent mechanism. Our findings have broad implications, since they are relevant for explaining the increased risk of infection (STIs) in older women, immunological control of gynecological cancers, and to more fully understand the immunology of pregnancy.  Figure S1a). Recognizing that resident T cells can be identified by the presence of CD69 and CD103 (Gebhardt et al., 2009), in preliminary studies we determined CD69 and CD103 expressions ( Figure S1b). CD69 was widely expressed on FRT T cells, while CD103 expression was more restricted. Since CD103 + T cells co-stained for CD69 (Figure S1b), consistent with previous reports (Posavad et al., 2017), we focused our studies on the CD103 + CD3 + T cell subset. Consistent with our previous observations (Rodriguez-Garcia et al., 2017), CD103 + T cells were present at the three sites analyzed with no significant differences between upper and lower FRT ( Figure 1b). To investigate potential changes after menopause in the percentage of CD3 + T cells that were CD103 + , results from Figure 1b were stratified according to menopausal status (characteristics of the patients shown in Table 1). The percentage of CD3 + CD103 + T cells was significantly increased in the EM from postmenopausal compared to premenopausal women (Figure 1c), with no changes observed in CX and ECX. These results are consistent with the CD103 mean fluorescence intensity (MFI) ( Figure S1c), showing no significant differences between tissues, but increased CD103 MFI in the EM from postmenopausal compared to premenopausal women. No differences were found between the proliferative or secretory phases of the menstrual cycle in the premenopausal group (not shown). This differential distribution between pre-and postmenopausal women represents a new example of site-specific changes in FRT T cells.
Since menopause is linked to the aging process, we asked whether age could influence CD103 + T cells, by calculating the correlation coefficient between age and the percentage of CD3 + T cells that express CD103 for each tissue. Recognizing that sex hormones regulate T cells in premenopausal women (Wira et al., 2015), only postmenopausal women were included in this analysis to eliminate the confounding effects of sex hormones. Unexpectedly, we found a negative correlation with aging in the CX and ECX (Figure 1d, top row), while the percentage of CD3 + CD103 + T cells remained unchanged in the EM. To determine whether this was a specific depletion of the CD103 + T cell subset, we also calculated the correlation coefficient between age and the percentage of total CD3 + T cells within the CD45 + immune cell population. As shown in Figure 1d (bottom row), no age-dependent changes were detected in any of the tissues for the total CD3 + T cell population. These findings suggest that after menopause, as women age, CD103 + T cell presence progressively decreases in the cervix (CX and ECX), but remains unaltered in the EM.
Since CD103 can be expressed on CD4 + and CD8 + T cells (Rosato et al., 2017), we investigated which T cell subpopulation was expressing CD103 in FRT tissues in a subset of the patients shown in Figure 1b (gating strategy shown in Figure S1a). The majority of T cells expressing CD103 were CD8 + T cells (Figure 1e    10% of CD103 + CD4 + T cells present. Importantly, menopausal status uniquely influenced CD103 + CD8 + T cells, but not CD103 + CD4 + T cells in the EM (Figure 1f). Consistent with the data presented in Figure 1c, no differences were found in CX and ECX between preand postmenopausal women for CD103 + CD8 + or CD103 + CD4 + T cells. it had no effect on overall T-cell proliferation. When SB431542 was present, proliferating CD103 + T cells were markedly reduced, but the CD103 À T cell population was increased (Figure 3a, right plot),

| Endometrial
suggesting that the induction of CD103 expression was selectively inhibited during the proliferation process. The significant reduction in CD103 + CD8 + T cells in the proliferating population observed after treatment with SB431542 ( Figure 3b) was not mimicked in CD4 + T cells, even though a small but significant increase in CD103-expressing cells was observed in proliferating CD4 + T cells (

| Menopausal status regulates induction of CD103 + T cells by endometrial DCs
Next, we asked whether menopausal status influences the ability of endometrial DCs to induce CD103 + T cells. For this purpose, we reanalyzed the data from Figure 2d based on menopausal status.
Since no differences were observed between CD1a + or CD14 + selected DCs, results from these two subsets were pooled to aging (gating strategy shown in Figure S1a). Figure 5a shows a significant progressive decrease in DC numbers as a function of age in the EM, CX, and ECX. Additionally, we found that DC number and CD103 + T cell percentage positively correlated in the CX (Figure 5b), but found no correlation in the EM or ECX. Recognizing that the decrease in total DC numbers could be a consequence of tissue atrophy with age, we quantified the absolute number of CD3 + T cells per gram of tissue, to understand whether decreases in cell numbers as a function of age is a general characteristic for all cell types in the FRT. As shown in Figure 5c, total numbers of CD3 + T cells were not affected by age, increasing the relevance of the decline in specific cell subsets.
These results indicate that aging is a critical regulator of DC number throughout the FRT and suggest that while DC number and CD103 + T cell percentage are associated in the CX, additional tissue-specific factors may be involved in the EM and ECX.

| DISCUSSION
Our study demonstrates the novel compartmentalization and regulation of CD103 + T cells and DCs in the FRT before and after meno-  Previous studies in the lung and the vagina have shown that subsets of DCs have the ability to induce CD103 expression on CD8 + T cells (Duluc et al., 2013;Yu et al., 2013), suggesting that CD103 induction is an "intrinsic" function of DCs (Yu et al., 2013). Our findings add to this information by demonstrating that endometrial DCs also induce CD103 expression on CD8 + T cells. However, an unexpected and novel observation from our study is that endometrial DCs' intrinsic ability to induce CD103 + CD8 + T cells is regulated by local changes in the FRT with menopause. The implications of these findings are that TRM presence in the EM before and after menopause may be controlled through the regulation of DC function. Importantly, while DC's ability to upregulate CD103 was affected by menopausal status, their ability to induce T-cell proliferation was not, suggesting a very selective regulation of DC function.
The selectivity of this mechanism is further supported by our previ- . Each dot represents a single patient; Spearman correlation extent to which changes in DC membrane-bound TGFb are involved in the differential ability of pre-and postmenopausal DCs to elicit CD103 expression on T cells. At this point, the mechanisms responsible for the selective TGFb signaling effects on CD8 + T cells but not on CD4 + T cells remain unclear; our findings of selective differential effects in the human FRT extend previously reported findings of selective effects on survival of CD8 + and CD4 + T cells in mice (McKarns & Schwartz, 2005). It also remains unclear whether the same mechanisms are responsible for the induction of CD103 + T cells in the CX and ECX. While we speculate the DCs from CX and ECX also induce CD103 expression on T cells, based on our preliminary results and previous publications of others with vaginal DCs (Duluc et al., 2013), whether upregulation at all sites in the FRT is mediated through TGFb signaling needs to be investigated, given the evidence in mice that nasal TRM induction was TGFb-independent, while lung TRM induction was TGFb-dependent (Pizzolla et al., 2017). Furthermore, whether DC function in the FRT is regulated in response to the presence or lack thereof of sex hormones and/or the local tissue environment remains to be determined.
Our findings also demonstrate that aging is a critical factor in determining DC numbers in the FRT. However, whereas declining DC numbers and CD103 + T cell percentage with aging correlated in the CX in postmenopausal women, no correlation was found in the ECX or the EM. These findings suggest that distinct tissue-specific factors regulate the presence and sustainability of CD103 + T cells in the different anatomical compartments of the FRT. Interestingly, the extracellular matrix has been described to change with age, including increased collagen stiffness, which may impair immune cell motility and survival (Moreau et al., 2017). While we observed a decrease in DCs and TRMs with age, no changes were observed for total T cell numbers per g of tissue, suggesting that specific immune cell subsets are more susceptible than others to age-related changes in the FRT.
It is possible that those cells that establish long-term residence in tissues, such as TRMs or DCs, are preferentially affected by agerelated changes in extracellular matrix composition with age. Beyond the potential role of the extracellular matrix, a recent study demonstrated that exogenous fatty acids were necessary for mice TRM survival in the skin (Pan et al., 2017). Future studies are needed to identify those factors, whether soluble or extracellular matrix related, that influence the human FRT tissue environment to preserve CD103 + T cell survival. Whether other immune cell types present in the FRT, such as NK cells or macrophages, also decrease with age remains to be elucidated.
Site-specific regulation of TRMs and DCs in the FRT is consistent with the contribution of each site to the reproductive function. We propose that, in premenopausal women, TRM presence is actively suppressed in the EM but not in the cervix, possible through a TGFb-dependent mechanism as a consequence of direct or indirect effects of sex hormones on endometrial DCs. Since the EM is the site for implantation, the regulation of TRMs with cytotoxic potential would be essential for successful pregnancy. In line with this hypothesis, a recent study found decreased expression of the tissue residency marker CD69 on CD8 + T cell populations in women with recurrent miscarriage when compared to controls (Southcombe et al., 2017). Whether implantation failure can be triggered by alterations in the resident T-cell populations identified in our study remains to be determined. In contrast to the EM, CD103 + T cells are abundant in the cervix from premenopausal women, consistent with higher exposure to pathogens and the need for protection from potential ascending pathogens (Posavad et al., 2017). Our findings are in agreement with our previous demonstration of increased CD8 + T cell cytotoxic activity in the cervix of premenopausal women when compared to the endometrium  and with mouse models demonstrating that cytotoxic T cells and DC function are regulated to prevent rejection during pregnancy (Erlebacher, 2013).
In postmenopausal women, we observed a progressive decline in TRMs in the cervix, possibly due to decreased numbers of DCs in the endocervix and/or unidentified tissue-specific factors in the ectocervix. In contrast to the cervix, TRMs increased in the EM following menopause, even though DC numbers declined with age. This could be due to an increased functional capacity of postmenopausal DCs as a consequence of the loss of suppressive mechanisms of DC function once pregnancy is no longer a primary function and possibly to offset the age-dependent loss of immune protection in the lower FRT. For example, others have shown that pelvic inflammatory disease, caused by ascending pathogens into the uterus and fallopian tubes, is rarely diagnosed in postmenopausal women (Jackson & Soper, 1999). Our results might offer a possible explanation for these observations, namely that TRM immune protection, possibly mediated by DCs, is enhanced in the uterus to counter any increase in upstream pathogens owing to altered cervical protection.

Our observation of lower numbers of CD103 + T cells in the CX
and ECX in older women correlates with epidemiological findings of increased susceptibility to certain STIs after menopause (Ghosh et al., 2013). Epidemiological studies demonstrate a second peak in the prevalence of human papillomavirus (HPV), a cause of cervical cancer, in women older than 45 years old (Gonzalez et al., 2010).
Since TRMs are implicated in HPV infection control (Cuburu et al., 2012;Gravitt, 2012), our finding of a decrease in cervical TRMs with aging suggests an explanation for increased HPV infection rates in older women. Our studies are consistent with our previous findings that CD8 + T cell cytotoxic activity in the EM is low in premenopausal women but markedly elevated following menopause   (Shin, Kumamoto, Gopinath & Iwasaki, 2016). Therefore, induced immune cell protection against STIs in humans would likely require a potent resident memory T-cell population (Iwasaki, 2010). Based on our findings, we speculate that despite vaccination, TRMs will not be equally induced and distributed throughout the FRT and will differ depending on women's age. For example, TRM endometrial protection might be impaired in younger women (premenopausal), while cervical protection will be suboptimal in older women (postmenopausal). As women age and remain sexually active, protection of the FRT at all ages is essential. Our findings are particularly relevant to HIV prevention, because most CD8 + T cell studies focus on the evaluation of blood T-cell responses, which do not correlate with the protective effects of resident T cells in the FRT (Shin et al., 2016;White et al., 2001).
In conclusion, we demonstrate that menopausal status and aging influence CD103 + CD8 + T cell induction and presence in the FRT.
Our findings offer a novel perspective to understanding immune pro-

| Tissue processing
Matched tissues from the endometrium (EM), endocervix (CX), and ectocervix (ECX) of the same patient were used whenever possible. In some cases, only endometrial tissue was provided by pathology. Vaginal tissues were not available. Tissues were processed to obtain a stromal cell suspension as described previously