Tissue-resident memory T (TRM) cells provide long-lasting immune protection. One of the key events controlling TRM cell development is the local retention of TRM cell precursors coupled to downregulation of molecules necessary for tissue exit. Sphingosine-1-phosphate receptor 5 (S1PR5) is a migratory receptor with an uncharted function in T cells. Here, we show that S1PR5 plays a critical role in T cell infiltration and emigration from peripheral organs, as well as being specifically downregulated in TRM cells. Consequentially, TRM cell development was selectively impaired upon ectopic expression of S1pr5, whereas loss of S1pr5 enhanced skin TRM cell formation by promoting peripheral T cell sequestration. Importantly, we found that T-bet and ZEB2 were required for S1pr5 induction and that local TGF-β signaling was necessary to promote coordinated Tbx21, Zeb2, and S1pr5 downregulation. Moreover, S1PR5-mediated control of tissue residency was conserved across innate and adaptive immune compartments. Together, these results identify the T-bet–ZEB2–S1PR5 axis as a previously unappreciated mechanism modulating the generation of tissue-resident lymphocytes.
Effective immune protection relies on the generation of long-lived memory T cells. During a typical immune response, naive T cells surveying lymphoid organs are activated by cognate antigen and differentiate into short-lived effector cells (SLECs) or long-lived memory T cells (Kaech and Cui, 2012; Omilusik and Goldrath, 2019). Following their activation, T cells exit lymphoid organs and infiltrate infected tissues where they can mediate pathogen clearance. Memory T cells persist after infection resolution to provide protection against subsequent encounters with the same pathogen. These memory cells can be partitioned into distinct subsets based on their migratory and functional properties. Nonrecirculating tissue-resident memory T cells (TRM cells) reside in a wide range of organs, such as the skin, gut, and lung. There, they form a defensive barrier, providing potent immune protection against a plethora of infections and cancer (Masopust and Soerens, 2019; Park et al., 2019; Szabo et al., 2019). After effector T cells enter inflamed tissues, multiple factors determine whether TRM cell precursors “stay” to receive local signals that promote TRM cell differentiation, or “go” and return to the circulation (Mackay et al., 2013). These decisions fundamentally shape the course of the immune response and the magnitude of protective immunity that ensues, yet the signals orchestrating these processes are not yet fully understood.
TRM cell differentiation is not a default pathway for T cells entering peripheral tissues but is instead a multistep process that requires the infiltration, retention, and long-term survival of T cells in the tissue. At epithelial surfaces, local tissue recruitment is achieved via the guidance of effector T cells through chemokine receptors, such as CCR9, CXCR3, and CXCR6 (Mackay et al., 2013; Masopust et al., 2010; Takamura et al., 2019). In contrast, tissue egress of effector and memory T cells is facilitated via the lymphatic system that is rich in chemoattractants, including CCL19/CCL21 (recognized by CCR7; Bromley et al., 2005; Debes et al., 2005) and sphingosine 1-phosphate (S1P; Skon et al., 2013). The sphingolipid S1P is highly abundant in blood and lymph and can bind to five different receptors (S1PR1–5; Baeyens and Schwab, 2020; Ishii et al., 2004; Kihara et al., 2014). Among these receptors, S1PR1 controls key aspects of T cell trafficking from the migration of naive T cells out of LNs to the egress of effector T cells from peripheral organs (Matloubian et al., 2004; Skon et al., 2013). At the transcriptional level, S1PR1 expression is induced by Kruppel-like factor 2 (KLF2; Carlson et al., 2006), and downregulation of Klf2 and S1pr1 are essential for TRM cell differentiation in a wide variety of organs (Skon et al., 2013). In addition, the activation-induced surface molecule CD69, which is found on most TRM cells, can directly bind to S1PR1 and promote its degradation, thereby antagonizing tissue egress and bolstering TRM cell differentiation in the skin (Mackay et al., 2015a; Shiow et al., 2006; Walsh et al., 2019).
While the role of S1PR1 in regulating T cell migration is well established, the contribution of the other S1PR family members is less clear (Baeyens and Schwab, 2020). Among S1PRs, we have previously observed that S1PR5 is expressed by memory CD8+ T cells and selectively downregulated in TRM cells (Mackay et al., 2013, 2016). Although S1PR5 has no ascribed function in T cells, it has been reported that S1PR5 mediates natural killer (NK) cell migration, promoting NK cell egress from the bone marrow (BM) and LNs (Mayol et al., 2011; Walzer et al., 2007). Mice carrying a mutation in Tbx21, the gene encoding the T-box transcription factor expressed in T cells (T-bet), exhibit reduced S1PR5 expression in NK cells, suggesting that T-bet induces S1pr5 (Jenne et al., 2009). In addition, unlike S1PR1, S1PR5 does not interact with CD69 (Jenne et al., 2009), together highlighting salient differences in the molecular regulation of these two receptors. Nonetheless, whether the control of S1PR5 expression is similar in CD8+ T cells and to what extent this receptor may regulate their trafficking remain unclear.
Here, we characterized the expression, regulation, and function of S1PR5 in effector and memory T cell subsets. We found that akin to S1PR1, downregulation of S1PR5 was required for efficient TRM cell differentiation. However, in contrast to S1PR1, which is uniformly expressed in naive and circulating memory T cells under the control of KLF2, S1PR5 expression was only induced following antigen experience and was predominantly driven by the transcription regulator zinc finger E-box binding homeobox 2 (ZEB2), which acts downstream of T-bet. We found that tissue-derived TGF-β was necessary to promote the downregulation of Tbx21 and Zeb2, and ultimately S1pr5, thereby hindering tissue traversion and promoting TRM cell formation. Moreover, we identified a similar role for S1PR5 in potentiating the development of certain tissue-resident innate lymphoid cell (ILC) populations. Collectively, our study identifies S1PR5 as a novel regulator of T cell trafficking that governs formation of tissue-resident lymphocyte populations.
Differential expression of S1PRs by circulating and resident CD8+ T cells
To dissect the roles of the different S1PRs in memory T cell generation, we first examined the pattern of S1PR expression in TRM cells and circulating memory T cell subsets. We began by comparing established transcriptomic profiles (Mackay et al., 2013) of antigen-specific CD69+CD103+CD8+ TRM cells isolated from the skin, lung, and small intestine after HSV-KOS, influenza virus (WSN.gB), or acute lymphocytic choriomeningitis virus (LCMV-Armstrong) infection, respectively, with their circulating effector memory T cell (TEM cell) and central memory T cell (TCM cell) counterparts. Whereas S1pr2, S1pr3, and S1pr4 were similarly expressed by resident and circulating T cell populations, S1pr1 and S1pr5 were selectively downregulated in TRM cells across all three tissues (Fig. 1 A). Consistent with the known role for KLF2 in driving S1pr1 expression (Carlson et al., 2006), Klf2 and S1pr1 transcripts were concordantly elevated in splenic naive T cells and TEM and TCM cells compared with TRM cells (Fig. 1 B). In contrast, S1pr5 was not expressed by naive CD8+ T cells (Fig. 1 B), suggesting that S1PR5 was likely regulated by factors other than KLF2. Analysis of analogous human CD8+ T cell populations revealed that S1PR5, KLF2, and S1PR1 were likewise extinguished in CD103+ skin TRM cells compared with circulating memory T cells in peripheral blood (Fig. 1 C), implying similar regulation of S1PRs in mouse and human CD8+ T cells.
To determine the kinetics of S1pr5 expression during memory T cell differentiation, we transferred congenically marked CD45.1+ gBT-I transgenic CD8+ T cells specific for the immunodominant HSV epitope (gB498–505) into C57BL/6 mice before HSV skin infection. While S1pr5 was not detected in naive gBT-I T cells, its expression was induced in splenic effector T cells as early as 4 d postinfection (dpi), increased over time, and was then maintained in circulating memory T cells for at least 30 d thereafter. Conversely, S1pr5 expression diminished upon T cell entry into the skin, with expression continuing to decline between 14 and 30 dpi (Fig. 1 D). We found that S1pr5 expression was extinguished before the upregulation of the integrin CD103 (Fig. 1 E), which is expressed following epidermal entry and indicative of full acquisition of the TRM cell program (Mackay et al., 2015b). Taken together, our data indicate that akin to S1PR1, S1PR5 is an S1PR expressed by circulating effector and memory T cells that is downregulated during TRM cell differentiation.
T-bet cooperates with ZEB2 to regulate S1PR5 expression in CD8+ T cells
Discordant patterns of S1PR5 and S1PR1 expression in naive T cells suggested a differential role of KLF2 in regulating S1PR1 and S1PR5. To examine this, we first employed ribonucleoprotein (RNP)-based CRISPR/Cas9 technology (Nüssing et al., 2020; Seki and Rutz, 2018) to disrupt Klf2 in in vitro–activated CD8+ T cells. While KLF2-ablated cells showed reduced S1pr1 expression in agreement with previous studies (Carlson et al., 2006; Takada et al., 2011), S1pr5 expression remained unchanged (Fig. S1 A). Interestingly, although KLF2 ablation did not affect S1pr5 transcript levels, overexpression of KLF2 in CD8+ T cells via retroviral transduction increased both S1pr1 and S1pr5 in vitro (Fig. S1 B). Previous studies have shown that forced KLF2 expression is sufficient to induce T-bet in CD4+ T cells (Lee et al., 2015) and that S1PR5 expression in NK cells is controlled by T-bet (Jenne et al., 2009), implying an indirect link between KLF2 and S1PR5. Indeed, ablation of T-bet in KLF2-overexpressing cells markedly reduced S1pr5 mRNA levels without changing S1pr1 expression, demonstrating that S1PR5 is upregulated via T-bet and that KLF2 does not directly drive S1PR5 (Fig. S1 B).
To investigate the role of T-bet in governing S1PR5 expression in vivo, we cotransferred congenically distinct naive OT-I transgenic CD8+ T cells specific for OVA (OVA257–264) and T-bet–deficient OT-I T cells (OT-I Tbx21−/−) into naive C57BL/6 mice before infection with a recombinant strain of HSV-expressing OVA (HSV-OVA). S1pr5 expression was abolished in T-bet-deficient OT-I T cells in the spleen 8 dpi, while S1pr1 and Klf2 expression remained unchanged (Figs. 2 A and S2 A). Furthermore, forced expression of T-bet by retroviral transduction in HSV-primed gBT-I T cells led to S1pr5 upregulation with no impact on Klf2 and S1pr1 expression (Figs. 2 B and S2 B). T-bet can promote the expression of transcription factors, including ZEB2, which can in turn regulate multiple gene targets to coordinate lymphocyte differentiation (Dominguez et al., 2015; Omilusik et al., 2015; van Helden et al., 2015). Consistent with this, we observed Zeb2 upregulation following retrovirus (RV)-induced T-bet expression (Figs. 2 B and S2 B) and reduced expression of Zeb2 in T-bet–deficient T cells responding to HSV infection, which correlated with a marked decrease in S1pr5 expression (Figs. 2 A and S2 A). We therefore asked whether S1PR5 was directly regulated by T-bet or induced indirectly via ZEB2. To this end, we used CRISPR/Cas9 to ablate Zeb2 in effector gBT-I T cells before transfer into HSV-infected mice. Critically, S1pr5 transcripts were substantially reduced in ZEB2-ablated cells, despite similar expression of Tbx21, Klf2, and S1pr1 compared with control T cells (Figs. 2 C and S2 C). This was in agreement with data indicating reduced T-bet binding to S1pr5 promoter in the absence of ZEB2 (Dominguez et al., 2015). Conversely, forced expression of ZEB2 in effector CD8+ T cells using retroviral transduction was sufficient to drive S1pr5 upregulation in vitro (Figs. 2 D and S2 D). Hence, ZEB2 appeared to be the primary regulator of S1pr5 in CD8+ T cells, with T-bet acting upstream of ZEB2. To corroborate this, we forced T-bet expression in ZEB2-ablated or control gBT-I T cells via retroviral transduction and transferred these cells into HSV-infected mice. While T-bet overexpression in gBT-I T cells potentiated SLEC formation, the differentiation of SLEC was reduced when T-bet expression was forced in ZEB2-ablated gBT-I T cells (Fig. S2 E), in agreement with the known role of ZEB2 in driving SLEC differentiation (Dominguez et al., 2015; Omilusik et al., 2015). Importantly, whereas forced T-bet expression increased S1pr5 expression in control gBT-I T cells, this effect was greatly diminished in the absence of ZEB2, although T-bet could still promote low levels of S1pr5 in ZEB2-ablated cells (Fig. 2 E). In contrast, ZEB2 overexpression in Tbx21−/− cells was sufficient to maintain S1pr5 expression in vitro (Fig. S2 F), reminiscent of observations in NK cells (van Helden et al., 2015). Overall, these findings support a model whereby T-bet indirectly regulates S1pr5 by activating expression of ZEB2, which then acts as the major proponent of S1pr5 induction.
Forced expression of S1PR5 perturbs T cell localization in secondary lymphoid organs
S1PR5 upregulation is required for NK cell trafficking from the BM and LNs to the periphery (Jenne et al., 2009; Mayol et al., 2011; Walzer et al., 2007). However, a potential role for S1PR5 in regulating T cell migration has not been explored. To this end, we used RVs to drive S1PR5 expression in CD8+ T cells, which yielded a fourfold increase in S1pr5 gene expression compared with effector CD8+ T cells primed by LCMV infection (Fig. S3 A). To test the impact of S1PR5 expression on CD8+ T cell migration, we transduced congenically marked effector CD8+ T cells with S1PR5 (S1PR5-RV) or control (Ctrl-RV) RVs and cotransferred these cells into naive mice. While forced expression of S1PR5 did not appear to influence T cell localization to the spleen, S1PR5-RV cells were severely underrepresented in LNs (Fig. 3 A). We hypothesized that the discrepancy between S1PR5 control of spleen and LN localization might be attributed to distinct tissue architecture in each lymphoid organ, with the spleen being highly vascularized compared with LNs. To address this, we performed intravascular labeling to distinguish cells located within blood vessels from those in the tissue parenchyma (Anderson et al., 2014). Relative to Ctrl-RV cells, S1PR5-RV cells exposed to the vasculature were increased in both number and frequency (Fig. 3 B and Fig. S3, B and C). Immunofluorescence staining confirmed that while similar numbers of Ctrl-RV and S1PR5-RV cells were observed in the spleen, the majority of Ctrl-RV cells localized to the T cell zone of the white pulp (WP), whereas S1PR5-expressing cells redistributed to the red pulp (RP) at higher frequencies (Fig. 3, C–E).
While S1PR5 has been described to drive cellular egress from lymphoid tissues (Jenne et al., 2009), it remained unclear whether S1PR5 expression could also limit T cell entry into tissues. To investigate this, we analyzed the ability of S1PR5-RV and Ctrl-RV cells to migrate to the spleen and LN 2 h after cell transfer and observed a reduction of S1PR5-RV in the splenic WP and LN parenchyma (Fig. 3 F). Overall, these findings indicate that S1PR5 expression can hinder the entry of CD8+ T cells in lymphoid tissues such as LNs and spleen WP and promote their relocalization into vascular beds.
S1PR5 expression controls bidirectional tissue trafficking and impairs TRM cell formation
TRM cells arise from precursor cells recruited to peripheral tissues during acute inflammation (Mackay et al., 2013; Sheridan et al., 2014). G-coupled protein receptors, including CCR7 and S1PR1, can promote egress of these cells to oppose TRM cell generation (Bromley et al., 2013; Mackay et al., 2013; Skon et al., 2013). To determine whether S1PR5 can also facilitate T cell migration from peripheral tissues and preclude TRM cell development, we injected effector CD8+ T cells transduced with S1PR5-RV or Ctrl-RV directly into the skin. This approach precipitates local TRM cell formation in the absence of local antigen recognition (Mackay et al., 2013). 8 d after intradermal injection, S1PR5-RV cells were numerically reduced in the skin but not the spleen compared with Ctrl-RV cells, suggesting that sustained S1PR5 expression promotes departure of effector T cells from the skin (Fig. 4, A and B). Corroborating this, culturing explanted skin containing control and S1PR5-RV cells cultured overnight revealed increased emigration of S1PR5-RV cells from the skin tissue (Fig. 4 C). Given our results showing that S1PR5 limited cellular homing to lymphoid organs (Fig. 3 F), we asked whether S1PR5 signals could also inhibit T cell entry into the skin. To address this, we forced S1PR5 expression in effector CD8+ T cells that were transferred i.v. into mice treated on the skin flank with the contact sensitizer 1-fluor-2,4-dinitrobenzol (DNFB) to induce T cell migration to this tissue (Frizzell et al., 2020). While equivalent numbers of S1PR5-RV cells were recovered from the spleen, reduced numbers of S1PR5-expressing cells were isolated from the skin as early as 4 d after transfer, a defect that was augmented over time (Fig. 4 D). To extend these observations to other organs, we forced S1PR5 expression in effector CD8+ T cells that were transferred i.v. (Fig. 4, E and F) and also examined P14 transgenic CD8+ T cells specific for the LCMV glycoprotein (gp33–41) that were transferred into LCMV-Armstrong–infected mice (Fig. 4 G). In these settings, TRM cells naturally develop in a wide range of tissues (Casey et al., 2012), and to induce T cell lodgment in the skin, mice were treated with the contact sensitizer DNFB (Frizzell et al., 2020). 8 d after transfer, S1PR5-overexpressing CD69+ T cells were comparatively reduced in the liver, salivary glands, small intestine intra-epithelial lymphocytes (SI-IELs), and skin. Furthermore, the expression of TRM cell–associated molecules, including CD69, CXCR6, and CD103, were diminished in S1PR5-overexpressing T cells (Fig. 4 F). This loss of S1PR5-RV cells did not appear linked to increased cell death in the tissue (Fig. S3 D) but likely reflects their relocalization to the vasculature (Figs. 3, B–D; and S3 C). Collectively, these data suggest that irrespective of the mode of TRM cell generation, S1PR5 expression hinders CD8+ TRM cell differentiation by dampening T cell extravasation and promoting T cell egress from nonlymphoid tissues.
While these findings showed that S1PR5 expression was sufficient to induce T cell exit from peripheral tissues, it was unclear how S1PR5 might intersect with other molecules coordinating tissue retention. CCR7 promotes memory T cell egress from the skin via lymphatics (Bromley et al., 2013), and CCR7 deficiency enhances skin TRM cell differentiation (Mackay et al., 2013). We therefore asked whether S1PR5 could interface with CCR7 signaling to influence CD8+ T cell tissue egress decisions. Consistent with our prior work (Mackay et al., 2013), Ccr7−/− effector T cells transduced with a Ctrl-RV displayed enhanced retention in the skin after intradermal injection compared with WT T cells (Fig. 4 H). Forced expression of S1PR5 in Ccr7−/− T cells restored their capacity to egress from the skin (Fig. 4 H), indicating that S1PR5 signaling can override loss of CCR7 to facilitate tissue egress of CD8+ T cells.
TGF-β enforces retention of TRM cell precursors in the skin via Zeb2 and S1pr5 downregulation
Tissue-derived cytokines, including IL-15 and TGF-β, play a crucial role in shaping multiple aspects of TRM cell differentiation (Mackay et al., 2013, 2015b; Schenkel et al., 2016; Zhang and Bevan, 2013). Among these, TGF-β has previously been shown to extinguish expression of Klf2 and S1pr1 (Skon et al., 2013), T-bet (Mackay et al., 2015b), and Zeb2 (Guan et al., 2018). To explore the impact of TGF-β signaling on S1PR5 expression, we cultured effector CD8+ T cells with TGF-β in vitro and found that this cytokine drives S1pr5 downregulation alongside Klf2, S1pr1, and Zeb2 (Figs. 5 A and S4 A). To assess the importance of TGF-β signaling in vivo, we transferred effector OT-I T cells that were either sufficient or deficient in the TGF-β receptor II (TGF-βRII; OT-I WT or Tgfbr2−/−) into HSV-OVA–infected mice. While TGF-βRII–deficient OT-I T cells were compromised in their ability to form TRM cells and almost entirely lacking from the skin, those that were recoverable from the skin epidermis 2 wk after cell transfer exhibited elevated expression of Klf2 and S1pr1 compared with OT-I WT T cells. Importantly, TGF-βRII–deficient cells also displayed increased expression of Zeb2 and S1pr5, suggesting that TGF-β signaling might also repress the ZEB2–S1PR5 axis to promote skin TRM cell formation (Fig. 5 B).
Given that ZEB2 can induce S1pr5 expression (Fig. 2, C and D) and that S1PR5 signaling drives T cell egress from the skin (Fig. 4, A–D), we reasoned that Zeb2 downregulation via TGF-β signaling may be necessary to promote efficient TRM cell formation. To address whether loss of ZEB2 might enhance skin TRM cell development, we transferred CRISPR/Cas9 Zeb2-ablated naive gBT-I T cells to mice that were subsequently infected with HSV. Consistent with previous reports (Dominguez et al., 2015; Omilusik et al., 2015), ZEB2 ablation led to a defect in SLEC differentiation without altering the number of circulating memory precursor cells (Fig. S4 B). In addition, ZEB2 disruption led to increased formation of CD69+CD103+ TRM cells in the skin 2 wk after HSV infection (Figs. 5 C and S4 C). To assess the relative importance of S1PR5 in this process, we cotransferred congenically distinct OT-I WT T cells (OT-I WT) or S1pr5-deficient OT-I T cells (OT-I S1pr5−/−) into CD45.1 recipient mice before HSV-OVA infection (Fig. 5 D). Akin to ZEB2 ablation, we found that S1PR5 deficiency led to enhanced retention of CD8+ T cells in the skin as early as 8 dpi, with an increased potential to form CD69+CD103+ TRM cells (Fig. 5, D and E; and Fig. S4 D). In contrast, equal proportions of WT and S1pr5-deficient CD8+ T cells localized to the spleen (Fig. 5, D and E; and Fig. S4 D). Together, our data highlight the importance of TGF-β signaling in downregulating Zeb2 and S1pr5 expression, which ultimately enforces the local retention of T cells in peripheral tissues and their differentiation into TRM cells.
S1PR5 disruption alters the tissue distribution of NK cells and ILC1
Tissue residency is not just a feature of CD8+ T cells but also a property shared by both innate and adaptive lymphocytes that use conserved mechanisms to differentiate and persist in peripheral tissues (Gasteiger et al., 2015; Mackay et al., 2016; Robinette et al., 2015). While S1PR5 controls the migration of NK cells from the BM and LNs to the peripheral circulation (Jenne et al., 2009; Walzer et al., 2007), the importance of this receptor in regulating the tissue retention of other ILCs is not known. To explore whether the ZEB2–S1PR5 axis may influence the distribution of additional populations of tissue-resident lymphocytes, we compared the expression of various molecules influencing tissue egress in NK cells and tissue-resident ILC1 in publicly available datasets (Fig. 6 A; Robinette et al., 2015). Analogous to CD8+ TRM cells, tissue-resident ILC1 displayed lower expression of Klf2, S1pr1, S1pr5, and Zeb2 compared with their circulating NK cell counterparts (Fig. 6 A). To understand the importance of S1PR5 during ILC1 differentiation in peripheral tissues, we generated mixed BM chimeras by transferring equal proportions of WT (CD45.1+CD45.2+) and S1pr5−/− (CD45.2+) cells to CD45.1+ hosts. In agreement with prior studies (Jenne et al., 2009; Mayol et al., 2011; Walzer et al., 2007), S1pr5−/− NK cells were found in lower frequencies within the blood and liver. In stark contrast, S1PR5 deficiency led to an increased accumulation of ILC1 in the small intestine and salivary glands (Fig. 6, B and C). Taken together, these results argue that S1PR5 commonly regulates the retention of multiple tissue-resident lymphocyte populations across both innate and adaptive cell lineages.
TRM cells are recognized as essential mediators of host protection against cancer and infections in the periphery (Park et al., 2019; Szabo et al., 2019), yet the processes regulating their generation are not fully understood. One of the key events controlling TRM cell commitment is the downregulation of molecules that promote tissue exit. In the current study, we identify the ZEB2–S1PR5 axis as a novel pathway balancing peripheral T cell trafficking and TRM cell formation, with S1PR5 downregulation being required for efficient TRM cell differentiation. We find that while S1PR5 has a complementary function to S1PR1, these two receptors display distinct expression patterns and transcriptional regulation. Despite this, local cues, such as TGF-β, act in an overarching manner to suppress both S1PR1 and S1PR5 expression and in doing so, coordinate tissue retention. As a consequence, while multiple independent molecular mechanisms converge to control T cell tissue exit, there appears to exist a unifying process that harmonizes these pathways and enforces TRM cell commitment.
It is known that S1P signaling regulates T cell migration and positioning in nonlymphoid organs (Baeyens and Schwab, 2020). We have previously shown that the downregulation of S1PR1 and S1PR5, but not other S1PRs is a conserved feature of the TRM cell differentiation program across distinct tissues (Mackay et al., 2013). Seminal work by Skon et al. (2013) showed that the loss of S1pr1 is required for CD8+ TRM cell development and further, that the forced expression of S1pr1 precludes TRM cell formation. Here, we extend this observation by showing that in addition to S1pr1, S1pr5 is selectively downregulated in TRM cells and similarly required for TRM cell formation. More importantly, we identified key differences between these receptors with regard to their transcriptional regulation, expression patterns, and impact on T cell trafficking over the course of CD8+ T cell differentiation. While S1pr1 expression is primarily regulated by KLF2 (Carlson et al., 2006), the loss of KLF2 had a negligible effect on S1pr5 expression. Instead, we observed that the induction of ZEB2 downstream of T-bet is necessary to promote S1pr5 expression. Interestingly, the sole expression of T-bet does not seem sufficient to induce S1pr5 since T-bet–expressing lymphocytes, including CD4+ and TCRγδ T cells, typically show negligible S1pr5 mRNA levels. This is likely because T-bet acts as a gradient (Dominguez et al., 2015; Joshi et al., 2007) whereby only high concentrations of T-bet can induce ZEB2 and, consequently, S1PR5. Together, this differential molecular regulation explains discordant S1PR1 and S1PR5 expression in naive CD8+ T cells, which express KLF2 (Carlson et al., 2006) but not ZEB2 (Dominguez et al., 2015; Omilusik et al., 2015). Importantly, while S1PR1 primarily acts by promoting T cell egress from peripheral tissues (Skon et al., 2013), we find that S1PR5 additionally affects T cells by hindering tissue infiltration and promoting the exit of interstitial T cells. Given that S1PR5 expression is limited to specific T cell subsets, S1PR5-controlled migration is likely restricted to effector and memory CD8+ T cells and may explain why ZEB2-dependant terminal effector T cells, which show the highest S1PR5 expression, are mostly confined to the vasculature and poorly infiltrate peripheral organs (Gerlach et al., 2016). Consequently, targeting S1PR5 or molecular components that induce this molecule may permit more selective manipulation of antigen-experienced T cells without impacting naive T cell trafficking.
After entry to peripheral tissues, effector T cells receive extrinsic cues that either promote their local retention or oblige their return to the circulation. While chemokine gradients can entice migration of TRM cell precursors toward epithelial surfaces (Mackay et al., 2013; Masopust et al., 2010; Takamura et al., 2019), other pathways, including CCR7, S1PR1, and S1PR5 signaling, actively drive their egress via lymphatics (Bromley et al., 2005; Debes et al., 2005; Ledgerwood et al., 2008; Skon et al., 2013). However, the integration of these signals and their impact on the decision to become tissue resident likely depend on the inflammatory context, tissue location, and cell type. For example, although S1PR1 degradation via CD69 complexing is required to enforce the local retention of CD8+ T cells in the skin, kidney, and lung (Mackay et al., 2015a; Takamura et al., 2016; Walsh et al., 2019), CD69 expression is dispensable for T cell retention in other tissues, such as the small intestine (Walsh et al., 2019). Differences observed in distinct tissue sites suggest a certain degree of redundancy with regard to egress mechanisms, which could be orchestrated by local cues from the microenvironment. Interestingly, the cytokines IL-15 and TGF-β, which drive keys aspects of TRM cell differentiation, survival, and function (Mackay et al., 2015b; Schenkel et al., 2016; Zhang and Bevan, 2013), are also essential to enforce the downregulation of tissue egress molecules. For instance, IL-15 induces the expression of Hobit, which in turn represses Ccr7 and S1pr1 (Mackay et al., 2016). Here, we found that TGF-β promotes the downregulation of both Klf2 and Zeb2, thereby suppressing S1pr1 and S1pr5, respectively, in the skin. While not all TRM cells rely on TGF-β for their development, it is unclear whether such TGF-β–independent TRM cells show a less stable form of tissue residency or might rely more stringently on other retention mechanisms, such as integrins (Christo et al., 2021; McNamara et al., 2017).
The shutdown of tissue egress signals is a key step toward the formation of TRM cells, yet it is not a feature unique to T cells. For instance, the downregulation of egress receptors, such as S1PR1, is widespread among innate lymphocyte populations isolated from peripheral tissues, including NK T cells, mucosal-associated invariant T cells, and ILCs (Huang et al., 2018; Mackay et al., 2016; Salou et al., 2019). Downregulation of tissue egress genes can be enforced via common mechanisms in innate and adaptive lymphocytes, with the transcription factors Blimp1 and Hobit directly repressing Ccr7 and S1pr1 in cells spanning both lineages (Mackay et al., 2016). Here, we found that extinguishment of S1pr5 is an additional conserved program acting to promote tissue residency in both ILCs and CD8+ T cells.
While tissue-resident cells provide critical protection against infections and malignancies, they can also be the driving cause of immune pathologies (Park and Kupper, 2015; Sasson et al., 2020). In such settings, therapeutic intervention aiming to selectively remove these deleterious T cells could be highly beneficial but may prove challenging given the durability of TRM cell populations in some tissues. For instance, skin TRM cells are numerically stable and not displaced over time (Park et al., 2018); furthermore, they are inaccessible to targeted depletion using monoclonal antibodies in some circumstances (Watanabe et al., 2015). Therefore, strategies aiming to remove TRM cells by promoting tissue exit via activation of the KLF2–S1PR1 or ZEB2–S1PR5 pathways may help to overcome these obstacles. On the other hand, while skin TRM cells are highly durable, those that reside in the lung decay over time (Slütter et al., 2017). Interestingly, lung CD8+ TRM cell attrition can be attributed in part to “retrograde migration” whereby long-term tissue-resident cells relocate from the periphery to the circulation at steady state, but the mechanisms facilitating this process are unknown (Stolley et al., 2020). It is possible that tissue egress gene pathways may be specifically reactivated in certain populations of TRM cells, such as those lacking expression of Id3 (Kurd et al., 2020; Milner et al., 2020). Interestingly, it has recently been shown that compared with Id3+ TRM cells, the Id3− population can progressively reexpress ZEB2 and appears to contract within the SI-IEL over time (Milner et al., 2020), potentially via an ZEB2–S1PR5 emigration axis. Elucidating the role of S1PRs in coordinating this process may provide new avenues to promote long-term local immunity for vaccination against mucosal pathogens. Furthermore, S1PR5 may serve as an exit strategy for TRM cells shown to reenter the circulation upon antigen restimulation (Behr et al., 2020; Fonseca et al., 2020; Stolley et al., 2020). Whether antigen recognition may lead to the induction of Zeb2 and S1pr5 in TRM cells to facilitate retrograde migration remains an open question.
Overall, we reveal that S1PR5 acts as a novel checkpoint that controls bidirectional lymphocyte tissue trafficking and impacts the formation of tissue-resident cells across both innate and adaptive cell lineages. Importantly, we demonstrate that tissue T cell retention is a highly coordinated process dictated by the integration of cytokine cues from the local microenvironment, which unanimously enforce the shutdown of multiple tissue egress genes, even when they are induced by distinct transcriptional regulators. Together, these signals enforce the establishment of tissue residency and promote TRM cell formation (Fig. S5). Further studies will be required to determine how S1PR5 interfaces or combines with other tissue egress factors to control peripheral lymphocyte migration and differentiation. Understanding these interactions will unveil how these pathways could be harnessed for therapeutic gain, such as to improve TRM cell formation during vaccination or elicit the departure of malevolent tissue-resident lymphocytes in the context of autoimmunity.
Materials and methods
C57BL/6, B6.SJL-PtprcaPep3b/BoyJ (CD45.1), B6.SJL-PtprcaPep3b/BoyJ × C57BL/6 (CD45.1 × CD45.2), gBT-I CD45.1, gBT-I GFP, P14 CD45.1, P14 Thy1.1, OT-I CD45.1, OT-I GFP, OT-I Tbx21−/− CD45.1, OT-I Tgfbr2f/f.dLck-cre CD45.1 (OT-I Tgfbr2−/−), OT-I S1pr5−/−, and S1pr5−/− mice were bred in the Department of Microbiology and Immunology at The University of Melbourne. Female mice were used for experiments at 6–12 wk of age. All animal experiments were approved by the University of Melbourne Animal Ethics Committee. S1pr5−/− mice were kindly provided by J. Chun. P14 mice express a transgenic TCR recognizing the LCMV glycoprotein-derived epitope gp33–41. gBT-I mice express a transgenic TCR recognizing the HSV glycoprotein-B–derived epitope gB498–505. OT-I mice express a transgenic TCR recognizing the OVA epitope OVA257–264. BM chimera were generated by irradiation of recipient mice (550 rad 3 h apart × 2) followed by reconstitution with 2–5 × 106 donor BM cells. Residual lymphocytes were depleted the next day (100 μg anti-Thy1 [T24] i.p.).
T cell transfer
Adoptive transfers of naive gBT-I, P14, or OT-I T cells were performed i.v. with LN suspensions. Naive gBT-I, P14, or OT-I T cells were transferred at a total number of 5 × 104 or 2.5 × 104 cells/population in cotransfer experiments, where cell types were transferred at a ratio of 1:1.
Infections and DNFB treatment
HSV infection was performed by scarification using 1 × 106 PFU of the KOS strain of the virus (HSV-KOS) or the KOS strain modified to express OVA protein (HSV-OVA). LCMV infection was performed by i.p. injection of 2 × 105 PFU of the Armstrong strain of LCMV. For DNFB treatment, mice were shaved and depilated before application of 15–20 μl DNFB (0.25%) in acetone and oil (4:1) to a 1.5-cm2 region of skin 3 d after LCMV infection.
Mouse tissue processing
Mice were i.v. injected with 4 μg biotin-conjugated anti-CD45 (30-F11) or CD8β (YTS156.7.7) 4 min before euthanasia, as indicated. Blood was collected via an incision in the submandibular region and then lysed using 1× RBC lysis buffer (eBioscience). Spleens and LNs were processed into a single-cell suspension using metal meshes. Femurs were flushed using a 23-gauge syringe filled with 1× PBS to obtain BM single-cell suspension. Livers were meshed through 70-μm cell strainers, and pellets were resuspended in 35% isotonic Percoll (GE Healthcare) before density gradient centrifugation (500 g, 20 min). BM, spleen, and liver RBCs were lysed using 1× RBC lysis buffer (eBioscience). Flank skin was shaved and depilated, and an area of 1–3 cm2 was excised. Skin was incubated in Dispase II (2.5 mg/ml; Roche) for 90 min at 37°C. Epidermal and dermal layers were separated, placed in collagenase III (3 mg/ml; Worthington) and DNase I (5 μg/ml; Roche), chopped into fine pieces, and further incubated for 30 min at 37°C. Digested skin was homogenized into a single-cell suspension and sequentially passed through 70 μm and 30 μm nylon mesh. Small intestine was cleared of luminal contents, and Peyer’s patches were excised. Intestines were longitudinally opened and cut into ∼1-cm fragments, which were incubated at 37°C for 30 min with lateral rotation (230 rpm) in 10% HBSS/Hepes containing dithioerythritol (0.15 mg/ml; Sigma-Aldrich). Intraepithelial lymphocytes were then purified using 44/70% Percoll gradient centrifugation. Kidneys and salivary glands were collected in collagenase III (3 mg/ml) and DNase I (5 μg/ml; Roche), chopped into fine pieces, and incubated for 30 min at 37°C. Digested pieces were homogenized and passed through a 70-μm cell strainer, and lymphocytes were purified using 44/70% Percoll gradient centrifugation.
Human tissue processing
Human peripheral blood and healthy skin were obtained with informed written consent from patients undergoing abdominoplastic surgery. This work was approved by the Ballarat Health Services and St. John of God Hospital Human Research Ethics Committee (ethics number HREC/15/BHSSJOG/5). Peripheral blood mononuclear cells (PBMCs) were prepared by Ficoll gradient centrifugation. T cells were isolated from the skin as previously described (Trubiano et al., 2020). Briefly, skin was incubated in Dispase II solution (2.5 mg/ml; Roche) overnight at 4°C. Epidermal and dermal layers were separated, placed in collagenase III (3 mg/ml; Worthington) and DNase I (5 μg/ml; Roche), chopped into fine pieces, and further incubated for 90 min at 37°C. Skin was homogenized into a single-cell suspension and filtered using 30-μm nylon meshes.
Flow cytometry and cell sorting
Mouse cells were stained at 4°C for 60 min with the following antibodies (all purchased from BD Biosciences, BioLegend, or Thermo Fisher Scientific): anti-B220 (RA3-6B2), anti-CD8α (53-6.7), anti-CD44 (IM7), anti-CD45.1 (A20), anti-CD45.2 (104), anti-CD49a (Ha31/8), anti-CD49b (DX5), anti-CD62L (MEL-14), anti-CD69 (H1.2F3), anti-CD103 (2E7), anti-CD127 (A7R34), anti-CXCR3 (CXCR3-173), anti-CXCR6 (SA051D1), anti-CX3CR1 (SA011F11), anti-KLRG1 (2F1), anti-Ly6C (HK1.4), anti-NK1.1 (PK136), anti-NKp46 (29A1.4), anti-TCRβ (H57-597), anti-TCRγδ (GL3), and anti-Vα2 (B20.1). PE-conjugated annexin V was purchased from BioLegend, and staining was performed according to the manufacturer’s instructions. In some experiments, cells were fixed and permeabilized using a FoxP3 transcription factor staining buffer set and stained with the following antibodies (purchased from BD Biosciences or Cell Signaling Technology): anti-Bcl2 (3F11) and anti-Bim (C34C5). Human cells were stained with the following antibodies (all purchased from BD Biosciences or BioLegend): anti-CD3 (UCHT1), anti-CD8α (SK1), anti-CD45RO (UCHL1), anti-CD69 (FN5D), and anti-CD103 (Ber-ACT8). Dead cells were excluded from analysis using DAPI (0.5 μM; BioLegend), Zombie Yellow, or Zombie NIR fixable live/dead (BioLegend). For flow cytometry experiments, samples were acquired on a five-laser BD LSRFortessa (BD Biosciences) or a five-laser Cytek Aurora analyzer. For cell sorting experiments, murine T cells from the spleen (CD8a+Vα2+CD45.1+/−) and skin (TCRβ+Vα2+CD45.1+/−) or human T cells from PBMCs (CD3+CD8α+CD45RO+) and skin (CD3+CD8α+CD45RO+CD69+CD103− and CD3+CD8α+CD45RO+CD69+CD103+) were sorted using a five-laser BD FACSAria III (BD Biosciences; >95% purity). Data were analyzed using FlowJo version 10 (TreeStar).
RV transduction of CD8+ T cells
RVs were produced using Plat-E cells (Cell Biolabs), which were transfected with pCL-Eco and pMSCV-IRES-GFP II (pMIG II)– or pMSCV-IRES-Thy1.1–based vectors. Briefly, Plat-E cells were seeded in 96-mm dishes at a density of 7 × 106 cells 12 h before transfection with 14 μg of pMIG II and 7 μg of pCL-Eco plasmid DNA using the CalPhos Mammalian Transfection Kit (Takara). Viral supernatants were harvested 48 h later and filtered (0.45 μm; Millipore). T-bet and KLF2 vectors have been previously described (Mackay et al., 2015b; Skon et al., 2013). S1pr5 and Zeb2 cDNA were cloned into pMIG II vector. Purified naive gBT-I, P14, or OT-I CD8+ T cells were in vitro activated with anti-CD3 (145-2C11) and anti-CD28 (37.51, 5 μg/ml for each; both from Bio X Cell) for 24 h and were “spinfected” with 0.5 ml of retroviral supernatant in 24-well plates coated with Retronectin (32 μg/ml; Takara). CD8+ T cells were further expanded for 3 d in the presence of IL-2 (25 U/ml; Peprotech). Transduction efficiency was determined by GFP expression. Cells transduced with an empty vector (Ctrl-RV) or overexpression vectors (Tbet-RV or S1PR5-RV) were mixed at a 1:1 ratio. 2 × 105 transduced cells of the relevant specificity were administered i.v. in mice that were infected with HSV 2 d before or LCMV 1 d before. In the absence of infection, 2–5 × 106 transduced cells were i.v. or intradermally injected into naive recipients. In some experiments, cells transduced with an empty vector (Ctrl-RV) or overexpression vector (ZEB2-RV) were maintained in culture for 3 d in the presence of IL-15 (10 ng/ml; Peprotech), and GFP+ cells were sorted by flow cytometry before quantification of gene expression using quantitative PCR (qPCR).
Skin explant migration assay
Flank skin was shaved and depilated, and an area of 1 cm2 was excised. Skin samples were cultured overnight in complete RPMI medium. The next day, cells that migrated out of the explant were collected from the culture supernatant, whereas cells that did not migrated from the skin tissue were isolated as indicated above. After processing, cells were enumerated by flow cytometry.
T cell culture with cytokines
Mice received naive P14 T cells and were infected with LCMV as described above. 7 dpi, P14 T cells were sorted from the spleen (CD8a+Vα2+CD45.1+) and cultured in the presence of IL-15 (10 ng/ml) and TGF-β (10 ng/ml) for 48 h before quantification of gene expression by qPCR.
CRISPR/Cas9 gene editing of CD8+ T cells
Single guide RNA (sgRNA) targeting Cd19 (5′-CCUGGCCUGGGAUUGCACGU-3′ and 5′-GAGAAGCUGGCUUGGUAUCG-3′), Klf2 (5′-CUGGCCGCGAAAUGAACCCG-3′ and 5′-UCCAUGGGAUUGGACGGUCU-3′), Tbx21 (5′-GGCAGUCACUGCAAUGAACU-3′ and 5′-GGUACUUGUGGAGAGACUGC-3′), Zeb2 (5′-ACUACUGGAAGACCGACAGG-3′ and 5′-CAGAGUCCAAUGCAGCACUU-3′), or scramble control (5′-GCACUACCAGAGCUAACUCA-3′) were purchased from Synthego (CRISPRevolution sgRNA EZ Kit). sgRNA/Cas9 RNPs were formed by incubating 1 μl sgRNA (0.3 nmol/μl) with 0.6 μl Alt-R S.p. Cas9 Nuclease V3 (10 mg/ml; Integrated DNA Technologies) for 10 min at room temperature. Naive or in vitro–activated (anti-CD3 and anti-CD28, 5 μg/ml each for 24–48 h) gBT-I T cells were resuspended in 20 μl P3 buffer (P3 Primary Cell 4D-Nucleofector X Kit S; Lonza), mixed with sgRNA/Cas9 RNP, and electroporated using a Lonza 4D-Nucleofector system (pulse code: DN100 for naive T cells and CM137 for activated T cells). Naive gBT-I T cells were rested for 10 min at 37°C before i.v. transfer, whereas activated gBT-I T cells were further expanded for 3 d in the presence of IL-2 (25 U/ml; Peprotech). 2 × 105 gBT-I T cells edited with control (sgCtrl) and Zeb2 (sgZeb2) guides were mixed at a 1:1 ratio and transferred i.v. Mice receiving edited naive T cells were infected with HSV 4 d after cell transfer, whereas mice that received edited in vitro–activated cells were infected with HSV 2 d before transfer. In some experiments, in vitro–activated P14 T cells were electroporated as mentioned above with control Cd19 (sgCd19) or Klf2 (sgKlf2) targeting guides in P4 buffer (P4 Primary Cell 4D-Nucleofector X Kit S). Cells were further maintained in culture for 3 d in the presence of IL-15 (10 ng/ml; Peprotech) and then performed quantification of gene expression using qPCR. In other experiments, in vitro activated P14 T cells were transduced as mentioned above with pMSCV-IRES-Thy1.1 empty vector (Ctrl-RV) or KLF2 overexpression vectors (KLF2-RV) and were subsequently nucleofected 2 d later with Cd19 (sgCd19) or Tbx21 (sgTbx21) targeting guides in P4 buffer (P4 Primary Cell 4D-Nucleofector X Kit S; Lonza). Cells were further maintained in culture for 3 d in the presence of rhIL-2 (100 U/ml, Peprotech), and Thy1.1+ cells were sorted by flow cytometry before quantification of gene expression using qPCR.
RNA was extracted from sorted samples using RNeasy Plus Micro Kit (QIAGEN) according to the manufacturer’s instructions. mRNA was converted into cDNA using High Capacity cDNA Reverse Transcription Kit or SuperScript IV VILO Master Mix (Thermo Fisher Scientific), and genes of interest were preamplified using TaqMan PreAmp Master Mix (Thermo Fisher Scientific). Gene expression was analyzed by real-time PCR using the StepOnePlus Real-Time PCR System (Thermo Fisher Scientific), TaqMan Fast Advanced Master Mix, and the following TaqMan probes (purchased from Thermo Fisher Scientific or Integrated DNA Technologies): mouse probes, Hprt Mm00446968_m1, Klf2 Mm01244979_g1, S1pr1 Mm00514644_m1, Tbp Mm00446973_m1, Tbx21 Mm00450960_m1, and Zeb2 Mm.PT.58.7239300. S1pr5 probe was custom made using the following primers: forward primer 5′-ACCAAGACTCCTCCAACA-3′, TaqMan probe 5′-AACCTTGGATCGCAGCTCTAGCC-3′, reverse primer 5′-GGGAGACAAGTGTTCTGATG-3′; and human probes, KLF2 Hs00360439_g1, S1PR1 Hs00173499_m1, S1PR5 Hs00928195_s1, UBC Hs00824723_m1. In some experiments, gene expression was assessed with an ABI 7000 sequence-detection system, and amplification was detected with PowerUp SYBR Green Master Mix (Applied Biosystems). The sequences of the primer pairs used were as follows: Klf2, forward 5′-ACCAACTGCGGCAAGACCTA-3′ and reverse 5′-CATCCTTCCCAGTTGCAATGA-3′; S1pr1, forward 5′-GTGTAGACCCAGAGTCCTGCG-3′ and reverse 5′-AGCTTTTCCTTGGCTGGAGAG-3′; S1pr5, forward 5′-GCCTGGTGCCTACTGCTACAG-3′ and reverse 5′-CCTCCGTCGCTGGCTATTTCC-3′; Tbx21, forward 5′-CAACAACCCCTTTGCCAAAG-3′ and reverse 5′-TCCCCCAAGCAGTTGACAGT-3′; Zeb2, forward 5′-CATGAACCCATTTAGTGCCA-3′ and reverse 5′-AGCAAGTCTCCCTGAAATCC-3′ or forward 5′-CCAGAGGAAACAAGGATTTCAG-3′ and reverse 5′-AGGCCTGACATGTAGTCTTGTG-3′; and Hprt, forward 5′-CATTATGCCGAGGATTTGGAA-3′ and reverse 5′-CACACAGAGGGCCACAATGT-3′. Cycle-threshold values were determined for genes individually, and gene expression was normalized according to the 2-dCt method to the housekeeping gene Hprt or Tbp for mouse samples or UBC for human samples.
cDNA library preparation, sequencing, normalization of sequenced reads, and analysis of differentially expressed genes have been described elsewhere (Christo et al., 2021). Heatmap representation of selected differentially expressed genes was generated using pheatmap. RNA sequencing data have been deposited at the National Center for Biotechnology Information Gene Expression Omnibus (GEO) public database under accession no. GSE178768.
Confocal microscopy and image analysis
Spleens were prepared for immunofluorescence staining as described previously (Kato et al., 2015). Briefly, spleens were fixed for 8 h at 4°C in PLP buffer (0.2 M NaH2PO4, 0.2 M Na2HPO4, 0.2 M L-lysine, and 0.1 M sodium periodate with 2% paraformaldehyde), washed twice in PBS, embedded in optimal cutting temperature compound (Tissue Tek; Sakura Finetek), snap frozen in liquid nitrogen, and stored at −80°C. Frozen tissues were sectioned to 12-μm thickness with a cryostat (CM3050S; Leica). Sections were fixed with acetone, blocked for 10 min with Serum-Free Protein Block (Dako), and stained with the following antibodies (all from BioLegend): anti-B220 (RA3-6B2) Pacific Blue, anti-GFP (FM264G) AF488, and anti-CD31 (390) AF647. Slides were mounted using ProLong Gold Antifade Reagent (Invitrogen). Mosaic imaging covering ∼1,600 × 3,700 μm were acquired with a 20× objective for each spleen on an LSM 700 confocal microscope. Images were processed using Imaris (Bitplane). Delineation of spleen WP and RP and quantification of retrovirally transduced GFP+ cells in these areas were performed semiautomatically using the Imaris “surface” tool.
All statistics analyses were calculated by performing unpaired t test, unpaired Mann–Whitney test, paired Wilcoxon test, one-way ANOVA test with Dunnet posttest, or two-way ANOVA test with Šidák posttest using Prism 8 (GraphPad Software). *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.
Online supplemental material
Fig. S1 shows the effect of KLF2 ablation or overexpression on the expression of Klf2, S1pr1, S1pr5, Tbx21, and Zeb2 in effector CD8+ T cells. Fig. S2 shows the effect of T-bet or ZEB2 ablation or overexpression on the expression of Klf2, S1pr1, S1pr5, Tbx21, and Zeb2 and SLEC differentiation in effector CD8+ T cells in various settings. Fig. S3 depicts the degree of S1pr5 overexpression in S1PR5-transduced cells and its effect on their vascular localization and regulation of apoptosis. Fig. S4 illustrates the impact of TGF-β signaling on the expression of various genes, including S1pr5, and shows the effect of Zeb2 and S1pr5 ablation of CD8+ T cell differentiation. Fig. S5 shows a model of how S1PR5 expression is regulated in CD8+ T cells and how it affects their trafficking and differentiation into TRM cells.
We thank Brooke Davies, David Freestone, and Natasha Zamudio for expert technical assistance. We thank Francis R. Carbone and Junyun Lai for scientific discussion and manuscript edits, Imran House for assistance with Cas9/RNP-related experiments (Peter MacCallum Cancer Centre, Melbourne, Australia), and Dr. Robert Sheen (St. John of God Hospital, Ballarat, Australia) for provision of surgical samples. We thank the Melbourne Cytometry Platform from the Peter Doherty Institute for assistance with cell sorting.
This work was supported by a Howard Hughes Medical Institute and Bill and Melinda Gates Foundation International Research Scholarship OPP1175796 to L.K. Mackay and Australian Research Council grant DP200102753 to L.K. Mackay and M. Evrard. M. Evrard is supported by the University of Melbourne McKenzie Postdoctoral Fellowship. S.L. Park is supported by a National Health and Medical Research Council Emerging Leadership Investigator grant. D.G. Pellicci is supported by a Commonwealth Serum Laboratories Centenary Fellowship. L.K. Mackay is a senior medical research fellow supported by the Sylvia and Charles Viertel Charitable Foundation.
Author contributions: Conceptualization, M. Evrard and L.K. Mackay. Investigation, M. Evrard, E. Wynne-Jones, C. Peng, Y. Kato, S.N. Christo, R. Fonseca, S.L. Park, T.N. Burn, and M. Osman. Formal analysis, M. Evrard, E. Wynne-Jones, C. Peng, Y. Kato, and S. Devi. Writing (original draft), M. Evrard and L.K. Mackay. Writing (review and editing), M. Evrard, S.N. Christo, R. Fonseca, S.L. Park, and L.K. Mackay. Visualization, M. Evrard and L.K. Mackay. Resources, J. Chun, S.N. Mueller, G. Kannourakis, S.P. Berzins, D.G. Pellicci, W.R. Heath, and S.C. Jameson. Funding acquisition, L.K. Mackay. Supervision, W.R. Heath, S.C. Jameson, and L.K. Mackay.
Disclosures: The authors declare no competing interests exist.
E. Wynne-Jones’s present address is Viral Immunity and Pathogenesis Unit, Laboratory of Clinical Immunology and Microbiology, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD.
Y. Kato’s present address is Center for Infectious Disease and Vaccine Research, La Jolla Institute for Immunology, La Jolla, CA.