Tim-3 drives tissue Treg effector function to promote neonatal immune tolerance and tumor-induced immune suppression

: Tissue regulatory T cells (Treg) are a key player in preventing immune pathology and shaping the immune suppressive tumor microenvironment (TME). The signaling molecules that uniquely regulate tissue Treg are not well understood. Here, we show that Tim-3 was predominantly expressed in Treg cells in both normal and tumor tissues. Notably, Tim-3 deficiency in Tregs led to the development of severe signs of autoimmune diseases in neonatal mice, indicating a crucial role of Tim-3 in establishing immune tolerance in early life. In addition, deletion of Tim-3 on Tregs resulted in reduced numbers and suppressor activities of tumoral Tregs and an increase in antitumor immune responses. A gene profiling study revealed the requirement of Tim-3 in upregulation of effector Treg genes encoding immune inhibitory molecules, chemokine receptors for tissue residency, chemokines, and study demonstrates the essential role of in driving tissue effector Treg-mediated immune suppression. to cell cell role + In fact, blockade of or inhibit CD8 + T cell immune responses depending on the experimental systems (5, 12, 13, 44, 45, 48-50). However, our study indicates that at least one way of synergy of PD-1 and Tim-3 blockade is through their direct function on Treg cells. Our study offers a simpler explanation for the synergy between PD-1 and Tim-3 blockade which is through a reduction of Treg function by Tim-3 mAbs and a decrease of CD8 + T cell exhaustion by PD-1 mAbs. By providing a new mechanism of synergy between PD-1 and Tim-3 blockade, our study will shed light on targeting Tim-3 for cancer immune therapy. Representative flow statistics of CD45 + , CD4 + and + lymphocytes IFN- g , granzyme B production showed (n=4). Data were presented as mean ± SEM, *P < 0.05, ** P < 0.01, *** P < 0.001, Student’s t test were performed. Data are representative three independent experiments.


Introduction:
Regulatory T (Treg) cells play a key part in preventing autoimmunity and maintaining immune tolerance in cancer and transplantation (1). Treg cells that are present in lymphoid tissues have been extensively investigated. Comparatively, Treg cells that reside in the non-lymphoid tissues, called tissue Tregs, are not well studied. Recent studies demonstrate that tissue Tregs, in small numbers, are found in many nonlymphoid healthy tissues, infected tissues, grafts, placenta, injured muscle, and tumors (2). It appears that tissue Tregs, similar to tissue conventional T cells, express genes that are upregulated in conventional effector, memory, and exhausted T cells.
They also express chemokine receptors and adhesion molecules unique to various tissues. Moreover, tissue Treg cells have a distinct T cell receptor (TCR) repertoire indicating a local clonal expansion of these cells (3,4). It is believed that these unique properties empower tissue Tregs to drive immune suppression in the tumor microenvironment (TME) (2). However, the specific molecular mechanisms that mediate the function of tissue Treg are not well understood.
Tim-3 was originally found upregulated in a Th1 cell line (5). Its expression was shown to be upregulated in chronically activated T cells upon multiple rounds of stimulation (6)(7)(8). It is upregulated and considered a marker for terminally differentiated T cells such as exhausted T cells (9)(10)(11)(12)(13). Besides conventional T cells, Tim-3 has recently been shown to be characteristically expressed in tissue Treg (2,(14)(15)(16)(17)(18). Tim-3 was first reported to be specifically upregulated in tumor infiltrating regulatory T cells and its expression has been associated with tumor progression in human lung cancer (14). Tim-3 was also expressed in tissue resident Treg in normal tissues such as fat tissues (2). Recent studies have demonstrated that Tim-3 + Treg have effector like phenotype and greater suppressive activities than Tim-3 -Treg (15)(16)(17)(18)(19). In addition, combination treatement of tumor-bearing mice with PD-1 mAbs and Tim-3 mAbs led to a decrease in Treg function along with increased effector T cell function (17). The exact function of Tim-3 on tissue Tregs, however, remains unresolved.
In this study, we are aiming to shed light on the functional role of Tim-3 on Tregs. To this end, we generated Foxp3 specific Tim-3 deficient mice. Using this model, we examined the role of Tim-3 in Treg-mediated prevention of autoimmunity and tumor immune suppression. We also determined the function of Tim-3 on Tregs in tumor tolerance and tumor checkpoint-blockade immunotherapy. Furthermore, we performed whole transcriptome analysis to comprehensively study the molecular mechanism underlying Tim-3's function in tissue Treg cells. Our study establishes the functional role of Tim-3 in tissue Treg cells and the significance for tumor immunotherapy targeting Tim-3 + Treg cells.

Mice with Tim-3 deficiency in Tregs developed severe signs of autoimmune diseases
We and others have previously demonstrated that Tim-3 is uniquely expressed on tissue regulatory T cells (2,14). Here, we also showed that Tim-3 was specifically up-regulated on skin tissue Treg cells from neonatal mice but was expressed at a minimal level in peripheral Treg cells (Fig1A). In order to elucidate the functional role of Tim-3 in Tregs, we generated Tim-3 flox/fox /Foxp3 YFP-Cre (hereafter referred to as T f/f F Cre ) mice to specifically delete Tim-3 in Treg cells. Interestingly, T f/f F Cre neonates quickly became runted around 14 days after birth and did not survive beyond 35 days (Fig. 1B). T f/f F Cre mice also had scaly skins and tails as well as deformed ears, similar to the phenotype of scurfy mice (Fig. S1A) (20). Histological analysis showed hyper inflammation and mononuclear cell infiltration in skins and lungs from 18-day-old neonates ( Fig. 1C and D and Fig. S1C). It was noted that lymph nodes were greatly enlarged in all T f/f F Cre mice around 18 days after birth (Fig. S1A). Flow cytometry analysis demonstrated much lower percentages of Tregs in the skin and spleen but not lymph node in T f/f F Cre mice compared to control wild type Tim-3 /Foxp3 YFP-Cre mice (T +/+ F Cre ) ( Fig. 1E and F).
The analysis also showed that T f/f F Cre mice had a higher percentage of both total immune cells and T cells in the skin when compared to wild type control mice (Fig1. G and H). In addition, the percentage of conventional CD4 + and CD8 + T cells were also increased in the spleen of T f/f F Cre mice compared to the percentage of those in control mice ( Fig. S1D and E). Consistent with tissue inflammation and lymphadenopathy, higher percentages of CD44 + and Ki67 + conventional T cells and lower percentages of CD62L + conventional T cells were present in the skin, spleen, and lymph nodes of T f/f F Cre mice compared to control mice, indicating dysregulated activation of T lymphocytes in T f/f F Cre mice (Fig1. G, H, and Fig. S1D-G). Despite a huge increase of absolute number of cells in lymph node (FigS1B) and hyper activation of conventional T cells( Fig. S1F and G), the percentage of CD4 + T cells was decreased in lymph node of T f/f F Cre mice. Collectively, these data suggest that Tim-3 expression on Tregs is crucial for maintaining immune tolerance homeostasis in neonates.

Tim-3 deletion in Tregs resulted in a reduction in the number of tissue Tregs
Although T f/f F Cre mice developed severe signs of autoimmune diseases around 18 days after birth, the neonates still looked normal around 10 days. It is understood that Treg cells start to accumulate in skin beginning around postnatal day 6 and peak around day 13 (21). We then examined peripheral and skin immune cells in T +/+ F cre and T f/f F Cre mice on postnatal day 10 by flow cytometry. The percentages of Tregs in spleens and lymph nodes were similar between T +/+ F cre and T f/f F Cre mice (Fig. S2A).
In contrast, the percentages of Tregs in skin were significantly reduced in T f/f F Cre mice compared to control mice ( Fig. 2A and B). In addition, Tim-3 was not significantly expressed on skin Treg cells from T f/f F Cre mice but expressed at high levels on those from control mice ( Fig. 2A and B). In contrast to a large increase in CD45 + immune cells in the skin of 18-day-old T f/f F Cre mice, there were significantly less CD45 + immune cells in the skin of 10-day-old T f/f F Cre mice compared with control mice ( Fig. 2A and B). However, both conventional CD4 + and CD8 + skin T cells expressed higher levels of activation markers such as CD44 and lower levels of naïve/ central memory T cells markers such as CD62L and IL7R ( Fig. 2A and B) in T f/f F Cre mice compared to control mice. In addition, conventional CD4 + and CD8 + T cells expressed higher levels of Ki67 in T f/f F Cre mice compared to control mice ( Fig.   2A and B), suggesting that these cells were highly proliferative. Consistent with these changes in the tissue conventional T cells, the percentage of effector/memory T cells was also increased in the spleen ( Fig. S2C and D) and slightly increased in lymph nodes ( Fig. S2B) from T f/f F Cre mice compared to control mice. No difference of thymic T cells was observed at this stage (data not shown). These data suggest that Tim-3 is required for the accumulation of tissue Tregs, and thus prevents uncontrolled activation and proliferation of T cells to maintain immune homeostasis at the early stage of life.

Ablation of Tim-3 in Treg cells increased the immune surveillance of tumor
Although T f/f F Cre mice did not survive more than 35 days, approximately 40% of Tim-3 flox/ko Foxp3 YFP-Cre (hereafter referred to as T f/-F Cre ) mice, which had deleted one allele of Tim-3 in all cells and had one floxed Tim-3 allele, survived to adulthood (Fig. 3A).
The lack of Tim-3 expression on some non-Treg immune cells in T f/-F Cre mice likely dampened the early lethality phenotype of T f/f F Cre mice. We then tried to determine Analysis of tumor immune infiltrates was performed when tumor sizes were 7~8 mm and similar T f/-F Cre and T +/-F Cre mice. The data indicated that both the percentage and total number of CD45 + lymphocytes in T f/-F Cre mice were significantly greater than those in control group. In addition, percentages of both CD4 + and CD8 + TIL cells were increased in T f/-F Cre mice when compared to control mice. (Fig. 3D and E).

Consistent with our findings in skin inflammation, percentages of tumor infiltrating
Tregs were reduced in T f/-F Cre mice ( Fig.3D and E). Notably, Tim-3 was completely absent on Treg cells but expressed on conventional CD4 + and CD8 + TIL in T f/-F Cre mice (Fig. 3D). In line with the reduction of Tregs cells, the effector molecules IFN-g in CD4 + T cells and granzyme B in CD8 + T cells were greatly increased in T f/-F Cre mice ( Fig. 3D and E). In addition, the percentage of "exhausted" PD-1 + Tim-3 + CD8 + T cells was reduced in T f/-F Cre mice. Consistent to flow cytometry data, the IFN-g ELISpot assay demonstrated that greater numbers of tumor antigen-specific T cells We also examined B16 melanoma, a poorly immunogenic tumor model, to further characterize the function of Tim-3 in tumoral Tregs. Tumor growth was similar between T f/-F Cre and control mice (Fig. S3E). However, more CD45 + lymphocytes and T cell infiltration were found in T f/-F Cre mice than control mice ( Fig. S3F-G).
Again, the percentage of Foxp3 + Treg cells was greatly reduced in T f/-F Cre compared to control mice ( Fig. S3F-G). Effector molecules of CD4 + TIL cells were also increased in T f/-F Cre mice ( Fig. S3F-G). In addition, CD44 and Ki67 expression of CD4 + TIL were increased, indicating enhanced activation and proliferation of CD4 + TIL in T f/-F Cre mice ( Fig. S3F-G). These data suggest that Tim-3 deficiency in Treg cells leads to reduced immune suppression in the TME and a profound increase in anti-tumor immune responses.

Tim-3 is required for the suppressor function of tumor infiltrating Tregs.
Tim-3 + tissue Treg cells have been shown to be more activated and suppressive in human cancer (15)(16)(17)19). Given our data that mice with Tim-3 deficiency in Tregs Consistent with the results from Tim-3 deficient mice, Tim-3 mAb treatment also resulted in a reduction of the percentage of tumoral Tregs. In addition, administration of Tim-3 mAb led to a decrease in CD44 and PD-1 expression and to an increase in IL7R expression in Tregs in the TME (Fig. 4F-G). These data suggest that Tim-3 is involved in the activation of Treg cells and also mediate immune suppressive functions of effector Tregs in tumor tissues.

Tim-3 is required for driving a network of genes responsible for the function of effector Tregs.
To gain a comprehensive understanding of the role of Tim-3 in Tregs in the TME, we performed RNA-sequencing analysis comparing Tim-3-/-Tregs and control Tregs in the TME. There were more than 700 genes that were differentially expressed between T f/-F Cre and control tumoral Tregs (Fig. 5A). Consistent with cytometric analysis, we found that Tim-3 deficient Treg cells expressed lower levels of CD44 but higher levels of IL7R and Tcf7, indicating Tim-3 is required for the effector phenotype of Treg cells (Fig. 5A, C). Gene enrichment analysis showed that control tumoral Tregs upregulated genes that were enriched with signature genes of Tim-3 + Tregs and genes highly expressed in Tim-3-/-Tregs were enriched with Tim-3 -Treg signature genes (15) (Fig. 5B). These data indicated that Tim-3 signaling was required for the generation and function of Tim-3 + Treg cells in the TME. Gene enrichment analysis also indicated that genes upregulated in Tim-3 deficient tumoral Treg cells were enriched with signature genes of naïve CD4 + T cells and naïve Treg cells (Fig. 5B), suggesting that Tim-3 is required for the expression of effector Treg genes. Gene Ontology analysis revealed several categories of genes, such as cell cycle/apoptosis/survival, chemotaxis/adhesion, helper T cell subset differentiation, immune suppression and signaling, were increased in the Tim-3 sufficient tumoral Tregs, whereas several categories of genes including antigen presentation, cell cycle/apoptosis/survival, chemotaxis and adhesion, cytokine, differentiation, metabolism, signaling and stress response were enriched in Tim-3 deficient tumoral Treg cells (Fig. S5). Therefore, these data suggested that Tim-3 signaling is involved in regulating multitudes of cellular processes.
Further study of immunological function of the differentially expressed genes yielded an in depth understanding of how Tim-3 drove Treg-mediated immune suppression (Fig. 5C). We first focused on genes that were expressed at higher levels in WT Tregs than Tim-3 deficient Tregs. The analysis indicated that Tim-3 was required for the expression of chemokine receptor and adhesion molecules, such as Cxcr6, Itgb8, Itgav, Cx3cr1, Ccr8, and Cd44, which were likely involved in tissue residence of the effector/memory Treg cells (22)(23)(24)(25). The reduced expression of these genes might explain the diminished number of Tim-3 deficient Treg cells in tumor tissues. It has been shown that the suppression function of Treg cells is dependent on their abilities to bring their target cells to proximity. This is controlled by chemokines such as Ccl3, Ccl4, Ccl1, Xcl1, and Ccl12 (15,26,27). Tim-3 was required for the upregulation of these chemokines in tumoral Tregs ( Figure 5C). The analysis indicated that Tim-3 was critical for the expression of an array of immune suppressive effector genes such as Calca, Fcrl6, Stab1, F2rl2, Il10, Ramp1, Cd200, Lag3, Lgals4, Fasl, Entpd1 (28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38). Interestingly, αvβ8 integrin mRNAs were also expressed at a higher level in Tim-3 sufficient Tregs, suggesting Tim-3 expression confers Treg cells the ability to activate TGF-β (39)(40)(41). Genes that were expressed higher in Tim-3 deficient Treg cells were mainly involved in function of the naïve and central/stem memory Treg cells, which was consistent with the idea that Tim-3 is involved in mediating the activation and function of effector Treg cells. Among these genes, several transcription factors, such as IL7R, Klf3, Tcf7, Twist2, Id3, Tsc22d3, Smad1, and Klf2, have been shown to be highly expressed in naïve and memory stem T cells (42). Higher expression of apoptotic and cell cycle inhibitor genes, such as Bcl2l15, Bcl2, Cdkn2b, Cdkn1c, and Cdkn2a, were also consistent with the resting phenotype (42).

Treg-specific deletion of Tim-3 gene synergized with PD-1 mAbs for tumor therapy.
It is thought that Tim-3 and PD-1 co-expression is the marker for exhausted CD8 + T cells and synergistic antitumor activities between PD-1 and Tim-3 mAbs are due to reversal of CD8 T cell exhaustion (12,13). Since we have shown that both Tim  Fig. 6A and Fig. S6A). Similar to what we observed using MC38 model, we also observed synergy between Tim-3 Treg deficiency and PD-1 mAbs in B16 tumor model (Fig. S6D). In addition, using Tim-3 f/f F CreErt2 mice, we also found that inducible Tim-3 deletion resulted in slower tumor growth and PD-1 mAbs treatment led to further slowed tumor growth (Fig.6B).
Consistent with reduced tumor growth, flow cytometry also showed higher number or percentages of CD45 + lymphocytes and IFN-g + CD4 + T cells in tumors isolated from

Discussion:
In this study, we have revealed a critical functional role Tim-3 plays in tissue Tregs.
Our data indicate that Tim-3 is important for suppressor activities of tumor tissue Tregs. The predominant expression Tim-3 in tissue Tregs is crucial for preventing autoimmunity in neonate mice and maintaining the immune suppressive TME in adult mice. In addition, Tim-3 is required for upregulating effector molecules such as chemokines and immune inhibitory molecules of tissue Tregs. Finally, the enhanced antitumor activities after combination of Tim-3 deficiency in Tregs and PD-1 blockade illustrate a new mechanism of synergy between Tim3 and PD-1 blockade.
The exact role of Tim3 in the T cell-mediated immune response has been complicated due to lack of clear understanding of its unique function in different T cell subsets. Besides conventional T cells, Tim-3 has recently been found highly expressed on tumor and normal tissue Treg cells (2,(14)(15)(16). Our finding that Tim-3 plays a critical role in tissue Tregs help explain the cellular mechanisms of many previous observations. One study shows that in a GVHD model, Tim-3 is induced on activated T cells. Tim-3 blockade by a Tim-3-Ig fusion protein or Tim-3 genetic deletion in donor T cells led to increased lethality (16). These data suggest that Tim-3 signaling inhibits GVHD. Although the authors have also showed in the same study that the effect of Tim-3 blockade on GVHD is dependent on donor Treg cells, they conclude that Tim-3's function in promoting cell death in Th1 cells is responsible for its suppressive effect on GVHD. However, in light of our current finding, it is also possible that Tim-3 is required for Treg function during GVHD and the lack of Tim-3 signaling in Tregs has led to an increase in GVHD. In another study, Tim-3 is shown to inhibit alloimmune responses (43). The authors have indicated that the effect of Tim-3 blockade in preventing transplantation tolerance was mediated by inhibiting immunosuppressive function of the alloantigen-primed Tregs (43). Although the authors of this study did not pinpoint activated Tregs as the cellular source of Tim-3, their result is consistent with the conclusion of our study. Therefore, many studies support the notion that Tim-3 mediates immune suppression through its direct functional effect on effector Treg cells.
Our previous study demonstrates that Tim-3 is specifically up-regulated in tumor infiltrating but not peripheral Treg cells (14). Many studies also show that Tim-3 is expressed on normal tissue Treg cells (2). In addition, Tim-3 + Tregs were found to have effector phenotype and higher suppressor activities (3,4,(15)(16)(17)(18)(19). In addition, combined treatement of tumor-bearing mice with PD-1 mAbs and Tim-3 mAbs led to a decrease in Treg function along with increased effector T cell function (17). Alought It has been demonstrated that PD-1 and Tim-3 mAbs have synergistic antitumor activities in many preclinical models (12,13). PD-1 and Tim-3 coexpression has been considered makers of exhausted CD8 + T cells and therefore PD-1 and Tim-3 blockade has been thought to reverse CD8 + T cell exhaustion (12,13). Although PD-1 mAbs has been shown to reverse T cell exhaustion, the role of Tim-3 in CD8 + T cells has been controversial (44)(45)(46)(47)(48). In fact, blockade of Tim-3 has been shown to promote or inhibit CD8 + T cell immune responses depending on the experimental systems (5,12,13,44,45,(48)(49)(50). However, our study indicates that at least one way of synergy of PD-1 and Tim-3 blockade is through their direct function on Treg cells. Our study offers a simpler explanation for the synergy between PD-1 and Tim-3 blockade which is through a reduction of Treg function by Tim-3 mAbs and a decrease of CD8 + T cell exhaustion by PD-1 mAbs. By providing a new mechanism of synergy between PD-1 and Tim-3 blockade, our study will shed light on targeting Tim-3 for cancer immune therapy.

Microsuppression Assay
Splenic CD4 + CD25 -T cells were FACS sorted from naïve mouse as responder cells and labeled with 5µM CellTrace Violet (Invitrogen). FACS-sorted CD4 -CD8splenocytes were irradiated with 3000 rad and used as antigen presenting cells (APC). Tumoral Tregs were sorted from T +/-F Cre and T f/-F Cre tumor bearing mice.  RNA-seq data were aligned to mouse reference genome GRCm38 using STAR.

RNA-seq profiling and gene expression analysis
Gene expression level was quantified and count expression matrices were generated using RSEM from aligned reads. Count per million (TPM) was used for further analysis. Package limma in R (51) was used to identify differentially expressed genes (DEGs) between T +/-F Cre and T f/-F Cre tumoral Tregs. DEGs with Benjamini-Hochberg adjusted P value <0.05 and absolute log2(fold change) (log2FC) ³ 0.5 were selected for further analysis. The signal-to-noise ratio was used for gene ranking and javaGSEA (52) was used to run GSEA with 1000 permutations to estimate P values.
Corrections for multiple tests were applied using the Benjamini-Hochberg procedure.
The GSEA maps were drawn by clusterProfiler(53) package in R. Heatmap figures were generated using the pheatmap package in R. All analyses were done in R v.3.6.0. The application Cluego (54) in Cytoscape v.3.7.1 was used for pathway enrichment analysis of DEGs.

Statistical analysis
Student's t-test and Log-rank test were performed with Graphpad Prism software,