Introduction

T cells play a crucial role in inducing immunity by recognizing peptides presented on MHC molecules through their TCR during pathogen infection. The strength of the TCR’s reactivity to the foreign antigen is contributed by both the affinity of the TCR for that antigen and the basal TCR signaling level of the T cell, which is determined by weak interactions between the TCR and self-peptides presented during T cell development (1, 2). Previous studies have shown that T cell self-reactivity with weak interaction between TCR and self-peptide-MHC (self-pMHC) not only instructs T cell development and activation thresholds, but also influences their functional response (2–4). It is noteworthy that the strength of the weak interaction can be evaluated through CD5, whose expression level is considered proportional to T cell self-reactivity in the developmental stage, and may reflect the responses of T cells to foreign-pMHC complexes in the future (2, 5, 6).

CD5 is a scavenger receptor cysteine-rich protein expressed on T and B cells that regulates TCR and B-cell receptor signaling (7, 8). It functions as a scaffold protein for the recruitment of phosphatase SHP-1 and E3-ubiquitin ligases CBL and CBLB, to reduce the phosphorylation of proximal TCR signaling molecule Zap70, thereby inhibiting TCR-mediated signaling (9). CD5, as a negative regulator for TCR signaling, also plays a crucial role in fine-tuning the process of positive and negative selection during thymocyte development (8, 10). Specifically, thymocytes with higher self-reactivity TCRs induce CD5 expression, which helps them adapt to strong TCR signaling and reduces the likelihood of negative selection. This, in turn, promotes thymocyte positive selection and broadens the affinity spectrum of the TCR repertoire during the TCR self-recognition process. Therefore, the expression of CD5 is not only a surrogate marker for the strength of TCR-self-peptide-MHC interactions during development but also reflects the intrinsic response set point of T cells to a signal (5, 10). This selection process enriches the mature CD4⁺ T cell repertoire with clones that exhibit increased responses to self-peptide-MHC, thereby displaying enhanced peripheral immune responses to pathogens and autoantigens in vivo (5, 11, 12). Not only in CD4⁺ T cells, previous study has also shown that CD5^hi population in CD8⁺ T cells, characterized by higher reactivity to self-peptide-MHC, are prone to positive selection (5, 13). In diabetes-prone NOD mice, though negative selection in central tolerance limits the early contribution of high-avidity T cells to disease development, prolonged inflammation in pancreatic islets still drives the avidity maturation of islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP)-specific CD8⁺ T cells, leading to increased cytotoxicity towards β cells (14–16). This suggests that mechanisms other than incomplete thymic negative selection are involved in generating autoreactive T cells. Here, we investigate whether the CD5-correlated self-pMHC reactivity, which attributes to TCR signal strength in autoreactive CD8⁺ T cells during thymic positive selection, contributes to the development of autoimmune disease.

The NOD mouse model is commonly used for studying autoimmune diabetes. NOD mice spontaneously develop autoimmune diabetes similar to human type 1 diabetes (T1D) with immune-mediated destruction of β cells in the pancreas (17–20). Additionally, genetic variation influences cell lineage decisions during T cell development, affecting positive and negative selection processes and contributing to the emergence of pathogenic CD4⁺ T cells in diabetes-prone NOD mice (16). Both CD4⁺ and CD8⁺ T cells play critical roles in pathogenic diabetogenesis, in which CD4⁺ T cells drive disease onset and CD8⁺ T cells mediate β cell destruction (21). However, little is known about the factors affecting CD8⁺ T cell activity, and the extent to which CD8⁺ T cell reactivity to self-pMHC during thymic positive selection affects its responses to autoantigens during autoimmune processes remains unclear.

We hypothesize that the spectrum of CD8⁺ T cell diabetogenicity correlates with its expression level of CD5, representing the gratitude of self-pMHC-mediated TCR signaling. In this study, we use Ptpn22 (22–24) and islet antigen-specific 8.3 TCR transgenic (NOD8.3) (25, 26) mice in addition to conventional NOD mice to investigate this hypothesis. The Ptpn22 gene encodes PEST domain-enriched tyrosine phosphatase (Pep), which negatively regulates TCR signaling. We have previously demonstrated that different modulation of Pep in effector and regulatory CD4⁺ T cells leads to attenuation of autoimmune diabetes in distal Lck promoter-driven Pep transgenic NOD mice (dLPC/NOD) (27). Here, we crossed dLPC/NOD to NOD8.3 transgenic mice to generate dLPC/NOD8.3 double transgenic mice and further investigate whether transgenic Pep-mediated protective potential in CD4⁺ T cells with lower TCR signaling (27) can also be applied to an autoantigen-specific CD8⁺ T cells with higher CD5 and TCR basal signaling in 8.3 TCR transgenic mice. Our data reveal that the activation and proliferation of CD8⁺ T cells from dLPC/NOD8.3 mice was not attenuated, compared to CD8⁺ T cells from NOD8.3 mice, suggesting that a dominant role of autoantigen reactivity in CD8⁺ T cell with high TCR basal signaling and CD5 overrides the transgenic Pep-mediated signaling attenuation during autoimmune diabetogenesis. Exploring a potential modulation of CD5-dependent TCR basal signaling and its impact on T cell reactivity may offer a promising therapeutic strategy for the treatment of T1D in the future.

Results

Phenotypic difference exists in naïve CD5^hiCD8⁺ and CD5 ^loCD8⁺ T cells of NOD mice

Given that CD5 can serve as a surrogate marker of TCR basal signal strength, which correlates with T cell’s self-recognition capability, our initial analysis aimed to determine whether naïve (CD44^loCD62L^hi) CD8⁺ T cells from NOD mice with different CD5 levels (CD5^hiCD8⁺ and CD5^loCD8⁺ T cells) display distinct levels of the TCR basal signal markers p-CD3ζ and p-Erk. The histograms and gMFI analysis revealed significantly increased levels of p-CD3ζ and p-Erk in CD5^hiCD8⁺ T cells, in comparison to CD5^loCD8⁺, while the amounts of p-CD3ζ and p-Erk in total naive CD8⁺ T cells were generally lied between the CD5^hi and CD5^lo subsets (Figure 1, A and B, and Supplementary Figure 1A) in both spleen and pancreatic lymph nodes (PLNs). This result implies an enhanced signaling potential and activation status of CD5^hiCD8⁺ T cells, compared to CD5^loCD8⁺ cells, in NOD mice. Since two critical T-box transcription factors, T-bet and Eomes, are regulators of differentiation and effector functions of CD8⁺ T cells (28–30), we further investigated whether intrinsic differences in self-recognition capabilities and subsequent functional heterogeneity between CD5^hiCD8⁺ and CD5^loCD8⁺ populations correlate the expression levels of T-bet and Eomes. Our results indicated higher levels of T-bet and Eomes in the naïve CD5^hiCD8⁺ population in both spleen and PLNs, compared to CD5^lo counterparts (Figure 1C and Supplementary Figure 1B). Elevated expressions of T-bet and Eomes in CD8⁺ T cells have been reported to play critical role in promoting T cell migration to inflamed tissues through induction of chemokine receptors (31, 32) in addition to their association with the differentiation of activated CD8⁺ T cells (28). These upregulated T-bet and Eomes in naïve CD5^hiCD8⁺ T cells (Figure 1C) were consistent with elevated levels of Granzyme B, TNF-α, IFN-γ and IL-2 (Figure 1D and Supplementary Figure 1C), indicating the potential of these cells in cytokine production and effector functions. Additionally, CD5^hiCD8⁺ subset expressed higher level of chemokine receptor CXCR3 (Figure 1E) than CD5^loCD8⁺ counterparts, a characteristic often associated with effector and memory T cell populations, implying a enhanced ability for migration and localization in response to specific chemotactic signals.

Figure 1

Building upon our observations of the phenotypic and functional differences between naïve CD5^hi and CD5^loCD8⁺ T cell subsets in NOD mice (Figure 1, A-D), as well as previous research highlighting T-bet- and Eomes-related regulatory mechanisms within CD8⁺ T cell populations (3, 28), we next sought to investigate whether CD5-associated self-recognition capabilities influence specific gene regulation in CD8⁺ T cells via distinct levels of T-bet and Eomes. It is well documented that T-bet and Eomes work together to control CD122 expression, a common β chain receptor for IL-2 and IL-15, which is crucial for the survival of IL-15-dependent memory CD8⁺ T cells (30, 33). Additionally, CD122, in cooperation with CD127 (IL-7Rα), represents essential receptors for IL-2, IL-15, and IL-7. They play a pivotal role in regulating CD8⁺ T cell subsets during the memory phase of the immune response (34–37), which is crucial for maintaining memory T cell homeostasis. As shown in Figure 1E, higher expression of CD122 and CD127 was observed in the CD5^hiCD8⁺ population, supporting that the overall phenotypic traits of naïve CD5^hiCD8⁺ T cells resemble those of memory CD8⁺ T cells (Figure 1E), particularly in their ability to maintain T cell homeostasis.

In addition to distinct expression patterns of these memory T cell-related homeostasis markers, like CD122 and CD127 (Figure 1E), we also identified higher expressions of other T cell activation markers, including CD69 (38, 39), PD-1 (40, 41), and KLRG-1 (37, 42), in naïve CD5^hiCD8⁺ T cells compared to their CD5^lo counterparts (Figure 1F). Similar to PD-1, the activation marker CD44, an adhesion molecule and marker for antigen-experienced T cells, was up-regulated in the CD5^hiCD8⁺ population (Figure 1F), as previously reported (3, 34, 43, 44). Additionally, CD5^hi cells also exhibit elevated expression of CD25 (IL-2Rα), supporting the presence of intrinsic phenotypic distinction between the CD5^hi and CD5^lo populations in NOD naïve CD8⁺ T cells (Figure 1F). Since the increased number of islet antigen IGRP-specific CD8⁺ T cells positively correlate with insulitis progression both in human T1D and NOD mice, we thus used MHC-tetramer technique to trace IGRP_206–214-specific CD8⁺ T cells in NOD mice. Our results revealed that CD5^hiCD8⁺ T cells compared to CD5^loCD8⁺ counterpart had significant increase in IGPR-tetramer staining level (Figure 1G), suggesting that CD5^hiCD8⁺ T cells with higher self-antigen recognizing ability contribute more to islet-destructive pathogenesis in T1D. Collectively, in line with previous findings, our results indicate that the phenotypic characteristics of naïve CD5^hiCD8⁺ T cells resemble those of effector/memory T cells and confer distinct functional capabilities in autoimmune diabetogenesis compared to CD5^loCD8⁺ cells.

Differential gene expression profiling of naïve CD8⁺ T cells stratified by CD5 levels identifies autoimmune disease-associated gene signatures

To unveil the underlying differences in phenotypic characteristics between CD5^hiCD8⁺ and CD5^loCD8⁺ T cells, we analyzed the gene expression profiling by sorting the 10% upper and lower CD5-expressing naïve CD8⁺ T cells from prediabetic mice at the age of 6 to 8 weeks old (Figure 2A). We performed differentially expressed genes (DEGs) analysis using multiple platforms and bioinformatic tools to compare the transcriptomic profiles of CD5^hi versus CD5^lo T cells. Our DEGs results showed that a total of 185 genes were found to be differentially expressed between CD5^hi and CD5^lo, with a fold change greater than 2 and a p-value less than 0.05 cutoffs; within these genes, 133 were upregulated and 52 were downregulated (Supplementary Table 1 and 2). The significantly upregulated DEGs revealed by the volcano plot (Figure 2B) in CD5^hi subset include memory T cell homeostasis marker Il2rb (CD122), T cell effector/memory function-associated markers Tbx21, Eomes, and chemokine receptor Cxcr3, consistent with the results obtained in Figure 1. Interestingly, the Xcl1 gene, which encodes the ligand XCL1 exclusively binding to XCR1 on dendritic cells and facilitating antigen-specific CD8⁺ T cell expansion, IFN-γ secretion and cytotoxicity, was also upregulated in the CD5^hi population (28, 45, 46), implying that CD5^hiCD8⁺ T cells are poised for efficient priming upon antigen encounter.

Figure 2

To further characterize DEGs for their functional implications, we performed gene set analysis based on the gene ontology (GO). A gene-concept network by CD5^hi versus CD5^lo DEGs was associated with the three GO terms: external side of plasma membrane (GO:0009897), T cell activation (GO:0042110), and cytokine production involved in immune response (GO:0002367) (Supplementary Figure 2A). Among the notable findings, genes such as Cd5, Cd83, Eomes, Xcl1, Tbx21, Il18rap, Klra1, Cxcr3, and Pdcd1 were significantly upregulated within CD5^hiCD8⁺ T cells, with fold changes equal to or greater than 2. Specifically, within the GO term cytokine production involved in immune response, genes such as Crlf2, Kit, F2rl1, Xcl1, Tbx21, Tnfrsf1b and Il18rap showed upregulated expression patterns. Overall, these results provide insights into the potential mechanisms underlying T cell activation and cytokine production in CD5^hi population.

To thoroughly analyze the intrinsic kinetics and expression profiles of CD5^hiCD8⁺ versus CD5^loCD8⁺ T cells under conditions resembling antigen encounter, we cross-referenced our findings with a dataset that delineates the functional relevance of ten gene clusters established in a prior study (47). This study identified core transcriptional signatures governing CD8⁺ T cell activation and effector/memory cell differentiation during infections, defining ten temporal gene-regulated patterns (clusters I∼X) in CD8⁺ T cells. Using a χ2 test to examine gene expression variations in these ten clusters between CD5^hi and CD5^lo cells (Table 1), we found that the CD5^hi subset has a significantly higher expression of genes linked to the initial cytokine or effector response (cluster I) and preparation for cell division (cluster II). Additionally, it exhibits a moderately higher expression of genes associated with cell cycle and division (cluster III), early effector and late memory (cluster V), short-term effector or memory (clusters IX), and late effector or memory (clusters X) in the CD8⁺ T cell response (Table 1 and Supplementary Figure 2B). Collectively, these data support that the CD5^hi T cells are better prepared for initial activation and late memory formation compared to the CD5^lo population.

Table 1

In addition, we performed gene set enrichment analysis (GSEA) on DEGs using the Broad Institute’s GSEA desktop software. GSEA analysis indicate that CD5^hi cells were enriched for GO terms such as T cell activation (GO:0042110), external side of the plasma membrane (GO:0009897), positive regulation of immune system process (GO:0002684), regulation of catalytic activity (GO:0050790) and other GO terms with similar biological process (BP) functions (Figure 2C, Supplementary Figure 2C and Supplementary Table 3), similar to the results in Supplementary Figure 2A. Collectively, these results of cluster analysis (Table 1) and GSEA (Figure 2C) indicate that CD5^hiCD8⁺ T cells in NOD mice possess a greater inherent potential for T cell activation and memory cell formation, which may contribute to the initiation and persistence of autoimmune diabetes.

Moreover, we analyzed DEGs by applying Kyoto Encyclopedia of Genes and Genomes (KEGG) database to further investigate whether CD5^hiCD8⁺ subset is enriched in any potential disease-related pathways. Strikingly, two pathways stood out: the inflammatory bowel disease (IBD)- and type 1 diabetes mellitus-associated pathways (Figure 2D). The enrichment results for IBD and TID pathways suggest that naïve CD5^hiCD8⁺ T cells with heightened self-reactivity have the intrinsic capacity to induce T1D and IBD. In summary, our data from gene-concept interactions, GSEA and KEGG pathways reveal distinct phenotypes in CD5^hiCD8⁺ subset, indicating their potential for autoimmune predisposition even in their naïve state.

CD5^hiCD8⁺ and CD5^loCD8⁺ T cell subsets have distinct activation and proliferation levels

To examine whether variations in self-recognition, as indicated by CD5 expression, determine diabetogenic CD8⁺ T cell proliferation after TCR stimulation, we isolated naïve CD8⁺ T cells and subjected them to 2 days of dose-dependent TCR stimulation using various concentrations of anti-CD3 in combination with a consistent concentration of anti-CD28 (Figure 3A). Subsequently, we characterized phenotypic and functional distinctions within the 5% upper and lower CD5-expressing subsets of CD8⁺ T cells. Unlike the lower 5% CD8⁺ groups showing moderate change, the upper 5% CD8⁺ groups exhibited much more significant increase in CD5 levels in TCR stimulation dose-dependent manner (Figure 3B). The CD5^hiCD8⁺ subset also showed higher p-CD3ζ abundance compared to CD5^loCD8⁺ subset across all TCR stimulation doses. Specifically, p-CD3ζ in CD5^hiCD8⁺ peaked at 1.25 µg/ml anti-CD3 stimulation, then gradually declining with increased TCR stimulation doses, possibly due to its faster kinetic expression compared to other T cell activation signaling molecule p-Erk (Figure 3, C and D). Higher p-CD3ζ and p-Erk expressions in NOD CD5^hiCD8⁺ cells compared to CD5^loCD8⁺ counterparts, consistent across TCR stimulation doses, align with the previous findings in B6 mice (2, 5). The upper 5% CD5^hi cells consistently exhibited elevated levels of T cell activation markers CD69 and CD44 (Figure 3, E and F) and showed dose-dependent enhanced proliferation under TCR stimulation (Figure 3G), along with their increased CD127 level (Figure 3H), suggesting heightened capabilities in memory T cell homeostasis.

Figure 3

In Figure 3, C-H, we exclusively evaluated the phenotypic and functional differences between CD5^hi and CD5^lo subsets from activated bulk CD8⁺ T cells following TCR stimulation. To investigate whether these distinctions post TCR activation stem from intrinsic differences between CD5^hi and CD5^lo cells, we specifically sorted the upper and lower 10% of CD5-expressing CD8⁺ T cells from prediabetic NOD mice aged 6 to 8 weeks, aiming to ascertain whether sorted naive CD5^hi and CD5^lo T cells maintain these intrinsic differences in their immune responses after TCR or downstream signaling stimulation (Figure 3, I-K). The results revealed that CD5^hi cells, after TCR stimulation, displayed increased levels of the T cell activation marker CD69 (Figure 3I), accompanied by higher Ki-67 expression and IFN-γ production (Figure 3J), reflecting their intrinsic distinctions in heightened effector and proliferation functions. Notably, in our T cell stimulation experiment comparing CD5^hi and CD5^loCD8⁺ T cells treated with anti-CD3/CD28 dynabeads, we observed superior activation potential in CD5^hi cells (Figure 3I). After 4 hours of stimulation, CD69 expression in CD5^hi cells was significantly higher than in CD5^lo cells, with a sharp distinction. As the stimulation time extended to 24 hours, CD69 levels in CD5^hi compared to CD5^lo remained higher, although the difference reduced, and CD69 expression in both populations became closer. In comparison, with PMA/Ionomycin treatment bypassing TCR signaling initiation, there were almost no distinctions in CD69 levels between CD5^hi and CD5^lo populations, implying cell activation differences between CD5^hi and CD5^loCD8⁺ T cells primarily through TCR-proximal signaling. The DNA analog [methyl-3H] thymidine incorporation assay further confirmed the heightened proliferation capacity in CD5^hi subset compared to CD5^lo subset (Figure 3K). Additionally, we compared CD5 levels between CD8⁺CD69⁺ and CD8⁺CD69⁻ populations in both the spleen and PLNs. In PLNs, a lymphoid tissue enriched with autoantigens, CD69⁺CD8⁺ T cells were skewed toward expressing higher CD5 levels compared to the counterparts in spleen (Figure 3L), implying that cells with high CD5 level display elevated activation potential upon encountering antigen. In summary, these results indicate that diabetogenic CD8⁺ T cells exhibiting increased activation and proliferation profiles in NOD mice tend to have elevated CD5-associated auto-immune responses.

The CD5 expression level in naïve CD8⁺ T cells is positively correlated with diabetes susceptibility in NOD mice

We further explored potential differences in in vivo immune response of CD5^hi versus CD5^loCD8⁺ T cells, especially their abilities to induce insulitis and diabetes through an adoptive T cell transfer experiment in NOD model. Following the similar sorting method as in Figure 3, I-K, the naïve CD5^hiCD8⁺ and CD5^loCD8⁺ cells from prediabetic NOD mice were then administered separately to immunodeficient NOD Rag1^−/− recipients via intraperitoneal injection. The diabetes incidence in Rag1^−/− mice was weekly monitored after transfer. Strikingly, the CD5^hi-transferred group exhibited markedly greater diabetes onset kinetics compared to CD5^lo group, with the appearance of diabetes as early as 4 weeks post-transfer. All 10 mice in CD5^hi group experienced earlier diabetes onset, in contrast to the CD5^lo-transferred group, where only 4 out of 11 mice exhibited diabetes after approximately 20 weeks post-transfer (Figure 4A, left panel). Moreover, histological analysis and insulitis scoring of pancreata after the 4-week transfer period revealed heightened severity of insulitis, characterized by profound lymphocytic infiltration in the CD5^hiCD8⁺ compared to CD5^loCD8⁺ group (Figure 4A, right panel). This suggests that CD5^hiCD8⁺ cells possess more potent diabetogenic properties during the immune responses in autoimmune diabetes.

Figure 4

Moreover, flow cytometry analysis revealed a significant increase in both the percentage and the number of CD8 T cells in the spleen and PLNs of the CD5^hi-transferred group compared to their CD5^lo counterparts (Figure 4B), implying superior in vivo proliferation by the CD5^hi subset, consistent with the observations in in vitro proliferation assay (Figure 3K). Additionally, the CD5^hi group exhibited higher percentages of CD8⁺ T cells producing TNF-α, IFN-γ and IL-2 (Figure 4C) and cells expressing the transcription factors T-bet, Eomes and the cytotoxic molecule Granzyme B (Figure 4D). In characterizing other key immune markers between CD5^hi and CD5^lo groups, as shown in Figure 1, E and F, we observed that CD5^hi cells-transferred group in PLNs exhibited high level of CD122 but not in spleen, while also maintaining high CD5 expression (Figure 4E). Furthermore, we assessed the composition of central memory T cells, defined by the CD44^hiCD62L^hi phenotype, and quantified the percentages of CD8⁺ T cells that recognize islet-specific antigen IGRP_206–214 by tetramer staining (IGRP⁺CD8⁺ T cells) in the PLNs (Figure 4F). This analysis underscores that central memory T cell population and the frequency of islet autoantigen-specific CD8 T cells are higher in the CD5^hi subset within the PLNs, implying more robust immune responses initiated by the CD5^hi cells.

The intrinsic high CD5-linked self-reactivity of CD8⁺ T cells attributed to thymic positive selection contributes to the peripheral poised autoimmune phenotypes

Based on our results, phenotypic and functional heterogeneity was observed within naive CD8⁺ T cells with different CD5 level (Figure 1 and 2), and the further sorted CD5^hi and CD5^lo cells displayed differential effector/memory functions (Figure 3 and 4). To validate that the differential autoimmune T cell responses in CD5^hi and CD5^loCD8⁺ T cells are linked to heterogeneity in TCR self-reactivity during positive selection, we examined whether the unique TCR basal signal strength and IL-2 production differs in the lower and upper CD5-expressing thymocytes in NOD mice. Indeed, from the immature CD4⁻CD8⁻ double-negative (DN), CD4⁺CD8⁺ double-positive (DP), to CD4⁻CD8⁺ (CD8SP) stages in 6-8-week-old NOD thymocytes, the high self-reactivity CD5^hi thymocytes exhibited elevated p-CD3ζ, p-Erk and IL-2 expression, supporting their characteristics of survival and proliferation (Figure 5A and Supplementary Figure 3, A and B). These findings are consistent with previous report (2) and may explain a higher CD8⁺ T cell expansion observed in CD5^hiCD8⁺ cell-transferred recipients compared to CD5^loCD8⁺ cell-transferred recipients in the adoptive transfer experiment (Figure 4B). We further explored the potential influence of thymic selection on peripheral T cell responses by using the islet antigen-specific 8.3 TCR transgenic NOD8.3 mice, characterized by higher CD5 expression, compared to that of NOD mice, and thus an enhanced level of self-recognition. NOD8.3 compared to NOD mice consistently showed elevated CD5 expression from DP to CD8SP stages (Figure 5B), and a significant increase in cell number in the CD8SP subset (Supplementary Figure 3, A and C), suggesting an elevated positive selection in NOD8.3 thymocytes. NOD8.3 mice, with elevated CD5 expression linked to basal TCR signal strength (10), showed significantly higher p-CD3ζ level in DP cells compared to NOD mice counterparts (Figure 5C and Supplementary Figure 3D). The increase of p-CD3ζ level in DP cells positively correlates with the alterations in thymocyte development reflected in increased cell number and percentage in the CD8SP subset transitioning from the DP stage of NOD8.3 mice (Figure 5D and Supplementary Figure 3C). Additionally, DN thymocytes, marked by CD44 and CD25, undergo crucial β-selection in DN3 (CD44⁻CD25⁺) and DN4 (CD44⁻CD25⁻), emphasizing the importance of DN stage in thymocyte development (48–51). NOD8.3 mice showed significantly higher cell numbers in DN4 (Figure 5E and Supplementary Figure 3E), suggesting a selection advantage in these stages due to bypassing the need for TCRβ rearrangement occurred in NOD cells.

Figure 5

Positive selection in the DP stage was further characterized by analyzing CD69 and TCRβ levels. In the pre-positive selection population (DP/TCRβ⁻CD69⁻), NOD8.3 cells showed significantly higher CD5 expression compared to NOD cells (Figure 5F, upper panel, and Supplementary Figure 3F), implying an enhanced TCR signal perception at the initiation of DP positive selection (52). Consistently, the frequency of CD69⁺TCRβ⁺ cells, representing post-positive-selection thymocytes (53), significantly increased in NOD8.3 compared to NOD mice (Figure 5F, lower panel), predisposing to an accelerated accumulation of CD8SP cells (Figure 5D and Supplementary Figure 3C). Additionally, to investigate whether the increased accumulation of CD8SP cell number in NOD8.3 mice is associated with elevated expressions of cytokine receptors crucial for T cell survival in DP stage, we assessed the expression levels of CD122, CD127, CD44 and CD25 in DP thymocytes. Indeed, these receptors showed higher levels in NOD8.3 compared to NOD DP thymocytes (Figure 5G), suggesting their contribution to cell survival into the CD8SP stage. Remarkably, NOD8.3 mice showed an increased percentage of memory-like CD122^hiCD44^hi cells (V_M; virtually memory) in CD8SP compared to NOD mice (Figure 5H and Supplementary Figure 3G), implying a direct transition to memory T cells upon entering peripheral tissues (54). Furthermore, CD5 expression patterns were compared across thymocyte stages among different NOD strains including NOD, BDC2.5 and NOD8.3 mice. NOD8.3 thymocytes consistently showed higher CD5 levels compared to two other strains from the DN to DP stages, and notably, in the CD8SP stage, NOD8.3 cells exhibited the highest CD5 expression, followed by NOD and then BDC2.5 cells (Supplementary Figure 4A). Interestingly, NOD8.3 cells displayed enhanced proliferation (Supplementary Figure 4, B and C), suggesting a potential link between perceived TCR basal signals during thymocyte selection and peripheral lymphocyte proliferation.

Downregulation of TCR signaling strength by transgenic phosphatase Pep expression does not confer the protection in dLPC/NOD8.3 mice

Based on the insights gained from thymocyte development in NOD, BDC2.5 and NOD8.3 mice, where the TCR signal strength influences intrinsic T cell response and subsequent effector function, we investigated whether the manipulation on TCR basal signal strength influences autoimmune potential of peripheral T cells in NOD mice. To address this issue, we utilized two transgenic mouse model, dLPC/NOD and dLPE/NOD mice, which express a T cell-specific phosphatase Pep to downregulate proximal TCR signaling (27). A significant decrease in autoimmune diabetes, in terms of a delayed onset, less insulitis severity and lower incidence of disease, was found in dLPC/NOD and dLPE/NOD mice compared to NOD controls (27). As expected, transgenic Pep expression in dLPC/NOD mice delayed the onset of diabetes and reduced insulitis. However, Pep overexpression in dLPC/NOD8.3 mice did not provide protection (Figure 6A), suggesting that intrinsic high CD5-associated self-reactivity in NOD8.3 T cells overrides the transgenic Pep-mediated protection observed in dLPC/NOD mice. Moreover, overexpression of Pep in dLPC/NOD and dLPC/NOD8.3 mice did not change the CD5 expression of T cells in the spleen compared to their controls, NOD and NOD8.3, respectively. Interestingly, we observed a transgenic Pep-mediated CD5 downregulation in CD8⁺ T cells in PLNs, enriched with islet-specific autoantigens, of dLPC/NOD mice compared to NOD controls (Figure 6B), suggesting that the transgenic Pep-mediated TCR signaling strength modulates CD5 expression in a context-specific manner. However, CD5 expression level in CD8⁺ T cells in PLNs of dLPC/NOD8.3 was similar to NOD8.3 mice (Figure 6B), suggesting that NOD8.3 T cells with an intrinsic high CD5-associated self-reactivity are more resistant to transgenic Pep-mediated change in TCR signaling and CD5 expression. These observations are further supported by our data that the CD5-correlated basal TCR signaling molecule, p-CD3ζ, was attenuated in dLPC/NOD mice but not in dLPC/NOD8 .3 mice, both in the spleen and PLNs (Figure 6C). Collectively, our results suggest that CD8⁺ T cells with intrinsic high CD5-linked self-reactivity are more resistant to transgenic Pep-mediated TCR signaling downregulation and sustain the diabetogenic susceptibility in NOD8.3 mice.

Figure 6

Consistently, isolated naïve CD8⁺ T cells (Supplementary Figure 5 A, left panel) from high CD5-expressing NOD8.3 mice exhibited superior proliferation compared to cells from NOD mice upon anti-CD3/CD28 or PMA/Iono stimulation (Supplementary Figure 5A, middle and right panel, respectively). Moreover, transgenic Pep expression attenuated CD8⁺ T cell proliferation in both dLPC/NOD and dLPE/NOD mice compared to NOD controls (Supplementary Figure 5B). However, transgenic Pep expression did not alter cell proliferation in dLPE/NOD8.3 T cells when stimulated with the autoantigen-specific self-peptide IGRP_206-214 (Supplementary Figure 5C), suggesting that a high TCR basal signal conferred by elevated intrinsic self-reactivity overrides the transgenic Pep-mediated reduction of T cell proliferation. Further analysis of p-Erk level also showed a significant decrease in dLPC/NOD compared to NOD mice. Similarly, transgenic Pep did not attenuate the p-Erk level in dLPC/NOD8.3 mice (Figure 6D), consistent with the observation of p-CD3ζ level in these Pep transgenic mice (Figure 6C). Additionally, CD69 and KLRG1 (Figure 6E) and other activation markers including CD44 and CXCR3 (Supplementary Figure 5D) were decreased in dLPC/NOD or dLPE/NOD mice compared to NOD controls. Similar to previous observations, transgenic Pep-mediated attenuation of T cell activation was abolished in dLPC/NOD8.3 mice, supporting again that an intrinsic high CD5-linked self-reactivity is able to confer the impact by transgenic Pep-mediated TCR signaling change.

In addition, transgenic Pep expression downregulated CD25 level in T cells of PLNs, not in spleen, in dLPC/NOD mice compared to NOD mice, whereas there was no significant change in CD25 expression between dLPC/NOD8.3 and NOD8.3 mice (Figure 6F). Moreover, memory T cell homeostasis molecules, such as CD122 and CD44, were also reduced in T cells of PLNs, not in spleen, in dLPC/NOD mice compared to NOD mice (Figure 6F and Supplementary Figure 5D), but were indistinguishable between dLPC/NOD8.3 and NOD8.3 mice (Figure 6F). Notably, transgenic Pep expression in dLPC/NOD mice significantly reduced production of effector T cell cytokines, such as IL-2, TNF-α and IFN-γ in CD8⁺ T cells compared to NOD mice (Figure 6G and Supplementary Figure 5E). Nevertheless, transgenic Pep-mediated reduction in cytokine production was abolished in dLPC/NOD8.3 mice (Figure 6G). Collectively, our results indicate that CD5-associated high self-reactivity is able to abolish transgenic Pep-mediated attenuation of effector/memory T cell functions in NOD8.3 mice.

Analysis of the CD5^hi population identifies TCR repertoires associated with autoimmune diseases

To investigate whether pathogenic feature in CD5^hiCD8⁺ population reflects in their TCR repertoire, we divided naïve CD8⁺ T cells into CD5^hiCD8⁺ and CD5^loCD8⁺ groups, conducted bulk RNA sequencing on two groups, and applied the MiXCR pipeline to analyze TCR repertoire composition in CD5^hi and CD5^lo groups (Figure 7A and Supplementary Figure 6). Analysis of T-cell receptor-α chain (TRAV and TRAJ) and T-cell receptor-β chain (TRBV and TRBJ) gene usages in CD5^hiCD8⁺ and CD5^loCD8⁺ T cells revealed increased utilization of TRAV13-2, TRAV16D-DV11, TRAV3-3, TRAJ32, TRBV14, TRBV16, TRBJ2-1 and TRBJ2-7 in CD5^hiCD8⁺ cells, with the exception of TRAV9-1, which exhibited lower utilization in CD5^hiCD8⁺ T cells (Figure 7, B-E). The CDR3 loops of TRA (CDR3α) and TRB (CDR3β), crucial for antigen recognition by TCRs on T cells, are shaped by the diverse recombination events that assemble TRA VJ gene segments and TRB V(D)J gene segments (55). The distinct utilization of V or J gene families of TCRα and TCRβ, as shown in Figure 7, B-E, may lead to variable CDR3 characteristics between CD5^hiCD8⁺ and CD5^loCD8⁺ T cells. Additionally, emerging evidence suggests a potential association of shorter CDR3 length in T1D patients (56). Therefore, we analyzed whether the CD5^hiCD8⁺ cells have similar pathogenic CDR3 patterns. Interestingly, examination of CDR3α and CDR3β length distribution in the TCR repertoire revealed a bias toward shorter length in the CD5^hi compared to the CD5^lo group, particularly with a significantly shorter CDR3β length (Figure 7F). Shared TCR pools were reported to be enriched in clonotypes with fewer insertions in CDR3 (57). To confirm this, the overlapped clonal count matrix for CDR3α or CDR3β motif was calculated among all groups (Figure 7G), indicating increased similarity between the CD5^hi groups based on its own pairing comparison. Strikingly, the IGRP-recognizing CDR3 motif was found in CDR3α of the CD5^hi group (Figure 7H), which has been identified in the TCR repertoire of T1D patients (58). Moreover, analysis of CDR3α and CDR3β amino acid characteristics revealed a hydrophobicity enrichment in TRA of CD5^hiCD8⁺ cells (Figure 7I), consistent with a previous study demonstrating that the generation of high self-reactive T cells was promoted by hydrophobic residues within the CDR3 region (59). Furthermore, in alignment with a previous study illustrating greater diversity in TCR repertoires from T1D patients than in healthy donors (56), CD5^hiCD8⁺ cells exhibited higher diversity in clonotypes compared to their CD5^lo counterpart (Figure 7J). Additionally, our results align with a previous study that employed high-throughput TCR-seq analyses in mice, identifying highly shared CDR3 sequences among mice characterized by restricted V and J segment usage (60). These CDR3 sequences, exhibiting lower nucleotide insertions and shorter lengths, were more abundant, particularly in individuals displaying autoimmune and allograft-related reactions, further affirming their association with self-antigen reactivity. In conclusion, these results suggest that CD5^hiCD8⁺ T cells, with elevated self-reactivity and increased clonal diversity, are intrinsically linked to the utilization of specific V and J genes within CD5^hi cells.

Figure 7

Discussion

In our study, CD5^hiCD8⁺ T cells, with heightened basal signals and increased self-ligand reactivity, show enhanced avidity for the autoantigen IGRP (Figure 8). These cells transition quickly into effector/memory cells upon encountering autoantigens, expressing markers like CD69 and CD44. Activation via CD122 and CD127 promotes cytokine secretion, contributing to inflammation and β cell destruction. Further investigation is needed to understand the role of these memory-like CD8⁺ T cells in autoreactivity. The loss of Pep-conferred protection in high CD5-expressing NOD8.3 mice highlights the significance of elevated TCR basal signals from positive selection in autoimmune responses, suggesting potential targets for treating T1D and related disorders.

Figure 8

In contrast to previous studies that compared TCR signaling in NOD and wild-type B6 mice (11), we investigate how thymic self-recognition impacts TCR signal strength and CD8⁺ T cell pathogenicity in NOD mice. Our research also differs from studies focusing on the CD4⁺CD5^low T-cell population (12), which resembles central memory T cells and is linked to spontaneous autoimmunity in NOD mice. Instead, we find distinct TCR signal strength variations between CD5^hi and CD5^lo cells, with CD5^hiCD8⁺ cells showing enhanced activation, proliferation and disease transfer potential. This discrepancy may arise from the higher CD5-expressing distribution in the naïve CD4⁺ T cells compared to the naïve CD8⁺ T cells, suggesting distinct TCR signaling sensitivity when stimulated and resulting in different functional activation thresholds between CD4⁺ and CD8⁺ T cells (61, 62). Our results reveal that CD5^hiCD8⁺ T cells exhibit elevated CD25 and CD122 expression compared to their CD5^loCD8⁺ counterparts, indicating increased sensitivity to IL-2 stimulation. While exposure to IL-2 may enhance cell expansion in CD8⁺ cells, it can lead to overstimulation in CD4⁺ T cells, resulting in activation-induced cell death. Surprisingly, clinical cases of type 1 diabetes reflect similar findings to our animal mouse model study. Children with T1D exhibit higher CD5 expression in CD8⁺ T cells compared to healthy controls (63). These observations underscore the complexity of immune responses in autoimmune diseases and warrant further exploration to unravel the underlying mechanisms and hold promise for therapeutic interventions in autoimmune diseases.

Nevertheless, the role of CD5 itself in shaping the distinctive functions of CD5^hi and CD5^lo cells remains largely unclear (64, 65). A previous research has indicated that CD5 serves as a scaffold protein, retaining the NF-κB inhibitor IκBα in the cytosol to prevent its inhibition of NF-κB signaling (66), though studies of Cd5^−/− mice do not appear to support this concept (67). This mechanism confers survival advantages to CD5^hi T cells. This observation may help elucidate the seemingly paradoxical result that CD5’s primary function is to attenuate TCR-proximal signaling; however interestingly, antigen-specific T cells with higher initial CD5 expression exhibit enhanced persistence as effector/memory cells following a peripheral challenge. Whether CD5^hiCD8⁺ T cells execute potent autoimmune responses through the modulation of NF-κB-dependent signaling needs further investigation.

Exploring how TCR composition contributes to the distinct responses between CD5^hi and CD5^lo cells is beyond the scope of our study. Nonetheless, prior research, which involved screening autoimmune disease databases such as arthritis, diabetes, etc., has revealed a trend in specific TCR repertoire usage characterized by germline-encoded TCRs with fewer nucleotide insertions and shorter CDR3 length (60). It is consistent with our results, demonstrating that CD5^hi T cells, skewed toward shorter CDR3 lengths in the TCR repertoire, have a higher potential for autoreactivity. Consequently, the high self-pMHC reactivity of T cells provides a stronger TCR basal signal, potentially priming T cells with specific TCR compositions for robust responses to foreign antigens in the future. From an evolutionary perspective, this TCR design may have initially conferred an advantage by enhancing immune responses against pathogen infections. However, it might have come at the cost of an increased risk of autoimmune diseases. A previous study explored the TCR repertoire of CD8⁺ T cells recognizing the islet-specific peptide IGRP_{206– 214} in both grafted and endogenous islet infiltrates, revealing a shared repertoire favoring specific gene segments and limited diversity and clonotypic expansion within IGRP_206–214-specific CD8⁺ T cells in NOD mice (58). Another study of effector/memory CD4⁺ and CD8⁺ T cells involved in autoimmune rejection of islet grafts in diabetic NOD mice demonstrated a comparable pattern of TCR repertoire diversity (68). Our analysis of TCR gene usage in CD5^hiCD8⁺ and CD5^loCD8⁺ T cells also revealed distinct patterns, suggesting potential functional differences between these subsets. The identified differences included increased usage of TRAV13-2, TRAV16D-DV11, TRAV3-3, TRAJ32, TRBV14, TRBV16, TRBJ2-1, and TRBJ2-7 in CD5^hi compared to CD5^lo cells, while TRAV9-1 was reduced in CD5^hi cells. Notably, previous studies identified specific gene segments associated with autoimmune diseases like Sjögren’s syndrome (TRAV13-2) (69), inflammatory bowel disease, ulcerative colitis (TRBJ2-1 and TRBJ2-7) (70, 71), and type 1 diabetes (TRBV14 and TRBV16) (72), underscoring the importance of understanding T-cell receptor dynamics in autoimmune pathogenesis across various disorders.

Several studies have highlighted the impact of TCR-self-pMHC interactions during positive selection on signaling molecules’ localization, such as p-Erk, particularly in B cells and CD4⁺ T cells (4, 73, 74). Similarly, our CD5^hiCD8⁺ T cells may consistently receive stronger signals, leading to better synchronization and activation of Erk at the cell membrane. However, despite using PMA/Iono to bypass proximal TCR signaling and elicit downstream responses, differences in T cell proliferation response persisted between CD5^hi and CD5^lo or between NOD and NOD8.3 CD8⁺ T cells. This suggests that TCR-self-peptide interaction influences T cells beyond just proximal TCR signaling intensity. There may be pre-existing transcriptional differences influenced by genes involved in T-cell receptor signaling established during thymic development among naïve NOD CD8⁺ T cells, affecting their effector lineage fate (75). Additionally, in a previous experimental autoimmune diabetes model, OVA antigen-specific CD8⁺ T cells with a virtual memory phenotype and high CD5 expression showed a low capacity to induce diabetes (76), contrasting with the pathogenic features of our CD5^hiCD8⁺ cells. This suggests that naïve CD5^hiCD8⁺ T cells in NOD mice may only partially resemble virtual memory T cells, posing intriguing questions requiring further investigation.

Materials and methods

Mice

The mice utilized in this study included NOD/Sytwu, NOD8.3, and NOD.BDC2.5 TCR transgenic mice, alongside Ptpn22-transgenic NOD mice, designated dLck–Ptpn22 C (dLPC/NOD) and dLck–Ptpn22 E (dLPE/NOD), previously generated by cloning Ptpn22 from NOD mouse splenocytes and microinjecting it into single-cell NOD embryos. The resulting dLck–Ptpn22 transgenic mice were genotyped by PCR and evaluated by Southern blot analysis for confirmation. Western blot analysis revealed the expression of transgenic Pep in thymocytes and splenic T cells of dLPC and dLPE mice, with a stepwise increase in Pep protein levels from nontransgenic littermates, dLPC/NOD to dLPE/NOD mice (27). The transgenic mice dLPC/NOD, dLPE/NOD, dLPC/NOD8.3 and dLPE/NOD8.3 used in the study were hemizygous for the Ptpn22 transgene (Supplementary Figure 7). dLPC/NOD8.3 and dLPE/NOD8.3 were established by mating dLPC/NOD or dLPE/NOD mice with NOD8.3 mice. NOD/Sytwu, NOD8.3 and NOD.BDC2.5 TCR transgenic mice were originally obtained from The Jackson Laboratory, while NOD/Rag1^−/− mice were procured from the National Laboratory Animal Center. The study adhered to institutional guidelines and was approved by the National Defense Medical Center Institutional Animal Care and Use Committee.

Sex as a biological variable

Our study exclusively examined female mice because the disease modeled is relevant in females.

Statistics

The log-rank (Mantel–Cox) test was employed for survival curve comparisons. Two-group comparisons utilized a 2-tailed Student’s unpaired t-test, two-sample comparison utilized Welch two sample t-test, and multigroup comparisons utilized one-way ANOVA with Tukey’s post-test. A significance level of P < 0.05 was considered significant.

Study approval

All protocols involving live animals adhered to institutional guidelines and received approval from the Institutional Animal Care and Use Committee at the National Defense Medical Center (Taipei, Taiwan).

Additional methods can be found in the Supplementary Material.

Supplementary material

Supplementary material provided with this study includes Supplementary Figures 1-7, Supplementary Tables 1-6 and Supplementary methods.

Data availability

The TCR repertoires of CD5^hi and CD5^lo, along with the data analyzed using R and the corresponding codes, are available in the following GitHub repository: https://github.com/Chia-Lo/TCR-signal-strength-modulates-antigen-specific-CD8-T-cell-pathogenicity-in-non-obese-diabetic-mice.git

Author contributions

CLH and LTY performed experiments and analyzed data. CLH, LTY, YWL and JLD performed experiments. LTY gave advice. CLH and HKS wrote the manuscript.

Acknowledgements

This work was supported by the Ministry of Science and Technology of the Republic of China (MOST 109-2320-B-400-018-MY3, MOST 110-2320-B-400-011-MY3), National Science and Technology Council (NSTC 112-2320-B-400-026-MY3) and Tri-Service General Hospital (TSGH-C02-112029, TSGH-C03-113037, VTA112-T-1-1, VTA113-T-1-1). The authors acknowledge the technical services provided by Instrument Center of National Defense Medical Center.

Abbreviations

• dLPC/NOD: Lck distal promoter– Ptpn22 C line in NOD background

• dLPE/NOD: Lck distal promoter– Ptpn22 E line in NOD background

• NOD: non-obese diabetic