T cells are present in non-diabetic islets and accumulate during aging

Background The resident immune population of pancreatic islets has roles in islet development, beta cell physiology, and the pathology of diabetes. These roles have largely been attributed to islet macrophages, comprising 90% of islet immune cells (in the absence of islet autoimmunity), and, in the case of type 1 diabetes, to infiltrating autoreactive T cells. In adipose, tissue-resident and recruited T and B cells have been implicated in the development of insulin resistance during diet-induced obesity and aging, but whether this is paralleled in the pancreatic islets is not known. Here, we investigated the non-macrophage component of resident islet immune cells in islets isolated from C57BL/6J male mice during aging (3 to 24 months of age) and following diet-induced obesity (12 weeks 60% high fat diet). Immune cells were also examined by flow cytometry in cadaveric non-diabetic human islets. Results Immune cells comprised 2.7 ± 1.3% of total islet cells in non-diabetic mouse islets, and 2.3 ± 1.7% of total islet cells in non-diabetic human islets. In 3-month old mice on standard diet, B and T cells each comprised approximately 2-4% of the total islet immune cell compartment, and approximately 0.1% of total islet cells. A similar amount of T cells were present in non-diabetic human islets. Islet T cells were comprised of CD8-positive, CD4-positive, and regulatory T cells. Interestingly, while islet B cells and macrophage numbers were unaltered by age, the number of islet T cells increased linearly (R2=0.9902) with age from 0.10 ± 0.05% (3 months) to 0.38 ±0.11% (24 months) of islet cells. This increase was uncoupled from body weight, and was not phenocopied by a degree similar weight gain induced by high fat diet in mice. Conclusions This study reveals that T cells are a part of the normal islet immune population in mouse and human islets, and that they accumulate in islets during aging in a body weight-independent manner. Though comprising only a small subset of the immune cells within islets, islet T cells may play a role in the physiology of islet aging.


RESULTS 69
Glucose tolerance and insulin secretory capacity increase in advanced-age mice. We first examined 70 the impact of aging from 3 to 24 months on glucose metabolism in male C57BL/6J mice. Body weight 71 increased with age up to 12 months, but was reduced in 24-month old mice (36.1 ± 4.1 g), to a level 72 comparable to 6-month old mice (35.4 ± 2.1 g) ( Figure 1A). A decline in body weight at 2 years of age is 73 consistent with previously published observations across many inbred mouse strains (33). Non-fasted and 74 6-hour fasted blood glucose levels were not altered by aging ( Figure 1B,C). Aging to 6 and 12 months 75 corresponded with moderately impaired glucose tolerance relative to 3-month old mice ( Figure 1C-D), 76 and aging to 24 months substantially lowered glucose excursions compared to all other age groups, 77 including 3-month old mice. This finding is consistent with previous reports that, correlating with body 78 weight, glucose tolerance is improved in advanced-aged male mice (34,35). Insulin tolerance tests 79 indicated that insulin sensitivity cannot account for the improvement in glucose tolerance in 24-month old 80 mice ( Figure 1E). Fasted plasma insulin levels increased with age up to approximately 2.7 fold at 12 81 months ( Figure 1F), consistent with previous observations in both mice and humans (34,36). In 24-month 82 old mice, fasted insulin levels were similar to those of 3-month old mice. Ex vivo insulin secretion from 83 isolated islets in low glucose conditions tended to increase with age up to 12 months, and glucose-and 84 KCl-stimulated insulin secretion increased with age, with 24-month old mice displaying the highest 85 stimulated responses compared to all other age groups ( Figure 1G,H). Islet insulin content was not altered 86 in aged mice ( Figure 1I). The increased glucose-and KCl-stimulated insulin secretion in 24-month old 87 mice suggests that stimulus-coupled insulin secretion is augmented with age, potentially accounting for 88 improved glucose tolerance in advanced-aged mice, and supporting previous findings that aged mice have 89 increased insulin secretion independent of insulin content (37,38). 90 T cells accumulate in aging islets. We next assessed the effect of aging on islet-resident immune 91 populations. Islets from C57BL/6J mice of the four age groups were isolated and hand-picked to purity. 92 Islets from a minimum of five mice were pooled per sample, to obtain a sufficient number of islet 93 immune cells, and dispersed into single-cell suspensions for analysis by flow cytometry (Figure 2A). Islet 94 cells were gated on singlets and viability prior to analysis of immune cell populations ( Figure 2B). 95 Immune cells (CD45+) accounted for 2.7 ± 1.3% of viable islet cells in 3-month old mice to 3.5 ± 0.9% in 96 24-month old mice (Fig. 2C). As we were initially interested in resident islet B-cell populations, islet cells 97 were additionally stained for CD19, along with CD23 and CD5 to differentiate B-cell subsets. 98 Approximately 90% of islet immune cells were negative for CD19 and CD5 ( Figure 2B), consistent with 99 reports that the vast majority of resident islet immune cells are macrophages (10,11,39). CD19+ cells 100 were negative for CD5, consistent with B2 cells (Figure 2B), and present at a frequency of approximately 101 0.1% of islet cells, which was not altered by age ( Figure 2D). 102 Surprisingly, in addition to B cells, we observed that islets contained a distinct population of 103 CD19-CD5+ cells ( Figure 2B). CD19-CD5+ cells comprised 3.9 ± 0.9% of islet immune cells in 3-month 104 old mice, had a side scatter profile (SSC low ) consistent with lymphocytes, and accumulated with age 105 ( Figure 2E-G). As CD5 expression is limited to B1a cells and T cells, we suspected that these were islet-106 resident T cells. Indeed, CD19-CD5+ cells expressed the T cell co-receptor, CD3, in all age groups 107 ( Figure 2F). Islet T cells showed a positive linear correlation with age, comprising 0.10 ± 0.05% of islet 108 cells at 3 months and rising approximately 4-fold to 0.38 ±0.11% of islet cells at 24 months of age 109 (R 2 =0.9902, p=0.0049) ( Figure 2G-H). This increase was due to a specific accumulation in T cells, as the 110 frequency of T cells relative to islet immune cells also increased with age from 3.9 ± 0.9% of CD45 cells 111 to 11.0 ±1.8% of CD45+ cells (R 2 =0.9059, p=0.048) ( Figure 2I-J). 112 We next assessed the characteristics of islet T cells by FACS sorting islet cells from an additional 113 cohort of 3-and 24-month old mice. Single, viable islet cells were gated on CD45 and subsequently on 114 CD19 and CD3 staining ( Figure 3A). Again, CD3+ cells were significantly increased in the islets of 24-115 month old mice compared to 3-month old mice (0.61 ± 0.08% vs 0.08 ± 0.03% of islet cells) ( Figure 3A-116 B) whereas there were no differences in islet B cells or CD3-CD19-cells in islets with age ( Figure 3B). 117 There was no increase in CD3+ cells in the spleens of aged mice, consistent with previous reports (30) (Supplemental Figure 1A). Cell profile analysis (Nsolver4.0) of sorted islet CD3+ cells confirmed a 119 statistically significant T cell transcript profile (Supplemental Figure 1B). Furthermore, islet CD3+ cells 120 expressed multiple T cell specific transcripts, including Cd3d, Cd3e, Cd3z, Cd4 and Cd8, none of which 121 were significantly altered by age ( Figure 3C), suggesting that similar islet T cell subsets persist across age 122 groups. We subsequently examined islet T cell subsets from approximately 1 year-old mice by flow 123 cytometry ( Figure 3D). CD19-CD3-cells comprised the majority of islet immune cells, and were positive 124 for CD11b, consistent with islet macrophages. Islet CD3+ cells contained a mixture of CD8+, CD4+ and 125 regulatory T cell subsets: 29.3 ± 0.3% were CD8+ and 38 ± 3% were CD4+, and 18.4 ± 0.3% of CD4+ 126 cells were FoxP3+, comprising 7.1 ± 0.6% of the islet T cell population. An additional 32 ± 2% of T cells 127 were double negative (CD8-CD4-). This was distinct from splenocyte T cell subsets, which were 128 comprised of 45.9% CD8+ cells and 42.7% CD4+ cells, 10.8% of which were FoxP3+ (Supplemental 129 Figure 1C). 130

T cells are present in non-diabetic human islets. To determine whether the presence of T cells in non-131
diabetic islets was generalizable to humans, we performed flow cytometric analysis on human islets from 132 3 cadaveric, non-diabetic human donors (Supplemental Table 2). Similar to mice, all human islet 133 preparations contained a small proportion of CD45+ cells, comprising approximately 2.3 ± 1.7 % of islet 134 cells ( Figure 3E). Furthermore, all samples contained a population of CD3+ cells within pancreatic islets, 135 at a frequency of 0.070 ± 0.043% of total live islet cells ( Figure 3E), comparable to frequencies we 136 observed in young adult mice. 137

Islet T cells are not increased by diet-induced obesity.
To determine whether the accumulation of intra-138 islet T cells was specific to aging, or associated generally with obesity or increasing body weight, we next 139 examined islet T cells in diet-induced obese mice. Mice were fed a high fat diet (HFD) for 12 weeks (up 140 to 4 months of age) and compared to age-matched, low fat diet (LFD)-fed mice. This duration of HFD 141 increased body weight by 25% relative to LFD-fed controls (39 ± 3 vs 31 ± 2 g) ( Figure 4A), a 142 comparable increase to the ~28% increase observed in 6-month relative to 3-month old mice (34 ± 2 g vs 144 (43 ± 6 g) and 24-month (36 ± 4 g) old aged-mice ( Figure 1A). Non-fasted and fasted blood glucose were 145 not altered by HFD relative to LFD ( Figure 4B-C), though glucose tolerance was robustly impaired, as 146 expected ( Figure 4C-D). HFD-fed mice were also insulin resistant compared to LFD-fed controls ( Figure  147 4E-G), and had a ~2.6 fold increase in plasma insulin levels ( Figure 4H), similar to the ~2.7 fold increase 148 in insulin in 12-month relative to 3-month old mice ( Figure 1F). Thus, HFD achieved similar increases in 149 body weight and insulin levels to 12 months of aging but resulted in a more dramatic impairment in 150 glucose tolerance. T and B cells were present in islets at frequencies of ~1% of immune cells in LFD-151 control mice, comparable to frequencies in 3-month old mice ( Figure 4I-J). Despite comparable increases 152 in body weight and insulin to 12 months of aging, HFD-fed mice showed no increase in islet CD3+ cells. 153 The frequency of CD45+ cells in islets, as well as the number of CD19-CD3-cells was also unaffected by 154 12 weeks of high fat diet ( Figure 4J). Collectively, these data show that T cells are present in islets in the 155 absence of autoimmune diabetes, and that there is a distinct increase in islet T cells during aging that 156 cannot be accounted for by obesity. 157

DISCUSSION 158
In this study we found that T cells contribute to the normal islet immune cell repertoire in non-159 diabetic mice and humans. While macrophages comprise the majority of islet immune cells (10)(11)(12)(13)(14)40), B 160 and T cells each comprise up to approximately 5% of islet immune cells in young adult mouse islets. Islet 161 T cells were comprised of a mixture of CD4+, CD8+ and FoxP3+ subsets. We also found that islet T cells 162 accumulated within islets during aging in mice. Islet T cells increased by 4-to 8-fold from 3 to 24 months 163 of age, and this increase was specific to T cells, with no changes in islet macrophages or B cells. 164 Moreover, an age-related accumulation of T cells was not observed in spleen, suggesting this 165 phenomenon is islet-specific. In addition, the particular frequencies of T cell subsets present within islets 166 appeared distinct from those in the spleen. Finally, islet T cells were not increased by similar increases in 167 body weight brought about by diet-induced obesity in young mice, suggesting that islet T cell accumulation is a distinct process of aging. Thus, we postulate that T cells play a role in islet biology (in 169 the absence of autoimmune diabetes), and particularly, in adaptive changes to the pancreatic islet during 170 aging, though a functional role for islet T cells was not tested in this study. 171 Our findings contrast some studies that have claimed a lack of T cells in non-diabetic mouse 172 islets. Though scarce relative to islet macrophages, our study clearly shows a small, yet persistent is observed as a percent of total islet cells. Despite the paucity of T cells within pancreatic islets, these 180 cells may still produce sufficient levels of local cytokines within the islet to influence the islet 181 environment. Such is the case for islet macrophages, comprising only 2-10 cells per islet, which are the 182 major source of islet IL-1β (6,15), and play key roles in islet development and beta-cell adaptation and 183 expansion (15)(16)(17)(18)(19)40,44). Islet-resident group 2 innate lymphoid cells, another rare immune population in 184 islets, have also been shown to influence insulin secretion, indirectly via islet macrophages (42). These 185 examples underline how a small number of cells can have a substantial impact on islet physiology. 186 The increase in T cellsbut not other immune cellswithin aging islets, suggests these cells 187 may play a specific aging-related role in pancreatic islets. Furthermore, the aging-induced accumulation 188 of islet T cells is not recapitulated by a similar degree of weight gain in diet-induced obese mice, and thus 189 is not driven by obesity. This parallels the divergent immune cell profiles in adipose tissue of aged versus 190 obese mice (31). Similarly, the metabolic outcomes of aging and diet-induced obesity in mice are 191 different, with aging resulting in increased insulin secretion and ultimately improved glucose tolerance in 192 advanced age mice, whereas diet-induced obesity impairs insulin secretion and glucose tolerance. Thus, the immunologic and metabolic state of aging is distinct from that of obesity, and it is plausible that, like 194 in adipose tissue, T cell accumulation in islets contributes to the regulation of glucose homeostasis during 195 aging. The function of islet T cells during aging, was not examined in this study due to the large number 196 of mice that would be required, but warrants further investigation. Indeed, the correlation between the 197 progressive increase in insulin secretion during aging, observed in this study and others (34,35,37,38,45), 198 and the progressive accumulation of islet T cells reported here, is intriguing. Interestingly, in pre-diabetic 199 NOD mice, the autoreactive T cells that home to islets have been shown to promote beta cell proliferation 200 through production of cytokines IL-2, IL-6, IL-10, CCL3, and CCL5 (46). The rise in islet T cells may 201 also be indicative of an immune response to senescent beta cells, which accumulate in insulin resistant 202 states including aging and obesity (47). 203

CONCLUSIONS 204
Collectively, this study demonstrates that T cells are part of the normal immune population of pancreatic 205 islets in non-diabetic mice and humans, and that their numbers accumulate during aging in mice. The and dispersed for flow cytometry as previously published (10). Briefly, islets were harvested from mice 233 and hand-picked to purity, and subsequently exocrine-free islets were selected and dispersed with 0.02 % 234 trypsin for 3 min at 37 °C. Dispersions were stopped by addition of 7 mL of FACS buffer containing 2% 235 FBS on ice. Dispersed cells were incubated with FcR block (Thermo Fisher Scientific), 10 min prior to 236 addition of antibodies (Supplemental Table 1 Table 2) were supplied by the Alberta Diabetes Institute (Edmonton, Canada). Upon receipt, islets were 264 hand-picked to 99% purity, and were cultured in CMRL (supplemented with 100 U/mL penicillin, 100 265 µg/mL streptomycin, 0.05 mg/mL Gentamicin, and 2 mmol/L glutamax) containing 11 mM glucose at 266 ITTs) data were analyzed by two-way repeated measures ANOVA with Geisser-Greenhouse correction 276 for non-sphericity. Data in all figures and text are represented as mean ± SD unless otherwise specified. 277

DECLARATIONS: 286
Availability of data and materials: Open source data on each human islet preparation can be obtained at 287 www.isletcore.ca. 288 Competing interests: HCD is a scientific advisor for and owns restricted shares in Integrated 289 Nanotherapeutics. CBV is a director of and owns restricted shares in Integrated Nanotherapeutics. SM, 290 SSL and FP have no potential conflicts to disclose.

. Characterization of islet T cells in non-diabetic mice and humans.
A-C) Islets were isolated from 3-and 24-month old mice. Data represent 3 samples per age group, with islets from 5-7 mice pooled per sample. A) Islet cells gated on FSC, SSC, viability (7AAD-) and CD45+, representative of 3 independent samples. B) Quantification of CD45+ cells, and immune cell subsets in islets from 3-and 24month old mice, data represent mean ± SD, and were analyzed by unpaired T-test. C) Transcript abundance of sorted CD3+ islet cells, in 3-month samples (black) and 24-month samples (blue), data represent mean ± SD. D) Islets were isolated from 10-month old mice (10 mice pooled per sample) for . Diet-induced obesity does not cause T cell accumulation in islets. C57BL/6J mice were fed a high fat diet (HFD, purple squares) or low fat diet (LFD, grey circles) for 9-12 weeks from 6 weeks of age. Non-fasted body weight (A) and blood glucose (B) after 9 weeks of diet. Glucose tolerance tests (C-D, n=5-7), shown as glucose traces (C) and incremental AUC (D), and insulin tolerance tests (E-G, n=3-5) at 10 and 11 weeks of diet, shown as raw blood glucose (E), delta blood glucose (F), and net AUC relative to baseline (G). H) Fasted plasma insulin at 11-12 weeks of diet. Data represent mean ± SEM, and were analyzed by Student's T-test with Welch's correction (A,D,G), Mann-Whitney test (B,H), or two-way repeated measures ANOVA with Sidak's multiple comparisons test (C,E,F). I-J) Islets were isolated from LFD-and HFD-fed mice after 12 weeks of diet (n=3 samples, 5 mice pooled per sample), dispersed, and gated on FSC, SSC, viability (7AAD-) and CD45+, and subsequently CD19 and CD3. I) Representative data for LFD-and HFD-mouse islets. J) Quantification of islet immune cell populations as a function of viable islet cells, data represent mean ± SD and were analyzed by Mann-Whitney test. Numbers in FACS plots represent the percent of cells in each selection as a function of the parent population.