Senescent T Cells Predict the Development of Hyperglycemia in Humans

Immunosenescence is the progressive impairment of the immune system thought to underlie age-related comorbidities (1,2). Change in T-cell immunities is a notable feature of immunosenescence. There is accumulating evidence that senescent T cells are involved in the pathogenesis of cardiovascular diseases (CVDs), including atherosclerosis, acute coronary syndrome, and hypertension (3–7). We reported that hypertensive patients had an increased frequency of senescent CD8⁺ T cells in peripheral blood, which exhibited CD28 loss and acquisition of CD57 on their surface (8). CD28 loss is a prominent change associated with human aging and is caused by the repetitive antigenic stimulation of T cells. CD57 expression during the late stage of T-cell differentiation might be a distinct measure of senescence in T cells (9). Compared with CD28⁺ or CD57^null T cells, CD28^null or CD57⁺ T cells produce more proinflammatory cytokines and exert greater cytotoxicity (10).

Substantial overlap and interconnectivity exists between the etiology and pathophysiology of diabetes and CVD, theorized as the “common soil” hypothesis (11). Despite evidence that immunosenescence can lead to CVD, there are no studies evaluating whether increased T-cell senescence leads to hyperglycemia development in humans. Our study’s aim was to investigate the relationship between senescent T cells and glycemic status using a cross-sectional cohort. Furthermore, we studied the longitudinal impact of senescent T cells on hyperglycemia development in participants with normoglycemia.

In this prospective longitudinal study, 805 Koreans registered in the Yonsei Cardiovascular Genome cohort or the Cardiovascular and Metabolic Disease Etiology Research Center-High risk cohort (clinicaltrials.gov: NCT02003781) were recruited via the outpatient clinic of Severance Cardiovascular Hospital from January 2011 to April 2016. Type 2 diabetes was defined as: fasting plasma glucose level ≥126 mg/dL; hemoglobin A_1c (HbA_1c) ≥6.5% (48 mmol/mol); or history of insulin or oral hypoglycemic agent administration. Participants without diabetes were categorized as either normal glucose tolerance (NGT), defined as fasting plasma glucose levels <100 mg/dL and HbA_1c <5.7% (39 mmol/mol) or prediabetes, defined as fasting plasma glucose levels 100–125 mg/dL or HbA_1c 5.7–6.4% (39–46 mmol/mol) (12). Hyperglycemia included prediabetes and type 2 diabetes (13). Hypertension was defined as systolic blood pressure >140 mmHg or diastolic blood pressure >90 mmHg over three visits before the use of antihypertensives. Coronary artery disease was defined as one or more lesions with a >50% diameter reduction by coronary angiography. Patients with any of the following conditions were excluded: significant systemic disease, debilitating malignant disease, severe hypertension (>200/140 mmHg), estimated glomerular filtration rate <30 mL/min/1.73 m², and history of overt chronic inflammatory disease and/or receiving anti-inflammatory medications. This study was approved by the Yonsei University College of Medicine Institutional Review Board, and study procedures were in accordance with institutional guidelines (Institutional Review Board number 4-2018-0190). Participants provided informed consent before enrollment.

Following an overnight fast, blood tests for biochemical measurements were performed. Serum glucose, insulin, and HbA_1c were measured using the Hitachi 7600 analyzer (Hitachi Ltd.), immunoradiometric assay (Insulin-IRMA; DIAsource, Louvain-la-Neuve, Belgium), and immunoassay by an Integra 800 CTS (Roche Diagnostics), respectively. We examined medical records of participants with normoglycemia who had at least one clinic visit for any reason from the date of registry enrollment until December 2016. Of 222 NGT participants, follow-up blood glucose data were available in 149 (88 men and 61 women). Based on fasting plasma glucose or HbA_1c levels, participants with baseline NGT were defined as hyperglycemia progressors when those met diagnostic criteria for prediabetes or diabetes in at least one time interval.

Peripheral blood mononuclear cells (PBMCs) were isolated using Ficoll-Hypaque (GE Healthcare, Uppsala, Sweden) density gradient centrifugation and immediately stained for flow cytometry analyses. PBMCs were incubated with directly conjugated monoclonal antibodies for 20 min at 4°C using anti-CD3 (Horizon V500), anti-CD4 (phycoerythrin [PE]-Cy7), anti-CD8 (allophycocyanin [APC]-H7), anti-CD19 (PerCP-Cy5.5), anti-CD28 (APC) (all from BD Biosciences, San Jose, CA), and anti-CD57 (eFluor 450; BioLegend, San Diego, CA). Multicolor flow cytometry was performed using an LSR II instrument (BD Biosciences) and analyzed using FlowJo software (Tree Star, San Carlos, CA). The gating strategy is provided in Supplementary Fig. 1. FlowJo software autogating was performed on CD28 and CD57 T cells. Frequency of CD8⁺CD57⁺ or CD8⁺CD28^null T cells and CD4⁺CD57⁺ or CD4⁺CD28^null T cells was expressed as a percentage of the entire population of CD8⁺ and CD4⁺ T cells.

Cytomegalovirus (CMV) serostatus was evaluated using a chemiluminescent microparticle immunoassay (Abbott Laboratories, Chicago, IL), and a titer of ≥6.0 antibody units/mL of IgG was considered CMV IgG-seropositive. PBMCs were stimulated with overlapping peptides from CMV pp65 (0.6 nmol of each peptide/mL) (Miltenyi Biotec) for 6 h in the presence of PE-conjugated anti-CD107a (BD Biosciences). After 1 h of incubation, brefeldin A (GolgiPlug; BD Biosciences) and monensin (GolgiStop; BD Biosciences) were added to accumulate cytokine proteins intracellularly. Following surface staining with anti-CD3 (Horizon V500), anti-CD4 (PerCP-Cy5.5), anti-CD8 (APC-H7), anti-CD28 (Horizon V450), and anti-CD57 (APC), the cells were fixed and permeabilized using a Fixation/Permeabilization Buffer Kit and further stained for intracellular cytokines with anti–interferon-γ (FITC) and anti–tumor necrosis factor-α (PE-Cy7) (both from BD Biosciences). All samples were assessed using an LSR II Flow Cytometer (BD Biosciences), and the data were analyzed using FlowJo software.

Omental adipose tissue from consenting patients without acute infection or receiving immune-modulating medications was obtained during abdominal surgery (e.g., hepatectomy, colectomy, or cholecystectomy due to cancer). In order to identify infiltrated CD8⁺ T cells adjacent to macrophages within adipose tissues, immunohistochemistry was performed with formalin-fixed, paraffin-embedded sections using the primary antibodies of CD68 (Thermo Fisher Scientific, Waltham, MA) and CD8 (Abcam, Cambridge, U.K.) and the secondary antibody of Polink DS (GBI Labs, Bothell, WA). Next, double immunofluorescence staining was performed to colocalize CD57⁺ with CD8⁺ T cells. Primary antibodies used were CD57 (HNK-1, Leu-7, and MA5-11605; Thermo Fisher Scientific) and CD8 (ab4055; Abcam). Secondary antibodies used were Alexa Fluor 488 (A21042; Life Technologies, Carlsbad, CA) and Alexa Fluor 647 (A21245; Life Technologies). Slides were examined on a Zeiss AXIO Imager A1 & HBO100 (Zeiss, Oberkochen, Germany) using AxioVision software (Zeiss).

All statistical analyses were performed using SPSS version 23.0 for Windows (IBM Corp., Armonk, NY). All P values of <0.05 were considered significant.

Participants’ baseline characteristics are summarized in Table 1. According to the glycemic status, age, BMI, and HOMA of insulin resistance (HOMA-IR) were increased, whereas HOMA of pancreatic β-cell function (HOMA-β) was decreased. CD8⁺CD57⁺ and CD8⁺CD28^null T cells were significantly increased in patients with prediabetes and type 2 diabetes compared with those with NGT (Fig. 1A and B). However, senescent CD4⁺CD57⁺ or CD4⁺CD28^null T-cell frequency was not significantly different (Fig. 1C and D). In Supplementary Fig. 2, representative flow cytometry plots present CD57 and CD28 expression in the CD8⁺ T-cell subset in the groups. In simple correlation analyses, fasting plasma glucose levels correlated significantly with frequency of CD8⁺CD57⁺ and CD8⁺CD28^null T cells (Supplementary Table 1). There was a positive correlation between CD8⁺CD57⁺ T-cell frequency and HOMA-IR. Logistic regression analyses were conducted to determine odds ratios for predicting hyperglycemia (Table 2). After adjustment for traditional diabetes risk factors, frequencies of both CD8⁺CD57⁺ and CD8⁺CD28^null T cells were significantly associated with hyperglycemia in the cross-sectional data set (N = 805). When divided into two groups based upon tertiles of CD8⁺CD57⁺ T-cell frequency in peripheral blood, HOMA-IR but not HOMA-β was significantly different in the higher (second to third) tertiles compared with the lower (first) tertile (Supplementary Table 2). When a cutoff of ≥2.34 for HOMA-IR was chosen (14), CD8⁺CD57⁺ T-cell frequencies were independently associated with insulin resistance (Supplementary Table 3).

Of 149 NGT participants, 58 developed prediabetes or type 2 diabetes during follow-up (2.32 ± 1.45 years). NGT participants were divided into two groups based upon tertiles of CD8⁺CD57⁺ or CD8⁺CD28^null T-cell frequency: low (first to second tertile) and high (third tertile). In the high CD8⁺CD57⁺ T-cell group, 57% (20 out of 35) developed hyperglycemia compared with 33% (38 out of 114) in the low group. We observed a significant difference in incidence rates among participants with a high versus low frequency of CD8⁺CD57⁺ T cells (Kaplan-Meier and log-rank test, P = 0.001) (Fig. 2A). However, this difference was not found for CD8⁺CD28^null T cells (Fig. 2B). A Cox regression model (Fig. 2C) revealed that, after adjusting for traditional risk factors for diabetes, the hazard ratio for developing hyperglycemia per percentage of CD8⁺CD57⁺ T cells was 1.785 (95% CI 1.298–2.455).

In humans, CMV is known as an important antigen for repetitive T-cell stimulation and is involved in the accumulation of CD28^null or CD57⁺ senescent T cells (9). Therefore, we analyzed CMV-specific antigen reactivity of T cells. Proinflammatory and cytotoxic functions of T cells were evaluated and compared among 56 baseline NGT participants who did versus did not develop hyperglycemia (Supplementary Table 4). All participants were seropositive for CMV in the current study (data not shown). CMV pp65-specific CD107a-expressing CD8⁺ T cells, which represent degranulation of cytotoxic proteins such as perforin and granzymes (15), were more frequently observed in those who did compared with those who did not progress to hyperglycemia.

Adipose tissue is a major target of insulin action and immune cell reservoir (16). Using immunohistochemistry, we identified infiltrated CD8⁺ T cells adjacent to macrophages within adipose tissues (Supplementary Fig. 3A). Furthermore, immunofluorescent analyses showed histological evidence of greater CD8⁺CD57⁺ T-cell infiltration in omental adipose tissues of patients with prediabetes or type 2 diabetes compared with those with NGT (Supplementary Fig. 3B and C).

The current study demonstrates that increased senescent CD8⁺ T cells in peripheral blood are independently associated with prevalence and incidence of prediabetes or type 2 diabetes. Furthermore, increased frequency of CD8⁺CD57⁺ T cells was associated with insulin resistance, and we observed histological evidence of CD8⁺CD57⁺ T-cell infiltration in visceral adipose tissues (VATs) of patients with prediabetes or type 2 diabetes.

Although type 2 diabetes pathogenesis is not fully understood, insulin resistance is the hallmark (17). Obesity is associated with chronic low-grade inflammation in VATs and a sustained whole-body proinflammatory state (18). Infiltration of T cells into adipose tissue has been extensively reported (19,20). Nishimura et al. (21) found large numbers of CD8⁺ effector T cells in epididymal adipose tissue in mice fed a high-fat diet. Immunologic/genetic depletion of CD8⁺ T cells lowered macrophage infiltration and adipose tissue inflammation and ameliorated systemic insulin resistance. In a human study that included patients with diabetes, CD4⁺ lymphocytes in VAT biopsies correlated significantly with BMI (19). In the current study, senescent CD8⁺ T cells and not CD4⁺ T cells were significantly associated with present and future hyperglycemia.

Common persistent viral infections (especially human CMV) are attributed to increasing CD8⁺CD28^null (CD8⁺CD57⁺) T-cell populations with age (9) and have been shown to express perforin, granzymes, and granulysin, with high cytotoxic potential (8,22,23). This finding aligns with our previous observation that CMV pp65-specific interferon-γ–, tumor necrosis factor-α–, and CD107a-expressing cells were more frequently observed in CD8⁺CD57⁺ T cells compared with CD8⁺CD57⁻ T cells (24). Previous studies demonstrated that proinflammatory cytokines could promote insulin resistance (25). Correspondingly, we found that increased frequency of CD8⁺CD57⁺ T cells was independently associated with insulin resistance measured by HOMA-IR. In addition, among participants with baseline NGT, CMV pp65-specific CD107a-expressing CD8⁺ T cells were more frequently observed in those who did versus did not progress to hyperglycemia. How senescent T cells interact with other immune cells within target tissues (including adipose) in hyperglycemia development remains to be investigated. Although our findings suggest that senescent T cells might be involved in the early development of diabetes, we were unable to elucidate whether senescent T cells are indicators or pathogenic players for hyperglycemia.

In conclusion, present and future hyperglycemia (prediabetes or type 2 diabetes) were associated with increased senescent CD8⁺ T cells in peripheral blood. Larger prospective studies and sophisticated mechanistic experiments are warranted to confirm whether modulation of immunosenescence is a potential new therapeutic target of diabetes.

Funding. This work was supported by grants from the Korean Health Technology R&D Project, Ministry of Health and Welfare, Republic of Korea (HI13C0715 to S.P. and HI17C0913 to Y.-h.L.) and by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (NRF-2015R1A2A2A01007346 to S.P. and NRF-2016R1A5A1010764 to Y.-h.L.). This work was also supported by the KAIST Future Systems Healthcare Project from the Ministry of Science, ICT and Future Planning.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.-h.L. and S.R.K. wrote the manuscript, analyzed data, and performed the statistical analysis. D.H.H., Y.D.H., J.H.K., and S.H.K. contributed to acquisition of data. H.T.Y., C.J.L., and K.H.K. analyzed data and performed the statistical analysis. B.-H.M. and D.-H.K. performed histologic analyses. J.W.C. provided critical review, advice, and consultation throughout. Y.-h.L., S.R.K., D.H.H., H.T.Y., Y.D.H., J.H.K., S.H.K., C.J.L., B.-H.M., D.-H.K., K.H.K., J.W.C., W.-W.L., E.-C.S., and S.P. contributed to critical revision of the manuscript and read and approved the final submitted version of the manuscript. W.-W.L., E.-C.S., and S.P. are the guarantors of this work and, as such, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.