Type 2 diabetes is associated with the accumulation of senescent T cells

Summary Type 2 diabetes is a global health priority, given that it is driven, in part, by an ageing population, the role of immune senescence has been overlooked. This is surprising, as the functional impairments of senescent T cells show strong similarities to patients with hyperglycaemia. Immune senescence is typified by alterations in T cell memory, such as the accumulation of highly differentiated end‐stage memory T cells, as well as a constitutive low‐grade inflammation, which drives further immune differentiation. We show here in a preliminary study that people living with type 2 diabetes have a higher circulating volume of senescent T cells accompanied with a higher level of systemic inflammation. This inflammatory environment drives the expression of a unique array of chemokine receptors on senescent T cells, most notably C‐X‐C motif chemokine receptor type 2. However, this increased expression of migratory markers does not translate to improved extravasation owing to a lack of glucose uptake by the T cells. Our results therefore demonstrate that the presence of senescent T cells has a detrimental impact on immune function during type 2 diabetes.


Introduction
Type 2 diabetes (T2D) is a metabolic disorder characterized by an impaired glucose homeostasis due to insulin resist ance, associated with a chronic lowgrade systemic inflam mation [1]. The global disease burden in 2014 was estimated to be approximately 8% of the adult population and has been predicted to rise steeply in the coming years [2]. Common comorbidities of T2D include increased risk of cardiac disease, alongside poor wound healing and premature onset of various agerelated disorders [1,3,4]. Furthermore, the immense burden of mortality which accompanies T2D and its complications was estimated to account for 6.8% of global deaths in 2010 [5]. The pandemic nature, alongside its discernible morbidity, exposes the mammoth threat to public health that T2D truly poses.
Early studies proposed that dysfunctional innate immu nity was the cause of the chronic lowgrade inflammation seen in T2D, leading many groups to concentrate on the contribution of the macrophage to diabetic disease advance ment [6]. However, more recently there has been heightening support advocating the role of T cell dysfunc tion in the pathological course of T2D. The exact mecha nism in which this systemic inflammation is potentiated is still unclear; however, one probable candidate is thought to be cellular senescence [9]. Cellular senescence forms a pillar of biological ageing and is implicated in various agerelated disorders, including T2D [7]. Cellular senes cence has been well described in fibroblast models and demonstrates cellular characteristics such as a stable cell cycle arrest, upregulation of cyclindependent kinase inhibitors, marked alterations in cellular pathway activity and transcriptional changes, including the production of a proinflammatory senescenceassociated secretory phe notype (SASP) [8].
Senescenceassociated changes are typified in the T cell compartment by the accumulation of highly differentiated endstage memory T cells, as well as a constitutive low grade inflammation [9]. Furthermore, recent evidence implicates the systemic milieu seen in T2D as a senescence promoting factor as well as potentiating diabetic disease ORIGINAL RESEARCH: UNDERPINNING `TRANSLATING IMMUNOMETABOLISM' progression and inflammation [7]. This theory has also gained traction in studies observing the effects of met formin, where both murine [10] and human studies [11] have shown a survival advantage with metformin treatment, theorized to be a consequence of inhibiting the SASP production by modulating the activation of the nuclear factor kappa B (NFκB) proinflammatory pathway [12]. Senescent T cells can be defined on the basis of surface marker expression, such as the loss of the costimulatory molecules CD27 and CD28 or the homing marker CCR7 and the reexpression of CD45RA, together with elevated levels of viralspecific makers CD57 and killer cell lectin like receptor G1 (KLRG1) [13,14]. While senescent T cells show proliferative impairment their ability to secrete pro inflammatory cytokines such as interferon (IFN)γ and tumour necrosis factor (TNF)α are retained [14][15][16].
The role played by senescent T cells in T2D has been overlooked. This is surprising, as the immune impairments inherent to ageing also resemble those of chronic hyper glycaemia: poor control of infections and reduced vac cination responses together with elevated inflammatory activity [17,18]. We show here in a preliminary study that people living with T2D have an increased number of senescent T cells in both the CD4 + and CD8 + T cell compartments. These senescent T cells show impaired migratory capacity despite having an upregulated chemokine receptor expression. Therefore, our results demonstrate that, rather than being a benign presence, the presence of senescent T cells have a detrimental impact on immune function during T2D.

Glucose and lipid uptake assays
Glucose uptake was assessed in CD27/CD45RAdefined T cell subsets; PBMCs were incubated with antiCD3 (1 µg/ ml) for 15 min at 37°C in complete media before being transferred into PBS for a further 15 min incubation prior to the addition of 100 µM 2NBDG. The PBMCs were then incubated with the 2(N(7nitrobenz2oxa1,3diazol4yl) amino)2deoxyglucose (2NBDG) together with antiCD3 for 30 min before being analysed by flow cytometry.

Transwell migration assay
Human umbilical vein endothelial cell (HUVEC) mon olayers were grown to confluence on Transwell mem branes (Corning, New York, NY, USA) in the presence of 10 ng/ml IFNγ (R&D Systems, Abingdon, UK). PBMCs from healthy donors or T2Ds were placed in M199 medium (Sigma, Poole, UK) in the top well and the chemoattractant in the bottom well, which was 20% autologous donor sera. Cell migration was assessed after incubation at 37°C for 4 h; each condition was set up in duplicate Transwells. Migrated T cells were then col lected from the top and bottom wells, respectively, and stained with phenotypical markers as described above.

Statistical analysis
GraphPad Prism or ncss version 11 was used to perform statistical analysis. Statistical significance was evaluated using the Student's ttest or a repeatedmeasures analysis of variance (anova) with the Tukey correction used for post-hoc testing. Differences were considered significant when P < 0·05.

The diabetic milieu alters the normal balance of T cell subsets
Chronic inflammation and immune dysfunction have been thought to play a role in the pathogenesis of diabetes. To investigate this relationship we examined differential ORIGINAL RESEARCH: UNDERPINNING `TRANSLATING IMMUNOMETABOLISM' leucocytes in individuals aged > 55 years with or without T2D, controlling for gender and age as a continuous covariate while excluding metastatic disease. We found a higher absolute white cell count, including neutrophils and eosinophils, but evidence of lymphopenia in the diabetic cohort of patients (Table 1). These data suggest that T2D patients have differential leucocyte counts that are char acteristic of raised levels of systemic inflammation [20].
To determine how our observed changes in the white cell count related to the T cell population, we performed phenotypical analysis of CD4 + and CD8 + T cells in a subgroup of people living with T2D aged > 55 years recruited from the DARE database. We used flow cytom etry to divide the T cell pool into four subpopulations based on their expression of surface receptors CCR7 and CD45RA. These subpopulations can be defined as naive (CCR7 + CD45RA + ), central memory (CM; CCR7 + CD45RA -), effector memory (EM; CCR7 -CD45RA -) and effector memory CD45RA reexpressing cells (EMRA; CCR7 CD45RA + ) and represent successive stages of T cell differentiation [21], with the EMRA T cells being the senescent subset (Fig. 1a,b) [16]. In people living with T2D we found a significant reduction in the naïve pool in both CD4 + and CD8 + populations; this was accompanied by a significant rise in the more differentiated EM and senescent EMRA populations of the CD4 + subset and CM and EMRA populations of the CD8 + T cells compartment compared to agematched controls (Fig. 1c,d). Overall, we show that T2D is a factor that drives the premature ageing of T cells, increas ing T cell differentiation and senescence.
A wide variety of stressors including inflammation are known to cause premature senescence. As already stated, the T2D milieu, together with the presence of senescent cells, both contribute to the generation of an inflammatory envi ronment. The presence of systemic inflammation is expressed here as a significantly raised neutrophil count (Table 1).

Altered expression of surface chemokine receptors on T cells from people living with T2D
The vast effects of systemic inflammation on normal T cell physiology is largely uncharacterized, as a characteristic feature of chronic inflammation is the anomalistic  infiltration of lymphocytes into nonlymphoid tissues [22]. We therefore examined the expression of migration associated chemokine receptors on T cells using flow cytometry from people living with T2D. We analysed the CXCR2 chemokine receptor, classically thought to be located on the neutrophil cell membrane and expressed scarcely in T cells [23]. We found it to be significantly overly expressed in both the EM and EMRA populations of CD4 + (Fig. 2a,c) and the EMRA population (Fig. 2b,d) of CD8 + T cells of people living with T2D compared to healthy controls. The CX3CR1 chemokine receptor, involved in the infil tration and proinflammatory polarisation of leucocytes, was increasingly expressed as T cells gained effector func tionality ( Fig. 2e-g). Furthermore, we observed a distinct increase in expression of CX3CR1 in CD4 + EM and EMRA populations (Fig. 2e,g) and the CD8 + naive, EM and EMRA T cell populations of people living with T2D (Fig.  2f,h). Taken collectively, these data suggest an abnormal promigratory phenotype of CD4 + and CD8 + T cells in T2D, although the extent of which this is attributed to chronic inflammation and or abnormalities in the serum glucose is unclear.

T cells from people living with T2D accrue functionally impaired migration
Migration plays an integral role in the normal immune function of lymphocytes. To functionally correlate our flow cytometry data, which suggest a promigratory phe notype in T cells isolated from people living with T2D, Transwell chemotactic assays were performed. HUVECs were activated using IFNγ and migration was assessed in response to 20% autologous serum (Fig. 3 a,b). We used autologous sera to replicate in-vivo conditions more effectively rather than observing the migration in response to one chemokine. We found that the CD4 + CM subset and the N and CM CD8 + subsets from patients living with T2D showed increased migratory capacity (Fig. 3c,d). However, both the CD4 + and CD8 + EM and EMRA diabetic subsets showed significantly impaired migration (Fig. 3c,d). Given our chemokine phenotyping, together with diabetic sera containing a higher concentration of chemokines (data not shown), this was an unexpected result.
In the context of a disease propagated by inefficient cellular glucose absorption, an abnormality in glucose uptake may limit T cell migration, despite a functional gain in promigratory receptors [24]. We used the fluo rescent glucose analogue 2NBDG in a glucosefree buffer to test whether the impaired migration arose as a con sequence of an energy deficit in the diabetic T cells (Fig.  4a,b). We found that all the CD4 + and CD8 + subsets displayed a lack of glucose uptake compared to the age matched T cells (Fig. 4c,d). This lack of glucose uptake probably generates an energy imbalance in the T cells from people living with T2D which gives rise to their decreased migratory ability, despite their upregulated chemokine receptors, and may be a cause of their func tional impairment.

Discussion
Since the seminal paper by Hayflick and Moorhead, evi dencing the finite replicative ability of fibroblasts in culture [25], it is now known that cellular senescence can also be induced in response to different stressors such as oxi dative stress or oncogenic activation [8]. In addition to growth arrest, senescent cells adopt several unique iden tifying characteristics, including the accumulation of DNA damage foci, and an upregulation of cell cycle inhibitors and reactive oxygen species [26]. Also prominent is the production of proinflammatory factors termed the senes cenceassociated secretory phenotype (SASP), where senes cent cells promote the development of chronic lowgrade sterile inflammation [27]. It has been postulated that the presence of senescent cells may be the causative agent  Descriptive leucocyte data for healthy patients and patients with type 2 diabetes. Data were controlled for gender and age as a continuous covariate, patients with metastatic disease were excluded. All values mean proportion (95% confidence interval) or n (%). ORIGINAL RESEARCH: UNDERPINNING `TRANSLATING IMMUNOMETABOLISM' in the development and progression of T2D contributing to tissue dysfunction; however, the diabetic milieu could be permissive to the development of senescent cells [7]. Whether or not senescence is the initial driver of T2D does not negate the existence of a detrimental feedback loop where metabolic dysfunction promotes cellular senes cence that contributes to the worsening of disease [8]. Cellular senescence during diabetes has been wellchar acterized in endothelial cells, renal mesangial cells, adipose derived stem cells and fibroblasts [28][29][30]. However, very little work has been carried out in the immune compart ment. Studies have eluded to an increase in senescent CD4 + T cells in young patients with T2D [31] and CD8 + senescent cells in older adults with T2D [32], although in both papers cells were defined as being senescent using only the loss of the costimulatory marker CD28. While the loss of CD28 is often associated with senescence, CD28 T cells are a heterogeneous mix of two distinct T cell populations: EM and EMRA. Here we conclusively dem onstrate that while people living with T2D exhibit lym phopaenia the remaining T cells present are the senescent EMRA subset of CD4 + and CD8 + T cells.
Senescent T cell have been well documented to be a proinflammatory subset [14] secreting their own unique array of proinflammatory products as part of their SASP [33], reinforcing both the senescent and inflammatory phenotype of these cells. The inflammatory environment of T2D, evidenced here as a measurable rise in serum markers of inflammation, further contributes to the gen eration of senescent T cells. Inflammation also changes the expression of chemokine receptors [34], and we found that senescent T cells from people living with T2D acquired Fig. 2. People living with type 2 diabetes (T2D) show increased proinflammatory chemokine receptor expression. Flow cytometry plots showing the expression of CXCR2 on both CD4 + (a) and CD8 + (b) CD45RA/CCR7 defined T cells from healthy controls and patients with T2D aged > 55 years. Together with graphs showing cumulative CXCR2 data from CD4 + (c) and CD8 + (d), T cells from nine healthy and nine T2D subjects. Representative CX3CR1 plots from both CD4 + (e) and CD8 + (f) T cell subsets and graphs for CD4 + (g) and CD8 + (h) T cells also from nine healthy and nine T2D individuals aged > 55 years. Graphs shows the mean ± standard error of the mean (s.e.m.) with Pvalues calculated using a ttest. migratory receptors not normally associated with T cells, such as CXCR2. The chemokine receptor CXCR2 is a typical feature of the neutrophil surface membrane, ena bling their migration from the bone marrow to the periph ery [35]. The relationship between CXCR2 and senescence has previously been explored, CXCR2 knockout studies using fibroblasts have shown mitigation of both replicative and oncogeneinduced senescence while also diminishing the DNA damage response [36]. Furthermore, ectopic expression of CXCR2 results in premature senescence via a p53dependent pathway [36]. Therefore, CXCR2 appears to be vital in initiating and sustaining senescence; whether the increased expression of CXCR2 observed on T cells merely reinforces the senescent phenotype or plays a migratory role requires further investigation. The ligand of the chemokine receptor CX3CR1, frac talkine, is a known mediator of chemoattraction, adhesion and activation of CX3CR1expressing cells [37]. This migra tion axis has been considered to be the mechanism of lymphocyte migration through the endothelium of inflamed tissues, a prominent feature during chronic inflammation [38]. Given the inflammatory nature of T2D we also moni tored the expression of CX3CR1 on senescent T cells. We showed increased levels of chemokine receptor CXCR3 on late differentiated CD4 + and CD8 + T cells, as well as the CD8 + naive population of people living with T2D. Dietinduced obesity, like T2D, is also associated with systemic inflammation [39]. The preferential migration into inflamed nonlymphoid sites associated with increases in CX3CR1 have also previously been demonstrated in CD4 + T cells of mice fed a highfat diet [40]. We postulate that the combination of systemic inflammation and the abnor malities in the extracellular environment, including sup raphysiological glucose levels, to be the cause of the increased chemokine receptor expression in T cells.
Despite the phenotypical changes in the latestage T2D T cell population, suggestive of an increased migratory potential, we paradoxically show an impaired functional Fig. 4. Impaired glucose uptake in T cells isolated from people living with type 2 (T2D) diabetes. Flow cytometry plots showing glucose uptake, assessed using the fluorescent glucose analogue 2NBDG in CD4 + (a) and CD8 + (b) CD45RA/CC27defined T cell subsets from healthy controls and patients with TD2 aged > 55 years. Graphs showing glucose uptake from CD4 + (c) and CD8 + (d) T cells from five healthy and five T2D subjects. Graphs shows the mean ± standard error of the mean (s.e.m.) with Pvalues calculated using a ttest.
ORIGINAL RESEARCH: UNDERPINNING `TRANSLATING IMMUNOMETABOLISM' migration in this subgroup of T cells. The inability of inflammatory terminal stage T cells to migrate suggests their sequestration in the peripheral circulation. T cell migration is centred around the production of adenosine triphosphate (ATP) during aerobic glycolysis to provide a rapid source of ATP for cytoskeletal rearrangement [41]. Previous studies have shown that interference with the glycolytic pathway results in constrained migration of CD4 + and CD8 + T cells [42]. We show a large reduction in the glucose uptake across CD4 + and CD8 + T cells in T2D, suggesting an insulin resistant phenotype in the circulating T cell pool. Various large cohort studies outline the increased morbidity associated with T2D compared to the nondiabetic population [43][44][45]. Despite our grow ing insight into the underlying pathogenesis of T2D com plications, current therapies are aimed at optimizing glycaemic control. There is increasing evidence that direct T cell dysfunction during T2D is responsible, in part, for classical diabetic complications, including increased rates of infection and impaired wound healing [46,47]. As T cell migration is vital in both the control of infection and the woundhealing process, we identify a functional impairment that may contribute to comorbid disease.
Overall, our results therefore demonstrate that an increased presence of senescent T cells has a detrimental impact on immune function during T2D. Given that many of the complications occurring during T2D have been difficult to treat through glucose control alone, targeting senescent cells may be a more effective treatment option.