Chronic Hyperglycemia Drives Functional Impairment of Lymphocytes in Diabetic INSC94Y Transgenic Pigs

People with diabetes mellitus have an increased risk for infections, however, there is still a critical gap in precise knowledge about altered immune mechanisms in this disease. Since diabetic INSC94Y transgenic pigs exhibit elevated blood glucose and a stable diabetic phenotype soon after birth, they provide a favorable model to explore functional alterations of immune cells in an early stage of diabetes mellitus in vivo. Hence, we investigated peripheral blood mononuclear cells (PBMC) of these diabetic pigs compared to non-diabetic wild-type littermates. We found a 5-fold decreased proliferative response of T cells in INSC94Y tg pigs to polyclonal T cell mitogen phytohemagglutinin (PHA). Using label-free LC-MS/MS, a total of 3,487 proteins were quantified, and distinct changes in protein abundances in CD4+ T cells of early-stage diabetic pigs were detectable. Additionally, we found significant increases in mitochondrial oxygen consumption rate (OCR) and higher basal glycolytic activity in PBMC of diabetic INSC94Y tg pigs, indicating an altered metabolic immune cell phenotype. Thus, our study provides new insights into molecular mechanisms of dysregulated immune cells triggered by permanent hyperglycemia.


INTRODUCTION
Diabetes mellitus is a major risk factor regarding the outcome of bacterial and viral infections (1). Among diabetic patients, increased numbers of highly susceptible individuals suffering from tuberculosis (2) or influenza (3) are common. Recently, meta-analyses revealed an enhanced severity and mortality of COVID-19 infections in people with diabetes mellitus (4,5). Undoubtedly, there is a link between the disease and an impaired immune system, but the molecular mechanisms involved in altered immune cell function in diabetic patients are still unknown. Dysfunctional T cells may play a pivotal role in immunological impairment in diabetes mellitus (6). However, the precise effects of chronic hyperglycemia on immune cell function are still uncertain since contradictory findings of both hyperresponsive (7) and attenuated (8,9) T cells were reported in type 2 diabetic patients. While numerous studies linked hyperglycemia to enhanced T cell activation and proliferation (7), others suggest that an abnormal glucose homeostasis promotes an insufficient T-cell response via an increased frequency of senescent T cells with proliferative impairment (8,9). Thus, it has yet to be determined whether dysfunctional T cells are impaired or highly activated in diabetes mellitus and whether hyperglycemia causes immune cell dysregulation.
Alterations of cellular mechanisms and molecular networks in disease states can be analysed by changes in proteome composition (10). Still little is known about accurate protein profiles of T cells affected by hyperglycemia in vivo. Therefore, mass spectrometry-based studies are a powerful tool for obtaining comprehensive information regarding cellular pathology (11) and enable hypothesis-generating approaches on the protein level to understand the molecular basis of altered proteome composition and regulation caused by disease. Moreover, recent advances in the field of immunometabolism recognized cellular metabolism as a primary driver and regulator of immune cell function, leading to a wide variety of functionally different immune cells (12). Since there is still a lack of knowledge about hyperglycemia-associated alterations of immune cell metabolic phenotypes, there is an urgent need to understand whether and how high glucose levels affect the metabolic phenotype of the cells and thereby change their immune function (12).
Diabetic INS C94Y transgenic (tg) pigs were generated as a large animal model of permanent neonatal diabetes mellitus (13). These pigs are characterized by impaired insulin secretion with consecutive hypoinsulinemia and increased fasting blood glucose levels (13). Early on, INS C94Y tg pigs show a stable diabetic phenotype and mirror several disease-associated alterations of diabetes mellitus as seen in humans such as cataract, retinopathy, impaired myocardial function and regeneration (14,15). Since early stages of disease in diabetic patients are often unperceived for a long time, initial immunological alterations are difficult to observe due to the latent disease process. Thus, this diabetic pig model enables the exploration of deviant immune cell function and underlying mechanisms in an early stage of diabetes mellitus with hyperglycemia and hypoinsulinemia, offering valuable insights into disease pathogenesis. Hence, we used PBMC of young INS C94Y tg pigs and non-diabetic wild-type littermates to investigate the impact of a permanent early-life hyperglycemic condition on immune cell function.

Animal Model and Sample Preparation
In this study, PBMC of 23 sex-matched diabetic INS C94Y tg pigs and 31 non-transgenic wild-type littermates at the age of 12 weeks were used. INS C94Y tg pigs were generated previously as described (13). First, young pigs were weighed. For sampling, pigs were fasted overnight and heparinized venous whole blood was collected. Blood glucose levels were determined immediately using a Precision Xceed glucometer with Precision XtraPlus test strip (Abbott) and insulin levels were measured by an enzymelinked immunosorbent assay (ELISA) for humans in serum (Mercodia, Uppsala, Sweden). Species reactivity with porcine insulin was specified as 93% by the manufacturer. Values below the quantification limit of the assay (≤ 1 mU/L) were arbitrarily set to 0. Body weight, blood glucose levels and serum insulin levels were in accordance with values from earlier characterizations of this large animal model (13). Due to the lack of insulin as growth factor and anabolic hormone, INS C94Y tg pigs exhibit a significantly (*p < 0.05) lower body weight (27.7 ± 5.4 kg) than wild-type littermates (34.9 ± 10.1 kg; Figure  1A). Blood glucose levels were significantly (***p < 0.001) elevated in diabetic pigs (15.8 ± 4.7 mmol/L) compared to wild-types (4 ± 0.7 mmol/L; Figure 1B) and serum insulin levels of INS C94Y tg pigs (6.47 ± 1.08 mU/L) were significantly reduced (***p < 0.001) compared to non-transgenic wild-types littermates (17.18 ± 9.54 mU/L; Figure 1C). All animals were housed under controlled conditions, had free access to water and were fed a commercial diet. PBMC were isolated by density gradient centrifugation (RT, 500 x g, 25 min, brake off) with Pancoll separating solution (PAN-Biotech, Aidenbach, Germany), and restored in PBS (pH 7.4) or RPMI medium (PAN-Biotech, Aidenbach, Germany), supplemented with 10%

Magnetic Activated Cell Sorting for CD4 + T Cells
Briefly, a total of 6 x 10 7 cells was incubated in staining buffer with mouse anti-pig CD4a (clone MIL17, Bio-Rad AbD Serotec, Puchheim, Germany, 1:50) at 4°C for 20 min. Staining buffer contained phosphate-buffered saline (pH 7.2) and was supplemented with 2 mM EDTA and 0.5% bovine serum albumin (BSA). In the next step, cells were resuspended in 480 ml staining buffer before adding 120 ml anti-mouse IgG 2a/b MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany) for an incubation period of 15 min. In further steps, BSA in staining buffer was omitted to prevent interference with mass spectrometry. Magnetic separation was performed using LS columns (Miltenyi Biotec, Bergisch Gladbach, Germany). Magnetically-labelled CD4 + T cells were retained in the magnetic field, while unwanted cells were eliminated by three washing steps. Positive CD4 + T cell fraction was eluted by removing the column from magnetic field and flushing with staining buffer. 6 x 10 5 positive selected cells were pelleted and stored at −20°C until filter-aided sample preparation (FASP). The isolation of porcine CD4 + T cells routinely achieved > 90% purity, confirmed by flow cytometry.

Mass Spectrometry and Data Analysis
Purified CD4 + T cells of four diabetic INS C94Y tg pigs and five littermate wild-types were analyzed. Samples were temporarily frozen after preparation until mass spectrometry could be performed. CD4 + T cell pellets were lysed in urea buffer (8M in 0.1M Tris/HCl pH 8.5) and 10 mg total protein of each sample was proteolyzed with LysC and trypsin by a modified filter-aided sample preparation (FASP) as described (17). Acidified eluted peptides were analyzed in the dataindependent acquisition mode on a Q Exactive HF-X mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) online coupled to an ultra-high-performance liquid chromatography (UHPLC) system (Ultimate 3000, Thermo Fisher Scientific). The spectral library was generated directly in Spectronaut Pulsar X (Biognosys, Schlieren, Switzerland; version 12.0.20491.17.25792) as described (18). Spectronaut was equipped with the Ensembl Pig Database (Release 75 (Sscrofa10.2), 25,859 sequences, https://www.ensembl.org). Peptide identification was filtered to satisfy an FDR of 1% by the mProphet approach with q-value cut-off at 0.05 (19). For subsequent data evaluation, statistical analysis was performed on log2 transformed normalized abundance values using Student's t-test. Differences in protein abundance with p < 0.05 were considered significant. Among all differentially abundant proteins, those with INS C94Y /wt ratio ≥ 2.0 were considered for possible biological relevance and used in further experiments. The heatmap of hierarchical cluster analysis was created with open source software Cluster 3.0 and was illustrated via Java TreeView (version 1.1.6r4, http://jtreeview.sourceforge. net). GraphPad Prism Software (version 5.04) was used to design Volcano plot, and pathway enrichment analysis was done with open source software Reactome (Pathway Browser version 3.7, Reactome database release 74, https://reactome.org).

Measurement of Oxygen Consumption Rate and Extracellular Acidification Rate by Seahorse XFe Analyzer
Metabolic phenotypes of PBMC from 15 wild-types and 10 INS C94Y tg pigs were determined in five independent experiments, using a Seahorse XFe Analyzer (Agilent Technologies, Waldbronn, Germany) measuring oxygen consumption rate (OCR), which is attributed to mitochondrial respiration and extracellular acidification rate (ECAR), which can be related to glycolysis (20). Duplicates (technical replicates) were generated for each animal. Mean values of technical replicates were then used for further statistical analysis. In accordance with the manufacturer's instructions, sterile XF assay buffer (Seahorse XF RPMI Medium supplemented with 10 mM glucose, 2 mM Lglutamine, and 1 mM pyruvate, pH 7.4; Agilent Technologies, Waldbronn, Germany) was used for experiments. Prior to the start of the assay, sensor cartridges (Agilent Technologies, Waldbronn, Germany) were prepared adding oligomycin, FCCP and rotenone & antimycin A. A total of 1 x 10 6 PBMC was seeded in 24-well XF24 cell culture microplates (Agilent Technologies, Waldbronn, Germany), while four wells were kept free from cells as background correction. Baseline OCR and ECAR were measured before adding oligomycin, FCCP and rotenone & antimycin A. OCR was reported in units of pmol/minute and ECAR in mpH/minute.

Statistical Analysis
Kolmogorow-Smirnov (KS) test was performed for determination of Gaussian distribution. If KS test indicated p < 0.05 (no normal distribution), Mann-Whitney U test was used for statistical analysis, while Student's t-test was used if KS test was p > 0.05 (normal distribution). In both tests, statistical probabilities were considered significant at p < 0.05. Significances are indicated by asterisks with *p < 0.05, **p < 0.01, and ***p < 0.001.

Reduced Proliferative Capacity of Lymphocytes in Diabetic INS C94Y Transgenic Pigs After Polyclonal Stimulation With Phytohemagglutinin
For assessment of the proliferation response in vitro, we used B and T cell mitogen PWM (21) and the specific T cell mitogens ConA (22) and PHA (23) to stimulate lymphocytes of diabetic INS C94Y tg pigs and wild-type littermates. Interestingly, while no differences could be observed between the two groups after stimulation with PWM ( Figure 2A) and ConA ( Figure 2B), lymphocytes of diabetic pigs (n=11) revealed a significantly 5fold decreased proliferation rate in response to PHA compared to wild-types (n=13) ( Figure 2C; ***p < 0.001).

Lymphocyte Subpopulation Ratio Did Not Differ Between Wild-Types and Diabetic Pigs
Next, we examined whether the decreased ability to proliferate correlated with altered leucocyte populations in diabetic INS C94Y tg pigs. Relative numbers of peripheral blood mononuclear cells were counted via DiffQuick stained blood smears. Wild-type (n=6) and INS C94Y tg pigs (n=6) displayed no significant difference in relative PBMC counts (Supplementary Figure 1). Next, lymphocyte subpopulations were examined (Supplementary Figure 2). There were no significant differences in the percentage of B or T cell population between wild-types and diabetic pigs ( Supplementary  Figures 2A, B). Moreover, measurements of T cell subsets CD4 + (Supplementary Figure 2C) and CD8a + (Supplementary Figure  2D) as well as pig-specific CD4 + CD8a + double positive ab T cells (Supplementary Figure 2E) and SWC5 + gd T cells (Supplementary Figure 2F) revealed no differences between both groups.
CD4 + T Cells of Diabetic INS C94Y Transgenic Pigs Showed Divergent Proteome Profile Which Associates to Metabolic Pathways In order to gain deeper insights into the observed proliferation on molecular level, we characterized the proteomes of porcine CD4 + T cells, which we hypothesized to be the key drivers of impaired proliferation response. Purified CD4 + T cells were examined with differential proteome analyses, using data-independent acquisition LC−MS/MS. A high-resolution proteome was obtained, with a total of 3,487 identified proteins, of which 2,704 were quantified with at least two unique peptides (Supplementary Table 1). Among these, 80 proteins were significantly different in abundance (*p < 0.05) (Supplementary Table 2, Figure 3A).
Pathway analysis of these 80 proteins via the Reactome database (24) revealed that enriched pathways differed substantially between groups (Supplementary Table 3, Supplementary Figure 3). Most significantly enriched pathways in cells from wild-types mainly associated to processes involved in DNA repair, cell cycle and chromatin remodeling ( Table 1, Supplementary Figure 3A), whereas cells from INS C94Y tg pigs predominantly associated to signal transduction, immune system and metabolic pathways ( Table 2, Supplementary Figure 3B). Among the latter, lipophagy was one of the most significantly enriched pathways ( Table 2). This is especially interesting, since lipophagy is linked to metabolic dysfunction of immune cells (25,26), and it was recently shown that functionally different cell subtypes can be identified solely by their metabolic phenotype (12).

Increased Basal Glycolytic Activity and Mitochondrial Respiration in Peripheral Blood Mononuclear Cells of Diabetic Pigs
Since Reactome pathway analysis revealed association to metabolic pathways, especially lipophagy ( Table 2), we were interested whether the metabolic phenotype of PBMC from INS C94Y tg pigs differs from wild-types. Interestingly, the basal glycolytic rate of PBMC from diabetic pigs was significantly higher compared to wild-type PBMC  ( Figure 4A; *p < 0.05), indicating increased lactate production and glycolytic activity of diabetic PBMC. Furthermore, analyses of OCR revealed a significant increase in oxygen consumption at all measured time points in diabetic pigs. Compared to wild-type PBMC, a rise of basal, ATP-linked and maximal mitochondrial respiration (**p < 0.01) as well as spare respiratory capacity (*p < 0.05) was present in PBMC from diabetic pigs ( Figure 4B). These differences indicated a fundamentally different metabolic phenotype of PBMC from wild-types (n=15) and diabetic INS C94Y tg pigs (n=10).

Diabetes-Associated Protein ANXA1 Was Highly Abundant in CD4 + T Cells of Diabetic INS C94Y Transgenic Pigs
After pathway interpretation of the proteome data from porcine CD4 + T cells, we subsequently focused on the analysis of single proteins. From the 80 significantly differing proteins, we were especially interested in two candidates with potential biological relevance (INS C94Y tg/wt ratio ≥ 2), namely MYL4, a regulatory light chain of myosin that binds actin and regulates ATPase activity (27) and ANXA1 (Supplementary Table 2, Figure 3B). ANXA1, a phospholipid-binding protein, was previously described as a regulator of inflammatory cells with antiinflammatory and pro-resolving properties (28). It is widely expressed in tissues and has been found in a soluble form in biological fluids (29). ANXA1 showed highest statistical significance in protein abundance differences between wild-types and diabetic pigs (***p < 0.001). Interestingly, higher levels of ANXA1 were also already described in serum and plasma of type 1 (30) and type 2 diabetes mellitus patients (31). Therefore, we subsequently focused our candidate analyses on ANXA1.

DISCUSSION
To assess the impact of permanent hyperglycemia on immune cell function in vivo, we characterised lymphocytes of a transgenic pig model for permanent neonatal diabetes mellitus.  We observed significant changes in body weight, blood glucose and serum insulin levels in these diabetic pigs at the age of 12 weeks. Compared to non-transgenic wild-type littermates, significant differences in proliferative response, CD4 + T cell proteomes and the metabolic phenotypes were already detectable in this early diabetic pig model. Our findings point to main distinct changes in the T cell compartment triggered by early-life high blood glucose and indicate that several immune alterations already occur in the early stage of diabetes mellitus. Importantly, our study was performed with diabetic pigs lacking insulin supplementation. The stable diabetic phenotype with permanently high blood glucose and low serum insulin levels makes the INS C94Y tg pig an attractive model for translational diabetes research. Since insulin is a major growth factor and anabolic hormone (32), body weight of diabetic pigs was markedly reduced compared to wild-type pigs due to lower growth rate, as previously described by Renner et al. (13). Investigating immunological properties of this diabetic pig model, we found no differences of lymphocyte proliferation between controls and diabetic pigs after polyclonal stimulation with B and T cell mitogen PWM and T cell mitogen ConA.
However, after polyclonal stimulation with T cell mitogen PHA, we detected a distinct decreased proliferation of lymphocytes from diabetic pigs, which points to an impaired immune response selectively induced by this mitogen. This fits in well with our pathway analysis data, which showed distinct clustering of high abundant proteins from diabetic CD4 + T cells to immune system and signal transduction processes. Interestingly, a decreased response of T cells to PHA-stimulation was already demonstrated in diabetic patients (33,34). The significantly reduced proliferation to PHA was correlated with a decreased interleukin (IL)-2 production (34). The cytokine IL-2, as well as its IL-2 receptor, is a cytokine crucial for proliferation processes in human T cells (35) and, thus, is enhanced in these cells after PHA stimulation leading to cell proliferation (36). These findings were supported and extended by another transcriptomic study of PHA-activated human CD4 + and CD8 + T cells with decreased gene expression of IL-2 and IL-2 receptor in diabetic type 2 patients (37). Furthermore, T cells from these patients exhibited differences in expression of additional genes and gene products like lower expression of genes of insulin signaling pathway and enzymes of the glycolytic pathway (37), indicating that an impaired immune response in the diabetic context induced by PHA is affected by more than IL-2 signaling. Little is known about the precise signaling pathways (37), and intracellular processes stimulated by PHA and ConA in humans and pigs. However, the impaired immune response selectively induced by PHA in hyperglycemic conditions, which is not addressed by polyclonal T cell mitogen ConA (22), points to PHA-associated cell signaling pathways specifically altered in the diabetic condition. In murine T cells, ConA was shown to bind to surface glycoproteins and with high affinity to the costimulatory molecule CD28, leading to enhanced T cell proliferation (38). If mechanisms hold true for porcine T cells, hyperglycemia does not affect CD28 signaling in diabetic pigs. In contrast, PHA is able to mimic the stimulation of T cells via crosslinking of T cell receptor and CD3 complex, which is followed by a sustained calcium influx as an important signal for T cell proliferation (23). In the Jurkat T cell line, high glucose levels were shown to inhibit T cell activation via an increase in non-enzymatic glycation of concentration-regulating calcium channels and delayed specific activation of these cells with anti-CD3 monoclonal antibody (39). Since alterations of calcium homeostasis were shown in various cell types of diabetic patients (40), insufficient calcium-mediated signaling in T cells of diabetic pigs might point to impaired PHA-induced proliferation. Thus, the immune cell calcium homeostasis of diabetic INS C94Y tg pigs merits further investigations in future studies to examine possible correlations to the impaired T cell response in the diabetic cases observed in our study.
To gain deeper insights into hyperglycemia -associated molecular changes of CD4 + T cells and their impact on cell function, we analyzed the cell-derived proteome from diabetic pigs and wild-types as a hypothesis-generating approach for the dysregulated immune responses. Notably, we obtained a comprehensive proteomic data set for porcine CD4 + T cells with a high resolution of 2,704 proteins identified with ≥ 2 peptides. Among these, a total of 3% (80 proteins) showed significantly different protein abundances between groups, indicating distinct proteome differences in CD4 + T cells from early stage diabetic pigs. While many studies of immune cells from diabetic type 2 patients were performed by a 2Dgelelectrophoresis approach, only significantly altered spots were analysed subsequently with mass spectrometry (10,41). To our knowledge, there is no in-depth proteomic fingerprint of CD4 + T cells in hyperglycemic conditions at initial stage of nonimmune-mediated cases of diabetes mellitus. Interestingly, the protein with the highest statistical significance and a 2-fold higher abundance in diabetic pigs was ANXA1. This protein identification corresponds to findings in type 1 (30) and type 2 (31) diabetic patients, where ANXA1 was previously found with increased levels in serum and plasma, pointing to a pivotal, disease-associated alteration of this protein. However, the biological consequence of elevated ANXA1 in diabetes mellitus remains unexplained (28). While human and mouse neutrophils, monocytes and macrophages constitutively contain high levels of ANXA1 in their cytoplasm, T cells express the protein at lower levels (42). Following cell activation, ANXA1 is readily mobilized to the cell surface where it interacts with its receptor named formyl peptide receptor 2 (FPR2) in a paracrine/autocrine manner (43). Early studies in mice have shown that ANXA1 acts as a molecular T cell tuner by increasing the strength of T cell receptor signaling and T cell activation (44). Overexpressed ANXA1 induced the differentiation of naïve T cells to proinflammatory Th1 cells with a high expression of INFg (45), promoting an inflammatory milieu. Thus, the higher abundance of ANXA1 on the surface of CD4 + T cells of diabetic pigs might indicate an enhanced T cell activation with functional importance of a pro-inflammatory immune response. However, also the opposite effect was shown in ANXA1deficient mouse CD4 + T cells (46). These cells increased their activation, proliferation and inflammation in the absence of ANXA1, suggesting that this protein rather has an antiinflammatory role and attenuates an exacerbated inflammatory response (46). One possible reason is that externalized ANXA1 regulates the extracellular regulated protein kinases (ERK)/ mitogen-activated protein kinases (MAPK) signaling pathway via binding to its formyl peptide receptor (FPR) (30). This pathway is important for T cell activity and strongly contributes to T cell activation and proliferation (47) and was shown to be activated in hyperglycemia (30). Treatment of Anxa1 −/− mice with human recombinant ANXA1 attenuated inflammatory processes in tissues by reducing MAPK signaling (30). Thus, a higher abundance of ANXA1 in diabetic CD4 + T cells may temper signaling to protect the cell from hyperresponsiveness initiated by chronic hyperglycemia and, thus, exert anti-inflammatory effects. Although we cannot yet define the precise role of ANXA1 in CD4 + T cells in the porcine diabetic model, the identification of ANXA1 in humans and pigs, independent of etiology, is an important finding which underscores a crucial role of this protein in T cell impairment accompanying diabetes mellitus. It furthermore highlights the potential translational quality of this animal model for diabetes research, which is, in our opinion, recommended for future studies to examine the role of this molecule for the observed dysregulated immune responses.
Pathway analysis of our proteome dataset revealed distinctly different patterns of over-represented pathways between CD4 + T cells from diabetic pigs and wild-type littermates. Proteins with higher abundance in wild-types, hence lower abundance in diabetic pigs, mainly clustered to pathways involved in gene regulation, which may explain the decreased proliferation capacity observed in immune cells from diabetic pigs. Proteins with increased abundance in diabetic pigs, on the other hand, mainly associated to immune system, signal transduction and metabolic pathways. Interestingly, one of the most overrepresented pathways associated to altered proteins in CD4 + T cells of diabetic pigs was lipophagy. Lipophagy is linked to metabolic dysfunctions of cells (25,26), and thus, pointed to an altered metabolic phenotype in diabetic pigs. We therefore characterized the metabolic immune cell phenotype of PBMC, since cellular metabolism can be a primary driver and regulator of immune cell function (12). Metabolic properties of diabetic PBMC in this study revealed an increase in mitochondrial respiration and basal glycolytic activity. We found significant increases in mitochondrial oxygen consumption rate (OCR) in PBMC of diabetic pigs. Interestingly, in line with our findings, increased oxygen consumption was also demonstrated in PBMC of type 2 diabetes patients (48). Previously, this group also measured a higher production of reactive oxygen species (ROS) in these cells and hypothesized mitochondrial damage in diabetic PBMC caused by enhanced oxidative stress. An association of hyperglycemia to glucose-mediated increase in ROS may activate the "dangerous metabolic route in diabetes" through diacylglycerol (DAG), protein kinase C (PKC) and NADPH oxidase [reviewed in (49)]. This results in intracellular accumulation of ROS, which has cytotoxic effects on the cell, possibly affecting proliferation ability of cells, as described in our studies. Since ROS synthesis is essential for physiological immune activation of lymphocytes but lead to cellular damage if concentration surpasses (50), this parameter should be determined in PBMC of diabetic pigs, in order to evaluate whether immune cells are impaired by oxidative cell stress. On the contrary, mitochondrial respiration with increased oxygen consumption is used by immune cells to supplement increased glycolysis, which is the main metabolic pathway fueling effector function upon T cell activation (51). Besides increased oxygen consumption rates, we also found a higher basal glycolytic activity in early diabetic pigs, suggesting that hyperglycemia induced metabolic adaption to local conditions of immune cells and promoted a more activated status in T cells (52). A rapid increase in glycolysis is an important determinant in cellular energy production and required for effective cell proliferation, size expansion, and differentiation (51). Hence, it is also possible that PBMC in young diabetic pigs are highly metabolically active to challenge hyperglycemic conditions. Our findings indicate that metabolic disturbances occurring at early stage diabetes mellitus, as observed in INS C94Y tg pigs, likely cause metabolic reprogramming in PBMC and may also influence their activation status. Since further studies are absolutely required for a better understanding of bioenergetic profiles of PBMC in diabetes mellitus, diabetic pigs enable to define mitochondrial properties and cellular energy production in these cells at initial stage of disease.
Taken together, our findings point to an early, fundamental dysregulation of immune cells when exposed to prolonged high blood glucose. Chronic hyperglycemia leads to profound molecular and functional changes of T cells reasoned by proteomic and metabolic alterations, and, moreover, affects T cell immune responses in vitro. Diabetic pigs qualify for additional translational experiments to explore the crucial role of altered ANXA1 levels in adaptive immunity in diabetes mellitus. The altered metabolic properties of immune cells in young diabetic pigs provided novel information on distinct immune cell metabolism at initial stage of disease. The exact meaning of this significantly altered metabolic phenotype of the immune cells is not known so far, thus, INS C94Y tg pigs are a valuable source to further analyze respective changes for the diagnosis, or treatment or even prevention of early stage diabetes mellitus and its immune system-related complications.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by the Government of Upper Bavaria. All co-authors critically read the manuscript and approved the final version to be published. CD is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors contributed to the article and approved the submitted version.

FUNDING
This work was supported by grants from the Deutsche Forschungsgemeinschaft SPP project 2127 (DFG DE 719/7-1 to CD, HA 6014/5-1 to SH) and by the German Center for Diabetes Research (82DZD00802 to EW and SR).