Increased Blood Levels of Growth Factors, Proinflammatory Cytokines, and Th17 Cytokines in Patients with Newly Diagnosed Type 1 Diabetes

The production of several cytokines could be dysregulated in type 1 diabetes (T1D). In particular, the activation of T helper (Th) type 1 (Th1) cells has been proposed to underlie the autoimmune pathogenesis of the disease, although roles for inflammatory processes and the Th17 pathway have also been shown. Nevertheless, despite evidence for the role of cytokines before and at the onset of T1D, the corresponding findings are inconsistent across studies. Moreover, conflicting data exist regarding the blood cytokine levels in T1D patients. The current study was performed to investigate genetic and autoantibody markers in association with the peripheral blood cytokine profiles by xMap multiplex technology in newly diagnosed young T1D patients and age-matched healthy controls. The onset of young-age T1D was characterized by the upregulation of growth factors, including granulocyte macrophage-colony stimulating factor (GM-CSF) and interleukin (IL)-7, the proinflammatory cytokine IL-1β (but not IL-6 or tumor necrosis factor [TNF]-α), Th17 cytokines, and the regulatory cytokines IL-10 and IL-27. Ketoacidosis and autoantibodies (anti-IA-2 and -ZnT8), but not human leukocyte antigen (HLA) genotype, influenced the blood cytokine levels. These findings broaden the current understanding of the dysregulation of systemic levels of several key cytokines at the young-age onset of T1D and provide a further basis for the development of novel immunoregulatory treatments in this disease.


Introduction
As cell-signaling molecules, cytokines play integral roles in the development and activation of immune cells. Much attention has been devoted to exploring their role in autoimmune diseases, including type 1 diabetes (T1D). Several disease-promoting cytokines and chemokines

Study population
This study included 36 newly diagnosed young T1D patients (median age 10.5 years; interquartile range [IQR] 5.2-12.9 years; 17 boys/19 girls) and 20 controls (median age 14.6 years; IQR 6.7-20.3 years; 8 males/12 females). All patients were recruited from November 2008 to October 2011 at Tartu University Hospital and Tallinn Children's Hospital. Diagnostic criteria for T1D were based on the classification of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus [16]. Data about concomitant autoimmune diseases (autoimmune thyroiditis, Graves' and Addison's diseases, celiac disease, vitiligo, autoimmune liver and rheumatic diseases, multiple sclerosis) and other diseases were available. No autoimmune diseases were recorded except two persons with autoimmune thyroiditis and one without diagnosis but with high anti-TPO antibodies in T1D group. None of individuals in studied population has a history of infections during last month. Peripheral blood was obtained less than 1 week after diagnosis. All T1D patients were on insulin treatment during the blood collection period (mean 0.72, range 0.15-1 U/kg of day). C-peptide values in T1D group were 0.132±0.077 nmol/L (mean±SD). The control group consisted of young healthy blood donors and children who visited Tartu University Hospital with minor surgical problems. Subjects in the control group did not have diabetes or abnormal fasting blood glucose levels, and there was no suspicion of any inflammatory process in these individuals.
All blood samples were collected in the morning before meals and EDTA-treated plasma from the blood was aliquoted and stored at -80°C. Samples did not go through additional freezethaw cycles before analysis. The study was approved by the Research Ethics Committee of the University of Tartu (protocols 163/T-6 from 24.09.2007 and 179/M-29 from 16.02.2009). All patients, their parents, and/or their guardians signed a written consent form before participation.

Autoantibodies and human leukocyte antigen (HLA) genotyping
In all patients and controls, the presence of the main diabetes-associated antibodies and HLA class II alleles was determined. Autoantibodies against 65-kDa glutamic acid decarboxylase (GADA), protein tyrosine phosphatase (IA2A), and zinc transporter 8 (ZnT8A) were measured by commercial ELISA kits (RSR Ltd., Cardiff, UK), in accordance with the manufacturer's protocol. The cut-off levels were !5 U/ml for GADA, !15 U/ml for IA2A and ZnT8A. These tests are performed routinely with specificity of 96-99% and sensitivity of 66-74%, as confirmed by the Islet Autoantibody Standardization Program (IASP) in 2012.
For determination of HLA DQA1-DQB1 genotypes and DRB1 Ã 04 subtypes, polymerase chain reaction (PCR)-based lanthanide-labeled oligonucleotide hybridization and timeresolved fluorometry were used [17,18]. Combinations of HLA DRB1-DQA1-DQB1 alleles were divided into five groups on the basis of risk for T1D development: high-, moderate-, slightly increased-, neutral-, and decreased-risk groups [18]. General medical information, autoantibodies, and HLA data for the study groups are presented in Table 1.

Cytokines
Cytokine levels in EDTA-treated plasma from controls and T1D patients were measured by the xMAP Technology on Luminex 200 (Luminex Corp., Austin, TX). Levels of 20 different cytokines were determined with the Milliplex MAP High Sensitivity Human Cytokine kit  Table 2). Cytokine levels were analyzed in accordance with the manufacturer's protocol, with levels below the detection limit being imputed as 10% less than the minimum detectable concentration limit, as calculated by the manufacturer's protocol.

Statistical analysis
The R version 3.0.1 language and environment (Free Software Foundation, Boston, MA) and GraphPad Prism 5 (GraphPad Software, La Jolla, CA) software packages were used for statistical analyses and figure preparation. For descriptive statistics, medians and IQRs are reported.
As the concentrations of cytokines do not assume a normal distribution, the non-parametric, pairwise Spearman's rank correlation was used to assess the correlation between cytokine levels. The Mann-Whitney U-test (two-tailed) was used to compare the characteristics of the two study groups and Kruskal-Wallis Rank Sum Test for more than two groups. A p-value less than or equal to 0.0025 after Bonferroni correction was considered statistically significant.

Correlation of cytokine concentrations
Three cytokines, IL-6, IFN-γ, and IL-10, were measured simultaneously with High Sensitivity Human Cytokine kit and Human Th17 Magnetic Bead Panel kit. No strong correlations (rho < 0.7) were found between the concentrations of these cytokines when measured by the two assay kits in the whole study group, but correlation was statistically significant (Spearman's rank correlation test: rho = 0.45, p = 0.0005 for IL-6 measured by Th17 kit versus High Sensitivity kits; rho = 0.60, p = 1.4 × 10 −6 for IFN-γ measured by Th17 kit versus High Sensitivity kits, and rho = 0.56, p = 9.9 × 10 −6 for IL-10 measured by Th17 kit versus High Sensitivity kits). The High Sensitivity Human Cytokine kit provided higher percentages of the detected cytokines ( Table 2) and was used in the subsequent experiments.

Relationship between cytokine levels in T1D patients and controls
In T1D patients, 20 cytokines from different functional groups were mutually correlated with each other. For example, granulocyte-macrophage colony-stimulating factor (GM-CSF) showed strong correlations (rho > 0.7) with the lymphoid hematopoietic growth factor IL-7 and with the T-cell growth factors IFN-γ and IL-2. An important mediator of the inflammatory response, IL-1β, showed strong correlations with several growth factors, namely GM-CSF, IL-7, IFN-γ, and IL-2. Interestingly, the well-known proinflammatory cytokines, IL-6, TNF-α, and IL-8, demonstrated weak correlations with other investigated cytokines, including IL-1β. As expected, IL-17A, IL-17E, IL-17F, IL-21, IL-22, and IL-23 were correlated strongly with each other, forming a large Th 17 cytokine cluster. IL-27, which prevent excessive T cell activity and limit pro-inflammatory cytokine production, was also very strongly correlated with the Th 17 cytokine cluster in T1D (Table 3, part A).
In the control group, strong correlations were detected among GM-CSF, IL-7, IFN-γ, and IL-1β. IL-7 was strongly correlated with inflammatory marker IL-6 and T cell growth factor IL-2, which, in turn, were correlated with each other. There was also a strong correlation between IL-2 and IL-1β. In addition, IL-6 was correlated with IL-12. The neutrophil chemotactic factor IL-8 revealed a strong correlation with Th 17 lymphocyte cytokines IL-17F and IL-21. IL-4 showed a strong correlation with another Th 2 cytokine, IL-13, and with the Th 17 cytokines IL-17A and IL-21. Moreover, IL-13 was correlated strongly with IL-22 and IL-7. Most of the Th 17 cytokines were correlated strongly with each other (Table 3, part B).
Influence of blood sampling time and diabetic ketoacidosis (DKA) on cytokine levels in the T1D group Almost half (44%) of investigated children suffered from DKA at the time of T1D diagnosis. As accompanying metabolic imbalance and insulin treatment could influence the production of several cytokines, we analyzed whether the time since diagnosis affected the cytokine concentrations. The IL-7, TNF-α, IL-8, and IL-13 levels showed a tendency (p < 0.05 but > 0.0025) to decrease with an increasing time gap between diagnosis and blood sampling (Fig 1). Moreover, T1D patients with DKA had a tendency for higher IL-8 (Mann-Whitney U-test, U = 59, p = 0.0038) and IL-10 levels (Mann-Whitney U-test, U = 66, p = 0.0088) compared to T1D patients without DKA (Fig 2).

Diabetes-related autoantibodies and cytokine levels
Most of the T1D patients (Table 1) had more than one diabetes-specific autoantibody: 36% had two and 53% had three. Only two T1D patients, neither of whom had DKA, had no autoantibodies. Two other patients with T1D, one with and one without DKA, had only one autoantibody, which was GADA in both cases.
In the control group, two individuals had low titers of GADA and one individual had a low titer of ZnT8A. All of these autoantibody-positive controls were classified in the low-risk group on the basis of HLA haplotype. Different laboratories, including ours, have reported low titers of diabetes-associated antibodies among healthy persons [19][20][21].

Diabetes-related HLA haplotypes and cytokine levels
Most of the examined T1D patients were categorized as having a moderate or high risk for T1D (28% and 25%, respectively) on the basis of HLA haplotypes (Table 1). Seven T1D patients (19%) were categorized as having a decreased risk. In the control group, only two individuals (10%) were classified as being HLA risk for T1D. Besides, we detected no statistically significant differences or tendencies in cytokine levels between T1D patients in the high-risk HLA group compared to T1D patients in all other risk groups (data not shown).

Discussion
In general, T1D is considered to be a Th1-type autoimmune disease caused by pancreatic attack by autoreactive T cells. Various inflammatory cells producing different proinflammatory cytokines could also be involved, and pancreatic β-cell destruction accompanies the inflammatory response (insulitis) within the islets [22]. Destructive insulitis is associated with elevated levels of Th1 cytokines (IL-2, IL-12, and IFN-γ) and proinflammatory cytokines (IL-1β, IL-6, TNF-β, and IFN-α) in animal models [1,23]. However, despite evidence for the upregulation of several of the aforementioned cytokines at [1,24] and before [4,25] T1D onset in humans, the published results are inconsistent across studies [3]. In our study population, we demonstrated the differences in proinflammatory cytokine (IL-1β and IL-8) levels between young newly diagnosed T1D patients compared to age-matched healthy controls. The highest level of proinflammatory cytokines was observed in patients exhibiting IA-2 and Zn-T8 autoantibodies. These signs of activation of the innate immune system are partly consistent with reports of IFN-α/γ and IL-1β pathway activation associated with altered Toll-like receptor responsiveness and enhanced nuclear factor (NF)-κB signaling in the dendritic cells (DCs) and monocytes of newly diagnosed T1D patients [9,26]. However, in contrast to the decreased IL-6 levels reported in these previous publications, we found no changes in IL-6 or TNF-α levels in our T1D patients with young age of onset.
In terms of pathogenesis, we did not detect an obvious imbalance of Th 1 /Th 2 polarization towards prominent activation of Th 1 . A recent publication showed that a dominant Th 1 -associated immune profile in the prediabetic phase could switch to a Th 3 -associated profile, with a burst of inflammatory cytokines, immediately before clinical onset of T1D [27]. These results suggest the consequences of an imbalance of the innate immune system, which would trigger islet disturbances and apoptosis that, in turn, could lead to clinical onset of the disease. Metabolic disorders could complicate the early clinical situation of T1D to which implies a systemic elevation of IL-8 in T1D patients with DKA in our study. Several ongoing clinical trials are investigating the effect of proinflammatory cytokine blockade in subjects with recent-onset T1D, demonstrating the interest in regulating the innate immune system in this disease [28].
Perhaps the most surprising and statistically significant (p < 0.005) result of the current study was the upregulation of the growth factors GM-CSF and IL-7 in the peripheral blood of T1D patients. Although GM-CSF is best known for its role in myeloid differentiation, it is also a potent growth factor for monocytes, macrophages, and DCs. Previous reports have proposed that the increased GM-CSF levels in nonobese diabetic (NOD) mice and T1D patients may represent the organisms' efforts to compensate for the defective responses of the hematopoietic cells (including bone marrow-derived DCs and pancreatic macrophages) to this growth factor  [29,30]. Supporting this suggestion, recent studies have demonstrated that GM-CSF/IL-3-deficient mice develop insulitis, precipitated by the administration of anti-CTLA-4 blocking antibodies, with destruction of insulin-producing β-cells and compromised glucose homeostasis [31]. Defects in the phagocytosis of apoptotic cells by macrophages might contribute to autoimmune diabetes by decreasing the production of immunoregulatory cytokines and increasing the production of proinflammatory cytokines [32]. However, we observed the joint elevation of GM-CSF and IL-10 in the blood of T1D patients, which may reflect the activation of protective immune mechanisms. Repeated treatment of NOD mice with GM-CSF was shown to prevent the development of insulitis by inducing tolerogenic DCs, which sustained the persistent suppressive function of IL-10-producing T reg cells [33]. Interestingly, GM-CSF was more effective when administered at later stages of insulitis. Similarly, the therapeutic potential of GM-CSF in human T1D could be speculated [34]. IL-7 is a major homeostatic cytokine for several cell types of the immune system. Overexpression of this growth factor or its receptor has been associated with the severity of autoimmune disease in animal models [35] and humans [36,37]. A study proposed that the provision of exogenous or lymphopenia-induced endogenous IL-7 promotes the expansion of self-reactive clones, even in the presence of T reg cells, thereby explaining the relevance of IL-7 in the development of diabetes [38]. However, a recent study reported the existence of diabetes-suppressive IL-17-expressing DCs that were capable of promoting the maturation of IL-7-responsive CD 25+ CD 127+ T reg cells [39]. In these T cells, IL-7 maintains the expression of FoxP3 and CTLA4, which could represent an additional non-IL2-dependent compensatory mechanism for T reg cell survival and functional activity.
We discovered an increased level of IL-27 in the peripheral blood of newly diagnosed T1D patients compared to healthy individuals. To our knowledge, this is the first study demonstrating the up-regulation of IL-27 in human T1D. Previously, this phenomenon has been described in granulomatous diseases [40], inflammatory bowel disease [41], and multiple sclerosis [42]. IL-27 is a pleiotropic cytokine of the IL-12 family with both inhibitory and activating functions on innate and acquired immunity. IL-27 is secreted by activated antigen presenting cells (macrophages, DCs) and demonstrates inhibitory effects on the development of Th 1 , Th 2, and Th 17 cells, as well as on the expansion of T reg (reviewed in [43]).
The role of IL-27 in autoimmune diabetes has been insufficiently investigated and the results obtained have been inconsistent. Blockade of IL-27 delays the onset of diabetes in NOD mice [44]. However, it was recently reported that IL-27 can inhibit streptozotocin-induced hyperglycemia and pancreatic islet inflammation in an animal model and therefore could represents a potential novel therapeutic approach for T1D [45]. In humans, the association of IL-27 polymorphisms with T1D has been reported in genome-wide association studies [46], but these results were not confirmed by others [47].
IL-27 suppresses effector Th 17 cells and promotes the generation of type 1 regulatory T (Tr1) cells, which, in turn, could dampen autoimmunity and tissue inflammation by secreting the immunosuppressive cytokine IL-10 [48]. We found that the IL-27 level correlated very strongly (rho > 0.9) with the levels of several Th 17 cytokines, but relatively weakly with IL-10 levels (rho = 0.39). Moreover, the Th 17 cytokines IL-17A, IL-21, and IL-23 and the regulatory cytokine IL-10 demonstrated notable upregulation in the T1D group compared to the control group. Taken together, these findings could support the hypothesis that several regulatory mechanisms, in particular those acting via IL-27 and IL-10, attempt unsuccessfully to dampen the harmful effects of Th 17 immunity in T1D patients. Not only Th 1 , but also Th 17 cells and hyperfunction of proinflammatory cytokines may play detrimental roles at the onset and during metabolic decompensation in recent-onset young-age T1D [1,49].
Limitations of this study include the assessment of cytokine concentrations in plasma alone. As peripheral blood cells from the current study groups were not available to us, we could not detect the spontaneous or stimulated production of cytokines by distinct immune cells. However, we believe that careful profiling of circulating cytokines in blood provides valuable data about the (dys)regulation of the immune system in vivo. The peripheral blood is arguably more appropriate for clinical purposes, especially for large cohorts with limited volumes of material (e.g., from children).
We stress the importance of the correct preparation and storage of plasma samples, as well as the avoidance of freeze/thaw cycles before the cytokines are studied. For some biomarkers, the measured level may either decrease or increase several times after repeated freeze/thaw cycles [50]. Another strength of this study was the ability to measure multiple cytokines simultaneously with the same small amount of probe by a highly sensitive method. This approach is an important element for the correct measurement of low-concentration cytokines in human serum, which often require highly sensitive assays for detection. The careful selection of the target patient population was also important, because the variability of insulin treatment length and the concurrence of other autoimmune/inflammatory diseases may significantly affect the results. The same consideration also applies to the choice of control children, who are the most difficult study group in humans.

Conclusions
Our findings broaden the current understanding of the dysregulation of systemic cytokine levels at the onset of young-age T1D. This dysregulation includes the upregulation of growth factors (GM-CSF and IL-7) and proinflammatory factors (IL-1β but not IL-6 or TNF-β) of the innate immune system, as well as Th 17 cytokines and regulatory cytokines (IL-10, IL-27).