Intestinal Dysbiosis in Autoimmune Diabetes Is Correlated With Poor Glycemic Control and Increased Interleukin-6: A Pilot Study

Intestinal dysbiosis associated with immunological deregulation, leaky gut, bacterial translocation, and systemic inflammation has been associated with autoimmune diseases, such as type 1 diabetes (T1D). The aim of this study was to investigate the intestinal dysbiosis in T1D patients and correlate these results with clinical parameters and cytokines. The present study was approved by the Barretos Cancer Hospital (Process number 903/2014), and all participants have signed the informed consent in accordance with the Declaration of Helsinki, and answered a questionnaire about dietary habits. Stool samples were used for bacterial 16S sequencing by MiSeq Illumina platform. IL-2, IL-4, IL-6, IL-10, IL-17A, TNF, and IFN-γ plasma concentrations were determined by cytometric bead arrays. The Pearson’s chi-square, Mann–Whitney and Spearman correlation were used for statistical analyses. Alpha and beta diversities were conducted by using an annotated observed taxonomic units table. This study included 20 patients and 28 controls, and we found significant differences (P < 0.05) among consumption of vegetables, proteins, milk and derivatives, spicy food, and canned food when we compare patients and controls. We detected intestinal dysbiosis in T1D patients when we performed the beta diversity analysis (P = 0.01). The prevalent species found in patients’ stool were the Gram-negatives Bacteroides vulgatus, Bacteroides rodentium, Prevotella copri, and Bacteroides xylanisolvens. The inflammatory interleukin-6 was significantly increased (P = 0.017) in patients’ plasma. Furthermore, we showed correlation among patients with poor glycemic control, represented by high levels of HbA1C percentages and Bacteroidetes, Lactobacillales, and Bacteroides dorei relative abundances. We concluded that there are different gut microbiota profiles between T1D patients and healthy controls. The prevalent Gram-negative species in T1D patients could be involved in the leaky gut, bacterial translocation, and poor glycemic control. However, additional studies, with larger cohorts, are required to determine a “signature” of the intestinal microbiota in T1D patients in the Brazilian population.

Intestinal dysbiosis associated with immunological deregulation, leaky gut, bacterial translocation, and systemic inflammation has been associated with autoimmune diseases, such as type 1 diabetes (T1D). The aim of this study was to investigate the intestinal dysbiosis in T1D patients and correlate these results with clinical parameters and cytokines. The present study was approved by the Barretos Cancer Hospital (Process number 903/2014), and all participants have signed the informed consent in accordance with the Declaration of Helsinki, and answered a questionnaire about dietary habits. Stool samples were used for bacterial 16S sequencing by MiSeq Illumina platform. IL-2, IL-4, IL-6, IL-10, IL-17A, TNF, and IFN-γ plasma concentrations were determined by cytometric bead arrays. The Pearson's chi-square, Mann-Whitney and Spearman correlation were used for statistical analyses. Alpha and beta diversities were conducted by using an annotated observed taxonomic units table. This study included 20 patients and 28 controls, and we found significant differences (P < 0.05) among consumption of vegetables, proteins, milk and derivatives, spicy food, and canned food when we compare patients and controls. We detected intestinal dysbiosis in T1D patients when we performed the beta diversity analysis (P = 0.01). The prevalent species found in patients' stool were the Gram-negatives Bacteroides vulgatus, Bacteroides rodentium, Prevotella copri, and Bacteroides xylanisolvens. The inflammatory interleukin-6 was significantly increased (P = 0.017) in patients' plasma. Furthermore, we showed correlation among patients with poor glycemic control, represented by high levels of HbA1C percentages and Bacteroidetes, Lactobacillales, and Bacteroides dorei relative abundances. We concluded that there are different gut microbiota profiles between T1D patients and healthy controls. The prevalent Gram-negative species in T1D patients could be involved in the leaky gut, bacterial translocation, and poor glycemic control. However, additional studies, with larger cohorts, are required to determine a "signature" of the intestinal microbiota in T1D patients in the Brazilian population.
Keywords: type 1 diabetes, dietary habits, intestinal dysbiosis, inflammatory cytokines, interleukin-6, glycemic control inTrODUcTiOn Type 1 diabetes (T1D) or autoimmune diabetes is a chronic disease mediated by immune reactions against pancreatic beta cells, resulting in insulin dependence to regulate blood glucose con centrations (1). The T1D pathogenesis involves the interaction of genetic and environmental factors, such as viral infections, vitamin deficiencies, and intestinal dysbiosis (2). According to the International Diabetes Federation, more than 96,000 children and adolescents under 15 years will be diagnosed with auto immune diabetes annually worldwide (3).
The gut microbiota might modulate the T1D pathogenesis via two mechanisms. In the first step, an impaired tolerance pro cess in infancy can predispose to develop autoimmune diseases, such as T1D, and might induce autoreactive T cell activation and autoantibodies (4). At the second step, intestinal dysbiosis may lead children with genetic predisposition and positive auto antibodies to develop clinical disease (4).
Studies in animal models reported an increased relative abun dance of Bacteroides, Ruminococcus, and Eubacterium genera in biobreeding diabetesprone rats, and an increased abundance of Bifidobacterium and Lactobacillus in diabetesresistant rats (5). Also, the prevalence of the Bacteroidetes phylum members could promote increased intestinal permeability and precede the clini cal onset of T1D in animal models of the disease, in prediabetic patients and in diabetic subjects (6,7).
In T1D children, there is decreased gut microbiota diversity and abundance of mucindegrading and butyrateproducing members, and reduced Firmicutes/Bacteroidetes ratio, along with Lactobacillus, Bifidobacterium, and Prevotela species (8). Smaller relative abundance of lactateproducing bacteria, including the Bifidobacterium longum, and increased Clostridium, Bacteroides, and Veillonella species are equally detected (9).
Moreover, intestinal dysbiosis was detected in prediabetic children with genetic predisposition and beta cells autoantibodies (10). Increased Bacteroides dorei and Bacteroides vulgatus species in seroconverted T1D patients are registered 8 months prior to beta cell autoimmunity, suggesting that early dysbiosis may predict T1D in genetically predisposal subjects (11). Additionally, children with beta cell autoantibodies exhibited increased Bacteroidetes members and decreased lactate and butyrate producing bacteria (12). These studies support the hypothesis that there is a gut microbiome signature associated with T1D development in seropositive children (13).
Previous reports suggested the involvement of the gut mucosa in the pathogenesis of islet autoimmunity in T1D (14). Thus, the modulation of interactions between commensal microbiota and gutassociated lymphoid tissues may represent a means to affect the progression of the autoimmune diabetes (15). Based on previous studies that identify that dysbiosis may be strongly correlated with barrier disruption, bacterial translocation, and autoimmune diseases development (16), we hypothesized that the prevalence of Gramnegative bacteria in the gut mucosa from T1D patients is greater in patients than in stool samples from controls, and positively correlated with poor glycemic control and systemic inflammatory cytokines. In the present study, we investigate the intestinal dysbiosis in T1D patients and correlated these results with clinical data and systemic inflam matory cytokines.

Patients and controls enrollment
Type 1 diabetes patients with fasting blood glucose greater than 126 mg/dL (3) at diagnosis were enrolled by the physician from the endocrinology department from Board of Health from Barretos, Sao Paulo, Brazil, from June 1st, 2015 to July 30th, 2016. A total of 20 patients, 14 females and 6 males (mean age ± SD = 23.1 ± 8.6 years), were included in the present study.
Twentyeight healthy subjects, 18 females and 10 males (mean age ± SD = 25.2 ± 9.8 years), were enrolled in the present study that was performed in accordance with the recommendations of Ethics committee from Barretos Cancer Hospital. All par ticipants have signed the informed consent in accordance with the Declaration of Helsinki. The present study was approved by the Barretos Cancer Hospital (Process number 903/2014). After the consent, the peripheral blood was collected and stool samples were delivered within 5 days. Stool and plasma sam ples were stored at −80°C until DNA extraction and cytokine quantification.
At enrollment, all of the subjects answered a questionnaire about dietary habits, such as daily consumption of vegetables, fresh fruits, carbohydrates, proteins (meat/eggs), trans fat, milk and derivatives, spicy food, canned food, hot drinks (coffee/tea), and alcohol. Exclusion criteria include use of antiinflammatories, antibiotics, laxatives, vaccination, and corticosteroids in the last 30 days. Chronic diarrheas and gastrointestinal surgeries were also considered as exclusion criteria for patients and controls.
Anthropometric measurements and clinical data from T1D patients, such as weight, height, fasting blood glucose and gly cated hemoglobin (HbA1c), and disease duration were recorded. The BMI mean was 23.9 ± 3.6 (nine patients presented BMI < 25 and seven patients were overweighed, with BMI > 25). The fast ing blood glucose mean was 236.8 ± 135.9 mg/dL (five patients with controlled blood glucose < 126 mg/dL, 11 patients with uncontrolled blood glucose > 126 mg/dL, and four patients without this clinical data). The HbA1C mean was 9.8 ± 1.8% (13 patients with poor glycemic control, HbA1C > 7.5%, one patient with moderate glycemic control, HbA1C = 7%, and six patients without this clinical data). The disease duration mean was 14.0 ± 7.2 years. Demographic characteristics, anthro pometric measurements, and clinical data from T1D patients, were summarized in Table 1.   cytokine Determination by cytometric Bead array

Dna extraction and Bacterial 16s sequencing
Peripheral blood (10 mL) was collected from T1D patients and controls and plasmaEDTA was isolated by centrifugation at 1,372 g, for 10 min, 4°C. Cytokine detection was performed by cytometric bead array (Human Th1/Th2/Th17 Cytokine Kit, BD Biosciences, CA, USA). Plasma levels of IL2, IL4, IL6, IL10, IL17A, TNF, and IFNγ were determined by flow cytometer FACS Canto™ II (BD Biosciences). Analyses was performed by BDFCAP array™ software and data were expressed by conversion of the median fluorescence intensity in picogram per milliliter.

statistical analyses
Data extracted from the questionnaires containing dietary habits were analyzed by Pearson's chisquare. Comparisons between cyto kines plasma concentrations in T1D patients and controls subjects were performed by Mann-Whitney. Correlations among reads percentages of the intestinal microbiota, cytokines plasma concentrations, and clinical data were performed by Spearman correlation. Analyses of variance, diversity indexes, and alpha and beta diversities were performed by using annotated Operational Taxonomic Units table. Sequencing analyses of the bacterial 16S was performed as described in our previous work (18). P values less than 0.05 were considered statistically significant.

Detection of intestinal Dysbiosis in T1D Patients
To investigate the dysbiosis in T1D patients, we sequenced the 16S bacterial DNA stool samples from patients and controls and calculated alpha and beta diversities. According to the rarefac tion curves, there were no significant differences (P = 0.318) in richness and evenness between samples obtained for patients and controls (Figures 3A,B; Table 2). However, we detected significant differences (P = 0.01) between microbial community found in T1D patients and controls, when we performed the weighted and unweighted UniFrac metric with Bonferroni's correction analyses (Figures 3C,D).

DiscUssiOn
The gastrointestinal tract hosts approximately 100 trillions of bacteria that reside in mucosal surfaces and constantly interact with immune cells (14). This microbial community function as microbiological defense barrier, induce antimicrobial peptides secretion and immunological responses that increase mucosal and systemic immunity (19). Previous studies have focused on the role of commensal microbiota in health maintenance and disease development and environmental factors that influence its dynamics (20,21). The equilibrium between commensal microbiota and host is characterized by microbiota members that improve metabolism and protect against pathobionts and gut inflammation (22). Several evidences suggest that alterations in function and diversity of the gut microbiota might be linked to the development of autoimmune diseases, including T1D (23). The hypotheses proposed to correlate dysbiosis with auto immune diseases include bystander Tcell activation, molecular mimicry, amplification of autoimmunity by inflammatory mil leu, induced by dysbiotic microbiota, and recently proposed, the posttranslational modification of luminal proteins by enzymes from dysbiotic microbiota (24). This altered posttranslational modification of luminal proteins could produce neoepitopes that may become immunogenic and trigger systemic auto immunity and autoimmune diseases (24).
Recent reports have demonstrated the influence of dietary habits in the gut microbiota composition (25)(26)(27)(28). In the study from Wu and coworkers (2011), they evaluated the association between diet and gut microbiota composition, in a cohort of 98 healthy subjects. Authors reported that the Bacteroides genera was associated with animal protein and saturated fat consump tion, while Prevotella genera was linked to carbohydrates and simple sugar intake (27).
In another study, Yamaguchi and colleagues evaluated dietary habits, metabolic markers, and fecal microbiota in 59 T2D patients. In this study, they reported a correlation between high carbohydrates, fat, and protein consumption with increased counts of Clostridium clusters IV and XI. Bifidobacterium species, Lactobacillales, and Bacteroides species were inversely correlated with carbohydrate, protein consumption, and fasting blood glucose, respectively. Authors concluded that low consumption of proteins and carbohydrates favors a healthy gut microbiota and improve glucose tolerance in T2D patients (28). In our work, we detected an inverse correlation between fresh fruits intake and Bacteroides, B. vulgatus, and B. rodentium abundances, and between protein consumption with Clostridiaceae, Oscillospira, and Oscillospira eae relative abundances.
In the present study, we reported an intestinal dysbiosis in T1D patients, with significant differences in the diversity of the gut microbiota, represented by betadiversity analysis between patients and controls. Study from Murri and colleagues inves tigated the gut microbiota in 16 T1D children and 16 control subjects and demonstrated an increased numbers of Clostridium, Bacteroides, and Veillonella genera in patients when com pared with control counterparts. Furthermore, the Firmicutes/ Bacteroidetes ratio and relative abundances of Lactobacillus, Bifidobacterium, and Prevotela members were decreased in T1D patients (9). Moreover, authors showed correlations among Lactobacillus, Bifidobacterium, and Clostridium members with glucose plasma levels (9).
In this study, we detected the prevalence of gramnegative species in stool samples from our T1D patients, including B. vulgatus, B. rodentium, P. copri, and B. xylanisolvens, support ing our hypothesis and suggesting an increase in the bacterial translocation through the epithelial barrier, triggering systemic inflammation, increased oxidative stress, metabolic deregula tion, and beta cell destruction (14). Previous studies in animal models showed that gut microbiota translocation to pancreatic lymph nodes triggers NOD2 activation, Th1 and Th17 differen tiation, which contribute to inflammatory infiltrate inside the pancreatic islets and T1D development (29).
The adult healthy intestinal microbiota is composed by Firmi cutes, Grampositive members, and Bacteroidetes, Gram negative members (30). The prevalent species from Firmicutes phylum are Faecalibacterium prausnitzii and Eubacterium rectale/ Roseburia species (31). These bacteria produce shortchain fatty acids, including the butyrate, which suppress NFκB signaling in the intestinal epithelial cells (31,32). The Bacteroidetes phylum, the second most prevalent in the human intestine, is dominated by Bacteroides and Prevotella species (33). The Prevotella members are able to activate TLR2 receptors and induce Th17 CD4 T cell differentiation. The prevalence of Bacteroides and Prevotella spe cies is associated with gastrointestinal inflammation, triggered mainly by the inflammatory Th17 cytokines (34). Additionally, Prevotella species induce IL8 and IL6 release by epithelial cells, inducing Th17 immune responses and neutrophil recruit ment (34). Thus, inflammation of the gastrointestinal mucosa, induced by Prevotella species might promote dissemination of inflammatory mediators, barrier dysfunction, and bacterial translocation, which induce and amplify the systemic inflam mation (34).
Previous reports showed that T1D children with autoantibo dies have an increase in the Gramnegative Bacteroidetes members, reduction in mucindegradation and butyrateproducing species (10,35,36). Butyrate has an antiinflammatory effect, induces T regulatory cells in the gut mucosa, and enhances the barrier via tightjunctions expression (37,38). In addition, researchers demonstrated that T1D children exhibit diminished abundance of lactateproducing bacteria, including B. longum, subspecies infantis. Bifidobacterium species can exert several functions, such as carbohydrate fermentation, acetate and lactate generation, polyphenols and linoleic acids release, and antioxidant activities (39). They also play an important role in the gastrointestinal lym phoid tissue maturation in early life and offer protection against pathogens by bacteriocin release, decrease in luminal pH, and inhibition of epithelial adhesion (40). In agreement with these studies, we detected the prevalence of Gramnegative bacteria in stool samples from our T1D patients and decreased butyrate producing bacteria, such as Bifidobacterium and Roseburia species, and decreased Clostridium members that could induce antiinflammatory IL10 cytokine and T regulatory cells (41).
The germfree NOD mice have a similar metabolite profile to that of prediabetic children, with reduced glycemic control and deregulated immunologic and metabolic responses (42). The absence of a gut microbiota in this mice did not affect the diabetes incidence but promoted insulitis and increased levels of proinflammatory IFNγ and IL12, suggesting the important role of the microbiota in glucose metabolism (42). Our data sug gest that T1D patients with intestinal dysbiosis, with decreased abundance of beneficial microbes, such as Bifidobacterium and Lactobacillales members have poor glycemic control, represented by high levels of HbA1c. The standard method to monitor gly cemic control is the measurements of the HbA1c, and the good glycemic control is related to decreased microvascular complica tions in T1D patients (3,43,44). The poor glycemic control is defined as HbA1c between 59 and 106 mmol/mol or 7.5-12% (43). About 25% of T1D patients' population is in persistent poor glycemic control (44), which is in accordance with our present study, that 13 patients (65%) presented poor glycemic control. Our hypothesis relies on that intestinal dysbiosis and bacterial translocation can induce systemic inflammation and suppression of insulin receptors, promoting an increased glucose blood levels and high levels of HbA1c in T1D patients. Gram negativederived LPS binds to TLR4 and trigger the inflamma tory cascade, resulting in NFκB activation and secretion of inflammatory mediators, such as TNF, IL1, and IL6, which influence glucose metabolism and inhibit phosphorylation of insulin receptors (45)(46)(47).
In the present study, T1D patients presented increased inflam matory IL6 concentrations that inversely correlated with Ruminococcaceae family, an anaerobe microorganism that degrade complex carbohydrates, and are common in subjects with carbohydrateenriched diets (48). The inflammatory TNF inversely correlated with Clostridiaceae members that include Clostridium species, which can induce T regulatory cells in the gut mucosa. The Clostridium species, especially clusters IV and XIVa, are sporeforming members from the gut microbiota and can induce T regulatory cells differentiation in the gut mucosa (41).
Proposed mechanisms to explain the link between intestinal dysbiosis and T1D include the dysbiosisassociated immunologi cal deregulation, leading to destruction of βcells by autoreactive T cells (14,49). The second proposed mechanism correlates T1D with leaky gut, bacterial translocation, endotoxemia, and chronic lowgrade inflammation (12,50). The Gramnegative bacte rial species, prevalent in stool samples from our T1D patients, enhance gut permeability and increase bacterial translocation, which can lead to systemic inflammation and metabolic deregu lation (14,50).
Finally, we concluded that there are different gut microbiota profiles between T1D patients and healthy controls. The prevalent Gramnegative species in T1D patients could be involved in the leaky gut, bacterial translocation, and poor glycemic control. However, additional studies, with larger cohorts, are required to determine a "signature" of the intestinal microbiota in T1D patients in the Brazilian population.

eThics sTaTeMenT
The present study was performed in accordance with the recom mendations of Ethics committee from Barretos Cancer Hospital. All subjects provided written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the Barretos Cancer Hospital (Process number 903/2014). aUThOr cOnTriBUTiOns BH: T1D patients' enrollment, DNA extraction, cytokine deter mination, data acquisition, and manuscript writing; MG: controls' enrollment and sample collection; NR and NS: V3/ V4 amplification, library construction, and sequencing; JB: support for blood samples collection; EM: support to Illumina platform sequencing; JP: responsible for clinical data from T1D patients. WO and DP: bioinformatics analyses; VM: sample col lection, DNA quantification, and cytokine determination; GO: experimental design, data interpretation, manuscript writing, and revision.

FUnDing
The study was supported by Foundation for the Support of Research in the State of São Paulo (FAPESP, grant numbers #2016/502040 and #2018/137027) and School of Health Sciences from Barretos Dr. Paulo Prata (PAP#2016).