TLR5 influences the development of type 1 diabetes

Abstract

Introduction

In mammalian and human life, it is important that the immune system defends against microorganisms. Although there is a huge overlap, innate cells are good against bacteria, whereas T cells are good against viruses, mainly because of antibody production via T helper and B lymphocytes. Toll-like receptor 5 (TLR5) is a regulator; when it is highly expressed, T cells are inhibited, and innate cells are favored. In glucose-activated pancreatic islets, TLR5 gene expression has been found to be highly upregulated, and the islets may therefore be protected from T cell destruction resulting in autoimmune type 1 diabetes (T1D).

Research design and methods

We investigated mRNA from the islets of Langerhans in patients with newly diagnosed T1D for TLR5 gene expression, as well as Tlr5 and Cd3 expression in the whole pancreatic tissue of female diabetic non-obese diabetic (NOD) mice. Also, we examined for polymorphisms between TLR5, immunological parameters, and T1D.

Results

Islet mRNA for TLR5 was downregulated by one-third of patients with newly diagnosed T1D, compared with controls, and correlated inversely with T cell infiltration in the islets. Moreover, the association between TLR5 and T cells was supported by a corresponding correlation of Tlr5 and Cd3 expression in the pancreatic tissue of diabetic NOD mice. Regarding polymorphisms, two associations were found between TLR5 and monocytes. Also, a significant polymorphism was seen concerning TLR5 and T1D.

Conclusions

In the present study, we find a low expression of mRNA for TLR5 in patients with newly diagnosed T1D associated with enhanced T cell infiltration. T cells are important for autoimmune diseases, including T1D. We hope that the present findings may be influential for the understanding of how T1D develops.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• It is known that Toll-like receptor 5 (TLR5) expression increases when beta cells are stressed by a high activity. T cells are important for the development of autoimmune diseases, including type 1 diabetes (T1D), but a potential protective role of TLR5 on T cell-driven autoimmunity was unknown. We therefore aimed to investigate the role of TLR5 in the pathogenesis of T1D by examining (1) its expression in the islets of patients with newly diagnosed T1D, (2) its association with autoimmune T cell infiltration, and (3) relevant polymorphisms.

WHAT THIS STUDY ADDS

• We find a low expression of TLR5 in the islets of patients with newly diagnosed human T1D. Furthermore, an inverse correlation between TLR5 and the degree of insulitis is demonstrated for patients with newly diagnosed human diabetes, with a corresponding inverse correlation for Tlr5 and Cd3 in diabetic non-obese diabetic mice. Hence, we find an association between islet TLR5 and development of T1D.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Our findings add to the understanding of how T1D develops and focus on TLR5 as a key molecule. Although the number of patients is limited, our present findings are supported in an animal model for T1D and may help find new ways of preventing T1D.

Introduction

Mammalian organisms possess both the innate and the adaptive immune systems. Therefore, in each situation with an infection, the immune system may react with granulocytes or lymphocytes, or both. Overall, bacteria may be fought with the innate immune cells, and viruses with antibodies via T cell stimulation, which, on the other hand, gives risk of autoimmunity.¹

Toll-like receptor 5 (TLR5) is a component of the innate immune system, primarily stimulated by bacterial flagellin.² In 2011, we found that TLR5 is the only TLR molecule that is upregulated on beta cells following glucose stimulation.³ Initially, the purpose of this upregulation was unclear, but it was hypothesized to be a defense mechanism for beta cells, which are more vulnerable when being active.⁴ It has been shown that high TLR5 boosts the innate immune system and granulocyte activity.⁵ TLR5 expression may inhibit T cells, often through induction of regulatory T cells.6,8 In contrast, low TLR5 expression may likely stimulate T cells and promote autoimmunity; in non-obese diabetic (NOD) mice, TLR5 deficiency has been linked to increased risk of developing diabetes.⁹

In the present study, we aimed to examine the TLR5 expression in the islets of Langerhans in patients with newly diagnosed type 1 diabetes (T1D). Additionally, possible polymorphisms related to the TLR5 system were investigated.

Methods

RNA analysis of human pancreatic islets

Human pancreatic tissue used in this study was obtained from two cohorts: the Diabetes Virus Detection (DiViD) study,¹⁰ which included patients with newly onset T1D (disease duration: 35 days) aged 24–35 years (n=5); and the network for Pancreatic Organ Donors with Diabetes (nPOD) study,¹¹ which provided samples from a healthy control group (n=18). For dataset access requests, please contact lars.krogvold@gmail.com. The nPOD study involves pancreatic tissue from organ donors following accidental death, whereas the DiViD study includes samples from living individuals. The nPOD study samples consist of 18 control persons (9F/9M) aged 36.2±15.6 (mean±SD) years; 12 autoantibody-positive persons (5F/7M) aged 20.4±8.4 years; 19 persons (10F/9M) 5 years after diagnosis with T1D aged 33.2±17 years; and 8 persons (5F/3M) with type 2 diabetes (T2D) aged 39.8±13.4 years. Frozen sections of 25 randomly selected pancreatic islets were subjected to laser capture dissection, and islets from two to five sections were pooled. RNA extraction was done using the Arcturus PicoPure RNA Isolation Kit (Applied Biosystems, Grand Island, New York, USA). RNA quantification was performed using a Bioanalyzer 2100 instrument (Agilent Technologies, Santa Clara, California, USA). Gene expression analysis was conducted using Affymetrix expression arrays (GeneChip Human Gene 2.0 ST, Thermo Fisher) and normalized using global scaling. All tissue handling procedures were conducted in the same laboratory to ensure consistency in the comparison. Furthermore, the analysis was performed using the same equipment, and the RNA quality was certified by RNA Integrity Number (RIN) value measurements (>3.5).

RNA analysis of murine pancreatic tissue

Five-week-old NOD/MrkTac female mice (n=10) from Taconic (Germantown, New York) were group housed (three to four mice/cage) in a rodent Specific-Pathogen-Free (SPF) barrier facility. Measurements of tail blood glucose commenced two times per week from 10 weeks of age, and a mouse was considered diabetic when blood glucose levels exceeded 12 mmol/L on two consecutive days. Mice were euthanized when diagnosed as diabetic, and the whole pancreas was immediately dissected and snap frozen in TRIzol (Thermo Fisher). RNA isolation was performed as previously described.³ cDNA synthesis was performed using the High-Capacity cDNA Reverse Transcriptase Kit (Applied Biosystems, Foster City, California, USA), followed by quantitative PCR (qPCR) analysis of Tlr5 and Cd3 (Probe IDs: TLR5, Mm00546288_s1; Cd3, Mm01179194_m1; Applied Biosystems) using TaqMan gene expression assays on a Bio-Rad C1000 Touch CFX96 Real-Time System Thermal Cycler (Bio-Rad, Hercules, California, USA), as previously described.¹² Actb (Probe ID: Mm00607939_s1) was used as a reference gene, and statistical analysis was done on delta cycle threshold values.

Genetic association analysis

Genome-Wide Association Study (GWAS) data were retrieved from ref 13,17, and associations were analyzed for TLR5 with innate immune cell numbers and T1D defined by American Diabetes Association (ADA) criteria. Blood glucose levels were evaluated in individuals free of diabetes defined by physician diagnosis, medication use, or fasting glucose <7 mmol/L.

Statistical analysis

The statistical analysis for RNA expression was conducted using GraphPad Prism V.8.0.2 (GraphPad, La Jolla, California, USA). Data were tested for normality and equal variances using Anderson-Darling and Brown-Forsythe tests, respectively. For comparison between the human patient groups, an analysis of covariance adjusted for age and sex was employed, followed by Tukey’s multiple comparison test. A partial correlation analysis compared TLR5 expression and CD3 in the islets of patients with newly diagnosed T1D, adjusted for age. A Spearman correlation analysis was performed comparing Tlr5 and Cd3 expression in the NOD mice pancreas. Statistical significance was considered when p<0.05. All tendencies are given when p<0.1.

Ethics statement

All methods were carried out in accordance with relevant guidelines and regulations. Informed consent was obtained from all subjects and/or their legal guardians. The animal study was approved by the Animal Experimentation Inspectorate of the Ministry of Food, Fisheries, and Agriculture in Denmark (License No 2021-15-0201-00847). The experiments were conducted following the Danish Animal Experimentation Act (Consolidation Act No 474 of 15 May 2014) and the European Union (EU) Directive 2010/63/EU on protecting animals used in scientific research.

Data and resource availability

Data are being deposited with datadryad.org, https://doi.org/10.5061/dryad.j9kd51cpz. The protocols used can be obtained on request to the corresponding author.

Results

In patients with newly diagnosed T1D, the mRNA expression of TLR5 in the islets of Langerhans was significantly reduced by one-third compared with healthy control individuals (34.1±17.8 vs 50.5±11.2, p=0.029) (see figure 1). Furthermore, there was a strong tendency (p=0.059) for patients with newly diagnosed T1D to have reduced TLR5 expression compared with all the autoantibody-positive persons, and a minor tendency (p=0.088) compared with those with T2D, whereas individuals who were diagnosed with T1D >5 years ago were not significantly different from patients with newly onset T1D. Individuals with pre-diabetes and who were autoantibody positive had the same expression of TLR5 mRNA as the healthy controls (see figure 1). Hence, the reduction in TLR5 expression seems to occur when the insulitis becomes malignant during the transition to overt diabetes. Four of the autoantibody-positive persons had two autoantibodies, which were either Glutamic Acid Decarboxylase (GAD) antibody, Insulin Auto Antibody (IAA), Islet Antigen 2 (IA2) antibody or Zinc Transporter 8 (ZNT8) antibody, and their TLR5 values were 52.0±12.1. The other eight autoantibody-positive persons had GAD autoantibodies only, and their TLR5 values were 47.4±11.1. Hence, no difference was found between the two autoantibody-positive groups and control groups.

Figure 1. Toll-like receptor 5 (TLR5) gene expression. Boxes represent the 25% and 75% quartiles and whiskers indicate 1.5× the IQR. All significant p values and tendencies (p<0.1) are shown in the figure. The remaining group comparisons were non-significant. Taking all 12 autoantibody-positive persons together, they tended to be significantly higher than patients with newly diagnosed T1D (p=0.059). AB+, non-diabetic, autoimmune, single autoantibody-positive donors, network for Pancreatic Organ Donors with Diabetes (nPOD) (n=8); AB2+, non-diabetic, autoimmune, double autoantibody-positive donors, nPOD (n=4); CTR, non-diabetic controls (n=18); T1D (median 5 years), donors with intermediate-duration type 1 diabetes, nPOD (n=20); T1D (median disease duration, 35 days), donors with newly diagnosed type 1 diabetes, Diabetes Virus Detection study (DiViD) (n=5); T2D (median 2 years), donors with type 2 diabetes, nPOD (n=8).

Moreover, patients with T1D with disease duration of 5 years, along with patients with T2D, were also not significantly different from the healthy controls. There was found to be no correlation between TLR5 and C-peptide values in any of the groups. The insulitis scores for patients in the DiViD study have been previously reported.¹⁰ After adjusting for age, there was a significant inverse correlation, showing that the higher the TLR5 expression, the lower the T cell numbers in the islets (figure 2). This was supported by analyzing the gene expression of Tlr5 and Cd3 in the whole pancreatic tissue from newly diagnosed diabetic NOD mice, which similarly demonstrated a significant inverse correlation (figure 3). These qPCR data were obtained from female NOD mice that had already developed overt diabetes. Consequently, the observed inverse correlation between Tlr5 and Cd3 expression reflects the situation after disease onset and cannot establish whether reduced Tlr5 expression precedes or follows T cell infiltration. Longitudinal studies in pre-diabetic NOD mice and in vitro mechanistic experiments are required to determine causality.

Figure 2. Association between islet TLR5 expression and T lymphocyte numbers. Correlation of islet TLR5 gene expression in donors with newly diagnosed type 1 diabetes, Diabetes Virus Detection study DiViD (n=5), and intraislet CD3+ cells. Each islet is normalized to the area of an IEQ, defined as an islet with a diameter of 150 µm or an average cross-sectional area of 0.010936 mm². TLR5, Toll-like receptor 5.

Figure 3. Association between pancreatic Tlr5 and Cd3 gene expression in diabetic NOD mice. Correlation of pancreatic Tlr5 and Cd3 gene expression in female NOD mice (n=10) diagnosed with diabetes by blood glucose levels >12 mmol/L. ΔCt values, that is, Ct gene−Ct reference gene, are shown and correlated by Spearman analysis. ΔCt, delta cycle threshold; NOD, non-obese diabetic.

The analysis of polymorphisms associated with the TLR5 system is presented in table 1; in support of our findings, one single nucleotide polymorphism (SNP) was correlated with T1D (rs113292043). Interestingly, SNP rs5744168 was highly significantly associated with monocyte and neutrophil counts.

Table 1. Polymorphism related to TLR5.

Lead variant (CHR:POS:REF:ALT)	dbSNP	Phenotype	P value
1: 223285200:G:A	rs5744168	↓ Monocyte percentage ↓ Monocyte count ↑ Neutrophil percentage ↑ Neutrophil count	1.77e-16 3.21e-12 1.40e-4 6.83e-3
1: 223231657:C:T	rs113292043	↓ Monocyte count ↓ Monocyte percentage ↓ Type 1 diabetes	1.21e-14 1.24e-14 1.3e-2

Data retrieved from ref 14 16 17.

dbSNP, Single Nucleotide Polymorphism Database; TLR5, Toll-like receptor 5.

Discussion

In the present study, we have demonstrated a significant difference in TLR5 expression in the islets of Langerhans between patients with newly onset T1D and healthy controls, and the TLR5 expression correlated inversely to islet-infiltrating CD3 T cells in the same patients, as well as in diabetic NOD mice. Additionally, we identified a TLR5-related polymorphism associated with T1D. TLR5 expression was partially normalized in patients with T1D with a duration of 5 years. At that time, the destructive diabetogenic inflammation has largely completed, and the patients are most often in a stable condition when having an appropriate insulin treatment.

Gram-negative bacteria with flagellin stimulate TLR5, which enhances granulocyte activity to eliminate these bacteria. Experience with flagellin due to low hygiene standards seems natural in Russian Karelia, which also has low incidence of T1D.¹⁸ The hygiene theory is indeed all about exposure to activated bacteria, which, via TLR5-related mechanisms, may explain the lower incidence of autoimmunity.¹⁹

Bacterial infections are not an obvious job for T cells which are present to react against tumor cells and virus, and to aid the production of antibodies via T helper and B lymphocytes. Importantly, the development of T1D does require islet destruction by T cells; this is supported by many studies, for example, treatment of patients with T1D with CD3 antibodies is beneficial as it reacts against aggressive T cells.²⁰

Internal factors such as pregnancy can influence TLR5 gene expression, leading to its upregulation.²¹ This may take place in order to prevent T cells from destroying the fetus which expresses paternal antigens. Interestingly, the beta cells produce 40% more insulin during pregnancy, reflecting a high activity. We have previously demonstrated that such high activity is associated with an increased risk of developing T1D.^{4 22} Yet, despite the seemingly increased risk, there is no increase in the incidence of T1D during the first two trimesters.²² In the third trimester, there is a 3.8 times increased risk of developing T1D but not more, despite the even higher pressure on the beta cells during this time.²² These observations align well with our results, highlighting the potential implications the TLR5-regulated T cell activity may have on T1D.

A highly interesting point is whether the TLR5 expression can be influenced by exogenic factors. It has been described that bacteria can invade the ductus pancreaticus and be found in the pancreas of humans.²³ These include Staphylococcus aureus,²³ which has been reported to inhibit the expression of TLR5.²⁴ Fusidic acid, an antibiotic targeting Staphylococcus, has also been shown to reduce the incidence of diabetes in BioBreeding (BB) rats.²⁵ Additionally, Helicobacter pylori warrants mention, as its presence has been shown to increase T1D incidence in humans by 1.77 times²⁶ and is able to inhibit TLR5 by masking its expression.²⁷ H. pylori is often present in the stomach and duodenum, but no studies have examined whether it is present in the pancreas. Thus, highly relevant exogenous factors related to T1D risk exist, which can affect TLR5 expression.

Polymorphisms of the TLR5 gene have been associated with innate immune responses in other inflammatory diseases as well, including systemic lupus erythematosus.²⁸ Notably, the SNP rs5744168 of TLR5 generates a stop codon that could affect TLR5 function,²⁹ whereas rs113292043 is an intergenic variant with no known functional consequences. In addition to T1D, we have also shown that the same polymorphisms correlate with reduced number of monocytes, increased number of neutrophils, and rs113292043 with reduced T1D risk.

It should be acknowledged that the DiViD material consisting of living patients is very small (n=5). This restricts the generalizability of the findings, and confirmation in larger, multicenter cohorts is needed. The connection between TLR5 and T cells may not be finally settled, but the results in this study strengthen the association. It is also supportive that an effect of low TLR5 has been found to enhance development of diabetes in NOD mice.⁹ Furthermore, a higher diabetes incidence has been described in germ-free male NOD mice that have had no opportunity for experiencing TLR5 stimulation.³⁰

In conclusion, a treatment strategy against early development of T1D could involve flagellin to increase TLR5 expression, and thereby naturally inhibiting T cells and potentially avoiding autoimmunity.

Data availability statement

Data are available upon reasonable request.