Insulin receptor expressing T cells appear in individuals at risk for type 1 diabetes and can move into the pancreas in C57BL/6 transgenic mice

Abstract

Insulin receptor (IR) expression on the T cell surface can indicate an activated state; however, the IR is also chemotactic enabling T cells with high IR expression to physically move toward insulin. In humans with type 1 diabetes (T1D) and the non-obese diabetic (NOD) mouse model, a T cell-mediated autoimmune destruction of insulin producing pancreatic β-cells occurs. In previous work, when purified IR positive and IR negative T cells were sorted from diabetic NOD mice and transferred into irradiated nondiabetic NOD mice, only those that received IR⁺ T cells developed insulitis and diabetes. Herein, peripheral blood samples from individuals with T1D (new-onset to 14 yrs of duration), relatives at high-risk for T1D, defined by positivity for islet autoantibodies, and healthy controls were examined for frequency of IR⁺ T cells. High-risk individuals had significantly higher numbers of IR⁺ T cells as compared to those with T1D (P <0.01) and controls (P<0.001); however, the percentage of IR⁺ T cells in circulation did not differ significantly between T1D and control subjects. With the hypothesis that IR⁺ T cells traffic to the pancreas in T1D, we developed a novel mouse model exhibiting a FLAG-tagged mouse IR on T cells on the C57BL/6 background, which is not susceptible to developing T1D. Interestingly, these C57BL/6-CD3FLAGmIR/mfm mice showed evidence of increased IR⁺ T cell trafficking into the islets compared to C57BL/6 controls (P<0.001). This transgenic animal model provides a novel platform for investigating the influence of IR expression on T cell trafficking and the development of insulitis.

Introduction

In persons with type 1 diabetes (T1D) and the non-obese diabetic (NOD) mouse model of disease, autoimmune mechanisms are key to the disorder’s development (1, 2). NOD mice develop a mononuclear cell infiltrate restricted to the pancreatic islets, termed insulitis, which is associated with selective destruction of the insulin-producing β-cells (3). Splenic T cells obtained from diabetic NOD mice are capable of transferring disease to irradiated young (6–8 weeks old) nondiabetic NOD recipients (4, 5). Pancreatic tissue from humans with T1D share a cellular lymphocytic infiltrate in the islets, albeit with markedly lower frequency and intensity than in NOD mice (6). Recent studies have identified islet infiltrating T cells expressing T Cell Receptors (TCRs) specific for known islet-antigens (e.g., insulin, proinsulin, glutamic acid decarboxylase (GAD) in both humans and NOD mice (7, 8) highlighting the antigen-specific autoimmune component to β-cell destruction (9–11).

MHC class I expression is increased on pancreatic β-cells in both NOD mice and humans with T1D (12, 13). MHC class II expression has not been observed on pancreatic β-cells in NOD mice (13) and transgenic expression of MHC class II on pancreatic β-cells does not lead to hyperglycemia (14). Recently low MHC class II expression has been reported on murine pancreatic β-cells, but only in the CD4⁺ pathogenic T cell NOD 4.1 mouse model and after transfer of BDC2.5 T cells into NODscid mice (15). When Hamilton-Williams et al. (16) used transgenic NOD mice that lacked MHC class I expression specifically on pancreatic β-cells, both CD8⁺ and CD4⁺ T cells still migrated into the pancreas. However, the insulitis was benign and did not progress to β-cell destruction without the presence of MHC class I on the pancreatic β-cells. Serreze et al. demonstrated that splenocytes obtained from diabetic NOD mice could transfer diabetes into the NOD-scid.β₂M^null mice (MHC class I negative), whereas prediabetic NOD splenocytes required MHC class I expression to transfer diabetes (17). Therefore, initiation of autoimmune diabetes requires MHC class I-dependent T cell responses, but once activated, those T cells can destroy pancreatic β-cells in an MHC class I-independent manner (17). Recently Sandor et al. (18) published that the presence of an antigen is not necessary for T cell trafficking into the islet, but depends on T cell chemokine receptor signaling. However, there may be other antigen independent avenues for migration into the pancreas. Thus, questions remain as to how T cells migrate into the islets if the T cells are not attracted by signals resulting from T cell-MHC/antigen-specific interactions.

The insulin receptor (IR) represents a key molecule that binds insulin and mediates a variety of cellular responses including glucose transport, endocytosis, cell differentiation and cell proliferation (19). IR is not detectable on resting or naive T cells (20) but is expressed on the surface of activated immune T cells in humans and rodents (21, 22). Thus, when T cells are activated with mitogens, glucocorticoids, allogeneic cells or their cognate antigens, IR expression is increased on the cell surface (20, 23–25). Moreover, in humans with diabetic ketoacidosis, an inflammatory state, oxidative stress or increased cytokine production, such settings lead to the upregulation of IR on T cells (26). The presence of the IR on T cells also plays an important role in hormonal modulation of lymphocyte immune function, enhancing cytotoxic T cell function, permitting mature cell differentiation, and maintaining the activated state of the lymphocyte, after mitogen or antigen challenge, when insulin is present (27).

A less understood function of the IR is its role in chemotaxis. The only endogenous source of insulin are the β-cells that exist within the pancreatic islets (28). It has been shown that mitogenic phytohemagglutinin (PHA) activated human T cells (29) and IR transfected Chinese Hamster Ovary (CHO) cells display increased chemotaxis toward insulin in vitro (30). Therefore, cells with high density expression of IR are capable of physically moving towards a concentrated source of insulin. In previous studies we demonstrated that approximately 20% of T cells in diabetic NOD mice express IR (31). Furthermore, transfer of purified IR⁺ T cells caused both insulitis and diabetes in young non-diabetic irradiated NOD recipients, whereas transfer of IR⁻ T cells caused neither insulitis nor diabetes (31). In NOD mice, IR⁺ T cells accumulate in the islets before overt diabetes development, indicating that IR expression may be an early marker for pathogenic T cells (31). Thus, IR expression could potentially be used to distinguish T cells capable of reaching the islet potentially based on chemotaxis toward insulin emanating from the pancreas. To test this hypothesis, we examined IR expression on T cells from the peripheral blood of healthy controls, individuals with T1D, and relatives at high-risk for T1D development indicated by the presence of disease-associated autoantibodies (32–34). In addition, to better understand potential mechanisms of action, we tested whether T cells engineered to express exogenous FLAG-tagged mouse IR (mIR) could migrate into the pancreas in the C57BL/6 mouse strain, which is not susceptible to diabetes.

Materials and Methods

Study Sites, Groups, and Autoantibody Analysis

There were two study sites for human subject analysis, the University of Toledo and the University of Michigan with both having Institutional Review Board (IRB) approval. Demographic and clinical information are provided for both sites in Table I.

Table I.

Groups, Age, and % of IR⁺ T cells.

Controls			Type 1 Diabetes				High Risk Relatives
ID#	age	% IR+ T	ID#	age	duration	% IR+ T	ID#	age	% IR+ T
UT CNT1	26	2.40	UT TID 1	10	Day 1	0.50	UT HRR 1	16	4.50
UT CNT2	42	1.80	UT T1D 2	11	Day 1	2.50	UT HRR 2	14	8.20
UT CNT3	36	0.70	UT T1D 3	5	Day 1	0.10	UT HRR 3	40	1.50
UT CNT4	40	0.40	UT T1D 4	12	Day 1	0.10	UT HRR 4	40	3.00
UT CNT5	32	1.80	UT T1D 5	10	Day 1	2.40	UT HRR 5	39	5.40
UT CNT6	48	0.80	UT T1D 6	7	10 days	1.8	UT HRR 6	13	1.80
UT CNT7	33	1.00	UT T1D 7	7	13 days	3.0	UT HRR 7	13	1.90
UT CNT 8	10	2.10	UT T1D 8	12	Established	0.60	UT HRR 8	54	6.10
UT CNT 9	38	0.10	UT T1D 9	11	4 years	3.9	UT HRR 9	28	1.40
UT CNT 10	36	13.2	UM T1D 11	13	11 years	0.31	UT HRR 10	58	2.40
UT CNT 11	46	0.30	UM T1D 13	10	8 years	0.05	UT HRR 11	44	1.40
UM CNT 39	14	0.14	UM T1D 14	17	6 years	0.26	UT HRR 12	12	2.00
UM CNT40	14	0.12	UM T1D 15	14	14 years	0.14	UT HRR 13	31	2.60
UM CNT41	12	0.68	UM T1D 16	15	7 years	0.14	UT HRR 14	19	27
UM CNT42	8	0.12	UM T1D 17	14	7 years	0.14
UM CNT43	15	0.03
UM CNT44	12	5.89
UM CNT46	16	4.85
UM CNT47	13	0.13
UM CNT48	10	0.21
UM CNT49	14	0.11
UM CNT50	10	0.20
UM CNT51	16	0.04
UM CNT52	17	0.03
UM CNT53	14	0.17
UM CNT54	9	0.14
Mean	22.35	1.44	Mean	11.20		1.06	Mean	30.07	4.94
SEM	2.54	0.54	SEM	0.81		0.32	SEM	4.11	1.71
range	8–48	0.03–13.2	range	5–17	Day 1 −14 yr	0.05–3.9	Range	12–58	1.4–27

At the University of Toledo site there were three groups of individuals analyzed: 1) individuals with T1D 2) relatives of individuals with T1D who had tested positive for one or more autoantibodies and 3) normal controls without diabetes. Autoantibodies to islet cytoplasmic antigens (ICA), insulin, glutamic acid decarboxylase (GAD), tyrosine phosphatase-related islet antigen 2 (IA-2), and zinc transporter 8 (ZnT8) were measured by radioimmunoassay as previously described (32) and subjects positive for at least one autoantibody were defined as having high-risk for developing T1D. Approximately 7 mL of peripheral blood, collected in yellow top vacutainers containing acid citrate dextrose (ACD) buffer (Becton Dickson Vacutainer system, Franklin Lakes, NJ) were received from the Cleveland Clinic Foundation, the University of Florida Diabetes Institute, the University of Toledo, and the Indiana University School of Medicine. IRBs were in place at each institution and prior to sample collection, each subject provided written informed consent with assent also provided by participating children. Peripheral blood mononuclear cells (PBMC) were isolated by density gradient centrifugation using Ficoll-Paque plus (Amersham Pharmacia Biotech AB, Piscataway, NJ) and washed. Mouse monoclonal antibody (mAb) to alpha chain of the human IR (hIR; MAB1137 clone (47–9) Millipore Sigma St Louis, MO) which does not crossreact with rodent IR or human insulin-like growth factor 1 receptor (IGF1R) (35), was titrated for staining the hIR by comparing the NIH3T3 cells (IR⁻ fibroblast cell line) against HIR3.5 cells (NIH3T3 fibroblasts transfected to express the α and β chains of hIR; estimated expression of 10⁵ hIR molecules per cell (36)). The secondary antibody utilized was donkey anti-mouse IgG (H+L) conjugated with PE (Jackson ImmunoResearch Westgrove, PA). Cells were also stained with anti-CD3-FITC (Coulter Immunology, Brea, CA) and collected on a Coulter Epics Elite flow cytometer (Hialeah, FL), which was calibrated before each use with fluorescent beads. T cells were identified by gating for lymphocytes using forward and side scatter and then gating on CD3⁺ T cells using FlowJo software.

The University of Michigan site examined two groups: pediatric controls and pediatric individuals with T1D. Freshly isolated PBMCs were stained with mAb reactive against human cell surface markers: anti-CD3-PE/Cy7 (Clone: SK7; BD Biosciences, San Jose, CA), and anti-human CD220 (hIR)-BV421, an anti-IR alpha chain antibody, (clone: 3B6/IR Fisher Scientific, Hanover Park, IL). Staining was detected using a BD Biosciences LSRII flow cytometer, which was calibrated weekly with fluorescent beads, in the Vision Core of the Kellogg Eye Center at the University of Michigan. Flow cytometric data were analyzed using FlowJo version X.0.6 software (TreeStar, Ashland, OR). Dead cells were identified by staining with 7AAD and excluded from further analysis. Lymphocyte (small, non-granular cells) gating based on forward scatter versus side scatter and singlet gating using forward scatter height versus forward scatter area were performed on all samples prior to analysis of cell surface markers.

Plasmid validation

Recombinant DNA research was approved by the Institutional Biosafety Committee at the University of Toledo (Toledo, OH). Insertion of the PPT-3xFLAG-mIR gene (approximately 4.1kb) from pCMV9–3xFLAG-mIR into pNeZB containing a CD3δ promoter and enhancer (a gift from Dr. Nancy Lee, Mayo Clinic, MN) (37) formed the pNeZBCD3–3xFLAG-mIR plasmid used previously in Friend Virus B (FVB) mice (38). CD3⁺ T cell restricted expression was validated by transfection into CD3⁻ human embryonic kidney cell line (HEK 293) (ATCC, Manassas, VA) and a CD3⁺CD8⁺ 4A7.7.15 hybridoma T cell clone (39).

Generation of C57BL/6-CD3FLAGmIR/mfm mice

The non-autoimmune C57BL/6 mouse strain was used for these experiments because they share a MHC Class I molecule D^b with NOD mice that carry the K^dD^b genotype. The C57BL/6-CD3FLAGmIR/mfm (BL/6-CD3FLAGmIR) transgenic mice were made by pronuclear injection at the Ohio State University (Genetically Engineered Mouse Modeling Core; Columbus, OH) and transplantation into pseudopregnant C57/BL6 NT (BL/6) females using DNA from the pNeZBCD3–3xFLAG-mIR plasmid after deletion of all bacterial genes as previously done on the FVB background (38). Founder mice were identified by reverse transcriptase PCR (RT-PCR) using the FLAGmIR and murine β-globin primers shown in Table 1. The specificity of the PCR product was confirmed by gel electrophoresis. Mice were categorized as transgene positive (Tg^+/−) or transgene negative (Tg^−/−) and bred to establish a transgene homozygous line (Tg^+/+). The results presented in this manuscript are based on data obtained from homozygous mice. Mice were housed at the University of Toledo (Toledo, OH), under specific-pathogen-free conditions, in the Department of Laboratory Animal Resources. All animal experiments were performed according to approved guidelines of the National Institutes of Health and protocols approved by the Institutional Animal Care and Use Committee (#N105461).

Sex as a Biological Variable

The prevalence of T1D is essentially equal in human males and females (40). Therefore, all individuals on a random basis, who signed the consent form or gave an assent, if children, provided the blood samples. Each group in the clinical study contained both male and female participants.

Due to the delay of FLAG⁺ cells in the pancreas in the transgenic female mice as compared to males, the results reported herein were obtained from male BL/6-CD3FLAGmIR mice and age-matched BL/6 male controls.

Immunofluorescence Staining (IF)

Spleen, pancreas, and thymus tissue was harvested from 10–50 wk old BL/6 and BL/6-CD3FLAGmIR mice in 5 wk increments and fixed in Z-fix (Anatech Ltd, Battle Creek, MI) or 10% formalin w/v (Fisher Scientific, Fair Lawn, NJ) at 4°C overnight. Liver, lung, heart, and kidney tissues were isolated at 15, 25 and 35 weeks of age and prepared as immediately above. Slides were de-paraffinized and rehydrated before antigen retrieval, by microwaving in 10mM citrate buffer pH 6.0. Sections were stained overnight at 4°C with anti-FLAG FITC (1:100 Sigma, St Louis, MO), anti-CD3 APC (1:100, eBioscience, San Diego, CA), anti-insulin produced in guinea pig (1:1500, Sigma, St Louis, MO), washed and stained with secondary anti-guinea pig IgG Texas Red (1:500, Invitrogen) for 1–2 hours at room temperature. After washing, sections were mounted with DAPI-Fluoromount G (Sigma, St Louis, MO) following dehydration and imaged on a Nikon TS fluorescent microscope.

Histological Microscopy

Pancreas, liver, lung, heart, and kidney tissue were harvested from 15, 25, and 35 week old male BL/6 and BL/6-CD3FLAGmIR mice and fixed in formalin at 4°C overnight. Sections were labelled for Hematoxylin and Eosin (H&E) after de-paraffinization and rehydration of tissue slides. After washing, sections were mounted with Fluoromount G mounting medium (Sigma, St Louis, MO) and imaged on a Nikon TS fluorescent microscope on brightfield setting.

Immunohistochemistry (IHC)

Pancreas tissue was harvested from 15, 25 and 35 week old male BL/6 and BL/6-CD3FLAGmIR mice and fixed in formalin at 4°C overnight. Slides were prepared and processed as described above in IF staining. After antigen retrieval, tissue sections were blocked with 1%BSA in tris buffer solution (TBS) containing 10% FBS. Tissue sections were labelled with biotinylated anti-mouse CD3 (1:50, eBioscience, San Diego, CA) at 4°C overnight. Streptavidin-HRP (1:100, Pharmingen, San Jose, CA) labeling was performed to detect CD3⁺ labelled cells. DAB substrate (Vector Laboratories, Burlingame, CA) was used to visualize the protein-antibody complex. Images were captured using Olympus VS120 slide scanner (Olympus, Center Valley, PA). ImageJ software was used to determine the number of CD3⁺ T cells per islet and the percentage of islets with insulitis, which was defined as the presence of six or more CD3⁺ T cells per islet (41).

mRNA Analysis by Quantitative Real Time PCR (qRT-PCR)

Islets were obtained from 15, 25 and 35 week old male BL/6 and BL/6-CD3FLAGmIR mice by digesting the pancreas with 1mg/mL Collagenase P (Sigma, St Louis, MO) (42). Islets were handpicked under the microscope and incubated overnight at 37°C in RPMI 1640 medium (Hyclone, Logan, UT) supplemented with Glutamine (Hyclone, South Logan, UT), penicillin-streptomycin (Mediatech Inc, Manassas, VA), 10% fetal bovine serum (FBS) (Atlanta Biologicals, Flowery Branch, GA), Sodium pyruvate (Life Technology, Grand Island, NY), non-essential amino acids (Mediatech Inc, Herndon, VA), 2-mercaptoethanol (Life Technology, Grand Island, NY). The complete RPMI 1640 medium also contained 5mM D-glucose for the islets. BL/6-CD3FLAGmIR and BL/6 mouse islets were lysed and total mRNA purified using RNeasy mini kits (Qiagen Hilden, Germany). cDNA was synthesized using M-MLV Reverse Transcriptase (ThermoFisher Scientific, Fair Lawn, NJ) and quantitated via the BioSpec-nano (Shimadzu Biotech, Columbia, MD). qRT-PCR was performed using SYBR green master mix (Bio Rad, Hercules, CA) and a CFX96 system Thermocycler (Bio-Rad, Hercules, CA) to identify the presence of CD3, CD4, CD8, or 3xFLAGmIR. Mouse Glyceraldehyde-3 phosphate dehydrogenase (GAPDH) was used to normalize the relative amount of mRNA using the delta Ct method for quantitative analysis (43). Primers for qRT-PCR are reported in Table 2.

Table II.

Primers used for qRT-PCR.

Forward FLAGmIR	5’CCATGTCTGCACTTCTGATCCTAGCTCTT 3’
Reverse FLAGmIR	5’CCATAGACACGGAAGAGAAGCAGGTAATC 3’
Forward mBeta-Globin	5’ CCAATCTGCTCACACAGGATA 3’
Reverse mBeta-Globin	5’ CCTTGAGGCTGTCCAAGTGAT 3’
Forward mGAPDH	5’AAGGTCATCCCAGAGCTGAA 3’
Reverse mGAPDH	5’ ATGTAGGCCATGAGGTCCAC 3’
Forward mCD3	5’ AGAGCAGCTGGCAAAGGTGGTGTC 3’
Reverse mCD3	5’ CAGCCATGGTGCCCGAGTCTAGC 3’
Forward mCD4	5’ GAGAGTCAGCGGAGTTCTC 3’
Reverse mCD4	5’ CTCACAGGTCAAAGTATTGTTG 3’
Forward mCD8	5’ CAGAGACCAGAAGATTGTCG 3’
Reverse mCD8	5’ TGATCAAGGACAGCAGAAGG 3’

Chemotaxis Studies

Spleens from BL/6 and BL/6-CD3FLAGmIR 25 wk old mice were harvested and made into single cell suspensions. Splenocyte Fc receptors were blocked with TruStain anti-CD32 and CD12 (1:200, Biolegend, San Diego, CA) in 1% BSA in PBS for 30 minutes on ice. Cells were centrifuged, re-suspended, and stained with anti-CD3 APC antibody (1:100, eBiosciences, San Diego, CA). CD3⁺ T cells and non-T lymphocytes, consisting of B cells and NK cells, were sorted using a FACS Aria (BD Bioscience, San Diego, CA) (44). Five thousand CD3⁺ T cells or non⁻T lymphocytes were transferred to the top chamber of a 96-well transmembrane chemotaxis plate (Essen Bioscience, Ann Arbor, MI) coated with 10μg/mL fibronectin (Sigma St Louis, MO) in RPMI 1640 complete media as described above, but containing only 0.05% FBS. No insulin, 50nM insulin (Sigma, St Louis, MO) or 10 islets (isolated from 10wk old male BL/6 and BL/6-CD3FLAGmIR mice) (42) were added to the bottom chambers. In the case of the islets, 5 mM D-glucose was added to RPMI 1640 complete media containing 0.05% FBS. 50nM insulin is close to physiological concentration since the intrapancreatic vasculature concentration of insulin is 10⁻¹⁰ to 10⁻⁸ mol/L (45). Additionally, based upon a dose response curve of 25nM, 50nM, 100nM, 200nM and 400nM of insulin, 50nM of insulin provided the best chemotaxis of transgenic macrophage depleted spleen cells expressing IR. The plate was placed into the incubator, and frames were captured using the IncuCyte HD system (Essen Bioscience, Ann Arbor, MI) at 2 hour intervals from 16 separate 950 × 760μm regions per top and bottom of each well using a 10× objective. Cultures were maintained at 37°C in an incubation chamber throughout the 72 hours, and run in quadruplicate. The IncuCyte real-time imaging system was used to measure chemotaxis using a non-labeled cell phase object count normalized to the top initial value.

Passive Insulin Release Assay

For the insulin secretion assay, islets were isolated from 10wk old male BL/6 and BL/6-CD3FLAGmIR mice, as described (42). After handpicking under the microscope, islets were cultured in RPMI 1640 complete media with 0.05% FBS and 5mM D-glucose, followed by overnight incubation at 37°C. The following day 10 size matched islets from each group were picked into 48 well tissue culture plates in 200μl of fresh islet complete media. The islets were cultured for a total of 72 hours with 25μl of media removed every 24 hours and frozen. Insulin levels in the conditioned media were then determined using ultra-sensitive mouse insulin ELISA (Alpco, Salem, NH, USA).

Blood Glucose Monitoring for Fasting and Non-fasting Populations

Murine blood glucose levels were monitored in both fasting and non-fasting populations of male, wild type and transgene positive mice and organized into age groups 10–15 weeks, 20–25 weeks, 30–35 weeks, and 40–45 weeks. For the fasting population, food was removed after 5pm and the animals were tested for blood glucose levels using an Alpha Track rodent glucometer (Abbott Laboratories, Chicago, IL) by 9 am the following day. Non-fasting random blood glucose readings were taken by glucometer during the day. The overall average blood glucose readings by age group were compared statistically to determine any significant effects of the transgene on blood glucose levels.

Statistics

Clinical experimental data, are presented as mean ± SEM. A limited number of outliers existed in both the control group and the high-risk relative group. Thus, results were analyzed using nonparametric methods that are more robust to extreme values. Wilcoxon rank sum test was employed for two-group comparisons. Kruskal-Wallis (KW) rank sum test, followed by pairwise Wilcoxon rank sum tests with Bonferroni correction was performed for comparisons across three groups. Assuming an effect size of 0.55 for one-way ANOVA analysis, the sample size needed for each group was 12 to achieve 80% power at type I error rate 0.05. According to Andrews (46), the asymptotic relative efficiency (ARE) of one-way ANOVA relative to KW is 0.955, therefore, the sample size for each group needed to achieve the same power for KW was 13.

For the animal experimentation, all data are presented as mean ± SEM and were analyzed using the unpaired student’s t-test, one-way ANOVA followed by Tukey’s multiple comparison test, or two-way ANOVA followed by Tukey’s multiple comparison test depending on the number of variables.

For both human and murine studies, P ≤ 0.05 was considered significant. All tests were performed using GraphPad Prism 5.04 software.

Results

IR expression on T cells in human samples

IR expression on human T cells was studied in PBMC obtained from three groups of individuals 1) healthy controls (n=26), 2) individuals with T1D (n=15), and 3) autoantibody positive relatives of individuals with T1D who have high-risk of developing T1D, (n=14) with data combined from two studies conducted at the University of Toledo and University of Michigan (see Table 1). Representative flow cytometry histograms and gated quadrants are shown for a control, a high-risk relative with two autoantibodies, a new-onset with T1D (day 1), and an individual that with T1D for 6 years (Supplemental Figure 1A–D). IR expression on human T cells differed significantly across the three groups (overall p-value p<0.001, KW test). Pairwise comparisons demonstrated that, autoantibody positive relatives had significantly higher numbers of circulating IR⁺ T cells compared to individuals with T1D (p<0.01) or the control group (p<0.001). There was no difference in the number of IR⁺ T cells in controls compared to individuals with T1D (p=1.0), (Figure 1A). When high risk relatives with multiple autoantibodies (red dots for 2 autoantibodies, and a blue dot for an individual with 5 autoantibodies, Figure 1A) were statistically analyzed compared to those with one antibody (black dots) and there was no significance difference (p=.87)

Figure 1: Insulin receptor expression on the T cell surface in human peripheral blood.

A. Percentage of Insulin receptor positive (IR⁺) CD3⁺ T cells in the human peripheral blood in controls (open circles, n=26), individuals with T1D (gray circles, n=15), and high risk relatives expressing autoantibodies (closed circles, red closed circles denote those individuals with 2 confirmed autoantibodies and the blue closed circles denotes an individual with 5 autoantibodies, n=14). B. Percentage of IR⁺CD3⁺ T cells in human peripheral blood, a comparison between individuals with new onset T1D (less than two weeks, closed circles, n=7) and individuals with established T1D (closed squares, n=8).

The T1D group consisted of seven new-onset individuals (duration1–13 days) (Table 1) and eight with established T1D (duration 4–14 years) (Table 1). There was no significant difference in peripheral blood levels of IR+ T cells between new-onset T1D vs established T1D (p-value =.38) (Figure 1B).

C57BL/6-CD3FLAGmIR/mfm (BL/6-CD3FLAGmIR) mouse model of IR overexpression on T cells

To understand the capacity of mIR overexpressing T cells to migrate towards pancreatic islets in vivo, the C57BL/6-CD3FLAGmIR/mfm (BL/6-CD3FLAGmIR) mouse model was generated using the pNeZBCD3–3xFLAG-mIR plasmid, as previously described (38). The mIR gene was tagged on the N terminal of the alpha chain gene with 3XFLAG to distinguish it from the endogenous gene and protein, and the mIR α and β chain genes were engineered behind the CD3 promoter and enhancer genes (38). To determine if the pNeZBCD3–3xFLAG-mIR plasmid was restricted by the CD3 promoter and enhancer genes, the plasmid was transfected into a CD3⁺ and a non T cell line. An empty plasmid pcDNA3.1, pNeZBCD3mIR-3XFLAG (with CD3 restriction), or pCMV9mIR-3XFLAG (mIR genes without CD3 restriction) were transfected into HEK293 cells, which do not express CD3. When the HEK293 cells were labeled with an antibody against the FLAG tag, only the positive control transfected cells expressing the unrestricted pCMVmIR-3XFLAG showed labeling (Figure 2A). However, in the CD3⁺CD8⁺ T cell hybridoma clone 4A7.7.15 (39), when the CD3 restricted pNeZBCD3mIR-3XFLAG was transfected, the FLAG tagged mIR was expressed on the cell surface (Figure 2B). Therefore, the CD3 promoter and enhancer were capable of facilitating cell surface expression of exogenous 3xFLAGmIR limited to T cells. The founders were screened using a primer including the 3XFLAG and the beginning of the mIR transgene (Figure 2C and Table 2).

Figure 2. Transgenic BL/6-CD3FLAGmIR Murine Model Generation.

A. HEK293 cells transfected with empty vector pcDNA3.1 (black line), pNeZBCD3–3xFLAG-mIR vector (dotted grey line), and pCMV9–3xFLAG-mIR vector (solid grey line). Each histogram represents HEK293 cells 24 hours post transfection, stained with Biotinylated anti-Flag monoclonal antibody as primary and Streptavidin-PE as secondary. B. 4A7.7.15 T cell hybridoma cells transfected with pNeZBCD3–3xFLAG-mIR vector to test CD3 regulated expression of mouse insulin receptor. The black filled histogram represents anti-FLAG FITC antibody stained mock transfected cells and the open histogram represents vector transfected cells stained with anti-FLAG FITC antibody. C. RT-PCR genotyping gel illustrating PCR product results from three transgene negative and three transgene positive mice showing the presence of beta globin in all samples, but the presence of the FLAGmIR transgene only in the transgene positive animals. D. IF staining of spleen tissue from 15 wk old mice for CD3⁺ T cells (magenta) and FLAGmIR (green) transgene expression with FLAG tagged mIR expression only in the transgenic mice. Scale bar: 300μM. E. IF staining of thymus tissue from 15 wk old mice for CD3⁺ T cells (magenta) and FLAGmIR (green) transgene expression. Double positive cells only appear in the transgenic mice. All IF staining is representative of at least three mice from that age group. Scale bar: 300μM.

The CD3 promoter and enhancer driven FLAG-tagged mIR expression should be detected in organs containing T cells in the BL/6-CD3FLAGmIR transgenic mice but not in control BL/6 mice. Indeed, FLAG-tagged mIR⁺ T cells were observed in stained spleen and thymus sections of BL/6-CD3FLAGmIR mice, an example of which is shown at 15 weeks of age, whereas splenic and thymus sections from control BL/6 mice of a similar age were negative (Figure 2D and 2E). The overlay in the BL/6-CD3FLAGmIR mice confirms that FLAGmIR staining co-localized with CD3 staining indicating that CD3⁺ FLAG tagged mIR⁺ T cells are in the spleen and thymus (Figures 2D and 2E). Neither CD3⁺mIR⁺ T cells nor morphological changes were observed in lung, heart, kidney, or liver tissue of BL/6-CD3FLAGmIR or BL/6 mice (Figure 3).

Figure 3. Morphology study of organs in male BL/6-CD3FLAGmIR and BL/6 mice by H&E and immunofluorescence (IF) staining at 25 wk of age.

Sections of liver, lung, heart and kidney tissue stained with H&E, anti-FLAG (green), anti-CD3 (red), and overlay with nuclei staining (blue). n=3 animals per group. Scale bar: 300μM.

IR⁺ T cells migrate towards the source of insulin

It is established that T cells with increased IR on their cell surface traffic towards insulin (29). To validate migration in vitro, CD3⁺ T cells from BL/6 and BL/6-CD3FLAGmIR splenocytes were sorted from non-T lymphocytes, consisting of B lymphocytes and NK cells, and placed in a chemotaxis chamber (44). Transgenic CD3⁺ T cells migrated to insulin at a significantly increased rate compared to transgenic non-T lymphocytes, but when no insulin was present the transgenic CD3⁺ T cells did not move (Figure 4A). CD3⁺ T cells and non-T lymphocytes cells from control BL/6 mice did not migrate towards insulin (Figure 4B). When transgenic CD3⁺ T cells were compared with wild type BL/6 CD3⁺ T cells, the transgenic T cells migrated significantly faster toward the 50nM insulin (Figure 4C). Due to the ability of the transgenic T cells to migrate towards insulin, we hypothesized that transgenic CD3⁺ T cells would similarly move toward islets in vitro. To examine this, flow cytometry purified CD3⁺ T cells from BL/6 and BL/6-CD3FLAGmIR were placed in a chemotaxis chamber, and pancreatic islets were used as a source of insulin in the lower chamber. Only CD3⁺ T cells from BL/6-CD3FLAGmIR mice migrated towards the islets, while CD3⁺ T cells from BL/6 mice did not (Figure 4D). In addition, we confirmed that insulin secretion from islets of 10 week old wild type BL/6 mice and BL/6-CD3FLAGmIR transgenic mice was similar at 24, 48, and 72 hours (Figure 5). Overall, this indicates that the transgenic CD3⁺IR⁺ T cells have an affinity to move towards sources of insulin in vitro.

Figure 4. Chemotaxis of IR⁺ T cells towards the source of insulin.

A. Chemotaxis of CD3⁺ T cells and non-T lymphocytes cells of 25 wk old BL/6-CD3FLAGmIR transgenic mouse from top chamber towards the insulin (50nM) in the bottom chamber. B. Chemotaxis of CD3⁺ T cells and non-T lymphocytes cells of 25 wk old BL/6 wild type mouse from top chamber towards the insulin (50nM) in the bottom chamber. C. Chemotaxis of CD3⁺ T cells of BL/6 and BL/6-CD3FLAGmIR mice from top chamber towards the insulin (50nM) in the bottom chamber. D. Chemotaxis of CD3⁺ T cells of BL/6 and BL/6-CD3FLAGmIR mice from top chamber towards the pancreatic islets in the bottom chamber. Graph represents number of non labelled cells in the top chamber normalized to the top initial value. A representative study from n=3 experiments. Two way ANOVA: *P≤0.05.

Figure 5. Passive Insulin Secretion.

Insulin secretion over time from 10 islets of wild type and transgenic mice. 10 islets were isolated from the 10 week old wild type and transgenic mice. This shows the amount of insulin secreted from 10 islets over 72 hours in triplicate wells. This shows no significant difference between the insulin secreted from islets obtained from wild type mice or the transgenic mice at any time point. Each dot represents an individual mouse.

Insulitis in Transgenic BL/6-CD3FLAGmIR mice by IR⁺ T cells

We next sought to examine the potential ability of CD3⁺ T cells in transgenic BL/6-CD3FLAGmIR mice to move into the islets in vivo. H&E staining of 15 weeks old male BL/6-CD3FLAGmIR pancreas indicated mild insulitis in the islet, while the BL/6 mice islets were clear (Figure 6A, see arrows for insulitis areas). In addition, FLAG-tagged mIR⁺CD3⁺ T cells were observed in the pancreas of BL/6-CD3FLAGmIR mice at 15weeks of age by IF and IHC staining, whereas pancreatic tissue sections from BL/6 mice were negative at 15 weeks (Figure 6B and 6C).Co-staining for insulin confirmed the presence of FLAG-tagged mIR⁺ T cells within the pancreatic islets (Figure 6D, overlay). There were, on average, 5–7 T cells per islet in the transgenic mice at 15 weeks of age. That number of FLAG-tagged mIR⁺ T cells within pancreatic islets increased with murine age and was significantly different from the control BL/6 mice (Figure 7A). In addition, the percentage of islets with insulitis was significantly greater in transgenic animals than in wild type BL/6 mice (Figure 7B). Although significant, the observed insulitis did not significantly alter blood glucose levels between transgenic and wildtype mice (5–45 weeks of age) upon fasting and non-fasting blood glucose monitoring, suggesting that the observed insulitis was benign (Figure 8). BL/6-CD3FLAGmIR murine islets also showed a significant increase in CD3, FLAG, and CD8 mRNA levels compared to BL/6 islets at 25 and 35 weeks of age, but there were no changes in CD4 mRNA expression at any age tested (Figure 7C–F).

Figure 6. FLAGmIR expressing T cell infiltration into the pancreatic tissue of BL/6 and BL/6-CD3FLAGmIR mice.

A. H&E staining of pancreatic tissue from 15 wk old mice illustrating the infiltration of cells in the islet of the transgenic mice. Pancreas of BL/6 (Scale bar: 200μm at 6.6x and 50μm at 15x) and BL/6-CD3FLAGmIR (Scale bar: 200μm at 6.6x) were stained with H&E. B. IF staining of pancreatic tissue from 15 wk old mice illustrating the presence CD3⁺ T cells (magenta) and FLAGmIR (green) transgene expression in the islet of the transgenic mice. Scale bar: 300μM. C. IHC labeling of pancreatic tissue sections from 15 wk old mice representing CD3⁺ T cell infiltration in the islet of the transgenic mice. Scale bar 50μM. D. IF staining of pancreatic tissue from 25 wk old mice demonstrating the presence of FLAG⁺ cells (green) within the insulin stained (red) β-cells in the transgenic mice. Scale bar: 50μM for BL/6 mice images and 100μM for BL/6-CD3FLAGmIR mice images.

Figure 7. Quantitative evaluation of T cells infiltrating into the transgenic mice islets.

A. Quantitative analysis of total number of CD3⁺ T cells per islet based on immunohistochemistry CD3 labeling of 15 wk (n=3), 25 wk (n=4) and 35 wk (n=5) old mice pancreatic tissues. −/− indicates BL/6 mice and +/+ indicates BL/6-CD3FLAGmIR mice. Unpaired student t test: ***P≤0.001. B. Percentage of islets with insulitis based on immunohistochemistry CD3 labeling of 15 wk (n=3), 25 wk (n=4) and 35 wk (n=5) old mice pancreatic tissues. Insulitis is defined as the presence of six or more CD3⁺ T cells per islet. −/− indicates BL/6 mice and +/+ indicates BL/6-CD3FLAGmIR mice. Unpaired student t test: *P≤0.05, ***P≤0.001. C, D, E, F. qRT-PCR analysis of BL/6 and BL/6-CD3FLAGmIR mice pancreatic islet mRNA expression of CD3, FLAGmIR, CD4 and CD8 normalized against the control gene (GAPDH) at 15 wk (n=3), 25 wk (n=3) and 35 wk (n=4) of age. −/− indicates BL/6 mice and +/+ indicates BL/6-CD3FLAGmIR mice. One way ANOVA: *P≤0.05, **P≤0.01

Figure 8. Fasting and non-fasting blood glucose levels.

Blood Glucose Measurement showing no hyperglycemia or significant difference between BL/6 (Black bars) and BL/6-CD3FLAGmIR (White bars) male mice from 5–50 weeks of age in fasting or non-fasting conditions (n=3 readings from 5 mice per group). One way ANOVA, *p≤0.05.

Discussion

Although multiple studies demonstrate that β cell death activates islet-resident antigen-presenting cells (APCs) in a pathogenic manner to cause inflammatory cytokine production and trafficking of β cell antigens to the pancreatic lymph nodes (PLN), there are minimal studies exploring the initial role of T cells in this process (47–49). In order for β cell destruction to occur, T cells must first traffic to islets from the blood stream, via PLN. Our studies explore the hypothesis that insulin produced by pancreatic β cells acts as a chemoattractant for T cells escaping thymic selection with increased IR on their cell surface. A recent publication from Tsai et al. (50) showed that under proliferating or activating conditions, the lack of IR reduced the inflammatory effects of both CD4⁺ and CD8⁺ T cells. They concluded that IR deficiency resulted in impaired T cell inflammatory cytokine production, effector differentiation, proliferation, and potentially migration/recruitment to target organs, including the pancreas (50). Upon T cell activation and recruitment to the islet, islet inflammation, and islet infiltration, additional cells such as, mononuclear phagocytic cells (MNPs) and B cells are recruited, (51). Once islets become inflamed and IFNγ is produced, changes in expression of proinflammatory cytokines, endothelial vascular adhesion molecules, and the chemokine gradient secreted by β cells, are required for further T cell recruitment via chemokine receptor signaling and trafficking (52, 53). Interestingly, Sandor et.al. (54) concluded that antigen presentation is not actually required for the entry of T cells into the inflamed islets but instead, required for the long-term retention and restimulation of pathogenic T cells within the islets. For progression to T1D from the insulitis stage in the NOD mouse, there is a requirement of “avidity maturation” of islet-specific cytotoxic CD8⁺ T cells (55). An acquired predominant trait, as autoantibody-positive at-risk first degree relatives advance toward T1D, is resistance to suppression of CD4⁺ T effector cells (56).

Our studies suggest that the high-risk relative group had significantly higher numbers of IR⁺ T cells compared to individuals with T1D and as compared to control subjects. Resting T cells do not display IR, but when activated with anti-CD3 and anti-CD28, IR is expressed on the T cell surface (22, 50, 57). Activated human T lymphocytes display chemotactic activity towards porcine insulin (29). Insulin is also used in human wound healing treatment because insulin induces chemotaxis of monocytes/macrophages in a dose-dependent and insulin receptor expression-dependent manner (58). Therefore, we hypothesized that the IR⁺ T cells in high-risk individuals are likely capable of chemotaxis toward insulin in vivo. IR⁺ T cells can traffic to the pancreas in the NOD model of T1D (31) as well as in transgenic FVB mice expressing FLAG tagged IR on T cells (38). If the IR⁺ T cells do migrate to the pancreas as a part of T1D pathogenesis, this may explain why circulating IR⁺ T cell frequencies are significantly lower in individuals with T1D versus autoantibody positive subjects.

There is a strong correlation between the number of autoantibodies and increased risk for T1D development (33, 59–61). Specifically, high risk relatives of individuals with T1D with two autoantibodies were demonstrated to have a 25% 5-year risk of developing T1D, while those with three autoantibodies had a risk of 40% 5-year risk, and those with four autoantibodies had a 50% 5-year risk in the diabetes prevention Trial-1 (DPT-1) (62). However, in our study there was no statistically significant difference between the percent of IR+T cells in autoantibody positive pre-diabetics when one or multiple autoantibodies were present. The reasons for this observation are unclear, but need to be subject to future investigations.

When comparing the numbers of IR+T cells between individuals with recent-onset T1D and individuals with established T1D, there was not a significant difference. Damond et al. (63) by analyzing human pancreatic organ donor tissue samples using imaging mass cytometry (IMC) found that recent-onset T1D patients retained a near-normal fraction of β-cells, in a large proportion of islets, whereas pancreata from individuals with longer duration of T1D were mostly devoid of β-cells. Therefore, pancreas histology studies have demonstrated that individuals with new onset T1D and individuals with established T1D are significantly different (63).

To confirm the hypothesis that T cells can migrate to the pancreas via the IR, we engineered a novel mouse model with FLAG tagged mIR expression restricted to the T cell surface under the control of the CD3 promoter and enhancer on the C57BL/6 non-autoimmune genetic background. In the BL/6-CD3FLAGmIR transgenic mice, IR⁺ T cells were found in the thymus and the T cell areas of the spleen, and FLAG staining co-localized with CD3 staining indicating that the cells were CD3⁺IR⁺ T cells. Purified CD3⁺ T cells from the transgenic BL/6-CD3FLAGmIR mice were capable of migration toward insulin in vitro, in agreement with previous studies with activated T cells and cells transfected with IR (29, 30). Furthermore, purified CD3⁺ T cells from the transgenic BL/6-CD3FLAGmIR mice were also capable of moving toward isolated islets, confirming a previous study with thioglycolate activated macrophages from the peritoneal cavity (64).

Transgenic BL/6-CD3FLAGmIR mice had evidence of insulitis in the pancreas by H&E, IF, and IHC staining by 15 weeks of age. Transgenic T cell movement was specific to the pancreas, since the heart, liver, lung and kidney were devoid of CD3⁺IR⁺ T cells in both transgenic and BL/6 mice. This study confirms and extends our previous efforts wherein insulitis was observed by H&E staining only, in 6 week old transgenic mice engineered expressing FLAG tagged IR⁺CD3⁺ T cells on a non-autoimmune FVB background (38).

According to the consensus paper by Campbell-Thompson, et al, the presence of six or more CD3⁺ T cells per islet and three insulitic islets per section is defined as insulitis in humans (41). As noted previously, while islets from diabetic NOD mice frequently have severe insulitis (65), in human studies of pancreata from organ donors with T1D, the islet lymphocytic infiltration is more subtle (65, 66). The BL/6-CD3FLAGmIR transgenic mice also had mild insulitis. Furthermore, patients with recent-onset and even long-standing T1D often have residual insulin containing islets or insulin positive single cells, although the insulin positive β-cell mass is markedly reduced (41). Similarly, the islets in the Bl/6-CD3FLAG mouse model still express insulin even though they are infiltrated (Figure 5D). CD8⁺ T cells were present in the islets of the transgenic mice, as determined by qRT-PCR analysis, whereas CD4⁺ T cells were not. CD8+ T cells are the known majority of cells present in both new-onset and long-term human T1D insulitis as well as in pancreas transplant rejection (67–70).

Although longitudinal studies of subjects progressing through the stages of pre-T1D (33, 59) have been explored, additional studies are needed to determine the temporal dynamics of IR expression on T cells in circulation and to evaluate IR expression on circulating T cells as a potential biomarker to track time to T1D development. Interestingly, while naive T cells specific for β cell antigens are commonly found in subjects with no signs of β cell autoimmunity (71), in patients with T1D, autoreactive T cells show signs of previous antigen encounter, such as telomere shortening, activation in the absence of co-stimulatory signals, and the expression of the memory marker CD45RO, highlighting the fact that T cell infiltration to β cells is not occurring in isolation to confer T1D progression (72, 73). Regardless of this complex progression of disease, including multiple chemokines and cytokines, such as IL-2, IL-6, IL-17, and IL-21 (56), the chemotaxis of IR⁺ T cells to the pancreatic islets is of key importance, and blocking it may prevent movement into the islet and further insulitis progression to overt T1D. To this end, we worked to develop a transgenic mouse that artificially and stably exhibits T cell-specific IR over-expression providing a platform that could be utilized for the development and study of therapeutics to control or limit T cell migration into the pancreas as a means to potentially prevent T1D. Additional studies determining the effect of T cell migration on the functionality of β-cell physiology using this transgenic animal model could further expand our understanding of human T1D.

Key Points.

High-risk relatives of T1D have significantly more IR⁺ T cells than T1D or controls.

A new transgenic mouse has insulitis in the pancreas via IR on T cells, but not T1D.