Abstract/Flash Poster Session B

Background and Aims

Studies of pancreatic tissue gathered soon after onset of Type 1 diabetes have revealed the existence of two, age-related, immune phenotypes in inflamed islets (CD20Hi & CD20Lo). Since these also vary according to the proportion of residual insulin containing islets retained at diagnosis, we have now determined whether they display differential pancreatic proinsulin processing and secretion.
Materials and Methods 
The subcellular localisation of proinsulin and insulin was monitored in a blinded manner in pancreatic tissue from 21 patients diagnosed within 1y of T1D, at increasing ages (6 < 7y, 11 7-12y; 6 >13y) plus 8 controls. In parallel, 90 minute plasma C-peptide and proinsulin levels were measured after a mixed meal tolerance test in 191 T1D patients (87 diagnosed< 7y and 84>13y) studied >5y (median 13.3y) post diagnosis.
Results

Two distinct patterns of proinsulin localisation were observed in residual insulin-containing islets of children diagnosed < 7y and those >13y. Abnormal proinsulin/insulin colocalisation was found in almost all islets in the < 7y (CD20Hi) group (mean Manders overlap coefficient: 0.72±0.03 vs0.28±0.01 in controls;p< 0.0001), while most islets (78%) of those diagnosed at >13y (CD20Lo) did not display this phenomenon (overlap coefficient: 0.38±0.03;p< 0.0001). The distribution of proinsulin in islets of children diagnosed between 7-12y (3 CD20Lo; 5 CD20Hi) segregated precisely with their immune phenotype, where highest rates of colocalisation was restricted to the CD20Hi group.
Subjects with longstanding T1D (>5y) diagnosed < 7y had lower C-peptide levels than those diagnosed >13y (median(IQR)< 3(< 3-< 3)v34.5(< 3,151)pmol/l; p< 0.0001). In contrast, the proinsulin/C-peptide ratio (PI/CP) was increased markedly in the < 7y compared to >13y (0.18(0.10,0.31)v0.01(0.009,0.10)nmol/mmol;p< 0.0001).
Conclusion

Two distinct endotypes of T1D exist, which correlate with age at diagnosis. These differ according to their insulitic & proinsulin processing profiles, the proportion of residual insulin-containing islets and circulating C-peptide and proinsulin concentrations. These factors must be considered when designing immunotherapeutic strategies for T1D.

Background: The risk of Type 1 Diabetes (T1D) comprises genetic and environmental components, which both could be mediated by DNA methylation (DNAm). Previous studies established associations between DNAm and T1D GWAS variants, but it is unclear whether these associations indicate causation or were raised due to confounding factors or even reverse causation, i.e. whether causal variants first introduce gene expression changes which in turn cause DNAm changes and not the other way around. We systematically investigated the causal roles of DNAm in mediating common genetic risk of T1D in a large-scale population sampled longitudinally.

Methods: We first assessed whether DNAm and T1D share a common genetic influence at genome-wide scale by performing an enrichment analysis using summary statistics. Subsequently, at T1D loci we regressed 65 top GWAS nucleotide polymorphisms (SNPs) with Illumina 450k MethylationBeadChip data in the ARIES cohort where blood DNAm were measured at birth (n=844), childhood (n=846) and adolescence (n=907). Next, joint likelihood mapping was performed to identify loci where the causal variants for DNAm and T1D were shared. At these loci bi-directional two-sample Mendelian Randomization was used to assess the causal roles of DNAm. Finally, we investigated whether DNAm influence local gene expression using whole blood data from the GTEx consortium.

Results: There was a common genetic influence for both DNAm and T1D across the genome, implying that DNAm could be either on the causal pathway to T1D or a non-causal biomarker of T1D genetic risk. We found 159 proximal SNP-cytosine phosphate guanine (CpG) pairs (cis), and 7 distal SNP-CpG associations (trans) at birth, childhood, and adolescence, but DNAm did not play causal roles at most of these loci. At 5 loci, ITGB3BP, AFF3, PTPN2, CTSH and CTLA4, DNAm is potentially mediating the genetic risk of T1D mainly by influencing local gene expression.

Background: Insulin is a target of autoimmunity that destroys beta cells in type 1 diabetes (T1D). In the NOD mouse model of T1D, insulin delivered to the nasal mucosa induced regulatory T cells and decreased diabetes incidence. We reported two human trials demonstrating intranasal insulin was safe and elicited immune responses and tolerance to insulin. Here, we present intranasal insulin trial II (INIT II), which aimed to determine if intranasal insulin delayed or prevented the onset of T1D in individuals with sub-clinical disease.
 
Methods: Initially, >10,000 T1D relatives were screened; 2.2% were positive for autoantibodies ≥ 2 islet antigens (insulin, GAD65, IA-2). Following ‘staging’ to establish normal beta-cell function, 110 seropositive relatives (71 males, 39 females; median ages 11.4, 12.7 years) were randomized to 1.6mg and 16mg insulin dose and placebo arms. Their estimated 5-year risk of diabetes was 40%; the trial was powered to detect a decrease of 50%. Recruitment was slower than anticipated and the ‘low dose’ dose arm was ceased after it had enrolled 18 participants, but these continued to be followed. Treatment was self-administered as two 100μl spray doses per nostril, daily for one week then weekly for 1 year. Participants were followed 3-monthly in the first year, then 6-monthly for a further 4 years, or until diabetes.
 
Results: There were no serious treatment-related adverse events. All participants will complete 3-year follow-up by 01/10/18, at which time the median follow-up will be 57.5 months. Intranasal insulin led to a dose-related increase in serum insulin antibody concentrations, which peaked and then declined during ongoing treatment in the first year. This response is consistent with induction of immune tolerance but was not associated with change in rate of progression to diabetes. Results for the primary outcome, diabetes, will be presented.
 

Background
Multiple genetic factors associated with an elevated risk of type 1 diabetes (T1D) are thought to affect the development and maturation of immune cell subsets. T1D-associated polymorphisms may affect the function of the immune system by altering the frequency or phenotype of immune cell subsets. 
Methods
The subjects analysed in the study were participants in the Finnish DIPP (Diabetes Prediction and Prevention) study and the cohort comprised 86 subjects positive for at least one biochemical autoantibody, 75 subjects who had progressed to T1D and 231 autoantibody-negative control subjects. These children were screened for the presence of T1D-predisposing genotypes within the HLA region. Additionally seven T1D-associated single nucleotide polymorphisms in six non-HLA genes, INS (rs689), PTPN22 (rs2476601), IL2RA (rs12722495 and rs2104286), PTPN2 (rs45450798), CTLA4 (rs3087243) and ERBB3 (rs2292239) were determined by qPCR. Several immune cell subset frequencies were analysed in blood samples by flow cytometry. The effects of the genetic markers on immune cell subset frequencies were assessed using analysis of covariance with the age of the children as a covariant.
Results
The most significant finding was the association of the PTPN22 rs2476601 risk allele with elevated total TREG (CD4+CD25+CD127-FOXP3+) frequencies (p < 0.001) and in particular naïve TREG (CD4+CD25+CD127-CD45RO+FOXP3+) frequencies (p < 0.001). These results remained significant when the clinically healthy subjects (p = 0.001) and children with T1D (p = 0.009 and p = 0.025 respectively) were analysed separately. These results will be validated in a separate cohort of healthy children with different PTPN22 rs2476601 genotypes.
Conclusions
Our results support the hypothesis that T1D-predisposing genetic polymorphisms affect T cell development and/or differentiation during childhood. Further studies are needed to clarify the mechanistic basis of these associations and their relevance for the pathogenesis of T1D.

Background
Coxsackievirus B (CVB) viruses are associated with Type 1 diabetes (T1D), however, whether they are causal remains unknown. A strategy to determine their involvement in T1D could involve vaccination of at-risk individuals with a vaccine targeting the six known CVB serotypes (CVB1-6) and successive monitoring of disease incidence. Currently, no commercially available CVB vaccines exist, thereby rendering this option unobtainable. Previously, we demonstrated that a CVB1 vaccine is immunogenic and protects against acute CVB1 infections and CVB1-induced diabetes in mice. Cross-protection against other CVB serotypes does not occur, necessitating the need for a hexavalent CVB1-6 vaccine. Here, we preclinically test a hexavalent CVB vaccine in murine models.
Methods
CVB1-6 serotypes were inactivated in formalin to produce the hexavalent non-adjuvanted CVB vaccine. NOD and SOCS-1-tg NOD mice (a CVB-induced diabetes model), were vaccinated 3 times and then challenged with CVBs. Weight and blood glucose (used to monitor diabetes onset) were measured regularly. Neutralising antibodies in sera were measured by standard neutralization assay. Viraemia and viral dissemination after CVB infection were assessed by standard plaque assay and histological analysis.
Results
The CVB1-6 vaccine had no detrimental effects on weight or blood glucose. Neutralizing antibodies against all six CVB serotypes were induced with equivalent neutralising capacities after three vaccinations. The CVB1-6 vaccine protected against acute CVB infection and prevented CVB-induced diabetes in the SOCS-1-tg mouse model for virus-induced diabetes. As far as we are aware, this is the first time a vaccine targeting the six CVB serotypes has been produced and shown to be effective in preventing infection. This preclinical proof-of-concept study provides the basis for the further development of a hexavalent CVB vaccine for use in humans to determine whether CVBs are involved in T1D, and if they were found to be casual, could provide a preventative treatment for T1D.

Background Sphingolipids play important roles in beta cell physiology, by regulating proinsulin folding and insulin secretion and in controlling apoptosis, as studied in animal models and cell cultures. Here we investigate whether sphingolipid metabolism may contribute to the pathogenesis of human type 1 diabetes and whether increasing the levels of the sphingolipid sulfatide would prevent models of diabetes in NOD mice.

METHODS We examined the amount and distribution of sulfatide in human pancreatic islets by immunohistochemistry, immunofluorescence and electron microscopy. Transcriptional analysis was used to evaluate expression of sphingolipid-related genes in isolated human islets. Genome-wide association studies (GWAS) and a T cell proliferation assay were used to identify type 1 diabetes related polymorphisms and test how these affect cellular islet autoimmunity. Finally, we treated NOD mice with fenofibrate, a known activator of sulfatide biosynthesis, to evaluate the effect on experimental autoimmune diabetes development.

RESULTS We found reduced amounts of sulfatide, 23% of the levels in control participants, in pancreatic islets of individuals with newly diagnosed type 1 diabetes, which were associated with reduced expression of enzymes involved in sphingolipid metabolism. Next, we discovered eight gene polymorphisms (ORMDL3, SPHK2, B4GALNT1, SLC1A5, GALC, PPARD, PPARG, and B4GALT1) involved in sphingolipid metabolism that contribute to the genetic predisposition to type 1 diabetes. These gene polymorphisms correlated with the degree of cellular islet autoimmunity in a cohort of individuals with type 1 diabetes. Finally, using fenofibrate, which activates sulfatide biosynthesis, we completely prevented diabetes in NOD mice and even reversed the disease in half of otherwise diabetic animals.

Background: Defining clinical endpoints is challenging in type1 diabetes (T1D) trials. While stimulated C-peptide following mixed meal tolerance test (MMTT) is the gold standard, it is cost- and time- intensive to measure. Furthermore, it fails to capture the clinical benefit expected with beta cell preservation. Measures of glycemic control and insulin usage are clinically important surrogates, but are difficult to interpret individually. 
The BETA-2 score is a validated measure of islet transplant success that includes beta cell function and glycemic control. It is calculated from a single blood sample and includes fasting C-peptide, A1c, fasting blood glucose and insulin dose. The aim of this study was to assess the BETA-2 score as a potential T1D trial endpoint.
Methods: Data from the proinsulin peptide trial was retrospectively analyzed. The trial included newly diagnosed T1D subjects aged 18-35 years. Subjects were randomized to 3 groups: treatment with proinsulin peptide (low frequency every 4 weeks, n=10 or high frequency every 2 weeks, n=9) or placebo (n=8). The trial was designed to assess safety/tolerability rather than treatment differences, however, AUC C-peptide following MMTT was a secondary outcome. The BETA-2 score was calculated at baseline, 3, 6, 9 and 12 months where data was available.
Results: The BETA-2 score correlated with AUC C-peptide at all time points (r=0.721-0.910, P<0.001). Although there was no treatment effect on AUC C-peptide, BETA-2 score was different over time between groups (P=0.031). BETA-2 measured at 3 and 6 months correlated significantly with AUC C-peptide at 12 months (3 months: r=0.632, P=0.001; 6 months: r= 0.780, P<0.001).
Conclusion: The BETA-2 score appears to correlate well with AUC C-peptide, be responsive to treatment effect, and predictive of AUC C-peptide at 1 year. It may therefore be useful as a surrogate endpoint in T1D intervention trials allowing for smaller and shorter duration trials.

Background:Preclinical studies suggest that blockade of pro-inflammatory IL-17/IFN-γ secreting T cells inhibits the T1D autoinflammatory response. We assessed the safety and optimal dosing of ustekinumab (a monoclonal antibody targeting the IL-17/IFN-γ pathway) for the treatment of adult recent-onset T1D in a phase I/II open-label clinical trial (NCT02117765). 
Methods:We enrolled 20 patients within 100 days of T1D diagnosis, aged 18-35 years, and with a peak C-peptide of >0.2nmol/l on MMTT. Subjects were divided into four subcutaneous dosing cohorts: i)45mg-weeks 0/4/16, ii)45mg-weeks 0/4/16/28/40, iii)90mg-weeks 0/4/16 and iv)90mg-weeks 0/4/16/28/40. The primary endpoint was safety (rate, frequency and severity of adverse events). We also measured the baseline-adjusted change in 2-h AUC C-peptide response to MMTT, insulin use/kg and HbA1c at 1 year. We performed flow cytometric analyses before and after study drug administration to assess changes in T helper cell subsets. We also assessed the changes in Proinsulin and GAD65 antigen-specific T cell subsets by flourospot. 
Results:Ustekinumab-treated patients had 13 adverse events (1/13 was possibly attributed to study drug). At 1 year, the highest dosing cohort (correlating to highest drug levels) had the most subjects that met clinical responder status (HbA1c < 6.5% with insulin use/kg/day < of 0.5) and the smallest decline in C-peptide AUC (0.1pmol/mL). In addition, ustekinumab administration significantly reduced the percentage of Th17, Th1 and Th17.1 cells from baseline and this reduction is specific to Proinsulin antigen-specific cells.
Conclusion: Our study suggests that the highest ustekinumab dosing schedule will be safe and optimal to prevent decline in C-peptide AUC, induce clinical response and reduce proinsulin antigen specific Th17/Th1 subsets. This pilot has provided a rationale for several placebo-controlled phase II/III efficacy studies using highest dosing approved in children (USTEKID) and adults (UST1D2).

Background: We have identified a common defect in the Tregs of patients with certain autoimmune and allergic diseases: diminished inhibition of desensitization of their Treg IL-2 receptors (IL-2R) in response to low dose IL-2. Desensitization, or “turning off the IL-2R” occurs when the SOCS3 cullin ring ligase (CRL) ubiquitinates the IL-2R beta chain associated pJAK1. CRL ubiquitination of pJAK1 leads to its degradation, thus halting IL-2R signaling and pSTAT5 expression. The cullin5 component of the SOCS3 CRL must be post translationally modified (neddylated) at a specific lysine residue to activate its CRL function. The pSTAT5b isoform of STAT5 is required for the transcription of the Treg centric transcriptome including FoxP3 and CD25 (and other genes required for Treg function). The ubiquitin E3 ligase GRAIL is constitutively expressed in normal mouse and human Tregs and inhibits IL-2R desensitization by mono-ubiquitinating cullin5 at the exact lysine residue that must be neddylated to allow the CRL function. Thus, ubiquitination of cullin5 by GRAIL competes with neddylation and delays IL-2R desensitization. This prolongs expression of pSTAT5 in “normal” Tregs in response to low dose IL-2 activation.
Results: We have identified this same defect in inhibition of desensitization of the IL-2R on Tregs of human T1D patients, and have restored function of their Tregs in vitro by inhibition of IL-2R desensitization using a Nedd8-activating-enzyme (NAE) inhibitor to inhibit cullin5 neddylation. Using the same NAE inhibitor in combination with low dose IL-2, we were able to delay the progression to hyperglycemia more efficiently in 12-week old NOD mice that were treated for 21 days with the combination, than with low dose IL-2 as a single agent.