Critical Amino Acid Variants in HLA-DRB1 and -DQB1 Allotypes in the Development of Classical Type 1 Diabetes and Latent Autoimmune Diabetes in Adults in the Japanese Population

The effects of amino acid variants encoded by the human leukocyte antigen (HLA) class II on the development of classical type 1 diabetes (T1D) and latent autoimmune diabetes in adults (LADA) have not been fully elucidated. We retrospectively investigated the HLA-DRB1 and -DQB1 genes of 72 patients with classical T1D and 102 patients with LADA in the Japanese population and compared the frequencies of HLA-DRB1 and -DQB1 alleles between these patients and the Japanese populations previously reported by another institution. We also performed a blind association analysis with all amino acid positions in classical T1D and LADA, and compared the associations of HLA-DRB1 and -DQB1 amino acid positions in classical T1D and LADA. The frequency of DRß-Phe-13 was significantly higher and those of DRß-Arg-13 and DQß-Gly-70 were significantly lower in patients with classical T1D and LADA than in controls. The frequencies of DRß-His-13 and DQß-Glu-70 were significantly higher in classical T1D patients than in controls. The frequency of DRß-Ser-13 was significantly lower and that of DQß-Arg-70 was significantly higher in LADA patients than in controls. HLA-DRß1 position 13 and HLA-DQß1 position 70 could be critical amino acid positions in the development of classical T1D and LADA.


Study Participants
This is a retrospective study. We identified 185 unrelated Japanese patients with classical (n = 72) and slow-onset (n = 113) T1D who visited the internal medicine departments of participating hospitals between 2001 and 2010. The participating hospitals were Ichinomiya Municipal Hospital (Ichinomiya, Japan), Kyoritsu General Hospital (Nagoya, Japan), and Okazaki City Hospital (Okazaki, Japan), all in the Aichi Prefecture. All patients fulfilled the World Health Organization criteria for diabetes [10]; we have reported some of their clinical and immunogenetic characteristics [2,11]. We excluded 11 patients with slow-onset T1D whose onset of diabetes occurred at age <30 years. Thus, 72 unrelated patients with classical T1D and 102 unrelated patients with slow-onset T1D comprised the final classical T1D and LADA cohort.
These frequencies of HLA-DRB1 and -DQB1 alleles were compared with those of other Japanese populations, as reported by the Japanese Society of Histocompatibility and Immunogenetics [2,9]. This study was approved by the Ethics Committee of Aichi Prefectural University (approval code: 1-43, 26 February 2021) and was conducted in accordance with the Declaration of Helsinki.

Measurements
The titers of glutamic acid decarboxylase (GAD) antibody (GAD-Ab) and insulinomaassociated antigen-2 antibody (IA-2Ab) were determined as described previously [12]. The cutoff values for GAD-Ab and IA-2Ab were 1.5 and 0.4 U/mL, respectively. Urinary and serum C-peptide levels were determined using a commercially available enzyme immunoassay kit (Eiken C-Peptide Kit; Eiken Chemical, Tokyo, Japan).
HLA-DRB1 sequence-based typing (SBT) was performed by directly sequencing DRB1 exon 2 using the AlleleSEQR DRB1 Typing Kit (Atria Genetics, San Francisco, CA, USA) according to the manufacturer's instructions, as described previously [2]. HLA-DQB1 SBT was carried out by direct sequencing of DQB1 exon 2 and 3 using the AlleleSEQR DQB1 Typing Kit (Atria Genetics) according to the manufacturer's instructions, as described previously [2]. Ambiguous genotyping samples were identified using high-resolution polymerase chain reaction-sequence-specific amplification and heterozygous ambiguity resolution primer's methods. The BIGDAWG software, version 2.5, implemented as the bigdawg R package (GitHub, San Francisco, CA, USA) [13], was used for the analysis of amino acids encoded by the HLA-DRB1 and -DQB1 alleles.

Statistical Analysis
Study results are presented as means ± standard deviation or percentages with numbers. SPSS 26.0 software (IBM Corp., Armonk, NY, USA) was used for statistical analyses. Chi-square tests based on 2 × 2 contingency tables and Fisher's exact probability tests were used to compare allele frequencies. p values were corrected for the number of different alleles tested using the Benjamini-Hochberg method (denoted as Pc) [14]. Pc < 0.05 was considered statistically significant.

Clinical Data
The median age of patients with classical T1D (29 men and 43 women) was 30 years (range 5-81) at the onset of diabetes. The duration of diabetes and body mass index (BMI) were 9.8 ± 9.9 years and 20.6 ± 3.3 kg/m 2 , respectively. The positive GAD-Ab and IA-2Ab rates were 60.9% and 42.9%, respectively; 21 classical T1D patients (29.2%) had no positive results for GAD-Ab or IA-2Ab. Urinary and serum 2-h C-peptide levels were 2.6 ± 3.5 nmol/day and 0.15 ± 0.24 nmol/L, respectively. All classical T1D patients were treated with insulin and insulin dosage was 0.69 ± 0.30 U/kg/day.
The median age of patients with LADA (55 men and 47 women) was 52 years (range 31-80) at the onset of diabetes. The duration of diabetes and BMI were 9.5 ± 8.7 years and 23.1 ± 4.3 kg/m 2 , respectively. The positive GAD-Ab rate was 100%. Urinary and serum 2-h C-peptide levels were 10.8 ± 12.8 nmol/day and 0.92 ± 0.96 nmol/L, respectively. The rate of LADA patients treated with insulin was 69.6% and insulin dosage was 0.50 ± 0.25 U/kg/day.
To the best of our knowledge, there is no report investigating the association of LADA with the amino acid variants in HLA-DRB1 and -DQB1 allotypes. The existence of DQα-Arg-52 and the absence of DQβ-Asp-57 did not confer susceptibility to LADA, as shown in classical T1D [18]. In the present study, DRß-Phe-13 encoded by the HLA-DRB1*09:01 allele and DRß-Arg-13 encoded by the HLA-DRB1*15:01 allele, respectively, confer susceptibility to and protection against LADA. Previous studies demonstrated that the HLA-DRB1*09 allele confers susceptibility to LADA in Asians [2,11,18,19], but not Caucasians [19,20]. The HLA-DRB1*15 allele confer protection against LADA both in Asians [2,11] and in Caucasians [20]. Although DRß-Ser-13 is not encoded by alleles, which confer susceptibility to and protection against LADA in the present study, it confers protection against LADA. DRß-Ser-13 is encoded by the HLA-DRB1*03:01 allele, which confers susceptibility to LADA in Caucasians [19,20] and in Asians other than the Japanese [18,19]. DRß-Ser-13 is also encoded by the HLA-DRB1*11:01 and -DRB1*14 alleles, which were found to have a protective trend against LADA in the present study and the previous study [19]. Desai et al. demonstrated that the HLA-DRB1*11:01 allele confers protection against LADA in Caucasians [20].

Conclusions
HLA-DRβ1 position 13 and HLA-DQß1 position 70 could be critical amino acid positions in the development of classical T1D and LADA. DRß-Phe-13 confers susceptibility to classical T1D and LADA, and DRß-Arg-13 and DQß-Gly-70 confer protection against the diseases. In addition, DRß-His-13 and DQß-Glu-70 confer susceptibility to classical T1D, and DRß-Ser-13 and DQß-Arg-70 confer protection against and susceptibility to LADA, respectively. Such novel alleles could guide autoantigen discovery and tolerance immunotherapies. Further studies are required to determine the underlying mechanisms of these differences. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.