Polymorphisms in intron 1 of HLA-DRA differentially associate with type 1 diabetes and celiac disease and implicate involvement of complement system genes C4A and C4B

  1. Ozkan Aydemir  Is a corresponding author
  2. Jeffrey A Bailey
  3. Daniel Agardh
  4. Åke Lernmark
  5. Janelle A Noble
  6. Agnes Andersson Svärd
  7. Elizabeth P Blankenhorn
  8. Hemang M Parikh
  9. Anette-G Ziegler
  10. Jorma Toppari
  11. Beena Akolkar
  12. William A Hagopian
  13. Marian J Rewers
  14. John P Mordes
  15. TEDDY Study Group
  1. University of Massachusetts Chan Medical School, United States
  2. Department of Pathology and Laboratory Medicine and Center for Computational Molecular Biology, Brown University, United States
  3. Department of Clinical Sciences, Lund University CRC Skane University Hospital, Sweden
  4. University of California, San Francisco, United States
  5. Drexel University College of Medicine, United States
  6. Health Informatics Institute, Morsani College of Medicine, University of South Florida, United States
  7. Institute of Diabetes Research, Helmholtz Zentrum München, Klinikum rechts der Isar, Technische Universität München, and Forschergruppe Diabetes e.V., Germany
  8. Research Centre for Integrative Physiology and Pharmacology, Institute of Biomedicine, and Centre for Population Health Research, University of Turku, Finland
  9. Department of Pediatrics, Turku University Hospital, Finland
  10. National Institute of Diabetes and Digestive and Kidney Diseases, United States
  11. Pediatrics, Indiana University School of Medicine, United States
  12. Department of Medicine, University of Washington, United States
  13. Barbara Davis Center for Childhood Diabetes, University of Colorado, United States
3 figures, 10 tables and 2 additional files

Figures

Figure 1 with 12 supplements
Cox Proportional Hazard (PH) regression results for all tested outcomes.

Log hazard ratios (log(HR)) and 95% confidence intervals of the tri-single-nucleotide polymorphism (SNP) 101 haplotype as well as the known risk factors first-degree relative (FDR) and sex from the Cox PH models for outcomes type 1 diabetes (T1D) (A), islet antigen (IA) (B), insulin autoantibody (IAA)-first (C), glutamic acid decarboxylase autoantibody (GADA)-first (D), celiac disease (CD) (E), and celiac disease autoimmunity (CDA) (F), using the entire cohort (left) or only the DR3-DQ2 homozygote individuals (right). Dashed vertical line at 0 indicating an HR of 1 (log(HR)=0), i.e., no effect on risk. Left side of the vertical line indicates reduced risk vs increased risk on the right side. Whiskers indicate 95% CI around HR. The model assesses the independent risk/protection afforded by each covariate compared to the baseline for categorical covariates FDR and sex for which the baselines are having no FDR and female sex, respectively. Tri-SNP 101 is modeled numerically, so the HR reported is per each additional 101 allele.

Figure 1—figure supplement 1
Complete model output for type 1 diabetes (T1D) outcome.

HR values and 95% CI are plotted for each covariate included in the model. Vertical line marks the HR = 1 (no change in risk). Nobservations = 7638, Nevents = 395.

Figure 1—figure supplement 2
Complete model output for type 1 diabetes (T1D) outcome using only DR3-DQ2 homozygotes.

HR values and 95% CI are plotted for each covariate included in the model. Vertical line marks the HR = 1 (no change in risk). Nobservations = 1589, Nevents = 36.

Figure 1—figure supplement 3
Complete model output for islet antigen (IA) outcome using all samples.

HR values and 95% CI are plotted for each covariate included in the model. Vertical line marks the HR = 1 (no change in risk). Nobservations = 7614, Nevents = 855.

Figure 1—figure supplement 4
Complete model output for islet antigen (IA) outcome using only DR3-DQ2 homozygotes.

HR values and 95% CI are plotted for each covariate included in the model. Vertical line marks the HR = 1 (no change in risk). Nobservations = 1585, Nevents = 121.

Figure 1—figure supplement 5
Complete model output for glutamic acid decarboxylase autoantibody (GADA)-first outcome.

HR values and 95% CI are plotted for each covariate included in the model. Vertical line marks the HR = 1 (no change in risk). Nobservations = 7614, Nevents = 382.

Figure 1—figure supplement 6
Complete model output for glutamic acid decarboxylase autoantibody (GADA)-first outcome using only DR3-DQ2 homozygotes.

HR values and 95% CI are plotted for each covariate included in the model. Vertical line marks the HR = 1 (no change in risk). Nobservations = 1585, Nevents = 85.

Figure 1—figure supplement 7
Complete model output for insulin autoantibody (IAA)-first outcome.

HR values and 95% CI are plotted for each covariate included in the model. Vertical line marks the HR = 1 (no change in risk). Nobservations = 7614, Nevents = 313.

Figure 1—figure supplement 8
Complete model output for insulin autoantibody (IAA)-first outcome using only DR3-DQ2 homozygotes.

HR values and 95% CI are plotted for each covariate included in the model. Vertical line marks the HR = 1 (no change in risk). Nobservations = 1570, Nevents = 29.

Figure 1—figure supplement 9
Complete model output for celiac disease (CD) outcome.

log(HR) values and 95% CI are plotted for each covariate included in the model. Vertical line marks the log(HR)=0 (no change in risk). Nobservations = 6530, Nevents = 608.

Figure 1—figure supplement 10
Complete model output for celiac disease (CD) outcome using only DR3-DQ2 homozygotes.

log(HR) values and 95% CI are plotted for each covariate included in the model. Vertical line marks the log(HR)=0 (no change in risk). Nobservations = 1364, Nevents = 298.

Figure 1—figure supplement 11
Complete model output for celiac disease autoimmunity (CDA) outcome.

log(HR) values and 95% CI are plotted for each covariate included in the model. Vertical line marks the log(HR)=0 (no change in risk). Nobservations = 6557, Nevents = 1282.

Figure 1—figure supplement 12
Complete model output for celiac disease autoimmunity (CDA) outcome using only DR3-DQ2 homozygotes.

log(HR) values and 95% CI are plotted for each covariate included in the model. Vertical line marks the log(HR)=0 (no change in risk). Nobservations = 1370, Nevents = 526.

C4 gene expression values with respect to tri-single-nucleotide polymorphism (SNP).

Count per million (CPM) values in 129 DR3 homozygous individuals showing decreasing C4A and increasing C4B gene expression as tri-SNP 101 allele count increases. Each point represents the median CPM value of multiple samples from one individual. Boxes represent the interquartile range (IQR) and midlines mark the median value.

Figure 3 with 1 supplement
Unique sequence read coverage in C4 region and copy number calls.

Uniquely mapping read coverage from whole genome sequencing (WGS) data of 188 homozygous DR3-DQ2 individuals. C4A and C4B genes share extensive sequence identity along the genes except a ~3 kilobase region indicated with boxes. Reads mapping to these regions were used to estimate C4A (left column) and C4B (right column) copy numbers per sample. Samples were sorted based on C4A copy numbers. A maximum value of 4 was used for the heatmap to moderate high outlier values.

Figure 3—figure supplement 1
Sequence read coverage in C4 region.

(A) Normalized read coverage from whole genome sequencing (WGS) data of 188 homozygous DR3-DQ2 individuals. Reduced coverage of C4 genes relative to the flanking regions indicates the presence of frequent gene deletions. Total C4 copy numbers per sample indicated on the left side of the heatmap were estimated based on the average coverage per sample in C4 region demarcated with blue boxes below the heatmap, including both C4A and C4B but excluding reads mapping to intronic HERV insertion. A maximum value of 4 was used for the heatmap to moderate high outlier values. (B) Histograms and the kernel density estimates (kde) of the sample distribution based on average read coverage for C4 (blue boxes below heatmap) and copy number invariant flanking regions (gray boxes below heatmap). Flanking regions show a tight distribution with a peak around 2 indicating normal/diploid status. C4 coverage distribution shows three peaks around 1, 1.5, and 2; most likely corresponding to 2, 3, and 4 total copies. (C) Distribution of unique coverage corresponding to C4A and C4B genes (see Figure 3). Major C4A peak at 0 coverage indicating C4A null samples and peaks around 0.3 and 0.6 are likely 1 and 2 copies of C4A, respectively. C4B distribution shows a single null sample as well as 2 peaks around 0.3 and 0.6.

Tables

Table 1
Distribution of human leukocyte antigen (HLA) and Tri-single-nucleotide polymorphism (SNP) haplotypes among The Environmental Determinants of Diabetes in the Young (TEDDY) samples.
Tri-SNP Genotype
010/010010/101101/101000/010001/010000/101000/000000/001001/101011/101010/011All
HLA
DR3/DR362348117713211002101616 (20.8%)
DR3/DR44082550319830011113022 (38.9%)
DR4/DR41480533931000001531 (19.7%)
DR4/DR814001297501710001334 (17.2%)
DR1/DR400015143002000160 (2.1%)
DR4/DR1300005600100057 (0.7%)
DR3/DR921600000000018 (0.2%)
DR4/DR913100000000014 (0.2%)
DR4/DR4*030 X/020 X400000000004 (0.1%)
DR4/DR4*030 X/0304300000000003 (0.0%)
All1986
(25.6%)
2920
(37.6%)
1183
(15.2%)
1383
(17.8%)
217
(2.8%)
42
(0.5%)
17
(0.2%)
5
(0.1%)
3
(0.0%)
2
(0.0%)
1
(0.0%)
7759 (100%)
Table 2
Descriptive characteristics of children with respect to the islet autoimmunity (IA) outcome in number (percentage).
IA
NoYesAll
Sex
 Female3388 (49.5)390 (45.5)3778 (49.0)
 Male3457 (50.5)468 (54.5)3925 (51.0)
POP
 EUR6122 (89.4)796 (92.8)6918 (89.8)
 AMR647 (9.5)56 (6.5)703 (9.1)
 AFR76 (1.1)6 (0.7)82 (1.1)
HLA Type
 DR1/DR4142 (2.1)18 (2.1)160 (2.1)
 DR3/DR31486 (21.7)121 (14.1)1607 (20.9)
 DR3/DR42599 (38.0)418 (48.7)3017 (39.2)
 DR4/DR1347 (0.7)10 (1.2)57 (0.7)
 DR4/DR41371 (20.0)158 (18.4)1529 (19.8)
 DR4/DR81200 (17.5)133 (15.5)1333 (17.3)
Country
 US2885 (42.1)297 (34.6)3182 (41.3)
 SWE2019 (29.5)290 (33.8)2309 (30.0)
 FIN1482 (21.7)214 (24.9)1696 (22.0)
 GER459 (6.7)57 (6.6)516 (6.7)
FDR
 06155 (89.9)691 (80.5)6846 (88.9)
 1690 (10.1)167 (19.5)857 (11.1)
All6845 (88.9)858 (11.1)7703 (100)
Table 3
Descriptive characteristics of children with respect to type 1 diabetes diagnosis (T1D) outcome in number (percentage).
T1D
SexNoYesAll
 Female3597 (49.2)181 (45.6)3778 (49.0)
 Male3709 (50.8)216 (54.4)3925 (51.0)
POP
 EUR6552 (89.7)366 (92.2)6918 (89.8)
 AMR674 (9.2)29 (7.3)703 (9.1)
 AFR80 (1.1)2 (0.5)82 (1.1)
HLA Type
 DR1/DR4146 (2.0)14 (3.5)160 (2.1)
 DR3/DR31571 (21.5)36 (9.1)1607 (20.9)
 DR3/DR42797 (38.3)220 (55.4)3017 (39.2)
 DR4/DR1351 (0.7)6 (1.5)57 (0.7)
 DR4/DR41457 (19.9)72 (18.1)1529 (19.8)
 DR4/DR81284 (17.6)49 (12.3)1333 (17.3)
Country
 US3034 (41.5)148 (37.3)3182 (41.3)
 SWE2204 (30.2)105 (26.4)2309 (30.0)
 FIN1591 (21.8)105 (26.4)1696 (22.0)
 GER477 (6.5)39 (9.8)516 (6.7)
FDR
 06560 (89.8)286 (72.0)6846 (88.9)
 1746 (10.2)111 (28.0)857 (11.1)
All7306 (94.8)397 (5.2)7703 (100)
Table 4
Descriptive characteristics of children with respect to glutamic acid decarboxylase autoantibody (GADA)-first appearing antibody outcome in number (percentage).
GADA first
NoYesAll
Sex
 Female3601 (49.2)177 (46.3)3778 (49.0)
 Male3720 (50.8)205 (53.7)3925 (51.0)
POP
 EUR6565 (89.7)353 (92.4)6918 (89.8)
 AMR678 (9.3)25 (6.5)703 (9.1)
 AFR78 (1.1)4 (1.0)82 (1.1)
HLA Type
 DR1/DR4157 (2.1)3 (0.8)160 (2.1)
 DR3/DR31522 (20.8)85 (22.3)1607 (20.9)
 DR3/DR42828 (38.6)189 (49.5)3017 (39.2)
 DR4/DR1355 (0.8)2 (0.5)57 (0.7)
 DR4/DR41472 (20.1)57 (14.9)1529 (19.8)
 DR4/DR81287 (17.6)46 (12.0)1333 (17.3)
Country
 US3036 (41.5)146 (38.2)3182 (41.3)
 SWE2164 (29.6)145 (38.0)2309 (30.0)
 FIN1622 (22.2)74 (19.4)1696 (22.0)
 GER499 (6.8)17 (4.5)516 (6.7)
FDR
 No6531 (89.2)315 (82.5)6846 (88.9)
 Yes790 (10.8)67 (17.5)857 (11.1)
All7321 (95.0)382 (5.0)7703 (100)
Table 5
Descriptive characteristics of children with respect to insulin autoantibody (IAA)-first appearing antibody outcome in number (percentage).
IAA first
NoYesAll
Sex
 Female3636 (49.2)142 (45.4)3778 (49.0)
 Male3754 (50.8)171 (54.6)3925 (51.0)
POP
 EUR6624 (89.6)294 (93.9)6918 (89.8)
 AMR685 (9.3)18 (5.8)703 (9.1)
 AFR81 (1.1)1 (0.3)82 (1.1)
HLA Type
 DR1/DR4151 (2.0)9 (2.9)160 (2.1)
 DR3/DR31578 (21.4)29 (9.3)1607 (20.9)
 DR3/DR42870 (38.8)147 (47.0)3017 (39.2)
 DR4/DR1352 (0.7)5 (1.6)57 (0.7)
 DR4/DR41471 (19.9)58 (18.5)1529 (19.8)
 DR4/DR81268 (17.2)65 (20.8)1333 (17.3)
Country
 US3082 (41.7)100 (31.9)3182 (41.3)
 SWE2215 (30.0)94 (30.0)2309 (30.0)
 FIN1597 (21.6)99 (31.6)1696 (22.0)
 GER496 (6.7)20 (6.4)516 (6.7)
FDR
 No6600 (89.3)246 (78.6)6846 (88.9)
 Yes790 (10.7)67 (21.4)857 (11.1)
All7390 (95.9)313 (4.1)7703 (100)
Table 6
Descriptive characteristics of children with respect to celiac disease diagnosis (CD) outcome in number (percentage).
CD
NoYesAll
Sex
 Female3413 (48.2)365 (59.2)3778 (49.0)
 Male3673 (51.8)252 (40.8)3925 (51.0)
POP
 EUR6325 (89.3)593 (96.1)6918 (89.8)
 AMR681 (9.6)22 (3.6)703 (9.1)
 AFR80 (1.1)2 (0.3)82 (1.1)
HLA Type
 DR1/DR4155 (2.2)5 (0.8)160 (2.1)
 DR3/DR31308 (18.5)299 (48.5)1607 (20.9)
 DR3/DR42805 (39.6)212 (34.4)3017 (39.2)
 DR4/DR1354 (0.8)3 (0.5)57 (0.7)
 DR4/DR41444 (20.4)85 (13.8)1529 (19.8)
 DR4/DR81320 (18.6)13 (2.1)1333 (17.3)
Country
 US2955 (41.7)227 (36.8)3182 (41.3)
 SWE2049 (28.9)260 (42.1)2309 (30.0)
 FIN1594 (22.5)102 (16.5)1696 (22.0)
 GER488 (6.9)28 (4.5)516 (6.7)
FDR
 No6447 (96.4)496 (80.7)6943 (95.1)
 Yes242 (3.6)119 (19.3)361 (4.9)
All7086 (92.0)617 (8.0)7703 (100)
Table 7
Descriptive characteristics of children with respect to celiac disease autoimmunity (CDA) outcome in number (percentage).
CDA
NoYesAll
Sex
 Female2555 (47.2)735 (56.8)3290 (49.0)
 Male2860 (52.8)559 (43.2)3419 (51.0)
POP
 EUR4855 (89.7)1237 (95.6)6092 (90.8)
 AMR515 (9.5)55 (4.3)570 (8.5)
 AFR45 (0.8)2 (0.2)47 (0.7)
HLA Type
 DR1/DR4134 (2.5)8 (0.6)142 (2.1)
 DR3/DR3870 (16.1)530 (41.0)1400 (20.9)
 DR3/DR42146 (39.6)503 (38.9)2649 (39.5)
 DR4/DR1346 (0.8)4 (0.3)50 (0.7)
 DR4/DR41130 (20.9)193 (14.9)1323 (19.7)
 DR4/DR81089 (20.1)56 (4.3)1145 (17.1)
Country
 US2232 (41.2)471 (36.4)2703 (40.3)
 SWE1588 (29.3)479 (37.0)2067 (30.8)
 FIN1262 (23.3)272 (21.0)1534 (22.9)
 GER333 (6.1)72 (5.6)405 (6.0)
FDR
 No5118 (96.5)1119 (86.9)6237 (94.6)
 Yes186 (3.5)169 (13.1)355 (5.4)
All5415 (80.7)1294 (19.3)6709 (100)
Table 8
Previously published Genome-wide association studies (GWAS) associations.

Single-nucleotide polymorphisms (SNPs) used as covariates in our CoxPH analysis. Celiac disease (CD), celiac disease autoimmunity (CDA), type 1 diabetes (T1D), and islet antigen (IA) columns indicating whether the SNP has been shown to be associated with that outcome and hence used in the model (yes) or not (no). Statistically significant hazard ratios for the associated outcome are also provided under HR columns, protective associations in bold.

SNPLocusCDCDAT1DIAPublicationHR CDHR CDAHR T1DHR IA
rs4851575IL18R1, IL18RAPyesnononoSharma et al., 20161.45
rs114569351PLEK,
FBXO48
yesnononoSharma et al., 20162.64
rs12493471CCR9,
LZTFL1,
CXCR6
yesnononoSharma et al., 20161.40
rs1054091RSPH3,
TAGAP
yesnononoSharma et al., 20161.59
rs72704176ASH1LyesnononoSharma et al., 20162.26
rs3771689BAZ2ByesnononoSharma et al., 20160.56
rs13014907ZNF804AyesnononoSharma et al., 20162.46
rs11739460TCOF1yesnononoSharma et al., 20161.41
rs77532435GRB10yesnononoSharma et al., 20162.05
rs6967298AUTS2yesnononoSharma et al., 20160.61
rs61751041LAMB1yesnononoSharma et al., 20162.23
rs2409747XKR6yesyesnonoSharma et al., 20161.581.37
rs12990970NPM1P33,
CTLA4
noyesnonoSharma et al., 20160.76
rs11709472LPPnoyesnonoSharma et al., 20160.80
rs72717025FCGR2AnoyesnonoSharma et al., 20161.84
rs114157400BANK1noyesnonoSharma et al., 20161.62
rs117561283IFNGnoyesnonoSharma et al., 20161.81
rs8013918FOSnoyesnonoSharma et al., 20160.80
rs73043122RNASET2,
MIR3939
nonoyesnoSharma et al., 20183.35
rs113306148PLEKHA1,
MIR3941
nonoyesnoSharma et al., 20183.06
rs428595PPIL2nonoyesyesSharma et al., 20183.422.46
rs1004446INSnonoyesyesKrischer et al., 20170.550.67
rs2476601PTPN22nonoyesyesKrischer et al., 20171.911.73
rs2292239ERBB3nonoyesyesKrischer et al., 20171.681.45
rs3184504SH2B3nononoyesKrischer et al., 20171.40
rs9934817RBFOX1nononoyesSharma et al., 20182.66
rs11705721PXK, PDHBnononoyesSharma et al., 20181.41
Table 9
Estimated C4A gene copy number with respect to tri-single-nucleotide polymorphism (SNP) haplotype.
tri-SNP 101
C4A copy number012
000107
104410
2997
3101
Table 10
Estimated C4B gene copy number with respect to tri-single-nucleotide polymorphism (SNP) haplotype.
tri-SNP 101
C4B copy number012
0100
15344
2417119
3022

Additional files

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Ozkan Aydemir
  2. Jeffrey A Bailey
  3. Daniel Agardh
  4. Åke Lernmark
  5. Janelle A Noble
  6. Agnes Andersson Svärd
  7. Elizabeth P Blankenhorn
  8. Hemang M Parikh
  9. Anette-G Ziegler
  10. Jorma Toppari
  11. Beena Akolkar
  12. William A Hagopian
  13. Marian J Rewers
  14. John P Mordes
  15. TEDDY Study Group
(2026)
Polymorphisms in intron 1 of HLA-DRA differentially associate with type 1 diabetes and celiac disease and implicate involvement of complement system genes C4A and C4B
eLife 12:RP89068.
https://doi.org/10.7554/eLife.89068.2