Impact of diet and genes on murine autoimmune pancreatitis

Received: 29 January 2020 | Revised: 29 May 2020 | Accepted: 5 June 2020 DOI: 10.1111/jcmm.15540 ORIGINAL ARTICLE Impact of diet and genes on murine autoimmune pancreatitis Robert Jaster1 | Yask Gupta2 | Sarah Rohde1 | Luise Ehlers1 | Horst Nizze3 | Artem Vorobyev2,4 | Ralf J. Ludwig2 | Saleh M. Ibrahim2 Department of Medicine II, Division of Gastroenterology, Rostock University Medical Center, Rostock, Germany 1 2 Lübeck Institute of Experimental Dermatology and Center for Research on Inflammation of the Skin, University of Lübeck, Lübeck, Germany 3 Institute of Pathology, Rostock University Medical Center, Rostock, Germany 4 Department of Dermatology and Center for Research on Inflammation of the Skin, University of Lübeck, Lübeck, Germany *Correspondence Robert Jaster, Department of Medicine II, Division of Gastroenterology Rostock University Medical Center ErnstHeydemann-Str. 6, Rostock 18057, Germany. Email: [email protected] Funding information Deutsche Forschungsgemeinschaft, Grant/Award Number: JA 819/5-1 and IB 24/9-1; Research Training Group 'Genes, Environment and Inflammation', Grant/ Award Number: GRK 1743/2 Abstract The impact of environmental factors, such as diet, and the genetic basis of autoimmune pancreatitis (AIP) are largely unknown. Here, we used an experimental murine AIP model to identify the contribution of diet to AIP development, as well as to finemap AIP-associated genes in outbred mice prone to develop the disease. For this purpose, we fed mice of an autoimmune-prone intercross line (AIL) three different diets (control, calorie-reduced and western diet) for 6 months, at which point the mice were genotyped and phenotyped for AIP. Overall, 269 out of 734 mice (36.6%) developed AIP with signs of parenchymal destruction, equally affecting mice of both sexes. AIP prevalence and severity were reduced by approximately 50% in mice held under caloric restriction compared to those fed control or western diet. We identified a quantitative trait locus (QTL) on chromosome 4 to be associated with AIP, which is located within a previously reported QTL. This association does not change when considering diet or sex as an additional variable for the mapping. Using wholegenome sequences of the AIL founder strains, we resolved this QTL to a single candidate gene, namely Map3k7. Expression of Map3k7 was largely restricted to islet cells as well as lymphocytes found in the exocrine pancreas of mice with AIP. Our studies suggest a major impact of diet on AIP. Furthermore, we identify Map3k7 as a novel susceptibility gene for experimental AIP. Both findings warrant clinical translation. KEYWORDS autoimmune pancreatitis, diet, gene-environment interactions, Map3k7, MRL/MpJ mice, QTL mapping, susceptibility genes 1 | I NTRO D U C TI O N diagnosis is of particular importance since AIP patients, in contrast to patients with other forms of CP, usually respond well to Autoimmune pancreatitis (AIP) is a rare but clinically relevant steroid treatment.1,2 The current concept of AIP pathophysiology form of chronic pancreatitis (CP). AIP patients may present differentiates two subtypes of AIP, AIP type 1 and type 2, which with obstructive jaundice due to the formation of inflammatory are distinguished by their pathogenesis: AIP type 1 is the pancre- pseudotumours, a finding that may evoke the differential diagno- atic manifestation of IgG4-related disease and is characterized sis of pancreatic cancer and result in surgical treatment. Correct by dense infiltrates of IgG4-positive plasma cells. In AIP type 2, Ralf J. Ludwig and Saleh M. Ibrahim are equal contributors. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2020 The Authors. Journal of Cellular and Molecular Medicine published by Foundation for Cellular and Molecular Medicine and John Wiley & Sons Ltd. J Cell Mol Med. 2020;00:1–9. wileyonlinelibrary.com/journal/jcmm | 1 2 | JASTER ET Al. which is frequently associated with inflammatory bowel disease, 2 | M ATE R I A L S A N D M E TH O DS granulocytic epithelial lesions are the key pathognomonic finding, whereas IgG4-expressing plasma cells are lacking. Common his- 2.1 | Animals tological features of both subtypes are a periductal fibrosis and the presence of periductal lymphoplasmacytic infiltrates.1-3 AIP MRL/MpJ, NZM2410/J, BXD2/TyJ and CAST/EiJ parental mouse shares with other autoimmune diseases basic immunological char- strains were intercrossed at an equal strain and sex distribution as acteristics, such as presence of autoantibodies and involvement described of autoreactive T cells. Furthermore, adoptive transfer of splenic tercross line (AIL). As previously reported, AIL mice develop AIP, 20 leucocyte subpopulations (specifically, CD44 high memory T cells) which is most likely due to the inclusion of the AIP-susceptible from MRL/MpJ mice with AIP into healthy mice induces AIP in the MRL/MpJ mice.16,18-20 For the mapping study herein, mice of the recipients. 4,19,20 to generate an advanced autoimmune-prone in- 15th, 18th-20th generation were used. To study the impact of diet 4 Environmental and nutritional influences are well-established on AIP, after weaning, AIL mice were randomly allocated to one key factors in the pathogenesis of pancreatitis. Of note, chronic alco- of the following diets: control mouse chow, caloric restriction and hol abuse represents the most common cause of CP in general, and western diet, as described elsewhere.19 Regarding randomization: hypertriglyceridemia is associated with an increased risk of acute Offspring mice were transferred into separate cages after weaning pancreatitis.5 The role of such factors in the development of AIP has, at 3-4 weeks. Each cage contained mice of either gender and was however, not been systematically studied yet. Therefore, we here randomly allocated at a 1:1:1 to one of the three different diets: con- also addressed the contribution of diet, as an every-day environmen- trol mouse chow, caloric restriction and western diet (all mice of one tal factor, on AIP development. cage received the same diet, which was selected at random). At the Regarding genetic associations, genome-wide association age of 6 months, a skin biopsy (for genotyping) and the pancreas studies (GWAS) and candidate gene-based approaches have (for histological analysis) were obtained (Figure 1A). A total of 734 identified several susceptibility loci for human AIP (largely, type mice lived until that age; of these, 461 were female and 273 were 1), including HLA DRB1*04:05-DQB1*04:01, 6 FCRL3,7 CTLA4 8,9 male. A total of 84, 276 or 101 female mice were in the caloric re- and KCNA3.10 Targeted sequencing has suggested additional ge- striction, control chow or western diet arms, respectively. For males, netic associations of AIP, specifically CACNA1S, SMAD7, TOP111 this amounted to 79, 140 or 54 mice (Table 1). Unequal sample size 12 and CALCB. Noteworthy, specific PRSS1 mutations (PRSS1_IVS distribution is due to the cage-wise randomization and the death of 2+56_60 delCCCAG and PRSS1_p.Leu81Met) that were suggested mice during the 6-month observation period. Mice were held under to cause ectopic trypsinogen activation have also been implicated specific pathogen-free conditions at 12-hour light/dark cycle at the into the pathogenesis of human AIP type 1.13,14 Further insights animal facility of the University of Lübeck, Germany. Animal experi- into the genetic architecture of AIP have been obtained using the ments were conducted according to the European Community rules MRL/MpJ mouse AIP model. At an age of approximately 6 months, for animal care, approved by the respective governmental adminis- predominantly female MRL/MpJ mice spontaneously develop AIP tration (Ministry for Energy, Agriculture, the Environment and Rural that resembles important histopathological features of human Areas, file number 27-2/13) and performed by certified personnel. 15,16 AIP. Application of polyinosinic:polycytidylic acid (poly I:C) further accelerates and enhances the disease.17,18 Previously, we generated an autoimmune-prone advanced intercross mouse 2.2 | Genotyping and QTL mapping (AIL) line, to study the impact of genetics and diet on complex (inflammatory disease) traits in the mouse, including AIP.19 In brief, The tips of the tails of AIL mice were collected during mouse sam- 3 parental mouse strains that spontaneously develop different pling at month 6 and subjected to the isolation of genomic DNA as autoimmune diseases—MRL/MpJ (AIP), NZM2410/J (lupus) and reported.19 In brief, the DNeasy Blood & Tissue Kit (Qiagen GmbH, BXD2/TyJ (arthritis)—and CAST/EIJ mice, which are not prone Hilden, Germany) was employed according to the manufacturer's 20 protocol, and DNA samples were stored at −20°C until further use. These AIL mice have been used for QTL mapping of several com- DNA samples from 734 mice were analysed by MegaMUGA geno- to any autoimmune disease, were subjected to crossbreeding. plex traits, including murine AIP.19,20 Regarding AIP, 5 associated typing array covering 77 800 markers throughout the mouse ge- QTL were identified, located on chromosomes 2, 4 (two QTLs), 5 nome. 21 Genotyping was performed at Neogen/GeneSeek (Lincoln, and 6. 20 The identified QTL, however, encompassed numerous NE, USA). Non-informative SNPs were filtered out using plink and genes, due to the fact that the study was performed in mice of the applying the following criteria: minor allele frequency (maf) >0.05, 4th generation, where relatively few crossing over had occurred. missing genotype probability <0.1 and common homozygous SNPs To fine map these QTLs, preferably to the single gene level, we among the founders. This approach yielded 55,458 informative SNPs here used mice of the 15th, 18th-20th generation. In addition, we for further downstream analysis. 22 The probabilistic reconstruction took advantage of recently published whole-genome sequences of AIL mouse genome in term of four founder strains was performed (WGS) of the AIL founder strains using the R package of HAPPY23,24 as described before.19 Briefly, ciations of AIP. 19 to fine map the genetic asso- the posterior probability that each mouse was in one of the four | JASTER ET Al. 3 F I G U R E 1 Diet modulates the spontaneous development of autoimmune pancreatitis (AIP) in an advanced intercross outbred mouse line (AIL) (A) A total of 734 outbred AIL mice were fed 3 different diets (caloric restriction, control or western diet) for 6 months. Thereafter, mice were genotyped and H&E-stained sections from the pancreata were evaluated for the presence and (if present) severity of AIP. (B) Pancreatic sections were stained with H&E and subjected to the evaluation of pathological changes on a semi-quantitative scale from 0 to 4. Stage 0: healthy; stage 1: small infiltrate of mononuclear cells (arrow); stage 2: large periductal focus of mononuclear cells; beginning parenchymal destruction; stage 3: severe inflammation and more extended parenchymal destruction; and stage 4: organ-wide inflammation; large-scale destruction of acini and partial replacement by adipose and fibrotic tissue. (C) When stratified for sex, no difference in the AIP severity was noted. Graph shows all individual values as dots, whereas the red line indicates the median. Statistical analysis was performed using rank-sum test. (D) By contrast, stratification for the different diets showed that mice fed calorie-reduced diet had a significant lower AIP severity compared to control and western diet-fed mice. Graph shows all individual values as dots, whereas the red line indicates the median. Statistical analysis was performed using one-way ANOVA (P < 0.0001) with Tukey's multiple comparison as post-test possible genotype states was estimated employing a hidden Markov covariates rather than additive covariate within the same function. model (HMM), and probabilities were converted to Rqtl2 object in R. Genome-wide and suggestive thresholds were estimated by tradi- Subsequently, a chromosome-wise kinship matrix, which represents tional permutation (1000)-based approach at 5% and 10% threshold. intra-individual relationship, was calculated using 'calc_kinship' func- The confidence interval for a QTL was described by 1.5 LOD drop tion with leaving out one chromosome at a time (R/qtl2 R package). 25 within 10 Mb from the peak SNP. To further fine map the QTL, WGS We estimated heritability of AIP using 'est_herit' function from the data from founder strains (variant call file format; vcf) were down- same R package, that is R/qtl2. For additive model, LOD scores were loaded from database.19 The annotation of SNPs and Indels within calculated by linear mixed model in which AIP score was regressed this file was performed using ENSEMBL VEP web server. 26 The to posterior probabilities with sex and diet as additive covariate and genomic region plots for the QTL were visualized and created using kinship as a random effect (scan1 function from Rqtl2). For QTL in- R ggplot2 teraction model, covariates diet and sex were considered interactive has been recently validated.9 27 and UCSC Genome Browser. 28 This analytical pipeline 4 | JASTER ET Al. Score pancreas Sex Diet 0 1 2 3 4 Scores 2-4 (%) Female Caloric restriction 60 7 14 3 0 20.2 79.8 Control chow 75 87 107 7 0 41.3 58.7 Western diet 43 11 32 14 1 46.5 53.5 Caloric restriction 56 6 14 3 0 21.5 78.5 Control chow 51 43 41 5 0 32.9 67.1 Male Both Scores 0-1 (%) Western diet 21 5 21 7 0 51.9 48.1 Caloric restriction 116 13 28 6 0 20.9 79.1 Control chow 126 130 148 12 0 38.5 61.5 Western diet 64 16 53 21 1 48.4 51.6 2.3 | Histology and immunohistochemistry Development of spontaneous AIP was assessed based on pancreatic histopathology. Therefore, paraffin-embedded pancreatic sections TA B L E 1 Pancreatic phenotypes of AIL mice. A total of 734 mice (461 females and 273 males, all from G15-19) were fed the indicated diets for 6 months, before phenotyping was performed by assessing pancreatic histopathology on a semi-quantitative scale from 0 (healthy pancreas) to 4 (most severe AIP). The unequal sex distribution is due to (i) the sex ratio at birth (which was in favour of females), and (ii) a higher mortality rate in male mice during the 6 months of the experiment 3 | R E S U LT S 3.1 | Caloric restriction reduces the prevalence and clinical severity of autoimmune pancreatitis (4 µm thick) were stained with haematoxylin and eosin (H&E), and subjected to the evaluation of pancreatic lesions on a semi-quantita16,18,20 tive scale from 0 to 4 as described before. Briefly, the stages At 6 months, 428 out of 734 mice (58.3%) presented with lymphocytic infiltrates in the exocrine pancreas. A total of 269 animals were defined as follows: 0, no pathological findings; 1, minimal focal (36.6%) displayed signs of parenchymal destruction and were clas- infiltration of periductal tissue with mononuclear cells but lack of sified as AIP stage 2 or greater. Both sexes were similarly affected parenchymal destruction; 2, presence of larger periductal foci of by AIP (Figure 1C). Prevalence of AIP stage 2 or greater amounted mononuclear cells along with beginning parenchymal destruction; 3, to 38.5% in female and 33.4% in male mice, and average AIP scores severe and multifocal periductal inflammation together with more were 1.1 ± 0.4 or 0.9 ± 0.4 for female or male mice, respectively. extended parenchymal destruction; and 4, comprehensive infiltra- AIP scores also showed no differences between both sexes when tion of pancreatic tissue with mononuclear cells, large-scale destruc- calculation was performed separately within each dietary group. tion of acini and (partial) replacement by adipose or fibrotic tissue When comparing AIP prevalence among the different diets, 71% (Figure 1B). All samples were assessed by microscopic analysis of at mice under caloric restriction were free of any AIP, whereas an least three tissue sections per sample by two independent investiga- AIP score of 0 was only observed in 30% and 41% of the mice fed tors and blinded before evaluation. control and western diet, respectively (P < 0.05; chi-squared test). Immunohistochemistry was performed on 4 µm thick paraf- Corresponding findings were made regarding AIP severity: In mice fin-embedded pancreatic sections. The sections were deparaffinized held at caloric restriction, the histological AIP score was 0.5 ± 0.03; and subsequently stained using the ImmPRESS—alkaline phospha- whereas the score reached 1.1 ± 0.03 or 1.2 ± 0.04 in mice fed tase detection system according to the instructions of the manu- control or western diet, respectively (Figure 1D). Strikingly, cases facturer (Vector Laboratories, Burlingame, CA, USA). For staining with parenchymal destruction (stages 2-4) were much less common of MAP3K7, a mouse protein-specific polyclonal antibody from in mice with caloric restriction than in animals on control chow or Biorbyt, Cambridge, UK (catalogue number: orb32004) was em- western diet (Table 1). Taken together, these findings point towards ployed. The slides were then counterstained with Mayer's hemalum a protective role of caloric restriction on the development and on solution, dehydrated by two short incubations in ethanol and xylene the progression of murine AIP. The principle effects of diets on the each and embedded in Pertex (MEDITE, Burgdorf, Germany). average scores were the same in females and males (data not shown). 2.4 | Statistical analysis 3.2 | Mitogen-activated protein kinase kinase kinase 7 (Map3k7) is associated with experimental autoimmune pancreatitis Unless stated otherwise, data were analysed employing IBM SPSS Statistics V25.0 and are presented as mean values ± SEM. Statistical tests are detailed in the figure legends. P values of < 0.05 were con- Next, we aimed to fine map previously reported QTLs of AIP, 20 pos- sidered statistically significant. We used Dsquare function from sibly identify so far unreported AIP-associated genes, and evaluate modEvA R package to estimate proportion of variance explained by the impact of gene-diet and gene-sex interactions in AIP. For fine diet (4.7%) and sex (0.3%) for pancreatitis score. mapping, we continuously intercrossed AIL mice leading to smaller | JASTER ET Al. 5 confidence intervals of QTL due to the meiotic crossing over. In the pancreata, MAP3K7 protein expression was largely restricted to 15th, 18th-20th AIL mouse generation, 734 mice were genotyped pancreatic islets (Figure 3). By contrast, pancreata from mice with and phenotyped for AIP severity. Considering diet and sex as addi- AIP showed additional expression by the pancreas-infiltrating mono- tive variables, we identified a QTL at the genome-wide significance nuclear cells (Figure 3). level on chromosome 4 (LOD = 6.44, Confidence interval (CI) = 3132.6 Mb, Figure 2A-B), which was located within the previously identified QTL for AIP (Table 2). We next considered diet or sex 4 | D I S CU S S I O N as interactive covariates for the QTL mapping, allowing to identify gene-diet and gene-sex interaction: The QTL on chromosome 4 Autoimmune disorders are multifactorial diseases that are influ- was confirmed at the genome-wide significance level (LOD = 7.68, enced by the interplay of environmental, for example dietary, and CI = 31.39-34.8 Mb) when diet was taken as an interactive variable. genetic factors.19 In the case of AIP, both aspects are still relatively When using sex as an interactive variable, the QTL was significant at poorly understood. We herein addressed this knowledge gap and the suggestive level (LOD = 6.89, CI = 31-36.1 Mb, Figure 2B). We show that diet has a vast impact on AIP development. More specifi- also mapped a suggestive QTL interacting with diet on chromosome cally, compared to control or western diet, caloric restriction halved 4 (LOD = 6.76) at CI of 135.5-136.4 Mb (Table 2), suggesting that both AIP prevalence and severity. Regarding the genetic AIP associa- diet can potentially unmask novel genetic associations also in AIP. In tion, we fine-mapped Map3k7 as an AIP-associated gene and dem- addition, 6.8% or 0.46% of AIP phenotypic variance are explained by onstrate its expression in the leucocyte infiltrate within the pancreas diet or sex, respectively. of mice with AIP. Still, the fine-mapped QTL, located on chromosome 4, contained Our genetic studies were based on the report of five, rather large several genes. Hence, for further fine mapping, we first correlated QTLs that were mapped within in AIL mice in the 4th generation, 20 (Spearman rank correlation) the posterior probability of the peak including one QTL located on chromosome 4 (termed AIP2). In order SNP (UNC7008858) for the individual founder strains to the AIP to obtain smaller confidence intervals of the QTLs, we took advan- score. Here, BxD2/TyJ and MRL/MpJ strains negatively correlated tage of meiotic crossing overs across generations. By continued in- (ρ = −0.18, P < 0.01) and positively (ρ = 0.14, P < 0.01) with AIP score. tercrossing of AIL mice for 20 generations and use of WGS data of The other two strains CAST/EiJ (ρ = 0.02, P = 0.57) and NZM2410/J the founder strains, we were here able to confirm and to narrow (ρ = 0.0007, P = 0.98) showed no correlation with the AIP score down the size of one of these QTLs, namely AIP2. (Figure 2C). Hence, as indicated by the haplotype effects, the ge- The herein identified QTL within the AIP2 locus encodes for one netic association of the AIP score is due to the genetic differences protein-coding gene, Map3k7 (Tak1). By contrasting the genomic between the BxD2/TyJ and MRL/MpJ strains within this locus. Next, variations of the AIL founder strains, the variability within this QTL we used the recently published WGS of the 4 AIL founder strains 19 to identify polymorphisms (SNPs and Indels) discriminating between derived from the AIP-prone MRL/MpJ strain, whereas the Map3k7 allele of BXD2/TyJ mice was associated with less AIP. BxD2/TyJ and MRL/MpJ strains. We identified 2,994 variants be- The MAP3K7 protein represents an essential signalling interme- tween the two strains among which most of the variants were either diate in tumour necrosis factor, interleukin 1, and Toll-like receptor intergenic regions (38%), intronic (36%) or located within non-coding signalling pathways. The member of the mitogen-activated protein transcripts without any predicted severe consequences. For pro- kinase (MAPK) kinase kinase family transmits upstream signals tein-coding genes, we did not identify any non-synonymous variants from the receptor complexes to the downstream MAPKs and to with severe consequences. However, we detected multiple variants the NF-κB pathway. 29-31 Whereas Map3k7 has not been associated on predicted upstream/downstream region and importantly in 3’ with AIP previously, mutations within this gene are associated with UTR region of the full-length protein-coding gene Map3k7, whereas other experimental inflammatory diseases in the mouse, including for the partial protein-coding gene Bach2 variants in predicted up- murine autoimmune myocarditis,32 encephalomyelitis33 and type stream region were found (Figure 2D). The importance of variants in 1 diabetes.33 In the latter two models, an inhibitor of MAP3K7/ 3’ UTR has been established previously. Hence, we identify Map3k7, TAK1, 5Z-7-oxozeaenol (OZ), attenuated progression of the disease also known as transforming growth factor-beta-activated kinase 1 and will therefore be of interest in the context of experimental AIP (Tak1), spans from 31,96 to 32,02 Mb, to be potentially associated as well. The mechanisms of OZ action are apparently complex and with AIP. may involve impairment of dendritic cell maturation and a Th1 to Th2 cytokine shift.34 However, when mapping 54 physiological and pathophysiological phenotypes, including ANA production and 3.3 | Map3k7 is expressed by infiltrating mononuclear cells of mice with experimental autoimmune pancreatitis lupus susceptibility, no association with Map3k7 was found, pointing towards a selective role of this gene in AIP pathogenesis.18 In addition to AIP2, we previously have identified additional QTLs for AIP located on chromosomes 2, 5 and 6.20 These previ- To obtain insights into a potential contribution of Map3k7, we stud- ously reported QTLs were, however, not replicated within the cur- ied its protein expression in the context of murine AIP. In healthy rent study. This is likely because of the epistatic interactions or gene 6 | JASTER ET Al. environmental interactions. It should also be noted that generation microenvironment may have altered the association of AIP with host 4th AIL mice were housed at the University of Rostock, whereas later genetics. For example, when we introduced diet (microenvironment generations were kept at the University of Lübeck. These changes in modulator) as an interacting covariate, the highest LOD scores on | JASTER ET Al. 7 F I G U R E 2 Map3k7 is associated with experimental autoimmune pancreatitis (A) Manhattan plot for mapping of quantitative trait loci (QTLs) for murine autoimmune pancreatitis (AIP9. The plot shows chromosomes on x-axis and LOD (log of odd ratios) for each SNP (grey and black dots) on y-axis. LOD scores were calculated by regressing histological pancreatitis score with the estimated genotype posterior probability. Sex, diet and first principle component of kinship matrix were considered as additive covariate loci. Logarithmic p-values are plotted against genome location. The red line represents the genome-wide threshold (P < 0.05), and the blue line shows the suggestive threshold (P < 0.1) evaluated after 1000 permutations for histological pancreatitis score. (B) The scatter line plot shows LOD scores across chromosome 4 for three models (ie additive, interactive diet and sex). Use of either one of the interactive variables ('diet' or 'sex') considers the variability introduced by either of the variables, allowing to detect gene-diet and gene-sex interactions. The shaded region within the plot described confidence interval in Mb for each model (grey = AIP2, orange = genome-wide and blue = chromosome-wide). (C) The box plot shows ranked haplotype posterior probabilities (derived from happy R package using HMM) for Peak SNP (UNC7008858) across the four founder strains, when stratified for different AIP score. (D) The figure shows the fine-mapped AIP2 loci (using UCSC browser) on chromosome 4 and genes within this loci. The figure also shows all the SNPs and Indels (derived from WGS) different between BxD2/TyJ and MRL/MpJ, as Peak SNP from these strains was negatively and positively correlated with AIP score in previous figure chromosome 2 (LOD = 4.07) and 5 (LOD = 5.25) were observed at 65.9 the rate of AIP with parenchymal destruction in our model. Western (SNP = UNC3225113) and 78.7 (SNP = UNC9531776) Mb. The re- diet, on the other hand, did not significantly affect the course of spective regions fall within the confidence intervals of the previously the disease, although we noticed a tendency towards more severe reported QTLs on chromosome 2 (56.3-81.9 Mb) and chromosome stages of AIP (Table 1). The effects of diet were observed in both 5 (74.8-96.6 Mb) but did not pass the significance threshold. These sexes, which in this study did also not significantly differ with re- discrepancies were also reported in previous studies and should be spect to the overall occurrence of AIP. To the best of our knowledge, further investigated for AIP.19 AIP2 comprised a CI of 25.2 Mbp (Mbp this is the first study to show an effect of diet on the development of 7.1-32.3), contained a variety of genes and was therefore too large for AIP in either an experimental model or AIP patients. In a rat model the analysis of genetic traits at the level of individual genes. In con- of non-insulin-dependent diabetes with obesity, WBN/Kob-fatty trast, the herein fine-mapped QTL was 1.55 Mb only and contained rats, caloric restriction was found to be associated with less severe several predicted and pseudogenes, a full-length protein-coding gene signs of spontaneous CP and specifically diminished interlobular, in- (Map3k7) and a partial protein-coding gene (Bach2). The latter gene tra-lobular and intra-islet fibrosis. These findings, however, could be might be of interest for follow-up studies as well since recent data attributed to a unique genetic background, the homozygous pres- suggest that Bach2 repression is associated with clinical features of ence of the fa allele of the leptin receptor gene, and are also not advanced CP through an increased Th17 cell-induced inflammation,35 directly linked to AIP.36 Apart from AIP, growing evidence from ex- a subset of T cells that may also play a role in murine AIP. 4 perimental and clinical studies suggests that caloric restriction and Of note, the QTL effect observed in this study (Figure 2) cannot fasting-mimicking diets may positively impact autoimmune diseases be caused by a single SNP, as the effect (a) differs between strains such as systemic lupus erythematosus (SLE), rheumatoid arthritis, CAST/EIJ + NZM2410/J (both no effect) versus BXD2/TyJ + MRL/ multiple sclerosis and autoimmune diabetes.19,37,38 In addition, we MpJ, (b) but BXD2/TyJ + MRL/MpJ differ according to severity. The recently demonstrated that caloric restriction confers complete pro- contribution of different SNPs to this complex pattern represents a tection from the development of lupus nephritis in NZM2410/J mice, point of interest for follow-up investigations. whereas mice fed western diet showed an accelerated and aggra- Using the same autoimmune-prone AIL as in this study, we have vated clinical phenotype. Interestingly diet-induced changes in the recently shown that diet substantially contributes to the variability intestinal micro- and mycobiome preceded clinical onset, indicating of complex traits and, in addition, may unmask novel susceptibility that the protective effects of diet may be modulated by changes in QTLs. Strikingly, caloric restriction alone was sufficient to over- the intestinal microbiome.19 Although it is well known that the type come the genetic susceptibility of the mice to develop lupus. This and levels of nutrients can influence the generation, survival and observation prompted us to ask whether diet might also influence function of lymphocytes, the mechanistic links between caloric re- the development of autoimmune pancreatitis in genetically suscep- striction and autoimmunity are far from being understood. Given the tible mice. The results of this study show that, compared to control potential clinical implications of dietary interventions, our results chow or western diet, long-term caloric restriction roughly halved encourage further studies in the context of AIP. 19 TA B L E 2 Pancreatitis Score QTLs identified by fine mapping. Genome-wide threshold for LOD (0.05), suggestive threshold (0.1). In addition, 6.8% or 0.46% of AIP phenotypic variance are explained by diet or sex, respectively Chr CI (Mb) Peak SNP Peak Pos (Mb) Pheno var % LOD Model Significance 4 31-32.56 UNC7008858 31.94 3.96 6.44 Additive Genome 4 31.39-34.8 UNC7008858 31.94 4.70 7.68 QTL x Diet Genome 4 135.55-136.48 UNC8290092 135.77 4.15 6.76 QTL x Diet Suggestive 4 31-36.12 UNC7008858 31.94 4.23 6.89 QTL x Sex Suggestive Abbreviations: Chr, chromosome; CI, confidence interval; pheno var, phenotypic variability. 8 | JASTER ET Al. F I G U R E 3 MAP3K7 is expressed in the lymphocyte infiltrates of experimental autoimmune pancreatitis Pancreatic tissues from mice with AIP stage 0 (healthy) and stage 3 (severe AIP) were stained with anti-MAP3K7. Arrows point to positively stained pancreatic islets and organ-infiltrating mononuclear cells, respectively. The photographs are representative for n ≥ 5 mice per group In summary, this investigation has identified a novel susceptibil- Investigation (supporting); Methodology (supporting); Resources ity gene of murine AIP, Map3k7, and a potent protective environmen- (equal); Supervision (equal); Validation (equal); Visualization (sup- tal factor, dietary restriction. As the MAP3K7 protein is in principle porting); Writing-original draft (supporting); Writing-review & edit- druggable39 and dietary interventions are possible, these findings ing (supporting). may shape the future management of AIP patient. DATA AVA I L A B I L I T Y S TAT E M E N T AC K N OW L E D G E M E N T S The data that support the findings of this study are available from We gratefully acknowledge the excellent technical assistance of the corresponding author upon reasonable request. Mrs Katja Bergmann. This work was supported by the Deutsche Forschungsgemeinschaft (DFG JA 819/5-1 and IB 24/9-1) and the ORCID Research Training Group 'Genes, Environment and Inflammation' Robert Jaster https://orcid.org/0000-0002-8220-4570 (GRK 1743/2). REFERENCES C O N FL I C T O F I N T E R E S T None to declare. AU T H O R C O N T R I B U T I O N Robert Jaster: Conceptualization (lead); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (lead); Resources (equal); Supervision (equal); Validation (equal); Visualization (equal); Writingoriginal draft (lead); Writing-review & editing (lead). Yask Gupta: Data curation (lead); Formal analysis (supporting); Methodology (equal); Validation (supporting); Visualization (supporting); Writing-original draft (supporting); Writing-review & editing (supporting). Sarah Rohde: Investigation (equal); Methodology (equal); Validation (supporting); Writing-original draft (supporting); Writing-review & editing (supporting). Luise Ehlers: Investigation (equal); Methodology (supporting); Validation (supporting); Writing-original draft (supporting); Writing-review & editing (supporting). Horst Nizze: Investigation (equal); Methodology (supporting); Writing-original draft (supporting); Writing-review & editing (supporting). Artem Vorobyev: Investigation (equal); Methodology (supporting); Validation (equal); Writing-original draft (supporting); Writing-review & editing (supporting). Ralf J Ludwig: Conceptualization (supporting); Data curation (supporting); Formal analysis (supporting); Funding acquisition (supporting); Investigation (supporting); Methodology (supporting); Resources (supporting); Supervision (equal); Validation (equal); Visualization (lead); Writing-original draft (equal); Writing-review & editing (equal). Saleh Ibrahim: Conceptualization (equal); Data curation (supporting); Formal analysis (equal); Funding acquisition (equal); 1. Uchida K, Okazaki K. Clinical and pathophysiological aspects of type 1 autoimmune pancreatitis. J Gastroenterol. 2018;53:475-483. 2. Hart PA, Zen Y, Chari ST. Recent advances in autoimmune pancreatitis. Gastroenterology. 2015;149:39-51. 3. Kamisawa T, Chari ST, Giday SA, et al. Clinical profile of autoimmune pancreatitis and its histological subtypes: an international multicenter survey. Pancreas. 2011;40:809-814. 4. Bischof J, Müller S, Borufka L, et al. Quantitative trait locus analysis implicates CD4⁺/CD44high memory T cells in the pathogenesis of murine autoimmune pancreatitis. PLoS One. 2015;10:e0136298. 5. Weiss FU, Laemmerhirt F, Lerch MM. Etiology and risk factors of acute and chronic pancreatitis. Visc Med. 2019;35:73-81. 6. Kawa S, Ota M, Yoshizawa K, et al. HLA DRB10405-DQB10401 haplotype is associated with autoimmune pancreatitis in the Japanese population. Gastroenterology. 2002;122:1264-1349. 7. Umemura T, Ota M, Hamano H, et al. Genetic association of Fc receptor-like 3 polymorphisms with autoimmune pancreatitis in Japanese patients. Gut. 2006;55:1367-1368. 8. Umemura T, Ota M, Hamano H, et al. Association of autoimmune pancreatitis with cytotoxic T-lymphocyte antigen 4 gene polymorphisms in Japanese patients. Am J Gastroenterol. 2008;103:588-594. 9. Chang MC, Chang YT, Tien YW, et al. T-cell regulatory gene CTLA-4 polymorphism/haplotype association with autoimmune pancreatitis. Clin Chem. 2007;53:1700-1705. https://doi.org/10.1373/clinc hem.2007.085951 10. Ota M, Ito T, Umemura T, et al. Polymorphism in the KCNA3 gene is associated with susceptibility to autoimmune pancreatitis in the Japanese population. Dis Markers. 2011;31:223-229. 11. Fujibayashi S, Sasajima J, Goto T, et al. A high-throughput sequence analysis of Japanese patients revealed 11 candidate genes associated with type 1 autoimmune pancreatitis susceptibility. Biochem Biophys Rep. 2016;6:76-81. | JASTER ET Al. 12. Liu QC, Chen F, Wu CY, et al. CALCB splice region pathogenic variants leading to plasma cell neurotropic enrichment in type 1 autoimmune pancreatitis. Cell Death Dis. 2017;8:e2591. 13. Gao F, Li YM, Hong GL, et al. PRSS1_p.Leu81Met mutation results in autoimmune pancreatitis. World J Gastroenterol. 2013;19:3332-3338. 14. Gao F, Li Y, Wang C, et al. Identification of a novel frame-shift mutation in PRSS1 gene in Han patients with autoimmune pancreatitis. Curr Mol Med. 2014;14:340-348. 15. Kanno H, Nose M, Itoh J, et al. Spontaneous development of pancreatitis in the MRL/Mp strain of mice in autoimmune mechanism. Clin Exp Immunol. 1992;89:68-73. 16. Sorg H, Lorch B, Jaster R, et al. Early rise in inflammation and microcirculatory disorder determine the development of autoimmune pancreatitis in the MRL/Mp-mouse. Am J Physiol Gastrointest Liver Physiol. 2008;295:G1274-G1280. 17. Qu WM, Miyazaki T, Terada M, et al. A novel autoimmune pancreatitis model in MRL mice treated with polyinosinic:polycytidylic acid. Clin Exp Immunol. 2002;129:27-34. 18. Schwaiger T, van den Brandt C, Fitzner B, et al. Autoimmune pancreatitis in MRL/Mp mice is a T cell-mediated disease responsive to cyclosporine A and rapamycin treatment. Gut. 2014;63:494-505. 19. Vorobyev A, Gupta Y, Sezin T, et al. Gene-diet interactions associated with complex trait variation in an advanced intercross outbred mouse line. Nat Commun. 2019;10:4097. 20. Asghari F, Fitzner B, Holzhüter SA, et al. Identification of quantitative trait loci for murine autoimmune pancreatitis. J Med Genet. 2011;48:557-562. 21. Morgan AP, Fu CP, Kao CY, et al. The mouse universal genotyping array: from substrains to subspecies. G3: Genes - Genomes Genetics. 2015;6:263-279. 22. Purcell S, Neale B, Todd-Brown K, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81:559-575. 23. Mott R, Talbot CJ, Turri MG, et al. A method for fine mapping quantitative trait loci in outbred animal stocks. Proc Natl Acad Sci USA. 2000;97:12649-12654. 24. Belheouane M, Gupta Y, Künzel S, et al. Improved detection of gene-microbe interactions in the mouse skin microbiota using high-resolution QTL mapping of 16S rRNA transcripts. Microbiome. 2017;5:59. 25. Broman KW, Gatti DM, Simecek P, et al. R/qtl2: software for mapping quantitative trait loci with high-dimensional data and multiparent populations. Genetics. 2019;211:495-502. 26. McLaren W, Gil L, Hunt SE, et al. The ensembl variant effect predictor. Genome Biol. 2016;17:122. 9 27. Ito K, Murphy D. Application of ggplot2 to pharmacometric graphics. CPT Pharmacometrics Syst Pharmacol. 2013;2:e79. 28. Tyner C, Barber GP, Casper J, et al. The UCSC Genome Browser database: 2017 update. Nucleic Acids Res. 2017;45:D626-D634. 29. Ninomiya-Tsuji J, Kishimoto K, Hiyama A, et al. The kinase TAK1 can activate the NIK-I kappaB as well as the MAP kinase cascade in the IL-1 signalling pathway. Nature. 1999;398:252-256. 30. Sato S, Sanjo H, Takeda K, et al. Essential function for the kinase TAK1 in innate and adaptive immune responses. Nat Immunol. 2005;6:1087-1095. 31. Takaesu G, Surabhi RM, Park KJ, et al. TAK1 is critical for IkappaB kinase-mediated activation of the NF-kappaB pathway. J Mol Biol. 2003;326:105-115. 32. Blyszczuk P, Müller-Edenborn B, Valenta T, et al. Transforming growth factor-β-dependent Wnt secretion controls myofibroblast formation and myocardial fibrosis progression in experimental autoimmune myocarditis. Eur Heart J. 2017;38:1413-1425. 33. Lu L, Zhang X, Tong H, et al. Central administration of 5Z-7oxozeaenol protects experimental autoimmune encephalomyelitis mice by inhibiting microglia activation. Front Pharmacol. 2017;8:789. 34. Cao H, Lu J, Du J, et al. TAK1 inhibition prevents the development of autoimmune diabetes in NOD mice. Sci Rep. 2015;5:14593. 35. Sasikala M, Ravikanth VV, Murali Manohar K, et al. Bach2 repression mediates Th17 cell induced inflammation and associates with clinical features of advanced disease in chronic pancreatitis. United European Gastroenterol J. 2018;6:272-282. 36. Akimoto T, Terada M, Shimizu A, et al. The influence of dietary restriction on the development of diabetes and pancreatitis in female WBN/Kob-fatty rats. Exp Anim. 2010;59:623-630. 37. Constantin MM, Nita IE, Olteanu R, et al. Significance and impact of dietary factors on systemic lupus erythematosus pathogenesis. Exp Ther Med. 2019;17:1085-1090. 38. Choi IY, Lee C, Longo VD. Nutrition and fasting mimicking diets in the prevention and treatment of autoimmune diseases and immunosenescence. Mol Cell Endocrinol. 2017;455:4-12. 39. Guan S, Lu J, Zhao Y, et al. TAK1 inhibitor 5Z-7-oxozeaenol sensitizes cervical cancer to doxorubicin-induced apoptosis. Oncotarget. 2017;8:33666-33675. How to cite this article: Jaster R, Gupta Y, Rohde S, et al. Impact of diet and genes on murine autoimmune pancreatitis. J Cell Mol Med. 2020;00:1–9. https://doi.org/10.1111/ jcmm.15540