Parent-of-origin effects implicate epigenetic regulation of experimental autoimmune encephalomyelitis and identify imprinted Dlk1 as a novel risk gene

Abstract

Parent-of-origin effects comprise a range of genetic and epigenetic mechanisms of inheritance. Recently, detection of such effects implicated epigenetic mechanisms in the etiology of multiple sclerosis (MS), a chronic inflammatory disease of the central nervous system. We here sought to dissect the magnitude and the type of parent-of-origin effects in the pathogenesis of experimental neuroinflammation under controlled environmental conditions. We investigated inheritance of an MS-like disease in rat, experimental autoimmune encephalomyelitis (EAE), using a backcross strategy designed to identify the parental origin of disease-predisposing alleles. A striking 37-54% of all detected disease-predisposing loci depended on parental transmission. Additionally, the Y chromosome from the susceptible strain contributed to disease susceptibility. Accounting for parent-of-origin enabled more powerful and precise identification of novel risk factors and increased the disease variance explained by the identified factors by 2-4-fold. The majority of loci displayed an imprinting-like pattern whereby a gene expressed only from the maternal or paternal copy exerts an effect. In particular, a locus on chromosome 6 comprises a well-known cluster of imprinted genes including the paternally expressed Dlk1, an atypical Notch ligand. Disease-predisposing alleles at the locus conferred lower Dlk1 expression in rats and, together with data from transgenic overexpressing Dlk1 mice, demonstrate that reduced Dlk1 drives more severe disease and modulates adaptive immune reactions in EAE. Our findings suggest a significant epigenetic contribution to the etiology of EAE. Incorporating these effects enables more powerful and precise identification of novel risk factors with diagnostic and prognostic implications for complex disease.

Figure 1. Backcross design.

Schematic illustration of the experimental set-up used to breed reciprocal crosses that enabled identification of parent-or-origin effects. EAE-susceptible DA and -resistant PVG strains were backcrossed to heterozygous F1 (DAxPVG) hybrids to generate two experimental populations, DABC and PVGBC, respectively. Within each population, two reciprocal crosses were established by breeding F1 hybrids to DA mothers (DAxF1) or fathers (F1xDA) in DABC and PVG mothers (PVGxF1) or fathers (F1xPVG) in PVGBC. Transmission of autosomal chromosomes (one pair of autosomes is represented by vertical lines) and mitochondria (represented by circle) is shown in the upper panel. Transmission of sex chromosomes is shown in the lower panel. DA and PVG alleles are indicated with D and P, respectively, and the maternally and paternally inherited alleles are indicted with m and p, respectively.

Figure 2. Accounting for parent-of-origin effects improves disease inheritance models.

Phenotypic variance (%) that can be explained by risk loci is greatly increased when parent-of-origin effects are included, supporting their contribution to EAE. The fit multiple-QTL model was used to validate the independent effects of each QTL (phenotype  =  QTL1 + QTL2 + … + QTLn +ε) (Table S1, Table S6). A) Mean variance of all phenotypes +/− SEM is presented for the entire DABC and as a sum of variance for each reciprocal cross (DAxF1) + (F1xDA). Due to the high severity in DABC males, the severity phenotypes are omitted. B) Mean variance of all phenotypes +/− SEM is presented for the entire PVGBC and as a sum of variance for each reciprocal cross (PVGxF1) + (F1xPVG).*, ** and *** indicate p<0.05, 0.01 and 0,001, respectively.

Figure 3. Maternal transmission of the disease-predisposing allele.

A) A logarithm of the odds (LOD) plot of chromosome 4 shows linkage for disease onset in F1xPVG offspring (peak at 144 Mb indicated by *) but not in PVGxF1 offspring, with the maternally transmitted DA allele predisposing for earlier onset of disease compared to the PVG allele. B) The same QTL at 144 Mb regulated onset of disease in DABC in F1xDA offspring but not in DAxF1 offspring, showing maternal transmission of the disease-predisposing DA allele in the independent DABC population (peak at 144 Mb indicated by *). An additional QTL that did not show parent-of-origin dependent effect was identified in DABC population (peak at 185 Mb indicated by **). Onset of EAE is shown as representative of all clinical phenotypes. LOD scores were generated using Haley-Knott regression (genome-wide p<0.05 thresholds for significant linkage were 2.6, 2.6, 2.8, 2.9, 2.8 and 2.8 for PVGBC (N = 471), PVGxF1 (N = 225), F1xPVG (N = 246), DABC females (N = 213), DAxF1 females (N = 105) and F1xDA females (N = 108), respectively). P-values given in allelic effect plots were calculated using Student's t-test (p-value <0.001  =  ***, ns  =  not significant). DA and PVG alleles are indicated with D and P, respectively, and the maternally and paternally inherited alleles are indicted with m and p, respectively.

Figure 4. Paternal transmission of the disease-predisposing allele.

A) A logarithm of the odds (LOD) plot of chromosome 6 shows linkage for disease onset in PVGxF1 offspring (peak at 131 Mb indicated by *) but not in all PVGBC or F1xPVG offspring, with the paternally transmitted PVG allele predisposing for earlier onset of disease compared to the DA allele. B) The same QTL at 131 Mb displayed evidence for linkage in DABC females in DAxF1 offspring but not in F1xDA offspring, showing paternal transmission of the disease-predisposing PVG allele in the independent DABC population. Onset of EAE is shown as representative of all clinical phenotypes. LOD scores were generated using Haley-Knott regression (genome-wide p<0.05 thresholds for significant linkage were 2.7, 2.8 and 2.7 for combined PVGBC (N = 471), PVGxF1 (N = 225) and F1xPVG (N = 246), respectively; nominal p<0.05 threshold for evidence of linkage was 1.3 in DAxF1 (N = 105)). P-values given in allelic effect plots were calculated using Student's t-test (p-value <0.05  =  *, p-value <0.001  =  ***, ns  =  not significant). DA and PVG alleles are indicated with D and P, respectively, and the maternally and paternally inherited alleles are indicted with m and p, respectively.

Figure 5. Paternally transmitted PVG allele at the Dlk1 locus predisposes for lower Dlk1 expression.

Dlk1, Rtl1 and Dio3 mRNA expression in spleen tissue collected A) 21 days after EAE induction in PVGBC population (N = 372), B) 35 days after EAE induction in DABC population (N = 347), and in C) three weeks old naïve DA, PVG and reciprocal F1 hybrid rats (N = 8–9 rats/group). A and B) Rats stratified according to the genotype at the peak of the QTL on chromosome 6 (131 Mb) showed that the paternally inherited PVG allele predisposes for lower expression of Dlk1 compared to the DA allele, whereas no effect was observed on Rtl1 and Dio3 expression. C) Likewise, F1 hybrids with paternally inherited PVG alleles demonstrated lower Dlk1 expression compared to DA alleles, whereas no effect was observed on Rtl1 and Dio3 expression. Relative Dlk1 expression was calculated in relation to the mean of housekeeping gene, hypoxanthine phosphoribosyltransferase (Hprt), using 2-ΔΔCt method and normalized within each group. P-values were calculated using Student's t-test (p-value <0.05  =  *). DA and PVG alleles are indicated with D and P, respectively, and the maternally and paternally inherited alleles are indicted with m and p, respectively.

Figure 6. Transgenic overexpression of Dlk1 modulates EAE severity and adaptive immune responses.

A) Dlk1 mRNA expression in immune tissues in Dlk1 transgenic and wild type littermate control mice measured using TaqMan PCR. Relative Dlk1 expression was calculated in relation to the mean of housekeeping gene, beta-2 microglobulin, using 2-ΔΔCt method (N = 3–5 mice/group). B) Daily EAE scores in Dlk1 transgenic (N = 24) and wild type littermate control (N = 20) mice after immunization with MOG demonstrate that the higher levels of Dlk1 in transgenic mice are protective against EAE. Mean EAE clinical score (and the standard error of the mean) is given for each day post immunization (p.i.). The figure represents pooled data from three separate experiments. Mann-Whitney U-test was used to compare the daily EAE scores (p-value <0.05  =  *). C) Flow cytometry analysis of percentages of activated T helper cells (CD3+CD4+CD44+) and B cells (CD3-CD45R+) in spleen tissue 25 days after EAE induction, ex vivo and after 48 hour of restimulation with MOG (MOG recall). Transgenic Dlk1 mice displayed lower percentage of activated T cells and B cells compared to wild type littermate controls (N = 7/group). D) Flow cytometry analysis of numbers of differentiated IFN-γ producing Th1 cells (details in Materials and Methods). Transgenic Dlk1 mice showed a lower number of Th1 cells in comparison to wild type littermates (N = 2/group, three independent experiments). Green and orange histograms show IFN-γ producing cell population in control and fully Th1 differentiated conditions, respectively.