Investigation of the possible effect of intragenic MEFV gene CpG island methylation on mRNA transcription and pyrin localization

Familial Mediterranean Fever (FMF) (OMIM no: 249100), is an inflammatory disease with recurrent fever attacks, prevalent in populations of Mediterranean descent. MEditerranean FeVer (MEFV) gene was identified as the gene of FMF by both International and French FMF Consortia. Up to 315 variations were defined to date, five of which (c.2080 A>G, p.Met694Val; c.2082 G>A, pMet694Ile; c.2177 T>C, p.Val726Ala; c.2040 G>C, p.Met680Ile and c.442 G>C, pGlu168Gln) were reported to be recessively inherited and cause FMF. On the other hand, 5 to 20% of patients, where FMF is not prevalent, were reported to carry no variations on their MEFV gene, suggesting that MEFV variations may not be the only cause of FMF. Recently, MHC Class I Polypeptide-Related Sequence A (MICA) gene was proposed as a modifier gene which may have an epistatic interaction with MEFV, however no significant results were obtained. Therefore, disruptions leading to changes in MEFV expression, independent of gene variations were thought to possibly play a role in FMF pathogenesis. MEFV is mainly expressed in white blood cells, as well as in synovial fibroblasts and dendritic cells. Studies have shown that MEFV expression in FMF patients were lower than healthy controls, being even lower during attack periods, and suggested that this decrease was due to inflammation. Previously, we observed expression levels in leukocytes isolated from 51 FMF patients and 11 healthy controls, with low MEFV expression in patients compared to healthy controls (p=0.031). On the other hand, another study group reported higher MEFV expression levels in FMF patients' peripheral blood mononuclear cells (PBMC) when compared with healthy controls' samples. Epigenetic mechanisms are one of many mechanisms that modulate gene expression. MEFV gene contains a 998 bp CpG island covering its whole second exon (NC_000016.10, 3254057 – 3255054). We compared our MEFV expression data with methylation levels of MEFV gene's CpG island in our study group, and observed a negative correlation between methylation and expression levels in both groups (cor=-0.29, P=0.041), of which the patient group's have been higher (cor=-0.36, P=0.035). Alternative spliced forms of MEFV mRNA were shown in previous studies. MEFV-fl is coding a full-length protein, whereas MEFV-d2 codes a protein lacking the second exon, with MEFV-8ext, MEFV-4a and MEFV-2a coding proteins containing intron fragments. As we did not assess different transcripts in our previous expression studies, we analyzed the expression levels of most common transcripts, MEFV-fl and MEFV-d2 in our study group. We found that MEFV-d2 was highly expressed in patients, while it is almost inexistent in the control group (p=0.026). Increase in MEFV-d2 expression proportional to increase in MEFV levels urged us to think that MEFV gene's CpG island methylation might play a role in the splicing of its second exon. Initially, we investigated possible MEFV variations that might lead to alternative splicing. Previously c.910G>A variant (rs75977701) was suggested to cause the splicing of MEFV gene's second exon. However, this rare variant was not detected in our previous FMF cohort, and was also absent in databases, like Infevers, where FMF related variations are submitted; eliminating it as a potential candidate. We first generated an in vitro model to test the possible role of methylation on MEFV gene in alternative splicing of its second exon, and showed that it leads to the splicing of the gene construct containing the whole second exon, using pSpliceExpress gene cassette. Then, we constructed different cell culture models to analyze the cells' endogenous MEFV expression under different conditions like cell differentiation, methylation, demethylation, activation and prevention of histone deacetylation. To this aim, HL-60 promyelocytes were subjected to dimethylsulphoxide (DMSO) for differentiation into neutrophil-like cells, methanol to increase global methylation, 5-aza-2′deoxycytidine (deazacytidine) to decrease global methylation, phorbol mystrate acetate (PMA) and lipopolysaccharide (LPS) for activation and trichostatin A (TSA) for prevention of histone deacetylation. We analyzed the expression levels of MEFV transcripts containing lacking the second exon in these cell models, compared to untreated controls, and observed that the transcript lacking the second exon was highly expressed in neutrophil-like cells (2 fold increase, p=0,0005), activated cells (2,5 fold increase, p=0,0034) and methylated cells (2,5 fold increase, p< 0,0001), and less expressed in demethylated cells (1,7 fold decrease, p=0,0126). We did not observe any significant expression variation in MEFV transcripts in cells treated with TSA, suggesting that histone acetylation do not have an effect on alternative splicing of MEFV gene's second exon, at least in our cell models. We also analyzed the methylation levels of first 61 CpGs of MEFV gene's CpG island in these cell culture models. Although one-way ANOVA test results demonstrated that we obtained significant results (Kruskal-Wallis test, p= 0,0184, Kruskal-Wallis statistic=9,567), individual t-tests performed between different models showed no significant differences possibly due to high variation among biological repeats. Similarity between methylation and expression levels observed in neutrophil-like and methylated cells suggested that methylation might play a role in the differentiation of the cells to neutrophil-like cells. We analyzed if methylated cells differentiate into neutrophil-like cells and found no sign of differentiation, suggesting that methylation might not be the cause but may be the consequence of differentiation. Recent studies have shown that methylation might play a role in alternative splicing through two mechanisms: by modulating the elongation rate of RNA Polymerase II, or by recruiting splicing factors to weak exons leading to their recognition by transcription machinery. Several proteins were defined to play a role in these two mechanisms. We incubated methylated and non-methylated gene fragments containing MEFV gene's second exon with nuclear proteins and separated proteins that bind the fragments. Differential proteins were assessed through SDS-PAGE analysis and sent to mass spectroscopy (MS) analysis. Although several proteins were found to bind non-methylated fragments and not to methylated ones, we could not obtain high scores in MS analyses. Two potential zinc-finger proteins were identified and should be further analyzed, along with other potential proteins using chromatin immunoprecipitation type analyses. Many diseases, like neurological disorders and cancer, have been linked to changes in protein subcellular localization. The protein coded by MEFV-fl transcript, pyrin-fl, was reported to be generally in the cytoplasm of various cells, and pyrin-d2 coded by MEFV-d2 transcript was found to shuttle between nucleus and cytoplasm. Therefore, we analyzed the localization of these two isoforms in neutrophil-like and activated cells and compared with untreated HL-60 cells. We observed pyrin-fl in the nucleus of all cells. On the other hand, pyrin-d2 was observed in the nucleus of HL-60 cells and in the cytoplasm of neutrophil-like and activated cells. There is still a debate on whether pyrin is an anti-inflammatory or a pro-inflammatory protein. Studies demonstrated that Apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC) protein interacts with pyrin to play a role in the development of inflammation. Furthermore, ASC can also interact with pyrin-d2, which prevents its translocation to the nucleus under inflammatory conditions. However, pyrin-d2 is unable to bind to 14.3.3 proteins that were shown to suppress inflammation, due to the lack of the residues needed for this interaction. We propose a new mechanism, which might lead to FMF formation in the light of the results presented hereby, combined with literature findings. In this mechanism, high methylation levels lead to an increase in the expression of pyrin-d2, which is kept in the cytoplasm by ASC, and cannot be blocked by 14.3.3 proteins, thus leading to inflammation. This model could present a new mechanism leading to FMF disease, independent of gene variations.