Familial Mediterranean fever (FMF) is an autosomal recessive disorder characterized by episodes of fever with serositis and/or synovitis, and a massive influx of granulocytes into the affected anatomic compartments. Some patients also develop systemic AA amyloidosis, which can lead to renal failure and death. FMF is frequent among non-Ashkenazi Jews, Armenians, Arabs, and Turks. Our laboratory set out several years ago to identify the gene causing FMF (designated MEFV) by positional cloning. In 1992 we mapped MEFV to chromosome 16p13.3; in 1997 we identified the FMF gene itself. MEFV is comprised of 10 exons spanning about 14 kb of genomic DNA; it encodes a 3.7 kb transcript observed predominantly in granulocytes. Pyrin, the 781 aa predicted protein product, is homologous to a number of transcription factors and nucleic acid- binding proteins. Pyrin shares with some of these proteins an approximately 400 aa tripartite motif consisting of a B-box zinc finger, an alpha-helical (coiled-coil) domain, and a domain of unknown function designated B30.2. Pyrin also has a bZIP basic DNA-binding motif and a Robbins-Dingwall nuclear targeting signal. We originally described 3 conservative missense mutations (M680I, M694V, and V726A) in the B30.2 domain that were present in FMF patients, but not in a large panel of normal chromosomes. Microsatellite and single nucleotide polymorphism (SNP) haplotype analysis demonstrated ancestral relationships among populations that have been separated for many centuries, suggesting that these mutations are very old. During the 1998 reporting period, we have focused on 5 areas: a) assessing the spectrum of MEFV mutations and associated clinical phenotypes; b) studies of MEFV expression in leukocyte development and activation; c) development of antibodies to facilitate direct studies of the pyrin protein; d) cloning the murine homolog of MEFV and establishing a knockout mouse model of FMF; and e) initiating therapeutic trials for patients not responding fully to colchicine. Progress in each of these areas is summarized below. Mutational analysis: A large percentage of the carrier chromosomes in our initial panel of non-Ashkenazi Jewish and Arab families were attributable to the 3 aforementioned mutations. We subsequently focused our sequencing efforts on the small number of patients in whom we had not identified mutations, and an ethnically much more diverse cohort of 274 new referrals. Earlier this year we reported 4 novel mutations (E148Q, K695R, A744S, and R761H), and we have subsequently submitted a manuscript with 2 additional novel mutations (P369S and a second nucleotide substitution causing M680I). Altogether, there are now 16 known MEFV mutations in FMF; 14 are missense mutations and the other 2 are short in-frame deletions. Eleven fall within the B30.2 domain, while one each is within or near the bZIP basic domain, B-box zinc finger, and coiled-coil domain. Among the mutations identified after the cloning of MEFV, E148Q is the most common, occuring in several ethnic groups. SNP haplotype analysis strongly suggests a common ancestral origin among E148Q chromosomes. Our studies revealed a higher than expected number of mutation-positive patients of Ashkenazi Jewish and Italian ancestry, underscoring the need for clinical awareness in these populations. A survey of DNAs from approximately 200 anonymous but asymptomatic Ashkenazi Jewish subjects from New York demonstrated a carrier frequency of 21% in this population. MEFV mutations, particularly E148Q, P369S, and K695R, are probably not fully penetrant among the Ashkenazim. Our work also provides important data on the sensitivity and specificity of genetic testing for FMF in the U.S. MEFV Expression Studies: Northern blot and RT-PCR have confirmed that MEFV is expressed in peripheral blood neutrophils but not lymphocytes. We have also found message by RT-PCR in peripheral blood eosinophils. We have observed considerable variation in MEFV expression in monocytes, with some normal subjects exhibiting high levels. Normal human bone marrow, either by in situ hybridization or RT-PCR, exhibits MEFV message. The myelomonocytic HL60 cell line showed no expression at baseline, but MEFV message was induced when cells were stimulated by retinoic acid or alkaline pH to differentiate towards neutrophils or eosinophils, respectively. Treatment with PMA, which drives HL60 cells towards monocytic differentiation, was associated with low-level MEFV expression. Cultures of CD34+ human pleuripotent progenitors with cytokines that induce myeloid differentiation was associated with message induction at relatively early stages. RNase protection assays demonstrated marked increases in MEFV expression in monocytes within 1 hour of stimulation with LPS and the Th1 cytokine interferon gamma. These effects were blocked by the Th2 cytokine IL4. Expression studies with stimulated granulocytes are in progress. Antibodies against pyrin: Polyclonal rabbit antibodies have been raised against several pyrin peptides. One of these, directed against residues 435-452, identifies a band of the predicted size (about 85 kDa) on Western blots from in vitro transcribed and translated MEFV, and from a cell line transfected with an MEFV expression construct. This antibody also detects an 85 kDa band in lysates from neutrophils and activated monocytes. The activity of this serum for monocyte (but not granulocyte) lysates can be adsorbed with recombinant pyrin protein. Microsequencing is in progress to confirm the identity of the granulocyte band. Mouse model: Using PCR primers derived from human MEFV, we have cloned the mouse homolog. The exon-intron structure of mouse Mefv is very similar to the human gene. There are several splice variants in the mouse, but the predominant form encodes a predicted protein of 797 aa. Like human pyrin, the predicted mouse protein has a Robbins-Dingwall nuclear targeting sequence, a bZIP basic domain, a B-box zinc finger, and an alpha-helical domain. Due to a frame shift early in exon 10, the B30.2 domain is truncated. At the protein level, there is 47.6% identity and 65.5% similarity between human and mouse proteins.
Showing the most recent 10 out of 35 publications
