In the zebrafish gene, as in the mouse, JAK inhibitor zMsi1 has two alternative exons as confirmed by sequencing results from several clones amplified using two sets of cloning primers ( Fig. 1A). The full length putative protein sequences are highly conserved with other species including mouse (83%) and human (84%). The extent of sequence conservation is high throughout the length of the protein, with the two RRM domains (black bars) exhibiting

an even higher degree of sequence identity with mouse (RRM1; 89%, RRM2; 94%) and human (RRM1; 89%, RRM2; 95%). From the results of a BLAST search of the zebrafish genomic sequence using the zebrafish Msi1 cDNA as a query, the Msi1 gene was found to have 14 exons on chromosome 8, and three putative zebrafish Msi1 transcripts were identified ( Fig. 1B). The shorter form of zebrafish Msi1 (zMsi1S) skips exons 4 and 11 (in red), producing a 2303 nucleotide sequence that is translated into a 330 amino acid polypeptide with Bleomycin a predicted molecular weight of 35.9 kDa. The longer form of Msi1 (zMsi1L) skips exon 4 (in red), producing a 2360 nucleotide sequence that is translated into 349 amino acids with a predicted molecular weight of 38.0 kDa. Thus, zMsi1L is 19 amino acids longer than zMsi1S. Approximately half of the zMsi1 cDNA is zMsi1L (six out of 13 clones) and the remainder is primarily zMsi1S (five out of 13 clones). A minor splicing variant is the zMsi1L + 35 bp

clone, which contains exon 4 and produces a 129 amino acid polypeptide caused by a premature stop codon in exon 6 following the inclusion of exon 4 ( Fig. 1C and Supplementary Fig. 1). In database searches, only the transcript variant for zMsi1S was detected. The nucleotide sequence of the zMsi1 cDNA was compared with its orthologs in human, mouse, rat, chick, Xenopus, C. elegans, ascidian (H. roretzi and C. intestinalis) and Drosophila. The protein sequences of Msi1 and Msi2 from eight species were multiply aligned using the UPGMA method with the Genetyx software. A phylogenetic tree was inferred Lck by neighbor-joining from a gapped alignment and the values on the

tree nodes are neighbor-joining bootstrap values. In addition to homology to the mouse and human, the phylogenetic analysis of zebrafish Msi1 revealed high sequence homology across an array of species ( Fig. 2) especially for RRMs ( Table 1). These results suggest that Msi family genes first diverged with branching from vertebrate and invertebrate lineages, and that the branching between mammalian and teleost Msi1 happened after the segregation of the Msi1 and Msi2 genes. Comparing the genomic sequence of Musashi family genes with those of several other species, the exon–intron structures of the genes were found to be mostly preserved between the human, mouse and zebrafish (Fig. 3). The human MSI1 consists of 15 exons spanning a 28-kb genomic region on chromosome 12. The mouse Msi1 also consists of 15 exons spanning a 25-kb genomic region on chromosome 5.