We identified Adam33 during the positional cloning of the mouse mahogany mutation, since Adam33 lies on the same bacterial artificial chromosome (BAC) that contains Attractin (Atrn), the gene mutated in mahogany16. This BAC, CITB 389B9, also contains Gfra4, Sialoadhesin (Sn), Cdc25b, Cenpb, and 7 clusters of expressed sequence tags (ESTs), represented by IMAGE clones 386736,388220,775311,614556,427645, 807697 and 514017.

The full-length Adam33 cDNA sequence was predicted using Genscan, GRAIL, and sequence data from EST clones, to identify exons within a ~48 kb contig of genomic DNA (Genbank AF155960). The predicted cDNA sequence was confirmed by sequencing RT-PCR products amplified from adult brain RNA. Adam33 lies centromere-distal to Atrn and Gfra4 and centromere-proximal to Sn and Cdc25b (Figure 1A). Beginning with the putative translational initiation site, the Adam33 cDNA consists of 22 exons spanning 12671 bp of genomic DNA and corresponds to a 2771 bp cDNA of which 2391 bp are protein-coding (Figure 1B, 2). The predicted 797 amino acid protein is most closely related to Xenopus laevis X-ADAM13 (44% identity, 58% similarity) and X-MDC13 (43% identity, 57% similarity), and to mouse ADAM12 (38% identity, 51% similarity) and ADAM19 (39 % identity, 53% similarity).

The hallmark motifs of an ADAM protein include a signal sequence, pro-domain, metalloprotease, disintegrin-like, and cysteine-rich domains, an EGF-like repeat, a single transmembrane domain, and a cytoplasmic tail, which sometimes contains proline-rich regions implicated in SH3 binding. The mouse ADAM33 protein contains all of these motifs (Figure 2) and based on sequence, the metalloprotease domain of mouse ADAM33 is likely to be functional. The carboxy-terminal intracellular region contains one proline-rich region but no consensus sites for SH3-binding domains, as found in Xenopus ADAM 13.

Northern blot hybridization using a probe corresponding to the end of the Adam33 coding region and the beginning of the 3'UTR detects a 2.7 kb RNA in adult brain, heart, kidney, and lung, and a 2 kb RNA in testis (Figure 3A). The size of the 2.7 kb RNA corresponds exactly to the prediction based on sequence, but the molecular basis for the testis-specific 2 kb transcript is not known (attempts to amplify the cDNA from testis using primers in exon 1 and the 3'UTR were unsuccessful). Using a more sensitive assay for expression based on semi-quantitative RT-PCR revealed expression in all samples examined (Figure 3B), including total RNA from mouse embryos on days 8.5, 9.5, 12.5 and 19 of development (Figure 3B).

The expression pattern of Adam33 in the adult brain was also examined by in situ hybridization of coronal brain sections (Figure 4). A striking pattern of hybridization was observed in the granule later of the dentate gyrus, the pyramidal layer of the hippocampus (Figure 4B), and the granule layer of the cerebellum (Figure 4E). The expression of Adam33 in two of the primary locations of granule neurons in the brain is intriguing and it will be interesting to follow expression of this gene during neuronal migration; while Xenopus ADAM13 has been implicated in neural crest cell migration during embryonic development 18, perhaps ADAM33 plays a role in the migration of granule neurons in the mouse brain. In Drosophila, the ADAM protein kuzbanian (orthologous to mammalian ADAM 10) is required for proteolytic cleavage of notch and delta, associates with a number of molecules that have been implicated in axon guidance and migration, and has been shown to play a role in axon guidance 6192021.

The human ortholog of mouse Adam33 was identified by comparing the mouse cDNA and predicted protein to genomic sequence in the public genome database at http://www.ensembl.org and in high throughput human genomic DNA sequence (Genbank AC055771, AL356755, AL109804, AC017113; ^*see note). Gene order and orientation are conserved between this region of mouse chromosome 2 and human Chromosome 20p13, and human ADAM33 has the same transcriptional orientation as its closest neighbors, SN, which lies immediately upstream, and GFRA4, which lies immediately downstream. A bona fide human cDNA for ADAM33 has not been isolated (see below), and only some of the exons have been annotated correctly in human genomic sequence. However, comparison of human genomic sequence to the mouse Adam33 cDNA reveals an identical exon-intron structure and a putative 814 aa protein with 71% identity and 78% similarity to mouse ADAM33 and 37% identity and 61% similarity to X-ADAM13 (Figure 1C, Figure 2).

The ADAM33 designation was originally assigned to an ~3000 bp testis cDNA clone, DKFZp434K0521, whose full-length sequence (AL117415) was determined as part of a large-scale genome project. Comparison to the mouse cDNA and to human genomic DNA reveals that the testis cDNA clone represents residues 300–813 (with one frameshift) of our predicted human ADAM33 protein.

The most closely related homolog of mouse and human ADAM33 is X-ADAM13, which encodes a 3200 bp cDNA with an ORF of 915 amino acids 15. The overall level of similarity between mouse or human ADAM33 and X-ADAM13 is lower than that of other mammalian and Xenopus ADAM orthologs (Figure 1C). In addition, there are no putative SH3 binding sites in the cytoplasmic tail of the mouse or human proteins, whereas X-ADAM13 contains several. Like the mammalian ADAM33 proteins, X-ADAM13 has all the hallmark motifs of an ADAM protein, including potential metalloprotease, disintegrin and signaling activities. It is expressed during embryogenesis in somitic mesoderm and neural crest cells, suggesting a likely role in embryonic development. Mouse and human ADAM33 are also highly similar to X-MDC13, another Xenopus ADAM protein that is highly homologous to X-ADAM13 and is expressed in embryos and adult testes 22. Characterization of the Adam33 embryonic expression pattern may provide additional insight into cross-species relationships, however, it should be noted that much of the work on X-ADAM13 to date has focused on its putative SH3 binding sites to identify interacting proteins 23, and the absence of these sites in mouse and human ADAM33 proteins indicates that such interactions are likely not conserved in mammals. Recent work does indicate that the metalloprotease domain of X-ADAM13 is active and that this protein is required for normal cranial neural crest cell migration 18. As the metalloprotease domain of ADAM33 also matches the consensus sequence for an active metalloprotease, further study of this protein's potential catalytic activity and identification of its target proteins will be important in determining its biological role. In addition, availability of genomic sequence surrounding mouse Adam33 will facilitate cross-species comparisons to identify conserved regulatory sequences.