Two overlapping bacterial artificial chromosome clones (BACs) were isolated that spanned the Bm-mif-1 locus. The inserts of BMBAC01L03 and BMBAC01P19 were 28,757 base pairs (bp) and 64,685 bp, respectively, with 10,637 bp of overlap, yielding a contiguated sequence of 82,805 bp (Figure 1). AT content overall was 68.0%; exonic DNA had an AT content of 59.9% and intergenic and intronic DNA had AT contents of 69.3% and 70.4% respectively. The average predicted gene size was 4.7 kb (range 0.6-20 kb). The average distance between genes was 3.1 kb (range 0.3-10.5 kb), giving an average gene density of one gene per 6.9 kb. There was an average of 9.3 introns per gene, with an average intron length of 316 bp (range 48-2,767 bp). The C. elegans orthologs of the B. malayi genes (see below) had a mean length of 3.2 kb, with an average of 5.5 introns per gene (mean size of 142 bp). The B. malayi genes were longer as a result of increased mean length and number of introns. Comparison to C. elegans presumed orthologs (see below) showed that only 50% of C. elegans introns were conserved in B. malayi (29 of 56 introns), and 25% of B. malayi introns (29 of 107) were conserved in C. elegans (Table 1). Of the 12 predicted B. malayi genes, seven were tested and confirmed by cDNA-PCR, and alternatively spliced transcripts were identified for four. Five of the 12 genes had corresponding ESTs (Table 1).

All 12 predicted genes had C. elegans homologs, but putative orthology could only be assigned to 11 pairs (Figure 1, Table 1). Orthology definition is possibly problematic, as the complete genome sequence of B. malayi is not known, and it is thus possible that genes more similar to these C. elegans comparators could be present. We note, however, that no B. malayi EST-defined genes (23,000 ESTs defining approximately 8,300 genes [16]) have better matches to these C. elegans proteins (data not shown), and that orthology definition included coextension of the proteins, and conservation of intron position and phase (Table 1). The exception, BMBAC01L03.3, contained two domains, an amino-terminal LON ATP-dependent serine protease domain (domain PF02190) and an anonymous carboxy-terminal domain (PFB022940). Proteins predicted from the Arabidopsis thaliana (AAC42255.1), Mus musculus (NP_067424), and Homo sapiens (XP_0421219) genomes share this architecture, but there are no C. elegans proteins that have both domains.

Some genes were similar to hypothetical, functionally uncharacterized genes from C. elegans. BMBAC01P19.7a/b had multiple predicted transmembrane segments also found in a number of peptides from other species (PFB002843) and were most similar to C36B1.12 (60% identity). There is only one homolog of BMBAC01P19.3a in any organism -F43G9.4 from C. elegans. The amino termini of both BMBAC01P19.3a and F43G9.4 contained PWWP domains (PF00855). PWWP domains are found in proteins with nuclear location and roles in cell growth and differentiation [21,22]. PSORT profiling indicated that BMBAC01P19.3 and F43G9.4 were likely to have nuclear localizations. The amino terminus of BMBAC01L03.4 contains a dnaJ-like domain (PF00684). The dnaJ domain is found in 41 C. elegans proteins, but BMBAC01L03.4 showed highest identity (57%) to F39B2.10. Both proteins had the dnaJ domain at their amino terminus and shared a common position of the first intron in this region. The remainder of the protein was not conserved.

BMBAC01P19.1 encodes Bm-mif-1 (Figure 2) [20]. Mammalian MIF is a cytokine involved in inflammation, growth, and differentiation of immune cells [23]: B. malayi MIF-1 may have a role in immunomodulation of the host [20,24]. C. elegans has four MIF-like genes: Ce-mif-1 (Y56A3A.3), Ce-mif-2 (C52E4.2), Ce-mif-3 (F13G3.9), and Ce-mif-4 (Y73B6BL.13). Transgenic reporter and immunolocalization studies suggest that C. elegans MIFs may have roles in development and the dauer stage [13,25]. Bm-MIF-1 has highest pairwise similarity to Ce-MIF-1 (41% compared to 23-29% for the other three paralogues; Figure 2) [20], and phylogenetic analysis of over seventy MIF-like proteins from eukaryotes confirms this assignment (D.B.G. and M.L.B., manuscript in preparation). Comparison of Bm-MIF-1 to the C. elegans MIFs, a second B. malayi MIF (Bm-MIF-2), and human MIF-1 (Figure 2) revealed that Bm-mif-1 and Ce-mif-1 shared two intron/exon boundaries also found in vertebrate MIFs. One of these introns was also present in Ce-mif-3, but Ce-mif-3 and the other two C. elegans mif genes shared a set of introns not present in the mif-1 genes. Bm-MIF-1 and other filarial MIF-1 homologs contain a CXXC motif (single-letter amino-acid code) critical for the thiol-oxidoreductase activities of vertebrate MIF [26]. None of the C. elegans MIF homologs contained this motif.

Two clusters of three genes in close proximity are conserved. The first involves BMBAC01L03.2, .3 and .5. The C. elegans orthologs of these genes are F43G9.5, F43G9.4, and F43G9.3 respectively. F43G9.5 and F43G9.3 are divergently transcribed from a 631 bp intergenic region. F43G9.3 is followed by F43G9.4 in the same transcriptional orientation with 501 bp separating the genes. In B. malayi this local synteny is conserved, except that two additional genes - BMBAC01L03.3 and .4 - are found between BMBAC01L03.2 and .5.

The second cluster also involves three genes. Proteins predicted from both alternative transcripts of BMBAC01P19.2 were found to be homologous to large proteins from Homo sapiens (BAF180, AAG34760 [27]), Gallus gallus (JC5056 [28]), D. melanogaster (CG11375, AAF56339), and C. elegans (C26C6.1) (Figure 3). These proteins shared six bromodomains (PF00439), two BAH domains (bromo-adjacent homology, PF01426), a HMG box (high mobility group, PF00505), and an anonymous carboxy-terminal domain (PFB007669). The B. malayi, C. elegans, and D. melanogaster polybromodomain (PBR) proteins also contain two C2H2 zinc fingers. PBR proteins may be involved in chromatin-remodeling complexes. Bromodomains interact with acetylated lysine in histone complexes, while HMG boxes are found in chromatin proteins that bind to single-stranded DNA and unwind double-stranded DNA. Human BAF180 has been shown to localize to the kinetochores of mitotic chromosomes [27]. None of the vertebrate PBR homologs contains zinc fingers, which may indicate additional functions for the nematode and fly proteins.

Two conserved genes were identified immediately upstream from pbr-1 (Figure 3). BMBAC01P19.5 (named Bm-ubr-1 (upstream of pbr-1)) showed significant similarity only to T28F4.4 from C. elegans (27% identity). The protein encoded by BMBAC01P19.4 is homologous to C. elegans T28F4.5 (30% identity). Iterative searches of GenBank using PSI-BLAST [29] indicated that BMBAC01P19.4 and T28F4.5 belong to a group of small peptides that include human DAP-1 (death-associated protein). DAP-1 is a nuclear protein and positive regulator of interferon gamma-induced apoptosis in HeLa cells [30]. PSORT profiling indicated that both nematode proteins may have a nuclear localization. BMBAC01P19.2 (Bm-pbr-1) and BMBAC01P19.5 (Bm-ubr-1) are divergently transcribed and BMABAC01P19.4 (Bm-dap-1) is found in the large third intron of BMBAC01P19.5 in the same transcriptional orientation as BMBAC01P19.2 (Figure 3). In the C. elegans instance of the PBR cluster, C26C6.1 (Ce-pbr-1) and T28F4.4 (Ce-ubr-1) are also divergently transcribed from a 1,233 bp intergenic region. The third gene, T28F4.5 (Ce-dap-1) is found in the large third intron of T28F4.4 on the same strand as C26C6.1.

Comparison of the intergenic and upstream regions of both clusters, and of the orthologous gene pairs, did not reveal any clear motifs that might be involved in transcriptional regulation. In particular, the intergenic DNA between pbr-1 and ubr-1, and the first intron of ubr-1, had less than 30% pairwise identity throughout, and there were no stretches of greater identity. The AT richness of the B. malayi genome compared to C. elegans may obscure any conserved elements. No RNA-coding genes were found. Two B. malayi ESTs matched at > 99.5% identity to two regions of BMBAC01P19 separated by 200 bp that were not predicted to be part of a transcript (see Figure 1). These regions are downstream of gene BMBAC01P19.3, and may derive from alternative 3' untranslated regions: the furthest downstream match includes a good polyadenylation site. The 3' end of the cDNA determined for this gene may have derived from internal priming from an A-rich segment of the 3' untranslated region.

All of the C. elegans orthologs, except for Y56A3A.3 (Ce-mif-1, 41% identity to Bm-mif-1, on chromosome III), are located on chromosome I (Figure 4). F13G3.9 (Ce-mif-3, 23% identity to Bm-mif-1) is found on C. elegans chromosome I in close proximity to the orthologs of B. malayi genes BMBAC01P19.2, .4, and .5. This could suggest that our orthology assignment is wrong. As described above, however, Ce-mif-1 and Bm-mif-1 share two intron positions and are more similar to each other than either is to Ce-mif-3, which has one concordant intron position, and one discordant intron position. The conflict between location and structure could be due to a gene-conversion event in either lineage, or an event of directed movement or insertion.

Eight of the 10 remaining C. elegans orthologs lay within a 2.3 Mb region in the center of chromosome I (6.7-9 Mb) (Figure 4). The orthologs of the other two genes (BMBACoLo3.4 and BMBAC01P19.6) are found at the distal tip of chromosome I. While there has been extensive rearrangement of gene order, when compared to the C. elegans orthologs, 10 of the B. malayi genes were in the same relative transcriptional orientation. Examination of the boundaries of the C. elegans cluster and individual gene regions did not show any association with repeat-sequence classes, including those shown to be commonly associated with rearrangements between C. elegans and C. briggsae [6].

To ascertain whether the segment sequenced was representative of the relationship between the B. malayi genome and that of C. elegans, we surveyed the B. malayi BAC-end derived genome survey sequences (GSSs; J. Daub, C. Whitton, N.H., M. Quail and M.L.B., unpublished observations). There are over 18,000 GSSs from B. malayi, derived from three independent libraries. Each BAC-end sequence was compared to the C. elegans proteome (Wormpep [31]) and significant similarities recorded (BLASTX probabilities < e-8). The chromosomal position of each matching C. elegans protein was derived from Wormbase [32]. One hundred and sixty-four BACs had matches at both ends to C. elegans proteins under these conditions (summarized in Table 2, details in Table 3). We note that these matches are not necessarily to orthologs, as we have not carried out intensive analysis of each one, but random selection of genes should not yield greater linkage estimation despite the problem of gene families and domain matches. While much of the C. elegans proteome consists of protein families, very few of these have a chromosomally restricted distribution [33,34].