Genomics and biology of the nematode Caenorhabditis briggsae^*

A hybrid approach combining whole-genome assembly with whole-genome shotgun sequencing (WGS) was adopted to sequence the C. briggsae genome (Stein et al., 2003). The C. briggsae strain AF16 was targeted for physical map construction (Marra et al., 1997) and whole genome sequencing. A total number of 2.068 million reads, representing an approximately 10-fold coverage of the entire genome, were assembled into 5,341 contigs and linked together into 889 gapped supercontigs containing 105.6 Mbp of DNA sequence. This assembly (named cb25.agp8) consists of 463 sequence superconfigs placed into 142 FPCs and covers 102,431,873 bp and is publicly available at the WormBase FTP site. This assembly is estimated to cover 98% of the entire C. briggsae genome with a sequencing accuracy of approximately 99.98%. Therefore the C. briggsae genome size is approximately 104 Mbp, a size similar to but slightly larger than that of the C. elegans genome (102 Mbp, according to WormBase WS160). Recently, the construction of a genetic map using single nucleotide polymorphic (SNP) markers has positioned 100 Mb along the six chromosomes and refined the sequence map (Hillier et al., 2005).

An interesting difference between the genomes of C. briggsae and C. elegans, which was revealed by comparative genomic analysis, is that the difference in the genome size is mainly due to the larger non-genic repetitive DNA content in C. briggsae's genome, which accounts for 22.4% (or 23.3 Mbp) of the C. briggsae genome compared to 16.5% (or 16.5 Mbp) of the C. elegans genome (see Table 1). Notably, a single C. briggsae repeat family, Cb000047, is present more than 20,000 times and covers 4.7 Mbp of the C. briggsae genome, while it is present only a few hundred times in the C. elegans genome. Despite the striking similarities between the genomes and coding sequences of C. briggsae and C. elegans as summarized in Table 1, only a few orthologous repetitive elements were identified between these two sister species, suggesting that essentially all extant dispersed repeat elements postdate the divergence of these two species.

Before a detailed description of the comparative analysis between C. briggsae and C. elegans protein-coding genes are made, some caveats should be pointed out. First, while the C. briggsae genome is still at a draft stage, the C. elegans genome sequencing is complete. Second, while the C. briggsae gene set was generated entirely by computational gene finding, the C. elegans gene set was generated by computational gene finding, followed by extensive manual curation with various experimental evidences including ESTs, OSTs, and full length cDNAs. Third, while only a small number of laboratories use C. briggsae as a major experimental organism, a significantly larger number of laboratories use C. elegans, which not only provides more data for accurately curating C. elegans gene modes, but also creates a strong pressure to establish a high-quality C. elegans gene set. In fact, the C. briggsae gene set (the "hybrid" gene set) remained the same for three years following the publication of the C. briggsae genome analysis paper (Stein et al., 2003) until recently when WormBase started to curate C. briggsae genes (WormBase Consortium, personal communication). In the meantime, the C. elegans gene set has been improved and updated in WormBase in each of its tri-weekly releases.

As shown in Table 1, there are similar numbers of protein-coding genes in the genomes of C. briggsae and C. elegans. The unspliced lengths of protein-coding genes and the total length of the protein-coding genes, including the length of their introns, are roughly the same between the two species, further suggesting that the larger size of the C. briggsae genome is not contributed by the number or size of the protein-coding genes. The C. briggsae genome has contributed to the improved predictions of C. elegans protein-coding genes. Notably, C. briggsae gene prediction has helped to predict 1,275 new C. elegans genes (Stein et al., 2003). So far, about 300 genes of these have been confirmed by OSTs of the ORFeome project. Despite the similarity of the number of genes shared by the two species, C. briggsae has about 800 genes that have not be found in C. elegans, while C. elegans has 1,061 genes that have not been found in C. briggsae (Stein et al., 2003).

One way C. briggsae and C. elegans orthologs were identified was by searching for C. briggsae/C. elegans pairs that were each other's best BLASTP (Altschul et al., 1997) match in the opposite species. Synteny was subsequently used as a criterion to search for more orthologous pairs. These combined approaches produced altogether 12,155 ortholog pairs, or 62% of the C. briggsae gene set or 65% of the C. elegans gene set (Stein et al., 2003). After the publication of the initial C. briggsae genome analysis paper, a recently developed program InParanoid (O'Brien et al., 2005) was applied to identify orthologs between C. briggsae and C. elegans. The InParanoid database (April 2005 release), using a recent C. elegans gene set, lists 12,858 C. briggsae and C. elegans ortholog pairs. Therefore the numbers of orthologs obtained by these two different approaches agree well with each other. However, the number of identified orthologs will change over time as the genome assembly and gene annotation, especially for C. briggsae, improve.

Currently the C. elegans ortholog of a C. briggsae gene is displayed in the Homology section on its gene page in WormBase and vice versa. Users can follow the link to the gene page of the C. elegans ortholog of a C. briggsae gene. Right next to this ortholog link, there is another link termed “syntenic alignment”, which directs users to a corresponding “Synteny Browser” page (see Figure 1). The Synteny Browser provides a comparative view of two genomic regions, a C. elegans genomic region that contains the C. elegan ortholog of the C. briggsae gene, a C. briggsae genomic region that is syntenic to the C. elegans region. The C. briggsae genomic region contains the C. briggsae ortholog of the C. elegans gene (see Figure 1). Users can browse along the genomic sequences by modifying the genomic coordinates.

Before the C. briggsae genome sequencing project, the divergence of C. elegans and C. briggsae had been estimated to be 20–120 MYA (million years ago; Coghlan and Wolfe, 2002; Heschl and Baillie, 1990; Kennedy et al., 1993; Lee et al., 1992; Prasad and Baillie, 1989). The C. briggsae genome sequencing project helped to refine the divergence date to 80–110 MYA (Stein et al., 2003). The new estimation, based on 338 sets of orthologs, is more precise because of the large gene sample size and improved C. briggsae gene predictions and ortholog assignments.

There are more introns per gene in C. elegans than in C. briggsae (see Table 1). Comparative analysis, by aligning orthologous protein-coding genes between C. briggsae and C. elegans, revealed 6,579 species-specific introns among the 60,775 introns in the orthologous pairs of protein-coding genes. Roughly two-thirds (4,379) are C. elegans-specific, while one-third (2,200) are C. briggsae-specific, which suggest rates of intron gains or losses of about 0.5 per gene in the 80–110 million years since the split between C. briggsae and C. elegans. This rate is dramatically higher than that of mouse and human, which is 0.01 intron gains or losses per gene in 75 million years. Active intron gains and losses have also been observed by comparing many smaller gene sets in C. elegans and C. briggsae, for example, the chemosensory gene families (Robertson, 1998; Robertson, 2000; Robertson, 2001), a sets of five genes (four CPEB genes and fog-3; Cho et al., 2004), and another set of five genes (SSU rDNA, LSU rDNA, part of RNAP2, par-6, and pkc-3; Kiontke et al., 2004). These studies demonstrated dramatically higher rates of intron gain and loss in C. briggsae and C. elegans compared to that in mammals, and furthermore, significantly higher rates of intron loss compared to intron gain, which happen only rarely. Recently, a study identified 81 introns gained in C. elegans and 41 in C. briggsae (Coghlan and Wolfe, 2004). Among these 122 new introns, twenty-eight show significant DNA sequence similarity to other introns in the same genomes, including three that are similar to other introns in the same gene, suggesting that some of the introns were gained via reverse splicing of preexisting intron.

Whole genome sequence alignment of C. briggsae and C. elegans, using the pair-wise alignment tool WABA (Kent and Zahler, 2000), revealed that 50.1% of the C. briggsae genome aligns to 52.3% of the C. elegans genome (Stein et al., 2003). Merged WABA alignments, 4,837 in total, cover 84.6% of the C. briggsae genome and 80.8% of the C. elegans genome. Classification of these WABA alignments by searching for different types of breakpoints, which are defined as junctions between adjacent colinear blocks with C. briggsae supercontigs, resulted in 1,384 putative inversion events, 244 putative translocation events, and 2,735 putative transposition events between C. briggsae and C. elegans genomes. By examining C. elegans sequences, matching each end of the breakpoint junctions on C. briggsae supercontigs we found a clear bias toward rearrangements being within the same ancestral chromosome (see Table 2).

Essentially all (96%) of the approximately 1,000 identified trans-spliced operons in C. elegans (Blumenthal and Gleason, 2003) are conserved in C. briggsae (Stein et al., 2003), suggesting that operons are under strong purifying selection pressure. Recent high throughput functional genomic projects including microarray analysis (Kim et al., 2001) and SAGE (serial analysis of gene expression; McKay et al., 2003) have demonstrated that genes within operons in C. elegans have positively correlated gene expression (Chen and Stein, 2006; Lercher et al., 2003). Whether genes within operons in C. briggsae are positively correlated in gene expression remains to be examined.

Correlation of gene expression is not limited to genes within operons. Recently, we have demonstrated that expression of neighboring genes is also positively correlated (Chen and Stein, 2006). Using tissue-specific SAGE libraries generated by the C. elegans Gene Expression Consortium, Chen and Stein (Chen and Stein, 2006) found that divergent neighboring gene pairs and parallel neighboring gene pairs are positively correlated in gene expression, while convergent neighboring gene pairs show a greatly reduced correlation in gene expression that is lower than the background correlation (see Figure 2; Chen and Stein, 2006). All of these correlations are dependent on the distances between neighboring gene pairs, the correlations approach background as the distances become longer than 10kbp (see Figure 2). As extreme examples of neighboring genes, we analyzed genes that physically overlap in their gene models (introns, exons and UTRs). Nested gene pairs and convergent gene pairs demonstrated lack of correlation in gene expression, while piggyback gene pairs showed positive correlation in gene expression. Since no large-scale functional genomics projects such as microarray analysis or SAGE have been carried out in C. briggsae, correlation of gene expression for its neighboring genes and overlapping genes are unknown. However, similar types and extends of correlation in gene expression likely exist in C. briggsae because overlapping genes are well conserved between the two nematodes (Chen and Stein, 2006).

There are similar numbers of program-predicted non-coding RNA genes in C. briggsae (962 and 191 fragmentary matches) and C. elegans (838 and 212 fragmentary matches). A detailed list of various types of RNA genes is included in Table 1, which shows that both species have similar number of each type of non-coding RNAs. For example, both species have exactly 18 SL2 spliced leader genes, and 70 out of 105 identified C. elegans miRNAs have close homologs in C. briggsae.

Most gene families have similar sizes between C. briggsae and C. elegans, as suggested by TRIBE-MCL clustering analysis (Stein et al., 2003). However, there are exceptions in that some gene families are larger in C. briggsae than in C. elegans and vice versa. For example, there are many more putative chemosensory genes in C. elegans than in C. briggsae (Chen et al., 2005; Stein et al., 2003). Breaking down the putative chemosensory superfamily into smaller families, shows that some families have similar sizes in C. briggsae and C. elegans, while others, including the srab family (Chen et al. 2005b) and the srz family (Thomas et al., 2005), have twice as many genes in C. elegans than in C. briggsae. Such gene family size differences resulted largely from massive tandem duplications of individual genes in the C. elegans genome. Tandem duplication were also observed in C. briggsae but both the ratio and frequency of expansion was not as dramatic in C. briggsae as in C. elegans (Chen et al., 2005). Altogether, TRIBE-MCL clustering analysis demonstrated that for gene clusters with five or more C. briggsae and C. elegans genes, there were 202 clusters (corresponding to gene families) with gene number differences of at least 2-fold between C. briggsae and C. elegans (Stein et al., 2003).

The excretory system in C. elegans plays a critical role in osmoregulation. It is made up of just three cells: the canal cell, the duct cell and the pore cell. The duct and pore cells connect the canal to the outside world and help eliminate excess body fluid. In C. briggsae the excretory system looks morphologically identical, however C. briggsae animals have more anterior duct opening than C. elegans (Wang and Chamberlin, 2002). This difference in the placement of duct opening is the result of the altered expression of lin-48, a Drosophila ovo family member (Wang and Chamberlin, 2002). In C. elegans, lin-48 is expressed in the excretory duct cell, a subset of the hindgut cells and a few neurons. In contrast, C. briggsae lin-48 does not express in the duct cell due to an evolutionary change in the regulation of Cbr-lin-48. This demonstrates that evolutionary changes in the regulation of C. briggsae genes (or C. elegans genes) can give rise to morphological differences.

Regulation of sex-specific developmental programs is of prime importance for the propagation and diversity of multicellular organisms. Experimental findings in metazoans have revealed that, in spite of their critical role, many sex determination genes and genetic pathways have rapidly evolved. Because of their morphological diversity and two different modes of reproduction (hermaphroditic, e.g., C. briggsae and C. elegans, and gonochoristic, e.g., C. remanei, C. japonica and C. sp.CB5161), nematodes provide excellent opportunity to study the molecular basis of sexual differentiation and evolution. In this respect, the C. briggsae genome has facilitated identification of the sex-determination genes and study of their functional differences. Given that hermaphroditism has evolved independently in C. elegans and C. briggsae (Kiontke et al., 2004), it is conceivable that the underlying regulatory mechanisms have diverged between the two species. A survey of the C. elegans sex pathway genes in C. briggsae genome has revealed significant sequence divergence in some genes suggesting functional differences in the mode of sex determination. Consistent with this observation, FEM-3 (novel) and TRA-2 (transmembrane domain) proteins have rapidly evolved in C. briggsae and C. remanei (Haag and Ackerman, 2005; Haag et al., 2002). In C. elegans, the two proteins are known to physically interact with each other (Mehra et al., 1999). While this interaction appears to be conserved in C. briggsae, the two proteins function in strictly species-specific manners. Furthermore, RNA interference (RNAi)-mediated inactivation of the fem-1 and fem-2 genes in C. briggsae reveal functional differences from C. elegans (Stothard and Pilgrim, 2003). In a separate study, fog-2, which is required for hermaphrodite spermatogenesis in C. elegans, has been shown to be absent in C. briggsae (Nayak et al., 2005). Hence, the genome sequence of C. briggsae promises new understanding of the evolution of sex-determination pathways.

Nematode vulval development is an established model for comparative developmental biology. In C. elegans, six vulval precursor cells (VPCs) (Pn.p, where n=3–8) respond to a spatially graded inductive signal from the gonadal anchor cell (AC) and give rise to a 3°-3°-2°-1°-2°-3° cell fate pattern. While the 1° and 2° VPCs divide to give rise to vulval tissue, the 3° lineage cells fuse into the surrounding hypodermal syncytial cell (hyp7). The cell ablation experiments have shown that all six VPCs are equally capable of adopting 1° and 2° fates (Horvitz and Sternberg, 1991). For example, ablation of all VPCs but the P3.p, allows it to generate vulval progeny although it does not do so in an intact animal. By contrast, in C. briggsae P3.p does not appear to be competent by this assay (Delattre and Felix, 2001). However, in certain multivulva mutants of C. briggsae P3.p is capable of being induced and give rise to vulval progeny suggesting that negative regulation may restrict the fate of P3.p in wild-type animals (B. Gupta, unpublished). Thus, the competence of P3.p appears to have evolved between the two species.

Rudel and Kimble (Rudel and Kimble, 2001; Rudel and Kimble, 2002) have examined evolution of Notch-family members, lin-12 and glp-1, in Caenorhabditis nematodes (C. elegans, C. briggsae and C. remanei). Their results suggest that in C. briggsae lin-12 and glp-1 have acquired novel functions. Furthermore, there appears to be two lin-12-like genes in C. briggsae compared to only one in C. elegans. In C. elegans LIN-12 and GLP-1 are functionally interchangeable and in wild-type animals perform some overlapping functions, but they are uniquely required in some tissues (see LIN-12/Notch signaling in C. elegans). GLP-1 is necessary during early embryogenesis and later in the development and proliferation of the germline whereas LIN-12, but not GLP-1, is required during lateral signaling events in some somatic tissues such as the vulva. In C. elegans loss of function alleles of lin-12 exhibit protruding vulva (Pvul) phenotype as a result of the fate conversion of the 2° lineage cells P5.p and P7.p into the 1° type cell lineage (see LIN-12/Notch signaling in C. elegans). glp-1 loss of function alleles, on the other hand, do not affect vulval development. However, in C. briggsae, glp-1 is required for vulval development. Cbr-glp-1(RNAi) animals exhibit a multivulva phenotype suggesting that wild-type glp-1 functions to regulate the competence of the VPCs. Likewise examination of the role of lin-12 revealed that, unlike lin-12 in C. elegans, lin-12 in C. briggsae is essential for larval viability. Cbr-lin-12(RNAi) animals arrest during L₁ larval stage, a phenotype that is not observed in Cel-lin-12(RNAi) animals. Functional differences between lin-12 and glp-1 in C. elegans and C. briggsae are supported by the low level of amino acid conservation (68% and 61% identities, respectively) and may result from evolutionary changes in protein-protein interactions.

In a separate study, Kirouac and Sternberg (Kirouac and Sternberg, 2003) have examined regulation of the three orthologs that are expressed in vulval cells (egl-17, zmp-1 and cdh-3) in C. briggsae and have identified many key regulatory elements that are conserved between the two species. In spite of this conservation, the dynamics of the expression of these genes in vulval cells have evolved. Hence, for example, the C. elegans egl-17::GFP reporter, which is always expressed in vulC and vulD cell types in C. elegans, often fails to express in the same set of cells in C. briggsae. The expression pattern of the cdh-3::GFP reporter, on the other hand, is significantly diverged in C. briggsae (Kirouac and Sternberg, 2003). This finding is somewhat similar to the regulatory differences reported for the pes-1 and fkh-2 forkhead family members (Molin et al., 2000). While both these genes are functionally redundant in C. elegans and C. briggsae embryogenesis, C. briggsae fkh-2 has acquired additional novel function.

The male tail, involved in copulation, exhibits considerable differences in morphology among the Rhabditidae family (including C. elegans and C. briggsae). This variability is one criteria used by researchers to distinguish between and classify various nematodes (Fitch, 1997). C. briggsae's male tail has a subtle difference, relative to C. elegans, in its pattern of ray morphology, specifically there is a tendency of ray 3 to fuse with ray 4. Baird and colleagues (Baird et al., 2005) have used recombinant inbred lines (RILs) to map at least two loci (linked to Cbr-unc-22) that appear to play a role in this morphological variation.

Studies focusing on the ability of nematodes to survive in different environmental conditions have shown evolutionary differences between C. elegans and C. briggsae. Examination of the thermal sensitivity of C. elegans has revealed significant reduction in the brood size at temperatures above 25°C and almost complete sterility above 27°C (M. Ailion and J. Thomas, personal communication; B. Gupta, unpublished). By contrast, C. briggsae grows well at 27.5°C with no significant reduction in brood size (AF16: 200 +/− 35 compared to 234 +/− 25 at 20°C; Fodor et al., 1983). In fact, C. briggsae can grow and develop to adulthood at 30°C (M. Ailion and J. Thomas, personal communication; RJ, unpublished). These studies have been done using multiple isolates of C. elegans and C. briggsae and all isolates of each species behave similarly. The formation of dauer larva, a developmentally arrested diapause stage, is another temperature-dependent trait that exhibits differences between C. briggsae and other Caenorhabditis nematodes (M. Ailion and J. Thomas, personal communication; T. Inoue and P. Sternberg, personal communication). C. elegans when grown at temperature above 25°C forms dauers even when well-fed, and analysis of mutants show these temperatures to be a strong dauer-inducing condition. In contrast, well-fed C. briggsae does not form dauers at these temperatures and analysis of mutants have failed to detect temperature-induced dauer induction at temperatures above 25°C (T. Inoue, personal communication).

In multicellular organisms many cells and tissues exhibit morphological and functional asymmetries. C. elegans has been successfully used to study function of some of the genes that confer asymmetry on neural cells (Hobert et al., 2002). Ortiz and colleagues (Ortiz et al., 2006) have carried out a detailed study of the sequence and expression pattern of the guanylyl cyclase (gcy) genes in C. elegans and C. briggsae. A total of 23 (out of 33) C. briggsae gcy genes have clear one-to-one orthologs in C. elegans whereas the non-orthologous family members (10 in C. briggsae and 11 in C. elegans) appear to have arisen as a result of species-specific gene duplication events. The expression patterns of two gcy genes (gcy-4 and gcy-19) in C. briggsae ASE neurons (ASEL and ASER) show striking evolutionary differences that resulted from changes in cis-acting elements and trans-acting factors. Since the C. elegans ASE neurons exhibit functional left/right asymmetry and express distinct set of gcy genes, the mechanism controlling regulation of the ASE laterality appears to have evolved in C. briggsae.

Among different approaches to deliver double-stranded RNA (dsRNA) to intact C. elegans cells, RNAi by feeding (systemic RNAi) is a rapid method to study gene function (Timmons et al., 2001). While the RNAi by microinjection technique has been successfully used to examine function of some of the genes in C. briggsae (Stothard and Pilgrim, 2003), wild-type C. briggsae AF16 is resistant to systemic RNAi. This is due to the lack of a functional sid-2 gene since the AF16 worms expressing C. elegans sid-2 (Cel-sid-2) are susceptible to systemic RNAi (C. Hunter, personal communication). Hence, a C. briggsae transgenic strain carrying integrated Cel-sid-2 array can be used in functional genomic studies.

To facilitate genetic studies of C. briggsae, a large collection of mutant strains (more than 200) is freely available to the community (see Table 3). These strains not only allow genetic and molecular analysis of the involved loci but also serve as useful markers in phenotypic analyses and linkage mapping of new mutants. While most of these are curated in three different laboratories (B. Gupta, P. Sternberg and D. Baillie), some of the strains are available at the Caenorhabditis Genetics Center for centralized distribution. A comprehensive list of the mutants can be found at the C. briggsae website.

The phenotypic marker-based genetic linkage map is currently under construction (see Figure 3; also see the C. briggsae website; B. Gupta, T. Inoue, RJ, A. Mah, D. Ballie and P. Sternberg, unpublished). This has involved standard 2 and 3-factor genetic mapping of mutations that exhibit visible phenotypes and cloning of some of the loci. In parallel, R. Waterston and colleagues (personal communication) are working to construct a single nucleotide polymorphism (SNP)-based genetic linkage map of C. briggsae. This approach involves careful mapping of SNPs identified by partial sequencing of two C. briggsae isolates HK104 and VT847. The current version of the SNP-based genetic linkage map (v3.1) is based on the 248 genotyped SNPs that are roughly equally distributed over the six chromosomes. Integration of the SNP-based and mutant-based maps, which is ongoing (B. Gupta and R. Miller, unpublished), will allow researchers to place their gene of interest on different linkage groups, examine synteny with C. elegans, carry out epistasis and genetic interaction experiments. This will facilitate cloning of the C. briggsae genes, in particular those with the novel mutant phenotypes i.e., ones for which there are no candidate genes proposed from our knowledge of C. elegans, and study evolutionary changes in gene function.

Mos1 is a Drosophila transposon that has been successfully used in C. elegans to generate mutations (Bessereau et al., 2001). The advantage of this system is that it can be used as a sequence tag to rapidly identify the mutated gene by PCR and therefore circumvent the need for time-consuming genetic mapping and gene cloning experiments. Dr. Marie-Anne Felix has tested the applicability of the system in C. briggsae and successfully recovered mutations in three genes (http://snp.wustl.edu/snp-research/c-briggsae/mos-mutagenesis.html). While the experimental protocol is in the early phase of development in C. briggsae, it has potential of rapidly isolating alleles of the C. elegans orthologs (M.-A. Felix, personal communication; B. Gupta, unpublished).

The wild populations of C. briggsae and C. elegans serve as useful resources for studies involving natural variation and ecology of the species. Comparison of the genomic DNA sequences in these strains has shown greater nucleotide diversity in C. briggsae compared to C. elegans (Graustein et al., 2002; Jovelin et al., 2003). Currently more than 60 wild isolates of C. briggsae are available, they have been collected from various regions of the world (S. Baird and M.-A. Felix, personal communications). These include 57 new isolates that are part of a recent study looking at the pattern of polymorphisms in 63 strains across the three continents (Asia, Europe and America; Cutter et al., 2006). In addition, a large number of recombinant inbred lines (RILs) have been generated. These are derived from two sets of crosses between AF16, HK104 and VT847 (set 1: AF16 and HK104, set 2: AF16 and VT847; S. Baird, personal communication). Besides their use in mapping quantitative trait loci (QTLs), the RILs are playing key role in construction of the SNP-based genetic linkage map (R. Miller and S. Baird, personal communications).

The genomic clones of C. briggsae (in the form of BAC and fosmid clones) were generated to construct a high resolution physical map to help assemble the sequenced genome. The BAC clones are available from the Children's Hospital Oakland Research Institute.