Non-protein encoding DNA performs a variety of important biological functions (reviewed in [1]). However, many of the functions of non-coding DNA are poorly understood. One such non-coding DNA element is guanine-rich DNA, which has been characterized in several functional domains: the telomeres, the ribosomal DNA and, in mammals, the immunoglobulin heavy-chain switch regions [2]. These G-rich DNA elements all have stretches of consecutive guanines and have the capacity to form secondary structures such as G-quadruplex or G4 DNA by Hoogsteen bonding [3]. G4 DNA has been proposed to have multiple biological functions in vivo including the regulation of gene expression and chromosome dynamics [2]. It has been hypothesised that the G4 conformation in RNA transcripts derived from G-rich DNA may be targets of transcriptional regulation based on the findings that many factors associated with RNA processing, including hnRNP D, hnRNP A1, and nucleolin, bind G4 DNA through their conserved RNA-recognition motif and RNA-binding domain [4–6]. More direct evidence for G-rich DNA affecting the transcription regulation was the finding that the G-rich DNA segment in the human c-myc promoter, which could form G4 DNA in vitro, functions as a repressor element [7]. Many studies have also suggested a possible role for G-rich motifs in chromosome dynamics. Sen and Gilbert originally proposed that the G-rich telomeres and internal chromosomal motifs in homologous chromosomes may participate in pairing during meiotic prophase, based on their observation that G-rich motifs in DNA can form parallel four-stranded G4 DNA [8]. Many proteins involved in chromosome synapsis and recombination were also found to interact with the G-rich DNA motifs [9–12]. Furthermore, the synaptonemal complex lateral element component Hop1 plays key roles in meiotic chromosome pairing, promoting synapsis of double-stranded DNA helices in vitro via the formation of G4 DNA [13, 14]. In addition to the in vitro evidence, G4 DNA structures formed by G-rich DNA in the immunoglobulin switch region were shown to be present in vivo, and are proposed to play a role in the immunoglobulin class switch recombination [4, 15]. These findings support the possibility of G-rich DNA being involved in chromosome dynamics.

The probability of one stretch of 18 guanines occurring by chance in the 100 Mb AT-rich C. elegans genome (GC content 36%) [16] is approximately 1/6 to the 18^th power, 1 in 100 trillion. In the Caenorhabditis elegans genome there are approximately 400 such homopolymeric poly-G/poly-C tracts (G/C tracts). Thus, these tracts are greatly over-represented in the genome. A study by Denver and colleagues on homopolymeric nucleotide (HP) runs in C. elegans reported that the observed number of the G/C tracts is much greater than expected. While the number of A/T tracts declines steadily with the increase of the length as expected, G/C tracts do not display that trend [17]. Furthermore, retention of the tracts was found to be dependent on enzymatic activity as disruption of DOG-1, a protein with similarity to the human FANCJ helicase, caused deletions that initiated in G/C tracts with no less than 18 guanines [18]. Cheung et al. proposed that DOG-1 may prevent deletions of G/C tracts by unwinding G-rich secondary structures arising during lagging strand DNA synthesis [18]. Youds et al. demonstrated involvement of the homologous recombination repair pathway in the prevention of G/C tract deletions in the dog-1 mutant [19]. Based on the observations that G/C tracts are over-represented in the C. elegans genome and protected by enzymatic activity, it seems unlikely that the occurrence and maintenance of G/C tracts are by chance.

In this study, we characterize the frequency and distribution of G/C tracts in two species of nematode, C. elegans and C. briggsae and explore possible biological roles of these tracts in these two organisms.

Both the C. elegans and C. briggsae genomes contain many more G/C tracts than expected as not even one tract greater than 18 bps is expected to be found by chance in a genome of 100 Mb of DNA. Most G/C tracts are located in intergenic or intronic areas. Genes containing tracts are much larger than average and more G/C tracts are located on the gene-poor chromosomal arms. Thus, the tracts are located in regions with a low density of coding DNA. However, the fact that both the C. elegans and C. briggsae genomes contain a large number of these unusual homopolymeric tracts distributed across each chromosome and the existence of a conserved protein, DOG-1, that prevents the deletion of G/C tracts support the possibility that G/C tracts have biological significance.

Initially it would appear that the X chromosome in C. elegans is significantly different than the autosomes with regard to G/C tracts distribution. The X chromosome contains almost twice as many G/C tracts as the autosomes whereas both inverted and tandem repetitive sequences are more frequent on the autosomes than on the X chromosome [16]. For example, CeRep11, with 711 copies distributed over the autosomes, has only one copy located on the X chromosome [16]. Although the X chromosome has many more G/C tracts than the autosomes, it has approximately the same number of intragenic G/C tracts (24/109) compared to the autosomes (Avg. 21.4 each), which means the large number of G/C tracts on the X are due to a larger number of intergenic G/C tracts (85 vs. 36 of autosomes) (Table 3). Furthermore, the overall gene density of the X chromosome is lower than that of the autosomes [16]. Taken together with regard to the distribution of G/C tracts on X chromosome, it would appear that the X chromosome is organized like an autosomal arm. More G/C tracts were found to be located on the arms of the X chromosome even though the X chromosome does not have a meiotically defined central gene cluster, suggesting that there is no correlation between the meiotic pattern and G/C tracts. Recent study on C. briggsae genetic mapping using SNPs showed that C. briggsae has a similar meiotic pattern as C. elegans[28]. This finding, along with our observation of the uniform distribution of G/C tracts in C. briggsae, also suggests that G/C tracts are not correlated with meiotic recombination.

We examined in C. elegans the distribution of the tracts in the context of the level of gene expression. Based on SAGE analysis, we observed a variation of transcription levels that correlates with distance from the G/C tract. The average number of SAGE tags associated with G/C tract flanking genes is significantly lower than the number of SAGE tags associated with genes further from G/C tracts. This effect is similar to the suppression of human c-myc gene by its G-rich promoter [7]. Although no evidence for direct regulation by G/C tracts on specific genes was found in C. elegans, our observation showed that there is a correlation with the presence of a G/C tract and the level of expression. If G/C tracts played important gene specific cis-regulatory functions they should be conserved in C. briggsae but few are. Thus, we believe G/C tracts are unlikely to be regulating transcription in a gene specific manner. On the other hand, previous studies on regulatory elements in C. elegans muscle genes showed that regulatory elements are also highly represented in C. briggsae muscle genes but the conservation of individual sites is weak [33]. One of their interpretations was that a specific site in C. elegans could disappear over evolutionary time and reappear at a different position and retain its regulatory activity. Similarly, G/C tracts in C. elegans, although not conserved in position in C. briggsae, could be affecting regional gene expression. This may be due to changes in the local chromatin environment, perhaps as a result of the G/C tract that affects gene transcription, or that these chromatin environments may be more conducive to the generation or maintenance of G/C tracts.

The capability of G-rich secondary DNA structures to interact with each other using non Watson-Crick base pairing makes it an excellent candidate to coordinate chromosome dynamics such as pairing during meiosis. Sen and Gilbert originally proposed that G4 DNA might facilitate the pairing of homologous chromosomes without the need for testing homology at the sequence level [8]. There are several lines of evidence suggesting that the G/C tracts in C. elegans might have a role in meiotic chromosome pairing. Many pairing or recombination related components can interact with G-rich DNA secondary structures [9–14]. For example, the mismatch repair protein MutSα binds G4 DNA and promotes synapsis and recombination in the mammalian immunoglobulin switch regions [34] and in yeast, the meiotic synapsis protein Hop1 can promote the formation of G4 DNA and the synapsis of double-stranded DNA helices through the generation of G4 DNA [12–14]. The distribution of G/C tracts across the chromosomes in two species is compatible with a role for the tracts in chromosome alignment for pairing.

In many organisms, homolog pairing is achieved by double-stranded DNA breaks and Rad51-mediated strand invasion. Subsequent strand invasion can then align and pair homologous chromosomes before the formation of the synaptonemal complex. In C. elegans and D. melanogaster, this is not the case as alignment and synapsis occurs without the generation of DSBs (reviewed by Joyce and McKim [35]). Although the specific mechanisms that drive homologous chromosome pairing while preventing the pairing of non-homologous chromosomes are unknown, DNA domains in C. elegans chromosomes have been proposed to play key roles in homolog pairing. Genetic analysis of C. elegans chromosome rearrangements identified cis-acting regions for pairing [36] which occur at one 'end' of each C. elegans chromosome which were called the homolog recognition region (HRR) [37]. Recent reports showed that specific C2H2 zinc-finger proteins bind to these regions and mediate the chromosome synapsis [38, 39], however, they are not essential for homolog pairing [39, 40]. Once synapsis initiates, it is likely that sequences distributed along the chromosomes promote proper alignment. In the case of large deletions or insertions that disrupt alignment, pairing for recombination can reinitiate, thus, there must be some mechanism by which homologous chromosomes can be brought into register along the length of the chromosomes without stand invasion [41, 42]. Abundant G/C tracts distributed along the chromosomes with the ability to generate secondary G-rich structures that can pair with other G-rich structures on the homologs could function to align homologous chromosomes. The observation that yeast Hop1 protein is able to promote synapsis of double-stranded DNA helices by the formation of G4 DNA raised the possibility that Hop1 homologs in C. elegans (him-3 and the him-3 paralogs) might have similar capabilities. This speculation is supported by the fact that mutants of him-3 and him-3 paralogs htp-1, htp-2 were reported to all have chromosome pairing defects [43–45]. It will be interesting to investigate whether him-3 or the him-3 paralogs can interact with G/C tracts.

In this article, we reported our analysis on the G/C tracts in C. elegans genome and their conservation in another Caenorhabditis species, C. briggsae. Overrepresented G/C tracts are dispersed throughout the genomes of these two species, but the specific positions and the overall distribution patterns do not appear to be conserved. Along with the finding that G/C tracts are correlated with the levels of regional gene expression in C. elegans, we proposed that G/C tracts have possible biological roles in chromosome dynamics and/or gene expression regulation.