Gene duplications and genetic redundancy in C. elegans^*

The duplication of genes is of paramount importance in providing raw materials for the evolution of genetic diversity. It is therefore of interest to consider C. elegans gene duplications in the context of other sequenced genomes. Two recent estimates of the overall extent of gene duplications in C. elegans have come up with very similar figures. There are thought to be around 1,200 gene families containing two or more paralogues in C. elegans (Cavalcanti et al., 2003; Gu et al., 2002). The total number of genes in the paranome is around 6,000, or ∼ 32% of the genome, where the paranome is defined as the set of proteins that have one or more paralogues, that is, those that are not singletons. In addition, 7.1% of the duplicated genes in the worm are thought to have resulted from block duplications (duplication events involving more than one gene; Cavalcanti et al., 2003). Simple sequence duplications range in size from hundreds of bases to tens of kilobases and copies may be dispersed or clustered (The C. elegans Sequencing Consortium, 1998). Local clusters of duplicated genes are easily identified and there are 402 such clusters (where a cluster is defined as a group of N genes that are similar within a window of 2N genes along the chromosome and where N = 3 or more) distributed throughout the genome (The C. elegans Sequencing Consortium, 1998). The worm genome contains twice the number of local gene duplications as Drosophila (Rubin et al., 2000), and also differs from the fly genome in terms of the distribution of gene duplicates. Gene duplications are found randomly throughout the fly genome, whereas in the worm duplicated genes are mostly clustered in the recombinogenic segments of the autosomal arms (Rubin et al., 2000).

The rate of origin of gene duplicates in C. elegans over the past few hundred thousand years appears to be substantially higher than that for Drosophila or yeast (Lynch and Conery, 2000). Lynch and Conery (2000) propose a per-gene rate of duplication of 0.02 per million years for C. elegans, compared with a rate of 0.002 duplications per gene per million years in Drosophila and 0.008 in yeast. Gu et al. (2002) have reported exactly the same figure for C. elegans, compared with a rate of only 0.0014 for Drosophila (Gu, 2003).

Given the range of values predicted for different species, it has been proposed that 50% of genes in a genome would be expected to duplicate, on average, at least once on time scales of 35 to 350 million years (Lynch and Conery, 2000). Thus, even in the absence of direct amplification of the entire genome (polyploidization), gene duplication has substantial potential for generating substrates for evolutionary innovation. Indeed, it has been proposed that the rate of duplication of a gene is of the same order of magnitude as the rate of mutation per nucleotide site (Lynch and Conery, 2000).

What is the mechanism of gene duplication? Replication slippage and unequal exchange are often invoked as an explanation for closely spaced gene duplicates, but these mechanisms would be expected to give rise to tandem duplicates pointing in the same direction. A recent study found, however, that up to 69% of C. elegans duplicate genes reside in the inverse orientation, especially young duplicates (Katju and Lynch, 2003). Inversions are considered by these authors to be part and parcel of the original duplication event, rather than secondary rearrangements (Katju and Lynch, 2003). It has been hypothesized that inverted duplications could be generated by an illegitimate recombination event during DNA replication, involving strand switching by the DNA polymerase or strand misalignment-realignment. RNA-mediated transposition is thought unlikely to have played a significant role in gene duplication within the C. elegans genome because of the very small proportion of duplicate genes that lack introns relative to the original copy (Katju and Lynch, 2003). The mechanisms responsible for gene duplication are therefore, in general, unlikely to respect gene boundaries. This is borne out by the Katju and Lynch study, which analysed 290 gene pairs with ≤ 10% divergence at synonymous sites (Ks) within the C. elegans genome. They found that the average duplication span of 1.4 kb is less than the average gene length in C. elegans (2.5 kb), suggesting that partial gene duplications are frequent (Katju and Lynch, 2003).

Around half of the C. elegans gene duplicates with very low levels of synonymous site substitution were found to also contain unique coding sequence not present in the other copy, in addition to the region of close homology (Katju and Lynch, 2003). Thus, structural heterogeneity between duplicate genes is common. Chimeric duplicates may well have creative potential, especially when they act in conjunction with shuffling events.

What fate awaits duplicate genes? Three alternative theories have been proposed (Lynch, 2002; Lynch and Conery, 2000). The first is the nonfunctionalization of one copy by the accumulation of degenerative mutations. The second is neofunctionalization. In this scenario, one copy acquires a novel, beneficial function and becomes preserved by natural selection, with the other copy retaining the original function. A variation of this scenario is that the two genes could acquire divergent functions but maintain a functional overlap. Selection would then act on these divergent properties and indirectly maintain the functional overlap. Selection could also act on a newly emergent property unique to the combined action of two closely related genes, or on some enhanced efficiency or fidelity achieved by the combined action of two such genes (Thomas, 1993). In the third scenario, subfunctionalization, both copies become partially compromised by the accumulation of mutations, to a point where their total capacity is reduced to the level of the single-copy ancestral gene.

Whatever the mechanism, most models for conserving redundancy over the course of evolution require some degree of symmetry-breaking: equality amongst duplicated genes must be disrupted if they are to be preserved. The fate awaiting most gene duplicates in C. elegans, as well as in most other species studied, is likely to be silencing and ultimate loss, rather than preservation (Lynch and Conery, 2000). It has been estimated that the average half-life of a gene duplicate is around 4 million years (Lynch and Conery, 2000). The propensity of recently duplicated genes to become nonfunctional pseudogenes is borne out in a recent study by Mounsey et al. (2002), who found a lower than expected success rate in generating expression patterns for recently duplicated genes, suggesting that recently duplicated genes are less likely to be expressed. Given our current state of knowledge of gene function in C. elegans, it is impossible to estimate the actual proportion of duplicated genes in the genome that are redundant. This will become clearer as sensitized screens and combinatorial RNAi approaches seek to address the general problem of assigning gene function.

One recent study examined the relationship between the prevalence of gene duplications and ontogeny in Caenorhabditis species (Castillo-Davis and Hartl, 2002). It was found that genes expressed after embryogenesis had a significantly greater number of duplicates than those expressed early in development. This was found to be true in both the C. elegans and C. briggsae genomes. For example, 18.36% of early-expressed genes (n=1,280) were found to have detectable paralogs in the C. elegans genome versus 35.31% of late-expressed genes (n=1,014). For C. briggsae, the figures are even more striking, 6.7% (n=165) and 38.8% (n=237), respectively (Castillo-Davis and Hartl, 2002). Therefore, duplicated copies of early-expressed genes appear to be selectively lost. Based on earlier calculations of the rate of origin of gene duplicates in C. elegans (Lynch and Conery, 2000) and the divergence between C. elegans and C. briggsae, Castillo-Davis and Hartl estimated that more than 40% of all genes are expected to have duplicated at least once in both the C. elegans and C. briggsae lineages since their divergence (Castillo-Davis and Hartl, 2002). The proportion of duplicated genes in the late-expressed class thus falls close to this estimate, whereas the proportion of duplicated genes in the early-expressed class is much lower than expected. It is suggested that the selective loss of duplicates of early-expressed genes reflects developmental constraint (Castillo-Davis and Hartl, 2002). A related study found that gene duplications were more prevalent amongst the set of conserved, slowly-evolving genes versus the set of non-conserved genes in the C. elegans and S. cerevisiae genomes, suggesting that slowly evolving genes may be the main source of new genes in eukaryotic genomes (Davis and Petrov, 2004).

11 of 33 of the largest clusters of duplicated genes in C. elegans consist of olfactory receptors (Rubin et al., 2000). Olfactory receptors are seven-transmembrane G-protein coupled receptors (GPCRs), each of which specifically recognises a set of odorant and tastant chemicals, allowing the worm to sense and respond to its chemical environment. GPCRs comprise by far the largest multigene family in C. elegans, with over 1000 members, making up ∼ 5% of the genome (Bargmann, 1998), although around a third of this group of genes are thought to be pseudogenes (Bargmann, 1998). Thus many duplications in this large multigene family are fated to become nonfunctional. Several recent studies have charted the molecular evolution of some of these genes and found that processes of duplication, diversification and movement that have led to these large gene families are very much ongoing. For example, the 3126 bp inverse orientation duplication that duplicates the 5’ half of gene T08H10.2 to give rise to pseudogene T08H10.a is thought to have occurred very recently as the duplicated sequences are identical (Robertson, 1998). It has been suggested that strong selective pressures are at work for the continued functionality of these genes because the occurrence of synonymous changes in duplicated GPCR genes is 11-fold higher than the occurrence of amino acid substitutions or nonsynonymous changes (Robertson, 1998; Robertson, 2000). Indeed, it has been proposed that natural selection might somehow favour duplications of genes that are generally involved in responses to environmental stress and pathogens, in organisms facing a challenging and dynamic molecular environment (Lespinet et al., 2002). The massive lineage specific expansion of worm odorant/chemosensory receptors could be just one example of this.

Other large gene families in C. elegans include C-type lectins, hormone receptors, collagens and serine/threonine/tyrosine protein kinases. A full description of major protein-coding gene families, including a discussion of family size distribution, can be found in Genomic classification of protein-coding gene families. An example of a C. elegans transcription factor gene family in which duplications abound is the T-box family. There are 21 T-box genes in the C. elegans genome (WormBase release WS132, 2004), most of which lack clear orthologs in other species. There are 4 pairs of genes that are likely to have arisen from relatively recent duplication events and a functional analysis of 2 of these pairs has been reported. tbx-37 and tbx-38 (83% amino acid identity) have redundant functions in mesoderm induction in C. elegans embryos (Good et al., 2004), whereas tbx-8 and tbx-9 (59% amino acid identity), have overlapping, but probably not completely redundant, functions in embryonic morphogenesis (Pocock et al., 2004). The most recent duplication in the T-box gene family would appear to be the one that gave rise to the genes Y59E9AR.3 (tbx-30) and Y59E9AR.5 (tbx-30.1). These two sequences are 100% identical and situated in inverse orientations, ∼ 2 kb apart. Biological function has been ascribed to tbx-30 (Pocock et al., 2004), but it is unclear at present whether or not tbx-30.1 is expressed.

There are several other good examples of duplicated gene pairs that have undergone neofunctionalization. A recent one is the fbf-1/fbf-2 gene pair involved in germ line development in C. elegans. fbf-1 and fbf-2 encode closely related RNA binding proteins that reciprocally regulate the expression of one another in order to modulate the size of the germ line mitotic region (Lamont et al., 2004). Thus, there appears to be a benefit unique to the combined action of these two closely related genes; the duplication event in this case has provided an opportunity for fine-tuning of a developmental process.

A good potential example of the subfunctionalization of a duplicated gene pair in C. elegans is the case of the Notch-like receptors lin-12 and glp-1. In C. elegans, lin-12 and glp-1 posess both overlapping and separate biological functions. When both gene functions are removed, larval lethality results (the Lag – Lin And Glp phenotype; Lambie and Kimble, 1991). However, in C. briggsae and C. remanei, the Lag phenotype is seen when only lin-12 expression is silenced (Rudel and Kimble, 2002). It is suggested that ancestral functions may have been divided between lin-12 and glp-1 in C. elegans after duplication in such a way that their total capacity became reduced to the level of the single-copy ancestral gene. In C. briggsae and C. remanei, on the other hand, lin-12 appears to have retained this ancestral function following duplication (Rudel and Kimble, 2002).

An interesting example of the apparent gradual evolutionary demise of a duplicated gene is presented in the case of elt-4. elt-4 encodes a truncated GATA type zinc finger transcription factor of just 8.1kDa (72 amino acids), and is situated ∼ 5 kb upstream of the highly conserved elt-2, proposed to be the original copy of the duplicate pair (Fukushige et al., 2003). The bigger elt-2 gene is expressed in intestinal cells and is required for intestinal development (Fukushige et al., 1998). While it is clear that elt-4 is expressed in the intestine, no effect was found of deleting elt-4, in the gut or elsewhere, either alone or in combination with elt-2. Furthermore, experiments in yeast and in vitro could not demonstrate any role for ELT-4 in activating or repressing transcription, or indeed any specific DNA binding activity (Fukushige et al., 2003). The elt-2/elt-4 duplication event is thought to have occurred only in the C. elegans lineage, after the point at which C. elegans and C. briggsae diverged. Thus, the proposal is that although elt-4 may have conferred some selective advantage to C. elegans in the past (hence the high level of conservation of the zinc finger domain), its ultimate evolutionary fate will be disappearance from the C. elegans genome (Fukushige et al., 2003).

Genome sequence analysis is thus providing the community with far more than simply the informational content of genomes. Gene duplications are widespread in the C. elegans genome and provide raw material for evolutionary novelty. The data are a snapshot of evolutionary time, from which we can glimpse into the past as well as, perhaps, seek to prophesy the future.

I would like to thank members of my laboratory and Andreas Russ for useful discussions and the reviewers for helpful comments.