Genetic enhancers^*

A mutation in one gene that enhances or exacerbates the phenotype caused by a mutation in another gene is called an enhancer (not be confused with an enhancer DNA element, which is a cis-acting regulator of transcription). The enhancement must exceed what one would expect from a simple summation of the phenotypic effects of the two mutations on their own, and thus each mutation can be considered to be an enhancer of the other. When the double mutant exhibits a phenotype not apparent in either single mutant, the novel phenotype is referred to as synthetic. Although one should consider the possibility that mutations in two genes that have completely unrelated functions might overburden an organism and be unexpectedly deleterious even though the single mutants are relatively healthy, the identification of enhancers in C. elegans has in practice often led to the discovery of genes that interact in informative ways. Enhancement may arise from different types of genetic interaction. For example, two genes may provide a specific function redundantly, in which case the homozygous double mutant may exhibit a much stronger phenotype than either single mutant. Alternatively, mutations in two genes that are both required for the same function, say by encoding either components of a protein complex or different steps in a signal transduction pathway, may synergize and lead to an enhanced double mutant phenotype.

For two genes that exhibit enhancement, insight into the possible type of genetic interaction can sometimes be gleaned from the nature of the mutations affecting the two genes. For example, null mutations that enhance each other identify overlapping or partially redundant functions. Each mutation on its own may cause a phenotype, but the double mutant exhibits either a novel or a much enhanced phenotype relative to either single mutant (Figure 1). The overlapping functions may be homologous, or they may be independent functions that affect a specific biological process redundantly. Many important functions appear to be provided redundantly or at least partially redundantly, and screens for enhancement of null mutations are likely to become increasingly important for the elucidation of worm development and behavior.

Hypomorphic mutations that enhance each other, by contrast, may be in genes that work in the same pathway; the two wild-type genes may act in the same direction, either positively or negatively, to regulate a downstream function, in which case the two hypomorphic mutations can synergize to give a phenotype stronger than the phenotype caused by either mutation on its own (Figure 2). We emphasize that two mutually-enhancing hypomorphic mutations need not always reflect involvement of the two genes in the same pathway or protein complex; impairment but not elimination of two redundant pathways by hypomorphic mutations might result in an insufficient level of final product or output.

Intergenic noncomplementation refers to the situation in which two mutations are in different genes and fail to complement despite being fully recessive to their wild-type alleles when present by themselves; i. e., each single heterozygote has a wild-type phenotype, but the double heterozygote exhibits a mutant phenotype. Each mutation can be thought of as a genetic enhancer of the other mutation. Intergenic noncomplementation is generally rare, and only a few examples are known in C. elegans, but they are instructive. Various alleles of the genes sqt-1, sqt-2, sqt-3, and rol-8 are recessive to their wild-type alleles on their own but show intergenic noncomplementation in certain combinations to give a twisted cuticle, which may be left-handed or right-handed, depending on the specific mutations involved (Kusch and Edgar, 1986). The alleles that behave in this way are not null, and it was proposed that they produce proteins that interact to disrupt the cuticular architecture. It is therefore not surprising that sqt-1, sqt-2, sqt-3 and rol-8 all encode cuticular collagens (Kramer et al., 1988; Cox et al., 1989; van der Keyl et al., 1994).

Mutations in genes that affect chemical synapse function have also provided good examples of intergenic noncomplementation (Yook et al., 2001; see Complementation). For example, certain recessive mutations in unc-13 and unc-64, which encode proteins required for the exocytosis step of the synaptic vesicle cycle, fail to complement. As in the cuticle examples, noncomplementation does not occur when both mutations are null, which means that noncomplementation is not simply the result of reduced dosage of both gene products but depends on the poisoning effect of a partially functional protein. Noncomplementation was also observed between mutations in other genes involved in synaptic vesicle trafficking, including genes whose products are thought not to interact directly. Allele-specific intergenic noncomplementation between mutations in genes that act in a pathway for male tail ray development has also been reported (Ikegami et al., 2004).

Dominant enhancement that does not involve intergenic noncomplementation has been seen in an analysis of animals homozygous for a temperature-sensitive mutation in one of several mec genes that affect the response to light touch (Gu et al., 1996). It was found that many otherwise-recessive mutations in other mec genes dominantly enhance the mechanosensory defects conferred by the temperature-sensitive mutations. The results were consistent with a model in which the interacting genes encode components of a protein complex that transduces mechanosensory signals. Finally, a myo-3 null mutation dominantly enhances the phenotype of a homozygous unc-54 mutant (Waterston, 1989); myo-3 and unc-54 encode different forms of myosin heavy chain, both present in body wall muscle. The enhancement in this case is due to reduced dosage of myo-3 product; conversely, enhanced expression of MYO-3 suppresses the uncoordinated phenotype conferred by the loss of UNC-54 (Riddle and Brenner, 1978; Maruyama et al., 1989).

Only 20-35% of all C. elegans genes when inactivated by mutation give a lethal, sterile, or obvious visible phenotype (Johnsen and Baillie, 1991; Waterston et al., 1997; C. elegans Sequencing Consortium, 1998; Hodgkin and Herman, 1998; see Essential genes). The many genes with nearly silent null alleles may contribute in subtle but significant ways to the fitness of the worm in nature, perhaps under particular environmental conditions, but the mild effects of knockout mutations make it difficult to study the specific roles of these genes. An obvious challenge is to determine what all these additional genes are doing. Many genes with subtle effects contribute redundantly to visible or essential functions (Thomas, 1993; Cooke et al., 1997). Redundancy is apparent when the simultaneous inactivation of two genes results in a strong phenotype not seen with either single gene knockout. In this case, each mutation is acting as an enhancer (usually recessive) of the other mutation. Complete gene redundancy, in which two genes perform exactly the same role and either can be inactivated with no loss of fitness, may be rare (Brookfield, 1997), but partial gene redundancy seems to be common, and many examples, again generally involving mutually-acting recessive enhancers, have been studied in C. elegans, and several are cited in the following two sections. Possible selective advantages of gene redundancy have been discussed (Thomas, 1993; Cooke et al., 1997); one idea, for example, is that partially redundant genes may together increase the fidelity of a biological process. Whatever the benefit of redundancy for the organism, there are obvious benefits to the researcher in studying redundant genes in appropriate double (or multiple) mutants.

Null or near-null mutations were identified in three genes that encode different kinetic classes of acetylcholinesterase (Culotti et al., 1981; Johnson et al., 1981; Johnson et al., 1988). Single mutants and ace-1 ace-3 and ace-2 ace-3 double mutants exhibit wild-type or nearly wild-type behavior, but ace-1 ace-2 mutations enhance each other, causing a severe uncoordination in the homozygous double mutant. Finally, the ace-3 mutation is an enhancer in the triple mutant, which is inviable—a synthetic lethality. Clearly, certain acetylcholinesterase functions are provided redundantly.

The genes lin-12 and glp-1 were originally identified on the basis of mutations that led to cell fate transformations. The genes were subsequently shown to encode similar Notch-like transmembrane receptors. The single mutant phenotypes are distinct, but the lin-12 glp-1 double mutant exhibits a much more severe phenotype than that of either single mutant (Lambie and Kimble, 1991). It was concluded that lin-12 and glp-1 provide some developmental roles individually, which are apparent in single mutants, but that they also provide some functions redundantly, which is made evident by the synthetic lethality of the double mutant.

Many examples of enhancement between mutations in homologous genes are now known, largely as a consequence of the availability of the genome sequence, which has given a complete picture of duplicated genes and gene families (see Gene duplications and genetic redundancy in C. elegans). Often, a mutation is identified that gives a phenotype (perhaps a subtle one); the gene is cloned; a homolog in the worm genome is found; and then either a knockout mutation of the second gene is sought, or RNA interference (RNAi) is used to reduce its expression; the phenotypes of single gene knockouts are then compared to the double knockout. Table 1 is an incomplete list of examples in which enhancement has been found by this general strategy. In all of these cases, it was concluded that the functions of the two members of each pair of genes overlap at least partly.

There are also cases of functional overlap involving three homologous genes, as was seen above for the ace genes. For example, the three worm rac-like genes function redundantly in axon pathfinding (Lundquist et al., 2001), as shown by enhancement of axon defects in various mutant combinations or mutants treated by RNAi; and a combination of RNAi and mutation in a sensitized genetic background has been used to demonstrate that the genes apx-1, dsl-1, and lag-2 encode redundantly-acting ligands for LIN-12 signaling during vulva development (Chen and Greenwald, 2004).

Overlapping functions can also be supplied by nonhomologous genes, which can be assigned to distinct but functionally overlapping pathways when animals carrying null mutations in both exhibit a more severe or more highly penetrant mutant phenotype than that found for either single mutant. In this way, genes affecting engulfment of cell corpses have been assigned to two parallel pathways (Ellis et al., 1991), as have genes that regulate dauer larvae formation (Malone and Thomas, 1994), sex myoblast migration (Chen et al., 1997), entry into meiosis (Kadyk and Kimble, 1998; Hansen et al., 2004), embryonic morphogenesis (Harrington et al., 2002), and ventral axon guidance (Hao et al., 2001).

Enhancers of null mutations are perhaps most instructive when they uncover surprising interactions. The classic worm example is provided by the synthetic multivulva (SynMuv) genes (Ferguson and Horvitz, 1989). The Muv phenotype (see Vulval development) requires two SynMuv mutations, one in a SynMuv class A gene and the other in a SynMuv class B gene. Two overlapping but nonhomologous pathways repress Muv development, which occurs only when both pathways are inactivated. Evidence for a third overlapping SynMuv pathway was recently published (Ceol and Horvitz, 2004). The original SynMuv interaction was discovered when it was found that a SynMuv mutant carried two unlinked mutations. Additional SynMuv genes were then found by screening for new SynMuv mutants in a class A or class B mutant background.

Screens for enhancers (usually recessive) of null mutations using extrachromosomal arrays provide a powerful method for discovering unanticipated overlapping functions. For example, mutations were sought that on their own conferred no obvious phenotype but which in combination with a null allele of mec-8, which encodes a putative RNA splicing factor, were synthetically lethal (Davies et al., 1999). Four genes identified in this way have been characterized (Davies et al., 1999; Yochem et al., 2004). Similarly, mutations have been identified that are synthetically lethal with a mutation in lin-35 (Fay et al., 2002), a class B SynMuv gene that encodes the worm homolog of the retinoblastoma protein. The interacting genes have been implicated in control of cell proliferation (Fay et al., 2002), pharyngeal morphogenesis (Fay et al., 2003; Fay et al., 2004), larval development (Cui et al., 2004), and somatic gonad development (Bender et al., 2004). Synthetic interactions with lin-35 and with other SynMuv genes in the control of cell-cycle progression have also been demonstrated by double-mutant and RNAi combinations (Boxem and van den Heuvel, 2001, 2002).

RNAi feeding of unc-34 null mutants has identified a synthetically lethal interaction between unc-34 and wve-1 (Withee et al., 2004); these two genes encode actin-regulating proteins that play overlapping roles in embryogenesis. In a genome-wide screen using RNAi feeding, gene interactions have been identified that are involved in the response to DNA damage (van Haaften et al., 2004). Genome-wide RNAi feeding is likely to become increasingly popular for identifying genetic enhancers, and an approach that uses extrachromosomal arrays is illustrated in Figure 3.

Screens for recessive enhancers of partial loss-of-function mutations are often designed to identify additional components acting in the same pathway. For example, in a screen for enhancers of a weak glp-1 mutation affecting germline proliferation, novel alleles of two genes previously known to play essential roles in germline development were identified as well as mutations in five new genes (Qiao et al., 1995). One of the latter mutations was a weak allele of a gene called ego-1, which encodes a predicted RNA-directed RNA polymerase that functions both in germline development and in the process of RNA interference (Smardon et al., 2000). A screen for enhancers of defects in egg-laying and the excretory system caused by a hypomorphic lin-45 raf mutation yielded unusual alleles of several known Ras pathway genes, confirming that the screen worked as intended, and two new genes, eor-1 and eor-2 (Howard and Sundaram, 2002; Rocheleau et al., 2002). Apparent null alleles of eor-1 and eor-2 on their own (or in eor-1 eor-2 double mutants) cause only low penetrant defects, but they synergize with other Ras pathway mutations, suggesting that they play a modulatory role in Ras pathway signaling. Indeed, strong enhancement of null alleles of lin-25 and sur-2, which positively regulate Ras signaling, indicates that eor-1 and eor-2 provide a function that is functionally redundant with the function provided by lin-25 and sur-2 (Howard and Sundaram, 2002). In a screen for mutations that enhance a weak allele of mab-20, which encodes a semaphorin and plays a role in the formation of male tail rays, six genes were identified by enhancer mutations (Ikegami et al., 2004). One of these was the first known mutation in plx-2, which encodes a presumed receptor for MAB-20. Although mutations in plx-2 enhance a weak allele of mab-20, they do not enhance a null allele, indicating that plx-2 functions in the same pathway as mab-20. On the other hand, a putative null allele of plx-2, unlike mab-20 null, did not cause male ray defects on its own; genetic interactions between plx-2 and other genes argued for the existence of a parallel pathway in MAB-20 signal reception, involving an ephrin receptor.

Enhancer mutations can also be found among suppressor mutations (see Genetic suppression) when the suppressors are placed in a different genetic background. For example, suppressors of a lin-12 loss-of-function, not surprisingly, can enhance a lin-12 gain-of-function mutation (Sundaram and Greenwald, 1993). Conversely, sel-7 mutations, identified as suppressors of a lin-12 gain-of-function mutation, enhance a hypomorphic lin-12 mutation as well as a null allele of the presenilin gene sel-12 (Chen et al., 2004).

The phenotypic effects of some non-null alleles of unc-52 are greatly enhanced by loss-of-function mutations in mec-8 such that the mec-8; unc-52 double mutant exhibits the unc-52 null phenotype (Lundquist and Herman, 1994). The explanation for this result is that MEC-8 is required to promote specific alternatively-spliced mRNA isoforms that skip unc-52 exons that carry the enhanced mutations (Lundquist et al., 1996).

Enhancers have also been isolated in C. elegans that do not enhance the phenotype caused by a mutation in an endogenous worm gene; rather, they enhance a late-onset neuronal degeneration caused by an added transgene that expresses polyglutamine fragments (Faber et al., 2002). Recessive enhancer mutations in this case identified a gene that in its wild-type form can protect against polyglutamine toxicity. Thus, the examples of genetic enhancement in C. elegans are as multifarious as they are informative, and additional analyses should continue to enhance our knowledge of the genetic interactions within the worm genome.