NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Riddle DL, Blumenthal T, Meyer BJ, et al., editors. C. elegans II. 2nd edition. Cold Spring Harbor (NY): Cold Spring Harbor Laboratory Press; 1997.

Cover of C. elegans II

C. elegans II. 2nd edition.

Show details

Section IIIInformational Suppressors

Isolating and analyzing genetic suppressors are a fundamental part of“forward” genetic analysis. Suppressors are defined classically as mutations that correct the phenotypic defects of another mutation without restoring its wild-type sequence. Suppressors may be intragenic (affecting the same gene) or they may be extragenic (affecting a different gene). Extragenic suppressors are particularly useful during genetic analyses, because they often identify additional components of a biological system or process. Informational suppressors are those that alter the genetic apparatus with which genes are expressed.

Two distinct types of informational suppressors have been identified in C. elegans. First, at least 13 different amber suppressors have been described, most of which affect defined tRNA genes. Both genetic and molecular criteria indicate that such suppressors are analogous to classical amber suppressors of microorganisms. Second, mutations affecting seven smg genes eliminate the C. elegans system of nonsense-mediated mRNA decay. Because of their effects on mRNA abundance, smg mutations can either suppress or enhance the phenotypic affects of other mutations.

A. Amber Suppressors

Amber mutations introduce UAG nonsense codons within affected mRNAs. Amber suppressors decode such codons as“sense,” thereby restoring gene product and function. C. elegans is the only metazoan in which amber suppressors have been identified by traditional reversion analysis. Amber suppressors have been identified among the revertants of at least six different C. elegans genes. Literally hundreds of revertants of amber alleles affecting unc-15 , unc-13 , tra-3 , unc-51 , dpy-20 , and lin-1 have been tested for the presence of amber (or other informational) suppressors (Waterston and Brenner 1978; Waterston 1981; Hodgkin 1985; Kondo et al. 1990); 13 different suppressors have been identified from these studies.

1. Genetic Properties of Amber Suppressors

Table 1 summarizes genetic and molecular properties of C. elegans amber suppressors. Analysis of the suppressors listed in Table 1, and of amber mutations themselves, leads to the following conclusions concerning amber suppression in C. elegans:


Suppression is allele-specific, but not gene-specific. Amber alleles of many different genes are suppressed, but for any specific gene, only certain alleles (the amber alleles) are suppressed.


Suppression is dose-dependent. The efficiency of phenotypic suppression depends on how many copies of an amber suppressor are present in a strain. Suppressor homozygotes suppress more strongly than suppressor heterozygotes. Amber suppressors are thus semidominant.


Different suppressors have different strengths. The suppressors in Table 1 differ greatly in the degree to which they suppress known amber mutations. When strengths of suppression are compared, sup-7 is almost always the strongest suppressor, whereas sup-29 is the weakest (Kondo et al. 1990). Amber alleles of tra-3 provide the most sensitive test for amber suppression. sup-29 , for example, strongly suppresses amber alleles of tra-3 but does not detectably suppress amber alleles of several other genes (Kondo et al. 1988). The proportion of UAG nonsense codons that are translated as sense can be remarkably high. For example, when unc-15 (amber); sup-7 mutants are grown at 15°C (see below for temperature effects), the quantity of full-length paramyosin is 39−45% of that found in wild type (Waterston 1981).


The strength of suppression is often temperature-dependent. Phenotypic suppression by sup-5 , sup-7 , sup-21 , sup-28 , and sup-34 is stronger at 15°C than at higher temperatures (Waterston and Brenner 1978; Waterston 1981; Kondo et al. 1990). The effect of temperature can be substantial. unc-15 (amber); sup-7 mutants grown at 15°C accumulate almost twice as much full-length paramyosin as those grown at 20°C (20−25% of wild type compared to ∼10% of wild type, respectively) (Waterston and Brenner 1978; Waterston 1981).


The strength of suppression is tissue-specific. Amber mutations have been identified in a wide variety of C. elegans genes, including those expressed in the nervous system, muscle, hypodermis, and germ line. The suppressors in Table 1 show consistent tissue-specific differences when tests of suppression are performed against a battery of amber alleles. For example, sup-21 , sup-28 , sup-33 , and sup-34 are strong suppressors when tested against genes expressed in the hypodermis, but they do not suppress (or do so only weakly) when tested against genes expressed in the nervous system (Kondo et al. 1988, 1990).


Strong suppression is deleterious. Both sup-5 and sup-7 exhibit dose-dependent and temperature-dependent growth abnormalities that result from high levels of suppression. When grown at 15°C, sup-5 (e1464) has a longer generation time than wild type (7.5 days vs. 5.5 days) and exhibits increased sterility. sup-7 (st5), the strongest of the suppressors, cannot be grown as a homozygote for more than one generation at 15°C. Deleterious effects of strong amber suppression presumably result from inappropriate translation of natural UAG translation terminators.

Table 1. Genetic and molecular properties of amber suppressors.

Table 1

Genetic and molecular properties of amber suppressors.

Amber suppressors are valuable tools for interpreting genetic data. They are especially valuable because they provide molecular insights into gene function. Amber alleles of a gene (those phenotypically suppressed by amber suppressors) are usually null alleles, although there are occasional exceptions (Ferguson and Horvitz 1985; Hodgkin 1985, 1987a; Charest et al. 1990). Amber suppressors and amber alleles can be manipulated to confer conditional phenotypes to essential genes. Because amber suppressors are semidominant, varying the gene dose of a suppressor varies the amount of a suppressed gene product. Because the strength of amber suppression is tissue-specific, varying the suppressor varies the tissues in which suppression (and hence gene function) occurs. Amber suppressors can be effective markers to identify transgenic animals (Fire 1986). The toxicity of multiple copies of amber suppressors can be used to select for low-copy integrated transgenes.

2. Molecular Analysis of Amber Suppressors

Long before they were identified as tRNA genes, the genetic properties of sup-5 and sup-7 suggested strongly that they were analogous to classic tRNA-mediated suppressors of microorganisms (Waterston and Brenner 1978; Waterston 1981). Three lines of evidence demonstrate that this is true and that sup-7 is, in fact, an amber suppressor. First, a purified tRNA fraction from sup-7 mutants was shown to possess suppressor activity when microinjected into the gonad (Kimble et al. 1982) or when used to program an in vitro translation system (Wills et al. 1983). Second, unc-54 (e1300), a sup-7 -suppressible allele of a myosin heavy-chain gene, was shown to be an amber mutation (Wills et al. 1983). Third, sup-7 was cloned, sequenced, and shown to encode a tRNATrpUGG (Bolten et al. 1984). Further analysis of amber suppressors and of the tRNATrpUGG gene family demonstrated that 8 of the 13 amber suppressors described in Table 1 are mutations of tRNATrpUGG genes (Bolten et al. 1984; Kondo et al. 1988, 1990). All eight suppressors encode identical tRNAs in which a single C→T substitution changes the anticodon of a tRNATrp gene from 5′-CCA-3′to 5′-CUA-3′. The anticodon change thus allows mutant tRNAs to read the amber codon UAG. It is not certain, however, whether tryptophan is inserted at suppressed amber sites. Analogous tRNATrpUGG suppressors have been identified in E. coli (Hirsh 1971), but the anticodon change causes mischarging of tRNATrp with glutamine rather than tryptophan (Yaniv et al. 1981). It is unknown whether this is also true of C. elegans tRNATrp amber suppressors. Certain of the gene assignments of Table 1 are derived from molecular analysis of tRNATrpUGG genes. For example, five alleles of“ sup-21 ” were originally defined on the basis of their similar map positions (Hodgkin 1985). Subsequent molecular analyses of these alleles and of the tRNATrpUGG gene family (Kondo et al. 1990) demonstrated that these five alleles affect at least three distinct genes ( sup-21 , sup-28 , and sup-32 ). Similarly, alleles originally defined as affecting“ sup-22 ” proved to affect two distinct genes ( sup-22 and sup-29 ) (Kondo et al. 1988).

C. elegans contains 12 tRNATrpUGG genes, 8 of which have been mutated to suppressor alleles. When one tRNATrpUGG gene is converted to a suppressor allele, the remaining 11 genes provide sufficient tRNATrp for translation of UGG codons during translation. Thus, at least some of the 12 wild-type genes are functionally redundant with regard to their ability to translate UGG codons. Consistent with this, animals that are homozygous for probable null alleles of sup-7 are phenotypically wild type (Waterston 1981).

Of the 12 tRNATrpUGG genes, 4 have not been mutated to suppressor alleles. Such“silent” tRNATrpUGG genes may simply have been missed during reversion analyses performed to date or their pattern of expression may preclude isolating suppressor alleles. For example, rtw-3 and rtw-5 may be pseudogenes (Kondo et al. 1990). rtw-6 and rtw-7 may be expressed in insufficient or excessive quantities, such that suppressor alleles either have no detectable effect on phenotype or have such deleterious effects that suppressor alleles are dominant lethals.

Five amber suppressors in Table 1 are not alleles of tRNATrpUGG genes ( sup-22 , sup-23 , sup-32 , sup[st402], and sup[st414]; Hodgkin 1985; Kondo et al. 1990). Such suppressors are suspected, but not proved, to be mutations affecting other tRNA genes, such as tRNAGlnCAG or tRNALysAAG. An amber-suppressing tRNASer gene of Drosophila functions efficiently as an amber suppressor when transformed into C. elegans (Pilgrim and Bell 1993). Such manipulations of transgenic tRNA genes may eventually allow direct control of the inserted amino acid or of the resulting pattern of tRNA expression.

B. smg Suppressors

Amber suppressors are“textbook” examples of how alterations of the genetic apparatus can influence the phenotypic consequences of mutations in many different genes. In principle, any of the many steps required for gene expression might be targets for informational suppression. For example, modification of the spliceosome can overcome defects resulting from mutant splice sites (Parker et al. 1987). Genetic and molecular analyses of informational suppressors are valuable both because they identify basic mechanisms of gene expression and because they provide technical tools for unrelated investigations. The seven C. elegans smg genes are informational suppressors whose wild-type mode of action involves selective mRNA degradation.

Independent work in three different laboratories identified smg mutations as extragenic suppressors of mutations affecting either sex determination, developmental timing, or muscle filament assembly (Hodgkin et al. 1989). Subsequent investigations established that smg mutations eliminate a specific system of mRNA turnover (Pulak and Anderson 1993). All eukaryotes examined to date have a system that selectively degrades mRNAs containing premature stop codons (Peltz et al. 1994; Maquat 1995). This system, termed nonsense-mediated mRNA decay (NMD), has been proposed to protect cells against the deleterious effects of expressing nonsense-fragment polypeptides produced either by somatic mutation or by errors of“normal” gene expression ("mRNA surveillance"; Pulak and Anderson 1993). Genes required for nonsense-mediated decay have been identified in both yeasts (the UPF/NMD genes; Leeds et al. 1991; Peltz et al. 1994; Cui et al. 1995; He and Jacobson 1995; Lee and Culbertson 1995) and nematodes (the smg genes) (Pulak and Anderson 1993).

Six smg genes ( smg-1 through smg-6 ) were originally identified as allele-specific suppressors of mutations affecting either tra-3 , lin-29 , or unc-54 (Hodgkin et al. 1989). A seventh gene, smg-7 , has been identified more recently as a suppressor of unc-54 (B. Cali and P. Anderson, unpubl.). smg-suppressible alleles of numerous additional genes have been identified either by reversion analysis or by direct tests of suppression. These include alleles of unc-17 (M. Nguyen et al., pers. comm.), dpy-5 (Hodgkin et al. 1989), glp-1 (Mango et al. 1991), gon-2 (Y. Sun and E. Lambie, pers. comm.), pha-1 (Schnabel et al. 1991), tra-1 (Hodgkin et al. 1989; Zarkower et al. 1994), tra-2 (Hodgkin et al. 1989), unc-30 (R. Hoskins, pers. comm.), and unc-76 (L. Bloom and R. Horvitz, pers. comm.). In each case, it is known (or believed) that phenotypic suppression occurs because (1) the mutant mRNAs are unstable in a smg(+) background, (2) the mutant mRNAs are stable in smg(−) backgrounds, (3) the mutant mRNAs are translated in a manner that yields partially or fully functional gene product, and (4) the elevated levels of gene product in smg(−) backgrounds are sufficient to confer a phenotypic effect. This molecular explanation has been rigorously demonstrated, however, only for certain smg-affected alleles of unc-54 and tra-1 (Pulak and Anderson 1993; Zarkower et al. 1994). The steady-state levels of unc-54 nonsense-mutant mRNAs are about 5% of normal in a smg(+) background and about normal in smg(−) backgrounds. Nonsense mutations nearer the unc-54 normal translational terminator have a less pronounced effect on mRNA stability.

Table 2. Genetic properties of smg genes.

Table 2

Genetic properties of smg genes.

Table 2 summarizes genetic properties of the smg genes. Analysis of the suppressors listed in Table 2 and of smg-suppressible mutations, leads to the following conclusions:


smg genes act globally. smg-suppressible (see above) and smg-en-hanced (see below) alleles of many different genes have been identified, including genes expressed in the muscle, nervous system, hypodermis, pharynx, developing vulva, and germ line. The existence of smg-affected alleles indicates that the smg genes themselves function in all of those tissues. An allele that is suppressed (or enhanced) by any smg mutation is suppressed (or enhanced) by all smg mutations.


Nonsense-mediated mRNA decay is not essential. As measured by the steady-state abundance of unc-54 (nonsense) mRNA, nonsense-mediated mRNA decay is eliminated in all tested alleles of smg-1 , smg-2 , smg-3 , smg-4 , smg-5 , and smg-7 (Pulak and Anderson 1993). Thus, the system of nonsense-mediated decay is not essential, consistent with its proposed role in mRNA surveillance. Nonsense-mediated decay is reduced, but not eliminated, in the only tested allele of smg-6 .


smg mutants exhibit mild morphogenetic abnormalities. The designation“smg” (suppressor with morphogenetic effects on genitalia) describes the only conspicuous phenotype of smg mutants. Adult hermaphrodites exhibit protruding vulvae, and adult males have abnormal bursae. Despite these abnormalities, smg() hermaphrodites are egg-laying-proficient, and smg() males are cross-fertile, albeit with low efficiency. Young smg() males mate more efficiently than older males. The morphogenetic abnormalities of smg mutants suggest a subtle role for nonsense-mediated decay in the expression of normal genes. Both the suppression and morphological phenotypes of smg mutants are recessive, but certain synthetic dominant interactions between smg genes have been noted (Hodgkin et al. 1989).


At least three smg genes ( smg-1 , smg-2 , and smg-5 ) are nonessential. smg() single mutants, including alleles identified by either their suppression or morphological phenotypes, are viable, healthy, and exhibit only the mild abnormalities described above. smg alleles isolated in such screens, however, are of necessity viable and fertile, so this alone does not establish the null phenotype. Continuing molecular analyses of smg genes have identified known (or probable) null alleles of smg-1 , smg-2 , and smg-5 (S. O'Connor and P. Anderson; B. Carr and P. Anderson; K. Anders and P. Anderson; all unpubl.). Thus, smg-1 , smg-2 , and smg-5 are nonessential genes, and they appear by genetic criteria to function only in nonsense-mediated mRNA decay.


Certain smg genes may be essential. Alleles of smg-6 and smg-7 isolated in a“noncomplementation” screen are often lethal when homozygous (B. Cali and P. Anderson, unpubl.). Thus, smg-6 and smg-7 may be essential genes, performing vital functions in addition to their role in nonsense-mediated decay.


Certain smg genes exhibit a maternal effect. Suppression by smg-3 , smg-4 , and smg-6 can be affected by maternal genotype. Specifically, animals of genotype unc-54 (r293);smg(), when derived as offspring of smg/+ heterozygous mothers, are only partially suppressed. Suppression of lin-29 (n546) exhibits a similar maternal effect for smg-6 but not for smg-3 or smg-4 . Maternal effects have not been detected with alleles of smg-1 , smg-2 , and smg-5 .


smg-suppressible alleles are often, but not always, nonsense mutations. Although many smg-suppressible alleles are nonsense mutations (Mango et al. 1991; J. Hodgkin, pers. comm.), others are deletions or rearrangements affecting the 3′end of a gene (Pulak and Anderson 1988; Hodgkin 1993; Alfonso et al. 1994a; Zarkower et al. 1994). In cases of suppressed nonsense mutations, suppression likely occurs either because the nonsense-fragment polypeptide is partially active or because certain nonsense codons are occasionally mistranslated as“sense” in smg() backgrounds, similar to that observed in yeast (Peltz et al. 1994). In cases of smg-suppressible rearrangements, deletions or other mutations affecting the 3′UTR cause an upstream stop codon (which, in some cases, can be the normal translational terminator codon) to be recognized as“premature” by the NMD system, thereby resulting in degradation of the mutant message. Such mRNAs are unstable in smg(+) backgrounds but stable in smg() backgrounds.

C. smg -dependent Dominance

Although smg mutations were first isolated as informational suppressors of recessive mutant phenotypes, one of their most striking (and useful) genetic properties is that they are informational enhancers of many dominant mutant phenotypes. Mutations that are recessive or only weakly dominant in smg(+) genetic backgrounds can be strongly dominant in smg(−) backgrounds. smg mutations can enhance the heterozygous phenotypes of either dominant-negative or dominant gain-of-function alleles. Conditional dominance of this type was first noted during genetic mapping of smg genes and during tests of suppression (Hodgkin et al. 1989; Mango et al. 1991), but this phenomenon appears to be remarkably common. Of 14 tested unc-54 nonsense mutations, 4 are recessive when smg(+) and dominant when smg(−) (Pulak and Anderson 1993). Of 10 feminizing tra-1 alleles isolated in a smg(−) genetic background, 8 are smg-sensitive; i.e., expression of the mutant phenotype is greatly reduced in a wild-type background (Zarkower et al. 1994). Similarly, approximately 2/3 of dominant visible mutations isolated in a smg(−) background, affecting 12−14 different genes, become recessive (or only weakly dominant) when crossed into a smg(+) background (B. Cali and P. Anderson, in prep.). In smg(+) backgrounds, many of these mutations are recessive-lethals. Thus, smg(−) mutations can be valuable because they confer conditional viable phenotypes to recessive-lethal mutations.

smg-dependent dominance likely results from expression of nonsense-fragment polypeptides that have dominant-negative activities. Such polypeptide fragments are expected to be expressed at higher levels in smg(−) strains, because the nonsense-mutant mRNAs are present in greater abundance. The location of smg-dependent dominant mutations in the unc-54 myosin heavy-chain gene is striking. Only myosins truncated near the myosin head/rod junction exhibit synthetic dominance (Pulak and Anderson 1993). Genetic tests demonstrate that the nonsense-fragment polypeptides themselves are the disruptive gene product (B. Cali and P. Anderson, in prep.). In principle, smg-dependent dominance may be especially pronounced in proteins like myosin that interact in macromolecular complexes. Nonsense-fragment polypeptides that are sufficiently stable to assemble into a complex with wild-type proteins may sufficiently disrupt the function of the complex to result in a dominant mutant phenotype. The location of smg-affected alleles can often define functional domains of a protein (Mango et al. 1991; Zarkower et al. 1994).

D. Mechanism(s) of Nonsense-mediated mRNA Decay

Nonsense-mutant mRNAs are unstable in most, if not all, eukaryotes (Peltz et al. 1991; Sachs 1993). How are nonsense-mutant mRNAs targeted for selective degradation? The mechanisms are not understood in detail, but insights from yeast and from cultured mammalian cells provide some clues. Selective turnover of nonsense-mutant mRNAs is coupled to translation (Belgrader et al. 1993; Peltz et al. 1993) and is preceded by decapping of the mutant message (Muhlrad and Parker 1994). Nonsense-mutant mRNAs are not deadenylated prior to decapping, unlike many wild-type mRNAs (Shyu et al. 1991; Muhlrad and Parker 1994). Cis-acting elements that either promote (Peltz et al. 1993; Cheng et al. 1994; Zhang et al. 1995) or inhibit (Peltz et al. 1993) nonsense-mediated mRNA decay have been defined. Such elements likely provide the means by which normal stop codons are distinguished from premature stop codons, although details of how they function are unclear.

As discussed above, the C. elegans smg genes and the yeast UPF/NMD genes encode trans-acting factors required for nonsense-mediated mRNA decay. How similar are the yeast and nematode proteins? SMG-2 and UPF1 are strong sequence homologs, being approximately 50% identical over most of their length (Leeds et al. 1992; K. Anders et al., unpubl.). Despite these similarities, expression of UPF1 in worms and expression of smg-2 in yeast do not appear to rescue NMD function in appropriate mutants (K. Anders and P. Anderson, unpubl.). The sequences of UPF2, UPF3, smg-1, smg-5, and smg-7 do not reveal any further strong homologies among UPF and smg genes (Cui et al. 1995; He and Jacobson 1995; Lee and Culbertson 1995; S. O'Connor et al., unpubl.). This might indicate that additional yeast and nematode genes remain to be discovered or that components of the NMD system are not strongly conserved. Both SMG-5 and SMG-7 are novel proteins; SMG-1 is a large protein that contains a“PI-3/protein” kinase domain at its carboxyl terminus (S. O'Connor and P. Anderson, unpubl.).

An unresolved issue concerning nonsense-mediated mRNA decay is where within the cell turnover occurs. A requirement for translation would suggest that turnover occurs in the cytoplasm, but observations involving several mammalian genes suggest that certain mRNAs are degraded while associated with nuclei (either within the nucleus or possibly during transport to the cytoplasm) (Maquat 1995). It is unclear at present whether there are distinct cytoplasmic and nuclear mechanisms of NMD. The UPF1 protein of yeast is located predominantly, if not exclusively, in the cytoplasm and is associated with polysomes (Atkin et al. 1995). It will be important to determine where within the cell the smg gene products are located.

Copyright © 1997, Cold Spring Harbor Laboratory Press.
Bookshelf ID: NBK19998


  • PubReader
  • Print View
  • Cite this Page

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...