Mechanism and regulation of translation in C. elegans ^*

From studies in mammals, yeasts, and plants, it is known that the three steps of protein synthesis are catalyzed by three groups of proteins: initiation, elongation, and release factors (Hershey and Merrick, 2000). A different class of initiation factors (eIF1, eIF2, etc.) catalyzes each step of initiation (Figure 1). [A uniform nomenclature system for translation factors is used here (Clark et al., 1996)]. A ternary complex of eIF2•GTP•Met-tRNA_i binds to the 40S ribosomal subunit to form the 43S initiation complex. Recruitment of mRNA to the 43S initiation complex to form the 48S initiation complex requires eIF3, the poly(A)-binding protein (PABP), and the eIF4 proteins. eIF3 is a ~800-kDa multimer that is also required for Met-tRNA_i binding to the 40S subunit (molecular masses refer to the mammalian factors). PABP is a 70-kDa protein that specifically binds poly(A) and homo-oligomerizes. The eIF4 factors consist of: eIF4A, a 46-kDa RNA helicase; eIF4B, a 70-kDa RNA-binding and RNA-annealing protein; eIF4H, a 25-kDa protein that acts with eIF4B to stimulate eIF4A helicase activity; eIF4E, a 25-kDa cap-binding protein; and eIF4G, a 185-kDa protein that specifically binds to and co-localizes all of the other proteins involved in mRNA recruitment on the 40S subunit.

The 48S complex consists of the eIF4 factors plus PABP, eIF1, eIF1A, eIF3, eIF5, and the eIF2•GTP•Met-tRNA_i ternary complex bound to the 40S subunit (Figure 2). It scans until the first AUG in good sequence context is encountered. Scanning requires ATP hydrolysis by eIF4A and the presence of eIF1 and eIF1A. Then eIF5 and eIF5B stimulate GTP hydrolysis by eIF2, followed by 60S joining to form the 80S complex. The released eIF2•GDP is recycled to eIF2•GTP by the guanine nucleotide exchange factor eIF2B. The first elongator aminoacyl-tRNA is brought to the ribosomal A-site by eEF1, after which the first peptide bond is formed. This is followed by a cycle of GTP hydrolysis and exchange. Translocation is catalyzed by eEF2 with another cycle of GTP hydrolysis and exchange. When the ribosome reaches a termination codon, the release factor eRF1 catalyzes termination, and then the GTPase eRF3 ejects eRF1 from the ribosome.

Many components of the translational machinery have been identified in C. elegans, including rRNAs (Albertson, 1984; Ellis et al., 1986; Files and Hirsh, 1981), ribosomal proteins (Jones and Candido, 1993; Zorio et al., 1994; Gonczy et al., 2000), 5S RNA (Nelson and Honda, 1985), tRNA (Schaller et al., 1991; Tranquilla et al., 1982; Khosla and Honda, 1989; Lee et al., 1990), and aminoacyl tRNA synthetases (Amaar and Baillie, 1993; Gabius et al., 1983; Gonczy et al., 2000).

Of the 56 initiation factor (eIF), elongation factor (eEF), and release factor (eRF) polypeptides that we and others have identified (Table 1), only ~10% have been isolated and characterized. In the following overview, information not explicitly cited is taken from WormBase, release WS138.

eIF2 factors. Both eIF2 and eIF2A catalyze the binding of Met-tRNA_i to the 40S ribosomal subunit, the former requiring GTP and the latter, an AUG codon (Zoll et al., 2002). eIF2 is composed of three subunits, α, β, and γ. eIF2β has also been reported to bind mRNA and contributes, together with eIF2γ, to GTP and Met-tRNA_i binding. Inactivation of either of these genes is lethal. Surprisingly, eIF2α deletion mutants are viable but have defects in growth and larval development. Yeast eIF2A is not an essential protein, but deletion of C. elegans eIF2A is embryonically lethal. eIF2B is a heteropentameric complex that is essential in yeast and mammalian cells. In C. elegans, inactivation of either the β, γ (Kuwabara and Shah, 1994), or ε subunits causes growth defects, larval arrest, or embryonic lethality.

eIF4 factors. Two splice-variants of eIF4G, encoded by ifg-1, have been identified in C. elegans (Long et al., 2002; Kamath et al., 2003). Although IFG-1 shows low identity to human eIF4G, its role as a translation factor was confirmed by its retention on m⁷GTP-Sepharose (Keiper and Rhoads, unpublished data) and by its presence in 48S initiation complexes (Dinkova et al., 2005). ifg-1 inactivation results in developmental arrest and embryonic lethality (Long et al., 2002).

eIF4E is encoded by five genes in C. elegans, ife-1 through ife-5 (Jankowska-Anyszka et al., 1998; Keiper et al., 2000; Figure 3). The five proteins can be grouped into three classes based on amino acid sequence identity, cap-binding specificity, and knockout phenotype (Table 2). Some isoforms bind both 2,2,7-trimethylguanosine (TMG)-containing caps as well as 7-methylguanosine (MMG)-containing caps, whereas other isoforms bind only the latter. Cap selectivity is determined by the dimensions and flexibility of the cap-binding pocket (Miyoshi et al., 2002). IFE-1 is bound to P granules through an interaction with PGL-1 and is required for spermatogenesis (Amiri et al., 2001). In the case of IFE-4, inactivation of the gene produces an Egl phenotype, low brood size, and a defect in food sensation (Dinkova et al., 2005). The translational efficiency (polysomal distribution) of only ~1% of mRNAs is affected by ife-4 deletion, but diminished levels of the encoded proteins are consistent with the complex phenotype.

Nine C. elegans gene products are homologous to eIF4A (Table 1; Roussell and Bennett, 1992; Gonczy et al., 2000). The products of F57B9.6 and F57B9.3 are the most homologous to mammalian eIF4A-1 and eIF4A-2. Others (Y65B4A.6 and F33D11.10) may be involved in splicing or nonsense-mediated decay, similar to mammalian eIF4A-3. Some may play a specific role in germline cells (C07H6.5) or in nuclear export (C26D10.2, T26G10.1, T07D4.4, and F53H1.1). Knockout of any eIF4A-like gene except T07D4.4 causes reproductive defects or embryonic lethality.

Other initiation factors and interacting proteins. eIF1, eIF3, and eIF5 form a multifactor complex in yeast that is required for assembly of the 43S initiation complex. C. elegans eIF5 and eIF1 are essential for growth and embryonic development. There are C. elegans homologs for all 12 eIF3 subunits present in higher eukaryotes (Asano et al., 1997), whereas yeast eIF3 has only five subunits. Loss of eIF3a (C27D11.1) produces an Egl phenotype (Desai and Horvitz, 1989). Both eIF3a and eIF3d are involved in meiotic divisions (Gonczy et al., 2000). Another eIF3a homolog (F55H2.6) is possibly involved in mitochondrial morphology and distribution. Inactivation of eIF5B results in larval arrest and sterility (Maeda et al., 2001). C. elegans contains three PABP homologs. PAB-2 (F18H3.3) has the highest similarity to human cytoplasmic PABP. PAB-1 (Y106G6H.2) is essential for gonad development, and PAB-2 is apparently important for somatic development (Ciosk et al., 2004). PAB-3 (C17E4.5) is more similar to human nuclear PABP and is probably not involved in translation.

Elongation and termination factors C. elegans has two identical eEF1A homologs that are products of different genes, eft-3 (Seydoux and Fire, 1994) and eft-4 (Kamath et al., 2003). EFT-3 is required for embryonic viability, fertility, and germline maintenance. There are two homologs of eEF1B, one of which (F54H12.6) is essential for viability while the other (Y41E3.10) is not. The two eEF2 homologs, EFT-1 (Ofulue and Candido, 1992) and EFT-2 (Ofulue and Candido, 1991; Nollen et al., 2004), are expressed during all stages and are encoded by essential genes. eRF1 has two splice variants whereas eRF3 has only one. Both genes are essential for embryonic and larval development, growth, and locomotion.

Spliced leaders. A SL is added to mature mRNA via trans-splicing in nematodes and some other metazoans (Bektesh et al., 1988; Evans et al., 1994; Stover and Steele, 2001). The SL consists of a conserved 22-nt sequence (Conrad et al., 1991; Blumenthal and Steward, 1997). In C. elegans, either SL1 or SL2 (or a variant such as SL3, SL4, etc.) is added to 70% of pre-mRNAs (Blumenthal and Gleason, 2003). SL1 is usually trans-spliced to mRNAs transcribed from the first cistron in an operon or to monocistronic mRNAs, whereas SL2 and its variants are trans-spliced to products of downstream cistrons.

Caps. Those mRNAs that undergo trans-splicing carry a TMG-cap and SL, whereas those that undergo only. cis-splicing carry an MMG-cap and no SL (Van Doren and Hirsh, 1990). The small nuclear RNA that donates the SL contains a TMG-cap (Zorio et al., 1994; Evans et al., 1997). MMG- and TMG-caps are recognized by some but not all eIF4E isoforms (Table 2).

5'UTR. Regulatory features of the 5'UTRs of mRNAs in general involve length, secondary structure, upstream open-reading frames, and specific sequences that interact with RNA binding proteins (Gingras et al., 1999). The trans-splicing processing of many C. elegans mRNAs results in a relatively short 5'UTR in which the SL is located near the AUG codon (Blumenthal and Steward, 1997). A few C. elegans mRNAs, e.g., gna-2, contain long 5'UTRs that harbor upstream open-reading frames (Lee and Schedl, 2004), which have a profound effect on translation efficiency in other organisims (Morris and Geballe, 2000).

Effects of 5'-terminal structures on translational efficiency. Mechanisms of regulation mediated by 5'-terminal structures remain largely unknown for C. elegans. Both MMG- and TMG-containing mRNAs are found on polysomes (Liou and Blumenthal, 1990). In Ascaris lumbricoides, mutations in the SL1 sequence alter efficiency of translation (Maroney et al., 1995). TMG-capped mRNAs are poorly translated in mammalian cell-free translational systems (Darzynkiewicz et al., 1988), but a TMG-cap and SL stimulate translation in an Ascaris suum translation system (Lall et al., 2004). Interestingly, an optimal distance from SL to AUG for translational efficiency can be demonstrated in this system. The translational synergism between cap and poly(A) is greater for TMG than MMG. The well described translational regulator GLD-1 also interacts with 5'UTR sequences (Lee and Schedl, 2004).

3'UTR. Translational regulatory elements at the 3'UTR play important roles in C. elegans mRNA expression (see WormBook chapters on Translational control of maternal RNAs and RNA-binding proteins). The elements first described as DREs (direct repeat elements; Goodwin et al., 1993) and later re-named TGEs (tra GLI elements; Jan et al., 1997) are found within the 3'UTR of tra-2 mRNA and negatively regulate its expression in germ line and somatic cells. The trans-acting factor for TGEs was identified as GLD-1 (Jan et al., 1999). A 5-nucleotide sequence element in the 3'UTR of fem-3 mRNA, cuUCUUGu, also exerts translational regulation (Anderson and Kimble, 1997). Another kind of 3'UTR element that represses translation is the lin-4 complementary element (LCE) found in lin-14 and lin-28 mRNAs (Wightman et al., 1993; Seggerson et al., 2002). This element is bound by the microRNA lin-4 (Moss et al., 1997; see C. elegans microRNAs). Mutations in these 3'UTR elements disrupt the translational regulation of their mRNAs (see WormBook chapters referenced above).

poly(A) The poly(A) tract is added by a poly(A) polymerase after specific cleavage during mRNA splicing. The nuclear polyadenylation signal consists of an AAUAAA sequence 20-30 nt upstream of the cleavage site (Blumenthal, 1995; Hajarnavis et al., 2004). In C. elegans, GLD-2 is a putative catalytic subunit of cytoplasmic poly(A) polymerase that is likely recruited to mRNAs by interaction with RNA binding proteins such as GLD-3 (Wang et al., 2002). Cytoplasmic polyadenylation also requires a cytoplasmic polyadenylation element (CPE), located upstream of the polyadenylation site, which is recognized by specific proteins (CPEBs).

Effects of 3'-terminal structures on tranlational efficiency. Although there are many instances in which specific structures in the 3'UTR have been shown to affect translational efficiency in C. elegans and other organisms (Kuersten and Goodwin, 2003), the molecular interactions responsible for these effects are only partially understood. In Xenopus oocytes, CPEB binds and sequesters eIF4E through an intermediary protein, Maskin (Mendez and Richter, 2001). In Drosophila embryos, there is a similar interaction between the 3'UTR-binding factor Smaug and eIF4E, mediated by another protein, Cup (Nelson et al., 2004). However, in C. elegans the translational component(s) involved in GLD-1-mediated regulation remain unknown (Jan et al., 1999; Marin and Evans, 2003; Lee and Schedl, 2004). The other 3'-terminal element, the poly(A) tract, increases the rate of translational initiation in yeast and plants due to the binding of PABP to a specific site near the N-terminus of eIF4G (Tarun and Sachs, 1996; Le et al., 1997; see Figure 2). Poly(A) stabilizes the PABP•eIF4G•eIF4E complex, which in turn leads to enhanced translational re-initiation (Wakiyama et al., 2000). As discussed below, there are several regulatory mechanisms in C. elegans that involve changing the poly(A) length.

Covalent modification of a translation factor. The unfolded protein response (UPR) is a transcriptional and translational signaling pathway activated by the accumulation of unfolded proteins in the endoplasmic reticulum (ER; Zhang and Kaufman, 2004). This involves activation of an eIF2α kinase, PEK, causing global inhibition of translation initiation and allowing time to remedy the folding problem. C. elegans PEK-1 was expressed in yeast and found to inhibit growth by hyperphosphorylation of eIF2α and inhibition of eIF2B (Sood et al., 2000). UPR gene transcription and survival upon ER stress also requires ire-1-mediated splicing of xbp-1 mRNA (Shen et al., 2001). ire-1/xbp-1 acts with pek-1 in complementary pathways that are essential for worm development, survival, and ER homeostasis. Furthermore, pek-1 mutants have shortened life spans (Harding et al., 2003).

Sequestration of a translation factor. TOR is a highly conserved protein kinase that controls cell growth and division in eukaryotes. In mammals, mTOR regulates translation by phosphorylation of p70 S6 kinase (S6K) and the eIF4E-binding protein 4E-BP1 (Gingras et al., 2001). The latter event releases eIF4E from a sequestered form, making it available to bind eIF4G (Figure 2). Raptor is a mTOR-binding protein that is necessary for the mTOR-catalyzed phosphorylation of 4E-BP1 and S6K (Hara et al., 2002). In yeast, TOR also regulates translation through eIF4E by a mechanism involving the phosphatases Sit4 and PP2A and the phosphatase-binding protein Tap42 (Jiang and Broach, 1999). Tip41 negatively regulates TOR by binding and inhibiting Tap42 (Jacinto et al., 2001).

In C. elegans, cTOR (let-363) deficiency causes developmental arrest and intestinal atrophy (Long et al., 2002). The phenotype resembles that of RNAi knockdown of eIF4G, eIF2α, and eIF2β, but not S6K, Tip41, or Tap42. RNAi of Raptor (daf-15) yields an array of phenotypes resembling those of cTOR knockout. Deficiency of both let-363 (Vellai et al., 2003) and daf-15 (Jia et al., 2004) extends adult lifespan, while mutations in either gene result in dauer-like larval arrest (Jia et al., 2004). To date, a C. elegans 4E-BP1 homolog has not been identified. The protein phas-1 (WP:CE30964; Agostoni et al., 2002; another name for 4E-BP1 is PHAS-I) is unrelated to mammalian and Drosophila 4E-BP1.

The presence of IFE-1 in P granules (Amiri et al., 2001) and its absence from initiation complexes (Dinkova et al., 2005) may represent another regulatory mechanism involving sequestration of a translation factor. A number of mRNAs have been found in P granules, some of which have been shown to be translationally controlled (Schisa et al., 2001). P granules also contain four GLH RNA helicases similar to eIF4A (Gruidl et al., 1996; Kuznicki et al., 2000). In Drosophila, VASA is a germline-specific ATP-dependent RNA helicase, homologous to the GLH proteins, that is required for translation of at least two mRNAs, gurken and nanos, through interaction with eIF5B (Styhler et al., 1998; Johnstone and Lasko, 2004). Also, eIF5A is required for PGL-1 localization (Hanazawa et al., 2004). The presence of all these translational components in P granules may signify regulation of germline-specific mRNA translation.

The translation of some C. elegans mRNAs is regulated by 3'UTR-binding proteins that alter poly(A) length (Kuersten and Goodwin, 2003). There are four CPEB homologs, CBP-1, CPB-2, CPB-3, and FOG-1 (Luitjens et al., 2000). CPB-1 is essential for progression through meiosis and is present in germ cells just before spermatogenesis. CPB-1 physically interacts with FBF, another RNA-binding protein and 3' UTR regulator. Similarities between FOG-1 and the CPEB proteins of Xenopus and Drosophila suggest that FOG-1 controls germ cell fates by regulating the translation of specific messenger RNAs.

A specific mRNA that undergoes regulated changes in poly(A) length is tra-2 mRNA, whose translation must be repressed for male development. A possible mechanism of action is suggested by the observation that TGEs control the length of the poly(A) tract in a Xenopus system (Thompson et al., 2000). Similarly, a 5-nt element in the 3'UTR of fem-3 mRNA controls the length of the poly(A) tract (Ahringer and Kimble, 1991). fem-3 must be downregulated to allow the switch from spermatogenesis to oogenesis. FBF and Nanos-3 interact with each other and repress translation of fem-3 mRNA (Kraemer et al., 1999). GLD-3 also physically interacts with FBF to interfere with FBF binding to the 3'UTR of fem-3 mRNA (Eckmann et al., 2002). GLD-3 promotes the transition from mitosis to meiosis together with the putative GLD-2 poly(A) polymerase. FBF also binds specifically to elements in the 3'UTR of gld-3S mRNA and regulates gld-3 expression (Eckmann et al., 2004).

The translation of several other mRNAs is known to be regulated through the 3'UTR, but this cannot yet be explained within the framework of known translational control mechanisms. A 34-nucleotide region of the 3'UTR of glp-1 mRNA contains two regulatory elements, one that represses translation in germ cells and posterior cells of the early embryo, and one that inhibits repressor activity to promote translation in the embryo (Marin and Evans, 2003). GLD-1 binds to this repressor element. MEX-3 is also expressed in anterior blastomere cells and is required for repression of pal-1 mRNA translation (Hunter and Kenyon, 1996; Draper et al., 1996).

A different type of 3'UTR regulation is seen with the heterochronic genes, which are temporally controlled to specify the timing of developmental events. lin-14 and lin-28 encode proteins that are required for temporal execution of cell lineages during larval development. lin-14, lin-28, lin-42, and daf-12 mRNAs contain a conserved element in the 3'UTR, the loss of which causes a gain of function (Seggerson et al., 2002). let-7 encodes a temporally regulated 21-nucleotide RNA that is complementary to elements in heterochonic mRNA 3'UTRs and regulates their translation (Reinhart et al., 2000). A second regulatory RNA, lin-4, negatively regulates lin-14 and lin-28 through RNA-RNA interactions with their 3'UTR (Moss et al., 1997). lin-28 is repressed during normal development by a mechanism that acts on its mRNA after translation initiation (Seggerson et al., 2002).

As illustrated throughout this chapter, several key developmental pathways in C. elegans are predominantly governed by translational mechanism. However, translation per se has received little attention. Even though one can find C. elegans homologs for most of the RNAs and proteins that make up the translational machinery (Table 1), few of these have been isolated, functionally identified, or studied at the biochemical level. This deficiency obscures the molecular mechanisms responsible for well characterized physiological processes. Another shortcoming is the absence of a cell-free translation system, although such systems have been developed in two other nematodes, Ascaris lumbricoides (Maroney et al., 1995) and Ascaris suum (Lall et al., 2004). The most interesting unexplored questions may involve changes in the translation of individual mRNAs in various physiological conditions or after genetic manipulations. Given the highly developed state of C. elegans genomics, it can be expected that future application of computational tools, including data visualization (Trutschl et al., 2005), will help detect new instances of translational control.