eat mutants, which have feeding defective phenotypes, were isolated many years ago (Avery, 1993). However, eight eat mutants (representing six genes) are not known molecularly (Table 1). To find out these genes, we used whole-genome sequencing (Sarin et al., 2008). Because we had approximate locations from classical genetic mapping, we compared reference and mutant whole-genome sequencing results in the regions to which each gene was known to map.

eat-8(ad599) had deletions and eat-8(ad697) had a missense mutation in Y82E9BR.16. This gene encodes a transporter, a member of the solute carrier family 22 superfamily. Mutations affecting similar proteins are associated with increased risk for Crohn’s disease in Homo sapiens (Newman et al., 2005). We are cloning Y82E9BR.16 to verify its identity as eat-8.

Two independent eat-9 mutants, eat-9(ad1129) and eat-9(e2337), had nonsense mutations at different locations in the coding region of F09C3.2. The eat-9 mutant phenotype was rescued by introducing wild type F09C3.2. When rescued mutants lost wild type F09C3.2, the mutant phenotype was recovered (Fig 1). F09C3.2 encodes a member of the haloacid dehydrogenase (HAD) superfamily (Seifried et al., 2012). We are studying what kind of HAD are related to the irregular pumping phenotype of eat-9.

eat-10 had a frame shift mutation in grld-1, orthologous to human RBM15 and Drosophila NITO, which affects glutamate receptor expression and extends lifespan about 10% by RNAi (Wang et al., 2010; Samuelson et al., 2007). We are comparing the eat-10 mutant phenotype with the known grld-1 mutant phenotypes.

eat-15(ad602) had a deletion in K10D3.4, which encodes a serine protease inhibitor. We cloned and made transgenic worms with this gene and are planning to measure the pumping rate and growth rate and compare the mutant phenotypes.

eat-13(ad522) had a deletion of over 70 kbp, which removed at least 19 genes from R07B1.5 to tsp-21 on chromosome X. We are planning fosmid and cosmid rescue experiments to identify which genes are related to the mutant phenotype. eat-14(ad573) had a 50 kbp deletion which removed at least eight genes from rsd-3 to F54F7.2 on chromosome X. We will also perform fosmid rescue experiments on this mutant. It is of course possible that the mutant phenotypes depend on multiple genes.

In conclusion, we used whole-genome sequencing method to find candidate mutations in 8 eat mutant strains previously believed to define 6 genes. We need more results to verify some of these candidates. However, these results, together with the identification eat‑17 as F01G12.6 and T24D11.1 (see Straud and Avery, this issue WBG) are likely to complete the molecular identification of all the eat genes.

In the closing years of the last century, for no better reason than that it was easy to do, we undertook a screen for suppressors of eat‑5. EAT‑5 is an innexin necessary for coupling the anterior half of the pharynx to the posterior. In an eat‑5 null mutant the corpus still pumps, and the terminal bulb (TB) still pumps, but they are no longer synchronized, and TB pumping is. eat‑5 grows almost as well as wild-type on E. coli HB101, but on DA837 (an E. coli strain derived from OP50) eat‑5 L1s arrest for a variable time, up to many days.

We prepared F1 eggs by basic hypochlorite treatment of EMS-mutagenized eat‑5 parents, grew them to adulthood on HB101, then plated their F2 eggs on DA837, about 100/plate. Screen statistics were as follows:

¹ Mean number of phenotypically mutant F2s per mutation in the F1, (recessive) or (dominant).

² Probability that a mutation present in the F1 is detected, P=1-e^-μ.

³ Effective number of genomes screened, N=2F1P.

⁴ Mean number of times a mutation is isolated, if it is found at all, .

A mutant line was established from each plate that starved by day 13. In addition, we noticed one plate on which most of the worms were adults rather than L1s and established a line from it (ad1614), even though it didn’t starve until day 16. We thought this mutation might allow escape from the L1 arrest but have other adverse phenotypes that decreased its growth rate.

From these two screens we got a total of 13 sef (suppressor of eat five) mutations. 9 recessive mutations mapped to a single complementation group on LGI, provisionally called sef‑1. sef‑1 mutants had normally synchronized corpus and TB contractions. 3 mutations were dominant, and one of these mapped to LGIII. The remaining mutation, ad1614, was recessive but difficult to map because of the weak phenotype. It looked like something we had seen before. Although corpus and TB were still unsynchronized in eat‑5; ad1614 worms, TB pumping was about twice as fast as in eat‑5. eat‑5; slo‑1 worms have the same phenotype, because M4 synapses onto TB muscle are activated by loss of the K⁺ channel encoded by slo‑1 (Chiang et al., 2006).

We sequenced the genomes of 9 sef mutant strains: 5 carrying sef‑1 mutations, 3 carrying the dominant sef mutations, and ad1614. Unsurprisingly, the ad1614 mutant had a splice donor mutation in slo‑1. All 5 sef‑1 mutants had mutations in cfi-1 LGI. There were 3 distinct mutations: a splice donor mutation, a splice acceptor mutation, and a nonsense mutation. Based on linked SNPs, these 5 mutations comprise 4 independent events. CFI‑1 is a transcription factor expressed in pharyngeal muscle and other places (Shaham and Bargmann, 2002 ). One of many plausible hypotheses is that in the absence of CFI‑1 an innexin that can replace EAT‑5 is expressed in pharyngeal muscle.

The dominant mutations were most unexpected. All 3 (2 independent) carry the identical A24V mutation in dod‑6 LGIII. Little is known of dod‑6 function. It was identified as a gene downstream of daf‑16, and RNAi slightly decreases lifespan of daf‑2 mutants (Murphy et al., 2003). It encodes a tiny 60 amino acid protein. The N-terminus looks like a signal sequence. The C-terminus is similar to a Metridin-like ShK domain (Figure 1). The eponymous proteins are 33 amino acid sea anemone toxins that block voltage-gated K⁺ channels. A24 is near the end of the putative signal sequence, 2 amino acids upstream of the toxin domain.

Finding that the same screen yields loss-of-function mutations in a K⁺ channel and dominant mutations in something that looks like a K⁺ channel blocker suggests an obvious hypothesis: DOD‑6(A24V) blocks SLO‑1. It is not equivalent to complete loss of SLO‑1, since eat‑5; dod‑6 mutants are healthier and grow faster than eat‑5; slo‑1. The action of DOD‑6(A24V) might be limited to particular times or places, or might be something other than a simple block of all SLO‑1 isoforms, or might have other targets. The hypothesis that DOD‑6(A24V) blocks SLO‑1 makes many experimental predictions, some of which we plan to test.