The major pathway of mRNA decay in yeast initiates with deadenylation, followed by mRNA decapping and 5′-3′ exonuclease digestion. An in silico approach was used to identify new proteins involved in the mRNA decay pathway. One such protein, Edc3p, was identified as a conserved protein of unknown function having extensive two-hybrid interactions with several proteins involved in mRNA decapping and 5′-3′ degradation including Dcp1p, Dcp2p, Dhh1p, Lsm1p, and the 5′-3′ exonuclease, Xrn1p. We show that Edc3p can stimulate mRNA decapping of both unstable and stable mRNAs in yeast when the decapping enzyme is compromised by temperature-sensitive alleles of either the DCP1 or the DCP2 genes. In these cases, deletion of EDC3 caused a synergistic mRNA-decapping defect at the permissive temperatures. The edc3Δ had no effect when combined with the lsm1Δ, dhh1Δ, or pat1Δ mutations, which appear to affect an early step in the decapping pathway. This suggests that Edc3p specifically affects the function of the decapping enzyme per se. Consistent with a functional role in decapping, GFP-tagged Edc3p localizes to cytoplasmic foci involved in mRNA decapping referred to as P-bodies. These results identify Edc3p as a new protein involved in the decapping reaction.
IN eukaryotic cells, mRNA turnover and its regulation are essential determinants of gene expression. In Saccharomyces cerevisiae, a major pathway of mRNA turnover for both stable and unstable transcripts initiates with deadenylation of the 3′-polyadenosine [poly(A)] tail (Muhlrad and Parker 1992; Decker and Parker 1993). The deadenylation of transcripts is followed by the removal of the 5′ 7meGTP cap and, subsequently, exonucleolytic digestion of the transcript occurs in the 5′-3′ direction (Decker and Parker 1993; Hsu and Stevens 1993; Muhlrad et al. 1994, 1995; Beelmanet al. 1996). Several important proteins involved in different steps of the mRNA decay pathway have been characterized (Tucker and Parker 2000). The deadenylation step in the decay pathway requires Ccr4p and Pop2p, which are components of the major cytoplasmic deadenylase in S. cerevisiae (Daugeronet al. 2001; Tucker et al. 2001, 2002; Chenet al. 2002). After removal of the tail, the Dcp1p/Dcp2p complex decaps mRNAs at their 5′ end (Beelmanet al. 1996; Dunckley and Parker 1999; Steiger and Parker 2002; Steigeret al. 2003), which then allows Xrn1p, the major yeast 5′-3′ exonuclease, to rapidly degrade the body of the transcript.
Decapping is an important step in the mRNA decay pathway as it allows the final degradation of the mRNA and is regulated by a number of proteins (Muhlrad et al. 1994, 1995; Tucker and Parker 2000). Apart from the decapping enzyme, which consists of Dcp1p and Dcp2p, several proteins including Pat1p, Dhh1p, the Sm-like (Lsm) complex (Lsm1p-Lsm7p), and Edc1p and Edc2p are enhancers of the decapping rate (Hatfieldet al. 1996; Bonnerotet al. 2000; Bouveretet al. 2000; Tharunet al. 2000; Colleret al. 2001; Dunckleyet al. 2001; He and Parker 2001; Tharun and Parker 2001; Fischer and Weis 2002). Lsm1p is part of the seven-member Lsm complex that binds to mRNA and is required for efficient decapping (Tharunet al. 2000). Pat1p is known to interact with the Lsm complex and is an enhancer of decapping (He and Parker 2001). Dhh1p is a member of the DEAD-box helicase proteins and is also required for efficient decapping (Colleret al. 2001). Edc1p and Edc2p are related proteins that are known to affect the decapping enzymes directly and are both enhancers of decapping (Dunckleyet al. 2001; Schwartzet al. 2003; Steigeret al. 2003). Whether additional proteins are involved in decapping is not yet known.
One way to find new factors affecting mRNA decay is to utilize databases of protein-protein interactions on the basis of high-throughput two-hybrid screens, systematic mass spectrometry, and multidimensional protein identification technology screens to find proteins that show numerous interactions with known decapping factors. To do this, we examined the S. cerevisiae genomic-scale data and observed that the yeast open reading frame Yel015W showed numerous interactions with the Dcp1p/Dcp2p complex, Dhh1p, Xrn1p, Pat1p, and members of the Lsm complex (Fromont-Racine et al. 1997, 2000; Schwikowskiet al. 2000; Uetzet al. 2000; Hishigakiet al. 2001; Itoet al. 2001; Gavinet al. 2002; Giaeveret al. 2002; Hoet al. 2002). This in silico analysis predicts Edc3p to be involved in mRNA turnover. In this study, we present experimental evidence that although Edc3p is not required for decapping, Edc3p can stimulate the rate of decapping in vivo and is found in sites of decapping in the cytoplasm.
MATERIALS AND METHODS
Sequence analysis: The Edc3p homolog in S. pombe was used as a query to identify homologs in other organisms using the BLAST algorithm (Altschulet al. 1990). Protein sequences of Edc3p and its homologs were aligned using the CLUSTAL W program (Thompsonet al. 1994). The domains were identified using the BLOCKS program (Henikoffet al. 1995) and the position of the domains in the protein was used to draw a cartoon of the Edc3p and its homologs in Figure 1. The accession numbers for the proteins are listed in the legend to Figure 1.
Yeast strains and plasmids: All strains used in this study are listed in Table 1. The NEO deletion cassette from the genomic DNA of the commercially available yel015wΔ mutant strain (Research Genetics, Birmingham, AL) was PCR amplified using the oligonucleotides oRP 1202 (5′-GAA GCA TAT CGT AAG CAC AC-3′) and oRP 1203 (5′-GTG AGA CAC TGG CCT CGT CTG-3′). This amplified PCR fragment was used for transformation of the strains yRP840 and yRP841 to get yRP1745 and yRP1746 strains by homologous recombination.
Growth phenotype studies were done with the dcp2-7ski3Δ edc3Δ and dcp1-2 ski8Δ edc3Δ strains, which were obtained by crossing edc3Δ with dcp2-7 ski3Δ and dcp1-2 ski8Δ strains and dissection of the diploid strains. The triple mutants were then grown along with the wild type (WT) and single and double mutants at 24°, 30°, and 37° for 3 days.
The construction of the Lsm1p-RFP plasmid is described in Sheth and Parker (2003). The EDC3-GFP strain construction was done as described in Sheth and Parker (2003). The Edc3p-GFP strain was transformed with the Lsm1p-RFP plasmid for the colocalization studies.
RNA procedures: All RNA analyses were performed as described in Muhlrad et al. (1994). For half-life measurements cells were grown to midlog phase containing 2% galactose. Cells were harvested and transcription was repressed by the addition of media containing 4% glucose. Aliquots were taken over a brief time course and frozen. Yeast total RNA extractions were performed as described in Muhlrad et al. (1994) and Caponigro et al. (1993). RNA was analyzed by running 10 μg of total RNA on either a 1.5% formaldehyde agarose gel as in Figures 2 and 3 or a 6% polyacrylamide/7.5 m urea gel as in Figure 4. All Northerns were performed using radiolabeled oligo probes directed against the MFA2pG reporter (oRP140) and the 7S RNA (oRP100). For the CYH2 Northern blot, a random-prime radiolabeled CYH2 complementary DNA probe was used (Tharunet al. 2000). Half-lives were determined by quantitation of blots using a Molecular Dynamics (Sunnyvale, CA) Phosphorimager. Loading correction for quantitations was determined by hybridization with oRP100, an oligo directed to the 7S RNA, a stable RNA polymerase III transcript. Northerns shown in Figures 2, 3, and 5 were done at least three times.
A primer extension analysis was performed as described in Hatfield et al. (1996). Oligo oRP131, which is complementary to the 5′ end of the PGK1 transcript, was radiolabeled and added to 20 μg of total RNA. Extension was performed using superscribe reverse transcriptase (GIBCO BRL, Gaithersburg, MD). The reactions were analyzed on 6% polyacrylamide gels and visualized using autoradiographs. The primer extension data shown in Figure 6 are representative of three independent experiments.
Confocal microscopy: The cells were grown to an OD of 0.4 at 30° in synthetic medium containing 2% dextrose, washed, resuspended in a smaller volume of the media, and observed. Observations were made as described in Sheth and Parker (2003). The Edc3p-GFP strain with the LSM1-RFP plasmid was grown to an OD of 0.4 at 30° on synthetic media containing 2% dextrose without uracil. The cells were collected and observed as described above. All images shown in Figure 7 are representative of at least three independent observations.
Edc3p belongs to a novel class of conserved proteins: To evaluate the possible significance of Edc3p, we first examined databases for homologs in other organisms. By a series of BLAST (Altschulet al. 1990) searches we were able to identify related proteins in several other organisms, including Schizosaccharomyces pombe, S. castelli (not shown), S. kluyvri (not shown), Homo sapiens, mouse, Drosophila, and Anopheles (Figure 1). Comparative sequence analysis showed that the S. cerevisiae Edc3p showed 21-28% identity with its homologs in these organisms. Interestingly, all the homologs of Edc3p are proteins of unknown function.
To determine which regions of Edc3p might be important for its function, we looked for conserved regions by several approaches. When Edc3p and its homologs were aligned using BLAST, a global alignment algorithm, no striking regions of similarity were observed. However, on using the alignment algorithms, CLUS-TALW (Thompsonet al. 1994) and BLOCKS (Henikoffet al. 1995), which look for regions of similarity over short stretches in the proteins, the EDC3 class of proteins showed five conserved domains arranged in the same order in all the homologs except the S. pombe homolog, where domain V is rearranged and positioned after domain I (Figure 1). The conservation of these regions supports the identification of these proteins as being homologs and suggests that these are functionally significant regions within the Edc3p.
Edc3p is not rate limiting for mRNA decay: The extensive interactions/associations between Edc3p and different proteins involved in mRNA decay suggested that Edc3p might be involved in mRNA decay. To determine the role of Edc3p in mRNA decay, we created an edc3Δ strain, which grew normally at temperatures between 18° and 37°. We then analyzed the turnover of two reporter mRNAs in the edc3Δ strain. Analysis of mRNA degradation in these strains revealed no defect in the decay of either the unstable MFA2pG reporter mRNA (Figure 2) or the stable PGK1pG mRNA (data not shown). For example, the half-life of MFA2pG reporter mRNA was 4.6 min in the edc3Δ strain, which is not significantly different from the 4-min half-life of MFA2pG in the wild-type strain. This suggests that Edc3p either is not involved in mRNA decay or affects a step in decapping that is not rate limiting in vivo.
edc3Δ slows mRNA decay in strains compromised for the mRNA decay: Given the extensive interactions with decapping factors, we hypothesized that Edc3p was involved in mRNA decapping, but affected a step that was not rate limiting. This interpretation was consistent with previous results identifying multiple substeps in the decapping reaction (Schwartz and Parker 2000) and revealing that defects in the decapping activator proteins, Edc1 and Edc2, show a decay defect only when cells are partially compromised for mRNA decapping activity (Dunckleyet al. 2001). Therefore, to test if Edc3p has some role in mRNA decapping, we created double mutants where the edc3Δ was combined with several mutations affecting decapping at different steps in the process. Specifically, we combined edc3Δ with conditional alleles in subunits of the decapping enzyme, dcp1-2 and dcp2-7, which affect the actual activity of the decapping enzyme (Dunckleyet al. 2001; Schwartzet al. 2003). In addition, we created double mutants where the edc3Δ was combined with dhh1Δ, pat1Δ, or lsm1Δ. These latter mutations affect an early step in the decapping pathway that is distinct from the actual catalytic cleavage (Schwartz and Parker 2000; Schwartzet al. 2003; Sheth and Parker 2003).
We first analyzed the mRNA decay defects of the edc3Δ dcp1-2 and edc3Δ dcp2-7 mutant strains. These experiments were done at 24°, which is the permissive temperature for both dcp1-2 and dcp2-7 alleles. At this temperature, dcp1-2 and dcp2-7 alleles are very slightly defective in mRNA decay in vivo at the permissive temperature (Dunckleyet al. 2001; and Figure 3). When the edc3Δ was combined with either the dcp1-2 or the dcp2-7 background we observed a decrease in the rate of decay of the MFA2pG (Figure 3) and the PGK1pG mRNA (data not shown). In the dcp2-7 background, the edc3Δ slowed the MFA2pG mRNA decay by at least twofold so that the half-life for the MFA2pG mRNA was increased to 12 min compared to the 5 min in strains with just the dcp2-7 allele. The effect on MFA2pG half-life was not so dramatic (7.2 min) in the dcp1-2 strains deleted for the EDC3 gene compared to the dcp1-2 strain alone (4 min). These results indicate that when Dcp2p, and possibly Dcp1p, are partially defective, a role for Edc3p in mRNA decay can be observed. In contrast, when the edc3Δ was combined with the dhh1Δ, lsm1Δ, or pat1Δ, no affect on decay rate was observed (Figure 3). We interpret these results to indicate that Edc3p functions as an enhancer of mRNA decapping and genetically interacts most strongly with the actual decapping enzyme.
Loss of EDC3 slows growth in cells compromised for mRNA decapping: To obtain additional evidence that Edc3p affected decapping, we utilized a growth assay where the viability of the strains is dependent on efficient mRNA decapping. This assay is based on the observation that strains lacking both the major deadenylation-dependent decapping pathway and the alternative 3′-5′ degradation pathway are dead (Jacobset al. 1998; Van Hoofet al. 2000). This synthetic lethality can be made conditional by combining temperature-sensitive alleles of the DCP1 (dcp1-2) and DCP2 (dcp2-7) genes with strains having defects in the exosome-mediated 3′-5′ decay pathway such as ski3Δ or ski8Δ. For example, a dcp1-2 ski8Δ strain grows at 24°, but is dead at 33° (Jacobset al. 1998). Similarly, dcp2-7 ski3Δ grows at 24° and 30° but is dead at 37° (Dunckleyet al. 2001). These strains are conducive for studying effects of proteins that have a subtle effect on decapping. For example, if Edc3p reduces the function of the decapping enzyme, then edc3Δ should exacerbate the growth defect in the dcp1-2 ski8Δ and dcp2-7 ski3Δ strains and we would see a synthetic lethality at a temperature <33° or 37°.
We observed that combining the edc3Δ in strains with the dcp1-2 ski8Δ and dcp2-7 ski3Δ did lead to a clear exacerbation of the growth defects (Figure 4). At 24°, the dcp2-7 ski3Δ edc3Δ had an extremely slow growth phenotype compared to just the dcp2-7 ski3Δ. Moreover, the dcp2-7 ski3Δ edc3Δ failed to grow at 30° in comparison to the dcp2-7 ski3Δ that showed normal growth. Combination of edc3Δ with dcp1-2 ski8Δ also showed a growth defect phenotype at 30°, but almost normal growth at 24° in comparison to just the dcp1-2 ski8Δ strain that grows normally at both 30° and 24°. These growth defects correspond well with the mRNA decay phenotype of these mutants where Edc3p showed a greater defect when combined with the dcp2-7 allele than when combined with the dcp1-2 allele. These data provide a second line of genetic evidence for Edc3p affecting decapping.
Edc3p deletion affects the decapping step in mRNA decay: Mutations affecting the mRNA decay rate in yeast can be at any step during deadenylation, decapping, or 5′-3′ exonucleolytic degradation. The edc3Δ showed an mRNA decay defect phenotype only when combined in strains with either the dcp1-2 or the dcp2-7 mutation. Therefore, the most probable hypothesis is that Edc3p affects the decapping step of mRNA decay, which would be consistent with known Edc3p physical interactions. To determine what step in decapping Edc3p affects, we compared the decay of the MFA2pG mRNA in dcp2-7 strains to dcp2-7 edc3Δ strains on polyacrylamide Northern gels, where the rates of deadenylation and subsequent decay can be observed. On analysis of the MFA2pG mRNA, we observed that there was no substantial difference in the deadenylation rate of the MFA2pG mRNA among the wild type, edc3Δ, dcp2-7, or dcp2-7 edc3Δ strains (Figure 5). This indicates that Edc3p does not affect the deadenylation rate. However, we did observe that the decay of the deadenylated species was slower in the dcp2-7 edc3Δ as compared to dcp2-7 strains alone. This observation suggests that the Edc3p promotes either decapping or 5′-3′ exonucleolytic degradation in the dcp2-7 strain.
A defect in mRNA decapping can be distinguished from a defect in the exonucleolytic digestion by a primer extension assay. This assay takes advantage of the observation that strains defective in 5′-3′ degradation accumulate a decapped PGK1 transcript that is two nucleotides shorter at the 5′ end compared to that of transcripts that still have the cap (Hatfieldet al. 1996; Zuket al. 1999; Colleret al. 2001). Given this, we examined the 5′ ends of the PGK1 mRNA in edc3Δ, dcp2-7 edc3Δ, and various control strains including the xrn1Δ strain (lacking the 5′-3′ exonuclease) and dcp1Δ (lacking the decapping enzyme) strains. We observed that the dcp2-7 edc3Δ strain, where there is a defect in decay after deadenylation, contains a full-length transcript similar to only the wild-type and dcp1Δ strains. Further, the edc3Δ mutants did not have a defect exclusively in the 3′-5′ exonuclease activity. In contrast, the xrn1Δ strain shows the full-length transcript and a -2-nucleotide (nt) species (Figure 6). The above observations provide evidence that Edc3p affects the decapping step of mRNA degradation. However, it should be noted that we cannot rule out that Edc3p also affects the 5′-3′ degradation step.
Edc3p localizes to cytoplasmic foci similar to the P-bodies: It has been recently shown that the proteins involved specifically in mRNA decapping in yeast and mammals are concentrated in cytoplasmic foci called P-bodies (Ingelfingeret al. 2002; Lykke-Andersen 2002; Van Dijket al. 2002; Sheth and Parker 2003). Given this observation, if Edc3p is involved in the mRNA decapping step, we would predict that Edc3p would be present in P-bodies. To assess this possibility, we constructed a C-terminal green fluorescent protein (GFP) fusion of Edc3p and localized the protein in the cell by confocal microscopy. We observed that Edc3p-GFP is present in discrete foci in the cytoplasm similar to P-bodies (Figure 7A). The identity of these foci as P-bodies is confirmed by the colocalization of Edc3p-GFP with red fluorescent protein (RFP)-tagged Lsm1p (Figure 7B). Thus, like other proteins involved in mRNA decapping, Edc3p also localizes to the P-bodies. Interestingly, we observed a greater number of Edc3p-GFP foci as compared to the Lsm1p-RFP foci (Figure 7B). One possibility is that different forms of P-bodies exist, some of which are lacking Lsm1p, or another possibility is that we failed to detect Lsm1p in all P-bodies due to the lower signal from the Lsm1p-RFP fusion.
Additional supporting evidence that the Edc3p-GFP foci are P-bodies comes from the observation that trapping mRNAs in polysomes using cycloheximide inhibits decapping and leads to loss of P-bodies after treatment with the drug (Sheth and Parker 2003). Just like P-bodies, the Edc3p-GFP foci also disappeared rapidly on treatment of the cells with cycloheximide (Figure 7A).
Edc3p does not affect nonsense-mediated decay: The decapping complex is also involved in nonsense-mediated decay (NMD), which is characterized by deadenylation-independent decapping; therefore we asked if Edc3p plays a role in NMD (Wiluszet al. 2001; Lykke-Andersen 2002). To determine if Edc3p affects NMD, the levels of the CYH2 pre-mRNA in edc3Δ and dcp2-7 edc3Δ strains were analyzed. A known NMD substrate is the CYH2 pre-mRNA, which is inefficiently spliced and contains a pretermination codon (PTC; Heet al. 1993; Hilleren and Parker 1999). Analysis of CYH2 mRNA at steady state showed that the edc3Δ did not lead to increased levels of pre-CYH2 mRNA either in wild-type strains or in combination with dcp1-2 and dcp2-7 alleles (Figure 8). These results suggest that Edc3p does not play a significant role in NMD.
Several observations indicate that Edc3p has a role in mRNA decay and specifically enhances the function of decapping. First, Edc3p has been shown in numerous in vivo and in vitro genomic studies to physically interact with several mRNA decay factors (Fromont-Racineet al. 2000; Uetzet al. 2000; Itoet al. 2001; Gavinet al. 2002; Hoet al. 2002). Second, deletion of the EDC3 gene leads to a defect in mRNA degradation in strains carrying conditional alleles of Dcp1p and Dcp2p, even though the analysis is done at permissive temperatures where the dcp1-2 and dcp2-7 alleles have little or no effect on mRNA decay rates (Figure 3). Moreover, these edc3Δ dcp2-7 strains accumulated deadenylated full-length MFA2pG reporter mRNA compared to the WT, edc3Δ, and dcp2-7 allele alone (Figure 5). Third, edc3Δ exaggerated growth defects and caused synthetic lethality in strains containing the dcp1-2 and dcp2-7 alleles in combination with a block to the 3′-5′ decay pathway (Figure 4). Finally, an Edc3p-GFP fusion protein localizes in P-bodies (Figure 7), which are specialized cytoplasmic foci containing decapping proteins (Sheth and Parker 2003). These observations together argue that Edc3p interacts with Dcp2p and Dcp1p to stimulate the mRNA decapping rate.
In principle, Edc3p could enhance decapping by acting at any of several distinct steps in the decapping process. Although poorly understood, previous experiments have suggested that decapping will require at least three distinct, but possibly related steps: (1) loss of the translation initiation complex bound to the 5′ end of the mRNA, (2) assembly of a decapping complex and localization within P-bodies, and (3) the actual catalytic step of decapping (Schwartz and Parker 2000; Tharun and Parker 2001; Sheth and Parker 2003).
Three observations are consistent with Edc3p primarily affecting the final step of enzymatic decapping, either by stabilizing Dcp1p/Dcp2p or by affecting their function in some manner. First, the edc3Δ shows only synergistic effects with other defects in the decapping activity and does not show any phenotypic interaction with deletions in the LSM1, PAT1, and DHH1 genes, whose products are likely to affect the first two steps in the decapping process. Second, on the basis of copurification, Edc3p directly interacts with Dcp1p/Dcp2p (Fromont-Racineet al. 2000; Uetzet al. 2000; Itoet al. 2001; Gavinet al. 2002; Hoet al. 2002). Third, the edc3Δ does not have a decay defect and previous results suggest that the actual decapping step in decay is not normally rate limiting. Evidence that the enzymatic decapping step is not rate limiting in vivo in our growth conditions is that the dcp1-2, dcp2-7, edc1Δ, and edc2Δ mutations all strongly affect mRNA decapping activity in vitro, yet have little effect on overall decay rate in vivo (Dunckleyet al. 2001; Schwartzet al. 2003). However, it should be noted that decapping per se could be made the rate-limiting step in decapping in vivo (as in a strong dcp1 or dcp2 allele).
The Edc3p is part of an emerging set of proteins that perform the mRNA decapping step. Proteins involved in mRNA decay can be divided into three functional and phenotypic categories. The first class of proteins consists of Dcp1p and Dcp2p, which are required for the decapping enzyme to function, and strains lacking these proteins show a complete block to decapping (Beelmanet al. 1996; Dunckley and Parker 1999; Tharun and Parker 1999). The second group of decapping regulatory proteins (Lsm1-7p, Pat1p, and Dhh1p) appears to constitute a general activator of decapping complex. Strains lacking Lsm1-7p, Dhh1p, or Pat1p show a partial block to decapping for multiple mRNAs. Edc3p belongs to the third class of decapping factors, which, although they can affect the decapping process, are not normally rate limiting for decapping. The proteins Edc1p and Edc2p are also in this class.
Recent results suggest that the EDC family of proteins may also be important in the control of a subset of mRNAs or in the control of decapping under different conditions. For example, Edc1p has been shown to be important for growth during a shift from dextrose to glycerol, suggesting that it may be affecting specific mRNAs necessary for this carbon-source shift (Schwartzet al. 2003). In addition, recent results show that Edc3p specifically affects the decapping of the RPS28B mRNA, possibly as part of an autoregulatory loop (G. Badis, C. Saveanu, M. Fromont-Racine and A. Jacquier, unpublished results). Given these results, the emerging picture of the EDC family of proteins is that they will play a general role in assisting decapping and have been coopted for the control of specific mRNAs. An unresolved issue is how broad the specific role of these proteins will be in the cell.
We thank the Parker laboratory for helpful discussions; Ujwal Sheth for the Edc3p-GFP strain and the LSM1-RFP plasmid; and Kristian Baker, Jeffery Coller, Carolyn Decker, Tharun Sundaresan, and Daniela Teixiera for experimental and intellectual assistance. We also thank Alain Jacquier for sharing unpublished results. A National Institutes of Health grant (GM45443) and funds from the Howard Hughes Medical Institute supported this work.
Communicating editor: P. Anderson
- Received July 8, 2003.
- Accepted October 27, 2003.
- Copyright © 2004 by the Genetics Society of America