In an effort to identify homologues of NCoR/SMRT in the C. elegans
proteome, we performed BLAST and PSI-BLAST searches in multiple protein databases 
. Searches with human NCoR and SMRT sequences returned the sequence annotated as GEI-8 (UniProt GEI8_CAEEL, E value 2e-10), as the best hit. In the reciprocal BLAST, NCoR and SMRT appeared likewise as the best hits for GEI-8 within the human proteome. Although only a small fraction of the entire protein sequence (~7%) was retrieved by Blast searches, nearly complete protein sequences were recovered in PSI-BLAST after the third iteration. Six GEI-8-related proteins from other Nematoda species (C. elegans, C. brenneri, C. briggsae, C. remanei, C. japonica, Loa loa
and Brugia malayi
) were aligned and submitted as a query in PSI-BLAST (). Sequences were extracted from databases UniProt, Wormbase and Ensembl. Entries for C. japonica
and X. tropicalis
were corrected according to NCBI nucleotide sequences using the GeneWise program 
. An alignment of these nematode GEI-8-related proteins with human NCoR was obtained in the second iteration.
Comparison of N-terminal regions of GEI-8-related proteins to NCoR/SMRT.
Multiple sequence alignments resulting from PSI-BLAST were further improved using the profile-to-profile alignment method (PSI-Coffee) 
, however, its quality remained ambiguous in several regions across the protein. All NCoR homologues contain long stretches of low complexity (e.g. 23% of amino acids in GEI-8 or 13% in human NCoR1) that are variable in length. The well conserved N-terminal region from representative Metazoa/Fungi
NCoR/SMRT is shown in . The sequence conservation in the C-terminal domains is much lower; all sequences contain many insertions, deletions, prolines, serines and oligoGlu residues that vary between species. This C-terminal variability is evident even within the alignment of the GEI-8-related proteins from the phylogenetically related Caenorhabditis
species. We also used ClustalW2.0 for identification of putative interaction motifs near the C-terminus. NCoR and SMRT bind nuclear hormone receptors by NR-binding domains consisting of three and two CoRNR-box sequences respectively. The CoRNR-box sequence was previously defined as L.x.x.x.I.x.x.x.I/L 
; I/L.x.x.I/V.I 
; L/V.x.x.I/V.I 
. We identified two putative CoRNR-box like sequences in GEI-8 (). The predicted GEI-8 sequence also contains two glutamine rich regions 
that also might serve as interaction domains.
The most conserved N-terminal regions of the GEI-8 related sequences contain both the DAD and SANT domains with their location and the positions of the conserved helices shown in . We noted that GEI-8 and related sequences preserve all features known to be essential for correct functioning of NCoR/SMRT as an HDAC-dependent transcriptional corepressor 
(highlighted in ). These include the number of helices, their topology, the conserved amino acids needed for the integrity of the structure and for the interaction with HDAC and, most importantly, the K159 residue in the loop between helices H1 and H2 that is indispensable for the activation of HDAC3. The helix H0, known to be very irregular in human SMRT, is probably also present although it contains a two amino acid insertion between the second and third helical turn. Based on the sequence analysis, we concluded that GEI-8 bears all major features identified in other NCoR/SMRT orthologues in annotated genomes from other species and is the NR corepressor and NCoR/SMRT orthologue in C. elegans
The C. elegans gei-8
gene is located on chromosome III and gives rise to three predicted isoforms with mRNAs ranging from 5.3 to 5.6 kb (WormBase WS195). All predicted isoforms contain two SANT domains that could provide DNA and HDAC interaction functions ().
Using primers based on predicted cDNA sequences of gei-8
isoforms, we cloned three overlapping regions corresponding to gei-8
cDNAs and confirmed the expression of predicted isoform gei-8a
containing both SANT domains and two putative CoRNR-box like motifs ().
cDNA clones also revealed that exon 12 can be removed and exon 16 is modified by alternative splicing ()
; a spliced region of the same location and size as our cDNA clone was also detected by polyA mRNA expression profiling 
. Depending on the presence or absence of exon 12, the size of gei-8
cDNA is 5043 bp (gei-8d
) and 5292 bp (gei-8e
), giving rise to either a 1680 or 1763 amino acid long GEI-8 isoforms. We have not cloned the region containing the complete predicted protein encoded by inclusion of exon 16, however, polyA mRNA expression profiling data suggest that this variant is expressed. We confirmed the transcription of the gei-8a
5′ untranslated region (5′ UTR) and its trans-splicing to SL1 by PCR assays 
. Expression of gei-8b
was not detected using primers directed at predicted exons 1 to 3; our results are consistent with polyA mRNA expression profiling data generated by modENCODE ()
We quantified gei-8 expression in individual embryonic and larval stages by real-time qPCR using cDNA prepared from synchronized populations of wild-type animals. We separately analyzed a region common for all predicted isoforms (gei-8a, b, c) as well as a gei-8a-specific region. We detected expression after probing both regions in all developmental stages at constant relative levels with the exception of the fourth larval (L4) stage where we observed a 2-fold increase for both (). We concluded that gei-8a was expressed throughout development, with its late larval increase possibly reflecting expression in the maturing germline.
Normalized expression of gei-8a.
The spatial expression pattern of gei-8
was studied using three different gei-8::gfp
constructs based on the predicted start of transcription for gei-8b
(promoter 1), the detected start of transcription for gei-8a
(promoter 2), and an overlapping region covering both promoters (promoter 3) ()
. pPD95.69 and pPD95.67 promoterless, nuclear localization signal-containing vectors were used for the promoter 1 and promoter 2 constructs, respectively. Expression from promoter 3 was studied by the PCR fusion-based SOEing approach 
The promoter 1 reporter gene consisted of 1.8 kb upstream of the predicted gei-8b
start codon and 222 bps of predicted exon 1. Its expression started in embryos at the comma stage in a ubiquitous pattern and was present in all larval stages. In larvae, the expression was detected in pharyngeal and tail neurons, intestinal cells, egg-laying muscles and the anal depressor ()
. The promoter 2 reporter gene construct consisted of 2.3 kb upstream of the predicted gei-8a
start codon and included exon 1 and 64 bp of exon 2. The expression of this reporter gene was observed in all larval stages starting at the L1 stage and continuing through adulthood where expression was primarily observed in neurons of the pharyngeal nerve ring, head neurons, tail neurons and the egg-laying muscles.
The promoter 3 reporter gene construct contained 6.2 kb upstream of the predicted gei-8a
start codon, covering both promoter regions 1, 2 and exons 1, 2 and a part of exon 3; GFP sequences were derived from pPD95.75 by SOEing 
and did not contain a nuclear localization signal. Expression of this reporter gene started at the embryonic comma stage. Larval expression was detected in pharyngeal neurons, ventral and dorsal nerve cords, tail neurons, egg-laying neurons, and egg-laying muscles. In males, GFP was observed in male-specific tail ganglia and rays. Typical examples of GEI-8::GFP cell- and tissue-specific expression are shown in .
Taken altogether, our reporter gene expression results defined multiple and distinct cis-acting regulatory regions of gei-8
that drive similar expression patterns that are present throughout development and predominantly in neurons. Expression in the germline would not be revealed by this strategy because transgenes are usually silenced in the germline 
. However, we noted that gei-8
expression in the germline has been detected by Y. Kohara’s in situ hybridization results accessible in the Kohara in situ database NEXTDB (http://nematode.lab.nig.ac.jp
Analysis of gei-8 expression using transgenic lines.
Loss of gei-8 Results in Mutant Phenotypes
We obtained the VC1213 strain harboring a gei-8(ok1671)
deletion allele generated by the C. elegans
Knockout Consortium. The mutation was initially characterized as a 1095 bp deletion/45 bp insertion affecting exons 7 and 8 of gei-8a,
removing the intron between them. We verified the size and location of the deletion by PCR genomic amplification from mutant animals and showed that the inserted sequences are identical to a 45 bp region from exon 7 starting at position 1550 of the predicted gei-8a
isoform cDNA sequence. Sequencing the gei-8(ok1671)
cDNA revealed a stop codon present in the gei-8(ok1671)
transcript at position 663, giving rise to a predicted protein containing SANT1 and SANT2 domains, but missing the majority of the putative NR interaction sites at the C-terminus of the protein. The mutant mRNA was detected in homozygous gei-8(ok1671)
animals using RT-PCR at levels similar to wild-type animals, suggesting the premature stop codon may be bypassed in some transcripts by alternative splicing or that the premature stop codon is not efficiently recognized by nonsense mediated decay 
. Thus, truncated GEI-8 protein may be present in homozygous mutant larvae.
The homozygous gei-8(ok1671)
animals had obvious phenotypes, including a progressive defect in locomotion starting at the L2 stage that was marked by a delayed response to prodding and a low pharyngeal pumping rate ()
. Compared to wild-type animals of the same age, mutants were also characterized by a shorter maximum body length (750.25 µm, n
50.59 µm), a convoluted intestine, gonadogenesis defects including loss of the spermathecae, sterility, and arrest at the L4 stage of development (
C and D)
. After outcrossing the original mutant strain to wild-type animals, the heterozygous mutant strain segregated 26.2% (SD
2656) affected progeny as described ()
. To verify that the observed phenotypes were caused by the ok1671
deletion allele of gei-8
, we performed rescue using intact gei-8
genomic DNA. This method has been used previously to generate transgenic animals and to rescue mutant animals 
. Overlapping PCR regions containing a 6 kb putative promoter region plus the complete coding region of gei-8a
were injected into heterozygous gei-8(ok1671)
animals along with pRF4 injection marker, rollers were selected and their progeny were screened for locomotion defects as defined as impaired responses to prodding. The wild-type gei-8
genomic sequences were able to reduce the percentage of affected mutant progeny segregating from heterozygous hermaphrodites from 26.2% to 18.3% (SD
7883); this difference was significant using the Student's t
-test (p<0.001; SD
. Importantly, all other mutant phenotypes also showed improvement in the presence of wild-type genomic sequences leading us to conclude that most, if not all, of the defects we observed in gei-8(ok1671)
animals were due to disruption of GEI-8 activity.
Analysis of the pharyngeal pumping rate of gei-8(ok1671) mutant animals and controls.
Development of the germline in gei-8(ok1671) mutants and additional phenotypic changes induced by RNAi targeted against Y9C9A.16 (sqrd-2) in homozygous gei-8(ok1671) mutants.
Rescue experiment of gei-8(ok1671) with overlapping amplified regions of genomic DNA injected into the gonads of parents.
We scored 20 gei-8(ok1671)
mutant animals for germline development defects using Nomarski optics and DAPI (4',6-diamidino-2-phenylindole) staining of fixed animals. In 19/20 mutant animals examined, distal tip cell (DTC) migration stopped short, reaching only two thirds of it’s normal length of migration on the dorsal side of the animal ()
. In homozygous mutant animals, both gonad arms were underdeveloped, containing fewer meiotic nuclei and germ cells compared to wild-type and heterozygous gei-8(ok1671)
control animals. We also failed to detect spermathecae, sperm, or embryos in any mutant animals. We concluded that gei-8(ok1671)
mutant germlines are arrested at the L4 stage, before complete gonad elongation and spermatheca development, although some somatic markers of early young adult stages were already present (adult alae, adult vulva).
The arrested animals also had a shorter lifespan than wild-type controls. The average lifespan of gei-8(ok1671)
mutants at 20°C was 11 days (n
3.4), which was significantly shorter than the average lifespan of wild type controls (17.4 days, n
Two muscle-related phenotypes were observed in homozygous gei-8
) mutants; decreased locomotion on plates and decreased pharyngeal pumping rates. The locomotion defects we observed for gei-8(ok1671)
animals on plates prompted us to carry out a thrashing assay. When placed in liquid, wild-type animals bend back and forth moving their head and tail relative to the midbody of the animal in a thrashing motion that can be easily quantitated 
. In the gei-8(ok1671)
mutant strain, this natural thrashing behavior is impaired and deteriorated over the course of development. Unlike wild-type controls, homozygous gei-8(ok1671)
mutants at the L4 stage were not able to perform smooth thrashing. Instead, their movements were spastic and irregular, averaging only 0 to 6 bends per minute at the L4 stage compared to about 250 bends per minute for wild-type animals (n
10). Similarly, assays of pharyngeal pumping revealed irregular and reduced contraction rates in the homozygous mutants that became progressively worse with age. The average pumping rate in gei-8(ok1671)
homozygous animals was 31.8, 17.5 and 5.3 pumps per minute at L2, L3 and L4 stages, respectively (n
10 for each larval stage), compared to 250 pumps per minute for wild-type animals.
The movement and pharyngeal mutant phenotypes could be due to defects in the functioning of muscle, nerves, or both. To investigate muscle structure, we performed immunostaining using phalloidin and anti-MYO3 antibody directed against contractile apparatus components. Phalloidin stains actin filaments whereas the anti-MYO3 probe recognizes myosin heavy chain-3 
. Immunostaining revealed no obvious structural differences between gei-8(ok1671)
mutants and wild-type controls (not shown). Yamamoto et al. reported increased mitochondrial oxidative function in C. elegans
inhibition by RNAi 
. We confirmed that finding using MitoTracker Red to visualize the mitochondrial oxidative state; homozygous gei-8(ok1671)
mutants had an average mean density of staining that was more than 2.7 times greater (p<0.001) than that observed in wild-type larvae (Figure S2
). Elevated MitoTracker staining could also be visualized in heterozygous gei-8(ok1671)
mutants compared to wild-type N2 worms, but was not statistically significant in densitometric analysis of randomly selected progeny of heterozygous gei-8(ok1671)
animals with a normal phenotype (which included both heterozygous mutants as well as wild-type animals (Figure S3
The absence of obvious muscle defects in gei-8
mutants suggested that the locomotion and pharyngeal pumping defects might be due to problems in neurotransmission. We investigated synaptic transmission by assaying animal sensitivity to either aldicarb or levamisole 
. Aldicarb is a reversible acetylcholinesterase inhibitor that increases the accumulation of acetylcholine in the synaptic cleft causing whole body paralysis and inhibition of pharyngeal pumping. Homozygous gei-8(ok1671)
64) and wild-type animals (n
75) at the L4 stage were incubated on NGM plates with 1 mM aldicarb and scored over time for paralysis in three separate experiments. The onset of paralysis occurred significantly earlier in gei-8(ok1671)
mutants than in wild-type controls ()
. Levamisole is a cholinergic agonist that also results in animal paralysis. We performed two experiments with homozygous gei-8(ok1671)
40) and wild-type animals (n
40) at the L4 stage on NGM plates with levamisole at a concentration of 1 mM. As in the aldicarb assay, the onset of paralysis occurred significantly earlier in gei-8(ok1671)
mutants compared to controls ()
. Taken together, these results indicate that the gei-8(ok1671)
mutation results in abnormal cholinergic signaling, however, it does not distinguish between post-synaptic versus pre-synaptic transmission defects.
Analysis of neuromuscular function of gei-8(ok1671) mutant (VC1213).
gei-8 Loss of Function Leads to Transcription Deregulation
Effects of the gei-8(ok1671
) mutation on gene expression were studied with whole genome microarrays (Affymetrix). Changes in gene expression were defined as increased or decreased if statistically significant compared to wild-type controls in at least 2 out of 3 biological replicates. Deregulated genes were analyzed for Gene Ontology (GO) term enrichment and clustered according to functional classification using DAVID 6.7 
and KEGG pathway tools 
Expression microarray analysis revealed 756 probe sets with decreased expression, corresponding with 690 unique Wormbase IDs (Table S1)
. DAVID classification tools 
identified 645 IDs using medium classification stringency. GO analysis resulted in 32 clusters with an enrichment score greater than 2 and P<0.05. The list was enriched in spliceosome (29 genes), proteasome (13 genes), cysteine and methionine metabolism (7 genes), and RNA polymerase genes (6 genes) as identified by KEGG pathway analysis. Among specific genes involved are RNA polymerase II and III (Pol II subunits B4, B7, B9 and Pol III subunits AC2 and F09F7.3), spliceosome components (U1 to U6 snRNAs, hel-1
helicase and others), and proteasome subunits (pas-3
, rpt-1, rpt-2
). The most common functional categories over represented by the changes in gene expression were growth, embryonic or larval development and development of reproductive structures. Other clusters include multiple histones and histone-like genes, mitochondrial membrane proteins, sperm structural proteins and hedgehog-like family genes. Interestingly, the set of genes downregulated in gei-8
mutants included several genes required for proper muscle function, including unc-52
(myofilament assembly and/or attachment of the myofilament lattice to the cell membrane), unc-27
(troponin I family), unc-54
(muscle myosin class II heavy chain), pat-10
(body wall muscle troponin C), lev-11
(myosin light-chain), and tni-1
(troponin 1). It is unclear if such changes in muscle gene expression contribute to, or are the result of, the defective movement phenotypes we observed in gei-8(ok1671
) mutant animals. Depletion of NCoR1 function specifically in mouse muscle resulted in increased muscle mass and mitochondrial function 
, a phenotype opposite to what we observed in worms with reduced GEI-8 activity in all tissues.
Microarray analysis revealed 296 probe sets with increased expression, corresponding to 275 unique Wormbase IDs (Table S2
). GO analysis identified 7 clusters with an enrichment score greater than 2 and P<0.05. Enriched clusters included gene annotations for life span and aging, lipid transport and vitellogenin genes, stress response (heat shock and cellular stress), metabolic genes (sugar metabolism, glycolysis), and neuropeptide signaling (including genes coding for neuropeptide like proteins nlp-27
). The KEGG pathway analysis identified six groups including genes involved in glycolysis (8 genes), cystein methionine metabolism (4 genes), galactose metabolism (3 genes), pentose phosphate pathway (3 genes), fructose and mannose (3 genes) and tryptophan metabolism (3 genes).
One of the most significantly affected genes in the gei-8(ok1671)
homozygous mutants was Y9C9A.16, encoding a predicted mitochondrial sulfide:quinone oxidoreductase, which had an averaged 7.6-fold increase in expression compared to wild-type controls; this increase was confirmed by RT-qPCR. The Y9C9A.16 region is assayed by Affimetrix probe set 184710_at and, interestingly, includes three 21U-RNAs; 21ur-2020, 21ur-11733 and 21ur-9201. To determine if disruption of expression of Y9C9A.16 affected development, we performed RNAi targeted to the spliced mRNA covered by the Affymetrix probe set (184710_at) or only the regions that include 21ur-2020, 21ur-11733 and 21ur-9201. Progeny of parental animals injected with dsRNA targeting the specific regions were scored using Nomarski optics and fluorescent microscopy (DAPI stained). We were not able to identify any specific phenotype of Y9C9A.16 knockdown in wild type animals. However, because the expression from Y9C9A.16 showed a dramatic response to loss of GEI-8 activity, we thought there might be a biological connection between them. We predicted that knockdown of the expression from Y9C9A.16 locus in gei-8 (ok1671)
homozygous mutants might revert or modify some of the observed phenotypes; the latter was observed. RNAi-mediated knockdowns targeted to the region covered by the 184710_at probe set and the region containing 21ur-2020, 21ur-11733 and 21ur-9201 induced additional phenotypes in the gei-8(ok1671)
homozygous mutant background. Additional phenotypes included severe distal tip cell migration defects, irregular gonadal nuclei tumor like accumulation of germline cells and vulval protrusions observed in 13.9% of homozygous gei-8(ok1671)
animals treated with Y9C9A.16 RNAi (n
). Interestingly, Y9C9A.16 has a paralogue in the C. elegans
genome, the gene sqrd-1
uctase). This gene encodes a protein that is identical in size (361 aa) to Y9C9A.16 sharing 266 identical amino acids in its sequence although the genes share very little DNA homology. SQRD-1 expression is regulated by hif-1
in response to H2
S and HCN 
, is involved in innate immunity and is associated with numerous 21U-RNAs. RNAi targeted to unique regions of the sqrd-1
coding region, including four 21U-RNAs, resulted in changes in gonad arm migrations and an accumulation of germline cells (4.5% affected, n
198) that were similar, although less severe, as those observed after Y9C9A.16 RNAi. We concluded that the paralogues encoded by Y9C9A16 and sqrd-1
, and perhaps their associated 21U-RNAs, have overlapping roles during development of the germline that can be exacerbated by loss of GEI-8 activity.
Induction of additional gonad and body shape phenotypes in homozygous gei-8(ok1671) mutant worms by RNAi directed against sqrd-2 or sqrd-1.