The Caenorhabditis elegans Iodotyrosine Deiodinase Ortholog SUP-18 Functions through a Conserved Channel SC-Box to Regulate the Muscle Two-Pore Domain Potassium Channel SUP-9

Loss-of-function mutations in the Caenorhabditis elegans gene sup-18 suppress the defects in muscle contraction conferred by a gain-of-function mutation in SUP-10, a presumptive regulatory subunit of the SUP-9 two-pore domain K+ channel associated with muscle membranes. We cloned sup-18 and found that it encodes the C. elegans ortholog of mammalian iodotyrosine deiodinase (IYD), an NADH oxidase/flavin reductase that functions in iodine recycling and is important for the biosynthesis of thyroid hormones that regulate metabolism. The FMN-binding site of mammalian IYD is conserved in SUP-18, which appears to require catalytic activity to function. Genetic analyses suggest that SUP-10 can function with SUP-18 to activate SUP-9 through a pathway that is independent of the presumptive SUP-9 regulatory subunit UNC-93. We identified a novel evolutionarily conserved serine-cysteine-rich region in the C-terminal cytoplasmic domain of SUP-9 required for its specific activation by SUP-10 and SUP-18 but not by UNC-93. Since two-pore domain K+ channels regulate the resting membrane potentials of numerous cell types, we suggest that the SUP-18 IYD regulates the activity of the SUP-9 channel using NADH as a coenzyme and thus couples the metabolic state of muscle cells to muscle membrane excitability.

Mutations in the gene sup-18 suppress the muscle defects caused by gf mutations in these three genes, strongly suppressing the locomotory defects of sup-10(n983gf) mutants, partially suppressing the locomotory defects of the strong unc-93(e1500gf) mutants, the weak unc-93(n200gf) mutants and the strong sup-9(n1550gf)/+ heterozygous mutants, and suppressing only the lethality of sup-9(n1550gf) mutants [29,30] (also see Table 1 below). In this study we report that sup-18 encodes the C. elegans ortholog of mammalian iodotyrosine deiodinase/dehalogenase (IYD) [7,8,10]. Our findings suggest that SUP-18 is a functional regulator of the SUP-9/ SUP-10/UNC-93 two-pore domain K + channel complex in vivo and that IYD might function with two-pore domain K + channel complexes in mammals.
sup-18 encodes the C. elegans ortholog of mammalian iodotyrosine deiodinase sup-18 had previously been mapped to the interval between daf-4 and unc-32 on LGIII [30]. Using three-point mapping we further localized sup-18 to the interval between ncl-1 and unc-36 (see Materials and Methods) ( Figure 1A). Transgene rescue experiments with cosmids spanning the ncl-1-to-unc-36 interval and with smaller cosmid subclones identified a 4.5 kb minimal rescuing fragment from cosmid C02C2: as a transgene, this fragment restored the rubberband Unc phenotype to sup-18(n1010); sup-10(n983gf) mutants ( Figure 1A). This rescuing fragment contained

Author Summary
Iodotyrosine deiodinase (IYD) controls the recycling of iodide in the biogenesis of thyroid hormones that regulate metabolism. Defects in IYD result in congenital hypothyroidism, a multisystem disorder that can lead to growth failure and severe mental retardation. We identified the gene sup-18 of the nematode Caenorhabditis elegans as a regulator of the SUP-9/UNC-93/SUP-10 two-pore domain potassium channel complex and showed that SUP-18 is an ortholog of IYD, a member of the NADH oxidase/flavin reductase family. SUP-18 IYD is required for the activation of the channel complex by a gain-of-function mutation of the SUP-10 protein. SUP-9 channel activation by SUP-18 requires a conserved serine-cysteine-rich region in the Cterminus of SUP-9 and is independent of the function of the conserved multi-transmembrane protein UNC-93. We propose that SUP-18 uses NADH as a coenzyme to activate the SUP-9 channel in response to the activity of SUP-10 and the metabolic state of muscle cells.
a single predicted gene, C02C2.5 [www.wormbase.org]. We screened a mixed-stage cDNA library [38] using the smallest cosmid subclone with sup-18 rescuing activity and obtained a single partial cDNA of this predicted gene. We defined the structure of this gene from RT-PCR and RACE experiments (see Materials and Methods) ( Figure 1B). sup-18 encodes a predicted protein of 325 amino acids. This protein is the only C. elegans ortholog of mammalian iodotyrosine deiodinase (IYD), which belongs to the NADH oxidase/flavin reductase superfamily (Fig. 1C) [7][8][9][10]. IYD catalyzes the recycling of iodide by deiodinating 39-monoiodotyrosine and 39, 59-diiodotyrosine, the main byproducts in the process of thyroid hormone biogenesis [2][3][4][5]7,8]. The identity between SUP-18 and human IYD protein variant 2 (also named DEHAL1) [8] is 31% overall and 45% over the NADH oxidase/flavin reductase domain ( Figure 1C). Like IYDs of Drosophila, mouse and human, SUP-18 has a hydrophobic region that precedes the NADH oxidase/flavin reductase domain and might serve as a transmembrane domain.
SUP-18 and SUP-10 are similarly localized within muscles To examine the expression pattern of sup-18, we introduced the coding sequence of gfp between codons 88 and 89 of a genomic clone of sup-18, generating a sup-18 translation fusion transgene (see Materials and Methods). Similar to transgenic animals expressing a Psup-10::gfp translational fusion transgene, Psup-18::gfp transgenic animals displayed GFP fluorescence in body-wall ( Fig. 2A, D), defecation (Fig. 2B, E) and vulval muscles (Fig. 2C, F). In body-wall muscle cells ( Fig. 2A, D), the SUP-10::GFP and SUP-18::GFP fusion proteins both localized to cell-surface regions aligned with dense bodies, the functional analogs to vertebrate Zlines that connect the myofibril lattice to the cell membrane [39]. In addition to muscles, three neurons in the head of Psup-18::gfp transgenic animals also displayed GFP staining (I. de la Cruz and H. R. Horvitz, unpublished observations). We previously reported expression of a Psup-9::gfp reporter in the four SIA interneurons [28]. We stained the Psup-18::gfp transgenic animals with an anti-CEH-17 antibody, which labels the four SIA neurons and the ALA neuron [40], and found that the neurons expressing the SUP-18::GFP fusion protein were not the SIAs (I. de la Cruz and H. R. Horvitz, unpublished observations).
We generated a rabbit anti-SUP-18 antibody (see Materials and Methods). In immunostained animals, this antibody could detect overexpressed SUP-18 but failed to detect endogenous SUP-18, probably because of the low level of SUP-18 expression. We next generated transgenic animals co-expressing a Psup-10::gfp fusion transgene and sup-18 under control of a myo-3 promoter [41] and examined the subcellular expression of SUP-18 using the antibody and of SUP-10::GFP using GFP fluorescence. We found that SUP-10 and SUP-18 colocalize in subcellular structures, including the dense bodies in the body-wall muscles (Fig. 2G, H, I). Since GFP fusions to SUP-9 and UNC-93 localize similarly [28], this result suggests that SUP-18 colocalizes with a SUP-9/UNC-93/SUP-10 complex.

SUP-18 is a type-I transmembrane protein that can function independently of membrane anchoring
Mammalian IYD is a transmembrane protein [7,8]. The presence of a possible transmembrane domain in the predicted SUP-18 protein sequence (Fig. 1C) suggests that SUP-18 is also a transmembrane protein. To distinguish whether the NADH oxidase/flavin reductase domain of SUP-18 resides intracellularly or extracellularly, we generated transgenic animals expressing different SUP-18::b-galactosidase fusion proteins and assayed bgalactosidase activity in vivo in fixed animals (Fig. 3A). When bgalactosidase is localized intracellularly it is enzymatically active, whereas extracellular localization results in loss of b-galactosidase activity [42,43]. The use of b-galactosidase activity to elucidate the membrane topology of C. elegans proteins in vivo has been reported previously for the presenilin SEL-12 protein [44] and for the MEC-4 sodium channel subunit [45].
Fixed transgenic animals expressing b-galactosidase fused to either the C-terminal region of SUP-18 or immediately C-terminal to the putative transmembrane domain showed robust bgalactosidase activity (Fig. 3A). Introduction of a synthetic transmembrane domain [45] between SUP-18 and b-galactosidase in these chimeras eliminated b-galactosidase enzymatic activity, presumably because the membrane orientation of b-galactosidase had been reversed (Fig. 3A).
These results strongly suggest that SUP-18 is a transmembrane protein and that the NADH oxidase/flavin reductase domain of SUP-18 resides intracellularly. But they do not distinguish between a type-I transmembrane protein (single-pass transmembrane protein with the N-terminal domain located extracellularly) and   a cytoplasmic protein that simply localizes at the cell surface, e.g., by interacting with another membrane protein or by linking to a GPI anchor [46]. To test if the putative transmembrane domain of SUP-18 can indeed behave as a transmembrane domain, we inserted a signal sequence at the N-terminus of SUP-18 (see Materials and Methods). While a fusion containing the presumptive extracellular domain of SUP-18 but lacking the putative transmembrane domain resided intracellularly as expected, the introduction of a signal sequence led to its secretion and loss of bgalactosidase enzymatic activity (Fig. 3A). By contrast, when either the SUP-18 putative transmembrane domain or the synthetic transmembrane domain [45] was added to this SUP-18::bgalactosidase fusion, the enzymatic activity was restored. These results indicate that the putative transmembrane domain of SUP-18 can indeed function as a transmembrane domain and suggest that SUP-18 is likely a type-I integral membrane protein, like IYD.
To establish an assay for in vivo SUP-18 activity, we expressed the sup-18 coding sequence under the control of the myo-3 promoter [41] in sup-18(n1033); sup-10(n983) mutant animals. While sup-10(n983gf) mutant animals are defective in locomotion, double mutants carrying the sup-18(n1033) null mutation had improved locomotory rates (Fig. 3B). Expression of P myo-3 gfp in sup-18(n1033); sup-10(n983gf) animals had little effect on their locomotory rate, whereas expression of a P myo-3 sup-18(+) transgene restored sup-10(n983gf) paralysis (Fig. 3B). By contrast, expression of two P myo-3 sup-18 mutant constructs containing either the n1554 missense mutation or the n1010 mutation (which affects a conserved amino acid in the NADH oxidase/flavin reductase domain; Fig. 1C and Table 3) did not restore the rubberband Unc phenotype to sup-18(n1033); sup-10(n983gf) mutants (Fig. 3B).  We found that the mouse IYD gene could not substitute for sup-18 in vivo in restoring the rubberband Unc phenotype of sup-18(n1033); sup-10(n983gf) animals ( Figure 3B). We tagged mouse IYD with GFP at its C-terminus and found that C. elegans expressing the fusion protein displayed GFP fluorescence in bodywall muscle structures similar to that observed for the SUP-18::GFP fusion (I. de la Cruz and H. R. Horvitz, unpublished observations). These results suggest that mouse IYD had been expressed properly and that mouse IYD might be inactive or otherwise incapable of substituting for SUP-18 in C. elegans.
Interestingly, transgenic expression of the SUP-18 intracellular domain alone (amino acids 66-325) was sufficient to restore rubberband Unc paralysis to sup-18(n1033); sup-10(n983gf) animals, although the rescue was less robust than that conferred by full-length SUP-18 (Fig. 3B). This finding suggests that the extracellular and transmembrane domains of SUP-18 are not essential for its in vivo function and is consistent with the conclusion that the NADH oxidase/flavin reductase domain is intracellular.
We next tested if overexpression of sup-9(+), unc-93(+) or sup-10(n983gf) itself could enhance the sup-10(n983gf) defect as did overexpression of sup-18(+). Overexpression of these other genes under the control of the myo-3 promoter did not affect the locomotory rate of transgenic sup-10(n983gf) mutant animals compared to animals transgenic for lin-15 alone (Table 4). These results suggest that the activity of SUP-18 might be enhanced by increased expression, while increased expression of SUP-9, UNC-93 and SUP-10 does not increase the biological effects of these proteins.
The sup-9(n1435) mutation affects a conserved region in the C-terminal domain of SUP-9 We determined the sup-9 coding sequences in sup-9(n1435) mutants and identified a C-to-T transition within codon 292, leading to a serine-to-phenylalanine substitution within the predicted intracellular C-terminal domain of SUP-9 (Fig. 6A). Although SUP-9 is 41%-47% identical in amino acid sequence over its entire region to several TASK-family two-pore domain K + channels [28], the C-terminal cytoplasmic domain of SUP-9 is poorly conserved among these channels (Fig. 6A). However, the serine affected by the n1435 mutation is located in a small conserved stretch of amino acids with the sequence SxxSCxCY (Fig. 6A). We named this region the SC (Serine-Cysteine-rich)box. The residues in the SC-box do not correspond to any reported motifs, including phosphorylation sites, as defined by the protein motif database PROSITE [47]. Variations of the SC-box are found in the human TASK-1 and TASK-3 channels and in two Drosophila two-pore domain K + channels (Fig. 6A). We have not found an SC Box in other human two-pore domain K + channels (I. de la Cruz and H. R. Horvitz, unpublished observations) or in TWK-4 (C40C9.1), a C. elegans two-pore domain K + channel that is 41% identical to and the most closely related C. elegans channel to SUP-9 (Fig. 6A).
To further understand how its C-terminal domain affects SUP-9 activity, we deleted in the sup-9 cDNA the region encoding the SUP-9 C-terminal cytoplasmic domain. We also replaced this region with the corresponding region of twk-4, which encodes a two-pore domain K + channel without an SC-box, or of TASK-3, a mammalian homolog that contains an SC-box (Fig. 6). Deletion of the SUP-9 C-terminal domain caused suppression of both the sup-10(n983gf) and unc-93(e1500gf) sup-18(n1030) mutant phenotypes, suggesting that the truncated form of SUP-9 acts as a dominant-negative protein. Interestingly, both the sup-9::twk-4 and sup-9::TASK-3 fusion transgenes suppressed the sup-10(n983gf) egglaying defect (Fig. 6B) but failed to suppress that of the unc-93(e1500gf) sup-18(n1030) mutants (Fig. 6C), suggesting that these fusion transgenes act similarly to sup-9(n1435) and affect rubberband Unc mutants in a gene-specific manner.

Discussion
sup-18 encodes a transmembrane protein orthologous to mammalian iodotyrosine deiodinase Two-pore domain K + channels are widely expressed and play important roles in regulating resting membrane potentials of cells [15,17]. However, very little is known about protein factors with which these channels interact. We previously identified UNC-93 and SUP-10 as presumptive regulatory subunits of the SUP-9 twopore domain K + channel. We now suggest that SUP-18 also regulates the SUP-9/UNC-93/SUP-10 channel complex.
sup-18 encodes the C. elegans ortholog of mammalian iodotyrosine deiodinase (IYD), which belongs to the NADH oxidase/ flavin reductase superfamily [7,8]. By oxidizing NADH using flavin mononucleotide (FMN) as a cofactor, IYD catalyzes the recycling of iodide from monoiodotyrosine and diiodotyrosine, two major byproducts in the synthesis of thyroid hormones [7,8]. Lack of IYD function can lead to congenital hypothyroidism [12,13]. In C. elegans, no SUP-18 function besides regulating the SUP-9 channel has been identified. The enzymatic activity of SUP-18 remains to be defined.
Little is known about the metabolism and function of iodide in nematodes. The C. elegans genome contains two genes, ZK822.5 and F52H2.4, that encode homologs of the mammalian sodium/ iodide symporter, which enriches iodide in the thyroid cells by active membrane transport [48]. The presence of both SUP-18 IYD and sodium/iodide symporter-like proteins suggests that iodide functions biologically in C. elegans. Although iodide appears not to be an essential trace element in the culture medium of C. elegans [49], it is possible that residual iodide in components of that medium can provide sufficient nutritional support for survival. C. elegans lacks homologs of mammalian iodothyronine deiodinase (I. de la Cruz, L. Ma and H. R. Horvitz, unpublished observations), enzymes that remove the iodine moieties from the precursor thyroxine (T4) and generate the more potent thyroid hormone 3, 5, 39-triiodothyronine [50], which suggests that thyroid hormones might not be synthesized in C. elegans.
IYDs across metazoan species share a similar enzymatic activity in reductive deiodination of diiodotyrosine [51], and it seems likely that SUP-18 acts similarly in C. elegans. Like mammalian IYDs, SUP-18 contains a presumptive N-terminal transmembrane domain that is required for full activity. Interestingly, the SUP-18 intracellular region lacking the transmembrane domain could still partially activate the SUP-9 channel, suggesting that membrane association is not absolutely required for SUP-9 activation by SUP-18. Membrane association is important for mammalian IYD enzymatic activities [5,52,53].
The presence of a transmembrane domain suggests that SUP-18 IYD might interact with other transmembrane proteins. The genetic interactions we observe between sup-18 and the genes that encode the SUP-9/UNC-93/SUP-10 two-pore domain K + channel complex support this hypothesis. Based on expression studies, we conclude that SUP-18 and SUP-10 localize to similar subcellular structures within muscle cells, further supporting the idea that SUP-18 and the channel complex interact physically. We found that transgenic expression of the SUP-18 intracellular domain could enhance the expression of the rubberband phenotype, suggesting that plasma membrane localization is not essential for SUP-18 function. Nonetheless, the expression of the full-length SUP-18 was more potent than the expression of the SUP-18 intracellular domain in rescuing the rubberband Unc phenotypes of sup-18(lf); sup-10(n983gf) mutants, suggesting that the presence of a transmembrane domain in SUP-18 IYD could enhance the activity of SUP-18 by targeting SUP-18 to the plasma membrane.
The crystal structure of mouse IYD reveals that eight residues contact the FMN cofactor: R96, R97, S98, R100, P123, S124, T235 and R275 [54]. Except T235, which is replaced by a serine in SUP-18, these residues are completely conserved ( Figure 1C, yellow boxes). Furthermore, the sup-18(n1010) missense mutation leads to an S137N substitution at the position equivalent to the mouse S98 residue, likely disrupting the binding of FMN. This high degree of conservation at the cofactor binding site suggests that SUP-18 likely retains the ability to bind FMN and likely has a catalytic activity.
Three IYD missense mutations that cause hypothyroidism (R101W, I116T, and A220T) affect residues that are conserved in SUP-18 [12,55] (Fig. 1C, red boxes). A fourth human mutation replaces F105 and I106 with a leucine [8]. The phenylalanine at position 105 is conserved in SUP-18 (Fig. 1C). The conservation of residues associated with IYD function supports the hypothesis that SUP-18 regulates the SUP-9 two-pore domain K + channel complex via an enzymatic activity. The SUP-18 substrate remains to be elucidated.
That SUP-18 might function as a NADH oxidase/flavin reductase raises the intriguing possibility that SUP-18 might couple the metabolic state of muscle cells with membrane excitability. Mammalian K v b voltage-gated K + channel regulatory subunits [56], which belong to the aldo-keto reductase superfamily [57,58], have similarly been proposed to couple metabolic state with cell excitability based on indirect evidence. K v b2 has an NADP + cofactor bound in its active site and a catalytic triad spaced appropriately to engage in enzymatic activity [58]. Although suggestive of an enzymatic activity, no substrate has been reported for K v b subunits. While K v b2 knockout mice have seizures and reduced lifespans, mice carrying a catalytic null mutation in K v b2 have a wild-type phenotype, suggesting that if an enzymatic activity for K v b2 exists, it is functionally dispensable in vivo [59]. By contrast, the predicted catalytic mutation sup-18(n1010) behaves like a null mutation in its inability to activate the SUP-9 channel, even though the SUP-18(n1010) protein is synthesized and localized normally to the cell surface of muscle cells (I. de la Cruz and H. R. Horvitz, unpublished observations). Five other sup-18 mutations affecting highly conserved residues in the NADH oxidase/flavin reductase domain also behave like null mutations, consistent with the hypothesis that SUP-18 enzymatic activity is essential for its function.
sup-18(lf) mutations define a new class of gene-specific suppressors of the rubberband Unc mutants sup-18(lf) mutations strongly suppress sup-10(n983gf) mutants and weakly suppress unc-93(e1500gf) mutants. Certain specific mutations of sup-9, including n1435, n4259, n4262, and n4269, act similarly to sup-18(lf) and are strong suppressors of sup-10(n983gf) mutants and weak suppressors of unc-93(e1500gf) mutants. Together these sup-9 mutations and sup-18(lf) mutations represent a novel class of mutations that exhibit gene-specific suppression of the rubberband Unc mutants and are distinct from another class of gene-specific suppressors we identified previously, mutations in three splicing factor genes that strongly suppress unc-93(e1500gf) and sup-10(n983gf) but do not obviously suppress unc-93(n200gf) or sup-9(n1550gf) [60][61][62]. The difference between sup-18(lf) and sup-9(n1435, n4259, n4262, n4269) mutations and the splicing factor mutations in their patterns of suppressing the rubberband Unc mutants suggests that these two classes of suppressors function by distinct mechanisms. SUP-18 Interacts with a Two-Pore Domain K + Channel SUP-18 is an activator of the SUP-9 two-pore domain K + channel SUP-9 is closely related to the subfamily of two-pore domain K + channels that include human TASK-1 and TASK-3 [28]. TASK-1 is activated by multiple factors, including extracellular pH [22,23,63], inhalational anesthetics such as halothane [24] and oxygen [64]. TASK-1 is directly inhibited by sub-micromolar levels of the cannabinoid neurotransmitter anandamide [65] and by neuromodulators such as thyrotropin releasing hormone (TRH) [27]. A histidine residue in the first P-domain of TASK-1 modulates its sensitivity to pH [66], while a six amino acid stretch following its fourth transmembrane domain is required for both halothane activation and TRH suppression [24,67]. Deletion of the TASK intracellular C-terminal domain, which contains the SC-box, does not change its basal activity or activation by halothane [24,67], suggesting that the TASK-1 C-terminal domain and probably the SC-box represent an activation region that is required by some types of channel activator (e.g., human IYD) but not by others (e.g., halothane and pH). It remains to be determined whether IYD is involved in the inhibition of TASK-1 channel activity by TRH.
From our genetic analysis of the sup-9(n1435) mutation and sitedirected mutagenesis of sup-9, we have defined the SC-box, a domain of SUP-9 required for SUP-10(n983gf)-specific activation. The importance of the SC-box in mediating this activation is supported by the results of a genetic screen in which we isolated additional sup-9 mutations (Fig. 7) that act like sup-9(n1435) and cause distinct amino acid changes in (n4259 (S292A), n4262 (S294A)) or near (n4269 (L303P)) the SC-box. Although conserved in the human TASK-1 and TASK-3 channels (Fig. 6A), no function has yet been assigned to the SC-box. Our analyses suggest that the SC-box and the C-terminal domain of SUP-9 likely mediate the functional interaction between SUP-9 and SUP-10/ SUP-18 but are dispensable for interaction with UNC-93. We found that replacing the C-terminal domain of SUP-9 with the corresponding region of TWK-4 (which lacks an SC-box) or of TASK-3 (with an SC-box) makes the fusion channels behave like SUP-9(n1435), consistent with the model that the SC-box is required for SUP-9 activation by SUP-10(n983gf) and SUP-18 (based on the TWK-4 data) and suggests that SC-box-dependent activation requires one or more nearby residues in the C-terminal domain (based on the TASK-3 data). The unc-93(e1500gf) mutation results in a glycine-to-arginine substitution at amino acid 388 in one of the putative transmembrane domains [33], suggesting that the UNC-93(gf) protein activates SUP-9 through an interaction involving transmembrane domains, without a need for the SC-box or the rest of the cytoplasmic domain.
We describe three important properties of the unusual sup-9(n1435) mutation. First, SUP-9(n1435) channels cannot be activated by SUP-10(n983gf). Second, SUP-9(n1435) channels are insensitive to SUP-18 activity. Third, SUP-9(n1435) channels can be activated by UNC-93(e1500gf). The existence of a channel mutation that is insensitive to both SUP-18 and SUP-10(n983gf) suggests that these two inputs act through a common pathway. A mutant channel that can be activated by UNC-93(e1500gf) but not by SUP-10(n983gf) suggests that there is an independent pathway for SUP-9 activation by UNC-93.
We propose a model to explain the functional interactions between SUP-18 and SUP-9/UNC-93/SUP-10 (Fig. 8). In this model, SUP-10 and UNC-93 have an essential role in and are both required for activating SUP-9 channel, since the n1550 gf mutation in sup-9 is completely suppressed by sup-10(lf) and unc-93(lf) mutations [38]. SUP-18 activates SUP-9 only weakly and relies on SUP-10 for this activation (Fig. 8). SUP-10(n983gf) enhances the activity of SUP-18 and results in over-activation of SUP-9 by SUP-18. Our model is consistent with the genetic and molecular evidence described in this and previous studies [28][29][30][31]33] and should provide a framework for understanding the interactions of SUP-18 and the SUP-9/UNC-93/SUP-10 channel complex. Our results do not distinguish whether SUP-18 regulates the SUP-9/UNC-93/SUP-10 complex via a direct physical interaction or indirectly through an unknown factor or factors.
In short, we identified SUP-18 IYD as a functional regulator of the SUP-9/UNC-93/SUP-10 two-pore domain K + channel complex. We also defined an evolutionarily conserved serinecysteine-rich domain, the SC-box, in the C-terminal region of SUP-9 and showed that this region is required for activation of the channel by SUP-18. Since IYD is likely to be an NADH oxidase/ flavin reductase that uses the ubiquitous energy carrier molecule NADH as a coenzyme, our study suggests that IYD might couple cellular metabolic state to two-pore domain K + channel activities. Future molecular analyses should reveal the mechanism underlying the interaction between the SUP-9 two-pore domain K + channels and SUP-18 IYD.

Isolation of partially suppressed unc-93(e1500gf) mutants
Since lf mutations in sup-10 completely suppress the paralysis of unc-93(e1500gf) mutants [31], we reasoned that partial lf mutations of sup-10 would partially suppress the unc-93(e1500gf) locomotory phenotype. To isolate such partial lf sup-10 mutations, we Figure 8. A model for activation of the SUP-9 channel by multiple subunits. In this model, SUP-10 and UNC-93 act independently of SUP-18 to activate SUP-9. In addition, SUP-10 enhances SUP-18, which activates SUP-9 through a distinct pathway. SUP-10(n983gf) over-enhances SUP-18, which over-activates SUP-9 and leads to paralysis. The widths of the arrows pointing at SUP-9 are representative of their relative strengths in sustaining gf activity, with thicker arrows representing a larger contribution. doi:10.1371/journal.pgen.1004175.g008 performed an EMS F 2 genetic screen for partial suppressors of the locomotory defects of unc-93(e1500gf) mutants. From 17,500 haploid genomes screened, we isolated over 30 strong suppressors and seven weak suppressors. We assigned two of the seven weak suppressors, n4025 and n4026, to the sup-10 locus by complementation tests and three others to the unc-93 locus. All seven were saved for future analyses.
RT-PCR was performed on cDNA from the wild-type N2 strain using the primers 59-TTGAAAACCCCTGTTAAATAC-39 and 59-CGAGTTTCTAATAAAAATAAACC-39. PCR products were cloned into pBSKII (Stratagene), and their sequences determined. 59 and 39 RACE were performed using the corresponding kits (Gibco).

Molecular biology
Genomic subclones of cosmid C02C2 were generated in pBSKII (Stratagene). The subclones, in the order shown in Figure 1, spanned the following sequences (Genebank Ac- All PCR amplifications used in plasmid constructions were performed using Pfu polymerase, and the sequences of their products were determined. The P myo-3 sup-18 vectors for ectopic expression of wild-type or mutant sup-18 alleles were generated by PCR amplification of the respective coding regions from sup-18 cDNAs using primers that introduced NheI and SacI sites at the 59 and 39 ends, respectively, and cloned into vector pPD95.86 (from A. Fire). P myo-3 sup-18(intra) was similarly constructed, except that the 59 primer began at codon 66 of sup-18. The gfp-tagged version of this vector was created by PCR amplification of the gfp coding sequence from vector pPD95.77 (from A. Fire) and subcloned into P myo-3 sup-18(intra) just prior to the start codon of the sup-18 sequence. P myo-3 mIYD (mouse IYD) was generated by PCR amplification of the coding region of the mouse cDNA (Gene Bank AK002363) with 59 and 39 primers containing NheI and EcoRV sites, respectively, and subcloning the PCR products into pPD95.86 at the NheI and SacI (blunted) sites. P myo-3 mIYD::gfp was generated by a similar strategy using a 59 primer containing an NheI site and a 39 primer that did not include the stop codon of mIYD but instead contained a BamHI site. The myo-3 promoter from pPD95.86 was subcloned into pPD95.77, such that upon subcloning of the mIYD PCR fragment into the NheI and BamHI sites of the vector the myo-3 promoter drove expression of the mIYD gene fused in-frame at its 39 end to gfp.
The sup-18::gfp genomic fusion was constructed by introducing SphI sites at the ends of a gfp cassette by PCR amplification of plasmid pPD95.77 (from A. Fire) and subsequent subcloning into the single SphI site contained within a 9.1 kb PstI genomic sup-18 rescuing fragment. The resulting fusion contained 6.5 kb of promoter sequence, the entire sup-18 coding region with gfp inserted between the transmembrane and NADH oxidase/flavin reductase domains and 1.1 kb of 39 UTR and downstream sequence.
The sup-10::gfp fusion used in colocalization studies was constructed by subcloning a 7.3 kb MfeI genomic fragment from cosmid C27G6 containing sup-10 into the EcoRI site of pBSKII. A 6.4 kb Pst I fragment was subcloned from this vector into pPD95.77, which contained 3.5 kb of promoter sequence and the sup-10 coding region. Using PCR, we introduced a SalI site immediately preceding the stop codon of sup-10 to create an inframe fusion with the gfp coding sequence.
sup-18::b-galactosidase fusions were created by PCR amplification of 1869 bp of 59 sup-18 promoter sequence and subcloning the product into the SphI and PstI sites of pPD34.110 (from A. Fire) to generate P sup-18 TM-b-Gal, which contains a synthetic transmembrane sequence [45] followed by the b-galactosidase coding sequence [72]. sup-18 genomic coding sequence spanning codons 1-42, 1-70 and 1-301 were PCR-amplified from the minimal rescuing fragment with 59 and 39 primers that contained PstI and BamHI sites, respectively, and subcloned into these sites in P sup-18 TM-b-Gal. The synthetic transmembrane domain was deleted from these plasmids by excising the KpnI fragment containing this domain. A signal sequence [73] was inserted into these vectors using standard PCR techniques.

Body-bend assay
Young adults were individually picked to plates with HB101 bacteria, and body-bends were counted for one minute using a dissecting microscope as described [74].

Antibody and immunostaining
A GST::SUP-18(N) fusion protein was expressed in E. coli and the insoluble protein was purified by SDS-PAGE and used to immunize rabbits. Antisera were purified by binding to the MBP::SUP-18 protein immobilized on nitrocellulose strips and elution with 100 mM glycine-HCl (pH 2.5). This antibody could detect SUP-18 overexpressed in the body-wall muscles (Fig. 2H) but failed to detect endogenous SUP-18.
For immunofluorescence experiments, worms at mixed stages were fixed in 1% paraformaldehyde for 2 hrs at 4uC and permeabilized as described [75]. For colocalization studies, transgenic worms were stained with primary antibodies at 1:200 dilution and a secondary goat-anti-rabbit antibody conjugated with Texas Red (Jackson Labs). Worms were viewed using confocal microscopy.

Transgenic animals
Germline transformation experiments were performed using standard methods [71]. Transgenic strains carrying the lin-15(n765ts) mutation contained the coinjection marker pL15EK(lin-15(+)) at 50 ng/mL [70], and transgenic animals were identified by their non-Muv phenotype at 22.5uC. The dominant rol-6 plasmid [71] was used at 100 ng/ml during cosmid rescue experiments, and transgenic animals were identified by their Rol phenotype. The dominant myo-3::gfp fusion vector pPD93.97 (from A. Fire) was used where indicated at 80 ng/ml, and transgenic animals were identified by GFP fluorescence. Experimental DNA was injected at 30-50 ng/ml.

Isolation of novel sup-9 alleles
One plausible genetic strategy for isolating sup-9 alleles similar to sup-9(n1435) would be to perform an F 2 screen for suppressors of the sup-10(n983gf) locomotory defect and then test these suppressors for their effects on the locomotory defect of unc-93(e1500gf) mutants. Most sup-9 alleles isolated from such a screen would be typical lf alleles rather than rare alleles that would result in a SUP-9 protein specifically impaired in activation by SUP-10(gf) and SUP-18(+). We therefore opted for an alternative strategy based on the semidominance of the sup-9(n1435) mutation. While sup-9 null mutations, such as n1913, recessively suppress the locomotory defects of sup-10(n983gf) mutants, sup-9(n1435) caused a strong semidominant suppression (Fig. 5C). As two-pore domain K + channels are homodimers [66,76], this semidominance likely reflects the formation of nonfunctional heterodimers composed of n1435 and wild-type SUP-9 proteins. The strength of this semidominance (,23 vs. ,5 bends/minute for sup-9(n1435)/+; sup-10(n983gf) vs. sup-10(n983gf) mutants, respectively) formed the basis of an F 1 screen for suppressors of the sup-10(n983gf) locomotory defect.
sup-10(n983gf) L4 hermaphrodites were mutagenized with EMS, and approximately 550,000 F1 progeny (1.1610 6 genomes) were screened for improved locomotion on agar plates. From 89 candidate suppressors, 35 mutants retested in the next generation, representing at least 31 independent isolates. To quantify the semidominant character of these mutants (sup(new)), wild-type males were crossed with homozygous mutant hermaphrodites to generate sup(new)/+; sup-10(n983gf)/0 males, and their locomotory rate was scored. Because sup-10 is on the X chromosome, this strategy generates males hemizygous for sup-10(n983gf) while heterozygous for autosomal mutations, providing a convenient assay of semidominance. Four mutations completely suppressed the rubberband Unc phenotype of males, with locomotory rates very similar to that of wild-type animals (,33 bends/minute). We reasoned that these four mutants were likely lf alleles of sup-10, as such animals would be hemizygous for sup-10. We confirmed this assignment by determining the sequences of the sup-10 locus and found mutations in all four strains (I. de la Cruz and H. R. Horvitz, unpublished observations). For the remaining strong mutants, we performed complementation tests with sup-9, sup-18 and unc-93 strains and identified 11 semidominant alleles of sup-9 (see Results).