GAS-1 : A Mitochondrial Protein Controls Sensitivity to Volatile Anesthetics in the Nematode Caenorhabditis elegans

Standard C. elegans nomenclature is used throughout. Italicized lowercase three-letter codes followed by a hyphen and a number designate gene names (e.g., gas-1). This can also refer to animals mutant in that gene. A Roman numeral after a gene name refers to the chromosome to which that gene has been mapped. A specific version of a genetic mutation (i.e., an allele) is designated by one or two italicized letters followed by a number without a hyphen, such as fc21. A phenotype is referred to by a nonitalicized spelling of the gene three-letter code (with or without a number), with the first letter capitalized, such as Gas-1 or Unc. The protein product of a gene is designated by the nonitalicized gene name written in all capital letters, such as GAS-1.

Basic techniques of genetics and worm culture were as described by Brenner [20]and Herman. [21]Nematodes were grown on agar plates using an Escherichia coli lawn as a food source. In general, nematodes were grown at 20 [degree sign]C, although sometimes gas-1 was tested for temperature sensitivity at 25 [degree sign]C. Several nematode strains were obtained from the Caenorhabditis Genetics Center in Minneapolis, Minnesota. These included the wild-type strain N2, unc-7(e5)(unc indicates an uncoordinated phenotype), mnDp1 (X,V);unc-3, mnDf1, unc-3, and several mutant strains containing lethal genes. lin-15(n309) was a gift of Paul Sternberg (Pasadena, CA). gas-1(fc21) was isolated by screening for increased sensitivity to enflurane after N2 was mutagenized with ethylmethanesulfonate, as described previously. [14,20]Procedures for exposing nematodes to anesthetics, scoring their responses, and measuring the anesthetic concentrations have been described in detail previously. [12,14] 

Fragments of DNA from the C. elegans genome have been cloned into cosmids. (Cosmids are modified bacteriophages in which large pieces of foreign DNA are inserted between two ends of the [small lambda, Greek] phage chromosome. The foreign DNA, in this case C. elegans genomic fragments, can then be grown in large quantities in bacteria.) Cosmids from the region of the gas-1 gene were provided by Alan Coulson (Cambridge, UK). Cosmid DNA was prepared using the Plasmid Maxiprep Kit (Qiagen, Chatsworth, CA), and DNA concentrations were measured by comparison with standards analyzed on agarose gels. Mutant rescue was done by the techniques described by Mello et al. [22,23]The mutant rescue assay involves microinjection of DNA into the gonad of mutant worms, where it can be taken up by developing oocytes. If the injected DNA contains a wild-type copy of the mutated gene, then some of the offspring can be stably transformed to wild-type; that is, they are “rescued.”

The gas-1(fc21) has an abnormally low brood size and gonads that are difficult to microinject. Therefore, test DNA was injected into a worm of the following genotype: mnDp1 (X,V);unc-7gas-1(X). mnDp1 (X,V) is a genetic balancer that carries a translocation of part of the right arm of the X chromosome to chromosome V [24,25]; it contains wild-type copies of gas-1 and unc-7 (Figure 1). Therefore, this animal is wild-type in movement and anesthetic sensitivity; it also has injectable gonads and a normal brood size. The translocation is only present in one copy (two copies are lethal). Therefore, one third of the self-progeny do not contain the duplication and are of the genotype unc-7gas-1. These animals are uncoordinated in air, hypersensitive to volatile anesthetics, and have few offspring.

The following steps were taken for mutant rescue:(1) 10 [micro sign]g/ml test DNA (cosmid or restriction fragments) and 100 [micro sign]g/ml plasmid pRF4 were coinjected into the gonads of mnDp1 (X,V);unc-7gas-1 (X) hermaphrodites. pRF4 contains the dominant mutation rol-6(su1006), which causes worms to roll about their longitudinal axis. This serves as an easily visible marker for animals carrying successfully injected DNA and thus as a control for successful injections. It also facilitates the formation of extrachromosomal arrays of DNA that are a combination of test and marker DNA. Transgenic offspring were selected by picking rolling offspring of microinjected animals. In this first generation of offspring (F1), rolling animals were either wild-type (containing the balancer duplication) or Unc-7 (not containing the duplication). (2) F1 Unc-7 rollers were kept separate from each other. Those lines that continued to roll in the second generation (F2) and subsequent generations were evaluated for rescue. Rescued worms met two criteria: They move in 2% halothane (a concentration immobilizing gas-1 but not N2 or rol-6 animals), and they have an increased number of progeny compared with gas-1. Nonrolling siblings of rolling animals were also checked for sensitivity to halothane. (3) Conversely, F1 wild-type rollers cannot be evaluated directly. However, in lines that continued to have rolling offspring, the Unc-7 rollers arising by loss of mnDp1 (X,V) were collected and evaluated as described for the first-generation rollers. Rescue is regarded as proof of the presence of the wild-type gene in the test DNA. However, nonrescue does not rule out that the wild-type gene was injected, because it might not be properly expressed or distributed to all cells (mosaic).

mnDp1 (X,V);unc-7gas-1 animals were made by mating mnDp1 (X,V);unc-3(X) males with unc-7gas-1 hermaphrodites. Wild-type F1 offspring were collected. In the next generation, wild-type F2 animals were individually isolated. Animals that produced only wild-type and Unc-7Gas-1 progeny (i.e., no Unc-3 animals) were kept. The genotype of these animals was checked by mating to unc-7gas-1 animals.

Genetics. In general, genetic mapping was done by standard techniques described by Brenner. [20]We performed complementation tests between gas-1 and several let (lethal) genes close to gas-1. If first-generation cross-progeny of two recessive mutations are wild-type, then the genes are said to complement. With few exceptions, this indicates that they are different genes. We again took advantage of the translocation mnDp1 (X,V). Meneely and Herman [24]had placed mnDp1 (X,V) over many recessive lethal mutations on the right end of the X chromosome, each of which was linked to an unc-3 mutation, which is also covered by the duplication (Figure 1). In general, the strains we used were of the genotype mnDp1 (X,V);unc-3let-x. The let genes tested in this manner were let-3, let-10, let-11, let-14, let-15, let-16, let-18, let-33, let-35, let-36, let-37, let-38, let-39, let-40, and let-41. N2 males, which are X/O, were crossed with each of these strains to generate wild-type males of the genotype mnDp1 (X,V);unc-3let-x/O. These were mated to unc-3gas-1 hermaphrodites. gas-1 complements a let gene if Unc-3 F1 animals have normal anesthetic sensitivity and normal brood size.

Genomic DNA and RNA Preparation. Worms were grown in 200-ml liquid cultures (S-basal, cholesterol, E. coli)[20]and shaken vigorously at 20 [degree sign]C until the bacteria were consumed (approximately 1 week). Worms were harvested by centrifugation. They were cleaned from remaining bacteria by spinning the resuspended worms on a 32%(wt/vol) sucrose cushion. After washing with S-basal twice, genomic DNA was extracted from the nematodes with the Puregene DNA Isolation Kit (GENTRA Systems, Research Triangle Park, NC). Alternatively, total RNA was prepared using TRIzol (Gibco/BRL, Gaithersburg, MD) according to the manufacturer's protocol.

Sequencing. Total RNA from gas-1 or wild-type worms was transcribed into cDNA using M-MLV (Moloney-Murine Leukemia Virus) Reverse Transcriptase (Gibco/BRL, Gaithersburg, MD) primed with random hexanucleotides. From the pool of cDNAs, the coding region of gas-1 was amplified selectively with the Expand High Fidelity polymerase chain reaction (PCR) system (Boehringer-Mannheim Biochemicals, Indianapolis, IN). The necessary primers (Genosys, The Woodlands, TX) were designed to fit the predict first and last exon. Subfragments of 250 to 300 base pairs (bp), overlapping by approximately 50 bp, were generated from the larger PCR product using the same PCR system with nested primers. Purified subfragments (QIAquick PCR Purification Kit; Qiagen, Chatsworth, CA) were bidirectionally sequenced by the Sequencing Facility of the Genetics Department of Case Western Reserve University. The 3'- and 5'-UTR of the mRNA were determined by RACE-PCR (rapid amplification of cDNA ends) as described by Frohman. [26]The resulting DNA fragments were cleaned and sequenced as described before. Fragments of genomic DNA were PCR amplified, purified, and sequenced using the same primers as for cDNA sequencing.

Sequences were analyzed for possible genes with the program Genefinder (http://dot.imgen.bcm.tmc.edu:9331/gene-finder/gf.html) and compared with the National Institutes of Health database using BLAST. [27]PC/Gene 6.85 (Intelligenetics, Inc., Oxford Molecular Group, Beaverton, OR) was used for prediction of transit peptides. Peptides included in the sequence comparison were selected and aligned by the program MaxHom at phd@EMBL-Heidelberg.de.[28]They are subunits of the following dehydrogenases: mitochondrial complex I from bovine heart (P17694) and from Neurospora crassa (P22142), cell membrane NADH dehydrogenase from Paracoccus denitrificans (P29916), chloroplast NAD(P)H dehydrogenase from Marchantia polymorpha (P12131) and from Zea mays (P25709), and NADH dehydrogenase from E. coli (P33600).

gas-1 was previously mapped to the X chromosome to the right of the gene unc-7 (Figure 1A). [18]We finely mapped gas-1 relative to two genes in this region, unc-7 and lin-15 (unc for uncoordinated, lin for lineage defective) by mating gas-1 males to the double mutant unc-7lin-15. F2 animals were screened for Lin non Unc animals and Unc non Lin animals. Such animals could only arise if a recombinant event occurred between unc-7 and lin-15. If gas-1 lay to the right of lin-15, then essentially all Unc non Lin animals would be Gas-1. Conversely, if gas-1 lay to the left of unc-7, then essentially all Lin non Unc animals would carry the gas-1 mutation. If gas-1 lay between unc-7 and lin-15, then the gas-1 mutation will sometimes go with Unc, sometimes with Lin. The frequency with which gas-1 is found with the other single mutations will reflect the genetic distance, along the unc-7 lin-15 span, where gas-1 is located. We found that 17/20 Unc non Lin animals were Gas-1 (hypersensitive to halothane) and 3/15 Lin non Unc animals were Gas-1. Thus, gas-1 was placed 83% of the distance from unc-7 toward lin-15 [20](Figure 1B).

fc21 was shown previously to be a temperature-sensitive hypomorphic (i.e., partial loss of function) lethal mutation. [18]Briefly, the data supporting this were that fc21/fc21 animals had approximately 50 offspring at 20 [degree sign]C but 5 to 10 offspring at 25 [degree sign]C (200-300 offspring is normal for the wild-type, N2) and many dead eggs. fc21 was placed over a deletion of the region, mnDf1. At 20 [degree sign]C, these animals had 5 to 10 offspring, whereas at 25 [degree sign]C they were sterile. Because fc21 represented a change in a lethal gene, we performed complementation tests with the known lethal genes in the region from unc-7 to lin-15. As noted in the materials and methods section, the let genes tested in this manner were let-3, let-10, let-11, let-14, let-15, let-16, let-18, let-33, let-35, let-36, let-37, let-38, let-39, let-40, and let-41 (Figure 1). All of these strains complemented gas-1. Thus, gas-1 probably represented a new lethal gene from this region.

In the unc-7 lin-15 interval, the Caenorhabditis Sequencing Consortium has already established a physical map consisting of cosmid and YAC (yeast artificial chromosomes) contigs. (A contig is a contiguous overlapping set of DNA fragments from wild-type worms propagated in bacteria or yeast.) Because gas-1 mapped within this contig, we used the genetic mapping data noted before to direct our search to the right end of the unc-7 lin-15 region. Cosmids from this region were used for mutant rescue experiments to identify the presence of the intact gas-1 gene, as described in Materials and Methods. In mutant rescue, only positive results are meaningful, because there are multiple possible explanations for the lack of expression of an injected piece of test DNA. Therefore, the appearance of even one stable rescued line is sufficient to equate a piece of DNA with a given gene.

The cosmids C40C3 and K09A9 each separately rescued gas-1, narrowing the locus of gas-1 to their overlap (Figure 2A). Of the stable (F2) rolling lines, 7 of 10 were rescued by C40C3 and 2 of 5 were rescued by K09A9 (in contrast to 18 stable lines of other cosmid injections that failed to rescue gas-1). Because K09A9 is completely sequenced, it was possible to design restriction fragments containing candidate genes and test them for mutant rescue. Two nested fragments successfully rescued the Gas-1 phenotype (Figure 2): a 9kb BsiW1 fragment and a 4.6kb Sful fragment. One of two stable lines arising from injection of the BsiWi fragment and one of one stable line arising from the injection the Sful fragment rescued gas-1. Rolling offspring of animals injected with fragments that did not contain the 4.6kb Sful fragment were never rescued (0 of 27 stable lines). Subsequent generations of these rescued stable lines continue to display rescue of the mutant phenotype. In addition, the double mutant, rol-6;gas-1, constructed via conventional mating, is not rescued. Thus, the rol-6 gene is not sufficient to rescue the gas-1 phenotype. Rescue can only occur as a result of the addition of the wild-type gene. The smallest rescuing fragment, the 4.6kb Sful fragment, thus contains the wild-type gas-1 gene. For this fragment, sequence analysis with Genefinder software predicted only one complete gene, designated K09A9.5 (Figure 2B).

To verify the computer prediction of the gene's structure, and to identify the nature of the gas-1(fc21) mutation, we sequenced the mRNA after reverse transcription and PCR, as described in Materials and Methods. The wild-type K09A9.5 message is spliced together from six of the predicted seven exons (Figure 2B). Starting at the first ATG codon, the K09A9.5 cDNA sequence translates into a preprotein of 482 amino acid residues (Figure 3). The N-terminal 31 residues fulfill the requirements for a transit peptide directing the preprotein into the mitochondrion. [29]In the mutant gas-1, the K09A9.5 mRNA differs only in a single nucleotide from the wild-type. A transition from G to A leads to a missense mutation. This point mutation was confirmed by sequencing the corresponding part of genomic DNA from gas-1 mutants. The arginine in position 290 of the wild-type protein is changed to lysine in the gas-1 mutant (Figure 3). A small open reading frame on the counterstrand is affected by the same mutation (Figure 2B). It is possible, although unlikely, that this open reading frame is the first exon for the gene K09A9.4. K09A9.4 can be ruled out as gas-1, because most of its exons are missing on the rescuing 4.6kb Sful fragment. Furthermore, the mutation has no effect on the peptide sequence of K09A9.4 (the amino acid coded for remains a leucine). In summary, (1) the smallest rescuing fragment carries only one complete gene (K09A9.5), (2) a missense mutation occurs in this gene in fc21, and (3) the same mutation is meaningless for the counterstrand. We conclude that K09A9.5 is the gas-1 gene.

The amino acid sequence of GAS-1 is similar to a family of proteins that are subunits of electrogenic dehydrogenases: mitochondrial NADH dehydrogenase, chloroplast NAD(P)H dehydrogenase, and bacterial NADH dehydrogenase (Figure 3). GAS-1 is most homologous (68% identical amino acid residues) to the “49-kDa (IP) protein,” a subunit of the mitochondrial NADH dehydrogenase, isolated from bovine heart. [30](IP indicates that it copurifies with the iron sulfur protein, a subcomplex carrying iron sulfur centers.) Most of the differences are found in the N-termini of the two proteins (which is typical for the entire family); from amino acid 100 on they share 75% identical residues. The degree of identity is greater than for all other pairwise comparisons, including 49-kDa (IP) homologues found in NADH dehydrogenase preparations from N. crassa [31]and P. denitrificans. [32] 

Based on the close homology of gas-1 to known sequences of the 49-kDa protein, we conclude that GAS-1(+) is a true 49-kDa (IP) protein and that it forms part of the worm's respiratory chain NADH dehydrogenase. (Because the homologues of the 49-kDa (IP) subunit of the bovine heart mitochondrial NADH dehydrogenase lack a more descriptive name, we refer to them here as 49-kDa (IP) proteins, although their molecular weights differ from species to species). The arginine mutated in gas-1 is strictly conserved throughout the protein family across all phyla and all enzyme variations (Figure 3), suggesting that arginine is absolutely required in this position for proper NADH dehydrogenase function. Thus, although the change from arginine to lysine does not change the electric charge of the peptide, it may alter protein function because lysine has a slightly smaller side group than does arginine. This minor change in one protein of a presumably large protein complex (at least 41 different subunits in some eukaryotes) causes a major change in the animal's phenotype. Somehow then this amino acid must play an important role in the function of the 49-kDa protein and, in turn, in the function of complex I.

Most of the current knowledge about eukaryotic mitochondrial NADH dehydrogenase stems from investigations of the enzyme from bovine heart and N. crassa. [33,34]The mitochondrial NADH dehydrogenase is the first enzyme complex of the respiratory chain (complex I, Figure 4). It oxidizes NADH (which is produced predominantly in the Krebs cycle or during [small beta, Greek] oxidation of fatty acids) and transfers the electrons to ubiquinone, which is reduced to ubiquinol (NADH:ubiquinone-oxidoreductase). Coupled to this redox reaction is the electrogenic transport of three to five protons (per electron transferred) across the inner mitochondrial membrane. [34]Ubiquinol passes electrons on to the downstream elements of the respiratory chain (complex III = ubiquinol:cytochrome C oxidoreductase, cytochrome C, and complex IV = cytochrome C oxidase). Complex III and complex IV also pump protons. Thus, NADH dehydrogenase contributes directly and indirectly to the proton motive force needed for oxidative phosphorylation. Besides its key role in energy metabolism, ubiquinol has a second important function as an antioxidant. [35]It protects membrane components from oxidative destruction by quenching free radicals that arise as inevitable by-products of respiration.

No definitive function has been assigned to the 49-kDa (IP) proteins. [34]However, they are common to both the complicated eukaryotic complex I and to the much simpler enzyme in Paracoccus (15 subunits only), which catalyzes the same reaction as the eukaryotic homologue. A knockout mutant of the 49-kDa (IP) gene in Neurospora lacked NADH dehydrogenase activity completely because the “matrix arm” of the enzyme complex failed to assemble. [36]The matrix arm contains the binding site for NADH and all but one of the redox centers. Thus, 49-kDa (IP) proteins seem to be essential for the core function of the eukaryotic enzyme complex.

The mutant gas-1 gene product causes the changes in sensitivity of C. elegans to volatile anesthetics. The molecular data reveal that a conserved amino acid is changed in fc21. The genetic data argue that the fc21 allele is a hypomorph, a genetic change that decreases function of the gas-1 gene. Together these two lines of data suggest that complex I function is decreased in fc21. Intriguingly, several reports document that one of the effects of volatile anesthetics is the reversible inhibition of cellular respiration. [37-40]Cohen [37]found that, at 25 [degree sign]C, 0.54 mM halothane (2.5 minimum alveolar concentration) decreased complex I-dependent respiration by 50% in isolated mitochondria. Harris et al. [38]also found that high concentrations of halothane (0.7 mM at 25 [degree sign]C) inhibit mitochondrial function. Nahrwold and Cohen [39]found that 1.5% isoflurane caused an approximate 20% decrease in mitochondrial complex I function; approximately 3.5% isoflurane was necessary to obtain a 50% reduction (the temperature was not given). The enzyme complex most sensitive to volatile anesthetics in these studies is complex I. With increasing concentrations of anesthetics, other components of the respiratory chain are also inhibited. [38,39]Because the GAS-1 homologues function in a complex shown to be anesthetic sensitive, it seems unlikely that the gas-1 mutation somehow makes complex I a de novo target of volatile anesthetics. Furthermore, if inhibition of complex I by volatile anesthetics causes the anesthetized state (either directly or indirectly), then fc21 cannot be a total loss of function, because mutant worms are not anesthetized in air. Rather, they are vigorous movers, albeit with a shortened life span and decreased brood size. This corresponds with our earlier findings consistent with fc21 being a hypomorphic mutation.

Given these data, fc21 may cause worms to be hypersensitive to volatile anesthetics by further decreasing activity of complex I, which is ordinarily a target for inhibition by volatile anesthetics. In this case, the changes in the GAS-1 protein directly affect anesthetic sensitivity. An increased sensitivity to volatile anesthetics is seen because mitochondrial function is now easier to disrupt by volatile anesthetics; that is, a lower concentration of volatile anesthetic is necessary to cause the acute decrease in complex I function that is necessary for the anesthetized state. We are doing studies to determine whether mitochondrial function is more sensitive to volatile anesthetics in gas-1 mutants than in wild-type animals.

Alternatively, changes in gas-1 may indirectly affect sensitivity to volatile anesthetics. Several possibilities exist for secondary effects of GAS-1 mutations. A decreased activity of complex I may alter the availability of adenosine triphosphate in cells. A myriad of cellular functions then may be affected, any of which may change an anesthetic target. It is not clear how such a state relates to that of acute hypoxia, but it is worth noting that very low oxygen tension values also decrease minimum alveolar concentration in dogs. [41]A decreased availability of high-energy phosphates may decrease posttranslational phosphorylation of proteins, which are the direct target of volatile anesthetics. Phosphorylation modulates the response of several ligand-gated ion channels to their natural ligands, such as glutamate and GABA. [42]This may tie the current study to those that implicate such channels in the control of anesthetic response. [43,44]However, the effects of phosphorylation of ion channels on anesthetic sensitivity have not been studied.

In addition to controlling respiration, complex I also generates ubiquinol, which quenches free radicals and limits lipid peroxidation (Figure 4). The damage initiated by lipid peroxidation could also make other targets more susceptible to volatile anesthetics. Oxidative damage to membrane components has been noted during the aging process, [45]when sensitivity to volatile anesthetics also increases. [46,47]It is worth noting that sensitivity to volatile anesthetics increases as the gas-1 animals age; that is, adults are considerably more sensitive than young larvae. However, the effects of lipid peroxidation on anesthetic sensitivity have not been studied.

Because we use immobility as an end point, fc21 may mediate a paralysis that is not directly analogous to the phenomenon of general anesthesia in humans. For example, gas-1 might be expressed only in muscle cells, and its inhibition specifically causes muscle paralysis. The cellular expression pattern of gas-1 is not yet known, so this cannot be ruled out. However, given that 49-kDa (IP) proteins are essential to NADH-dehydrogenase (complex I) function, GAS-1 (or a homologue thereof) also will be present in mitochondria of the nervous system. Whatever the pattern of expression of gas-1 in C. elegans, in mammals homologues of GAS-1 undoubtedly play a crucial role in respiration in the nervous system. The degree of similarity between the C. elegans 49-kDa (IP) protein and its mammalian counterpart is extraordinary. This makes it compelling that any interactions between volatile anesthetics and the GAS-1 protein are likely to be found in the similar systems present in mammals.

We cannot differentiate by what mechanism gas-1 exerts its effects on anesthetic sensitivity. However, the same mitochondrial complex previously shown to be affected by volatile anesthetics in vitro is identified as important in a whole animal. Because most neurophysiologic functions are energy dependent, it is likely that many will be affected by defects in mitochondrial function. Even if anesthetics have multiple sites of action, a mitochondrial mutation could affect some or all of these multiple sites. The fact that gas-1 mutations override changes in anesthetic sensitivity caused by other mutations is consistent with this possibility. This report represents the first molecular identification of a gene product affecting sensitivity of an intact animal to all volatile anesthetics.

The authors thank Paul Sternberg for providing lin-15, and Alan Coulson and Steve Jones for sharing unpublished data about the sequence of the cosmid, K09A9, and Howard Nash and Helmut Cascorbi for their continued support and critical discussions.