Volatile general anesthetics inhibit neurotransmitter release by an unknown mechanism. A mutation in the presynaptic soluble NSF attachment protein receptor (SNARE) protein syntaxin 1A was previously shown to antagonize the anesthetic isoflurane in Caenorhabditis elegans. The mechanism underlying this antagonism may identify presynaptic anesthetic targets relevant to human anesthesia.
Sensitivity to isoflurane concentrations in the human clinical range was measured in locomotion assays on adult C. elegans. Sensitivity to the acetylcholinesterase inhibitor aldicarb was used as an assay for the global level of C. elegans neurotransmitter release. Comparisons of isoflurane sensitivity (measured by the EC50) were made by simultaneous curve fitting and F test as described by Waud.
Expression of a truncated syntaxin fragment (residues 1-106) antagonized isoflurane sensitivity in C. elegans. This portion of syntaxin interacts with the presynaptic protein UNC-13, suggesting the hypothesis that truncated syntaxin binds to UNC-13 and antagonizes an inhibitory effect of isoflurane on UNC-13 function. Consistent with this hypothesis, overexpression of UNC-13 suppressed the isoflurane resistance of the truncated syntaxins, and unc-13 loss-of-function mutants were highly isoflurane resistant. Normal anesthetic sensitivity was restored by full-length UNC-13, by a shortened form of UNC-13 lacking a C2 domain, but not by a membrane-targeted UNC-13 that might bypass isoflurane inhibition of membrane translocation of UNC-13. Isoflurane was found to inhibit synaptic localization of UNC-13.
These data show that UNC-13, an evolutionarily conserved protein that promotes neurotransmitter release, is necessary for isoflurane sensitivity in C. elegans and suggest that its vertebrate homologs may be a component of the general anesthetic mechanism.
AT clinical concentrations, volatile general anesthetics (VAs) such as isoflurane have multiple electrophysiologic effects that depress overall nervous system activity and likely contribute to their mechanism of action.1By clinical concentrations, we mean concentrations of anesthetic that are in the range used in human clinical practice; 2 minimal alveolar concentration of isoflurane produces an aqueous concentration of 0.62 mm. Therefore, we operationally define anesthetic concentrations less than 0.6 mm as clinical concentrations. One of the actions of clinical concentrations of VAs is inhibition of neurotransmitter release.2The mechanism of this inhibition is poorly understood. Release of glutamate and γ-aminobutyric acid (GABA) from rat cortical synaptosomes is inhibited by VAs, and inhibition of sodium channels blocks the effect of VAs on 4-aminopyridine-evoked release but not on basal release.3,4VAs more efficaciously inhibit glutamate release compared with GABA release from synaptosomes. Sodium channel blockade does not explain the differential inhibition by VAs of the basal release of glutamate and GABA.3In rat hippocampus, VAs have been shown to inhibit glutamatergic transmission by a primarily presynaptic mechanism.5Subsequent studies confirmed a presynaptic VA action in the hippocampus and attributed approximately a third of the inhibition of glutamate release to a reduction in the action potential and, by default, the remainder of the effect to downstream targets such as the transmitter release machinery.6,7As was found with synaptosomes, VAs selectively inhibited glutamate versus GABA release.6
Consistent with a presynaptic anesthetic mechanism in the nematode Caenorhabditis elegans , we found in a screen through existing C. elegans mutants that mutations reducing levels of neurotransmitter release conferred hypersensitivity to the VAs halothane and isoflurane8and mutations increasing transmitter release conferred resistance.9–11These results could be explained if VAs inhibited excitatory neurotransmitter release. Indeed, isoflurane and halothane both produced behavioral and pharmacologic effects resembling mutants with reduced excitatory neurotransmission8in that anesthetized animals moved sluggishly and were resistant to the acetylcholinesterase inhibitor aldicarb. Aldicarb increases the steady state level of acetylcholine at the neuromuscular junction and thereby produces a depolarizing neuromuscular blockade; mutations or drugs that reduce acetylcholine release confer resistance to aldicarb.12
In testing all available viable alleles of genes known to regulate neurotransmitter release in C. elegans , we found one mutation in the C. elegans syntaxin-1A gene unc-64 that had an unexpected phenotype. unc-64(md130) (indicates a strain carrying the md130 mutation in the unc-64 gene) had reduced excitatory neurotransmission by behavioral, aldicarb sensitivity, and electrophysiologic measurements,8,13yet it was highly VA resistant.8For example, its isoflurane EC50was more than fivefold that of wild type, making unc-64(md130) fully resistant to isoflurane concentrations in the clinical range. unc-64(md130) is 30-fold less sensitive to isoflurane than other unc-64 reduction-of-function alleles; therefore, isoflurane resistance is not a general property of reduction of syntaxin function. The md130 mutation disrupts an intron donor splice sequence, resulting in a reduced level of full-length syntaxin and the production of novel truncated syntaxins.8By Western blot, the relative ratio of full-length to truncated syntaxin is approximately 4:1 (Barbara Scott, B.S., and C. Michael Crowder, M.D., Ph.D., Department of Anesthesiology, Washington University, St. Louis, MO, written communication, March 2007). Therefore, if truncated syntaxins are actually responsible for the anesthetic resistance, a relatively low total concentration of truncated syntaxin can antagonize anesthetic action.
Despite our previous report of isoflurane binding to syntaxin,14several genetic results argue against the most direct model where the isoflurane resistance of the md130 mutation is due to deletion of the isoflurane binding sites on syntaxin. First, the isoflurane resistance phenotype of md130 is semidominant.8Second, an unc-64 null mutation has no anesthetic phenotype as a heterozygote.8That is, unlike with the md130 mutation in the background, one copy of wild-type syntaxin confers normal isoflurane sensitivity; therefore, the md130 mutation does not behave genetically as if the isoflurane resistance is due to loss of a binding site. Third, structural and cell biologic studies strongly support a model where the portion of syntaxin deleted by the md130 mutation is absolutely required for the normal function of syntaxin.15Therefore, in unc-64(md130) , which expresses both wild-type and truncated protein, the loss of binding of isoflurane to the truncated form would have no consequence because it cannot serve as a functional syntaxin anyway. Further, the remaining wild-type syntaxin in the mutant would still bind isoflurane as usual and should be affected normally. Rather than the md130 mutation deleting an isoflurane binding site, the most parsimonious hypothesis is that these truncated syntaxins act essentially as VA antagonists against the anesthetic target. Here we show that truncated syntaxins do in fact antagonize isoflurane action, identify the likely protein target for truncated syntaxin, find that this protein is necessary for VA sensitivity, and show that VAs act to alter its synaptic localization. Therefore, this protein fits criteria for a functional VA target.
Materials and Methods
C. elegans Strains and Transformants
A list of strains used in the work is given in table 1. N2 var Bristol was the wild-type strain and the genetic background for all mutants.16,tom-1(ok285) was obtained from the C. elegans knockout consortium; ok285 is a 1,580-bp tom-1 deletion, which removes all or part of four exons in the center of the gene.17,tom-1(ok285) unc-13(e376);unc-64(js115);oxIs34 was constructed by selecting Unc non-Dpy progeny segregating from dpy-5(e51) +e376 /+ok285 + hermaphrodites and homozygosing for ok285 e376 . unc-64(js115);oxIs34 [unc-64 (L166A/E167A);Pmyo-2 ::GFP]18males were then crossed with ok285 e376 , and the best-moving Unc green second-generation progeny were clonally passaged. The presence of the homozygous tom-1(ok285) deletion was confirmed by polymerase chain reaction amplification of the mutant gene; the homozygous js115 mutation was confirmed by outcrossing from oxIs34 and observing segregation of dead larval progeny on all second-generation broods. tom-1(ok285) unc-13(s69);unc-64(js115);oxIs34 was constructed similarly and was a gift from Janet Richmond, Ph.D. (Department of Biologic Sciences, University of Illinois Chicago, Chicago, IL). For heterozygous unc-64(md130) animals, non-Unc non-Bli animals segregating from md130+/+bli-5 were used. mdIs3 [unc-13(+);snb-1 ::GFP] is an integrant of mdEx43 19and was a gift from Kenneth G. Miller, Ph.D. (Program in Molecular and Cell Biology, Oklahoma Medical Research Foundation, Oklahoma City, OK). mdIs3;gcEx45 [pTXmd130] and mdIs3;gcEx55 [pTX1-107] were constructed by crossing gcEx45 or gcEx55 males into mdIs3 and passaging progeny with both green fluorescent protein (GFP) cotransformation markers until homozygous for mdIs3 . unc-13(s69);oxIs78 [myr::unc-13 S::GFP;ccGFP] was a gift from Erik Jorgensen, Ph.D., and Kim Schuske, Ph.D. (Department of Biology, University of Utah, Salt Lake City, UT), and was generated by injection and integration of KP28020; the presence of the myristoylation consensus sequence was confirmed by sequencing the integrated transgene. The truncated syntaxin transformants were generated by gonad injection of 50–100 ng/μl of the particular pTX plasmid along with 40 ng/μl of the hypodermal GFP coinjection marker pPHgfp-121into N2 and selection of stably transformed lines.
The parent plasmid for all syntaxin plasmid constructs was pTX21, which contains a 12-kb Pst I–MfeI rescuing genomic fragment from pTX2013inserted into pBluescript (KS−) and includes all known unc-64 transcripts and transcriptional regulatory regions. pTXmd130was generated by polymerase chain reaction amplification from unc-64(md130) genomic DNA and insertion of an Nhe I–Nsi I fragment spanning the md130 mutation into Nhe I–Nsi I cut pTX21. pTX1–258, pTX1–227, pTX1–158, pTX1–106, pTX1–86, and pTX1–64were all made by one or two rounds of oligonucleotide-directed mutagenesis (Stratagene Quickchange; Stratagene, Inc., La Jolla, CA) of pTX21 to generate a stop codon at the desired site. All plasmid constructs were sequenced to confirm the mutation.
Behavioral and Drug Assays
Locomotion was measured on 2% agar plates as the fraction of animals that dispersed in 40 min from the center of a 9.5-cm plate to the edge that was seeded with bacteria (the dispersal index),10by the rate of body bends,9or by the speed of movement across an agar plate as described previously.22,unc-64(md130) is resistant in all three assays (Laura B. Metz, B.S., Department of Anesthesiology, Washington University, written communication, March 2007). However, the slow and slowest strains did not move well enough to disperse to the edge of the dispersal plates, and the slowest strains had weak and infrequent body bends that produce excessively variable results in body bend assays. Therefore, we used the relatively easy body bends assay for slow strains and the highly quantitative but tedious speed assay for the slowest strains. The dispersal assay was used for normally moving strains. Aldicarb sensitivity was measured by the rate of paralysis on 0.35 mm aldicarb–containing agar plates.23Isoflurane was delivered and the concentrations were measured as described previously.24Locomotion as measured by one of the three assays was plotted against anesthetic concentration; concentration–response data were fitted to a modified Hill equation to calculate the EC50(the anesthetic concentration at which the reduction in locomotion was half maximal) and slope.24
Scoring of Unc-13::GFP Puncta Density
L4 larval stage animals were exposed on agar plates to 0 or 1 μg/ml phorbol-12-myristate-13-acetate for 1 h. Subsequently, the plates were placed into glass chambers containing 0 or 2 vol% isoflurane for an additional 1 h and then transferred rapidly by platinum wire onto a thin agarose pad into a fresh drop of M9 buffer16containing 0 or 1 mm isoflurane (≈ 1.9 vol% isoflurane) and 10 mm azide. Agar pads were rapidly sealed with a glass coverslip. The dorsal nerve cord was imaged within 5 min of removal of worms from the glass chamber. Preliminary experiments showed that isoflurane without azide remained effective at producing sluggish movement under these conditions for at least 5 min (C. Michael Crowder, M.D., Ph.D., Department of Anesthesiology, Washington University, written communication, March 2007). The Images were obtained on a Zeiss Axioskop 2 (Carl Zeiss Microimaging, Inc., Thornwood, NY) using a 63× Plan-Apo 1.4na oil immersion lens and a Chroma 41017 Endow GFP filter set and captured using a Retiga Exi charge-coupled device camera coupled to Q-Capture Pro software (QImaging, Inc., Surrey, British Columbia, Canada) with identical capture settings for all animals. Image files were coded and scored using Scion NIH image software (Scion Corporation, Frederick, MD) by an observer blinded to condition. For each animal, puncta were manually counted along the longest length of dorsal nerve cord in the focal plane. The distance of dorsal nerve cord scored was measured by conversion of pixel number to micrometers. Puncta density was then expressed for each animal as puncta/100 μm dorsal nerve cord.
All statistical comparisons were made using GraphPad Prism 4 software (GraphPad Software, Inc., San Diego, CA). EC50s were compared for statistical differences by simultaneous curve fitting as described by Waud25using GraphPad Prism 4. For a particular strain, EC50s were estimated by pooling all of the data for that strain; the error values after the EC50values are the errors of the fit. Locomotion rates were compared by two-sided t test. GFP puncta density was compared by two-sided t test. The threshold for statistical significance for all tests was set at P < 0.01.
To test whether the truncated syntaxin was responsible for the VA resistance phenotype and, if so, to define the structural requirements for its antagonism, we generated a series of plasmids expressing truncated syntaxins under the native unc-64 promoter and tested the isoflurane sensitivity of the transformed animals (fig. 1). As expected, transformation of wild-type C. elegans with pTXmd130, which contains the original md130 mutation, conferred isoflurane resistance (figs. 1B and I). pTX1–258, which expresses a truncated syntaxin longer than that produced by md130 and without the additional amino acids produced by read through into the intron, also conferred isoflurane resistance (figs. 1C and I). Wild-type animals transformed with pTX1–227, which introduces a stop codon immediately 5′ to the splice donor site mutated by md130 , were also isoflurane resistant (figs. 1D and I). The isoflurane resistance produced by both pTX1–258and pTX1–227demonstrates conclusively that truncated unc-64 syntaxin does antagonize VA action and that the novel amino acids produced by the original md130 mutation are not required for VA antagonism. Transformation with plasmids expressing increasingly smaller truncated syntaxins, pTX1–158and pTX1–106, showed that a fragment encoding only the HABdomains of syntaxin was sufficient to produce resistance (figs. 1E, F, and I). However, further truncation removing all or half of HB, pTX1–64, or pTX1–86abolished the VA resistance–conferring activity (figs. 1G–I). Therefore, C-terminally truncated syntaxin does indeed antagonize VA action and only a relatively small N-terminal fragment is sufficient for this action. This syntaxin fragment does not include the soluble NSF attachment protein receptor (SNARE) domain, encoded by residues 185–255, that is thought to interact with the majority of syntaxin-interacting proteins.
We have previously shown that mutations that increase transmitter release in C. elegans produce resistance to VAs.9,11Therefore, an alternative to the hypothesis that the truncated syntaxin is a direct VA antagonist is that it indirectly antagonizes VA action by increasing synaptic transmitter release. We tested this hypothesis by measuring the locomotion and aldicarb sensitivity of three truncated syntaxin-expressing constructs that confer isoflurane resistance. Hyperactive locomotion and aldicarb hypersensitivity are indicative of increased neurotransmitter release.12None of the transformants were significantly hyperactive or hypersensitive to aldicarb (figs. 2A and B). Movement data for the hyperactive mutant goa-1(sy192) is included as a positive control for hyperactivity.11Data for the original unc-64(md130) , which is aldicarb resistant due to reduced levels of wild-type syntaxin,8,13and slo-1(js379) , which has a loss of function mutation in the BK Ca2+-activated K+channel that negatively regulates transmitter release,9,26are included for positive controls for aldicarb resistance and hypersensitivity, respectively. Therefore, we conclude that increased release of neurotransmitter is unlikely to be the mechanism whereby the truncated syntaxins produce VA resistance. We hypothesize that the truncated syntaxin interacts with another protein that regulates VA sensitivity and thereby antagonizes VAs.
With what might the truncated syntaxin interact to antagonize VA action? We first considered whether the putative interacting protein also interacted with wild-type syntaxin. If so, wild-type syntaxin should compete with truncated syntaxin and restore VA sensitivity. Indeed, animals with one copy of the md130 mutation and one wild-type allele were less resistant than homozygous md130 animals (figs. 3A and B). In addition, transformation of unc-64(md130) with a plasmid containing full-length syntaxin (pTXfl) partially rescues the VA resistance of md130 . Therefore, wild-type and truncated syntaxins seem to compete for interaction with the putative protein controlling VA sensitivity.
In vertebrates, two proteins that have been shown to interact with the N-terminal two helices of syntaxin are mUNC-13 and N-type calcium channels (fig. 4A).27,28,C. elegans has one mUNC-13 homolog (UNC-13) and one N-type calcium channel (UNC-2). If one of these is the relevant VA target, a null mutant should be VA resistant. Both UNC-2 and UNC-13 promote transmitter release in C. elegans by a mechanism at least partially conserved in mammals.29–34UNC-2 is thought to promote transmitter release by supplying calcium to the transmitter release machinery.29The molecular mechanisms whereby UNC-13 promotes transmitter release are more complex. UNC-13 acts in steps after docking of synaptic vesicles with the presynaptic membrane and somehow prime vesicles so that they are competent for fusion.31UNC-13’s activity to promote vesicle fusion is partially dependent on interaction with syntaxin and may involve conversion of syntaxin from a closed conformation where the H3 helix containing the SNARE domain interacts tightly with its HABCdomains to an open conformation where the H3helix is free from its self-interactions and available for binding to the SNARE domains of other SNARE proteins.18However, catalyzing the transformation of syntaxin from a closed to open conformation does not seem to be the only function of UNC-13.35We tested two severe unc-13 alleles19and one null unc-2 allele.29,36To test the essentially paralyzed unc-13 (lf) animals, we placed the unc-13 mutations in the background of three other mutations, tom-1(ok285 null) , unc-64 (L166A/E167A), and unc-64(js115 null) . Both the tomosyn loss-of-function mutation, tom-1(ok285) , and the unc-64 (L166A/E167A) mutation, which favors the open conformation of syntaxin,18have been found to suppress partially the paralyzed phenotype of unc-13 (lf) and, when combined, suppress the paralyzed unc-13 (lf) phenotype adequately for anesthetic testing.17,37,38Tomosyn is thought to form a complex with syntaxin, preventing its binding to the vesicular SNARE synaptobrevin/vesicle-associated membrane protein and thereby reducing neurotransmitter release.17Open syntaxin may partially suppress unc-13 (lf) by supplying the product of one of the normal functions of UNC-13, which is to convert closed syntaxin to open syntaxin.18The unc-64(js115) mutation is included in the strain so that normal UNC-64 syntaxin does not compete with open syntaxin and thereby diminish the positive effect on locomotion of the unc-13 mutants.18
Both unc-13 alleles were highly isoflurane resistant (figs. 4B and C). The locomotion of the null unc-13 allele s69 was not significantly decreased by isoflurane at concentrations up to 4 vol%, a concentration that maximally reduced wild-type locomotion. The partial loss-of-function e376 allele was less resistant than s69 but still highly resistant compared with the wild-type strain or with the tom-1(ok285);unc-64(js115);oxIs34 genetic background control strain (figs. 4B and C). The unc-2 null mutant was fully sensitive to isoflurane, with an isoflurane EC50not significantly different from wild type (figs. 4B and C). These results show that UNC-13 is required for isoflurane sensitivity and that the UNC-2 calcium channel is not.
Therefore, UNC-13 is required for normal VA sensitivity, but is it the target of the truncated syntaxin as originally proposed? To examine this hypothesis, we tested the ability of overexpressed wild-type UNC-13 (produced by mdIs3 ) to suppress the VA resistance of the truncated syntaxin (fig. 5), the presumption being that UNC-13 when in excess of truncated syntaxin should restore VA sensitivity. Overexpression of full-length UNC-13 did fully suppress the isoflurane resistance of both pTXmd130and pTX1–107(figs. 5A and B). However, full-length UNC-13 overexpression (mdIs3 ) did not significantly change isoflurane sensitivity in a wild-type background (fig. 5B). These results are most directly explained if truncated syntaxin physically interacts with UNC-13 to block VA sensitivity, and when UNC-13 is expressed in excess of truncated syntaxin, it becomes available for inhibition by VAs. Alternatively, UNC-13 overexpression might prevent interaction of the truncated syntaxin with some other unknown protein that is essential for VA sensitivity.
If UNC-13 is a VA target in C. elegans , how might binding of VAs disrupt UNC-13 function? mUNC-13 (mammalian UNC-13) has been shown to bind to Rab3a-interacting molecule (RIM),39calmodulin,40and syntaxin27(fig. 5A), and these interactions promote synaptic transmission.34,35,40,41Loss-of-function mutants of mouse RIM or the C. elegans homolog UNC-10 drastically reduces neurotransmitter release, and mammalian RIM has been implicated in cerebellar long-term potentiation.42–45In mice, calmodulin binding to mUNC-13 has been implicated in activity-dependent synaptic facilitation.40Whether this function of calmodulin is conserved in C. elegans is unknown. In terms of VA action in C. elegans , a reasonable hypothesis is that VAs disrupt RIM or calmodulin binding to UNC-13 or directly alter RIM/calmodulin function in a UNC-13–dependent fashion. As for unc-13 , unc-10 null mutants move slowly, and as for unc-13 mutants, this sluggishness is suppressed by open syntaxin,45allowing testing of their anesthetic sensitivity. Both the unc-10(md1117 null); open syntaxin double mutant and the open syntaxin mutant alone were fully sensitive to isoflurane (fig. 5C). These results demonstrate that UNC-10 RIM is not required for isoflurane action and that isoflurane does not act by inhibiting the interaction of UNC-10 with UNC-13. In addition, the normal sensitivity of open syntaxin–bearing animals argues against a mechanism where VAs inhibit UNC-13’s promotion of the open form of syntaxin.18Calmodulin null mutants are lethal in C. elegans . To test the role of calmodulin, we measured the isoflurane sensitivity of animals transformed with a shortened form of UNC-13, UNC-13S (fig. 5A), which lacks the calmodulin- and RIM-binding domains and partially rescues locomotion and transmitter release.20,23Animals expressing UNC-13S were normally sensitive to isoflurane (figs. 5D and F), demonstrating that neither the UNC-10–binding nor calmodulin-binding domains are necessary for VA sensitivity.
Based on four previous observations, we considered the hypothesis that VAs might antagonize diacyl glycerol (DAG) binding to and/or activation of UNC-13. First, DAG enhances the transmitter release promoting activity of UNC-13 in C. elegans and in higher organisms.20,23,32Second, VAs and anesthetic alcohols have been shown to antagonize DAG-binding to and activation of mammalian protein kinase C (PKC), which contains a C1 domain homologous to that binding DAG in UNC-13.46,47Third, anesthetic alcohols have been shown to bind to the C1 domain near the DAG binding pocket.46Finally, we have previously shown that phorbol ester, a C1 domain agonist, antagonizes VAs in C. elegans .9
To test whether VAs might act to block DAG activation of UNC-13, we made use of the ability of myristoylated UNC-13S (myr-UNC-13S) to promote transmitter release in C. elegans in a DAG-independent manner.20,23If VAs act on UNC-13 to disrupt DAG binding or activation, myr-UNC-13S by targeting itself to the membrane in a DAG-independent manner should bypass the effect of VAs on UNC-13 and should therefore be VA resistant. Indeed, animals expressing myr-UNC-13S were highly isoflurane resistant compared with those expressing UNC-13S or full-length UNC-13-unc-13(s69 lf);mdIs3 ( figs. 5E and F). We considered the possibility that myr-UNC-13S is VA resistant because it has increased levels of neurotransmitter release compared with wild type. To test this hypothesis, we compared the aldicarb sensitivity of wild-type, unc-13 S, and myr-unc-13 S strains (fig. 5G). myr-unc-13 S animals had wild-type aldicarb sensitivity, whereas unc-13 S animals were aldicarb resistant. Therefore, a general increase in neurotransmitter release does not explain the resistance of myr-unc-13 S. Rather, the most direct explanation of these results is that myristoylation of UNC-13S confers isoflurane resistance by bypassing isoflurane antagonism of DAG-mediated membrane targeting of UNC-13.
To test directly whether isoflurane acts on UNC-13 to antagonize DAG-mediated membrane targeting, we observed the effect of isoflurane on localization of a UNC-13S::GFP fusion protein. This fusion protein has previously been shown to concentrate at synapses in a DAG-dependent manner, where it is visible as distinct puncta that colocalize with other presynaptic proteins in C. elegans dorsal and ventral nerve cords.20,23,38Therefore, the density of puncta along the nerve cord is an indicator of the relative amount of presynaptic membrane-localized UNC-13. Isoflurane significantly reduced both basal and phorbol ester–stimulated UNC-13S::GFP puncta in the dorsal nerve cord (fig. 6and table 2). Therefore, we conclude that isoflurane decreases DAG-mediated synaptic localization of UNC-13. Alternatively, isoflurane could decrease overall levels of UNC-13; our data do not exclude this possibility.
We have shown that a relatively short N-terminal syntaxin-1A fragment can fully antagonize the behavioral effects of clinical concentrations of the VA isoflurane in C. elegans . The antagonism is direct in that the truncated syntaxin had no apparent effect on locomotion or synaptic concentrations of acetylcholine as measured by aldicarb sensitivity. The specificity of the truncated syntaxin coupled with its dominant genetic behavior argues that the truncated syntaxin is acting essentially as a pharmacologic antagonist against the VA target. Working under this assumption, the question becomes, to what is the truncated syntaxin binding to antagonize VAs? The number of proteins that have been shown to bind to this N-terminal segment of syntaxin is few. Testing of mutants of two such candidates showed that UNC-13 but not the non–L-type Ca2+channel was essential for sensitivity to VAs in the clinically relevant concentration range. Consistent with UNC-13 as the target of the antagonistic activity of truncated syntaxin, overexpression of UNC-13 suppressed the VA resistance phenotype of truncated syntaxin. Finally, isoflurane seems to decrease synaptic localization of UNC-13, and animals with membrane targeted UNC-13 are isoflurane resistant. Therefore, UNC-13 is required for VA sensitivity in C. elegans , it is likely the protein to which the truncated syntaxin is binding, and its localization is inhibited by and important for isoflurane action. A schematic of our working model based on these data are shown in figure 7.
The questions now posed by our findings are several, including why has UNC-13 not been previously implicated in anesthetic action; is UNC-13 a direct anesthetic target; and what might be the role, if any, of UNC-13 orthologs in vertebrate anesthesia? Multiple reasonable explanations can be posited as to why no direct evidence till now has implicated UNC-13/mUNC-13 in anesthetic action. First, the primary methodology for most anesthetic mechanism studies is electrophysiology. While multiple electrophysiologic studies have shown that VAs do inhibit neurotransmitter release in various preparations,2no specific mUNC-13 inhibitors are available that might have implicated mUNC-13. Moreover, inhibition of UNC-13/mUNC-13 does not produce a distinct electrophysiologic phenotype that would clearly implicate these proteins. Likewise, binding studies that might have identified UNC-13/mUNC-13 as a VA target are severely limited by the low affinity and high membrane partitioning of VAs. Indeed, no synaptic protein has been specifically identified in binding studies on crude membrane preparations despite the presence of relatively abundant proteins such as GABAAreceptors, which are likely to be direct VA targets.1
UNC-13 has several properties suggesting that it is a direct and relevant anesthetic target in C. elegans . First, animals lacking UNC-13 are VA resistant. Resistance is neither a necessary nor a sufficient feature of an anesthetic target because of the possibility of multiple targets, genetic redundancy, and indirect effects. Nevertheless, the high-level resistance of the unc-13 mutants is consistent with the VA target being UNC-13. Second, a mutation in UNC-13 (a myristoylation sequence) that otherwise does not disrupt UNC-13 function confers isoflurane resistance, and UNC-13 synaptic localization seems to be decreased by isoflurane. These results are difficult to explain by an indirect model. One reasonable indirect model to consider is that VAs act to prevent production of DAG and thereby inhibit the function of UNC-13. Such an anesthetic mechanism would depend on UNC-13, be circumvented by myr-UNC-13S, and decrease UNC-13 synaptic localization. However, we have previously shown that a probable null mutation in egl-8 , which encodes the only known phospholipase Cβ acting upstream of UNC-13 to stimulate transmitter release, is not as VA resistant as the unc-13 or truncated syntaxin mutants.9Moreover, this mechanism does not explain the resistance produced by the truncated syntaxin. On the other hand, a model where the truncated cytoplasmic syntaxin competes with anesthetics for binding to UNC-13 is plausible and consistent with the existing data.
Is there any evidence that VAs bind to UNC-13 or its vertebrate homologs? Although no binding experiments have been reported with UNC-13/mUNC-13, VAs and anesthetic alcohols have been shown to inhibit PKC,48which has a C1 domain structurally similar to the C1 domain in UNC-13. Subsequent studies found that anesthetic alcohols do bind to the C1 domain of bovine PKC-α and compete for DAG binding.47More recently, a photoaffinity anesthetic alcohol labeled tyrosine 236 of the C1 domain of PKC-δ.46The x-ray crystal structure of the C1B domain of PKC-δ places tyrosine 236 approximately 10 Å from the DAG binding pocket46and provides a structural explanation for how anesthetics might inhibit DAG binding to the C1 domain. The structure of mUNC13-1 C1 domain has been compared with that of PKC-δ by solution nuclear magnetic resonance spectroscopy and found to be generally similar with conservation of the location of the homologous tyrosine.49However, a tryptophan residue was found to overlie the DAG-binding pocket in mUNC13-1 but not in PKC-δ. This structural difference was proposed to account for the lower affinity of mUNC13-1 for phorbol esters. Therefore, while anesthetic binding to the C1 domain of PKC-α and PKC-δ suggests that anesthetics are likely to also bind to the C1 domain of UNC-13/mUNC-13, the structural differences between the proteins offer the possibility that the affinities and/or effect of binding may differ between the homologous domains.
What might be the role of vertebrate UNC-13 homologs in general anesthesia? This question can be divided into two parts. First, in general, how might inhibition of neurotransmitter release contribute to general anesthesia? Second, specifically, what role might inhibition of vertebrate UNC-13 homologs play in the overall presynaptic anesthetic mechanism in vertebrates? Clearly, inhibition of excitatory neurotransmitter release could lead to an overall depression in nervous system function and contribute to a general anesthetic state. On the other hand, reducing inhibitory neurotransmitter release would counteract this effect and might actually increase arousal. In particular for VAs, a block of inhibitory release would likely reduce anesthetic efficacy because of the well-established postsynaptic VA action of potentiation of ligand gating of inhibitory GABAAand glycine receptors. However, as outlined in the introduction, at clinical concentrations, VAs including isoflurane selectively inhibit neurotransmitter release, reducing glutamate release significantly more than GABA release. Therefore, VAs could reduce excitatory neurotransmission by its presynaptic mechanism while at the same time, because of minimal effect on inhibitory neurotransmitter release, potentiate postsynaptic inhibitory GABA and glycine currents and thereby synergistically depress central nervous system activity. Besides this potential for synergy, in particular brain regions such as the hippocampus, presynaptic excitatory inhibition seems to be a predominant effect.5,6,50Therefore, presynaptic effects could be particularly critical for the amnestic effects of VAs.
As to the role of UNC-13 orthologs in presynaptic VA action in vertebrates, one must consider that VAs almost certainly act on other presynaptic targets besides UNC-13 homologs. Sodium channels have been strongly implicated as essential for a significant portion of VA presynaptic inhibition.3,4,51Therefore, unlike in C. elegans , which lacks sodium channels, UNC-13 orthologs are unlikely to be the sole presynaptic VA target in mammals. However, the mammalian UNC-13 homologs, mUNC13-1, 2, and 3, have interesting distinct functional roles that could account for the synapse-selective effects of VAs. Specifically, mUNC-13 isoforms are differentially expressed in GABA versus glutamate terminals. Release from the majority of glutamatergic terminals in mouse hippocampus requires the mUNC13-1 isoform, whereas mUNC13-1 and mUNC13-2 function redundantly in GABAergic release at least in the cerebral cortex and hippocampus.52In rat brain, mUNC13-1 is expressed throughout the central nervous system, whereas mUNC13-2 expression is restricted to the cerebral cortex and hippocampus. mUNC13-3 seems to be expressed exclusively in the cerebellum.53Intriguingly, mUNC13-1– and mUNC13-2–mediated release differ in their potentiation by DAG; mUNC13-1 is less efficaciously potentiated.52These previous observations coupled with the results reported here suggest that mUNC13-1, the closest homolog to C. elegans UNC-13, may be more sensitive to VAs because its weak DAG potentiation is more efficaciously blocked by VAs compared with that of mUNC13-2. The availability of mouse knockout strains for each of the mUNC13 isoforms will allow testing of this hypothesis.
The authors thank Owais Saifee, M.D., Ph.D., and Mike Nonet, Ph.D. (Department of Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, Missouri), for their seminal contributions to this work; Janet Richmond, Ph.D. (Department of Biologic Sciences, University of Illinois Chicago, Chicago, Illinois), for sharing of unpublished results and strains; and Kim Schuske, Ph.D., and Erik Jorgensen, Ph.D. (Department of Biology, University of Utah, Salt Lake City, Utah), for the unpublished oxIs78 array.