Analysis of Caenorhabditis elegans acetylcholine synthesis mutants reveals a temperature-sensitive requirement for cholinergic neuromuscular function

Abstract

In Caenorhabditis elegans, the cha-1 gene encodes choline acetyltransferase (ChAT), the enzyme that synthesizes the neurotransmitter acetylcholine. We have analyzed a large number of cha-1 hypomorphic mutants, most of which are missense alleles. Some homozygous cha-1 mutants have approximately normal ChAT immunoreactivity; many other alleles lead to consistent reductions in synaptic immunostaining, although the residual protein appears to be stable. Regardless of protein levels, neuromuscular function of almost all mutants is temperature-sensitive, i.e., neuromuscular function is worse at 25° than at 14°. We show that the temperature effects are not related to acetylcholine release, but specifically to alterations in acetylcholine synthesis. This is not a temperature-dependent developmental phenotype, because animals raised at 20° to young adulthood and then shifted for 2 h to either 14° or 25° had swimming and pharyngeal pumping rates similar to animals grown and assayed at either 14° or 25°, respectively. We also show that the temperature-sensitive phenotypes are not limited to missense alleles; rather, they are a property of most or all severe cha-1 hypomorphs. We suggest that our data are consistent with a model of ChAT protein physically, but not covalently, associated with synaptic vesicles; and there is a temperature-dependent equilibrium between vesicle-associated and cytoplasmic (i.e., soluble) ChAT. Presumably, in severe cha-1 hypomorphs, increasing the temperature would promote dissociation of some of the mutant ChAT protein from synaptic vesicles, thus removing the site of acetylcholine synthesis (ChAT) from the site of vesicular acetylcholine transport. This, in turn, would decrease the rate and extent of vesicle-filling, thus increasing the severity of the behavioral deficits.

Introduction

Choline acetyltransferase (ChAT, EC 2.3.1.6) is the enzyme that synthesizes the neurotransmitter acetylcholine (ACh). ChAT enzymatic activity in rat brain was first described almost 80 years ago (Nachmansohn and Machado 1943), and ChAT has long been considered the definitive marker of the cholinergic neuronal phenotype (Wu and Hersh 1994). The availability of specific antibodies to the enzyme has permitted extensive immunohistochemical mapping of cholinergic pathways in many vertebrate and invertebrate nervous systems (Eckenstein and Thoenen 1982; Gorczyca and Hall 1987; Duerr et al. 2008). In addition, in vitro mutagenesis studies of the Drosophila and rat enzymes have identified catalytically essential sequence motifs of the protein (Wu and Hersh 1995). Investigations of transcriptional regulation of ChAT have identified both positive and negative regulatory promoter elements controlling ChAT transcription and alternative splicing in vertebrates and invertebrates (Kitamoto and Salvaterra 1993, 1995; Shimojo et al. 1998; Mathews et al. 2015; Yan et al. 2016).

In the analysis of many complex biological systems, a genetic (i.e., mutational) approach can perturb individual system components specifically and reproducibly, and thus can be a powerful complement to biochemical and physiological approaches. However, only in humans, Drosophila, zebrafish, and Caenorhabditis elegans has it been feasible to identify or obtain individuals harboring mutations in the ChAT structural gene in order to evaluate the behavioral consequences of such mutations (Greenspan 1980; Rand and Russell 1984; Ohno et al. 2001; Joshi et al. 2018).

The congenital myasthenic syndromes (CMS) are a set of hereditary disorders of muscle weakness accompanied by a decrease in cholinergic neuromuscular transmission (Mora et al. 1987; Walls et al. 1993; Shen et al. 2011). The molecular lesions associated with many of these disorders have been traced to mutations in postsynaptic proteins (e.g., acetylcholine receptor subunits, rapsyn) and synaptic cleft proteins (e.g., acetylcholinesterase, collagen Q) (reviewed in Arredondo et al. 2015; Beeson 2016). There are also relatively rare CMS disorders associated with mutations affecting presynaptic proteins, including a specific type of CMS, termed “CMS with episodic apnea” (CMS-EA), most frequently caused by missense mutations in the human ChAT gene (Ohno et al. 2001; Arredondo et al. 2015; McMacken et al. 2018). More than 29 different ChAT missense mutations have been identified in CMS-EA patients, and most of the CMS-EA patients were shown to carry two different mutant ChAT alleles (i.e., they were heteroallelic). Moreover, in vitro analysis demonstrated that most of the mutant enzymes had altered K_m and/or V_max values (Ohno et al. 2001; Shen et al. 2011), and were therefore likely to contribute to behavioral dysfunction.

ChAT in Drosophila is encoded by the Chat gene (Greenspan 1980). Although most Chat alleles are lethal as homozygotes, three of these lethal alleles are temperature-sensitive: at 18°-19°, homozygotes (as well as heterozygotes containing these alleles in trans to a deficiency) develop normally, but when grown at 29°-30°, they die as late-stage embryos or early first instars (Greenspan 1980; Takagawa and Salvaterra 1996; Wang et al. 1999). Animals that are grown at 19° and subsequently transferred to 30° become paralyzed within 48–72 h, and die shortly thereafter (Greenspan 1980; Gorczyca and Hall 1987; Kitamoto et al. 2000). In vitro assays demonstrated that the ChAT activity present in homogenates derived from the temperature-sensitive mutants was more thermolabile than the wild-type enzyme and decreased irreversibly when incubated at 38° (Greenspan 1980; Salvaterra and Kitamoto 2001).

Caenorhabditis elegans ChAT is encoded by the cha-1 gene (Rand and Russell 1984). The gene was cloned and sequenced by Alfonso et al. (1994b), who also showed that it was part of a complex “operon” with the unc-17 gene, which is nested in the first intron of cha-1 (Alfonso et al. 1994a). unc-17 was shown to encode the synaptic vesicle acetylcholine transporter (VAChT) (Alfonso et al. 1993), and cha-1 and unc-17 transcripts are produced by alternative splicing of a common precursor transcript (Alfonso et al. 1994a). Subsequent investigations revealed that the nested gene structure of this “cholinergic locus” is conserved across phyla from insects to mammals (Bejanin et al. 1994; Erickson et al. 1994; Eiden 1998; Kitamoto et al. 1998).

In C. elegans, cha-1 mutations that eliminate all gene function (null mutations) are lethal: such mutant homozygotes are able to complete embryonic development and hatch, but do not move, eat, or grow, and they soon die (Rand 1989; Alfonso et al. 1994b). In addition, many viable cha-1 mutants have been isolated that contain low levels of residual enzyme activity. These mutants are small, slow-growing, uncoordinated, and resistant to inhibitors of cholinesterase (Rand and Russell 1984). Some of these mutants were reported to be temperature-sensitive, i.e., they were behaviorally normal at 16° but uncoordinated at 25° (Hosono et al. 1985; Clark et al. 1997), or else the mutant phenotype was considerably stronger at the higher temperature (Rand and Russell 1984; Rand 1989).

In the present study, we examine viable cha-1 mutants—those with some residual gene function and thus some residual ChAT activity. Most of these are missense mutations, and we show that although many of them result in significant decreases in neuronal ChAT immunoreactivity, there seems to be little correlation between the loss of ChAT immunoreactivity and the extent of behavioral impairment. We also demonstrate an unusual feature of cha-1, namely, that almost all hypomorphic cha-1 mutations confer temperature-sensitive development and behavior. We propose that, rather than a fortuitous set of mutations, which all lead to a temperature-sensitive ChAT protein, our data reveal a temperature-sensitive requirement for localized ACh synthesis.

Materials and methods

Growth and culture

Caenorhabditis elegans was grown on solid medium as described by Brenner (1974) modified by the addition of streptomycin and nystatin to reduce contamination, and the use of the streptomycin-resistant bacterial strain OP50-1 (Johnson et al. 1988). Except where specified, animals were grown at 20°.

Strains and origin of alleles

Many of the cha-1 alleles used in the present study have been described elsewhere (Rand and Russell 1984; Hosono et al. 1987; Rand 1989; Alfonso et al. 1994b). cha-1(md84) and cha-1(md88) were ethyl methane sulfonate-induced mutations isolated in a non-complementation screen for aldicarb resistance as previously described (Alfonso et al. 1993). The cha-1 alleles ky48, ky49, ky50, and ky52 were isolated by Scott Clark and Cori Bargmann (Clark et al. 1997), the ox52 allele was isolated by Erik Jorgensen, the y226 allele was isolated by Mike Nonet and Barbara Meyer (Zhao and Nonet 2000), and the ok2253 deletion allele was isolated by the C. elegans Deletion Mutant Consortium (2012). The unc-17(p279), unc-17(md1447), unc-41(e1162), and snt-1(md290) mutations have been described previously (Zhu et al. 2001; Nonet et al. 1993; Mathews et al. 2012; Mullen et al. 2012).

In a previous report (Rand 1989), the temperature-sensitive cha-1 mutations p1182 and p1186 were described. However, although the mutations gave rise to slightly different phenotypes, a subsequent review of the isolation notes suggested that p1182 and p1186 might not have arisen independently. During genetic construction of strains containing p1186, a somewhat more severe isolate was obtained, which closely resembled the phenotype of p1182 homozygotes; this “new” isolate was designated md39 and failed to complement cha-1. Sequence analysis of these mutations showed that all three were associated with the same cha-1 missense mutation. We therefore conclude that p1182, p1186, and md39 most likely represent the same original mutational event at the cha-1 locus, and in our studies, we report only the data with md39.

Sequencing of cha-1 mutations

The sequence of the wild-type cha-1 cDNA and genomic region has previously been reported (Alfonso et al. 1994a, 1994b). A set of overlapping 1–2 kb PCR fragments were synthesized, either by amplification from mutant-derived DNA samples or by direct “single-worm PCR” from individual mutant animals (Williams et al. 1992). Most mutations were then localized to a particular fragment using a heteroduplex mismatch-cleavage method described by Cotton et al. (1988) and Barstead and Waterston (1991), followed by sequencing with internal primers directly from the PCR product. A few of the mutations were analyzed by direct (“brute-force”) sequencing of overlapping fragments of PCR-amplified genomic DNA. All DNA sequencing was performed at the Oklahoma Medical Research Foundation DNA Sequencing Core Facility, using oligonucleotide primers obtained from Integrated DNA Technologies (Coralville, IA, USA).

Behavioral tests

Behavioral tests were performed on young adult animals (12–24 h after the onset of egg-laying) grown in synchronous culture at 20° and shifted to the assay temperature for 2 h before testing. Swimming rates were measured in 10 animals of each strain as described by Miller et al. (1996). Individual worms were placed in a 60-μl droplet of temperature-equilibrated M9 buffer on an agar surface. After 1 min for acclimatization, the number of changes in the direction of bending at the mid-body was recorded for 3 min. Pharyngeal pumping was observed at 100X and was quantitated by counting the number of pumps in 1 min for 50 animals per strain. Animals were chosen at random by assaying the first worm in the field of view (except for worms that had crawled off the bacterial food source), and each animal was removed from the plate after counting. A pump was defined as a closing and opening of the terminal bulb (not necessarily the jerking of the pharynx back into the gut).

Immunocytochemical procedures and microscopy

The isolation and properties of antibodies to the C. elegans ChAT and VAChT/UNC-17 fusion proteins have been described (Duerr et al. 2008). As published previously, the subcellular pattern of anti-ChAT immunoreactivity varies for different monoclonal antibodies to ChAT. Staining with mAb 1402, 1415, and 1432 shows a more synaptic pattern, with significant colocalization with the vesicular acetylcholine transporter (VAChT/UNC-17). mAb 1401 and 1414 label puncta and cytoplasm in neuronal somas, with less intense staining in those same synaptic regions. This was hypothesized to reflect different states of ChAT with different epitope availability; neither class of antibodies stains L1 lethal cha-1 deletion mutants. The pattern of immunoreactivity was evaluated for both classes of anti-ChAT mAbs, double-stained with either rabbit 383 anti-UNC-17 or chicken 96 anti-UNC-17 as a positive control. Nematodes were grown at 15°, 20°, or 23–25° (if viable); strains that were inviable at 23–25° were grown at 15° and transferred to 23–25° for 1 day. All strains were stained using a modified freeze-fracture procedure and methanol acetone fixation as previously published (Duerr et al. 1999; Mullen et al. 2006). Visual evaluation of relative levels of staining was done blind on groups of strains fixed and stained together. At least three animals per genotype per slide were evaluated from at least three different groups. Since relative intensity is difficult to evaluate by eye, strains were scored by whether specific neural features (e.g., the anterior dorsal sublateral nerve fibers) were labeled brightly enough to be detected with e.g., the 20× objective. For a selection of strains, confocal images were collected blind to genotype with identical parameters using a Zeiss LSM-510 scanning confocal microscope. Collection parameters were set so that the intensity of individual varicosities in wild-type worms was below maximum. Then maximum projections were used to compare these varicosity intensities in different strains.

Genome comparisons

Genomic sequences of cholinergic gene loci were identified by BLAST searches. Nematode sequences were then downloaded from either the Caenorhabditis Genomes Project (http://Caenorhabditis.org, accessed June 4th, 2021), WormBase (https://wormbase.org, accessed June 4th, 2021), or ParaSite (https://parasite.wormbase.org, accessed June 4th, 2021); genomic sequences from all other species were downloaded from GenBank (https://ncbi.nlm.nih.gov, accessed June 4th, 2021). Sequences were aligned using either MAFFT (Katoh and Standley 2013) or ClustalW (Thompson et al. 1994) and if necessary, predicted gene structures were corrected and adjusted by inspection.

Data availability

The authors affirm that all data necessary for confirming the conclusions of the article are present within the article text, figures, tables, and Supplementary Files. However, a number of strains used have been frozen for a very long time, during which they have been subjected to several thermal challenges. Some of our frozen stocks have been successfully thawed within the past year, but others may not be recoverable. We are therefore unable to certify the viability of specific strains from our freezers. However, many of the alleles described in this manuscript, e.g., cha-1 alleles b401, cn101, m324, md39, ok2253, p503, p1152, and y226, snt-1(md290), and unc-17(e1447) are available from the Caenorhabditis Genetics Center (https://cgc.umn.edu, accessed June 4th, 2021). In addition, alleles with ky and ox prefixes may be requested from the laboratories of origin. Finally, with current technology, it is now possible to reconstruct any desired missense mutation in a relatively short time.

Supplementary File S1 contains genomic flanking sequences for all the cha-1 alleles described in the present manuscript. Copies of Table 1 and Supplementary File S1 are being provided to WormBase. Supplementary File S2 provides the names of the 145 species used for the analysis in Figure 6. Supplementary File S3 is a protein sequence alignment of C. elegans ChAT with the 69-kDa isoform of human ChAT, including the locations of mutations in both enzymes, and other landmarks. Supplementary File S4 contains a list of the strains and alleles used in these studies, as well as a list of the Literature Cited in all four Supplementary Files. Supplementary material is available online at G3.

Table 1.

Properties of cha-1 mutants

Allele	Protein change a	Staining intensity b	Temperature-sensitive?	Notes	References c
N2 (+)	None	+++	No	Wild type
b401	S454L	+	Yesd		Rand and Russell (1984)
cn101	D430N	+	Yese		Rand (1989)
ky48	P99S	n.d.	Yese		Clark et al. (1997)
ky49	G316R	n.d.	Yese		Clark et al. (1997)
ky50	R168K	n.d.	Yese		Clark et al. (1997)
ky52	P97L	n.d.	Yese		Clark et al. (1997)
m324	deletion	n.d.	n.d.	156-bp deletion; lethal	Rogalski et al. (1988) and Alfonso et al. (1994b)
md39	A499D	+	Yesd	ts-lethal	Duerr et al. (2008)
md84	W266stop	n.d.	n.d.	Lethal	This study
md88	P104S	+++	Yesf		This study
md1067	none	+	n.d.	Tc1 insertion	Alfonso et al. (1994b)
md1143	none	±	n.d.	Tc1 insertion	Alfonso et al. (1994b)
md1162	none	+	n.d.	Tc1 insertion	Alfonso et al. (1994b)
md1422	none	+	Yesf	Tc1 insertion	Alfonso et al. (1994b)
ok2253	deletion	n.d.	n.d.	1712-bp deletion; lethal	C. elegans Deletion Mutant Consortium (2012)
ox52	G237E	n.d.	Yese		E. Jorgensen, unpublished
p503	A455T	++	±g		Rand and Russell (1984)
p1152	P584L	+++	Yesf		Rand and Russell (1984)
p1154	P104S	+++	±		Rand and Russell (1984)
y226	T557I	++	Yese	ts-lethal	Zhao and Nonet (2000)

Alleles are listed alphabetically; their positions in the gene and protein are shown in Figure 1. The mutations listed as “lethal” are unconditional lethal mutations, i.e., they are lethal at all temperatures. n.d., not determined.

^a Flanking genomic DNA sequences for each allele are presented in Supplementary File S1.

^b Staining evaluated blind in adult worms after growth at 15°, 20°, or 25° or, for alleles md39 and y226, after growth at 15° followed by 1 day at 23–25°. Intensity was assessed by the finest processes detectable with a 20× objective: +++, sublateral nerve cords; ++, dorsal nerve cord; +, ventral nerve cord; ±, nerve ring.

^c References listed describe the isolation and phenotypes of the cha-1 mutants; all sequence data (except for ok2253) were obtained in the present study.

^d Temperature-sensitive growth (Rand and Russell 1984).

^e Mutant isolated on basis of temperature sensitivity.

^f Present study.

^g Although these mutants are not behaviorally temperature-sensitive, purified ChAT from p503 homozygotes is significantly more thermolabile than wild-type enzyme (Rand and Russell 1984).

Figure 6.

Phylogenetic Conservation of C. elegans ChAT Mutation Sites. (A) Sites of C. elegans cha-1 Missense Alleles: 145 different species were divided into four groups: the first group contained 56 other species from the genus Caenorhabditis; the second group (Other Nematodes) contained 39 representative species; the third group (non-nematode Invertebrates) included 32 species; and the fourth group (Chordates and Vertebrates) included 18 species (Supplementary File S2). Except for the group of Caenorhabditis species, the second, third, and fourth groups included representative species from diverse clades. The ChAT protein sequences within each species group were aligned with each other and with the C. elegans ChAT sequence. For each of the 12 sites corresponding to a cha-1 missense mutation (Table 1), we scored the number of species in each group containing the same amino acid at that site. If at least 85% (an arbitrary cutoff) of the species in a group matched the C. elegans amino acid, that amino acid was considered to be conserved. In the upper matrix , each cell represents a missense site evaluated in one of the four groups of species. The percent of species matching the C. elegans amino acid is shown, along with the fractional raw data count. Note that in some of the cells, the denominator of the fraction is less than the total number of species in the group—this is because some of the species have small assembly gaps within the ChAT genomic sequence. (B) Control: Conservation of Every 50th Amino Acid: A control set of 12 amino acids was chosen—every 50th position in the C. elegans ChAT sequence from #50 to #600. This set of amino acids was compared to the sets of aligned ChAT sequences already described, and the resulting matches are presented as above. For A and B, percentages >85% are shown in magenta. ^a The species in each group are listed in Supplementary File S2. ^b md88 and p1154 are independently isolated cha-1 mutations associated with exactly the same C>T transition mutation.

Results

Molecular analysis of cha-1 mutations

The amino acid sequence alterations associated with 13 independent viable cha-1 mutations are presented in Table 1 and depicted graphically in Figure 1. Also shown are the sites of the four previously described Tc1 transposon insertions and three lethal alleles (two deletions and a nonsense mutation) (Alfonso et al. 1994b). The genomic sequences flanking each of the cha-1 mutations are listed in Supplementary File S1. The mutation sites do not appear to be clustered in any particular part of the protein, although as described below, most of the amino acid substitutions affect phylogenetically conserved amino acids.

Figure 1.

cha-1 exon structure and cha-1 mutations. Shown are the splicing pattern and exon structure of the C. elegans cha-1 transcript (Alfonso et al. 1994b). Protein-coding regions are blue, and the 5ʹ- and 3ʹ-untranslated regions are gray. Also shown are the locations of missense mutations and their predicted amino acid substitutions (see Table 1) as well as the sites of the previously described Tc1 transposon insertions and m324 deletion (Alfonso et al. 1994b). The mutations shown in red are homozygous lethal alleles.

Immunostaining indicates that many cha-1 mutants have reduced levels of synapse-associated ChAT protein

We used anti-ChAT monoclonal antibodies to assess the degree of residual ChAT immunoreactivity in the viable cha-1 mutants. Strains were stained on at least three different occasions, using two different classes of anti-ChAT monoclonal antibodies with different specificities (Duerr et al. 2008), as well as polyclonal antibodies to the vesicular acetylcholine transporter VAChT/UNC-17 as a positive control (Duerr et al. 2008). As shown in Table 1 and Figure 2, some of the viable cha-1 mutants (md88, p1152, p1154, and y226) had approximately normal synaptic ChAT immunoreactivity (i.e., colocalized with the vesicular acetylcholine transporter), but the remaining alleles were associated with consistently decreased synaptic ChAT immunoreactivity. Interestingly, somatic or cytoplasmic levels of ChAT immunoreactivity were not necessarily affected in the mutants with lower levels of synaptic staining (Figure 3). In addition, all viable cha-1 mutants had normal immunostaining for UNC-17 (Figure 3). We note that even in the mutants in which synaptic ChAT protein was significantly less abundant than in wild-type animals, the remaining ChAT appeared to be properly trafficked and localized (Figures 2 and 3).

Figure 2.

Synaptic ChAT immunostaining at different temperatures. ChAT immunoreactivity shows similar patterns in wild-type and cha-1 mutants with the predominantly synaptic antibodies (mAb 1402 and 1414) at 15° (A, C, E) and 25° (B, D, F). (Duerr et al. 2008). Staining in wild type (A, B), cha-1(p503) (C, D), and cha-1(md39) (E, F). Overall intensity appears somewhat lower after 1 day at 25° in all strains; immunoreactivity in neuronal processes, such as the anterior sublateral nerve cords (arrows) is similar in wild type and cha-1(p503) and lower in cha-1(md39). Anterior is left; dorsal is up; NR, nerve ring; DNC, dorsal nerve cord; VNC, ventral nerve cord. Scale bar is 30 μm.

Figure 3.

Somatic ChAT and synaptic VAChT immunostaining at different temperatures. ChAT immunoreactivity shows similar patterns in wild type and cha-1 mutants with the predominantly somatic antibodies (Mab 1401, 1415, 1432). Color panels show partial colocalization of anti-VAChT (green) and anti-ChAT (red) in cholinergic neurons; black and white panels show ChAT alone for clarity. Young adults were kept 1 day at 15° (A, B, E, F, I, J) or 25° (C, D, G, H, K, L) (Duerr et al. 2008). Staining in wild type (A, B, C, D), cha-1(cn101) (E, F, G, H), and cha-1(y226) (I, J, K, L). Somatic staining is similar in all strains. Arrows point to characteristic ChAT immunoreactivity in processes in the isthmus of the pharynx, clearly seen in wild type. This staining was fainter in the cha-1 mutants at 25°; arrows indicate where the isthmus could be seen in the original images. Arrowheads point to neuronal cell somas in the ventral ganglia that show punctate ChAT immunoreactivity. Anterior is left; dorsal up; NR, nerve ring; DNC, dorsal nerve cord; VNC, ventral nerve cord. Scale bar is 20 µm.

cha-1 mutants have temperature-sensitive behaviors

The temperature-sensitive behavior of cha-1 mutants homozygous for the alleles b401, cn101, ky48, ky49, ky50, ky52, and y226 has already been reported (Rand and Russell 1984; Hosono et al. 1987; Clark et al. 1997; Zhao and Nonet 2000), and temperature-sensitive lethal alleles of cha-1 have also been described previously (Rand 1989). However, casual observation indicated that many other cha-1 mutants had more severe phenotypes at 25° than at 16°: they were smaller and more uncoordinated, and they grew more slowly at the higher temperature. Most ACh-releasing neurons in C. elegans are motor neurons (Duerr et al. 2008; Pereira et al. 2015); we therefore employed semi-quantitative assays to measure two types of neuromuscular function at different temperatures. One method measured the rate of pharyngeal pumping, and the other measured the rate of body flexion while the animals were swimming.

Our initial experiments involved animals grown at different temperatures and assayed at the growth temperature. A problem with this approach was that in order to evaluate animals at comparable developmental stages, the 25°-grown wild-type animals would be chronologically younger, and the 25°-grown cha-1 mutants would be considerably smaller, than the 14°-grown wild-type and mutant animals, respectively. However, we found that animals raised at 20° to young adulthood (i.e., 12–24 h after the start of egg-laying for wild-type animals), and then shifted for 2 h to either 14° or 25° had swimming and pharyngeal pumping rates that were essentially the same as animals grown and assayed at either 14° or 25°, respectively (pumping data shown as a control for a subsequent experiment).

We subsequently determined that within 30 min following the temperature shift, the animals’ behavior was essentially the same as that of animals raised at the final temperature. In fact, most of the time lag associated with the temperature shift appears to reflect the thermal equilibration time for the agar plate; in experiments where individual animals were transferred to pre-equilibrated plates, their behavior shifted far more rapidly (data not shown). Nevertheless, for consistency with our initial data sets, all of the behavioral assays presented here (except where noted) were performed with animals grown at 20°, and shifted to the assay temperature 2 h before measuring swimming rate or pharyngeal pumping.

We found that almost all of the cha-1 missense alleles were temperature-sensitive when homozygous; in at least one of the two assays their neuromuscular activity rate at 25° was lower than at 20°, and in many cases, lower than at 14° (Figure 4, A–F). This was in sharp contrast to wild-type animals, which swam and pumped approximately 50% and 85% faster, respectively, at the higher temperature (Figure 4, A–F). Because all the animals had been grown at 20°, except for the 2 h immediately prior to the behavioral assays, we concluded that the temperature-sensitive behaviors did not reflect altered neuronal or neuromuscular development at different temperatures, but rather temperature-sensitive neuronal function(s). Apparent exceptions to these patterns were animals homozygous for the cha-1 allele p503, and to a lesser extent p1154. The behavior of p503 homozygotes was close to that of wild-type animals in both assays at all three assay temperatures Figure 4, C and D). p503 is the mildest cha-1 mutant allele; homozygotes have approximately 5–10% the ChAT enzyme activity of wild-type animals and their behavior is superficially normal in most respects (Rand and Russell 1984). p1154 appears to be somewhere between a mild allele and a moderate allele; in the original report characterizing cha-1 mutants, the behavioral and developmental consequences of p1154 mutations were stated to be “less extreme than those of the other mutations” (Rand and Russell 1984).

Figure 4.

Behavioral analysis of cha-1 and other synaptic mutants. Synchronous cultures of all strains were grown at 20° and shifted to the desired assay temperature for 2 h before assay. Pharyngeal pumping (A, C, E, G) and swimming behavior (B, D, F, H) of cha-1 mutants demonstrate the behavioral effects of temperature. Pharyngeal pumping (I) and swimming behavior (J) of other synaptic mutants demonstrate increased rates at higher temperature. Pumping data represent the mean ± SD of 50 animals of each strain measured for 1 min each. Body bends were measured for 3 min for 10 animals of each strain, presented as the mean body bends/min ± SD. Statistical analysis utilized the Mann–Whitney U-test; for each behavioral test of each cha-1 strain (A–H), the 25° set of values was compared with either the 14° or the 20° data set (whichever had the higher mean value). For I and J, the 25° set of values was compared to the 14° data set. *P < 0.05; **P < 0.0005; ***P < 0.00005.

It is noteworthy that there was little or no correlation between the extent of behavioral deficit and the decrease in immunostaining for each mutant. For example, after 1 day at 23°, the staining in y226 and cn101 is similarly decreased in processes (Figure 3), but the behavior of y226 is significantly worse than that of cn101 (Figure 4, A–D).

Temperature-sensitive mutations are relatively uncommon (see Discussion), so we wondered if there was something unusual about this specific set of missense mutations, or if the temperature sensitivity was a property of this particular gene and protein. We therefore evaluated two Tc1 transposon insertion alleles, cha-1(md1422) and cha-1(md1067) (Figure 1). Tc1 transposons are frequently excised from mRNAs by standard splicing mechanisms; in such cases, however, the Tc1 element is usually excised imprecisely, leaving behind a small insertion, deletion, or substitution (Rushforth and Anderson 1996). Even when these small sequence changes preserve reading frame, protein function may be seriously compromised, leading to a strong mutant phenotype. This is the case with these two transposon insertion alleles; the temperature-sensitive behaviors of these Tc1 alleles (Figure 4, G and H), together with the (mostly) temperature-independent behaviors of p503 missense homozygotes (Figure 4, C and D) support our contention, discussed below, that the temperature-sensitivity of cha-1 mutants is not a function of missense mutations per se, but rather of strong hypomorphs.

Other synaptic transmission mutants are not temperature-sensitive

We also evaluated the effects of temperature on the phenotypes of four mutations affecting three synaptic genes: unc-17 (p279), unc-17(md1447), snt-1(md290), and unc-41(e1162). These mutants were chosen because their uncoordinated phenotypes are qualitatively similar to cha-1 and because the three genes have been cloned and their protein products are known to be required for proper release of ACh. The unc-17 gene encodes the synaptic vesicle acetylcholine transporter (Alfonso et al. 1993); the loading of ACh into synaptic vesicles is presumed to be the step which immediately follows synthesis of the neurotransmitter, and in most respects, unc-17 mutants have phenotypes quite similar to cha-1 mutants (Rand and Russell 1984). The snt-1 gene encodes the synaptic vesicle protein synaptotagmin, a protein necessary for the calcium-dependent release of all neurotransmitters (Nonet et al. 1993). The unc-41 gene encodes a homolog of stonin, a synaptic vesicle endocytosis adaptor protein required for proper synaptic vesicle cycling (Mullen et al. 2012).

Neuromuscular function of each of these mutants was considerably impaired compared to wild-type animals, yet swimming and pumping were noticeably better at 25° than at 14° (Figure 4, I and J). We conclude that the temperature sensitivity associated with cha-1 mutants is a specific attribute of cha-1, rather than a general phenotype associated with decreased cholinergic neuromuscular transmission.

Are the ChAT temperature effects reversible?

As stated above, 2 h after wild-type young adults grown at 20° were shifted to either 14° or 25°, their pumping rate was essentially the same as the rate of un-shifted animals grown at that temperature. Furthermore, we found that the effects of temperature shifts on pharyngeal pumping rate were reversible through several rounds of temperature shifts for wild-type C. elegans and for several cha-1 mutants we examined (Figure 5). This suggests that the decreased neuromuscular function of cha-1 mutants at elevated temperatures does not result from irreversible denaturation or degradation of the protein. For example, even though the md39 homozygotes were unable to move, feed, or grow after 1 h at 25°, 8 of the 10 animals we tested nevertheless recovered to their previous uncoordinated state when transferred back to 14°.

Figure 5.

Reversibility of temperature effects. (A) Synchronized cultures of wild type (N2) and the indicated cha-1 mutants were grown at 20° until they were young adults. They were then transferred to 25° for 1 h, followed by alternating incubations at 14° and 25° for 1 h each. Before each transfer, pharyngeal pumping rate was measured for 10 worms for 1 min each. Two different experimental protocols were employed: (1) we started with 100–200 synchronized young adult worms on a plate, and at each time point, we picked 10 worms to assay, and those worms were then discarded (solid lines); and (2) we picked 10 synchronized young adults to separate plates at the start of the experiment and followed the same 10 worms through each time point (dashed lines). Each data point represents the mean (±SD) of 10 worms assayed for 1 min each. Note: one of the 10 md39 homozygotes died during the first 25° incubation, and another one died during the second 25° incubation. (B) Controls. Synchronized cultures of wild-type animals were grown at either 14° or 25° until they were young adults, and pharyngeal pumping rate was measured. Each data point represents the mean (±SD) of 50 worms assayed for 1 min each.

Most cha-1 missense mutations conferring behavioral phenotypes affect phylogenetically conserved amino acids

We wondered if there was anything special about the specific sites of the ChAT missense mutations. We therefore compared the C. elegans ChAT amino acid sequence to the sequences from 56 other Caenorhabditis species, as well as sequences from 39 species representing 7 major nematode clades, 11 representative platyhelminth species, 9 insect species, 12 species from other invertebrate phyla, 3 lower chordates, and 15 vertebrate genomes (including humans), for a total of 145 species from 10 different phyla (details in Supplementary File S2).

Of the 12 missense mutation sites we have characterized in the C. elegans cha-1 gene, 11 affect amino acids that are conserved in all 56 Caenorhabditis species; 9 of the sites are conserved in almost all 39 nematode species; 9 are conserved among the 32 non-nematode invertebrate species, and 9 of the 12 are conserved among the 18 lower chordate and vertebrate species (Figure 6 and Supplementary File S2).

As a control, we examined a completely different set of 12 amino acids from C. elegans ChAT—we chose every 50th amino acid in the ChAT sequence and evaluated their conservation among the four sets of species already described (Figure 6 and Supplementary File S2). The results were quite different—only 5 of the 12 amino acids were conserved even within the genus Caenorhabditis; only 2 of the 12 were conserved among other nematodes, and none of them was conserved in any of the higher invertebrate and vertebrate phyla.

Discussion

Some cha-1 missense mutants have significantly decreased ChAT protein abundance, yet the level of residual protein is not correlated with the extent of behavioral impairment

This conclusion is based on the decreased synaptic ChAT immunostaining associated with many cha-1 missense mutants. For most mutant proteins with a reduced level of expression, it is often (correctly) assumed that the mutation perturbs protein folding and/or conformation, which leads to some type of instability and increased proteolytic turnover. However, this interpretation does not explain the relative stability of the mutant ChAT proteins once they have been trafficked to synapses—the temperature-dependent behavioral effects seem to be fully reversible (Figure 5), and therefore do not reflect denaturation and/or turnover of the protein at synapses. Although we do not understand the precise basis for the reduced levels of apparently stable mutant ChAT immunostaining, it seems plausible that some of the mutant proteins might require more time after synthesis to assume a stable conformation, and until they do, they would be subject to proteolytic attack and turnover.

Comparing the human and nematode ChAT proteins

Unfortunately, extrapolating from the features and regulation of human ChAT to the C. elegans enzyme has not been very informative. For example, of the seven amino acids in human ChAT that can be phosphorylated in vitro (Dobransky and Rylett 2005), only two have clear nematode counterparts (see Supplementary File S3). Similarly, although the C. elegans ChAT sequence has a number of basic amino acid clusters that might be involved in membrane interaction, these do not correspond to the location of the basic surface patches reported for the human enzyme (Cai et al. 2004). Furthermore, the little information available about possible thermal sensitivity associated with any of the human mutations is based only on in vitro assays of the human mutant proteins expressed in tissue culture. This is not at all surprising—temperature sensitivity is rarely considered in mammalian enzymology.

Most cha-1 hypomorphic mutants are temperature-sensitive

In general, temperature-sensitive mutants are quite rare—it is estimated that for most C. elegans genes, only about 4% of recovered hypomorphic mutants are temperature-sensitive (Hirsh and Vanderslice 1976). Although six cha-1 alleles (cn101, ky48, ky49, ky50, ky52, and y226) were identified in mutant screens designed to identify temperature-sensitive uncoordinated mutants, the rest were identified by their uncoordinated behavior at 20° or their resistance to cholinesterase inhibitors.

We also note that different cha-1 mutants displayed different types of temperature-sensitive behavior. Some cha-1 mutants, such as md39, display a dramatic change in behavior when the temperature is raised (Figure 4, A and B). Animals homozygous for other alleles, such as p1152 (Figure 4, E and F), have only a mild decrease in neuromuscular function (e.g., body flexions or pharyngeal pumping) when the temperature is raised. This is noteworthy because wild-type animals swim and pump at significantly higher rates at higher temperatures (Figure 4, A and B). Mutants in other synaptic transmission genes, such as unc-17(p279) and snt-1(md290) also have increased neuromuscular performance at higher temperatures (Figure 4, I and J). The snt-1 and unc-41 mutations used in Figure 4, I and J are presumed null alleles (Nonet et al. 1993; Mullen et al. 2012), and may therefore not be strictly comparable to hypomorphic cha-1 mutations; however, the unc-17 mutations we examined (p279 and md1447) are hypomorphic alleles (Alfonso et al. 1993; Zhu et al. 2001), and they, too, displayed somewhat improved neuromuscular function at elevated temperatures.

The temperature-sensitive phenotype of cha-1 mutants suggests a temperature-sensitive requirement for acetylcholine

Studies of temperature-sensitive mutants of numerous proteins from a large number of organisms have established some common themes: the mutations frequently affect hydrophobic amino acids that are “buried” within the protein structure. These amino acids are most often valine, leucine, isoleucine, methionine, phenylalanine, or tryptophan, and such amino acids are likely to be replaced with alanine, asparagine, aspartic acid, or proline (Poultney et al. 2011; Tan et al. 2014). Such patterns have led, with some success, to algorithm-based strategies designed to produce a temperature-sensitive variant of any given protein. However, it is noteworthy that an algorithm-derived list of 21 “desirable” mutagenic targets for C. elegans ChAT does not include any of the 12 amino acids that are altered in our set of ChAT missense mutations (Tan et al. 2014) (Supplementary File S3).

Traditionally, temperature-sensitive mutations have been interpreted as representing specific disruptions of stabilizing intramolecular interactions within a protein, so that at higher temperatures, the conformation of the mutant protein is less stable than the wild-type protein, and catalytic or binding activity is decreased. Thus, as indicated above, for most genes, temperature-sensitive alleles are relatively rare, and it is therefore striking that almost all cha-1 hypomorphic mutants should have some degree of temperature sensitivity. We have demonstrated that the temperature-sensitive behaviors associated with cha-1 mutants are not due to altered development of cholinergic neurons, nor are they a general feature of impaired release of ACh per se. We have also shown that the temperature-sensitive behaviors are not a specific feature of missense alleles; rather, they appear to be associated with functional deficits of strong cha-1 hypomorphic mutations. And finally, they do not follow the “standard” pattern of buried hydrophobic amino acid replacement. We are left with the conclusion that although they may lead to temperature-sensitive phenotypes, strong hypomorphic alleles of cha-1 are atypical temperature-sensitive mutations.

A thermal dissociation model of severe cha-1 phenotypes

We therefore propose that these mutants reveal an underlying temperature-sensitive step in cholinergic neuromuscular transmission. This step is presumably before the transport of ACh into synaptic vesicles and subsequent steps in ACh release, because deficiencies in vesicular acetylcholine transport (unc-17 mutants) and vesicular release (snt-1 mutants) do not lead to a comparable temperature-sensitive phenotype.

We suggest that a plausible mechanism might be based on an association of ChAT with synaptic vesicles. ChAT enzymatic activity recovered from tissue homogenates has been reported to be predominantly soluble in most organisms, including C. elegans, but membrane-bound and/or synaptic vesicle-associated forms are also present (Benishin and Carroll 1983; Rand and Russell 1985; Badamchian and Carroll 1985; Salem et al. 1993; Carroll 1994; Pahud et al. 1998). The membrane association of ChAT enzymatic activity from numerous sources is sensitive to high salt concentrations, indicating that the association is noncovalent (Fonnum 1968; Potter et al. 1968; Driskell et al. 1978; Carroll 1994). In addition, the crystal structure of human ChAT indicates a number of basic surface patches (not present in the closely related carnitine acetyltransferase), and it has been suggested that these might play a role in membrane association (Cai et al. 2004).

In contrast to data obtained with tissue homogenates, immunolocalization studies of ChAT in both C. elegans and Drosophila indicate predominantly synaptic, rather than cytoplasmic, localization (Pahud et al. 1998; Yasuyama et al. 2002; Duerr et al. 2008). Presumably, this discrepancy arises because there is usually salt present in most tissue extraction buffers (Fonnum 1968). In particular, the synaptic localization of ChAT in C. elegans as monitored by immunostaining requires the synaptic vesicle kinesin-like motor UNC-104 (Duerr et al. 2008), suggesting that proper trafficking and localization of ChAT depends critically on its association with synaptic vesicles (or with some other UNC-104-dependent vesicular cargo). These data complement long-standing studies in vertebrates indicating a functional coupling of ACh synthesis and vesicle loading (Collier 1969; Collier and Katz 1974; Carroll 1996). We also note that a functional as well as a physical association has been reported in vertebrates between the synthetic enzymes and the vesicular transporters for GABA and for dopamine (Chen et al. 2003; Jin et al. 2003; Buddhala et al. 2009; Cartier et al. 2010; Wang et al. 2011). Thus, association of neurotransmitter synthetic enzymes with synaptic vesicles and transporters may be a general feature of neurons in most organisms.

Since at least some of the ChAT protein in C. elegans (and probably in most nervous systems) appears to be physically, but not covalently, associated with synaptic vesicles, it seems plausible that the soluble and vesicle-associated forms of ChAT are in equilibrium, and presumably, such a physical association would be temperature-sensitive. Thus, at higher temperatures, a greater fraction of the ChAT protein would dissociate from vesicles, and therefore less ACh would be synthesized (and thus available) at the site of transport (Figure 7).

Figure 7.

A thermal dissociation model. According to this model, in wild-type C. elegans cholinergic neurons, some of the ChAT protein is physically, but not covalently, associated with synaptic vesicles. The vesicle-associated ChAT would therefore produce ACh in the immediate neighborhood of the vesicular ACh transporter (A); this would provide an efficient mechanism for rapid filling of synaptic vesicles. Since the ChAT-vesicle association is not covalent, it seems likely that the soluble and vesicle-associated forms of ChAT would be in equilibrium, and presumably, the extent of physical association would depend on the temperature. Thus, as ambient temperature is increased from 14° to 25°, less of the ChAT protein would remain associated with vesicles (B). However, under normal conditions, wild-type C. elegans have at least a 20-fold excess of ACh production capacity (Rand and Russell 1984), so that if even half of the vesicle-associated ChAT were to diffuse away from the vesicle, there would be almost no measurable decrease in the rate of vesicle-filling. However, severe hypomorphic (loss-of-function) cha-1 mutants have only a few percent of wild-type ACh production capacity remaining (C) (Rand and Russell 1984); this may be due to a decrease in ChAT protein at synapses, or to altered catalytic properties of the remaining enzyme (or both), but in any event, even at 14°, the ability to synthesize ACh and load it into vesicles is greatly reduced, which severely impairs overall cholinergic function. In such mutants, increasing the temperature would promote dissociation of some ChAT protein from synaptic vesicles (D), thus reducing even further the amount of ACh synthesized in the immediate vicinity of the transport sites. This, in turn, would decrease the rate and extent of vesicle-filling, thereby increasing the severity of the cha-1 behavioral deficits.

We have previously constructed heteroallelic mutant cha-1 animals with 3–5% of the wild-type ChAT activity, and we have shown that they are behaviorally wild type at 20° (Rand and Russell 1984); thus, in wild-type animals at 20°, there is a considerable excess of ChAT and also presumably, an excess of available ACh. Therefore, even if at high temperature, some of the wild-type ChAT were to dissociate from the immediate neighborhood of transport, there would still be enough locally synthesized ACh to saturate the transport sites. However, in cha-1 mutants, with the synthesis of ACh already compromised to the point where there are clear behavioral deficits, increasing the temperature would promote dissociation of ChAT protein from synaptic vesicles, thus removing the site of ACh synthesis (ChAT) from the site of vesicular ACh transport (VAChT; see Figure 7). This, in turn, would reduce even further local ACh available to synaptic vesicles, and increase the severity of the behavioral deficits.

More specifically, we would predict that each cha-1 hypomorphic mutant would be able to meet the demands of increased temperature until a mutation-specific “tipping point” temperature is reached. For each cha-1 mutation, the “tipping point” would depend on the abundance, specific activity, and altered substrate binding constants of the mutant ChAT enzyme as well as the kinetics of its dissociation from vesicles. Above that tipping temperature, the reduced amount of available local ACh would no longer saturate the VAChT/UNC-17 transport sites; further increase in temperature would not only increase the demand for ACh synthesis and release, but also decrease the locally available ACh supply, thus exacerbating the animals’ behavioral deficits. This is consistent with what we observe; for example, the rapid and complete reversibility of behavioral metrics following temperature shifts (Figure 5) is completely consistent with (and predicted by) a thermal dissociation model.

Another feature of the thermal dissociation model is that it provides a different set of molecular requirements to produce temperature-sensitive phenotypes: a cha-1 mutation does not need to cause temperature-sensitive catalysis or temperature-sensitive protein stability, nor does it need to disrupt any basic surface patches or putative phosphorylation sites of the protein. All that is necessary is that the mutant protein (1) has a very low specific activity; (2) be relatively stable once it is localized to synapses; and (3) its ability to associate and dissociate from vesicles be unimpaired. This is a less restrictive set of requirements and provides a plausible explanation for the relatively high incidence of temperature-sensitive cha-1 mutations.

It is also likely that a physical, noncovalent association between ChAT and synaptic vesicles could provide a sensitive site for regulation, and in vitro analysis of the phosphorylation of human ChAT reveals that phosphorylation at specific sites decreases the binding of the enzyme to membranes (Dobransky et al. 2001; Dobransky and Rylett 2005), which in turn would affect the amount of transmitter available for release. If this were also true in C. elegans, the interaction of ChAT with synaptic vesicles would provide both a temperature-independent mechanism (i.e., phosphorylation) as well as a temperature-dependent mechanism (discussed above) to regulate ACh availability in wild-type animals.

Funding

Our research was supported by research grants from the Oklahoma Center for the Advancement of Science and Technology (HN6-027 to J.S.D.) and the National Institute of General Medical Sciences (GM38679 to J.B.R.), and institutional support from Ohio University (to J.S.D.) and the Oklahoma Medical Research Foundation (to J.B.R.).

Conflicts of interest

The authors declare that there are no conflicts of interest.