Vitamin B12 produced by gut bacteria modulates cholinergic signaling

Abstract

A growing body of evidence indicates that gut microbiota influence brain function and behavior. However, the molecular basis of how gut bacteria modulate host nervous system function is largely unknown. Here we show that vitamin B12-producing bacteria that colonize the intestine can modulate excitatory cholinergic signaling and behavior in the host Caenorhabditis elegans. We find that vitamin B12 reduces cholinergic signaling in the nervous system through rewiring of the methionine (Met)/S-Adenosylmethionine (SAM) cycle in the intestine. We identify a conserved metabolic crosstalk between the Met/SAM cycle and the choline-oxidation pathway. We show that metabolic rewiring of these pathways by vitamin B12 reduces cholinergic signaling by limiting the availability of free choline required by neurons to synthesize acetylcholine. Our study reveals a gut-brain communication pathway by which enteric bacteria modulate host behavior and may affect neurological health.

There is growing interest in the relationship between gut microbiota, brain function and neurological disorders¹. Changes in gut microbiota have been linked to several neurological conditions including autism, anxiety, depression, schizophrenia, neurodegeneration, and migraine^2–4. However, it remains largely unclear whether imbalances in gut microbiota contribute to the neurological disorders or whether the disruption in gut microbial composition is a consequence of the disorder. While probiotics and psychobiotics are being explored to improve physiology and neural health, the cause-and-effect relationships are difficult to untangle⁵. Often, the effect of diet and microbiota on health and behavior are subtle and only become apparent when the host’s condition is challenged^6–10. Elucidating the molecular mechanisms underlying the effects of gut microbiota on host nervous system function poses several additional challenges. The human microbiome consists of trillions of microorganisms and hundreds different bacterial species¹¹ that could affect brain function of the host by modulating immune responses and through the production of highly diverse set of metabolites, and neurochemicals^12,13. Even when potential beneficial microbiota are identified, the complexity of the mammalian nervous system and the gut microbiome makes it exceedingly difficult to determine the effects of specific bacterial species on neural function^14,15.

The nematode C. elegans provides a powerful system to study the microbiota-gut-brain interactions because its nervous system and diet are relatively simple, well defined, and genetically tractable. In the wild, C. elegans feeds on diverse bacterial communities that can colonize its intestine^16–18. In the laboratory, C. elegans can be maintained on monoxenic bacterial culture, which simplifies mechanistic studies of microbial impacts on the host^19,20. Several studies have begun to investigate the effects of specific bacteria and their metabolites on C. elegans physiology, lifespan, and behavior^21–27. However, molecular mechanisms by which commensal gut bacteria modulate host neural function remain largely unknown.

In this study, we used unc-2/CaV2α(gof) mutants as a sensitized genetic background to facilitate the identification of bacteria that affect host neural function. unc-2(gof) mutants are hyperactive due to a gain-of-function (gof) mutation in the presynaptic voltage-gated calcium channel UNC-2/CaV2α²⁸. Similar mutations in human CACNA1A/CaV2.1α channel cause familial hemiplegic migraine type 1 (FHM1). The unc-2(gof) mutation, like CACNA1A FHM1 mutations, causes increased excitatory transmission and excitatory/inhibitory imbalance²⁸. Other neurological disorders with excitation/inhibition imbalance, such as autism, epilepsy, and migraine, have been associated with alterations in the gut microbiota^29–34. We find that B12-producing bacteria that colonize the intestine modulates excitatory signaling in C. elegans. We show that B12 reduces cholinergic signaling in the nervous system through metabolic crosstalk between B12-dependent Methionine/S-Adenosylmethinonine (Met/SAM) cycle and choline-oxidation pathway in the intestine. We find that B12 drives metabolic rewiring and reduces the availability of free choline which is taken up by neurons for the synthesis of acetylcholine. Our study reveals a molecular mechanism of gut-brain communication by which gut microbiota modulates host nervous system function and behavior and may explain the positive impact of B12 on neural disorders.

Results

Vitamin B12 producing bacteria modulate behavior

We surveyed bacterial diets for their ability to suppress the hyperactive behavior of unc-2(gof) mutants. unc-2(gof) mutants have an increased locomotion rate and display a striking clonic seizure-like behavior characterized by frequent jerky reversals²⁸. This behavioral hyperactivity can be easily quantified by the scoring reversal frequency with a multi-worm tracking system³⁵ and provides a simple and sensitive readout to study the effect of bacterial diets on host neural function. To rule out potential confounding effects of the bacterial diet on neuronal development, animals were grown on a standard laboratory Escherichia coli OP50 diet until the late 4th larval stage (L4). L4 animals were then transferred to plates with single bacterial diet. After 24 h, young adult unc-2(gof) animals were transferred to a thin lawn of OP50, to quantify reversals. We tested 40 bacterial strains of 20 different genera, most of which have been found to be associated with C. elegans in the wild^16–18. The selected strains are beneficial to C. elegans since they supported normal animal growth and did not induce the expression of the bacterial infection response gene, irg-1::GFP³⁶ (Extended Data Fig. 1a,b).

We found that 18 of the 40 bacterial diets significantly suppressed the hyper-reversals of unc-2(gof) mutants compared to OP50 (Fig. 1a). Unlike OP50, some of the bacterial species that suppressed hyper-reversals, such as Comamonas aquatica and Pseudomonas putida, are known to produce B12^19,37. Since B12 plays a crucial role in health and neural function³⁸, we screened all bacteria for B12 production using a Pacdh-1::GFP reporter strain. acdh-1::GFP expression is strongly repressed by dietary B12^19,39. All bacterial strains that suppressed Pacdh-1::GFP expression also suppressed unc-2(gof) reversals (Fig. 1a,b and Extended Data Fig. 1c). This suggests that B12 may be a key bacterial metabolite that suppresses the hyper-reversals of unc-2(gof) mutants. This is further supported by the observation that a Comamonas cbiAΔ strain, which cannot produce B12¹⁹, failed to reduce reversals of unc-2(gof) mutants (Fig. 1c and Extended Data Fig. 1d).

Fig. 1: B12 produced by gut bacteria suppresses hyperactive behavior of unc-2(gof) mutants.

a, Reversal frequency of unc-2(gof) mutants fed the indicated bacterial strains for 24 h (mean ± s.e.m., one-way ANOVA with Dunnett’s multiple comparison). b, GFP expression of Pacdh-1::GFP animals fed the indicated bacterial strains for 24 h. Shades of green represent relative GFP expression levels, “High” indicates strong fluorescence throughout the intestine as in the OP50 shown, “Low” indicates barely detectable fluorescence as in the Comamonas shown, “Moderate” indicates visible fluorescence but weaker compared to the GFP signal on OP50 (n = 3 biologically independent samples with similar results, see Extended Data Fig. 1c). c, Reversal frequency of wild-type and unc-2(gof) mutants fed OP50, Comamonas, Comamonas cbiAΔ (B12-), or OP50 supplemented with 64 nM B12 (mean ± s.e.m., two-way ANOVA with Sidak’s multiple comparison). d, Reversal frequency of unc-2(gof) mutants fed OP50 supplemented with the indicated concentrations (nM) of B12 (mean ± s.e.m., one-way ANOVA with Dunnett’s multiple comparison). e, Reversal frequency of unc-2(gof) mutants fed OP50 supplemented with 64 nM B12 for the indicated time (2G, 2 generations) and after transfer to OP50 (mean ± s.e.m., one-way ANOVA with Dunnett’s multiple comparison). f, Reversal frequency of unc-2(gof) mutants fed live and heat-killed OP50 ± 64 nM B12 (mean ± s.e.m., two-way ANOVA with Sidak’s multiple comparison). Source numerical data are available in source data.

Dietary vitamin B12 is sufficient to modulate behavior

B12 is synthesized exclusively by bacteria and archaea⁴⁰. Animals acquire B12 through symbiotic relationship with bacteria in their gut or through their diet^41,42. E. coli can import B12 from regular media, but the concentrations are much lower than those found in B12-producing Comamonas^19,43. Is dietary B12 is sufficient to change behavior? We found that supplementation of B12 (64 nM) to animals fed on either OP50 or Comamonas cbiAΔ suppressed reversals of unc-2(gof) mutants to a similar extent as Comamonas (Fig. 1c,d and Extended Data Fig. 1d,e). The suppression effect of B12 supplementation on reversals took effect within 16 h but disappeared a day after transfer to OP50 without B12 (Fig. 1e). B12 reduced reversals of unc-2(gof) mutants equally well when supplemented to live or dead OP50 (Fig. 1f and Extended Data Fig. 1f). This indicates that the B12 effect was not mediated by secondary alterations in bacterial metabolism and is sufficient to change host behavior.

Gut colonization by Comamonas modulates C. elegans behavior

To determine whether Comamonas establishes a gut microbiome, we examined the ability of Comamonas to colonize the intestine of worms. L4 worms were transferred from OP50 to different bacterial diets for 24 h after which animals were homogenized and the number of colony-forming units (CFUs) per worm was calculated^16,18. Comamonas efficiently colonized the intestine of worms with a density of ~ 2 x 10³ CFUs per animal, compared to ~ 80 CFUs for OP50 and ~ 2.7 x 10⁴ CFUs for the pathogenic bacteria Pseudomonas aeruginosa (Fig. 2a). Comamonas cbiAΔ colonization efficiency was similar to wild-type Comamonas suggesting that B12 production does not affect colonization efficiency (Fig. 2b). Comamonas colonization was stable for at least 3 days after the transfer to OP50 (Fig. 2c). Consistently, we found that unc-2(gof) mutants grown on live Comamonas still showed reduced reversals three days after a shift to OP50 (Fig. 2d). In contrast, sustained reversal suppression was not observed when unc-2(gof) mutants were fed killed Comamonas (Fig. 2d) or B12-supplemented OP50 (Fig. 1e). Thus, persistent changes in the behavior correlate with the presence of Comamonas in the worm intestine. To further test whether gut colonization by Comamonas can drive reversal suppression, we treated worms raised on live Comamonas with antibiotics. Kanamycin treatment effectively eliminated Comamonas from the worm gut (Fig. 2e and Extended Data Fig. 2a,b). The reversal suppression of unc-2(gof) mutants colonized with Comamonas was no longer observed after clearance of gut bacteria with kanamycin treatment (Fig. 2f, Extended Data Fig. 2c). Together, these results indicate that Comamonas colonization of the intestine can lead to persistent changes in behavior.

Fig. 2: Comamonas colonizes the C. elegans intestine and modulates behavior.

a, Bacterial colony forming units (CFU) per animal from unc-2(gof) mutants fed OP50, P. aeruginosa PA14, or Comamonas (mean ± s.e.m., two-tailed unpaired t-test). b, Bacterial colony forming units (CFU) per animal from unc-2(gof) mutants fed Comamonas and Comamonas cbiAΔ (mean ± s.e.m., one-way ANOVA with Dunnett’s multiple comparison). c, Bacterial CFU per animal from unc-2(gof) mutants fed Comamonas after transfer to OP50, measured over multiple days, as indicated (mean ± s.e.m., two-tailed unpaired t-test). d, Reversal frequency of unc-2(gof) mutants fed OP50 with 64 nM B12, Comamonas, or dead Comamonas after transfer to OP50, measured over multiple days, as indicated (mean ± s.e.m., one-way ANOVA with Dunnett’s multiple comparison). e, Antibiotics susceptibility of E. coli and C. aquatica treated for 24 h with the indicated antibiotics (1x concentrations: 50 μg/ml Ampicillin, 100 μg/ml Carbenicillin, 50 μg/ml Kanamycin, 50 μg/ml Chloramphenicol, 100 μg/ml Streptomycin, and 12.5 μg/ml Tetracycline). f, Bacterial CFU per animal from unc-2(gof) mutants fed Comamonas treated with 200 μg/ml kanamycin for the indicated time (mean ± s.e.m., one-way ANOVA with Dunnett’s multiple comparison). g, Reversal frequency of unc-2(gof) mutants colonized with Comamonas after kanamycin treatment for 24 h (mean ± s.e.m., two-way ANOVA with Sidak’s multiple comparison). Source numerical data are available in source data.

Vitamin B12 inhibits excitatory cholinergic signaling

How does B12 impact neural function and behavior? The increased reversals and hyperactivity of unc-2(gof) mutants is due, at least in part, to increased excitatory cholinergic neurotransmission²⁸. Consequently, unc-2(gof) mutants are hypersensitive to the acetylcholine esterase inhibitor, aldicarb. In the presence of aldicarb, acetylcholine accumulates at the neuromuscular junction (NMJ), which leads to muscle hyper-contraction and eventual paralysis⁴⁴. Wild-type animals completely paralyzed on 1 mM aldicarb plates within 100 min. In contrast, unc-2(gof) mutants that were raised on OP50 completely paralyzed within 40 min. B12-supplemented OP50 significantly slowed aldicarb induced paralysis of unc-2(gof) animals (Fig. 3a), suggesting that B12 inhibits cholinergic signaling in unc-2(gof) mutants. Since B12 has been implicated in monoamine synthesis⁴⁵, we also analyzed the impact of B12 on cat-1 unc-2(gof) double mutants. cat-1 encodes the C. elegans vesicular monoamine transporter required for octopamine, tyramine, dopamine and serotonin transport into vesicles⁴⁶. B12 supplementation still significantly reduced reversals in cat-1 unc-2(gof) mutants, suggesting that B12 effect on behavior is not mediated through changes in monoamine signaling (Extended Data Fig. 3a).

Fig. 3: B12 inhibits cholinergic signaling.

a, Percentage of paralyzed wild-type and unc-2(gof) mutants fed OP50 ± 64 nM B12 on 1 mM aldicarb (mean ± s.e.m., n = 4, two-way ANOVA with Tukey’s multiple comparison). b,c, 10 min tracks (b) and quantification of path length (c) of single young adult animal of wild-type and ace-1(p1000);ace-2(g72) double mutants fed OP50, OP50 with 64 nM B12, Comamonas, or Comamonas cbiAΔ for 24 h (mean ± s.e.m., n = 4, two-way ANOVA with Tukey’s multiple comparison). Scale bar, 5 mm. d, Quantification of convulsion phenotype of acr-2(n2420gf) mutants fed OP50, OP50 with 64 nM B12, Comamonas, or Comamonas cbiAΔ for 24 h (n = 5, one-way ANOVA with Dunnett’s multiple comparison). In violin plots, middle-dotted line shows median, upper and lower lines represent 1st and 3rd quartiles. e, Calcium transient in muscle of a freely moving animal expressing a Pmyo-3::GCaMP6 transgene. Pixel intensity (arbitrary units) and corresponding colormap are depicted in the color bar. f, Quantification of the mean GCaMP6 fluorescence in body wall muscles of wild-type, unc-2(gof), and acr-2(n2420gf) mutants fed OP50 ± 64 nM B12 (mean ± s.e.m., n = 4, two-way ANOVA with Tukey’s multiple comparison). g, Percentage of paralyzed wild-type animals fed OP50 ± 64 nM B12 on 0.01 mM aldicarb during swimming (mean ± s.e.m., n = 3, two-way ANOVA with Tukey’s multiple comparison). h, B12 reduces swimming-induced quiescence. Fraction of wild-type animals fed OP50 ± 64 nM B12 in quiescence (mean ± s.e.m., n = 3, two-way ANOVA with Tukey’s multiple comparison). Source numerical data are available in source data.

We further analyzed the effect of B12 on mutants with specific increases in cholinergic signaling. Mutants lacking the acetylcholinesterase ace-1 and ace-2 genes are hypercontracted and severely uncoordinated due to ACh build-up at NMJ⁴⁷. A B12-supplemented OP50 or Comamonas (C. aquatica) dramatically improved the severe locomotion defects of the ace-1;ace-2 double mutants (Fig. 3b,c and Extended Data Fig. 3b). Next, we analyzed the impact of B12 on mutants with a gain-of-function mutation in the ACR-2 acetylcholine receptor subunit (acr-2(gof)). ACR-2 is expressed in the cholinergic motor neurons in the ventral cord and regulates their excitability⁴⁸. acr-2(gof)) mutants display convulsive muscle contractions, as the result of increased excitatory cholinergic- and reduced inhibitory GABAergic- output onto body wall muscles⁴⁸. The imbalance of excitation/inhibition in acr-2(gof) mutants resembles that reported for gain-of-function mutations of nicotinic acetylcholine receptors (nAChRs) in frontal lobe epilepsy⁴⁹. B12-supplemented OP50 or Comamonas significantly reduced convulsive muscle contractions of acr-2(gof)) mutants, whereas a mutant Comamonas cbiAΔ failed to do so (Fig. 3d).

To further analyze the effect of B12 on cholinergic transmission, we performed Ca²⁺ imaging experiments using transgenic worms expressing a fluorescent Ca²⁺ indicator GCaMP6s in body wall muscles (Pmyo-3::GCaMP6)⁵⁰. Cholinergic motor neurons in the ventral cord synapse onto body wall muscles and calcium transients correlate with activity of cholinergic transmission at NMJ^51,52. In freely moving animals fed on OP50, the GCaMP6s fluorescence was higher in unc-2(gof) mutants compared to wild-type, consistent with an increase in cholinergic signaling (Fig. 3e,f). Similarly, increased excitability of cholinergic motor neurons in acr-2(gof) mutants induced increased calcium transients in body wall muscles (Fig. 3e,f). B12 supplementation had no obvious effects in wild-type animals, but significantly reduced GCaMP6 fluorescence in unc-2(gof) and acr-2(gof) mutants (Fig. 3e,f). Together, these data indicate that B12 reduces excitatory cholinergic signaling in mutants with E/I imbalance.

Vitamin B12 reduces swimming induced quiescence

In contrast to the cholinergic signaling mutants, B12 had no obvious effects on locomotion or aldicarb sensitivity of wild-type animals under standard conditions (Fig. 3a and Extended Data Fig. 3c–f). Since the effects of gut microbiota and individual nutrients can be revealed by environmental and genetic conditions of the host, we hypothesized that the effect of B12 in wild-type animals may become more apparent under conditions of increased cholinergic signaling. C. elegans movement is controlled by cholinergic transmission at NMJ that propels undulating body-wall muscle contractions along the anterior-posterior axis. When C. elegans transitions from a solid surface to liquid, it dramatically increases its undulation frequency as it switches from crawling to swimming⁵³. When exposed to 1 mM aldicarb in liquid, swimming animals became paralyzed within 20 min; compared to 100 min when crawling on agar plates (Extended Data Fig. 3g). The enhanced aldicarb sensitivity of wild-type animals in liquid is consistent with increased cholinergic signaling during swimming⁵⁴. We observed that B12 supplementation significantly suppressed the aldicarb sensitivity of swimming wild-type animals (Fig. 3g). High levels of cholinergic signaling has also been shown to induce bouts of quiescent behavior characterized by transient paralysis during prolonged swimming in liquid⁵⁴. We found that B12 supplementation also strongly suppressed the swimming-induced quiescence in wild-type animals (Fig. 3h). Together, these results suggest that B12 reduces cholinergic signaling in wild-type animals under conditions that increase acetylcholine release.

Vitamin B12 modulates behavior through the Met/SAM cycle

B12 is an essential cofactor for two metabolic enzymes in both C. elegans and mammals¹⁹: methionine synthase (metr-1) which converts homocysteine to methionine in the Met/SAM cycle and methylmalonyl-CoA mutase (mmcm-1) which is required for propionyl-CoA breakdown (Fig. 4a). We generated metr-1;unc-2(gof) and mmcm-1;unc-2(gof) double mutants to determine whether either of these B12-dependent pathways modulates cholinergic signaling. The suppression effect of B12 on reversals of unc-2(gof) animals was completely abolished by a metr-1, but not by mmcm-1 mutation (Fig. 4b). Notably, the metr-1 mutation further increased reversals of unc-2(gof) mutants, whereas the mmcm-1 mutation had no such effect (Fig. 4b). B12 also failed to suppress reversals by mutations in sams-1, the predicted C. elegans SAM synthetase in Met/SAM cycle, but not by mutations in pcca-1 or mce-1 in propionyl-CoA breakdown pathway (Extended Data Fig. 4a,b). Furthermore, the effect of B12 on aldicarb sensitivity was completely abolished in metr-1;unc-2(gof) (Fig. 4c) and sams-1;unc-2(gof) mutants (Extended Data Fig. 4c). metr-1 single mutants also displayed a marked increase in swimming-induced quiescence (Fig. 4d). In contrast to wild-type, B12 supplementation failed to suppress quiescence in metr-1 mutants (Fig. 4d). This indicates that the effect of B12 on behavior depends on metr-1 and Met/SAM cycle. Since B12 affected cholinergic signaling, we analyzed the effect of B12 supplementation on acetylcholine levels in worm extracts. We found that unc-2(gof) mutants had a marked reduction in acetylcholine levels in both HPLC-MS analyses (37%) and fluorometric assays (28%) when OP50 was supplemented with B12 (Fig. 4e and Extended Data Fig. 4d,e). In contrast, B12 supplementation did not affect acetylcholine levels in metr-1;unc-2(gof) mutants (Fig. 4e). As expected, B12 supplementation drastically increased the levels of methionine in both wild-type⁵⁵ and unc-2(gof) mutants (Extended Data Fig. 4f). Methionine levels were unaffected by B12 supplementation in metr-1 or metr-1;unc-2(gof) mutants (Extended Data Fig. 4f), consistent with METR-1’s crucial role in B12-dependent methionine synthesis in C. elegans¹⁹. Notably, the impact of B12 on methionine synthesis was much more pronounced in unc-2(gof) mutants (Extended Data Fig. 4f). Taken together, these data suggest that B12 reduces cholinergic signaling through cross-talk with methionine synthesis in the Met/SAM cycle.

Fig. 4: B12 dependent Met/SAM cycle acts in intestine and hypodermis to modulate behavior.

a, B12-dependent metabolic pathways, Met/SAM cycle and propionyl-CoA breakdown pathway are highly conserved in C. elegans. b, Reversal frequency of metr-1(ok521);unc-2(gof) or mmcm-1(ok1637);unc-2(gof) mutants fed OP50 ± 64 nM B12 (mean ± s.e.m., two-way ANOVA with Tukey’s multiple comparison). c, Percentage of paralyzed metr-1(ok521);unc-2(gof) mutants fed OP50 ± 64 nM B12 on 1 mM aldicarb (n = 3 (OP50) and n = 4 (OP50 +B12), two-way ANOVA with two-way ANOVA with Tukey’s multiple comparison). d, Percentage of quiescence of wild-type and metr-1(ok521) mutants fed OP50 ± 64 nM B12 (n = 3, two-way ANOVA with two-way ANOVA with Tukey’s multiple comparison). e, Quantification of acetylcholine in unc-2(gof) and metr-1(ok521);unc-2(gof) mutants fed OP50 ± 64 nM B12 for 24 h (mean ± s.e.m., two-way ANOVA with Tukey’s multiple comparison). f, Expression pattern of Pmetr-1::GFP. Upper panel: confocal slice showing strong expression in intestine and hypodermis (arrow). Lower panels: Z-projection of the head (left) and the tail (right) showing no detectable signal in neurons (n = 3 biologically independent samples with similar results). Scale bars, 25 μm. g, Reversal frequency of metr-1(ok521);unc-2(gof) mutants expressing metr-1 cDNA driven by Pmetr-1 (endogenous), Pelt-2 (intestinal), Pdpy-7 (hypodermal), Pmyo-3 (muscle), or Ptag-168 (pan-neuronal) promoter fed OP50 ± 64 nM B12 (mean ± s.e.m., two-way ANOVA with Tukey’s multiple comparison). Source numerical data are available in source data.

To determine the site of action for the Met/SAM cycle, we analyzed the metr-1 expression pattern using a Pmetr-1::GFP transcriptional reporter. The Pmetr-1::GFP fluorescence was observed in the intestine and hypodermis, but not in neurons (Fig. 4f). Expression of metr-1 in intestine (Pelt-2::METR-1) or hypodermis (Pdpy-7 ::METR-1), but not in muscle (Pmyo-3::METR-1) or neurons (Ptag-168 ::METR-1), was sufficient to restore the B12-dependent suppression of reversals in metr-1;unc-2(gof) mutants (Fig. 4g and Extended Data Fig. 4g). This indicates that B12 modulates cholinergic signaling cell non-autonomously through its role as a co-factor METR-1 and the Met/SAM cycle in intestine and hypodermis.

Next, we asked whether B12 suppressed reversals of unc-2(gof) mutants as the result of accumulation or depletion of metabolites from the Met/SAM cycle. METR-1 catalyzes the conversion of homocysteine to methionine, which can be further converted to SAM by SAMS-1^56,57 (Fig. 4a). Supplementation of methionine or SAM increased animal growth consistent with previously reports¹⁹, but did not suppress reversals of unc-2(gof) mutants (Extended Data Fig. 4h,i). Increased levels of homocysteine are associated with the symptoms of a variety of neurological disorders, including depression, autism and migraine^58–60. This finding has led to the “homocysteine hypothesis” which proposes that B12 deficiency and increased homocysteine levels underlie the etiology of these disorders⁴⁵. However, HPLC-MS analyses showed that B12 supplementation had no significant effect on homocysteine and S-adenosylhomocysteine levels in both wild-type and unc-2(gof) mutants (Extended Data Fig. 4j,k). Homocysteine supplementation also did not impact on reversals of unc-2(gof) mutants (Extended Data Fig. 4g) and failed to block the effect of B12 on animal growth and reversals (Extended Data Fig. 4h,i). Furthermore, mutations in cystathionine β-synthase gene, cbs-2, which encodes a homolog of human enzyme that converts homocysteine to cystathionine⁶¹, did not lead to an increased reversals in unc-2(gof) mutants (Fig. 4a and Extended Data Fig. 4a). In addition, B12 suppressed unc-2(gof) reversals in the cbs-2 mutant background (Extended Data Fig. 4a). Together, these results suggest that the effect of B12 on behavior is independent of its role in animal growth and is mediated indirectly by increased metabolic flux through the Met/SAM cycle.

Metabolic crosstalk: the Met/SAM cycle and choline metabolism

How does the B12-dependent Met/SAM cycle in intestine and hypodermis modulate cholinergic signaling in the nervous systems? In mammals, the Met/SAM cycle is tightly associated with choline-oxidation pathway^62–64. The choline-oxidation pathway converts choline into betaine, which in turn can serve as a methyl donor for the synthesis of methionine in Met/SAM cycle (Fig. 5a)^65,66. Choline is also the rate-limiting precursor of acetylcholine biosynthesis in the nervous system^67,68. To examine metabolic crosstalk between Met/SAM cycle and cholinergic signaling, we tested the effect of exogenous choline on reversals of unc-2(gof) mutants. Dietary choline supplementation completely abolished the effect of B12 on reversal suppression (Fig. 5b) but had no effect on reversals in the absence of B12 (Fig. 5b). Furthermore, in wild-type animals, B12 failed to suppress the swimming-induced quiescence in the presence of exogenous choline (Fig. 5c). Choline supplementation did not affect B12’s positive impact on animal growth (Extended Data Fig. 5a), consistent with our finding that B12 suppresses reversals independently of its role in animal growth. To determine if B12 affects choline levels, we measured the amount of free choline using an enzyme-based fluorometric assay. B12 supplementation led to a 25% decrease of free choline levels in unc-2(gof) mutants (Fig. 5d).

Fig. 5: B12 modulates excitatory cholinergic signaling through metabolic crosstalk between the Met/SAM cycle and the choline-oxidation pathway.

a, Metabolic network of Met/SAM cycle and choline metabolism. Red and blue indicate C. elegans and human metabolic enzymes involved, respectively. b, Reversal frequency of unc-2(gof) mutants fed OP50 ± 64 nM B12 with indicated concentrations of choline (mean ± s.e.m., one-way ANOVA with Tukey’s multiple comparison). c, Percentage of quiescence of wild-type animals fed OP50, OP50 with choline (30 mM), OP50 with B12 (64 nM), or OP50 with both B12 and choline (mean ± s.e.m., n = 3, two-way ANOVA with Tukey’s multiple comparison). d, Quantification of free choline of unc-2(gof) mutants fed OP50 ± 64 nM B12 for 24 h (mean ± s.e.m., two-tailed unpaired t-test). e, Co-expression pattern of Pmetr-1::GFP and Pchdh-1::mCherry in the intestine and hypodermis. chdh-1 is expressed in the intestine, hypodermis, and RIM neurons (n = 3 biologically independent samples with similar results). Scale bar is 100 μm. f, Reversal frequency of unc-2(gof) mutants subjected to RNAi knockdown of chdh-1 or alh-9 fed OP50 ± 64 nM B12 (mean ± s.e.m., two-way ANOVA with Tukey’s multiple comparison). Source numerical data are available in source data.

In mammals, choline is irreversibly oxidized to betaine in the liver through two sequential reactions catalyzed by choline dehydrogenase and betaine-aldehyde dehydrogenase⁶⁵ (Fig. 5a). The C. elegans genome contains orthologs for both choline dehydrogenase (chdh-1) and betaine-aldehyde dehydrogenase (alh-9) genes (Fig. 5a). Like metr-1, transcriptional reporters for both chdh-1 and alh-9 are expressed in the intestine and hypodermis (Fig. 5e and Extended Data Fig. 5b). We found that RNAi knockdown of either chdh-1 or alh-9 completely abolished the suppression effect of B12 on reversals of unc-2(gof) mutants (Fig. 5f). Notably, RNAi knockdown of chdh-1 further increased reversals (Fig. 5f), similar to metr-1 mutation (Fig. 4b). chdh-1 and alh-9 RNAi did not diminish the effect of B12 on animal growth of unc-2(gof) mutants (Extended Data Fig. 5c), providing further evidence that the effect of B12 on behavior is independent of its role in animal growth. These results suggest that B12 modulates behavior through metabolic crosstalk between the Met/SAM cycle and choline-oxidation pathway (Fig. 5a).

In vertebrates, betaine, the product of choline-oxidation pathway, and 5-meTHF can act as methyl donors for the synthesis of methionine from homocysteine through the activity of two closely related enzymes: betaine-homocysteine methyltransferase (BHMT) and methionine synthase (MS), respectively⁶⁹ (Fig. 5a). C. elegans, like other invertebrates, however, lacks a gene encoding BHMT⁷⁰. Phylogenetic analysis suggests that BHMT and methionine synthase have evolved from a common ancestor enzyme (Fig. 6a,b). We further characterize METR-1 methionine synthase activity, we purified the METR-1::GFP protein from worm lysates (Extended Data Fig. 6a). METR-1 converted homocysteine into methionine in the presence of either 5-meTHF or betaine as a methyl donor in in vitro assays (Fig. 6c). This suggests that invertebrate methionine synthases, like METR-1, can use both betaine and 5-meTHF as methyl donors. To further test this hypothesis, we analyzed the in vivo interaction between betaine and Met/SAM cycle. Betaine supplementation, like methionine or SAM supplementation (Extended Data Fig. 4h), increased animal growth in a dose-dependent manner (Fig. 6d). However, betaine supplementation failed to increase animal growth of metr-1 mutants (Fig. 6e). In mammals, betaine and 5-meTHF can complement each other to maintain flux through the Met/SAM cycle^65,66. Consequently, in mice, betaine supplementation can rescue the early postnatal mortality caused by 5-meTHF deficiency of methylenetetrahydrofolate reductase (Mthfr) knockout mice⁷¹. Interestingly, polymorphisms in the MTHFR gene in humans increases susceptibility to migraines^72,73. In C. elegans, deletion of mthf-1/MTHFR results in 5-meTHF deficiency and developmental arrest at early larval stages⁷⁴. We found that betaine supplementation significantly rescued the larval arrest of mthf-1 mutants: Only ~10% of mthf-1 mutants reached adulthood on OP50, whereas ~ 40% of mthf-1 homozygous mutants reached adulthood when betaine was supplemented (Fig. 6f). B12 plus betaine further rescued the development of mthf-1 mutants (Fig. 6g). Together, these results support a conserved metabolic link between betaine, the product of the choline-oxidation pathway, and the B12-dependent Met/SAM cycle.

Fig. 6: Betaine can act as a methyl donor in the Met/SAM cycle.

a, Schematic representation of domain architectures in methyltransferase proteins. HMT, homocysteine methyltransferases; BHMT, betaine-homocysteine methyltransferase; MS, Methionine synthase. b, Phylogenetic analysis of the methyltransferase enzymes. The tree was constructed with amino acid sequences of the homocysteine binding domains methyltransferase enzymes from bacteria, yeast, fly, plant and mammals (Methods). Black, blue, and red lines indicate HMT, BHMT, and MS enzymes, respectively. Maximum likelihood bootstrap values are shown at the relevant branches. c, in vitro enzyme activity of the purified METR-1 measured by the production of methionine from homocysteine in the presence of 5-methyltetrahydrofolate (5-meTHF) or betaine as a methyl donor (mean ± s.e.m., n = 3, two-way ANOVA with Tukey’s multiple comparison) (Methods). d, Growth rate as indicated by body length of wild-type fed OP50with the indicated concentrations of betaine (mean ± s.e.m., n = 3, one-way ANOVA with Tukey’s multiple comparison). e, Growth rate as indicated by body length of wild-type and metr-1(ok521) mutants fed OP50 with the indicated concentrations of betaine (mean ± s.e.m., n = 3, two-way ANOVA with Tukey’s multiple comparison). f, Percentage of mthf-1(gk465);bli-2(e768) mutants that reached adulthood on OP50 ± 75 mM betaine (mean ± s.e.m., two-tailed unpaired t-test). g, Percentage of mthf-1(gk465);bli-2(e768) mutants that reached adulthood on OP50, OP50 with B12 (64 nM), OP50 with betaine (75 mM), or OP50 with both B12 and betaine (mean ± s.e.m., one-way ANOVA with Tukey’s multiple comparison). Source numerical data are available in source data.

Recent studies in C. elegans have also identified a role for betaine in modulating behavior through the activation of betaine and choline-gated ion channels, ACR-23⁷⁵, LGC-41⁷⁶, and DEG-3⁷⁷. To test whether the B12 effect on behavior is mediated through neuronal betaine or choline signaling, we generated acr-23;unc-2(gof), lgc-41 unc-2(gof), and deg-3;unc-2(gof) double mutants. While an acr-23 mutation decreased- and an lgc-41 mutation increased-reversals of unc-2(gof) mutants, B12 significantly suppressed reversals in the absence of these betaine receptors (Extended Data Fig. 7a,b). Furthermore, unlike choline supplementation (Fig. 5b), betaine supplementation was unable to rescue the suppression effect of B12 (Extended Data Fig. 7c). This indicates that the main effect of B12 on behavior is not mediated by neural betaine signaling.

Gut-brain communication: choline and ACh synthesis

Crosstalk between the choline-oxidation pathway and Met/SAM cycle can impact levels of free choline^65,78. Choline supply for acetylcholine biosynthesis in cholinergic neurons occurs in part through the uptake of free choline by high-affinity choline transporters. Choline uptake is modulated to meet demands of increased ACh synthesis and release⁷⁹. Choline transporter (CHT) mouse knockouts have severe defects in cholinergic transmission resulting in early neonatal lethality indicating that choline uptake through CHT is a major source of choline for ACh biosynthesis. C. elegans has a single choline transporter cho-1, which is exclusively expressed in cholinergic neurons⁸⁰. cho-1 null mutants are viable, display normal locomotion on agar plates but fail to sustain swimming behavior in liquid⁸¹. Loss of cho-1 function reduced reversals and aldicarb sensitivity of unc-2(gof) mutants (Fig. 7a,b). This result suggests that free choline and choline transport is required to maintain elevated levels of cholinergic signaling in unc-2(gof) mutants. Therefore, we next tested whether choline uptake by CHO-1 in cholinergic neurons is required for the B12 effect on behavior. B12 supplementation did not further suppress reversals and aldicarb sensitivity of cho-1;unc-2(gof) mutants (Fig. 7a,b) but still stimulated animal growth (Extended Data Fig. 7d). Furthermore, we found that while choline supplementation fully rescued the reduced reversals of unc-2(gof) mutants induced by B12 (Fig. 5b), it had no such effect in the absence of cho-1 (Fig. 7c). These results support a gut-brain communication pathway in which B12-dependent metabolism in the intestine reduces excitatory cholinergic signaling in the nervous system by decreasing the availability of free choline required for the synthesis of acetylcholine (Fig. 7d).

Fig. 7: A neuronal choline transporter is required to mediate the effect of B12 on excitatory signaling.

a, Reversal frequency of unc-2(gof) and cho-1(tm373);unc-2(gof) mutants fed OP50 ± 64 nM B12 (mean ± s.e.m., one-way ANOVA with Tukey’s multiple comparison). b, Quantification of paralysis percentage of unc-2(gof) and cho-1(tm373);unc-2(gof) mutants fed OP50 ± 64 nM B12 on 1 mM aldicarb (mean ± s.e.m., n = 3, one-way ANOVA with Tukey’s multiple comparison). c, Reversal frequency of unc-2(gof) and cho-1(tm373);unc-2(gof) mutants fed OP50, OP50 with B12 (64 nM), or OP50 with both B12 (64 nM) and choline (30 mM) (mean ± s.e.m., one-way ANOVA with Tukey’s multiple comparison). d, Model: B12 produced by gut bacteria modulates excitatory neurotransmission of the host C. elegans. Metabolic crosstalk between the B12 dependent Met/SAM cycle and the choline-oxidation pathway in the intestine and hypodermis lowers the levels of free choline. The reduced choline availability by B12 limits acetylcholine synthesis in the neurons particularly under conditions of elevated acetylcholine release. Source numerical data are available in source data.

Discussion

Diet and gut microbiota play a critical role in neuronal health and disease. Elucidating the impact of individual nutrients and microbiota on host nervous system function and its underlying mechanisms is challenging^1,14,15. The complexity of the mammalian diet, microbiome and nervous system has limited the ability to define causal relationships between the microbiota and behavior. We leveraged the defined bacterial diet of C. elegans to identify microbiota that modulate host neural function and behavior. In a survey of bacterial diets, we find that microbiota that produce B12 and colonize the intestine, lead to persistent changes in C. elegans behavior. Our data indicate that B12 reduces excitatory cholinergic signaling through metabolic crosstalk between Met/SAM cycle and choline-oxidation pathway.

C. elegans like other animals, must obtain B12 through their diet or symbiosis with bacteria in their intestine^41,43,82. In nematodes, B12 regulates development, lifespan, lysosomal activity, gene expression and predatory behaviors^19,83–85. Human intestinal microbiota can also produce B12^41,86–88, although the consumption of animal products is likely to be the main source for this nutrient. B12 is an essential cofactor for two highly conserved metabolic enzymes methionine synthase and methylmalonyl-CoA mutase^19,43,89. Our data show that B12-dependent reduction of cholinergic signaling depends on methionine synthase, a key enzyme in the Met/SAM cycle. Previous studies have shown that B12-dependent stimulation of the Met/SAM cycle accelerates development through the synthesis of methionine and SAM¹⁹. Here we find that B12 regulates dynamic crosstalk between Met/SAM cycle and choline-oxidation pathway. This metabolic crosstalk can limit the availability of free choline, which is required for acetylcholine synthesis in neurons particularly under conditions where acetylcholine release is elevated.

In vertebrates, methionine is synthesized in the Met/SAM cycle from homocysteine by two related enzymes using 5-meTHF (methionine synthase) or betaine (BHMT) as methyl donors. C. elegans, like other invertebrates, lacks BHMT suggesting that they may rely on a single methionine synthase. Based on the similarity of methionine synthase and BHMT enzymes^69,90, and our enzymatic experiments, we hypothesize that invertebrate methionine synthases can use both betaine and 5-meTHF as methyl donors for methionine synthesis. Betaine is synthesized from choline in the choline-oxidation pathway. The metabolic connection between Met/SAM cycle and choline-oxidation pathway is well established in mammals^65,66. In C. elegans the choline-oxidation pathway genes alh-9 and chdh-1 are co-expressed with metr-1 in the intestine and the hypodermis. Our data indicate that crosstalk between choline-oxidation pathway and Met/SAM cycle impacts levels of free choline, the rate-limiting precursor in the synthesis of acetylcholine. We find that the inhibitory effect of B12 on cholinergic signaling is blocked by mutations in high-affinity choline transporter CHO-1. While we cannot exclude that B12 affects other transmitter systems, our findings are consistent with a gut-brain communication pathway in which dynamic crosstalk between B12-dependent Met/SAM cycle and choline-oxidation pathway decreases the availability of free choline required for the synthesis of acetylcholine in neurons (Fig. 7d).

The impact of B12 on cholinergic signaling and behavior is not obvious under normal conditions but becomes apparent under conditions where cholinergic signaling is increased, either genetically (unc-2(gof), acr-2(gof), and ace-1;ace-2 mutants) or during physical exertion (swimming). Under normal conditions, choline levels in cholinergic neurons can be sustained by re-uptake mechanism at the synapses^91,92. However, under genetic or behavioral conditions of increased acetylcholine release, reduced choline availability limits acetylcholine synthesis. This is consistent with observations in rodents where the synthesis and release of acetylcholine becomes strongly dependent on choline availability under conditions of increased acetylcholine release^93–95. B12 induced rewiring of choline metabolism may limit acetylcholine synthesis particularly under conditions of elevated acetylcholine release. Our findings provide insight why the relationship between diet, microbiota and the nervous system often only become apparent under conditions where the host is challenged either genetically or environmentally.

Alterations in gut microbiota have been associated with many human diseases^29–34. Human intestinal bacteria also produce B12^41,86–88. B12 is important for overall health, but particularly for the brain. A wide variety of neurological disorders are also commonly associated with B12 deficiency^38,96,97. B12 supplementation has been shown to alleviate the symptoms of several neurological disorders including autism, epilepsy, schizophrenia and migraine^98–100. However, the underlying mechanisms of the positive impact of B12 are largely unclear. Notably, these neural disorders, like unc-2(gof) and acr-2(gof) mutants, have been associated with a shift in excitatory and inhibitory balance toward excitation in the nervous system^101–103. As acetylcholine is a main excitatory neurotransmitter in the brain, it will be interesting to determine whether B12 dampens excitatory signaling in the nervous system and thus improves the symptoms of neurological disorders that are associated with excitatory and inhibitory imbalance.

Methods

Nematode strains

C. elegans strains were obtained from Caenorhabditis Genetics Center (CGC) or generated in this study (Supplementary Table 1). All strains were maintained at 22°C on nematode growth medium (NGM) plates seeded with Escherichia coli OP50 bacteria. Transgenic strains were generated by standard microinjection of the desired transgenes and a fluorescent co-injection marker (Punc-122::RFP), and at least two independent lines were tested.

Bacterial strains

The E. coli OP50 strain was used as the standard laboratory diet for C. elegans. Diverse species of bacterial strains used in this study are listed in Supplementary Table 2. All bacterial strains were grown overnight at 30°C in LB media (or mannitol media for Gluconobacter and Acetobacter). To kill bacteria, an overnight culture of OP50 (or Comamonas) grown in LB was pelleted, washed, concentrated ten-fold in water and incubated at 85°C for 30 min (1 h for Comamonas). Absence of live bacteria was confirmed by failure to grow on LB plates at 37°C overnight.

Metabolites supplementation

Stock solutions of B12 (3.2 mM), methionine (250 mM), homocysteine (500 mM), SAM (200 mM), folate (90 mM), choline (3.5 M), and betaine (5 M) were made in ddH₂O. For each experiment, each stock solution was diluted to the final concentration in NGM agar prior to plate pouring. Chemicals used in this study are listed in Supplementary Table 3.

Bacterial screen

All behavioral analysis was performed with young adults at room temperature (22–24°C). L4s grown on OP50 were transferred to plates seeded with the indicated bacterial strain to feed on the new bacterial diet and allow gut colonization. After 24 h, 20 young adults were transferred to an unseeded NGM plate and allowed to crawl away and remove remaining bacteria. To standardize behavioral conditions and eliminate any influence from bacterial density and viscosity on locomotion, animals were then transferred to a 60-mm NGM plate with a thin lawn of OP50. After 100s acclimation, animals were tracked using a Multi-Worm Tracker³⁵ for 3 min and their reversal frequency was analyzed using a custom Matlab scripts⁵⁰. Videos were recorded at 30 frames per second using a digital camera (AVT Pike F421-b) and Fire-i image acquisition software (Unibrain). The number of reversals per animal was calculated as the total number of reversals of the population divided by the average number of animals tracked. Analysis was limited to the animals that had been tracked for a minimum of 20 seconds. The acclimation and tracking times were determined based on previous studies^28,50.

Locomotion assay of ace-1;ace-2 mutants

L4s grown on OP50 were transferred to the plates seeded with OP50 ± 64 nM B12. After 24 h, a single young adult was placed on a thin lawn OP50 plate and allowed to acclimate for 100s. Total distance and duration travelled during a forward run was analyzed for 5 min using a Multi-Worm Tracker. Locomotion rate was calculated by distance traveled over time.

Convulsion phenotype of acr-2(gof) mutants

Convulsion phenotype analysis was performed as previously described⁴⁸. L4s grown on OP50 were transferred to plates seeded with OP50 ± 64 nM B12. After 24 h, a single young adult was placed on a plate with a thin lawn OP50 plate and allowed to acclimate for 3 min. Convulsion phenotype of the animal was then manually scored during 1 min as simultaneous contraction of both anterior and posterior body wall muscle.

Aldicarb resistance

Aldicarb assays were performed as described²⁸. In short, L4s grown on OP50 were transferred to plates seeded with OP50 ± 64 nM B12. After 24 h, 20 young adults were placed on an unseeded NGM agar plate with 1 mM aldicarb (Sigma-Aldrich). Animals were scored every 10 or 20 min as paralyzed if they did not move when gently touched with a platinum wire on the tail. Animals that crawled off the plate during the analysis were discarded from the analysis.

Quiescence phenotype

Quiescence assays were performed as described⁵⁴. L4s grown on OP50 were transferred to plates seeded with OP50 ± 64 nM B12 and 30 mM choline. After 24 h, 10 young adults were transferred into 24-well plates containing 500 μl M9 buffer in each well. Animals were manually scored every 5 min during 3 h with a dissecting microscope as quiescent if they did not move and maintained a straight rather than sinusoidal posture.

Body length measurement

The effect of bacterial diets or metabolites on animal growth was examined by measuring worm body length. L4s grown on OP50 were transferred to the plates seeded with the indicated bacterial strains ± metabolites and grown for 24 h. 20 young adults were imaged while freely moving with a digital camera (AVT Pike F421-b) using Fire-I image software (Unibrain). Their body length was quantified using Image J FIJI software.

Antibiotic susceptibility test for Comamonas

An overnight culture of OP50 or Comamonas grown in LB was pelleted, washed, and normalized to an OD600 of 1.0 with water and incubated at 85°C for 30 min (1h for Comamonas). Normalized cultures were next diluted 1:100 in 1 ml LB media dosed with one of seven antibiotics (ampicillin, carbenicillin, kanamycin, chloramphenicol, streptomycin, and tetracycline) at 1–10x working concentrations. Cultures were grown at RT with gentle rotation for 24 h, then 10 uL of each culture were plated on LB plates. Bacterial colonies were imaged after an overnight incubation at 37°C.

Bacterial colonization assay

To determine bacterial colonization of C. elegans, L4s grown on OP50 were transferred to plates seeded with OP50, P. aeruginosa PA14, or Comamonas. After one day, 10–20 young adults were picked and incubated in 1/50 dilution of bleaching solution at RT for 10 min to remove bacteria from worm surface. The animals were washed 4 times with cold M9, and the last wash was saved. The total number of animals was counted under dissecting microscope and lysed using 0.5 mm glass beads with vigorous vortexing in a bead-beater (Bullet Blender Gold, Next Advance Inc.) for 3 min at maximum speed. The lysates were diluted in M9 and plated on LB plates. Bacterial colonies were counted after an overnight incubation at 37°C, and the number of bacteria per animal was calculated. Absence of external bacterial contamination was confirmed by plating the last wash on LB plates. To evaluate the persistence of colonization, young adults grown on Comamonas were transferred to the plates seeded with OP50. After 1–4 days, 10–20 animals were washed, lysed, and plated on LB plates as described above, and the colonies of OP50 and Comamonas were separately counted according to their distinguishable colony size and morphology (Extended Data Fig. 2d). Bacterial colonies were identified by sequencing the 16S rRNA gene using 27f and 1492r primers¹⁰⁴ to confirm identity. To kill Comamonas bacteria in the worm gut, young adults grown on Comamonas were transferred to the plates seeded with OP50 + 200 μg/ml kanamycin. After 4–24 h, 10–20 animals were washed, lysed, and plated on LB plates as described above. The colonies of Comamonas were counted after 24 h to determine if all live bacteria were removed from the worm intestine.

Imaging and fluorescence quantification

To test the effects of individual bacterial strains on acdh-1 expression or immune response of C. elegans, L4s expressing Pacdh-1::GFP³⁹ or Pirg-1::GFP³⁶ grown on OP50 were transferred to the plates seeded with a different bacterial strain. After 24 h, young adults were mounted on 2% agarose pads containing 60 mM sodium azide. GFP images were acquired with an Axioimager Z1 microscope (Zeiss) at 10x or 20x magnification, and the maximum fluorescence intensity was quantified using Image J FIJI software. To determine expression pattern of metr-1 or chdh-1, young adults expressing either or both Pmetr-1::GFP and Pchdh-1::mCherry grown on the plates seeded with OP50 were imaged as described above.

Calcium Imaging

Young adults expressing a Pmyo-3::GCaMP6s::SL2::mCherry transgene were transferred to NGM plates seeded with a thin lawn of OP50 and allowed to acclimate for 5 min before imaging on an inverted fluorescent microscope (Axio Observer A.1, Zeiss) using a Hamamatsu ORCA-Flash4.0 camera equipped with an image splitter (OptoSplit II, Cairn Research Ltd.) with filters for simultaneous dual color imaging (49502 GFP/mCherry, Chroma Technology). Micro-manager software¹⁰⁵ was used to acquire images during 5 min while animals were freely moving on plate. Recordings were analyzed using custom Matlab scripts (MathWorks Inc.) as follows: The brightfield image in mCherry channel was used to generate a binary mask around the worm and the mean fluorescent intensity of the corresponding pixels in the GFP channel was measured. The mean intensity of the pixels outside of the mask were measured to determine the background. The values reported in the figures represent the mean intensity within the mask minus the background intensity.

Molecular biology

Standard molecular biology techniques were used. For Pmetr-1::GFP, a 2883 bp fragment upstream of the metr-1 start site was amplified from genomic DNA by PCR and cloned into pPD95.70 GFP plasmid. The plasmid was injected into N2 animals at 80 ng/μl and integrated by X-ray irradiation followed by outcrossing three times. For Pchdh-1::mCherry, a 1200 bp fragment upstream of the chdh-1 start site was amplified from genomic DNA by nested PCR using two pairs of primers and cloned into mCherry vector. The plasmid was injected at 80 ng/μl with Pmetr-1::GFP (80 ng/μl) into N2 animals. For rescue constructs, the metr-1 cDNA was amplified from a first-strand cDNA library and cloned behind the endogenous metr-1, intestinal elt-2, hypodermal dpy-7, muscle-specific myo-3, or Pan-neuronal tag168 promoter. The plasmid was injected at 5 ng/μl with the co-injection marker Punc-122::RFP (80 ng/μl) and pBSK (75 ng/μl) into metr-1(ok521);unc-2(zf35gf) mutants. Primers used in this study are listed in Supplementary Table 4.

Measurement of free choline and acetylcholine levels

Approximately 25,000 synchronized young adults grown in liquid S-complete medium with OP50 ± 64 nM B12 were washed, resuspended in 10% TCA, and frozen in dry ice/ethanol bath. The samples were sonicated for 90 sec (15s on/60s off pulse cycles at 20% power) on ice-water bath using ultrasonic processor (Misonix, XL2020) and centrifuged (16,000g, 30 min, 4°C). The pellets were resuspended in 1M NaOH and used to determine total protein concentrations using Bradford method (Thermo Scientific). Supernatant was extracted with ether to remove TCA, desiccated, and resuspended in assay buffer. The levels of free choline and acetylcholine were determined using a fluorometric Choline/Acetylcholine assay kit (Abcam) according to the manufacturer’s protocol. The values were normalized to total protein concentrations. We observed a slight increase of approximately 3–5% in the protein amount in the B12 sample, which may reflect the enhanced animal growth rate on B12. This increase did not have a significant impact on the observed effect of B12. The total amount of metabolites and proteins in each sample are shown in the source data.

Sample preparation for HPLC-MS

Worm pellets were lyophilized for 18–24 h using a VirTis BenchTop 4K Freeze Dryer. Dried pellets were transferred to 1.5 mL microfuge tubes and dry pellet weight recorded. Pellets were disrupted in a Spex 1600 MiniG tissue grinder after addition of three stainless steel grinding balls to each sample. Microfuge tubes were placed in a Cryoblock (Model 1660) cooled in liquid nitrogen, and samples were disrupted at maximum speed for six cycles of 30s. Pellets were transferred to 8 mL glass vials in 6 mL LC-MS grade methanol. Samples were stirred overnight at RT following the addition of a stir bar (VWR 58948–375). Glass vials were centrifuged (2750g, 5 min) in an Eppendorf 5702 Centrifuge (rotor F-35-30-17). 5 mL of the supernatant was transferred to a clean 8 mL glass vial and concentrated to dryness in an SC250EXP Speedvac Concentrator coupled to an RVT5105 Refrigerated Vapor Trap (Thermo Scientific). The residue was suspended in 150 μL methanol, briefly sonicated, transferred to a 1.5 mL microfuge tube and spun (20,000g, 5min) in an Eppendorf 5417R Centrifuge to sediment precipitate. The supernatant was transferred to an HPLC vial and stored at −20°C or analyzed immediately.

HPLC-MS Analysis

Normal-phase chromatography was performed using a Vanquish LC system controlled by Chromeleon Software 7.2.9 (Thermo Scientific) and coupled to an Orbitrap Q-Exactive High Field mass spectrometer controlled by Xcalibur software 4.1.31.9 (Thermo Scientific). Methanolic extracts prepared as described above (1 μL injection) were separated on an Agilent Zorbax RRHD HILIC Plus column (150 mm x 2.1 mm, particle size 1.8 μm) maintained at 40°C with a flow rate of 0.5 mL/min. Solvent A: 0.1% formic acid in water; solvent B: 0.1% formic acid in acetonitrile. A/B gradient started at 95% B for 2 min after injection and decreased linearly to 50% B at 20 min, followed by a linear decrease to 10% B at 22 min, held at 10% B until 25 min, followed by a linear increase to 90% B at 27 min, and finally held at 90% B for an additional 3 min to re-equilibrate the column. Mass spectrometer parameters: spray voltage (+3.5 kV), capillary temperature 380°C, probe heater temperature 400°C; sheath, auxiliary, and sweep gas 60, 20, and 1 AU, respectively. S-Lens RF level: 50, resolution 120,000 at m/z 200, AGC target 3E6, m/z range 70–700. HPLC-MS data were analyzed by Xcalibur Quan Browser (Thermo Scientific). Acetylcholine and amino acid standards were purchased from Sigma Aldrich.

RNAi experiments

RNAi bacteria clones obtained from the Ahringer library¹⁰⁶ were selected by ampicillin (100 mg/ml) and tetracycline (12.5 mg/ml) and verified by sequencing. The RNAi bacteria grown at 37°C overnight in LB with ampicillin were concentrated (4X) and seeded on RNAi NGM plates that contain 6 mM IPTG and 100 mg/ml ampicillin. Synchronized L1s were placed on RNAi plates and allowed to grow at RT. 20 RNAi-treated L4s were transferred to new RNAi plates ± 64 nM B12. Young adults were analyzed for behavioral and growth phenotypes after 24 h.

Phylogenetic analysis

Phylogenetic analysis was conducted for the homocysteine binding domains in the METR-1 and other homocysteine methyltransferase enzymes from bacteria (E. coli, Oceanobacillus iheyensis, Pelagibacter ubique, Sinorhizobium meliloti, Pseudomonas aeruginosa), fly (Drosophila melanogaster), yeast (Saccharomyces cerevisiae, Candida albicans), plant (Arabidopsis thaliana) and mammals (Homo sapiens, Mus musculus). The accession number of the protein sequences were obtained from the NCBI database and listed in Supplementary Table 5. The homocysteine binding domains were predicted using the PROSITE database (https://prosite.expasy.org)¹⁰⁷. The phylogenetic tree was constructed by the Maximum likelihood analysis using the NGPhylogeny system (https://ngphylogeny.fr)¹⁰⁸. Bootstrap values were given at relevant nodes.

In vitro activity assay of METR-1

Methionine synthase activity of purified METR-1 proteins was measured under standard assay conditions based on previous studies^109,110 . To isolate METR-1 proteins, approximately 100,000 synchronized young adult unc-2(gof);metr-1 animals ± (control) Pmetr-1::METR-1::GFP grown in liquid S-complete medium with Comamonas were harvested, washed, and immediately frozen in dry ice/ethanol bath. The frozen worm samples were lysed in lysis buffer (50 mM Tris-HCl pH 7.5, 100 mM NaCl, 50 mM KCl, 0.5 mM EDTA, 0.5 % NP-40, 2 mM PMSF, 1X protease inhibitor) by sonication on ice-water bath and centrifuged (16,000g, 30 min, 4°C). The protein concentration of supernatant was measured using the Bradford method, and about 40 mg of total proteins were incubated with GFP-Trap A beads (Chromotek) overnight at 4°C. The beads were washed 5 times by cold 50 mM Tris-HCl (pH 7.5) and incubated in 50 μl of B12-dependent methionine synthase or BHMT assay buffer for 30, 90, 180 min at 25°C. After the indicated incubation time, the methionine levels in 10 μl of supernatants were determined using a fluorometric Methionine Assay Kit (Abcam, ab234041) according to the manufacturer’s protocol. B12-dependent methionine synthase assay buffer contained 50 mM Tris-HCl (pH 7.5), 500 μM DL-homocysteine (Sigma), 50 μM methyl-cobalamin (Sigma), 250 μM 5-methyl-tetrahydrofolic acid disodium salt (5me-THF, Sigma), 250 μM S-(5’-Adenosyl)-L-methionine chloride dihydrochloride (SAM, Sigma), and 25 mM DTT. BHMT assay buffer contained 50 mM Tris-HCl (pH 7.5), 500 μM DL-homocysteine, 50 μM methyl-cobalamin, and 500 μM betaine (Sigma). When 5me-THF was used as a methyl donor instead of betaine in the BHMT assay buffer, the amount of methionine formed was almost negligible (Extended Data Fig. 6b), consistent with the requirement of SAM and a reducing agent for B12-dependent methionine synthase activity^109,110.

Western blot and Silver stain

Isolated METR-1::GFP proteins were detected by standard Western blot methods. Briefly, proteins eluted from the beads were separated on pre-cast 4%-10% polyacrylamide gels (Bio-Rad) and transferred to polyvinylidene fluoride membranes (Millipore). The membranes were probed with anti-GFP polyclonal antibody (1:1000, Abcam, #ab6556) and immunoreactivity was detected using a chemiluminescent reagent (Thermo scientific). Total individual proteins bound on the beads were analyzed using a Silver Stain Kit (Bio-Rad) according to the manufacturer’s protocol.

Rescue of developmental arrest caused by mthf-1 deletion

mthf-1(gk465) mutants were isolated in a screen for homozygous lethal mutations on chromosome II using a balancer strain DR2078 [bli-2(e768) unc-4(e120)/mIn1[mIs14 dpy-10(e128)]]⁷⁴. The balanced mthf-1(gk465) mutants (bli-2(e768) mthf-1(gk465) unc-4(e120)/mIn1[mIs14 dpy-10(e128)]) were maintained on OP50 for at least two generations without food deprivation. 4–5 L4 heterozygous mthf-1(gk465) mutants were transferred to plates seeded with OP50 ± 64 nM B12 and/or 75 mM betaine. Mutant animals were allowed to lay eggs for 20–24 h, after which adult animals were removed. The percentage of Unc mthf-1(gk465) homozygous animals were followed through development until larval death or reaching adult stage. WT heterozygous and Dpy homozygous animals were removed to avoid food depletion during the assay. mthf-1(gk465) homozygous mutants (bli-2(e768) mthf-1(gk465) unc-4(e120)) that reached adulthood were scored based on Bli Unc phenotype, specifically the Bli phenotype as it becomes apparent during adult stage. Only 10% of mthf-1(gk465) homozygous mutants reached adulthood on OP50. In the presence of B12 or betaine, 26% and 35% of mthf-1(gk465) homozygous mutants reached adulthood, respectively. 49% of mthf-1(gk465) homozygous mutants reached adulthood when supplemented with both B12 and betaine.

Statistics and Reproducibility

All data are represented in a format that shows data distribution clearly (dot plots) and all the graph elements (median, error bars) are defined in each figure legend. Sample size was not predetermined by any statistical tests. In all figures and figure legends, n represents the number of independent experiments unless specified in the figure legends, and the total number of animals analyzed is also specified. Statistical comparisons were performed using ANOVA with Dunnett’s, Tukey’s, or Sidak’s correction for multiple samples or using an unpaired Studenťs t test for two samples. The exact P-values were indicated in the relevant graphs. n indicates the number of biological replicates.

Extended Data

Extended Data Fig. 1. Effects of different bacterial diets on growth, immune response, and acdh-1 expression of C. elegans.

a, Growth rate as indicated by body length of unc-2(gof) mutants grown on the indicated bacterial strains for 24 h (mean ± s.e.m., n = 3, number of animals indicated, one-way ANOVA with Dunnett’s multiple comparison). b, Fluorescence values of immune reporter Pirg-1::GFP after 24 h exposure to indicated bacterial strains. Pathogenic strain P. aeruginosa 14 was included as a positive control (mean ± s.e.m., n = 3, number of animals indicated, one-way ANOVA with Dunnett’s multiple comparison). c, Representative DIC and fluorescence images of Pacdh-1::GFP animals fed different bacterial diets for 24 h. Shades of green represent relative GFP expression levels, “High” indicates strong fluorescence throughout the intestine as in the OP50 shown, “Low” indicates barely detectable fluorescence as in the Comamonas shown, “Moderate” indicates visible fluorescence but weaker compared to the GFP signal on OP50 (n = 3 biologically independent samples with similar results). Scale bar is 300 μm. d, Growth rate as indicated by body length of wild-type and unc-2(gof) mutants fed OP50, OP50 with 64 nM B12, Comamonas, or Comamonas cbiAΔ for 24 h (mean ± s.e.m., n = 3, number of animals indicated, one-way ANOVA with Dunnett’s multiple comparison). e, Reversals of wild-type and unc-2(gof) mutants fed OP50, Comamonas, Comamonas cbiAΔ, or Comamonas cbiAΔ with 64 nM B12 for 24 h (mean ± s.e.m., n and number of animals indicated, one-way ANOVA with Dunnett’s multiple comparison). f, Growth rate as indicated by body length of wild-type and unc-2(gof) mutants fed live or heat-killed OP50 ± 64 nM B12 for 24 h (mean ± s.e.m., n = 3, number of animals indicated, one-way ANOVA with Tukey’s multiple comparison). Source numerical data are available in source data.

Extended Data Fig. 2. Antibiotic susceptibility for Comamonas gut bacteria elimination.

a, Bacterial CFU per animal from unc-2(gof) mutants grown on Comamonas treated with the indicated concentrations of kanamycin for 24 h (mean ± s.e.m., n = 3, number of animals indicated, one-way ANOVA with Dunnett’s multiple comparison). b, Reversal frequency of unc-2(gof) mutants grown on OP50 with indicated concentrations of kanamycin for 24 h (mean ± s.e.m., n and number of animals indicated, one-way ANOVA with Dunnett’s multiple comparison). c, Bacterial colonies of OP50 and Comamonas isolated the worm gut. Bacterial colonies were isolated and identified by 16S rRNA gene using 27f and 1492r primers (Methods) (n = 21 independent experiments with similar results). Source numerical data are available in source data.

Extended Data Fig. 3. B12 reduces cholinergic signaling especially under conditions of increased acetylcholine release.

a, Reversal frequency of unc-2(gof) and cat-1(ok411) unc-2(gof) mutants fed OP50 ± 64 nM B12 (mean ± s.e.m., n and number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). b, Quantification of locomotion speed of wild-type and ace-1(p1000);ace-2(g72) mutants fed OP50, OP50 with 64 nM B12, Comamonas, or Comamonas cbiAΔ for 24 h (mean ± s.e.m., n = 4, number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). c-f, Quantification of reversal frequency (c), locomotion speed (d), head bending (e), and body bending (f) of wild-type and unc-2(gof) mutants fed OP50 ± 64 nM B12 for 24 h (mean ± s.e.m., n and number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). g, Quantification of paralysis percentage of wild-type animals fed OP50 on 1 mM aldicarb-containing NGM agar plates or M9 liquid buffer (mean ± s.e.m., n = 4 (crawling on agar), n = 5 (swimming on liquid), number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). Source numerical data are available in source data.

Extended Data Fig. 4. B12 regulates C. elegans behavior and growth through Met/SAM cycle.

a, Reversals of unc-2(gof), sams-1(ok2946);unc-2(gof), cbs-2(ok666);unc-2(gof), pcca-1(ok2282) unc-2(gof), or mce-1(ok243);unc-2(gof) mutants fed OP50 ± B12 for 24 h (mean ± s.e.m., n and number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). b, Growth rate of unc-2(gof), mmcm-1(ok1637);unc-2(gof), metr-1(ok521);unc-2(gof), sams-1(ok2946);unc-2(gof) mutants fed OP50 ± B12 for 24 h (mean ± s.e.m., n = 3, number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). c, Quantification of paralysis percentage of unc-2(gof) and sams-1(ok2946);unc-2(gof) mutants fed OP50 ± B12 on 1 mM aldicarb (mean ± s.e.m., n = 4, number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). d, Quantification of acetylcholine in unc-2(gof) mutants fed OP50 ± B12 for 24 h (mean ± s.e.m., n indicated, one-way ANOVA with Dunnett’s multiple comparison). e, Quantification of acetylcholine in WT and metr-1(ok521) mutants fed OP50 ± B12 for 24 h (mean ± s.e.m., n indicated, two-way ANOVA with Tukey’s multiple comparison). f, Quantification of methionine (Met) in wild-type, metr-1(ok521), unc-2(gof), metr-1(ok521);unc-2(gof) mutants fed OP50 ± B12 for 24 h (mean ± s.e.m., n and number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). g, Growth rate of metr-1(ok521);unc-2(gof) mutants expressing metr-1 cDNA driven by the indicated tissue-specific promoter fed OP50 ± B12 for 24 h (mean ± s.e.m., n = 3, number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). h-i, Growth rate (h) and reversals (i) of unc-2(gof) mutants fed OP50 with the indicated metabolites for 24 h (mean ± s.e.m., n = 3 (h), n indicated (i), number of animals indicated, one-way ANOVA with Dunnett’s multiple comparison). j-k, Quantification of homocysteine (Hcy) (j), S-adenosylhomocysteine (SAH) (k) in wild-type, metr-1(ok521), unc-2(gof), metr-1(ok521);unc-2(gof) mutants fed OP50 ± B12 for 24 h (mean ± s.e.m., n and number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). Source numerical data are available in source data.

Extended Data Fig. 5. Choline metabolism is linked to the Met/SAM cycle.

a, Growth rate as indicated by body length of unc-2(gof) mutants fed OP50 ± 64 nM B12 with 30 mM choline for 24 h (mean ± s.e.m., n =3, number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). b, Expression pattern of Palh-9::GFP in the intestine, hypodermis, and RIM neurons (n = 3 biologically independent samples with similar results). Scale bar is 100 μm. c, Growth rate as indicated by body length of unc-2(gof) mutants subjected to RNAi knockdown of chdh-1 or alh-9 fed OP50 ± 64 nM B12 for 24 h (mean ± s.e.m., n = 3, number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). Source numerical data are available in source data.

Extended Data Fig. 6. in vitro activity assay of METR-1.

a, SDS-PAGE of purified METR-1 proteins. Purity of immunopurified GFP-tagged METR-1 protein was analyzed with 4%-10% polyacrylamide gels electrophoresis under denaturing conditions. Panels show silver staining (left) and Western blot stained with a GFP antibody (right). Arrowhead indicates METR-1::GFP protein (molecular weight ~ 166 kDa). (n = 3 independent experiments with similar results) b, Methionine synthase activity of the purified METR-1::GFP protein was assayed in the presence of 5-methyltetrahydrofolate (5-meTHF) or betaine as a methyl donor ± SAM and DTT. Enzyme reaction was performed in 50 mM Tris-HCl buffer (pH 7.5) at 25°C for 6 h (Methods) (mean ± s.e.m., n and number of animals indicated, one-way ANOVA with Dunnett’s multiple comparison). Source numerical data and source blot images are available in source data.

Extended Data Fig. 7. A neuronal choline transporter is required to mediate the effect of B12 on excitatory transmission.

a, Reversal frequency of unc-2(gof), acr-23(ok2804);unc-2(gof), or lgc-41(sy1494) unc-2(gof) mutants fed OP50 ± 64 nM B12 for 24 h (mean ± s.e.m., n and number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). b, Reversal frequency of unc-2(gof) or deg-3(u701);unc-2(gof) mutants fed OP50 ± 64 nM B12 for 24 h (mean ± s.e.m., n and number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). c, Reversal frequency of unc-2(gof) mutants fed OP50, OP50 with B12 (64 nM), OP50 with betaine (75 mM), or OP50 with both B12 and betaine (mean ± s.e.m., n and number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). d, Growth rate as indicated by body length of unc-2(gof) and cho-1(tm373);unc-2(gof) mutants fed OP50 ± 64 nM B12 for 24 h (mean ± s.e.m., n = 3, number of animals indicated, two-way ANOVA with Tukey’s multiple comparison). Source numerical data are available in source data.

Data availability

All data files from this study have been uploaded to figshare and are freely available at: https://doi.org/10.6084/m9.figshare.21197953. Source data including metabolic data have been provided in Source Data. All other data and reagents generated in this study are available from the corresponding author, Mark J. Alkema (mark.alkema@umassmed.edu), on reasonable request.