Epigenetic constraints and enhancer innovation link neuronal plasticity to evolutionary adaptation

Significance

How do nervous systems evolve new cell types while maintaining core functions? This study reveals that small changes in gene regulation can enable neurons to adopt new roles, influencing behavior and evolution. Using nematodes, we show that some neurons naturally suppress serotonin uptake through epigenetic mechanisms, but this trait can be activated by environmental cues. In related species, a genetic change makes this trait permanent, altering egg-laying behavior and potentially offering an advantage in serotonin-rich environments. These findings illustrate how flexible gene regulation balances cellular stability with adaptive plasticity, providing a model for understanding the evolution of neural circuits and their responses to environmental challenges.

Abstract

How nervous systems balance the generation of robust neuron types with gene expression plasticity mechanisms and how these processes impact cell-type evolution are unknown. Here, we use Caenorhabditis species to study neuron-type robustness, plasticity, and evolution, using VC4 and VC5 cholinergic motoneuron types as models. In Caenorhabditis elegans, we found that epigenetic silencing through histone 3 lysine 9 methylation (H3K9me) is necessary to suppress the expression of the serotonin reuptake gene mod-5/Sert and a serotonergic phenotype in these cells. In contrast, we observed that VC4 and VC5 neurons in the Angaria group of Caenorhabditis species have evolved an intense serotonergic staining. This phenotype is caused by the emergence of a new enhancer in the mod-5/Sert locus, which has been recruited to the ancestral neuron-type gene regulatory network. Enhancer transfer from C. angaria is sufficient to impose a constitutive serotonergic fate in C. elegans. Remarkably, acquiring this new trait modulates egg-laying responses to high levels of exogenous serotonin, which can be found in specific nematode environments. Finally, we found that the repression of the serotonergic fate in C. elegans VC4 and VC5 neurons is indeed a plastic trait that can be adjusted in specific environmental growth conditions to elicit egg-laying behaviors similar to those observed in Angaria species. Our work uncovers a dual regulatory paradigm: Epigenetic constraints harmonize neuron-type robustness and plasticity, while enhancer co-option enables evolutionary innovation. This mechanistic plasticity directly links regulatory adaptation to behavioral diversification.

Nervous systems are remarkably complex, containing an extraordinary array of cell types with highly specialized functions. The generation of this diversity is a deeply stereotyped process in which all individuals from the same species give rise to an identical variety of neuron types. Nonetheless, the evolution of nervous systems requires the emergence of novel neuronal identities, cellular functions, and behaviors. Changes in the architecture of gene regulatory networks have been proposed to play a key role in this evolutionary process (1, 2). However, the mechanisms controlling the robustness of neuronal development and their impact on neuronal evolution have been challenging to delineate at the molecular level.

One example of neuronal diversity is found among serotonergic systems. In mammals, serotonin (5HT) is a widely distributed neurotransmitter controlling essential functions, including sensory processing, cognitive control, internal states, autonomic responses, and motor activity (3). 5HT neurons can be classified as bona fide 5HT-synthesizing, if they contain the enzymes required for the synthesis of 5HT, or 5HT-reuptaking, if they express SERT, the transporter necessary to upload 5HT from the extracellular space (3) (Fig. 1A). 5HT-synthesizing neurons are restricted to the raphe region, where they are born during embryonic development and perdure in adulthood (3). In contrast, 5HT-reuptaking neurons are found throughout the cortex, thalamus, and retina only during development due to transient SERT expression in these neurons (3). This ephemeral serotonergic feature has been involved in axon guidance, synapse formation, or cell survival (3, 4). However, mechanisms controlling the temporally restricted 5HT reuptake phenotype are not well understood (5).

Fig. 1.

Epigenetic silencing prevents VC4 and VC5 serotonergic phenotype in C. elegans. (A) Schema of the minimal set of proteins required for the synthesis (Left) or reuptake and package (Right) of 5HT. Black: C. elegans names, Gray: Human names. (B) Schematic representation of main components of Histone 3 lysine 9 methylation pathway in C. elegans. (C) VC4 and VC5 5HT staining quantification for H3K9me pathway mutants met-2(n4256), set-25(n5021), lin-61(n3809), and hpl-2(tm1489) (n = 84, 282, 67, 194, 318). Fisher’s exact test was applied for statistics (P value < 0.001, 0.808, <0.001, <0.001). (D) Schematic representation and representative micrographs of VC4 and VC5 5HT staining in wild type and H3K9me mutants. (E) Quantification of H3K9me pathway mutants rescue experiments using VC specific promoter shows epigenetic repression of serotonergic phenotype is cell autonomous. Mean values from three independent lines (n = 185, 171, 170, 120, 173, 143). Fisher’s exact test was applied for statistics (P value < 0.001, <0.001, <0.001). See Dataset S1 for source data.

In animals, the use of 5HT as a neurotransmitter dates to the bilaterian origin, being present in annelids, mollusks, arthropods, nematodes, and vertebrates (6). Assignments of neuronal homologies across species are required to study neuron-type evolution. Nevertheless, this process can be challenging in complex organisms due to significant anatomical differences. Advantageously, all nematode species in the Caenorhabditis genus share a common invariable embryonic lineage that gives rise to a strikingly identical number of neurons placed in the same body locations (7, 8). Thus, the genetic amenability of C. elegans and the existence of closely related Caenorhabditis nematodes offer a unique opportunity to investigate the generation of cell diversity and the emergence of novel neuronal features underlying key evolutionary steps. C. elegans contains multiple serotonergic neuron types that regulate sensory processing, motor function, and internal states (9–11). As in mammals, these cells can be classified as 5HT-synthesizing and 5HT-reuptaking neurons, although, in contrast to mammals, C. elegans 5HT reuptaking neurons are maintained throughout the life of the animal (12, 13). In addition to the well-characterized serotonergic system, two cholinergic motoneurons in C. elegans (VC4 and VC5 neuron types) have been reported to display weak 5HT staining (14, 15). However, this staining was described as sporadic and variable among individuals and experiments, contrasting with the stereotypical and robust classification of all other neuron types in C. elegans (16). This discrepancy prompted us to investigate the molecular mechanisms underlying the sporadic 5HT staining in VC4 and VC5 neurons and the evolvability of this phenotype.

Here, we found that active epigenetic silencing is employed by C. elegans VC4 and VC5 neurons to prevent a 5HT reuptake phenotype. In contrast, VC4 and VC5 neurons in Caenorhabditis species of the Angaria group exhibit intense and robust 5HT staining due to the emergence of a new enhancer driving constitutive mod-5/Sert expression in these cells. We also unraveled that the VC4 and VC5 5HT phenotype explains specific differences in egg-laying behavior between C. elegans and C. angaria species. Finally, we revealed that the repression of the VC4 and VC5 serotonergic phenotype in C. elegans is a plastic trait that can be modulated, and we identified specific environmental conditions that overcome epigenetic silencing of mod-5/Sert in this species, resulting in egg-laying behaviors like those found in C. angaria. In sum, our results highlight the importance of epigenetic repression in the canalization of neuron-type identities, illuminate specific gene regulatory changes driving the evolution of novel neuronal phenotypes and behaviors, and identify unique mechanisms of neuronal identity plasticity in response to environmental signals. These findings illustrate how the modulation of repressive mechanisms impacts developmental plasticity and how regulatory sequence evolution drives neuron type novelties in nervous systems.

Results

Epigenetic Silencing Restricts VC4 and VC5 Serotonergic Phenotype in C. elegans.

As previously reported (14, 15), weak 5HT staining in VC4 and VC5 neurons was observed in a low percentage of C. elegans (N2 strain) animals under our standard laboratory conditions (Fig. 1C), in contrast to robust staining found in all known 5HT synthesizing (ADF, NSM, and HSN) and reuptaking (AIM and RIH) neurons in C. elegans. The epigenetic mark Histone H3 lysine 9 methylation (H3K9me) is associated with heterochromatin and transcriptional silencing (17) (Fig. 1B). Previous reports indicate that H3K9me can be deposited in a cell-type-specific manner to repress expression of undesired effector features of related cellular lineages, like olfactory receptor repression in the diversification of mouse olfactory receptor neurons (17–22). We find that mutants for different H3K9me pathway components (23) (specifically alleles for mono- and dimethyl transferase met-2/Setdb1 and hpl-2/Hp1 and lin-61/L3mbtl2 protein silencers) show a strong, reproducible, and highly penetrant 5HT staining in VC4 and VC5 neurons (Fig. 1 C and D and Dataset S1). The VC4 and VC5 serotonergic phenotype is rescued with the reexpression of the corresponding gene under a VC neuron-specific promoter supporting a cell-autonomous role for H3K9me (Fig. 1E and Dataset S1). Next, we aimed to determine if VC4 and VC5 neuron types become 5HT synthesizing or reuptaking neurons in these mutant backgrounds. Wild type C. elegans VC4 and VC5 neurons constitutively express the vesicular monoamine transporter cat-1/Vmat (14). Of note, the function of this gene in the otherwise cholinergic VC4 and VC5 motoneurons remains undetermined. According to cat-1/Vmat expression, the acquisition of mod-5/Sert expression should be sufficient to acquire the serotonergic phenotype (Fig. 1A). Thus, we considered 5HT reuptake from nearby HSN 5HT synthesizing neurons the more plausible scenario. To test this hypothesis, we built double mutants for H3K9me pathway components together with mutants for transcription factors necessary for HSN differentiation, which are not expressed in VC4 and VC5 neurons [ast-1/Fli1 (24) and unc-86/Brn3c (25, 26)]. We find that these genetic backgrounds display a complete absence of VC4 and VC5 5HT staining, in agreement with the acquisition of a 5HT reuptaking phenotype (Fig. 2 A and B and Dataset S2). In complementary experiments, we used laser ablation to eliminate HSN neurons in lin-61/L3mbtl2 mutant animals at the first larval stage, before HSN maturation. HSN depletion leads to substantial defects in VC4 and VC5 5HT staining (Fig. 2 A and B and Dataset S2), further supporting a reuptake phenotype for these neuron types in H3K9me mutants. Interrogation of available chromatin immunoprecipitation and sequencing (ChIP-seq) data (27) reveals C. elegans wild type animals show high H3K9me marks and HPL-2/HP1 recruitment to the mod-5/Sert locus (Fig. 2C). Double mutant of hpl-2/Hp1 with mod-5/Sert alleles suppresses the hpl-2/Hp1 5HT staining phenotype confirming VC4 and VC5 neurons in H3K9me mutants reuptake 5HT in a mod-5/Sert-dependent manner (Fig. 2 D and E and Dataset S2). However, we failed to detect mod-5/Sert expression in VC4 and VC5 neurons in H3K9me pathway mutants using two different endogenously tagged strains (mod-5::T2A::NeonGreen or mod-5::SL2::StayGold2::H2B), (SI Appendix, Fig. S1A and Dataset S11), which could be due to low levels of expression. Alternatively, hpl-2 mutants could result in the indirect modification of mod-5 activity, for example by regulating its transport, localization, activity, or stability. Nevertheless, we found that induced mod-5/Sert expression in otherwise wild type VC4 and VC5 neurons is sufficient to drive intense 5HT staining (Fig. 2 F and G and Dataset S2). Altogether, our results support a model in which mod-5/Sert expression in VC4 and VC5 neurons is actively epigenetically repressed by the H3K9me pathway, suppressing the 5HT reuptaking phenotype.

Fig. 2.

mod-5/Sert expression is necessary and sufficient to provide VC4 and VC5 serotonergic phenotype in C. elegans. (A) Schematic representation of experimental approaches used to eliminate 5HT from HSN neurons in H3K9me mutants. (B) Quantification of VC4 and VC5 serotonergic phenotype in wild type, H3K9me mutants hpl-2(tm1489) and lin-61(n3809), and double mutants hpl-2(tm1489); ast-1(ot417), lin-61(n3809); unc-86(n846) and lin-61(n3809) with HSN laser ablation (n = 41, 78, 74, 18, 150, 54, 21). Fisher’s exact test was applied for statistics (P value < 0.001, <0.001, <0.001. (C) C. elegans larval stage L4 ChIPseq mono-, di-, and trimethylation levels and HPL-2 binding. mod-5 locus (marked with gray shadow) shows high me2, me3, and HPL-2 recruitment. (D) Schematic representation and micrograph of VC4 and VC5 5HT staining in double mutant H3K9me pathway and mod-5(knu383). (E) Quantification of (D), VC4 and VC5 serotonergic phenotype in H3K9me mutants requires mod-5 activity (n = 35, 80, 89, 82). Fisher’s exact test was applied for statistics (P value < 0.001). (F) Schematic depiction and representative micrographs of 5HT staining in C. elegans control animals and animals expressing mod-5/Sert cDNA under a VC specific promoter. Expression of mod-5/Sert is sufficient to drive serotonergic phenotype in C. elegans. (G) Quantification of (F) (n = 46, 19). Fisher’s exact test was applied for statistics (P value < 0.001). (H) Schematic depiction and representative micrographs of 5HT staining of double mutant H3K9me pathway and unc-4(gk705). (I) Quantification of (H) (n = 35, 31, 18, 80), Fisher’s exact test was applied for statistics (P value < 0.001). See Dataset S2 for source data.

Low UNC-4/UNCX Restricts VC4 and VC5 Serotonergic Phenotype in C. elegans.

Epigenetic modifiers act in concert with transcription factors (TFs) to activate or repress specific target genes (18). Thus, we next aimed to identify TFs involved in activating the VC4 and VC5 5HT reuptake phenotype in H3K9me pathway mutants. The homeodomain TF UNC-4, a homolog of vertebrate UNCX, is required for the terminal differentiation of VC4 and VC5 cholinergic neurons (28–30). We found that UNC-4/UNCX is also necessary for constitutive cat-1/Vmat reporter expression in VC4 and VC5 neurons (SI Appendix, Fig. S1 B and C). Analysis of the cat-1/Vmat minimal cis regulatory module (CRM or enhancer) active in VC4 and VC5 neurons revealed the presence of a functional homeodomain TF binding site, suggesting UNC-4/UNCX action could be direct (SI Appendix, Fig. S1D). In addition, VC4 and VC5 5HT staining is absent in unc-4/Uncx; lin-61/L3mbtl2 double mutants, revealing UNC-4/UNCX is necessary for serotonergic phenotype induction in H3K9me mutants (Fig. 2 H and I and Dataset S2). Considering these results, we next aimed to assess if increased UNC-4/UNCX levels could overcome H3K9me repression. We observed that UNC-4/UNCX overexpression through a heat shock inducible promoter is sufficient to induce 5HT staining in C. elegans VC4 and VC5 neurons (SI Appendix, Fig. S1 E and F). Similar to H3K9me pathway mutants, we failed to detect VC4 and VC5 mod-5/Sert expression using endogenously tagged mod-5::T2A::NeonGreen strain (SI Appendix, Fig. S1H). However, two pieces of evidence suggest UNC-4/UNCX overexpression acts through mod-5/Sert induction: First UNC-4/UNCX overexpression induced serotonergic phenotype is absent in mod-5/Sert mutants (SI Appendix, Fig. S1F), and second, high levels of UNC-4/UNCX are sufficient to induce detectable VC4 and VC5 GFP expression from the multicopy array mod-5prom3::gfp transgenic reporter which, under control conditions, is only expressed in NSM and ADF neurons that synthesize serotonin (SI Appendix, Fig. S1 G and H). Finally, we examined if serotonergic phenotype of H3K9me mutants is mediated through increasing UNC-4/UNCX levels. However, quantification of an endogenous fluorescence tagged unc-4/Uncx allele reveals a slight but significant decrease in UNC-4/UNCX levels in VC4 and VC5 neurons (SI Appendix, Fig. S1I). This observation is consistent with previous data showing that H3K9me restricts the binding of TF fate determinants to the regulatory regions of unwanted target effector genes but does not repress TF expression (18). Altogether, our results support a model in which C. elegans mod-5/Sert locus bears some latent enhancer activity in VC4 and VC5 neuron types that is actively prevented by two parallel mechanisms, restricted UNC-4/UNCX expression levels and epigenetic H3K9me repression.

VC4 and VC5 Serotonergic Phenotype in Angaria Group of Species.

The serotonergic phenotype of VC4 and VC5 neurons has been reported in other nematode species (31) but has not been systematically studied in the Caenorhabditis genus. We performed 5HT immuno-staining in 26 Caenorhabditis species (Fig. 3 A and B) and found highly penetrant and strong VC4 and VC5 5HT staining in C. angaria, as previously reported (31), and in four additional species of the Angaria group (Fig. 3 A and B and Dataset S3). These data suggest that VC4 and VC5 neurons acquired the ability to contain 5HT in the common ancestor of the Angaria group (synapomorphy). Next, we aimed to identify the molecular determinants of this acquired trait.

Fig. 3.

VC4 and VC5 neurons from Angaria group of species are serotonergic. (A) Schema of soma locations for serotonergic and VC neurons in Caenorhabditis species and phylogeny of Caenorhabditis species (adapted from http://caenorhabditis.org/). Heatmap quantification of serotonergic staining. VC4 and VC5 neurons display strong 5HT staining in all analyzed Angaria group of species. Additional VC neurons also display 5HT staining albeit more variable and at lower penetrance. n > 25 animals per specie. (B) Micrographs showing illustrative examples of 5HT staining in the vulva region of several species, including a lateral view of C. quiockensis displaying 5HT staining in VC1 to VC5. See Dataset S3 for source data.

C. angaria VC4 and VC5 Neurons Reuptake Serotonin from HSN Neurons.

To determine if, like in C. elegans H3K9me mutants, the serotonergic phenotype of C. angaria VC4 and VC5 neurons is due to 5HT reuptake, we eliminated the HSN neurons by laser ablation. C. angaria HSN ablated animals show reduced VC4 and VC5 5HT staining compared to mock ablated controls (Fig. 4 A and B and Dataset S4). We next generated C. angaria mod-5/Sert null animals [Cang-mod-5(vlc45)] (SI Appendix, Fig. S2A). We found a complete absence of 5HT labeling in VC4 and VC5 neurons in these mutants, which reinforced the view that these cells reuptake 5HT produced by HSN neurons (Fig. 4 A and B). Like in C. elegans, C. angaria VC4 and VC5 neurons stain for the UNC-17/VCHAT vesicular acetylcholine transporter, suggesting that they maintain their cholinergic identity while acquiring a serotonergic phenotype rather than undergoing a fate switch (SI Appendix, Fig. S2B). Dual neurotransmission is present in other C. elegans neuron types (14). Altogether, these experiments imply that VC4 and VC5 neurons in the Angaria group of species have acquired robust mod-5/Sert expression, which allows them to reuptake 5HT from nearby HSN neurons.

Fig. 4.

C. angaria VC4 and VC5 neurons are mod-5/Sert-dependent serotonin reuptakers. (A) Schematic depiction (Top) and representative micrographs (Bottom) of 5HT staining in C. angaria mock ablated animals, HSN ablated animals and C. angaria mod-5(vcl45) mutants. Loss of HSN or mod-5 gene abolish VC4 and VC5 5HT staining. (B) quantification of experiments in (A) (n = 17, 15, 80, 83). Fisher’s exact test was applied for statistics (P value < 0.001, <0.001). See Dataset S4 for source data.

Regulatory Regions of C. angaria mod-5/Sert Are Active in C. angaria and C. elegans VC4 and VC5 Neurons.

We next aimed to identify the molecular mechanisms driving the expression of mod-5/Sert in VC4 and VC5 neurons in Angaria group of species. Although we failed to generate stable transgenic lines in C. angaria, we were able to generate F1 transgenic animals. We found regulatory regions of C. angaria mod-5/Sert locus (Cang_mod-5 reporter) active in C. angaria VC4 and VC5 neurons, further supporting a cell-autonomous role for mod-5/Sert in the acquisition of the serotonergic phenotype (Fig. 5 A and C, SI Appendix, Fig. S3C, and Dataset S5). As expected from the lack of mod-5/Sert expression in C. elegans VC4 and VC5, none of the tested C. elegans_mod-5 regulatory regions are active in C. elegans in this neuron class (Fig. 5 A and B, SI Appendix, Fig. S3A, and Dataset S5).

Fig. 5.

cis regulatory changes underlie VC4 and VC5 serotonergic phenotype in Angaria species. (A) Schematics of cis regulatory modules used in the fluorescence reporter analysis. (B) C. elegans transgenic reporter animals are used to study mod-5/Sert regulatory region activity from different Caenorhabditis species: two non-angaria species C. elegans (Cel) and C. virilis (Cvir) and three Angaria species C. angaria (Cang), C. quiockensis (Cquio), and C. castelli (Ccast). mod-5/Sert CRMs from Angaria species but not from other species drive gfp expression in C. elegans VC4 and VC5 neurons. The graph represents the mean of at least two lines (n = 60, 156, 94, 94, 91). (C) C. angaria F1 transgenic reporter animals are used to study C. elegans and C. angaria mod-5/Sert regulatory region activity. C. angaria but not C. elegans reporter is active in C. angaria VC4 and VC5 neurons (n = 31, 28). (D) Schematic representation of mod-5/Sert locus in C. elegans and C. angaria. Cel mod-5(vlc47) corresponds to a CRISPR tagged T2A::NeonGreen allele, while Cel mod-5(syb7715) substitutes Cel intron1 sequence by Cang intron1. (E) Micrographs of the 5HT staining of (D) strains. (F) Quantification of (E) shows that intron 1 swap is sufficient to drive strong 5HT phenotype in C. elegans VC4 and VC5 neurons (n = 40, 40, 40).

Two alternative scenarios could explain mod-5/Sert expression in VC4 and VC5 neurons in C. angaria. First, changes in the trans-environment (e.g., differential expression or activity of TFs or epigenetic landscape in VC4 and VC5 neurons in C. angaria versus C. elegans) could be responsible for de novo activation of mod-5/Sert gene. Second, changes in the cis-regulatory regions of mod-5/Sert locus might lead to the emergence of a new enhancer that resulted in its activation by the ancestral VC4 and VC5 trans-environment. These possible scenarios make specific predictions for the outcome of heterologous gene reporter experiments. If trans-environment changes are mainly responsible for the phenotype, C. elegans regulatory region reporters (Cel_mod-5/Sert) would be activated in C. angaria VC4 and VC5 neurons while C. angaria reporter (Cang_mod-5/Sert) would be inactive in the context of C. elegans neurons. Conversely, if cis-regulatory changes are the main drivers of the phenotype, the Cang_mod-5/Sert reporter would be active in C. elegans VC4 and VC5 neurons and Cel_mod-5/Sert would not be active in C. angaria. Our heterologous reporter analyses show that Cang_mod-5/Sert fluorescent reporter is also active in C. elegans VC4 and VC5 neurons (Fig. 5B and Dataset S5), while the corresponding Cel_mod-5/Sert regulatory regions tested in C. angaria showed no detectable activity (Fig. 5C and Dataset S5). These results support a cis-environment model, as species origin of the regulatory regions dictated expression (Cang_mod-5/Sert ON; Cel_mod-5/Sert OFF) and not the species-specific VC4 and VC5 cellular context (trans-environment). Further supporting the cis-model, mod-5/Sert reporters from two additional Angaria species (C. castelli and C. quiockensis) show VC4 and VC5 activation in C. elegans transgenic animals, like the C. angaria construct, while an equivalent reporter from C. virilis, an Angaria outgroup species, does not (Fig. 5 A and B, SI Appendix, Fig. S3C, and Dataset S5). In summary, our results show that Angaria species share the presence of a CRM in the mod-5/Sert locus, which is constitutively activated in VC4 and VC5 neurons by TF(s) with expression shared between Angaria species and C. elegans.

C. angaria mod-5/Sert Enhancer Is Sufficient to Provide the Serotonergic Phenotype to C. elegans VC4 and VC5 Neurons.

We found that C. angaria mod-5/Sert CRM active in VC4 and VC5 neurons is located in the first intron of the locus (SI Appendix, Fig. S3C). In C. elegans, the homologous region (Cel_mod-5 intron1 reporter) is active in NSM and ADF but not VC4 and VC5, while the Cang_mod-5 intron1 reporter is active in NSM, VC4, and VC5 in C. elegans (SI Appendix, Fig. S3 A–C and Dataset S5). To test sufficiency of the C. angaria enhancer in its endogenous genomic context we modified the C. elegans mod-5::T2A::NeonGreen strain [mod-5(vlc47)] substituting the first intron by C. angaria intron 1 sequence [generating C. elegans mod-5(vlc47syb7715) referred to as Cel_mod-5^{C.ang_intron1}] (Fig. 5D). This intron swap is sufficient to drive the VC4 and VC5 robust serotonergic phenotype in C. elegans like in C. angaria (Fig. 5 E and F and Dataset S5), further supporting the cis-regulatory model. These results suggested that the ancestral Angaria species gained robust mod-5/Sert expression in VC4 and VC5 neurons through the emergence of an enhancer that overcomes H3K9me repressive activity in the cell, allowing its recruitment to the preexisting VC4 and VC5 gene regulatory network.

UNC-4/UNCX Directly Activates C. angaria mod-5/Sert Enhancer.

Next, we aimed to identify TFs responsible for activating the C. angaria mod-5/Sert enhancer. Since genetic manipulations in C. angaria are challenging and the Cang_mod-5 intron 1 enhancer is active in VC4 and VC5 neurons in both C. angaria and C. elegans, we hypothesized that the TFs responsible for its activation are conserved between both species. Based on this idea, we used C. elegans transgenics for the initial characterization of its regulatory logic. Consistent with our findings in C. elegans, where UNC-4/UNCX is necessary for C. elegans cat-1::gfp reporter expression in VC4 and VC5 neurons and for the serotonergic phenotype in H3K9me mutants (Fig. 2I, SI Appendix, Fig. S1 B–D), we observed that the Cang_mod-5::gfp fluorescence reporter is inactive in VC4 and VC5 neurons in C. elegans unc-4/Uncx(gk705) mutants, while reporter expression in NSM neurons is unaffected (Fig. 6 A and B and Dataset S6). Consistently, C. elegans mod-5^{C.ang_intron1} strain loses VC4 and VC5 5HT staining in unc-4(gk705) mutant background (Fig. 6 A and B). UNC-4/UNCX activation of C. angaria mod-5/Sert enhancer is likely direct because point mutations in predicted UNC-4/UNCX homeodomain binding sites (TAAT) abolished Cang_mod-5 reporter gene expression both in C. elegans and C. angaria transgenic animals (Fig. 6 C–F and Dataset S6). Among the three potential HD consensus binding sites found in the Cang_mod-5 minimal promoter, only HD3 matches the TAATT core for UNCX TF and shows the greatest effect when mutated (Fig. 6 C and D). These results suggest that cis-regulatory changes in mod-5/Sert intron 1 allowed the recruitment of this effector gene in Angaria species as a new robust direct target of UNC-4/UNCX, which provided a constitutive VC4 and VC5 serotonergic phenotype in these species. C. elegans mod-5/Sert intron 1 contains a higher number of TAATT sequences compared to C. angaria mod-5/Sert intron 1, thus we hypothesized additional motifs could be differentially present in Intron 1 of Angaria species compared to non-Angaria species that could be responsible for their constitutive activity in VC4 and VC5 neurons. Differential motif enrichment analysis failed to identify such motif/s. Alternatively, we directly investigated if UNCX motif from Angaria species could bear additional information. Using Cang_mod-5 HD3 site as reference, we could identify an “Angaria species extended motif” (SSTAATTYWT) (Fig. 6G). This motif is present in the Intron 1 sequence of 7 out of 7 analyzed Angaria species while only in 2 out of 40 non-Angaria Caenorhabditis species (Dataset S6). Motif prediction analysis from the extended motif reveals an overlapping POU HD site in addition to the UNCX motif (Fig. 6G). Three POU HD TFs are encoded in the C. elegans genome (unc-86, ceh-6, and ceh-18), none of them are expressed in mature VC4 and VC5, however, ceh-6 is expressed in C. elegans VC postmitotic immature neurons (32). ceh-6 TF is conserved in C. angaria, thus, both UNC-4 and CEH-6 could potentially cooperate in the activation of C. angaria mod-5 intron 1 enhancer.

Fig. 6.

UNC-4/UNCX directly activates C. angaria mod-5 enhancer. (A) Quantification of Cang_mod-5::gfp reporter expression and 5HT staining of Cel mod-5(vlc47syb7715) in wild type and unc-4(gk705) mutant background. UNC-4 is required for VC4 and VC5 serotonergic fate (n = 64, 76, 42, 44). (B) Micrographs of Cang_mod-5::gfp reporter expression and 5HT staining of Cel mod-5(vlc47syb7715) in wild type and unc-4(gk705) mutant background. (C) C. angaria mod-5 locus schema and location of the minimal CRM (mod-5prom9) driving expression in VC4 and VC5 neurons. mod-5prom9 contains three putative HD binding sites. Individual and combined HD point mutations were tested for its activity in C. elegans and C. angaria transgenic animals. (D) Quantification of mutated constructs in C. elegans reporter lines (n = 101, 99, 61, 65, 66). (E) Quantification of mutated constructs in C. angaria transgenic animals. (n = 28, 35). (F) Representative pictures of C. elegans VC4 and VC5 gfp expression of mod-5prom9 and HD3 mut. (G) Extended Angaria HD motif See Dataset S6 for source data and P values.

The VC4 and VC5 Serotonergic Phenotype in Angaria species Buffers Egg-Laying Induction by Environmental Serotonin.

Next, we aimed to identify the impact that the acquisition of the serotonergic phenotype by VC4 and VC5 neurons may have on behavior. In C. elegans, VC4, VC5, and HSN motoneurons are part of the egg-laying circuit (33). We found that, under laboratory conditions, the egg-laying behavior of C. elegans and C. angaria strains is significantly different (SI Appendix, Fig. S4 A–G). However, many factors could explain these differences and not just those related to the serotonergic phenotype of VC4 and VC5 neurons. Indeed, when we specifically tested the contribution of mod-5/Sert expression in VC4 and VC5, we found egg laying is similar in C. angaria between wild type and mod-5/Sert(vlc45) mutants while mod-5/Sert expression in C. elegans VC4 and VC5 leads to shorter egg laying window length that results in decreased number of laid eggs (SI Appendix, Fig. S4 B–G).

We reasoned that laboratory growth conditions, which vastly differ from those of the nematode’s natural habitat, could explain the absence of a differential phenotype. Caenorhabditis species have been isolated mostly from rotten fruits and flowers, where they are exposed to complex chemical and microbial environments (34). 5HT is found across all domains of life, including plants, bacteria, and fungi, and it is consistently implicated in interspecies crosstalk, including plant defense against insects (35–41). Considering that biotic interactions are often significant drivers of evolutionary change (42), we hypothesized that C. angaria may encounter habitats with high levels of environmental 5HT and wondered whether this feature might be related to the acquired VC4 and VC5 serotonergic phenotype (43). In the laboratory, C. elegans acute exposure to exogenous 5HT induces egg laying (25). Thus, to test the hypothesis that VC4 and VC5 neurons serotonergic phenotype may modulate behavior in high 5HT environments, we first compared egg-laying responses to acute treatment with exogenous 5HT between wild type C. elegans and C. elegans animals expressing mod-5/Sert in VC4 and VC5 neurons. Our results revealed that C. elegans egg-laying induction by exogenous 5HT is significantly attenuated by mod-5/Sert expression in VC4 and VC5 neurons (Fig. 7A and Dataset S7). Quantification of egg retention also shows increased number of eggs inside VC4/VC5 mod-5/sert expressing animals compared to controls (SI Appendix, Fig. S5A), discarding that the decrease in egg-laying induction by acute 5HT treatment is due to fertility defects in the transgenic animals. Conversely, C. angaria wild type animals are resistant to egg-laying induction by exogenous 5HT while C. angaria mod-5/Sert mutants lose this buffering capacity (Fig. 7A and Dataset S7). An extended analysis of egg-laying responses to exogenous 5HT in different Caenorhabditis species shows a strong anticorrelation between VC4 and VC5 5HT staining and egg-laying induction by exogenous 5HT (SI Appendix, Fig. S5B). Altogether, our results suggested that the ability of VC4 and VC5 neurons to reuptake 5HT can prevent egg-laying induction by exogenous 5HT.

Fig. 7.

The VC4 and VC5 serotonergic phenotype in Angaria species buffers egg-laying induction by exogenous serotonin. (A) Quantification of egg laying events upon 2 h exposure to external 5HT. Both, C. elegans control animals (N2) and transgenic animals with mod-5/Sert cDNA expression in VC neurons induce egg laying upon 5HT exposure (n = 43, 44, 43, 43) (P value < 0.001, <0.001), however, egg laying induction in mod-5/Sert transgenic animals is significantly lower than controls (P value < 0.001). C. angaria control animals (RGD1) do not increase egg laying when exposed to external 5HT while C. angaria mod-5 mutant animals show significant egg laying induction (n = 44, 44, 43, 43) (P value 0.329, <0.001). Mean and SEM are represented. Nonparametric Mann–Whitney test. (B) Schematic depiction (Top) and representative micrographs (Bottom) of VC4 and VC5 neuron 5HT staining in Cel mod-5(vlc47syb7715) animals, Cel mod-5(vlc47syb7715); bas-1(ad446) animals that lack 5HT biosynthesis and Cel mod-5(vlc47syb7715); bas-1(ad446) animals exposed to high levels of exogenous 5HT. (C) Quantification of (B) shows VC4 and VC5 neurons can uptake 5HT from exogenous sources (n = 40, 50, 51). Fisher’s exact test (P value < 0.001). See Dataset S6 for source data.

mod-5/Sert Expression in VC4 and VC5 Neurons in C. elegans Allows for Environmental 5HT Reuptake.

In C. elegans 5HT released from HSN neurons stimulates egg laying by directly modulating the functional state of the vulval muscles (44–47). However, how environmental 5HT induces acute egg laying in C. elegans is not known. We hypothesize that exogenous 5HT could similarly induce vulval muscle contraction and egg-laying behavior by reaching vulval muscle serotonergic synapses either by diffusion through the vulva slit or ingestion. We reasoned that in Angaria species, VC4 and VC5 neurons could not only reuptake 5HT from HSN synapses but also, if present, from external sources, thereby preventing excessive activation of neuromuscular synapses. To test this hypothesis, we took advantage of C. elegans mod-5^{C.ang_intron1} animals, which show intense VC4 and VC5 5HT staining under laboratory conditions, like C. angaria, and generated a double mutant with the bas-1/Aadc(ad446) allele. bas-1/Aadc codes for the amino acid decarboxylase enzyme, fundamental for endogenous 5HT biosynthesis (48). As expected, under standard laboratory conditions C. elegans mod-5^{C.ang_intron1}; bas-1/Aadc(ad446) animals show a complete loss of 5HT staining, including VC4 and VC5 neurons. However, these animals recover strong VC4 and VC5 5HT staining when exposed to exogenous 5HT (Fig. 7 B and C and Dataset S7), demonstrating that these neurons can uptake environmental 5HT. Together with our previous results, these observations suggest that the acquisition of mod-5/Sert expression in VC4 and VC5 neurons by the Angaria group of species had modified their egg-laying behavior, allowing them to buffer egg-laying induction produced by exogenous 5HT.

Environmental Serotonin Induces VC4 and VC5 Serotonergic Phenotype in C. elegans.

C. elegans nervous system can adapt to different environmental stimuli through neuron-type specific changes in gene expression (16). Considering the active epigenetic repression of mod-5/Sert locus in VC4 and VC5 neurons, we wondered if the serotonergic phenotype in C. elegans could represent a plastic trait and aimed to identify potential environmental triggers. As we determined that this trait in Angaria species helps prevent egg-laying induction by exogenous 5HT, we wondered if this environment could precisely trigger the VC4 and VC5 serotonergic phenotype in C. elegans. We found that C. elegans wild type animals grown for one generation in high levels of environmental 5HT show a significant increase in VC4 and VC5 serotonergic staining, both in categorical ON/OFF analysis and quantitative 5HT intensity measurements (Fig. 8 A–C, SI Appendix, Fig. S6 A and B and Dataset S8). Induction of C. elegans VC4 and VC5 5HT staining requires UNC-4/UNCX and MOD-5/SERT, like in C. elegans H3K9me mutants and wild type C. angaria (Fig. 8 B and C and SI Appendix, Fig. S6B). To assess how broadly distributed in the Caenorhabditis genus this plastic trait is, we exposed two additional species, C. briggsae and C. tropicalis, to high 5HT environments for one generation. We failed to detect VC4 and VC5 5HT staining in these animals, suggesting VC4 and VC5 plasticity to environmental 5HT is not a general feature of the Caenorhabditis genus (SI Appendix, Fig. S6 C and D). In addition, in contrast to C. elegans, constitutive cat-1/Vmat expression in VC4 and VC5 neurons is not detected in C. briggsae and C. tropicalis (49), consistent with the lack of serotonergic phenotype plasticity.

Fig. 8.

C. elegans VC4 and VC5 serotonergic phenotype is a plastic trait induced by environmental 5HT exposure that leads to modified egg laying responses. (A) Schematic representation of the experimental set up, worms are grown for one generation in control conditions or exposed to high levels of environmental 5HT. (B) Representative micrographs of VC4 and VC5 serotonergic staining in wild type, mod-5(knu383), or unc-4(gk705) animals under control conditions or grown for one generation in environmental 5HT. (C) Quantification of VC4 and VC5 serotonergic intensity phenotype. 5HT staining induction by environmental 5HT is significantly reduced in two different alleles of mod-5/Sert transporter and unc-4(gk705) mutants (n = 89, 72, 54, 50, 56 neurons). Mean and SEM are represented. Kruskal–Wallis nonparametric test with Dunn’s correction for multiple comparisons (P value < 0.001, <0.001, <0.001). (D) Schematic representation of the experimental set up, worms grown for one generation in control conditions or environmental 5HT are tested for acute egg laying responses (2 h) to exogenous 5HT. (E) Quantification of laid eggs. Wild type worms grown for one generation exposed to environmental 5HT (green) are able to buffer the egg laying induction upon acute 5HT exposure compared to controls (orange). This buffering effect is absent in mod-5(knu383) and reduced in unc-4(gk705) mutant animals (n = 72, 72, 48, 48, 47, 41); (P value < 0.001, 0.74, 0.015). Mean and SEM are represented. Mann–Whitney test. CTCF: Corrected Cell Total Fluorescence. See Dataset S7 for source data.

Finally, we tested the behavioral consequences of this form of plasticity in C. elegans. We found that C. elegans animals grown in high 5HT environments modified their egg-laying responses to acute 5HT treatment in a way analogous to the constitutive behavior observed in Angaria species, i.e., they buffer egg-laying induction by exogenous 5HT (Fig. 8 D and E and Dataset S8). Importantly, this attenuation response is absent in mod-5/Sert mutants and significantly reduced in unc-4/Uncx mutant animals grown under similar 5HT exposure conditions, further supporting the requirement of mod-5/Sert expression in VC4 and VC5 neurons to mediate this form of behavioral plasticity. In summary, prolonged exposure to 5HT (high 5HT environments) adapts C. elegans behavior to be able to buffer egg laying induction by acute exogenous 5HT, which require MOD-5/Sert function.

Discussion

During development, robust cell type differentiation needs to be in balance with the ability to drive adaptive changes in gene expression. What mechanisms underlie such trade-off between robustness and plasticity and how they impact the evolution of cell types is not well understood. We have found that in C. elegans, robust specification of VC4 and VC5 cholinergic neuron types requires H3K9me epigenetic signaling to suppress a 5HT reuptaking phenotype. H3K9me-mediated repression of undesired effector genes in the differentiation of both neuronal and non-neuronal tissues has been reported in the past (17). Here, we show that this process can also be plastic, as epigenetic repression of VC4 and VC5 5HT reuptake phenotype can be released when worms develop under particular environmental conditions that promote an alternative, yet highly robust, serotonergic phenotype. These findings demonstrate a mechanism for neuronal plasticity that is regulated through H3K9me epigenetic repression, akin to other plastic responses described in the immune system (50–52). The mechanisms that integrate environmental 5HT, H3K9me regulation and mod-5/Sert expression are currently unknown. Moreover, at this point, we cannot discard that in wild type C. elegans, low levels of mod-5/Sert expression in VC4 and VC5 are still present and provide 5HT staining when grown in high 5HT environments. Future research will address these questions.

Our results in C. elegans prompted us to systematically investigate the distribution of the VC4 and VC5 serotonergic phenotype in the Caenorhabditis genus. Under laboratory conditions, we found that the Angaria group of species share strong and constitutive 5HT staining in VC4 and VC5 neurons. This robust phenotype is provided by the emergence of an enhancer in the mod-5/Sert locus, which is active in VC4 and VC5 neurons, allowing for the robust 5HT reuptake phenotype. When placed in the C. elegans genomic context, this enhancer is sufficient to overcome H3K9me repression in VC4 and VC5 neuron development without the requirement of any environmental trigger and it is recruited as a direct target of the neuron-type gene regulatory network controlled by UNC-4/UNCX TF. Our motif analysis shows a predicted hybrid UNCX/POU consensus site present in Angaria species, in addition, CEH-6/POU TF is expressed in VC neurons at early larval stages (32), thus, we hypothesize CEH-6 might cooperate with UNC-4 to prevent H3K9me and activate Angaria mod-5 intron 1 enhancer. The absence of this consensus site in C. elegans might lead to H3K9me in the mod-5 locus and its repression. Furthermore, affinity of UNC-4/UNCX might be higher for the Angaria consensus than for other TAATT binding sites present in non-Angaria species. In conclusion, our results suggest that the addition of this cis regulatory module is sufficient to produce a shift from plastic to a constitutive trait. Our results provide an example of the evolution of neuron-type specific gene expression and neuronal function and describe its underlying gene regulatory mechanism, giving experimental support to previous models (1, 2).

Moreover, we identify behavioral differences that can be directly attributed to this new trait: the ability to block egg-laying induction by external sources of 5HT. We speculate that this neuronal function might be adaptative as tight temporal control of egg-laying and egg retention in harsh environments is important for embryonic viability and population competition (53). We propose that evolution of this constitutive trait in Angaria species could be driven, at least in part, by exposure to high levels of environmental 5HT found in some plants and fruits, which can be further increased as defensive mechanism upon insect feeding, or fungal infections (35–41). Indeed, some Angaria species have been isolated from fruits with high 5HT content, such as tomatoes (C. sp.8) or associated with the weevil pests of bananas, which are also high in 5HT (43). The specificity of the constitutive trait for Angaria species is intriguing as they share some natural habitats with other Caenorhabditis species (54). It is also known that the sensory behavior of C. elegans can be modulated by bacterially synthesized neurotransmitters (55), We speculate that the plastic responses to exogenous 5HT we found for C. elegans might be related to its generalist lifestyle and the fact that it is frequently found in slug’s intestines, where it can survive intestinal transit (56). This particular behavior has been suggested to be a dispersal mechanism. The intestine is the major source of 5HT in the body, including in snails and slugs (57, 58). Considering that phenotypic plasticity is favored in changing environments where predictive cues are present, it is tempting to speculate C. elegans VC4 and VC5 serotonergic plasticity might be key to induce egg retention when present in intestinal environments, which will favor offspring survival, and egg deposition in external habitats, promoting population dispersal and colonization of new food resources. The gain and subsequent loss of plasticity are proposed to have a role in the emergence of phenotypic novelty, a concept known as genetic assimilation (59). The lack of a plastic response to exogenous 5HT in C. briggsae and C. tropicalis suggests that C. elegans might have acquired this particular facultative feature. We cannot assess if a plastic state preceded constitutive VC4 and VC5 5HT staining in Angaria species. Nevertheless, constitutive serotonergic staining in VC4 and VC5 neurons is present in several non-Caenorhabditis nematodes of diverse origin (31), suggesting this phenotype has been gained or lost repeatedly, further supporting its evolvability. Importantly, we provide specific genetic mechanisms underlying both, trait plasticity (facultative epigenetic silencing) and canalization (enhancer emergence refractory to epigenetic silencing). In addition, our results are also in agreement with the genetic assimilation model as a mechanism for evolutionary innovations.

Finally, our results also bear biomedical implications. Sert genetic variants have been linked to pathological conditions such as depression, anxiety, or autism (60, 61). Importantly, there is also a key interplay between Sert genetic variants and environmental modulation (vulnerability or protection to disease) (62). The mechanisms directing dynamic MOD-5/SERT expression in cortical neurons are unknown, as well as whether environmental stimuli can modulate its developmental repression under physiological or pathological conditions. Indeed, mouse retinal ganglion cells, which transiently express SERT in development, reactivate its expression after optic nerve injury, increasing the vulnerability of these neurons (63). Importantly, the H3K9me pathway has also been genetically linked to psychiatric disorders (64, 65). In light of our results, we speculate that plastic H3K9me epigenetic silencing of Sert locus might be phylogenetically conserved in mammals, could potentially be environmentally modulated and involved in 5HT related neuropsychiatric disorders.

Materials and Methods

N2 (Bristol strain) was used as C. elegans wild type strain. All strains and Caenorhabditis species were obtained from the Caenorhabditis Genetics Center (CGC), except for mod-5(knu383) obtained from Knudra Transgenics and C. sp. 24 and C. sp. 39, which were kindly provided by Karin Kiontke at New York University. Primers are included in Dataset S10. A list of all strains is included in Dataset S11. See SI Appendix, Supplementary Material and Methods for details on immunohistochemistry, HSN laser ablations, 5HT treatments, bioinformatics analysis, generation of mutants and transgenics, egg-laying behavior analysis, fluorescence measurements and statistical analysis.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Dataset S05 (XLSX)

Dataset S06 (XLSX)

Dataset S07 (XLSX)

Dataset S08 (XLSX)

Dataset S09 (XLSX)

Dataset S10 (XLSX)

Dataset S11 (XLSX)

Data, Materials, and Software Availability

All data produced in this study included in the article and/or supporting information.