Temperature regulates synaptic subcellular specificity mediated by inhibitory glutamate signaling

Summary

Environmental factors such as temperature affect neuronal activity and development. However, it remains unknown whether and how they affect synaptic subcellular specificity. Here, using the nematode Caenorhabditis elegans AIY interneurons as a model, we found that high cultivation temperature robustly induces defects in synaptic subcellular specificity through glutamatergic neurotransmission. Furthermore, we determined that the functional glutamate is mainly released by the ASH sensory neurons and sensed by two conserved inhibitory glutamate-gated chloride channels GLC-3 and GLC-4 in AIY. Our work not only presents a novel neurotransmission-dependent mechanism underlying the synaptic subcellular specificity, but also provides a potential mechanistic insight into high-temperature-induced neurological defects.

Keywords:

Genetically modified animals – Glutamate – Chlorides – Interneurons – Neurons – Synapses – Test statistics

Introduction

Normal brain functions require precise synaptic connectivity among billions of neuronal and non-neuronal cells. Synaptic targeting happens not only at the cellular, but also at the subcellular level [1–3]. For example, in mouse cerebellum, basket neurons specifically form synapses at the axon initial segment of purkinje neurons [4]. Similarly, C. elegans specific AIY presynaptic region targets onto the RIA interneurons [5,6]. In the last couple of decades, studies have identified many genetic factors required for synaptic subcellular specificity, including secreted and adhesion molecules [4,6–16]. Additionally, synaptic development is also regulated by neural activity [17–19]. However, it is largely unknown whether environmental-dependent neuronal activity is involved in the synaptic subcellular specificity.

Temperature is a special environmental factor that can affect neuronal development and functions through activity-dependent manner [20–25]. Neuronal activity plays critical roles in neural circuitry development [18,19]. In vertebrates, neuronal activity is essential for synapse formation in the visual system [26–28]. In invertebrates, neural circuitry was traditionally thought to be hardwired and regulated by activity-independent mechanisms [29–34]. However, recent studies show that neural activity is involved in the circuit development and remodeling in Drosophila [35–38]. Similarly, in C. elegans, neuronal activity can modulate neurite growth and branching [39–42], cell fate determination [43], presynaptic remodeling and dendritic spine density [44,45]. However, it is unknown whether and how temperature or neuronal activity affects the synaptic subcellular specificity.

The nematode C. elegans AIY interneurons are part of the thermotaxis circuit [46–51]. In this circuit, sensory neurons such as AFD and AWC sense the thermal information and transmit it to the AIY interneurons through glutamatergic synapses [46,51–54]. The information is further passed from AIY to the next layer interneurons including RIA and AIZ [46,51]. Although the thermotaxis circuit is known for a long time, the detailed circuit connectivity is not completely understood, and the regulatory mechanisms underlying the circuit formation are largely unknown.

AIY forms stereotypic presynaptic distribution [5,6]. With this system, we previously found that the epithelial CIMA-1, a sialic acid transmembrane transporter, is required for maintaining the subcellular specificity of the AIY interneurons. In cima-1 loss-of-function mutants, ectopic synapses emerge in the AIY asynaptic region partially due to the posterior displacement of ventral cephalic sheath cells (VCSC) glial endfeet [55]. However, ablating the VCSC glia did not completely suppress the cima-1 ectopic synapses, suggesting that additional signals, most likely from the nervous system, are involved [55].

In this study, we showed that the AIY ectopic synaptic formation in cima-1 loss-of-function mutants requires the inhibitory glutamate signaling, which is mediated by the ASH expressed vesicular glutamate transporter EAT-4 and the AIY expressed pLGIC family glutamate gated chloride channels GLC-3 and GLC-4. Additionally, we showed that wild-type animals cultivated at high temperature display ectopic AIY presynaptic phenotype mimicking the cima-1 mutants. The glutamate transporter EAT-4 in ASH and the glutamate gated chloride channels GLC-3 and GLC-4 in AIY are required for both cima-1 and high-temperature-induced ectopic synapse formation in AIY neurons. Our study not only uncovers a novel role of the glutamatergic transmission in synaptic subcellular specificity, but also provides potential pathological insights into the high temperature-induced neurodevelopmental defects.

Results

Glutamatergic neurotransmission regulates the AIY presynaptic subcellular specificity

The C. elegans AIY neurons are a pair of bilaterally symmetric neurons in the head with stereotypical synaptic distribution: the ventral asynaptic zone 1 region, the synaptic-enriched zone 2 region, and the distal synaptic-scattered zone 3 region [5,6] (Fig 1A). The sialin homolog CIMA-1 in epidermal cells and the ADAMTS metalloprotease MIG-17 in muscles are required to maintain AIY presynaptic subcellular specificity mediated by the VCSC glia morphology during adult stage [55,56]. Incomplete suppression of the cima-1(wy84) ectopic synapses by VCSC glia ablation implies that neuronal signaling is involved in the synaptic subcellular specificity (see the model in Fig 1A and [55]).

Neuronal activity plays important roles in circuit formation [18,19]. To determine if the ectopic AIY presynaptic phenotype requires neuronal activity, we used synaptic transmission defects unc-13(e1091) mutants [57]. We found that unc-13(e1091) mutants displayed normal AIY presynaptic distribution (S1A–S1C and S1J Fig), consistent with previous findings that synaptic transmission is not required for normal synaptic formation [31,33,34].

Next, we asked if neurotransmission was required for the ectopic synaptic formation in cima-1(wy84) mutants. To address the question, we made cima-1(wy84);unc-13(e1091) double mutants, and found that unc-13(e1091) robustly suppressed the ectopic synapses in cima-1(wy84) mutants (90.19% of animals displayed ectopic synapses in cima-1(wy84) vs 23.38% in cima-1(wy84);unc-13(e1091) mutants, p<0.0001, Fig 1C–1E and 1O). Those data indicate that neurotransmission is required for the ectopic synapse formation in the cima-1(wy84) mutants.

To determine which type of neurotransmission is required, we blocked the glutamatergic, GABAergic, cholinergic, or dopaminergic neurotransmission via the following loss-of-function mutants: eat-4, unc-47, unc-17 and cat-2, which encode the vesicle glutamate transporter, vesicle gamma-aminobutyric acid γ (GABA) transporter, acetylcholine transmembrane transporter and dopamine biosynthetic enzyme respectively [58–61]. Consistent with that seen in unc-13(e1091) mutants, the AIY synaptic distribution was normal in those single mutants (S1C–S1J Fig), suggesting that those types of neurotransmission are not required for synaptic spatial specificity per se. Then, we tested their roles in the ectopic synaptic formation in cima-1(wy84) mutants. We found that the ectopic synapses were robustly suppressed only by eat-4(ky5) as assayed with both synaptic vesicle marker GFP::RAB-3 (90.19% of animals with ectopic synapses in cima-1(wy84) and 20.98% in cima-1(wy84);eat-4(ky5), p<0.0001. Fig 1F and 1O), and the synaptic active zone marker GFP::SYD-1 (Fig 1L–1O), but not by unc-47(n2409), unc-17(cn355), or cat-2(e1112) mutations (S2A–S2E Fig). The effect of eat-4(ky5) on suppressing the cima-1(wy84) ectopic synapses was validated with two additional loss-of-function eat-4(nj2) and eat-4(nj6) alleles [51] (Fig 1G, 1H and 1O). To exclude the possibility that the suppression of the cima-1 ectopic synapses is wy84 allele-specific, we tested another independently isolated cima-1 allele gk902655 that harbors a nonsense mutation at the R476 site [55,62]. Consistent with the cima-1(wy84) data, we found that both unc-13(e1091) and eat-4(ky5) suppressed the AIY presynaptic specificity defects induced by cima-1(gk902655) (72.65%, 9.37% and 12.11% of animals displayed ectopic synapses in cima-1(gk902655), cima-1(gk902655);unc-13(e1091) and cima-1(gk902655);eat-4(ky5) respectively, p<0.0001 for both comparison, Fig 1I–1K and 1O). These data suggest that the suppression of the ectopic synapses is not wy84 allele specific.

To further confirm the requirement of eat-4 for the AIY synaptic phenotype in cima-1 mutants, we quantified the expressivity of the ectopic synapses by measuring the ventral synaptic length and the ratio of the ventral to total synaptic length. In the cima-1(wy84) mutants, the ventral synaptic length and the ratio of the ventral to total synaptic length increased dramatically due to the formation of ectopic synapses (the length and the ratio are 8.65μm and 0.21 in WT; vs 16.68μm and 0.33 in the cima-1(wy84) mutants, P<0.0001. Fig 1M and 1N). Consistent with the penetrance data described above, both the ventral synaptic length and the ratio in cima-1(wy84) mutants were significantly suppressed by eat-4(ky5) (the length and the ratio are 16.68μm and 0.33 in cima-1(wy84); vs 9.38μm and 0.23 in the cima-1(wy84);eat-4(ky5) double mutants, P<0.0001. Fig 1P and 1Q).

Collectively, these data indicate that glutamatergic neurotransmission is required for the ectopic synaptic formation in cima-1(wy84) mutants.

eat-4 acts in the ASH neurons to regulate the AIY synaptic subcellular specificity

To understand where eat-4 acts to regulate the AIY synaptic subcellular specificity, we performed tissue-specific rescue by expressing eat-4 cDNA in different tissues or cell types. We found that eat-4 completely rescued and restored the ectopic synapses in cima-1(wy84);eat-4(ky5) double mutants when expressed in the nervous system with rab-3 promoter [63], or in the glutamatergic neurons with eat-4a promoter (S3A Fig, [64]), but not in the VCSC glia, epidermis, muscle, intestine or AIY interneurons with hlh-17, dpy-7, myo-3, ges-1 and ttx-3 promoters respectively [65–69] (Fig 2A and 2B). The data further support the hypothesis that glutamatergic neurotransmission is required for the ectopic synapse formation in cima-1(wy84) mutants.

To further determine the specific glutamatergic neuron(s) involved in the AIY ectopic synaptic formation in cima-1(wy84) mutants, we expressed eat-4 cDNA in the glutamatergic neurons previously identified as AIY synaptic partners including the presynaptic AUA (Pflp-8)[70], ASE (Pgcy-5)[71], AFD (Pgcy-8)[71], AWC (Pstr-2)[72], ASG and BAG (Peat-4b: 4454bp to 3554bp upstream regulatory sequence) [64] and the postsynaptic RIA (Pglr-3) neurons [5,73]. To our surprise, none of them rescued (Fig 2A and 2B). Then, we expressed eat-4 in twelve pairs of sensory neurons, including AWC, ASG, ASH, ASK, ADL, PHA and PHB seven pairs of glutamatergic neurons with odr-4 promoter [74]. Interestingly, this transgene fully rescued (Fig 2A and 2B). Finally, we tested the rescue in ASH, ASK or ADL with nhr-79 (or sra-6), sra-9 and srh-220 promoter respectively [75–78], but not in others because AWC and ASG were excluded previously and PHA and PHB are located in the tail, far away from AIY neurons. Interestingly, robust rescue was observed when eat-4 was expressed in the ASH, and to a less degree in ASK, but not in ADL neurons (Fig 2A and 2B). The data suggest that eat-4 acts mainly in the ASH to promote the AIY ectopic synapse formation in cima-1(wy84) mutants.

VGLUT overexpression leads to increasing glutamate loading in the synaptic vesicle and enhancing glutamate release in Drosophila [79,80] and vertebrates [81,82]. To determine whether overexpression of EAT-4/VGLUT is sufficient to induce the AIY ectopic synapses, we overexpressed the eat-4 cDNA in the ASH with different promoters, and found that they all robustly induced the ectopic synapses (Fig 2C). The data suggest that eat-4(OE) in the ASH is sufficient to induce the AIY ectopic synapse formation.

To further confirm the role of ASH neurons in regulating AIY synaptic specificity, we ablated the ASH neurons by expressing apoptotic protein caspase-3 [83]. We observed that ASH ablation partially but significantly suppressed the ectopic synapses induced by cima-1(wy84) (89.91% and 53.94% of animals with ectopic synapses in ASH-normal and -ablated animals respectively, p<0.0001. Fig 2D), and completely abolished the ectopic synapses induced by eat-4 overexpression in the ASH (43.97% and 14.09% of animals with ectopic synapses in ASH-normal and -ablated animals respectively, p<0.0001. Fig 2D). Those data further support that the glutamate required for the AIY ectopic synaptic formation is mainly from the ASH sensory neurons.

CIMA-1 regulates the AIY synaptic position mediated partially through VCSC glia [55,56]. To address if the VCSC glia is required for the glutamatergic signaling induced ectopic synapse formation, we ablated the VCSC glia in wild-type, cima-1(wy84) and eat-4(OE) animals. In wild-type animals, loss of the glia did not affect synaptic distribution (18.31% and 15.75% of total animals with ectopic synapses in wild type and glia-ablated animals respectively, p = 0.4788. S4A Fig). In cima-1(wy84) mutants, glia ablation partially suppressed the ectopic synaptic distribution (93.11% and 66.08% of total animals with ectopic synapses in glia-normal and -ablated animals respectively, p = 0.0023. S4A Fig), which is consistent with previous studies [55]. Interestingly, in eat-4(OE) (Peat-4a::EAT-4 transgenic) animals, glia ablation only slightly suppressed the ectopic synapses (62.19% and 53.37% of total animals with ectopic synapses in glia normal and ablated animals, p = 0.0047. S4A Fig). The data indicated that VCSC glia only contribute a little to the synaptic defect induced by eat-4(OE). In other words, eat-4(OE) promotes the AIY ectopic synaptic formation largely in a glia-independent manner.

To address when eat-4 acts, we quantified the AIY synaptic distribution at different developmental stages in eat-4(OE) animals. Interestingly, the ectopic synapses in eat-4(OE) emerged since the larval L1 stage (S5A–S5I Fig), unlike in cima-1(wy84) mutants which shows up only at adult stage [55]. Consistently, we found that the ventral synaptic length and the ratio of ventral to total synaptic length were significantly increased since the L1 stage (S5H–S5I Fig). Furthermore, the eat-4 embryonic expression supports its early role in AIY synaptic subcellular specificity (S3B and S3B’ Fig). These data collectively indicate that eat-4(OE) and cima-1(wy84) may promote the AIY ectopic synaptic formation through different mechanisms.

The synapses in zone 2 of wild-type animals are formed primarily onto the postsynaptic partner RIA [5]. To determine whether the ectopic synapses in eat-4(OE) are targeted to RIA, we simultaneously labeled RIA neurons and the AIY presynaptic sites, and found that the AIY ectopic presynaptic sites were only partially in apposition to the RIA neurons (S5J and S5K Fig), suggesting that some of the AIY ectopic synapses are not targeting onto RIA.

Glutamate-gated chloride channels GLC-3 and GLC-4 mediate the ectopic synapse formation

To address which glutamate receptor(s) is required, we analyzed all four types of glutamate receptors that have been identified in C. elegans including AMPA receptors, NMDA receptors, metabotropic G-protein-coupled receptors and glutamate-gated chloride channels (GluCls) (Fig 3A) [84–86]. Among eighteen loss-of-function receptors we tested, all of them displayed the normal AIY synaptic subcellular distribution (S6A–S6R Fig), suggesting that those receptors are not required for the AIY presynaptic subcellular specificity per se, which is consistent with the eat-4 loss-of-function phenotype seen above.

Then, we tested the roles of those receptors in suppressing cima-1(wy84) mutant phenotype. Interestingly, two glutamate-gated chloride channel mutants, glc-3(ok321) and glc-4(ok212) partially but significantly suppressed the cima-1(wy84) ectopic synapses formation as assayed with the synaptic vesicle marker GFP::RAB-3 (89.91% of animals with ectopic synapses in cima-1(wy84); 53.91% in cima-1(wy84);glc-3(ok321); 67.57% in cima-1(wy84);glc-4(ok212), p<0.0001 and p = 0.0029 as compared to cima-1(wy84) respectively. S7A–S7U Fig), while the rest mutant receptors did not. And glc-3(ok321);glc-4(ok212) double mutations completely suppressed the ectopic synapses in cima-1(wy84) mutations (89.84% of animals with ectopic synapses in cima-1(wy84), 19.13% in cima-1(wy84);glc-3(ok321);glc-4(ok212), p<0.0001. Fig 3B, 3C and 3F). The suppression effect by glc-3(ok321);glc-4(ok212) was confirmed with the active zone marker GFP::SYD-1 (89.66% of animals with ectopic synapses in cima-1(wy84); 28.76% in cima-1(wy84);glc-3(ok321);glc-4(ok212), p<0.0001. Fig 3D–3F). Consistently, both the ventral synaptic length and the ratio of ventral to total synaptic length in cima-1(wy84) mutants were robustly suppressed by glc-3(ok321);glc-4(ok212) double mutations (the length and the ratio are 16.76μm and 0.34 in cima-1(wy84); vs 9.77μm and 0.22 in cima-1(wy84);glc-3(ok321);glc-4(ok212) mutants, P<0.0001 as compared to cima-1(wy84). Fig 3G and 3H). The role of glc-3(ok321) and glc-4(ok212) in suppressing cima-1 was confirmed by cima-1(gk902655) allele (S7V–S7Z Fig). Together, the data suggest that the ectopic synapse formation in cima-1 mutants requires the glutamate-gated chloride channels GLC-3 and GLC-4.

Next, we tested whether the ectopic synapses induced by eat-4(OE) also requires GLC-3 and GLC-4. We found that either glc-3(ok321) or glc-4(ok212) partially suppressed the eat-4(OE)-induced ectopic synapses (18.07% of animals with ectopic synapses in wild type; 66.46% in eat-4(OE), p<0.0001 as compared to wild type; 45.98% in eat-4(OE);glc-3(ok321), p = 0.0061 as compared to eat-4(OE); 44.38% in eat-4(OE);glc-4(ok212), p = 0.0032 as compared to eat-4(OE). Fig 3I–3L and 3N). Notably, glc-3(ok321);glc-4(ok212) double mutations completely suppressed the ectopic synaptic formation induced by eat-4(OE) (22.95% of animals with ectopic synapses in eat-4(OE);glc-3(ok321);glc-4(ok212), p<0.0001 as compared to eat-4(OE). Fig 3J, 3M and 3N). These data collectively suggest that eat-4(OE) promotes the AIY ectopic synaptic formation through the glutamate-gated chloride channels GLC-3 and GLC-4.

GLC-3 and GLC-4 act cell-autonomously in AIY to promote the ectopic synaptic formation

To understand where GLC-3 and GLC-4 act to promote the AIY ectopic synapse formation, we firstly determined where they were expressed by generating transcriptional reporter Pglc-3::GFP and Pglc-4::GFP, co-labeled with the AIY reporter Pttx-3::mCherry [69]. We found that both Pglc-3::GFP and Pglc-4::GFP were expressed in head neurons including the AIY (Fig 4A–4B’’). We also noticed that both glc-3 and glc-4 were expressed since the embryo stage (S3C-S3D’ Fig), which is consistent with their role in mediating the ectopic synapse formation of eat-4(OE) animals at the L1 stage (S5A–S5I Fig). Next, we performed cell-specific rescue by driving glc-3 or glc-4 cDNA with AIY specific (ttx-3) promoter [69], with endogenous promoters as controls. We found that expressing glc-3 or glc-4 with AIY specific ttx-3 promoter rescued the corresponding mutants to the degree as with the endogenous promoters (Fig 4C). Additionally, we found that overexpressing glc-3 and glc-4 simultaneously in the AIY of wild-type animals induced the ectopic synapses in eat-4-dependent manner (Fig 4D–4F and 4H). Those data reveal that two glutamate-gated chloride channels GLC-3 and GLC-4 act cell-autonomously in AIY to modulate the synaptic subcellular specificity.

Given that GLC-3 and GLC-4 mediate inhibitory neurotransmission [51,87,88], we speculated that they induced the ectopic synapses through inhibiting AIY activity. To test this possibility, we expressed the gain-of-function potassium channel UNC-103(A334T) in AIY neurons. The gain of function UNC-103(A334T) can inhibit neuron excitability [89–92]. Indeed, the AIY-specific unc-103(gf) expression resulted in the AIY ectopic presynaptic formation in the zone 1 (Fig 4G and 4H), supporting the model that inhibiting the AIY activity is sufficient to induce the ectopic presynaptic assembly.

To directly examine if ASH-specific eat-4(OE) affects AIY activity, we recorded the calcium signaling in AIY with GCaMP6s[93]. We found while the amplitude of the automatic calcium oscillation was not affected, the frequency was dramatically reduced (Fig 4I–4K and S1 Video). These results support the model that the glutamate transmission from ASH promotes the AIY ectopic synaptic assembly through inhibiting its activity.

To further understand how GLC-3 and GLC-4 regulate AIY synaptic specificity, we determined GLC-3 and GLC-4 localization in AIY with AIY-specific mCherry::GLC-3 and mCherry::GLC-4 reporters. Interestingly, both GLC-3 and GLC-4 clusters largely overlapped with the synaptic marker GFP::RAB-3 in the zone 2 in wild-type or eat-4(ky5) mutants, and they were not present at the zone 1 region (Fig 5A–5B’ and 5F–5G’). In cima-1(wy84) or the ASH-specific eat-4(OE) animals, however, the GLC-3 and GLC-4 were also ectopically colocalized with the GFP::RAB-3 in the zone 1 region (Fig 5C, 5C’, 5E, 5E’, 5H, 5H’, 5J and 5J’). Loss of eat-4 suppressed the cima-1(wy84)-induced ectopic distribution of GLC-3 and GLC-4 as well as GFP::RAB-3 in the zone 1 (Fig 5D, 5D’, 5I and 5I’), suggesting that GLC-3 and GLC-4 probably act locally to promote presynaptic assembly.

ASH neurons are AIY presynaptic partners

Our above results demonstrate that overexpressing eat-4 specifically in ASH neurons promotes the AIY ectopic presynaptic formation through inhibiting its activity mediated by GLC-3/GLC-4 receptors. Those data implied that the ASH neurons most likely form synapses onto AIY, which was not reported previously [5,94]. To test this hypothesis, we examined electron microscopy (EM) reconstructions of three hermaphrodite nerve rings [5,95]. Interestingly, we found that ASH formed a chemical synapse onto AIY on one or both of the left-right pairs at the anterior region of zone 1, where the ectopic synapses begin to form in cima-1(wy84) or eat-4(OE) animals (Fig 6A and S2 Video) [5, 95]. And we found that the ASH neurons were extended posteriorly and aligned next to the AIY zone 1 in cima-1(wy84) mutants (S8A-S8B’’ Fig), which makes it possible for ASH to form extra synapses onto the AIY in the zone 1. Together, these data show that ASH neurons are AIY presynaptic partners, which suggests that the formation of the ectopic AIY presynaptic structure may be due to the ectopic synaptic connections between ASH and AIY.

High temperature alters synaptic subcellular specificity through glutamatergic signaling

To understand whether there is any physiological condition that can affect the AIY synaptic specificity, we tested the cultivation temperature since AIY is part of the thermotaxis circuit [46,51]. We examined the AIY presynaptic markers at a high physiological temperature (25°C) (Fig 7A). Wild-type animals can grow and reproduce normally at 25°C [96]. We found that the AIY morphology appeared largely intact at 25°C (S8A’ and S8C’ Fig). Interestingly, those animals displayed a highly penetrant ectopic synaptic structure as indicated by both GFP::RAB-3 and GFP::SYD-1 in the normally asynaptic zone 1 of AIY (GFP::RAB-3: 16.83 vs 79.67% at 22°C and 25°C respectively, p<0.0001; GFP::SYD-1: 15.79 vs 78.83% at 22°C and 25°C respectively, p<0.0001. Fig 7B–7F). Consistently, the ventral synaptic length and the ratio of the ventral to total synaptic length were increased at 25°C (8.53μm and 0.21 at 22°C; vs 16.88μm and 0.36 at 25°C, p<0.0001 for both comparisons. Fig 7G and 7H). The data indicate that high physiological temperature induces the ectopic synapses in AIY interneurons in wild-type animals.

Next, we asked whether the low temperature could inhibit the AIY ectopic synapses. To address this question, we quantified the AIY presynaptic phenotype in both wild-type and cima-1(wy84) animals at 15°C and 22°C. We found that the ectopic synapses were indeed reduced both in the wild-type and cima-1(wy84) animals at 15°C as compared to that at 22°C (S9A and S9B Fig). To exclude the possibility that the phenotypic difference was due to the slow development rate at 15°C, we also quantified the synaptic phenotype at the adult Day 2 stage and found similar results (S9A and S9B Fig). Those data indicate that high temperature promotes, while low temperature suppresses the AIY ectopic synaptic assembly.

To determine the temporal window required for high temperature to promote the AIY ectopic synapse formation, animals were shifted to 25°C during different developmental stages (S9C Fig). Interestingly, the AIY ectopic synapse formation required the high temperature treatment during developmental stages, with more robust effect during embryonic stages (S9D Fig). No ectopic synapse was observed when treating from the larval L4 stage (S9D Fig). The results suggest that the AIY ectopic synaptic formation induced by high temperature is development-dependent.

Given that glutamate signaling is required for the AIY ectopic synaptic formation in cima-1 mutants, we asked whether it was also required for the high temperature induced ectopic synapse formation. We examined the phenotype of eat-4(ky5) mutants at 25°C. Interestingly, eat-4(ky5) suppressed the AIY ectopic synapse formation at high temperature (82.43% and 20.81% of animals with ectopic synapses in wild-type and eat-4(ky5) mutants respectively, p<0.0001, Fig 7I, 7J and 7Q). The data demonstrate that glutamatergic neurotransmission is required for the AIY ectopic synaptic formation at high temperature.

Next, we determined whether GLC-3 and GLC-4 were required by examining the mutant phenotype at 25°C. Indeed, either glc-3(ok321) or glc-4(ok212) mutation partially, while the glc-3(ok321);glc-4(ok212) double mutations completely inhibited the ectopic synapses at 25°C (Fig 7K, 7L, 7M and 7Q). These results indicate that high temperature induces the AIY ectopic synaptic formation mediated by the glutamatergic GLC-3/GLC-4 receptors.

Given that the AIY ectopic synaptic formation in eat-4(OE) or cima-1 mutant animals require glutamate transmission from ASH neurons, we asked whether ASH neurons were also required for the high temperature induced AIY ectopic synaptic formation. Through cell specific eat-4 rescue experiments, we found that expressing eat-4 specifically in the ASH neurons significantly restored the ectopic synapses in eat-4(ky5) mutants at 25°C, which was more robust than that at 22°C (Fig 7N–7P and 7R), suggesting that ASH neurons are involved in high temperature induced AIY ectopic synaptic formation.

To visualize the anatomic relationship between ASH and AIY at high temperature, we labeled the ASH and AIY with cytoplasmic GFP and mCherry simultaneously, and found ASH process extended posteriorly alongside the AIY zone 1 (S8C-8C’’ Fig), suggesting that ASH could form synapses onto AIY in this region.

To address whether high temperature enhances the glutamate release from ASH, we quantified the intensity of the ASH VGLUT-pHluorin. PHluorin is a fluorescent protein quenched in acidic conditions such as inside the synaptic vesicle lumen [97]. We found that VGLUT-pHluorin intensity was enhanced in the ASH axon at high temperature, suggesting more glutamate vesicles releasing from ASH neurons (Fig 7S–7U). These results are consistent with the model that high temperature induces the AIY ectopic synaptic formation by enhancing the ASH glutamatergic neurotransmission.

Although 25°C is at the border line of the normal breeding temperature range (15–25°C), this could be a potential stress condition. To address if other stress conditions could induce the ectopic synapse formation, we tested the effect of osmotic and oxidative stresses on the AIY synaptic subcellular specificity, and found that animals treated with 200~500mM sorbitol or 0~10mM hydrogen peroxide displayed normal AIY synaptic distribution (S10 Fig), suggesting that the ectopic AIY synapses are not induced by general stresses.

Discussion

Our previous study identified that cima-1 in epidermis is required for the normal AIY presynaptic distribution. cima-1 functions partially through the VCSC glia [55]. In this study, we uncover an inhibitory glutamate signaling that is required for the cima-1(wy84)-induced AIY ectopic synaptic formation (Fig 8A). Furthermore, we show that eat-4(OE) or high temperature can trigger the glutamate signaling from ASH sensory neurons to promote the ectopic presynaptic formation, which is mediated by the inhibitory glutamate gated chloride channels GLC-3 and GLC-4 in the AIY interneurons. These findings describe a novel mechanism underlying synaptic subcellular specificity.

ASH neurons form inhibitory synapses onto the AIY

In this study, we demonstrate that ASH forms inhibitory synapses on the AIY interneurons. Four lines of evidence support this. First, ASH processes are aligned next to the AIY, which indicates ASH may form synapses onto AIY (S8 Fig). Secondly, through tissue-specific expression analysis, we showed that the glutamate required for the AIY ectopic synaptic formation is released from the ASH neurons and sensed by the GLC-3/GLC-4 receptors in the AIY. Thirdly, ASH specific eat-4(OE) reduces the frequency of AIY Ca²⁺ oscillation, indicating that ASH inhibits AIY excitability. Finally, the ASH-AIY synaptic connection was confirmed by electron microscopy reconstruction [95].

The next question is why expressing eat-4 in other AIY presynaptic glutamatergic neurons such as AFD and AWC does not rescue. There are two possibilities. First, the amount or frequency of the glutamate released from ASH could be much higher than from any of other AIY presynaptic neurons. Second, the ASH-AIY synapses, which localizes at the border of zone 2 and zone 1 in the wild-type animals, are closer to the ectopic synaptic sites in the zone 1 than those of AFD-AIY or AWC-AIY. Therefore, the glutamate from ASH can diffuse more easily to the zone 1 region where it probably locates the GLC-3/GLC-4 receptors and promotes the ectopic synaptic assembly.

In vertebrates, excitatory neuronal activity is well recognized for its role in modulating excitatory synapse formation, maturation and plasticity [98–100]. More recently, GABAergic activity was also found to regulate both inhibitory and excitatory synaptic development at early developmental stage through depolarizing the postsynaptic neurons [101,102]. However, our knowledge about the role of GABA activity in promoting synaptic formation is largely limited to the early developmental stage when GABA acts as an excitatory transmitter [103]. In this study, we demonstrated that an important role of the inhibitory ASH-AIY synaptic transmission in promoting ectopic excitatory presynaptic assembly in the postsynaptic AIY neurons. The future work should focus on understanding the underlying molecular mechanisms.

Pentameric ligand-gated ion channels regulate synaptic specificity

Glutamate signals promote the AIY ectopic synaptic formation through two pentameric ligand-gated ion channels GLC-3 and GLC-4, which are localized to the AIY presynaptic region, partially overlapping with the presynaptic marker RAB-3. Unlike a typical bipolar neuron, which assembles presynaptic and postsynaptic structures in axons or dendrites, AIY presynaptic and postsynaptic sites are overlapping along the single neurite in zone 2 and 3 regions [5]. The close anatomic relationship between postsynaptic and presynaptic sites may be helpful for the activity-dependent presynaptic assembly. Alternatively, GLC-3 and GLC-4 may also localize to the presynaptic sites. In this case, GLC-3 and GLC-4 may be activated by the glutamate spillover from adjacent synapses.

Glutamate spillover plays physiological or pathological roles [104–107]. The loss of astrocyte-like VCSC glia or glutamate reuptake transporter GLT-1 can alter the animal escaping or exploration behavior [107]. Increasing the extracellular level of glutamate may also result in neurotoxicity and degeneration [104,108]. Similar functions of glutamate present in mammals [105,106].

The inhibitory neurotransmitter receptors such as GABA receptors were also found in the excitatory presynaptic boutons in mammalian brain, where they play important roles in regulating synaptic transmission [109–112]. However, it is largely unknown if these presynaptic inhibitory receptors are involved in synaptic development or plasticity.

The closest related mammalian homologs of GLC-3 and GLC-4 are glycine receptors (GlyRs) [113–115]. GlyRs are one of the major inhibitory neurotransmitter receptors, involved not only in neuronal signaling processing, but also in neurodevelopment [116]. GlyRs regulate postsynaptic protein clustering in immature rat spinal neurons [117], and cortical interneuron migration in mouse [116]. Mutations of GlyRs are associated with a number of neurological disorders including hyperekplexia, temporal lobe epilepsy, chronic inflammatory pain, autism, etc, which makes GlyRs potential drug targets [118]. Given the functional conservation of pLGIC family receptors, C. elegans GLC-3 and GLC-4 may provide an excellent model to address the mechanisms underlying physiological and pathological roles of GlyRs.

Temporal regulation of spatial specificity

During embryonic development, the AIY presynaptic assembly in zone 2 region is mainly regulated by netrin/DCC secreted from the VCSC glia [6]. However, it is largely unknown how the zone 1 avoids synaptic assembly. In this study, we found that the amount of glutamate released from ASH is critical for the synaptic assembly in zone 1 region. Although glutamatergic neurotransmission from ASH is also required for the ectopic synapse formation in cima-1(wy84) mutants, we noticed that the synaptic subcellular defects are different between cima-1(wy84) and eat-4(OE) animals. The ectopic synapses appear since newly hatched larval L1 stage in eat-4(OE) animals and at the adult stage in cima-1 mutants [55]. Additionally, the VCSC glia contribute more to the synaptic defect of cima-1(wy84) than that of eat-4(OE). Those differences indicate that cima-1(wy84) and eat-4(OE) may regulate the synaptic subcellular specificity through different molecular mechanisms.

Environmental temperature affects synaptic subcellular specificity

In this study, we showed that the synaptic subcellular specificity was affected by temperature during developmental stages. Specifically, we showed that high temperature promoted the ectopic synaptic formation mediated by the vesicle glutamate transporter VGLUT/EAT-4 in ASH and glutamate receptors GLC-3/GLC-4 in AIY, while low temperature inhibited the ectopic synaptic assembly. This finding suggests that temperature modulates the synaptic subcellular specificity through glutamatergic neurotransmission. No ectopic synapse observed under osmotic or oxidative stresses suggests the synaptic specificity is not affected by general stresses.

The AIY interneurons are part of the thermosensory circuit involved in the thermotaxis behavior [46–51]. Previous studies have identified that AFD, AWC and ASI are major thermosensory neurons [46,51–54,119]. In this study, we found that ASH sensory neurons could sense the cultivation temperature and regulate the AIY synaptic subcellular specificity, suggesting that the ASH could be part of the thermosensory circuit, which should be further tested in the future.

Temperature is a common and vital environmental factor for many organisms. The nervous system is very sensitive to high temperature during embryogenesis [120]. High temperature often results in neurological disorders including neural tube defects, microcephaly, microphthalmia, microvascular abnormity in vertebrates [120]. In Drosophila, high temperature also induces neural developmental defects [21,23,121,122]. Temperature can modulate the nematode C. elegans thermotaxis behaviors and lifespan mediated by neuronal activity [123–125]. In our study, the effects of temperature on synaptic subcellular specificity provide an excellent model to address the mechanistic insights into the high temperature induced neurodevelopmental defects in vivo.

Materials and methods

Strains and cultivation

Strains were cultivated on OP50-seeded nematode growth medium (NGM) plates at 22°C unless specified [126]. Wild-type (WT) animals are Bristol strain N2. Mutant alleles used in this study include:

LGI: unc-13(e1091), avr-14(ad1302), glc-2(gk179), mgl-2(tm355), glr-3(tm6403)

LGII: cat-2(e1112), glc-4(ok212), nmr-1(ak4), glr-4(tm3239)

LGIII: eat-4(ky5), eat-4(nj2), eat-4(nj6), unc-47(n2409), glr-1(n2461), glr-2(ok2342)

LGIV: cima-1(wy84), cima-1(gk902655), unc-17(cn355), mgl-3(tm1766)

LGV: avr-15(ad1051), glc-1(pk54), glc-3(ok321), nmr-2(ok3324), glr-5(tm3506)

LGX: mgl-1(tm1811), glr-6(tm2729), glr-7(tm1824)

All worm strains used in this study are listed in the S1 Excel.

Plasmids and transgenic manipulations

Plasmids were made in the pSM or pPD49.26 by recombination [127]. The transgenic strains carrying extrachromosomal DNA arrays were generated using standard microinjection protocol [128]. The following plasmids were used as co-injection markers: Phlh-17::mCherry, Pmyo-3::mCherry, Punc-122::GFP, Punc-122::RFP or Plin44::mCherry. Unless otherwise stated in S1 Excel, the concentration of plasmids was injected at 20 ng/μl. The cDNA plasmids generated for use in this study (glc-3 cDNA, glc-4 cDNA), were cloned by RT-PCR from total RNA isolated from WT (N2) worms. The unc-103^A334T cDNA was amplified from the strain SQC0132 [yfhIx0132 (Punc-103::unc-103^A334T::GFP)] [92], which is a gift from Dr. Shiqing Cai. The eat-4 cDNA was cloned from the plasmid Peat-4a::eat-4 (cDNA)::GFP [51] from Dr. Ikue Mori. The Psra-6::caspase p12 and Psra-6::caspase p17 constructs were modified from plasmids DACR336(Pttx-3::caspase p12) and DACR335(Pttx-3::caspase p17) respectively through replacing the ttx-3 promoter with sra-6 promoter (4kb) by recombination[50,83].

The ASH-specific EAT-4(VGLUT)-pHluorin expression construct was created through inserting the PHluorin CDS into the Psra-6::eat-4a[129]. The PHluorin was inserted after the conserved glycine residue at position 106 of eat-4 A isoform cDNA by PCR primers which adds 42 bases (TCTACCTCTGGAGGATCTGGAGGAACCGGAGGATCTATGGGA) for the upstream linker, 45 bases (ACCGGTGGAGGAACCGGAGGAACCGGAGGA TCTGGAGGAACCGGA) for downstream linker, as previously described [129]. Forward primer to amplify PHluorin: 5’- GAGGATCTGGAGGAACCGGAGGATCTATGGGAAGTAAAGGAGAAGAACTTTTC-3’. Reverse primer to amplify PHluorin: 5’- CTCCAGATCCTCCGGTTCCTCCGGTTCCTCCACCGGTTTTGTATAGTTCATCC-3’. The vector were amplified from the Psra-6::eat-4 plasmids with forward primer: 5’-ACCGGAGGAACCGGAGGATCTGGAGGAACCGGAAAAGTTCAT ATGCATGAATTC-3’ and reverse primer: 5’-GATCCTCCGGTTCCTCCAGATC CTCCAGAGGTAGATCCGTATGGATCTGTATAATTTT-3’. All plasmids and primer information in this study were listed in the S2 Excel.

ASH and glia ablation

The two-component system of reconstituted caspase (recCaspase) [83] was driven by the sra-6 promoter, which specifically ablating the ASH neurons. Ablation was confirmed by lack of ASH specific marker (kyIs39) [78].

The two-component system of reconstituted caspase (recCaspase) [83] was driven by the hlh-17 promoter, which specifically ablating the CEPsh glia. Ablation was confirmed by lack of the CEPsh-specific marker(nsIs105) [107].

Electron microscopy analysis

Serial-section electron microscopy datasets were imported into CATMAID [130] to peruse. Each section containing AIY was examined to determine if contact was made with ASH, and if so, whether chemical synapses were present. Chemical synapses were defined as a presynaptic bouton containing a pool of synaptic vesicles as well as a dense presynaptic projection inside the membrane.

Special temperature treatment

Animals was transferred to 15°C or 25°C at specific time points as illustrated in the figures. The phenotype of next or the same generation was scored at the adult Day 1 or Day 2 stage. In these assays, animal synchronization was done through two steps. First, eggs were collected within one-hour time window; second, animals were synchronized at the L4 stage.

Osmotic stress assays

Young adults were grown on NGM agar plates containing 0mM(control), 200mM, 300mM, 400mM or 500 mM sorbitol seeded with OP50 until they reached the adult Day 1 stage when the synaptic phenotype was scored. The concentrations and methods were modified from the study of Chandler-Brown et al. [131].

Oxidative stress assays

Young adults were grown on NGM agar plates with OP50 and supplemented with S-basal buffer containing hydrogen peroxide (0.5mM, 2mM, 5mM,10 mM) at specific time points as illustrated in the figures. Animals were synchronized at the L4 stage and phenotypes were scored 24 hours later. The concentrations were modified from Lee et al. [132]. Animals can survive and reproduce at low concentration of H₂O₂ (0.5mM, 2mM) from the young P0 stage, but not at higher than 5mM. We also treated animals with high concentrations (5mM, 10mM) for shorter time (time window4 in the high temperature treatment).

Calcium imaging of AIY neurons

For in vivo calcium imaging, individual Day 1 (D1) adult hermaphroditic worms were immobilized with Polybead Microspheres 0.10μm (Polysciences) on 12% agarose pads. Fluorescent images were acquired using an Andor Dragonfly Spinning Disc Confocal Microscope with 60x objectives coupled with an ZYLA camera. GCaMP6s (in AIY) was excited by 488nm excitation wavelength lasers, and the mCherry control was imaged with 561 nm excitation wavelength lasers. The fluorescent signals of video were collected at the rate of 2 Hz[133].

For AIY GCaMP signals, the ROI is AIY neurite (Zone 2 and Zone 1). The relatively GCaMP signals for each data point were calculated as:

For peak frequency of AIY GCaMP was taken as F_n, which was calculated as:

For peak amplitude of AIY GCaMP was calculated as:

The data of fluorescence intensity was quantified with the ImageJ (Fiji).

Fluorescence microscopy and confocal imaging

Confocal images were acquired with an Andor Dragonfly Spinning Disc Confocal Microscope with 40x or 60x objectives. The fluorescently tagged fusion proteins GFP or mCherry was imaged with 488 or 561 nm excitation wavelength lasers, respectively. Animals were anesthetized with 50mM muscimol or Polybead Microspheres 0.10μm (the recorded about GCaMP and PHluorin). Images were processed with Imaris, ImageJ (Fiji) and Photoshop. All images are oriented anterior to the left and dorsal up.

Quantification and statistical analysis

To quantify the percentage of animals with ectopic synapses of AIY zone 1 at the adult stage, animals were synchronized at larva stage 4 (L4) and then we scored the phenotypes 24 hours later using a Nikon Ni-U fluorescent microscope with 40x objectives or Andor Dragonfly Spinning Disc Confocal Microscope with 40x objectives. For the larval phenotypes, synchronized eggs were cultivated for 12 and 48 hours to reach the middle stage of L1 and L4. At least three biological replicates were done for each quantification. For transgenic analysis, at least two independent transgenic lines were generated and quantified unless specified. The data of AIY ectopic synapses were blindly recorded. Other data were collected based on genotypes or treatments. All quantitative raw data are in S3 Excel.

For ASH EAT-4-PHluorin intensity, the ROI is ventral axon of ASH. The relatively PHluorin intensity for each data point were calculated as: F_PHluorin/F_mCherry.

The data of fluorescence intensity was collected using the ImageJ (Fiji).

Statistical analyses were conducted with GraphPad Prism software (version 6.01). The comparisons between two groups were determined by the unpaired t test, while multiple comparisons were analyzed with one-way analysis of variance with Dunnett’s multiple comparison test. Error bars represent the standard errors of the mean (SEM).

Supporting information

S1 Fig [a]
Neurotransmission are not required for synaptic subcellular specificity per se.

S2 Fig [wy84]
GABAergic, cholinergic and dopaminergic neurotransmissions are not required for the ectopic synaptic formation in .

S3 Fig [a]
The expression of begins at the embryo stage.

S4 Fig [a]
The AIY ectopic synapses induced by (OE) is largely independent of the VCSC glia.

S5 Fig [oe]
promotes the AIY ectopic synapse formation since L1 stage.

S6 Fig [a]
Glutamate receptors are not required for AIY synaptic subcellular specificity per se.

S7 Fig [a]
Glutamate-gated chloride channels GLC-3 and GLC-4 are required for the ectopic synapse formation in .

S8 Fig [wy84]
The ASH axons are extended posteriorly and overlap with AIY zone 1 in and 25°C treated wild-type animals.

S9 Fig [a]
Temperature alters the synaptic subcellular specificity.

S10 Fig [a]
Osmotic and oxidative stresses do not affect the AIY synaptic subcellular specificity.

S1 Excel [xlsx]
The detail information for strains used in this study.

S2 Excel [xlsx]
The primer sequence information.

S3 Excel [xlsx]
The archive of raw quantitative data.

S1 Video [mp4]
The AIY GCaMP fluorescent video in wild-type and Psra-6::EAT-4 transgenic animals.

S2 Video [mp4]
The 3D model showed the anatomic relationship between ASH and AIY.