The Input-Output Relationship of AIY Interneurons in Caenorhabditis elegans in Noisy Environment

Summary

Determining how neurotransmitter input causes various neuronal activities is crucial to understanding neuronal information processing. In Caenorhabditis elegans, AIY interneurons receive several sources of sensory information as glutamate inputs and regulate behavior by integrating these inputs. However, the relationship between glutamate input and the Ca²⁺ response in AIY under environmental noise, in other words, without explicit stimulation, remains unknown. Here, we show that glutamate-input fluctuations evoke a sporadic Ca²⁺ response in AIY without stimulation. To ensure that Ca²⁺ response can be considered AIY output, we show that the membrane-potential depolarization precedes Ca²⁺ responses in AIY. We used an odor as model stimulation to modulate the sensory inputs. Simultaneous imaging of glutamate input and Ca²⁺ response, together with glutamate transmission mutants, showed that glutamate-input fluctuations evoke sporadic Ca²⁺ responses. We identified the input-output relationships under environmental noise in vivo, and our results address the relationship between sensory-input fluctuations and behavioral variability.

Graphical Abstract

Highlights

• •Glutamate-input fluctuation evokes a sporadic Ca²⁺ response in AIY
• •The membrane-potential depolarization precedes Ca²⁺ responses in AIY
• •Ca²⁺ increases in AIY when the glutamate input decreases, and vice versa
• •Mutation of glutamate signaling abolished Ca²⁺ response in AIY

Neuroscience; Systems Neuroscience; Sensory Neuroscience

Introduction

Information processing from sensory inputs to regulate behavior in noisy environments is a fundamental topic in neuroscience. To improve our understanding of such information processing, the relationship between the noise level of the stimulation and animal perception has been studied in depth (Faisal et al., 2008, Nienborg and Cumming, 2009, Renart and Machens, 2014). However, at the neuronal level these relationships remain largely unclear. Although the input-output relationship in neurons has also been thoroughly investigated (Augustine et al., 2003, Cichon and Gan, 2015, Destexhe, 2011, Frick et al., 2004, Losonczy et al., 2008, Ogawa and Oka, 2015, Stuart and Spruston, 2015), it remains unknown how neurotransmitter inputs affect postsynaptic neuronal activity in vivo in noisy environments, because simultaneous measurement of input and output in vivo is technically difficult due to the complexity of the networks involved (Ogawa and Oka, 2015, Prešern et al., 2015, Stuart and Spruston, 2015).

The nematode Caenorhabditis elegans is a suitable model for investigating this question. It has a simple neuronal circuit (White et al., 1986), as well as optical transparency, making it well suited for visualization with various imaging techniques (Kerr and Schafer, 2006). The AIY interneurons of C. elegans are particularly suitable for this purpose; they receive various sensory inputs, including gustatory, olfactory, and thermal information (Chalasani et al., 2007, Clark et al., 2006, Satoh et al., 2014), which regulate the animal's behavior following integration (Kocabas et al., 2012, Li et al., 2014, Satoh et al., 2014). Furthermore, AIY interneurons show a sporadic Ca²⁺ response regardless of the presence of explicit stimulation (Chalasani et al., 2007, Clark et al., 2006), whereas sensory neurons show a deterministic Ca²⁺ response to environmental stimulation (Chalasani et al., 2007, Clark et al., 2006, Kato et al., 2014, Tsukada et al., 2016). Both glutamate receptor mutations and the ablation of sensory neurons have been shown to abolish the Ca²⁺ response to sensory stimulation in AIY (Chalasani et al., 2007, Clark et al., 2006). This suggests that glutamate input to AIY can evoke the sporadic Ca²⁺ response as a result of sensory input integration under natural noise.

We used several simultaneous fluorescence imaging techniques to identify the input-output relationship in AIY in a noisy environment in vivo. To ensure that Ca²⁺ spikes could be considered AIY output, we first showed that depolarization of the membrane potential precedes Ca²⁺ spikes, by simultaneously imaging both in AIY. We used an odor, isoamyl alcohol (IAA) as model stimulation to modulate the sensory inputs, and simultaneous imaging of glutamate input and the Ca²⁺ response showed that glutamate input decreases when the Ca²⁺ spikes occur, with or without odor stimulation, and vice versa. We also investigated these relationships in both the glutamate receptor and glutamate-defective mutants and showed that fluctuations in glutamate input evoke the sporadic Ca²⁺ response in AIY. As far as we know, this is the first report identifying the input-output relationship for fluctuations under natural environmental noise in vivo. Glutamate inputs represent sensory inputs, and AIY neurons have been reported to regulate behavior; our results thus suggest that fluctuation in sensory input induces behavioral variability.

Results

Membrane-Potential Spike Precedes Ca²⁺ Response in AIY

To quantify the input-output relationship in a single neuron in vivo, we investigated AIY interneurons in C. elegans. AIY neurons receive environmental information from sensory neurons via glutamate as a neurotransmitter and show a large sporadic Ca²⁺ response (hereafter, a “spike”) as a result of sensory integration (Chalasani et al., 2007, Clark et al., 2006). In C. elegans, Ca²⁺ is thought to be the main charge carrier in membrane-potential changes, due to the lack of voltage-gated Na⁺ channels in its neurons (Faumont et al., 2011, Faumont et al., 2006, Gao and Zhen, 2011, Goodman et al., 1998, Mellem et al., 2008, Shidara et al., 2013). Recently, the relationship between membrane potential and Ca²⁺ channels has been thoroughly investigated, and it has been shown that Ca²⁺ channels are necessary for membrane-potential spikes (Liu et al., 2018, Shindou et al., 2019). Ca²⁺ responses in AIY regulate the animal's behavior (Li et al., 2014), so the results of information processing should be expressed as membrane-potential changes. However, the relationship between Ca²⁺ and membrane potential in AIY remains unclear (Liu et al., 2018, Liu et al., 2009, Shidara et al., 2013), because it has not been measured simultaneously in AIY. We therefore measured both simultaneously, using a genetically encoded voltage indicator, ArcLight (Jin et al., 2012), and a genetically encoded Ca²⁺ indicator, R-GECO (Zhao et al., 2011) (Figure 1A). As previously suggested (Faumont et al., 2012), changes in membrane potential appeared as a spike-like response under ArcLight fluorescence, similar to a Ca²⁺ spike (Figure 1B). Importantly, these membrane-potential spikes were coincident with Ca²⁺ spikes. Membrane-potential changes were detected at the neurite and even at the soma, as reported in our earlier study (Shidara et al., 2013).

Figure 1.

Simultaneous Imaging of Membrane Potential and Ca²⁺ in AIY Interneurons

(A) Confocal image of the voltage indicator (ArcLight: green, left) and Ca²⁺ indicator (R-GECO magenta, middle). The merged image is on the right. Scale bar, 10 μm. A, anterior; L, left; P, posterior; R, right.

(B) Representative results of Ca²⁺ (red) and membrane potentials (black) at the soma (left) and neurite (right).

(C) Averaged Ca²⁺ spike (red) and membrane-potential spike (black) without (left) and with (right) odor. The shadows around the solid lines indicate the standard error of the mean (SEM) (no odor, N = 20, n = 16; odor, N = 10, n = 15).

(D) Membrane potential (left) and Ca²⁺ spike (right) intensities, with (filled boxes) and without (hollow boxes) odor. Box plots include the median (center line), quartiles (boxes), and range (whiskers). The statistical metrics are as follows: membrane potential, p = 0.38; Ca²⁺, p = 0.85; paired t tests (no odor, N = 20, n = 16; odor, N = 10, n = 15).

(E) Correlation between Ca²⁺ spike amplitude and membrane-potential spike amplitude. The red diamonds and pink dots are the responses in the presence and absence of odor stimulation, respectively. The statistical metrics are as follows: no odor, Pearson correlation coefficient (PCC) = −0.19, p = 0.48; odor, PCC = −0.40, p = 0.14 (no odor, N = 20, n = 16; odor, N = 10, n = 15).

To investigate the relationship between membrane-potential spikes and Ca²⁺ spikes, we collected and summarized the data (Figure 1C). Membrane-potential spikes preceded Ca²⁺ spikes, and their amplitudes were not modulated by odor presentation (Figure 1D). However, the amplitudes of Ca²⁺ spikes and membrane-potential spikes were not correlated (Figure 1E). Based on these results, membrane-potential changes precede Ca²⁺ changes. In subsequent analyses, therefore, we consider Ca²⁺ spikes to be AIY output.

Glutamate Input and Ca²⁺ Response to Odor Stimulation in AIY Interneurons

In AIY interneurons, glutamate input reflects environmental information from sensory neurons, and sporadic Ca²⁺ spikes occur as a result of sensory integration (Chalasani et al., 2007, Clark et al., 2006). These glutamate inputs emanate from gustatory, olfactory, and thermal sensory neurons (Chalasani et al., 2007, Clark et al., 2006, Satoh et al., 2014, White et al., 1986); therefore, in the absence of explicit changes of stimulation, glutamate inputs to AIY presumably reflect environmental noise, such as fluctuations in environmental factors such as temperature, local salt concentration, and olfactory molecules. AIY neurons receive inhibitory glutamate inputs via a glutamate-gated chloride channel, GLC-3 (Clark et al., 2006, Horoszok et al., 2001, Serrano-Saiz et al., 2013). AIY neurons exhibit Ca²⁺ spikes in response to odor stimulation because sensory neurons reduce their Ca²⁺ response and neurotransmitter release upon addition of odor (Chalasani et al., 2007, Ventimiglia and Bargmann, 2017) (Figure S1). Mutation of glc-3 or glutamate transmission abolishes Ca²⁺ spikes in response to odor in AIY neurons (Chalasani et al., 2007). Therefore, sporadic Ca²⁺ spikes in AIY could occur in response to variable glutamatergic inputs representing environmental noise. To demonstrate this directly, we simultaneously measured glutamatergic inputs and Ca²⁺ responses in AIY interneurons, using a genetically encoded glutamate indicator, iGluSnFR (Marvin et al., 2013), and a genetically encoded Ca²⁺ indicator, R-GECO (Zhao et al., 2011), respectively (Figure 2A).

Figure 2.

Simultaneous Imaging of Glutamate Inputs and Ca²⁺ Responses in AIY Interneurons

(A) Confocal image of the glutamate indicator (iGluSnFR: green, left) and Ca²⁺ indicator (R-GECO: magenta, middle) in AIY. The merged image is on the right. Scale bar, 10 μm. A, anterior; L, left; P, posterior; R, right.

(B) Averaged glutamate input (blue) and Ca²⁺ response (red) to odor. The shaded region indicates isoamyl alcohol (IAA) stimulation, and the shadows around the solid lines indicate the SEM.

(C) Averaged glutamate intensities before (pre-addition) and after (post-addition) odor addition (left). Glutamate intensities before (pre-removal) and after (post-removal) odor stimulation removal (right). The statistical metrics are as follows: left, p = 0.00010; right, p = 0.000086; paired t test (N = 15). ***p < 0.001.

As a previous study hypothesized that odor (IAA) stimulation could modify glutamate input to AIY (Chalasani et al., 2007), we selected IAA as model sensory stimulation. We measured glutamate input and the Ca²⁺ response with and without odor stimulation (Figure 2B). The glutamate input to AIY decreased following odor application and transiently increased following odor removal, in a deterministic manner (Figure 2C). These responses to odor correspond to the responses of sensory neurons (Chalasani et al., 2007, Shidara et al., 2017). Although there were deterministic changes in glutamate input, large Ca²⁺ spikes were observed even without odor, and often at the onset of odor stimulation, as reported previously (Chalasani et al., 2007). Odor application typically evoked Ca²⁺ spikes immediately in wild-type animals (see N2 in Figure 6C), suggesting that glutamate decreases, induced by odor application, evoke Ca²⁺ spikes. Moreover, odor presence increased the frequency of Ca²⁺ spikes (see N2 in Figure 6D). This suggests that reducing glutamate input via odor application causes Ca²⁺ spikes.

Figure 6.

Ca²⁺ Responses and Glutamate Inputs in eat-4 Mutants

(A) Averaged glutamate input (blue) and Ca²⁺ response (red) to odor. The shaded region indicates IAA stimulation and the shadows around the solid lines indicate the SEM (N = 12).

(B) Average glutamate intensity before (pre-addition) and after (post-addition) odor stimulation onset (left), and average intensity before (pre-removal) and after (post-removal) odor stimulation removal (right). Error bars indicate the SEM. The statistical metrics are as follows: left, p = 0.41; right, p = 0.055; paired t test (N = 12).

(C) Ca²⁺ spike probability following odor addition. Comparison between N2 animals and glc-3 mutants: p = 0.13; comparison between N2 animals and eat-4 mutants: p = 0.014. Fisher's exact test with Holm correction (N2, N = 15; glc-3, N = 12; eat-4, N = 12); *p < 0.05.

(D) Ca²⁺ spike frequency in N2 (filled), glc-3 (striped), and eat-4 mutants (hollow) with and without odor. Box plots indicate the median (center line), quartiles (boxes), and range (whiskers). The statistical metrics are as follows: N2 data comparison between no odor and later odor, p = 0.0045; glc-3 data comparison between no odor and later odor, p = 0.34; eat-4 data comparison between no odor and later odor, p = 0.34; Wilcoxon signed rank test. No odor data comparison between N2 and glc-3, p = 0.024; no odor data comparison between N2 and eat-4, p = 0.00044; later odor data comparison between N2 and glc-3, p = 0.014; later odor data comparison between N2 and eat-4, p = 0.000015; Wilcoxon rank-sum test. The Holm method was used for the correction (N2, N = 15; glc-3, N = 12; eat-4, N = 12). *p < 0.05, **p < 0.01, ***p < 0.001.

(E) Averaged glutamate responses coincident with Ca²⁺ spikes. Time 0 indicates the onset of the Ca²⁺ spike. All glutamate responses were normalized to zero at time 0. The purple, blue, and green lines indicate the responses under the no odor, early odor, and later odor conditions, respectively. The bars indicate the SEM (N = 12, no odor: n = 9; early odor, n = 3; later odor, n = 9).

(F) Glutamate input before Ca²⁺ spikes and at Ca²⁺ spike onset. “Pre” indicates the average intensity from −3 to −2 s in (E), and “onset” indicates the average intensity from −0.5 to +0.5 s. The statistical metrics are as follows: no odor: p = 0.60; early odor, p = 0.53; later odor, p = 0.50; paired t test (N = 12; no odor, n = 9; early odor, n = 3; later odor, n = 9).

(G) Averaged glutamate decreases (filled boxes) with and without (hollow boxes) Ca²⁺ spikes. Box plots indicate the median (center line), quartiles (boxes), and range (whiskers). The statistical metrics are as follows: no odor, p = 0.88; later odor, p = 0.48; Welch's t test (N = 12, no odor, spike, n = 9; no odor, no spike, n = 108; later odor, spike, n = 9; later odor, no spike, n = 100).

Glutamatergic Input Decreases when the Ca²⁺ Response Increases in AIY under Natural Noise, and Vice Versa

When quantifying the average response of glutamatergic input and Ca²⁺ in AIY to odor, we found that decreases in glutamatergic input appeared to be related to Ca²⁺ spikes (Figure 3A). To investigate this relationship further, we classified conditions into three categories: no odor, early odor, and later odor (Figure 3B); glutamatergic inputs were collected and aligned with the timing of Ca²⁺ spikes (Figure 3C). Glutamatergic inputs decreased before the Ca²⁺ spikes, regardless of odor presentation (Figure 3D). Decreasing levels of glutamate did not show a strong correlation with Ca²⁺ spike amplitude (Figure 3E) in any of the three conditions. These results imply that decreases in glutamatergic inputs are necessary to trigger Ca²⁺ spikes. We also compared the extent of glutamate decreases in the presence or absence of Ca²⁺ spikes (Figure 3F) and found that glutamate input decreased significantly when Ca²⁺ spikes occurred, with or without odor, which suggests that decreases in glutamate evoke Ca²⁺ spikes.

Figure 3.

Glutamate Input When the Ca²⁺ Spike Occurs

(A) Representative responses of Ca²⁺ and glutamate input (left). Magnifications of the dotted boxes showing the responses without (middle) and with (right) odor stimulation. The blue and red lines are the glutamate input and Ca²⁺ response, respectively. The shaded region indicates IAA stimulation.

(B) Definition of the response categories.

(C) The averaged glutamate responses coincident with Ca²⁺ spikes. Time 0 indicates the onset of Ca²⁺ spikes. All glutamate responses were normalized to zero at time 0. The purple, blue, and green lines indicate the responses under the no odor, early odor, and later odor conditions, respectively (defined in B). The error bars indicate the SEM (N = 15; no odor, n = 15; early odor, n = 11; later odor, n = 21).

(D) Glutamate inputs before the Ca²⁺ spike and at spike onset. “Pre” indicates the average intensity from −3 to −2 s in (C), and “onset” indicates the average intensity from −0.5 to +0.5 s. The statistical metrics are as follows: no odor, p = 0.010; early odor, p = 0.0024; later odor, p = 0.0037; paired t test (N = 15; no odor, n = 15; early odor, n = 11; later odor, n = 21). *p < 0.05, **p < 0.01.

(E) Correlation between glutamate decreases and Ca²⁺ increases. The purple dots, blue diamonds, and green triangles are correlations under the no odor, early odor, and later odor conditions, respectively. The statistical metrics are as follows: no odor, PCC = 0.58, p = 0.022; early odor, PCC = 0.080, p = 0.81; later odor, PCC = 0.069, p = 0.77 (no odor, n = 15; early odor, n = 11; later odor, n = 21).

(F) Glutamate decreases with and without Ca²⁺ spikes. The filled and hollow box plots indicate decreases with and without a spike, respectively. Box plots indicate the median (center line), quartiles (boxes), and range (whiskers). The statistical metrics are as follows: no odor, p = 0.0062; later odor, p = 0.037; Welch's t test (N = 15; no odor, spike, n = 15; no odor, no spike: n = 89; later odor, spike, n = 21; later odor, no spike, n = 69) *p < 0.05, **p < 0.01.

To support the idea that decreases in glutamate induce a Ca²⁺ response, we investigated whether Ca²⁺ increases when glutamate decreases. We collected and aligned Ca²⁺ traces with the timing of decreases in glutamatergic inputs (Figure 4A) and found that Ca²⁺ reliably increases when glutamatergic input decreases, with or without odor (Figure 4B). Moreover, the degree of increase was significantly larger than in the randomly resampled data (Figures 4C and 4D). These findings show that Ca²⁺ increases when glutamate decreases and support the idea that decreases in glutamate induce a Ca²⁺ response.

Figure 4.

Ca²⁺ Response to Decreases in Glutamate Input

(A) Ca²⁺ response to decreases in glutamate input. Time 0 indicates the time when the glutamate decreases. All Ca²⁺ responses were normalized to zero at time 0. The red and yellow lines indicate the response under the no odor and later odor conditions, respectively. The shadows indicate the SEM (for the Ca²⁺ response under later odor condition, the shadows are too small to see; N = 15; no odor, n = 43; later odor, n = 61).

(B) Ca²⁺ intensities before (“pre”; averaged from −3 to −2 s in A) and after (“onset”; averaged from −0.5 to +0.5 s in A) glutamate decrease. The statistical metrics are as follows: no odor, p = 0.014; later odor, p = 0.014; paired t test (N = 15; no odor, n = 43; later odor, n = 61). *p < 0.05.

(C) Ca²⁺ responses to decreases in glutamate input relative to randomly selected data with (right) and without odor (left). The gray lines indicate the averaged randomly selected data ± 3 SD (see Methods).

(D) Ca²⁺ increases with glutamate decreases (colored boxes) and by random selection (gray boxes). Box plots indicate the median (center line), quartiles (boxes), and range (whiskers). The statistical metrics are as follows: no odor, p = 0.045; later odor, p = 0.0015. Wilcoxon rank-sum test (N = 15; no odor, decrease, n = 43; no odor, random, n = 100; later odor, decrease, n = 61; later odor, random, n = 100). *p < 0.05, **p < 0.01.

The GLC-3 Receptor Primarily Mediates a Sporadic Ca²⁺ Response in AIY under Odor Stimulation

We demonstrated that glutamate decreases when Ca²⁺ increases, and vice versa. Previous research suggests that the glutamate-gated chloride channel, GLC-3, is crucial for the sensory-input-evoked Ca²⁺ response in AIY (Aoki et al., 2017, Chalasani et al., 2007, Horoszok et al., 2001, Ohnishi et al., 2011). Therefore, we investigated the relationship between Ca²⁺ and glutamate input in a glc-3 mutant. We detected the same type of changes in glutamate input following odor stimulation in glc-3 mutants as in wild-type (N2) animals (Figures 5A and 5B), and found that GLC-3 mutation had a limited effect on the Ca²⁺ response evoked by odor application (see glc-3 in Figure 6C). However, the Ca²⁺ spike frequency decreased significantly compared with N2, both with and without odor stimulation (see glc-3 in Figure 6D). Next, glutamatergic inputs were collected and aligned with the timing of Ca²⁺ spikes, as was done for the N2 animals (Figure 5C). Glutamate decreases were abolished under odor stimulation, but not in the absence of odor stimulation (Figure 5D). The extent of the decrease when accompanied by a Ca²⁺ spike was not significantly different from that without a Ca²⁺ spike under the odor stimulation, but was significantly larger in the absence of odor stimulation (Figure 5E). AIY neurons have a metabotropic glutamate receptor, MGL-1 (Jeong and Paik, 2017, Kang and Avery, 2009), which could play a role in initiating Ca²⁺ spikes. These results suggest that fluctuations in glutamate input via the GLC-3 receptor are important for the sporadic Ca²⁺ spikes.

Figure 5.

Ca²⁺ Responses and Glutamate Inputs in glc-3 Mutants

(A) Averaged glutamate inputs (blue) and Ca²⁺ responses (red) to odor. The shaded region indicates IAA stimulation, and the shadows around the solid lines indicate the SEM (N = 12).

(B) Glutamate intensities before (pre-addition) and after (post-addition) odor stimulation onset (left), and average intensities before (pre-removal) and after (post-removal) odor stimulation removal (right). The statistical metrics are as follows: left, p = 0.0030; right, p = 0.0018; paired t test (N = 12). **p < 0.01.

(C) Averaged glutamate responses coincident with Ca²⁺ spikes. Time 0 indicates the onset of Ca²⁺ spikes. All glutamate responses were normalized to zero at time 0. The purple, blue, and green lines indicate the responses under the no odor, early odor, and later odor conditions, respectively. Shadows indicate the SEM (N = 12, no odor, n = 15; early odor, n = 6; later odor, n = 20).

(D) Glutamate input before the Ca²⁺ spike and at spike onset. “Pre” indicates the average intensity from −3 to −2 s in (C), and “onset” indicates the average intensity from −0.5 to +0.5 s. The statistical metrics are as follows: no odor, p = 0.0012; early odor, p = 0.65; later odor, p = 0.89; paired t test (N = 12, no odor, n = 15; early odor, n = 6; later odor, n = 20). **p < 0.01.

(E) Averaged glutamate decreases with (filled boxes) and without (hollow boxes) Ca²⁺ spikes. Box plots indicate the median (center line), quartiles (boxes), and range (whiskers). The statistical metrics are as follows: no odor, p = 0.0027; later odor, p = 0.58; Welch's t test (N = 12, no odor, spike, n = 15; no odor, no spike, n = 114; later odor, spike, n = 20; later odor, no spike, n = 82). **p < 0.01.

Glutamate Input Evokes a Sporadic Ca²⁺ Response in AIY

We thus showed that the GLC-3 receptor plays a crucial role in the sporadic Ca²⁺ spikes. We used a glutamate-defective mutant, eat-4, to directly reveal the effect of glutamate input on AIY. EAT-4 is a vesicular glutamate transporter, and eat-4 mutants are defective in glutamate transmission (Chalasani et al., 2007, Lee et al., 1999). In contrast to the N2 animals and glc-3 mutants, the glutamate response following odor application and removal was abolished (Figures 6A and 6B). Moreover, the Ca²⁺ response to odor stimulation was also abolished (Figure 6C), and in fact the Ca²⁺ spike rarely occurred at all, with or without odor stimulation (Figure 6D). We collected and aligned glutamate input with the timing of Ca²⁺ spike initiation (Figure 6E), but no decreases were observed (Figure 6F), and the extent of decrease in the presence of a Ca²⁺ spike did not differ from that without a Ca²⁺ spike (Figure 6G). These results show that glutamate-input fluctuations evoke sporadic Ca²⁺ spikes in AIY neurons.

Discussion

In this article, we present two sets of simultaneous imaging investigations into the effects of environmental fluctuations on the interaction between Ca²⁺ and membrane potentials and Ca²⁺ and neurotransmitter (glutamate) inputs in AIY interneurons. By combining this approach with the use of mutant strains, we succeeded in determining the input-output relationship. We showed that membrane-potential spikes precede Ca²⁺ spikes in AIY, a previously unknown relationship. Moreover, decreases in glutamate in response to environmental fluctuations evoke sporadic Ca²⁺ spikes. We thus succeeded, for the first time, in identifying the relationship between input and output fluctuations in response to environmental fluctuations in vivo.

Glutamate inputs to AIY represent sensory neuronal outputs (Chalasani et al., 2007, Serrano-Saiz et al., 2013, White et al., 1986), and Ca²⁺ responses in AIY correlate with velocity and turn behavior (Li et al., 2014). Therefore, our results suggest that fluctuations in sensory input induce behavioral variability. It is difficult to address the relationship between sensory input and behaviors under environmental noise, because of the low signal-to-noise ratio (Faumont et al., 2012). Our results thus provide novel insights into the relationship between sensory-input fluctuation and behavioral variability.

In previous research, noise has been applied artificially (Nienborg and Cumming, 2009, Renart and Machens, 2014) and is thus not reflective of real environmental fluctuation. Controlling noise is an appropriate strategy for understanding perception, but it is not suitable for evaluating spontaneous variability. Here, we showed that Ca²⁺ spikes are caused by glutamate-input fluctuations. AIY neurons play an important role in sensory integration (Chalasani et al., 2007, Clark et al., 2006, Yoshida et al., 2012) and regulate behavior, and Ca²⁺ responses are correlated with reversal behavior and locomotion speed (Kocabas et al., 2012, Li et al., 2014). Therefore, changing behaviors under natural conditions are presumably caused by fluctuations in sensory input under environmental noise. Our approach of evaluating the input-output relationship in AIY under environmental fluctuations aids understanding of resting-state dynamics in complex neural circuits.

We identified the relationship between neurotransmitter inputs and the Ca²⁺ response. In previous research (Branco and Häusser, 2010, Callaway and Katz, 1993, Kramer et al., 2013), caged compounds and the stimulation of presynaptic neurons were used to investigate these relationships. However, such stimulation is often intense and artificial, and neurotransmitter inputs are not directly measurable. To address these shortcomings, we first demonstrated the relationship between neurotransmitter inputs and Ca²⁺ responses under naturally occurring physiological conditions, and then quantified the input-output relationships using simultaneous imaging of neurotransmitter inputs and neuronal activity.

Limitations of the Study

Here, we identified the relationship between glutamate input and the Ca²⁺ response. In the eat-4 mutants, however, the Ca²⁺ responses were rare but were not completely abolished. We acknowledge that other neurotransmitters, such as neuropeptides, also affect the Ca²⁺ response in AIY (Aoki et al., 2017, Kuhara et al., 2011, Rabinowitch et al., 2016). Although our results show that decreases in glutamate evoke Ca²⁺ responses in AIY under natural noise, cooperation between several neurotransmitters could modulate the responsiveness of AIY to glutamate inputs, which would modulate sensory information and the behavioral response.

In this study, we revealed the relationships with and without odor condition. However, it is difficult to generalize for the other sensory stimulation, and modulation of the other environmental modalities could have yielded different results. In addition to the different neurotransmitters involved, AIY neurons receive glutamate from several types of sensory neurons (Clark et al., 2006, Kuhara et al., 2011, Kunitomo et al., 2013, Serrano-Saiz et al., 2013, Wang et al., 2017, White et al., 1986). For example, high concentration of salt affects odor sensory neurons (Leinwand and Chalasani, 2013). So, the other sensory stimulation could change the input-output relationship in AIY.

We showed that mutation of glc-3 and eat-4 reduced the frequency of Ca²⁺ spikes. Because GLC-3 is a glutamate-gated chloride channel (Horoszok et al., 2001), ablation of the glutamate signal should depolarize the membrane potential of AIY neurons. However, shapes of Ca²⁺ spikes were identical (Figure S2). This indicates that AIY neurons retain the function of Ca²⁺ spike generation even without glutamate inputs. This suggests that the AIY neurons should have maintained their equilibrium potential even in the mutant animals because depolarization is necessary for generating Ca²⁺ spikes (Liu et al., 2018, Shindou et al., 2019). In the mutants, glutamate inputs could not cause the depolarization of membrane potential beyond the threshold in response to environmental fluctuations by glutamate inputs, so the Ca²⁺ spike frequency decreased. However, the mutations could change the physiological property in AIY; we could not dismiss these effects in the mutants.

We also noted that C. elegans does not have voltage-gated Na⁺ channels in its neurons (Faumont et al., 2011, Faumont et al., 2006, Gao and Zhen, 2011, Goodman et al., 1998, Mellem et al., 2008, Shidara et al., 2013). This fact indicates that the mechanism of spike generation should be different from that of the other animals. Therefore, it is difficult to directly extrapolate our conclusions beyond C. elegans.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: September 27, 2019