What can a worm learn in a bacteria-rich habitat?

Abstract

With a nervous system that has only a few hundred neurons, Caenorhabditis elegans was initially not regarded as a model for studies on learning. However, the collective effort of the C. elegans field in the past several decades has shown that the worm displays plasticity in its behavioral response to a wide range of sensory cues in the environment. As a bacteria-feeding worm, C. elegans is highly adaptive to the bacteria enriched in its habitat, especially those that are pathogenic and pose a threat to survival. It uses several common forms of behavioral plasticity that last for different amounts of time, including imprinting and adult-stage associative learning, to modulate its interactions with pathogenic bacteria. Probing the molecular, cellular and circuit mechanisms underlying these forms of experience-dependent plasticity has identified signaling pathways and regulatory insights that are conserved in more complex animals.

Keywords:

• Caenorhabditis elegans
• learning and memory
• behavioral plasticity

Caenorhabditis elegans senses and responds to diverse environmental cues

Animals live in different ecological niches that are characteristic of different chemical, physical and biological cues and have likely evolved sensorimotor systems that are able to detect and respond to the environmental conditions of their habitats. C. elegans feeds on bacteria and is often found in decaying fruits or other organic matters that are rich in bacteria (Felix & Duveau, 2012; Frezal & Felix, 2015; Samuel, Rowedder, Braendle, Felix, & Ruvkun, 2016). It navigates its environment by detecting and responding to various chemical cues, including odorants and salts, temperature, pheromones, gases, as well as mechanical stimuli [(Figure 1) and (Aoki & Mori, 2015; Bargmann, 2006; Brandt et al., 2012; Bretscher, Busch, & de Bono, 2008; Butcher, Fujita, Schroeder, & Clardy, 2007; Chalfie, 2009; Cheung, Cohen, Rogers, Albayram, & de Bono, 2005; de Bono & Maricq, 2005; Goodman et al., 2014; Goodman & Sengupta, 2019; Gray et al., 2004; Hallem et al., 2011; Hao et al., 2018; Jeong et al., 2005; Kaplan & Horvitz, 1993; Kim et al., 2009; Macosko et al., 2009; Pierce-Shimomura, Faumont, Gaston, Pearson, & Lockery, 2001; Reddy, Hunter, Bhatla, Newman, & Kim, 2011; Schafer, 2015; Srinivasan et al., 2008; 2012; White et al., 2007; White & Jorgensen, 2012)]. Some of the odorants that are attractive to C. elegans can be produced by plants and may serve as cues representing an environment that is abundant in bacteria. In addition, C. elegans is known to navigate within a thermal gradient and the ambient temperature significantly regulates the development and life span of a worm. The sensorimotor response to chemical cues and temperature have been extensively studied in C. elegans. For example, a few ciliated sensory neurons use G-protein coupled seven-transmembrane receptors and cyclic nucleotide-gated channels (CNGs), to detect and mediate responses to odorants. The calcium-permeable CNGs transform odorant information into intracellular signals, which produce intercellular signals to engage postsynaptic interneurons and downstream motor neurons to generate movement towards or away from the odorants (Bargmann, 2006; de Bono & Maricq, 2005). Similarly, the major sensory neurons, as well as their intracellular signaling pathways, that perceive and respond to external salt concentration, ambient temperature gradient, pheromones, gases and mechanical cues have been identified and characterized (Aoki & Mori, 2015; Bargmann, 2006; Brandt et al., 2012; Bretscher et al., 2008; Butcher et al., 2007; Chalfie, 2009; Cheung et al., 2005; de Bono & Maricq, 2005; Goodman et al., 2014; Goodman & Sengupta, 2019; Gray et al., 2004; Hallem et al., 2011; Hao et al., 2018; Jeong et al., 2005; Kaplan & Horvitz, 1993; Kim et al., 2009; Macosko et al., 2009; Pierce-Shimomura et al., 2001; Reddy et al., 2011; Schafer, 2015; Srinivasan et al., 2008; 2012; White et al., 2007; White & Jorgensen, 2012). The behavioral strategies and the underlying neural circuits through which C. elegans navigates a sensory environment are also intensively investigated (Aprison & Ruvinsky, 2019; Bargmann, 2006; Chalasani et al., 2007; de Bono & Maricq, 2005; Donnelly et al., 2013; Goodman & Sengupta, 2019; Gordus, Pokala, Levy, Flavell, & Bargmann, 2015; Gray, Hill, & Bargmann, 2005; Iino & Yoshida, 2009; Ikeda et al., 2020; Jang et al., 2012; Kaplan, Salazar Thula, Khoss, & Zimmer, 2020; Kato et al., 2015; Kunitomo et al., 2013; Li, Liu, Zheng, & Xu, 2014; Liu et al., 2018; Luo et al., 2014; Macosko et al., 2009; Mori & Ohshima, 1995; Pierce-Shimomura, Morse, & Lockery, 1999; Schafer, 2015; Tsalik & Hobert, 2003; Venkatachalam et al., 2016; Wen, Gao, & Zhen, 2018; White et al., 2007). The molecular, cellular and circuit bases for these sensorimotor responses provide the substrates for experience-dependent regulation. The studies that investigate various forms of learning in C. elegans have been reviewed elsewhere (Alcedo & Zhang, 2013; de Bono & Maricq, 2005; McDiarmid, Yu, & Rankin, 2019; Sasakura & Mori, 2013). Here, we will focus on several learning paradigms that regulate the interaction between C. elegans and pathogenic bacteria.

Figure 1. Diverse adaptive behaviors in response to environmental cues in C. elegans.

Environmental cues induce plasticity across different timescales

C. elegans displays both adaptation and habituation, two common forms of non-associative learning (Figure 1). C. elegans is attracted to several chemical odorants, such as isoamyl alcohol and benzaldehyde; however, prolonged exposure to these volatile chemicals reduces the sensory response to the odorants and generates adaptation that lasts for a couple hours (Colbert & Bargmann, 1995; Inoue et al., 2013; Kaye, Rose, Goldsworthy, Goga, & L’Etoile, 2009). It is shown that during adaptation the endogenous small RNA (endo-siRNA)-mediated regulation of gene expression in the sensory neuron that detects isoamyl alcohol and benzaldehyde downregulates a guanylyl cyclase that is critical for the G-protein coupled signaling pathway underlying the sensing of the odorants (Juang et al., 2013). These results reveal a novel function of endo-siRNA pathways in regulating gene expression in response to olfactory experience. In addition, the worm reverses from a mild mechanical stimulus that is delivered to its body or nose and senses the stimulus using receptor neurons, several of which contain distinct morphological features (Chalfie, 2009; Kaplan & Horvitz, 1993; Schafer, 2015). Tapping the cultivating plate also generates mechanical stimuli that trigger reversals. However, tapping for multiple times reduces the amplitude of the reversals (Rankin, Beck, & Chiba, 1990). This type of behavioral changes is analogous to habituation previously characterized in Aplysia and cats, where multiple stimulations with a benign mechanical stimulus lead to a reduction in response (Bailey & Chen, 1983; Spencer, Thompson, & Neilson, 1966). Repeated habituation training under certain conditions can generate memory that lasts for 24 h (Rose, Kaun, & Rankin, 2002).

In addition to non-associative learning, previous studies have shown that olfactory responses can be respectively enhanced or weakened by paring odorant exposure with the presence or absence of food, which presumably represents an appetitive or aversive environment (Figure 1). Various neuronal circuits and molecular pathways have been characterized in regulating these associative learning behaviors [(Alcedo & Zhang, 2013; de Bono & Maricq, 2005) and the references therein]. C. elegans also remembers the salt concentration under its cultivation condition and seeks this concentration when tested in a salt gradient after the training. However, if the worm is kept at a salt concentration in the absence of food, it avoids the concentration during the post-training rest (Kunitomo et al., 2013; Luo et al., 2014; Saeki, Yamamoto, & Iino, 2001; Tomioka et al., 2006). As a critical condition, the cultivation temperature significantly modulates the navigation of the worm in a temperature gradient (Aoki & Mori, 2015; Biron et al., 2006; Goodman et al., 2014; Goodman & Sengupta, 2019; Hedgecock & Russell, 1975; Mori & Ohshima, 1995). Some of these forms of behavioral plasticity resemble associative learning identified in vertebrate animals and in fruit flies. While a one-time massed training in these paradigms often generates a memory for a couple hours, spaced training can generate a long-term memory that lasts for 16 h (Kauffman, Ashraf, Corces-Zimmerman, Landis, & Murphy, 2010).

Experimental power of C. elegans facilitates dissection of plasticity mechanisms

The ease of using forward and reverse genetic approaches to characterize gene function in C. elegans and the knowledge of genetic identities and synaptic connections of the worm neurons facilitate studies on learning and behavioral plasticity in C. elegans in several important ways:

1. Because the worm neurons are defined in their genetic making and synaptic connection, we are able to identify the neurons where the gene products implicated in learning are generated and act, as well as their presynaptic and post-synaptic neurons. These analyses provide us with knowledge on neuronal circuits underlying various forms of learning behaviors.

2. By applying in vivo imaging and genetic manipulations, we can identify experience-dependent changes in the activity and the connection of the learning circuit that are correlated with behavioral changes and characterize the causality of these changes in generating learned behavior.

3. Once the key neurons underlying learning are identified, we can also analyze gene expression in these neurons in naive and trained animals in order to identify genes that display training-correlated changes in their expression and address the function of these molecules in learning.

4. Meanwhile, the ease of performing genetic analyses in the C. elegans nervous system also makes it feasible to conduct genetic screens in order to identify new functions of characterized genes and pathways in learning, as well as identify new genes with previously unknown functions.

Interactions with pathogenic bacteria that modulate behavior

C. elegans feeds on bacteria in the wild and laboratories. A wide range of different bacteria strains, including many in the Pseudomonas genus, are found to be associated with C. elegans isolated from its natural habitats (Felix & Duveau, 2012; Frezal & Felix, 2015; Samuel et al., 2016; Schulenburg & Felix, 2017). While some of these bacteria serve as food sources, others are pathogenic and kill C. elegans through infections or with secreted toxins [(Hoffman & Aballay, 2019; Irazoqui, Urbach, & Ausubel, 2010; Kim & Ewbank, 2018) and the references therein]. Because bacteria play a vital role in the development and survival of C. elegans, it is conceivable that C. elegans has evolved diverse strategies to mediate its interactions with the environmental bacteria.

Bacteria produce multiple types of sensory cues that can be used by the worm to detect and respond to the microbes. In addition to odorants, bacteria also produce water-soluble metabolites, generate or alter concentration of gases. The border and the texture of a bacterial lawn may also generate mechanical stimulation to moving worms. These bacteria-derived sensory cues act in a combinatorial manner to elicit behavioral responses in C. elegans [(Bargmann, Hartwieg, & Horvitz, 1993; Brandt & Ringstad, 2015; Bretscher et al., 2008; Calhoun et al., 2015; Cheung et al., 2005; Flavell et al., 2013; Kim & Flavell, 2020; Gramstrup Petersen et al., 2013; Gray et al., 2004; Ha et al., 2010; Hallem et al., 2011; Hao et al., 2018; Harris et al., 2019; Meisel, Panda, Mahanti, Schroeder, & Kim, 2014; Ooi & Prahlad, 2017; Pradel et al., 2007; Reddy et al., 2011; Rhoades et al., 2019; Sawin, Ranganathan, & Horvitz, 2000; Tran et al., 2017) and the references therein]. The diversity of the sensory cues is consistent with multiple signaling pathways that are identified to mediate bacteria-worm interactions.

Adult-stage learning of pathogenic bacteria

Some pathogenic bacteria, such as the Pseudomonas aeruginosa strain PA14, infect C. elegans after being ingested, which leads to a slow death of the worm over several days (Tan, Mahajan-Miklos, & Ausubel, 1999). Thus, pathogenic bacteria likely signal both food and danger to the worm. The odorants of several pathogenic bacteria, including PA14, are attractive to the worms that are cultivated standard conditions by feeding on E. coli OP50 at 20 − 22 °C (Ha et al., 2010; Jin, Pokala, & Bargmann, 2016; Zhang, Lu, & Bargmann, 2005). When newly transferred to a lawn of PA14, worms feed on the lawn. However, in the next few hours worms start to leave the lawn [Figure 2 and (Chang, Paek, & Kim, 2011)]. The virulence of the bacteria, as well as several bacteria-derived chemicals act together to repel the worms from the lawn. The mechanisms underlying the changes in behavior and physiology of the worms over this process are separately reviewed (Hoffman & Aballay, 2019; Kim & Flavell, 2020; Kim & Ewbank, 2018; Meisel & Kim, 2014).

Figure 2. Schematic diagrams showing assays for behavioral responses to PA14. (A) A small drop of supernatant of PA14 culture is used as the source of odorants to examine attractive steering movements, when a worm starts from a position relatively close to the odorant source. (B) To test the relative preference between the odorants of PA14 and the odorants of E. coli OP50, two airstreams saturated with the odorants of OP50 or the odorants of PA14 are used to deliver alternating stimuli to individual worms swimming in a small drop of buffer in an airtight chamber. (C, D) Two small drops of supernatant of bacterial culture are put on a plate immediately before the assay (C) or two small drops of bacterial culture are quickly air-dried before the assay (D) to measure the preference between the two odorant mixtures in a single worm (C) or a population of worms (D). In D, the plate does not contain peptone and therefore does not support growth of the bacteria. (E, F) A small plate with two bacteria lawns grown on the plate for a few hours (E) or for 24–48 hours (F) to be used as odorant sources. During cultivation, the bacterial lawns may produce cues diffused into the medium, produce or alter the concentration of gases in the lawn areas. (G, H) A bacterial lawn centered on a small plate (G) or completely covering a small plate (H) prepared by fully growing first at 37 °C and then at 25 °C to examine the lawn avoidance/occupancy or survival of the worms over time.

Do worms learn to associate the aversiveness of PA14 with sensory cues produced by the pathogenic bacteria to generate retrievable memory of the bacterium? This question can be addressed by testing the response to PA14-derived sensory cues in the naive, i.e. E. coli-raised, worms and the trained, i.e. PA14-fed, worms [(Ha et al., 2010; Jin et al., 2016; Liu et al., 2018; Zhang et al., 2005) and Figure 2(A–F)]. Previously, by probing the worms with an assay that resembles the chemotaxis assay on olfactory responses or with an assay that uses airstreams saturated with the odorants of tested bacteria, it is shown that after feeding on PA14 for 4–6 h, adult worms learn to reduce their preference for the odorants of the bacterium (Figure 2(A–D)) (Ha et al., 2010; Jin et al., 2016; Liu et al., 2018; Zhang et al., 2005). This type of learning in the adult C. elegans is contingent on the pathogenicity of the training bacteria and a serotonin signal. The training-dependent change in the olfactory response is specific for the odorants of the training bacteria (Figure 2(A–C)) (Choi, Liu, Wu, Yang, & Zhang, 2020; Ha et al., 2010; Jin et al., 2016; Liu et al., 2018; Zhang et al., 2005). Together, these results indicate a learned association between pathogenicity and the odorants of the training bacterium. Since pathogenic bacteria represent a critical constraint to the survival of C. elegans, it is conceivable that C. elegans has evolved the ability to associate the olfactory cues of some pathogenic bacteria with the virulence, which regulates subsequent interactions with the pathogens. The learned behavioral response is reversible, which suggests a temporary modulation of the nervous system (Ha et al., 2010; Jin et al., 2016; Liu et al., 2018; Zhang et al., 2005).

Several different methods have been used to analyze the changes in behavioral responses to pathogenic bacterium PA14 after feeding on PA14 [Figure 2 and (Ha et al., 2010; Hao et al., 2018; Horspool & Chang, 2017; Jin et al., 2016; Lee & Mylonakis, 2017; Liu et al., 2018; Ma, Zhang, Dai, Khan, & Zou, 2017; Meisel et al., 2014; Miller, Grandi, Giannini, Robinson, & Powell, 2015; Moore, Kaletsky, & Murphy, 2019; Ooi & Prahlad, 2017; Singh & Aballay, 2019; Wolfe et al., 2019; Zhang et al., 2005)]. Some of these methods differ in the types of the sensory cues that they examine, the spatial and temporal patterns of the cues, and the behavioral strategies that they measure (Figure 2). For example, the behavioral strategies used to distinguish between simultaneously present odorants from two bacterial strains (Figure 2(C–F)) are likely different from those used to respond to two alternating odorant stimuli (Figure 2(B)). The odorant gradient established by a point source is not linear (Tanimoto et al., 2017). Therefore, the size of the testing plate is important for the intensity and the spatial pattern of the bacterial odorants sensed by the worms (Figure 2(C and D)). The duration for which the testing bacteria were placed or grown on the testing plate or the temperature used to cultivate the testing bacteria generates different mixtures of the sensory cues (Figure 2(E and F)). In addition, separating the training process from the testing assay makes it possible to examine whether a retrievable memory is formed. Furthermore, the assay that measures the occupancy of a bacterial lawn grown on an assay plate likely measures sensory responses elicited by metabolites that are not volatile and mechanical cues, in addition to olfactory responses (Figure 2). While worms feeding on a lawn of pathogenic bacterium PA14 gradually leave the lawn over time (Figure 2(G and H)), it usually takes longer for the worms to significantly leave the lawn than to learn to reduce the preference for the odorants of PA14. Because different molecular and neuronal apparatus are employed to detect and generate behavioral responses to these various types of sensory information, studies using these different assay conditions highlight the robustness of the behavioral responses and potentially allow us to examine different pathways through which the worm interacts with pathogenic bacteria.

Imprinting

Interestingly, training the worms by feeding on pathogenic bacteria during the first larval stage (L1) for 12 h forms the aversive memory of the odorants of the pathogens that can be retrieved during the adult stage [Figure 1 and (Jin et al., 2016)]. This form of memory is comparable to the imprinted memory characterized in various vertebrate animals (Lorenz, 1935; Nevitt, Dittman, Quinn, & Moody, 1994; Wilson & Sullivan, 1994). The worms imprint not only the odorants associated with the aversive experience, but also the odorants associated with food sources to form a long-lasting appetitive memory [Figure 1 and (Remy & Hobert, 2005)]. Mapping the neural circuits for the imprinting of pathogen odorants and the retrieval of the aversive memory, which take place two days apart, show that different circuits subserve learning and retrieval (Jin et al., 2016). In addition to the odorants that often represent food, pheromones also signal significant environmental conditions, such as the density of the conspecifics, to the worm (Butcher et al., 2007; Jeong et al., 2005; Macosko et al., 2009; Srinivasan et al., 2008; 2012; White et al., 2007). Exposing the worms during the L1 stage to a repulsive pheromone enhances the avoidance of the pheromone during the adult stage by strengthening the synaptic connection between a pheromone-sensing neuron and its downstream motor neurons [Figure 1 and (Hong et al., 2017)]. Starvation during the L1 stage also profoundly alters the wiring of the nervous system by regulating neurotransmitters that respond to food availability (Bayer & Hobert, 2018). It is conceivable that during the early larval development when the nervous system is being formed, strong neuronal activities in response to environmental conditions reprogram the developmental process to generate persistent changes. A couple of studies show that harsh conditions during development systematically regulate gene expression and modulate the anatomy and activity of the nervous system, which produce behavioral changes that last into the adult stage. For example, when the worm density is high and food is relatively sparse, the larval worms can enter a diapause state, dauer, that halts the development for days until the conditions improve (Golden & Riddle, 1984). Adult animals that have experienced the dauer stage exhibit distinct behavioral traits, including those important for food seeking. The molecules that regulate chromatin structures and endogenous RNAi pathways mediate dauer formation, which potentially modulate the expression of genes underlying dauer-inducing changes in anatomy and behavior (Bharadwaj & Hall, 2017; Hall, Beverly, Russ, Nusbaum, & Sengupta, 2010; Ow, Borziak, Nichitean, Dorus, & Hall, 2018; Pradhan, Quilez, Homer, & Hendricks, 2019).

Intergenerational effects

Because pathogenic bacteria serve as food sources and critical survival constraints to the worm, it is plausible that exposure to the pathogens modulates the nervous system and the behavior of the progenies. A recent study shows that training adult hermaphrodites by feeding on PA14 for 4 h, which is known to produce a robust aversive memory that associates PA14 odorants with virulence in the adult mothers, increases the progeny’s preference for the PA14 odorants (Pereira, Gracida, Kagias, & Zhang, 2020). Many animals prefer the food that they are exposed to in utero (Liu & Urban, 2017; Nehring, Kostka, von Kries, & Rehfuess, 2015; Todrank, Heth, & Restrepo, 2011). These results suggest that with 4-h exposure the food response elicited by PA14 is significant in the hermaphrodite mothers and that the resulting signals modulate the progeny developing in the uterus. Increasing the duration of PA14 exposure to 8 h enhances the infection to the mothers (Troemel et al., 2006) and reduces the preference for PA14 in the progeny (Pereira et al., 2020). These findings suggest that longer exposure to PA14 induces a stronger response to the pathogenicity of PA14, which changes the response of the progeny to PA14 from attraction to avoidance. While 4 to 8-h parental training with PA14 produces robust aversive memory in the hermaphrodite mothers, their modulatory effects on the progeny are limited to the first generation of the offspring (Pereira et al., 2020).

Further increasing the duration of PA14 exposure to 24 h that starts at the L4 larval stage not only generates avoidance of PA14 in the exposed mothers, but also produces PA14 avoidance in the progeny for 4 generations through a separate mechanism (Moore et al., 2019). However, both the effect of 4-h training on the progeny and the multi-generational effect require the endo-RNAi pathway and the piRNA pathway (Moore et al., 2019; Pereira et al., 2020). Small RNA pathways also mediate olfactory adaptation by regulating the expression of genes critical for odorant sensation (Juang et al., 2013). However, these behavioral changes differ in their durations and modulatory effects, which suggest distinct mechanisms through which the underlying small RNA pathways alter the nervous system and behavior.

In addition to food-seeking related behavior, exposing the worms to pathogenic bacteria generates intergenerational effects on other physiological traits critical for survival. For example, exposure to Pseudomonas vranovensis, another bacterium pathogenic to C. elegans, generates multigenerational effects and enhances immune resistance to the pathogen in the offspring (Burton et al., 2019). Meanwhile, exposure to certain pathogenic bacteria for two consecutive generations induces formation of dauers, a dormant development stage that is highly resistant to environmental stresses (Palominos et al., 2017). These studies reveal multiple ways that the worm has evolved to adapt its development and function to the bacteria in its environment. Interestingly, pathogenic bacteria are not the only environmental conditions that impact the worm for multiple generations. It has been shown that parental experiences, including dietary restriction, osmotic stress, temperature changes, olfactory imprinting, and prolonged starvation, can regulate the physiology of the offspring, some of which last for several generations and are mediated by small RNA pathways (Burton et al., 2017; 2019; Das et al., 2020; Demoinet, Li, & Roy, 2017; Greer et al., 2011; Hibshman, Hung, & Baugh, 2016; Jobson et al., 2015; Klosin, Casas, Hidalgo-Carcedo, Vavouri, & Lehner, 2017; Ni et al., 2016; Palominos et al., 2017; Posner et al., 2019; Rechavi et al., 2014; Remy, 2010; Schott, Yanai, & Hunter, 2014). These studies together show that C. elegans has evolved diverse adaptive strategies to generate long-term plasticity that lasts for multiple generations. The short lifespan of C. elegans and the ease to conduct genetic experiments on it makes this line of research productive. It will be informative to compare the mechanistic insights identified in studies on intergenerational effects in different animals to understand the difference and similarity in the logic of these regulations.

Outlook

C. elegans lives in a bacteria-rich environment that represents a vast amount of opportunities and constraints to its survival, reproduction and evolution. We are at the beginning stage to understand its impressive adaptive responses encoded in a compact genome. In addition to its responses to pathogenic bacteria, several recent studies also reveal interesting interactions between C. elegans and its commensal bacteria species. These studies show that C. elegans utilizes neurotransmitters or vitamins produced by the environmental bacteria (O’Donnell, Fox, Chao, Schroeder, & Sengupta, 2020; Urrutia et al., 2020; Wei & Ruvkun, 2020) to maintain or modulate various physiological events and neural functions. These findings together with those investigating the interactions of C. elegans and pathogenic bacteria have established C. elegans as a promising system to probe mechanisms underlying gut-brain interactions. Together, these studies allow us to leverage the experimental powers provided by a model organism to investigate the function of the nervous system in an ethological and evolutionary context.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The work in the Zhang laboratory is supported by The National Institutes of Health [DC009852, R21MH117386, NS115484].