1-Undecene from Pseudomonas aeruginosa is an olfactory signal for flight-or-fight response in Caenorhabditis elegans.

Animals possess conserved mechanisms to detect pathogens and to improve survival in their presence by altering their own behavior and physiology. Here, we utilize Caenorhabditis elegans as a model host to ask whether bacterial volatiles constitute microbe-associated molecular patterns. Using gas chromatography-mass spectrometry, we identify six prominent volatiles released by the bacterium Pseudomonas aeruginosa. We show that a specific volatile, 1-undecene, activates nematode odor sensory neurons inducing both flight and fight responses in worms. Using behavioral assays, we show that worms are repelled by 1-undecene and that this aversion response is driven by the detection of this volatile through AWB odor sensory neurons. Furthermore, we find that 1-undecene odor can induce immune effectors specific to P. aeruginosa via AWB neurons and that brief pre-exposure of worms to the odor enhances their survival upon subsequent bacterial infection. These results show that 1-undecene derived from P. aeruginosa serves as a pathogen-associated molecular pattern for the induction of protective responses in C. elegans.

Introduction

Threat perception is one of the key drivers of behavioral and physiological responses in living organisms. Animals use their senses of vision, taste, auditory perception, and olfaction to perceive threats and engage in flight‐or‐fight mechanisms to improve survival. In case of infection, the sensing of pathogens via pattern recognition initiates physiological responses in the host including immune response to enhance survival. The chemical nature of pattern varies greatly from bacterial lipopolysaccharides to peptides or even volatile compounds (Stensmyr et al, 2012; Bufe et al, 2019). Volatiles are increasingly being recognized as a mode of communication between plants in response to herbivory and other threats (Erb et al, 2015). However, the contribution of volatiles in driving immune responses in animals remains poorly understood. In this study, we have used bacterivorous nematode Caenorhabditis elegans to identify volatiles that serve as molecular pattern to alter host behavior and immune response.

Caenorhabditis elegans forages for food bacteria in decaying organic matter (Schulenburg & Félix, 2017). A well‐developed chemosensory system including an odor sensory system enables worms to efficiently engage in food search behavior as well as avoid pathogens (Shtonda & Avery, 2006; Pradel et al, 2007). C. elegans shows chemotaxis response to a wide range of volatile molecules including alcohols, ketones, amines, aldehydes, organic acids, and aromatic and heterocyclic compounds (Bargmann et al, 1993), many of these are product of bacterial secondary metabolism indicating their role in sensory perception (Worthy et al, 2018; Foster et al, 2020).

Caenorhabditis elegans exhibits both fight and flight responses to Pseudomonas aeruginosa, a ubiquitous bacterium and an opportunistic human pathogen. Several studies on the interaction of C. elegans with P. aeruginosa have led to the elucidation of innate immune mechanisms that confer protection to the worms (Kim et al, 2002; Singh & Aballay, 2006; Estes et al, 2010; Irazoqui et al, 2010). The flight response, the ability of worms to avoid a lawn of P. aeruginosa, is a behavior termed “aversion” (Zhang et al, 2005; Styer et al, 2008). Anatomical changes in worm’s intestine during P. aeruginosa infection have been linked to aversion response of worms (Singh & Aballay, 2019). This behavior has also been linked to the sensation of water‐soluble secondary metabolites, pyochelin and phenazine‐1‐caboxamide of P. aeruginosa, by chemosensory neuron ASJ (Meisel et al, 2014). Evidence also suggests that odor sensory neurons regulate chemosensation and aversive olfactory learning during P. aeruginosa infection (Zhang et al, 2005) suggesting a possible involvement of volatile metabolites of P. aeruginosa in C. elegans response. The innate ability of worms to respond to the olfactory cues from P. aeruginosa was shown recently (Ooi & Prahlad, 2017). Additionally, genes involved in nitric oxide production in P. aeruginosa have been implicated in eliciting aversion in worms (Hao et al, 2018). However, the nature of P. aeruginosa volatile molecules which induce immune response by serving as microbe‐associated molecular pattern in C. elegans remains unclear.

In this study, we asked whether bacterial volatiles serve as molecular patterns. We utilized odor sensory mutants of C. elegans and specific volatiles from bacterium P. aeruginosa to study aversion response and immune response in the host. By analyzing volatile organic compounds produced by the bacterium, we identify 1‐undecene, an 11‐carbon olefin, as the aversive volatile signal. The volatile induces aversion response in C. elegans and calcium signaling in AWB odor sensory neurons. Finally, we show that 1‐undecene serves as a molecular pattern and induces upregulation of a subset of immune response genes specific to P. aeruginosa in worms, in AWB neuron‐dependent manner.

Results

Odor sensation controls C. elegans flight response to P. aeruginosa

In a systematic study of aversion response in C. elegans, we exposed adult worms to an old lawn of P. aeruginosa PA14 and scored lawn occupancy as a measure of aversion response every 4 h. The lawn occupancy of naive worms on the P. aeruginosa lawn reduced from 100% at 0 h to ~ 20% at 12 h of exposure indicating an increase in aversion with time (Fig 1A and B, Movie EV1A). To understand whether odor sensation in worms contributed to lawn leaving, we examined the aversion response of odr‐3(n2150) and odr‐3(n2046) olfaction defective mutants. ODR‐3 encodes a G protein alpha subunit required for normal chemotaxis to odorants (Bargmann et al, 1993; Roayaie et al, 1998). We found that odr‐3 mutants were defective in aversion response on P. aeruginosa lawn with high (60–90%) lawn occupancy even at 12 h of exposure (Fig 1A and B, Movie EV1B). Severely muted aversion response in odr‐3 mutant worms suggested that volatile cues from the P. aeruginosa lawn contribute to aversion response.

Figure 1

Odor sensation controls C. elegans flight response to P. aeruginosa

Time course of avoidance of P. aeruginosa lawn by N2, odr‐3(n2150), and odr‐3(n2046) worms. n ≥ 3 assays. ***P ≤ 0.001, ****P ≤ 0.0001 as determined by two‐tailed unpaired t‐test. Error bars indicate SEM.

Images of P. aeruginosa lawn at different time intervals showing avoidance response of N2 and odr‐3(n2150) mutant worms. Scale bar = 5 mm.

Kaplan–Meier survival curve of N2, odr‐3(n2150), and odr‐3(n2046) worms on partial lawn of P. aeruginosa at 25°C.

Kaplan–Meier survival curve of N2, odr‐3(n2150), and odr‐3(n2046) worms on full lawn of P. aeruginosa at 25°C.

Time required for 50% of worms to die (TD50) on partial lawn and full lawn of P. aeruginosa. Individual data points indicate replicates with ~ 100 worms each. ****P ≤ 0.0001, ns (not significant) P > 0.05 as determined by two‐tailed unpaired t‐test. n ≥ 3 assays. Error bars indicate SEM.

Schematic of young (8 h) and old (24 h) lawn choice assay in a tripartite plate and choice index. n ≥ 3 assays. Individual data points indicate replicates with ~ 40 worms each. *P ≤ 0.05 as determined by two‐tailed unpaired t‐test. Error bars indicate SEM.

Representative image of intestinal lumen width, indicated by arrows, in N2 worm exposed to P. aeruginosa for 12 h. Scale bar = 50 µm.

Quantification for the width of intestinal lumen in N2 exposed to E. coli and P. aeruginosa for 12 h. ns (not significant) P > 0.05 as determined by two‐tailed unpaired t‐test. n = 3. Error bars indicate SEM.

To understand whether olfaction‐dependent aversion response is protective, we analyzed the survival of N2 and odr‐3 mutant worms on a partial lawn (aversion allowed) and full lawn (aversion not allowed) of P. aeruginosa. While odr‐3 mutants were more susceptible to infection than N2 on the partial lawn, N2 and odr‐3 mutant worms had comparable survivals on the complete lawn of PA14 (Fig 1C–E), suggesting that olfaction‐dependent aversion response of worms enhances their survival. In a volatile‐driven choice assay between 8‐h (young) and 24‐h (old) lawn of P. aeruginosa, in a tripartite plate, we observed that N2 worms preferred the younger lawn over the older lawn, but this preference was absent in odr‐3(n2150) mutant (Fig 1F). This suggested that the older lawn of P. aeruginosa might produce volatile chemorepellents. It was shown recently that bloating in C. elegans gut upon long exposure of worms to P. aeruginosa leads to aversion response (Singh & Aballay, 2019). However, we found no bloating in N2 worms upon 12 h of exposure and the gut lumen width in E. coli and P. aeruginosa exposed worms were comparable (Fig 1G and H), suggesting that olfaction‐mediated aversion response seen at 12 h exposure is not linked to bloating in worms. Taken together, these results suggested that the volatiles produced by P. aeruginosa lawn induce a protective aversion response in olfaction‐competent C. elegans.

1‐Undecene from P. aeruginosa elicits aversive response in C. elegans

To identify the aversion‐inducing volatile(s) of P. aeruginosa, we performed solid‐phase microextraction (SPME) and gas chromatography–mass spectrometry (GC‐MS) analysis of volatiles. Volatiles were collected directly from the headspace between two plates containing P. aeruginosa lawns (schematic in Appendix Fig S1A). P. aeruginosa had six abundant volatiles, dimethyl sulfide (DMS), dimethyl disulfide (DMDS), pyrrole, 1,4‐dichlorobenzene, hexanoic acid, 2‐ethyl‐, methyl ester, and 1‐undecene (Fig 2A) that were absent in the media control and E. coli OP50 headspace (Appendix Fig S1B and C).

Figure 2

1‐Undecene from P. aeruginosa elicits aversive response in C. elegans

GC‐MS/MS profile of volatiles produced by 24‐h old lawn of P. aeruginosa PA14.

Chemotactic response of N2 worms to varying dilutions of 1‐undecene. n ≥ 3 assays. Error bars indicate SEM.

Omega turn in N2 worms under 1‐undecene exposure. Scale bar = 1 mm.

Schematic of PA14 (24 h) and undA mutant (24 h) bacterial lawn choice assay in a tripartite plate and choice index. n ≥ 3 assays. Error bars indicate SEM.

(i) Schematic of setup used to record the movement of 5 worms each under control and 1‐undecene odor exposure. (ii) Movement trajectories of N2 worms on E. coli OP50 lawn exposed to 1‐undecene odor for 30 min and 60 min, each color represents the trajectory of single worm. Scale bar = 10 mm.

We examined the chemotaxis of N2 worms to individual volatiles and found that only 1‐undecene was an aversive signal for worms (Fig 2B, Movie EV2A). Using a two‐plate arrangement, to ensure delivery of only volatile form of 1‐undecene, we confirmed that worms show aversion to 1‐undecene odor (Appendix Fig S2A). Most animals deploy various escape strategies upon sensing threat signals, C. elegans exhibits enhanced roaming and omega turns upon encountering aversive signals (Liu et al, 2018). To test whether 1‐undecene is indeed an aversive signal, we examined the locomotion of worms in presence of 1‐undecene. Worms executed omega bend in response to 1‐undecene in 16 out of 17 worms (Fig 2C, Movie EV3) consistent with the response to an aversive signal. Finally, we also examined undA mutant of PA14 (PA14_53120 locus), incapable of producing 1‐undecene. UndA is a non‐heme iron oxidase necessary for the biosynthesis of 1‐undecene in different P. aeruginosa strains (Rui et al, 2014). As expected, transposon insertion mutant of undA did not produce any 1‐undecene in GC‐MS analysis (Appendix Fig S2C). We performed a choice assay between odors of PA14 and undA lawn and found that worms preferred undA over PA14 (Fig 2D). We also examined the transcript levels of undA in 8‐h old and 24‐h old lawn of bacteria and found that undA expression was higher in the older lawn (Appendix Fig S2D). We also observed enhanced roaming by worms in an arena exposed to 1‐undecene odor compared to the arena without the repellent (Fig 2E and Appendix Fig S2B). In all, our analysis of volatiles from the headspace of aversion‐inducing P. aeruginosa lawns revealed that 1‐undecene produced by P. aeruginosa is an aversive signal for the host.

1‐Undecene is sensed by AWB odor sensory neurons of C. elegans

To test whether olfaction in worms is necessary for sensing 1‐undecene, we tested the chemotaxis response of various olfaction‐compromised mutants. To begin with, we examined the chemotaxis of N2 and odr‐3 mutant worms to 1‐undecene. We found that the sensing of 1‐undecene was dependent on ODR‐3 (Fig 3A, Movie EV2B). Further, to identify olfactory neurons necessary for sensing, we used mutants or ablation lines for odor sensory neurons AWA, AWB, and AWC of worms. We first examined lim‐4(ky403) and lim‐4(yz12) worms that lack functional AWB neurons (Sagasti et al, 1999) and found that they showed no response to 1‐undecene. However, AWC‐ablation worms (Beverly et al, 2011) and odr‐7(ky4) worms lacking functional AWA neurons (Sengupta et al, 1994), along with AWA‐ablation worms, had a normal aversion response to 1‐undecene, comparable to N2 worms. ODR‐3 is also expressed in non‐olfactory neurons such as nociceptive neuron ASH which can respond to odors (Yoshida et al, 2012). We also used ablation for ASH neurons to show that this neuron is not involved in response to 1‐undecene. (Fig 3B). These results suggested that the sensing of 1‐undecene is dependent on functional AWB neurons.

Figure 3

1‐Undecene is sensed by AWB odor sensory neurons of C. elegans

Chemotactic dose response of N2 and odr‐3(n2150) worms to 1‐undecene. n ≥ 3 assays. ns (not significant) P > 0.05, ***P ≤ 0.001, ****P ≤ 0.0001 as determined by one‐way ANOVA, followed by Dunnett’s multiple comparison test. Error bars indicate SEM.

Chemotaxis of N2, odr‐3(n2150), odr‐3(n2046), AWC(‐), lim‐4(ky403), lim‐4(yz12), odr‐7(ky4), AWA(−), and ASH(−) worms to 1‐undecene at 1:10 dilution. n ≥ 3 assays. ns (not significant) P > 0.05, ****P ≤ 0.0001 as determined by one‐way ANOVA, followed by Dunnett’s multiple comparison test. Error bars indicate SEM.

Average calcium responses of AWB::GCaMP3 worms exposed to 1‐undecene at indicated concentrations. Each worm was recorded for 180 s. Worms were under stimulus between 11–130 s, window shown in gray, and stimulus withdrawal beyond 130 s. Shaded regions around the curves represent SEM. n = 3. Two dashed box represents 10 s time window after stimulus addition and stimulus removal, respectively.

Average percentage change in ∆F/F for 10 s time window after stimulus addition and withdrawal (dashed boxes in C), each data point represents average data of ≥ 13 worms. **P ≤ 0.01, ****P ≤ 0.0001 as determined by one‐way ANOVA, followed by Dunnett’s multiple comparison test. Error bars indicate SEM.

Representative image of GCaMP3 fluorescence in AWB neuron during 1‐undecene exposure (129 s on x‐axis of Fig 3C) and after 1‐undecene withdrawal (131.2 s on x‐axis of Fig 3C). Scale bar = 10 µm.

Heat map of ratio ΔF/F (change in fluorescence to total fluorescence) in AWB::GCaMP3 neuron upon the addition (10 s) or the removal (130 s) of 1‐undecene. Each row represents an individual worm recorded for 180 s under 1:100 dilution of 1‐undecene, in 11–130 s window.

To further confirm the role of AWB neurons in sensing 1‐undecene, we examined the odor‐evoked calcium response of AWB neurons expressing genetically encoded calcium sensor GCaMP (Chalasani et al, 2007). In AWB neurons, we observed a dose‐dependent, stimulus‐withdrawal response within 1–2 s of removal of stimulus from the nose of the worm (Fig 3C and D), consistent with the requirement of AWB neurons in chemotaxis response (Fig 3B). Withdrawal‐induced calcium response in AWB neurons has been reported for other aversive volatiles (Ha et al, 2010). Stimulus‐withdrawal response was observed in AWB neurons in all the worms examined (Fig 3E and F). We observed none or very small calcium response to 1‐undecene withdrawal in AWA, AWCon, or AWCoff neurons (Appendix Fig S3) consistent with their dispensability in chemotaxis response. These results confirmed that AWB neurons are necessary for sensing and responding to 1‐undecene.

1‐Undecene odor induces pathogen‐specific immune response in C. elegans

To understand whether 1‐undecene serves as a microbe‐associated molecular pattern for C. elegans, we examined the activation of immune response in worms in the presence of only the volatile signals. We reasoned that sensing of P. aeruginosa‐associated molecular patterns would be integrated into mechanisms to activate innate immune responses specific to this pathogen. To test this, we analyzed the expression of P. aeruginosa‐specific immune response genes of irg family which are expressed under the control of ZIP‐2 transcription factor (Estes et al, 2010) or immune effector genes under the control of p38 MAP kinase and other immune response genes (Troemel et al, 2006). Interestingly, we found that worms pre‐exposed to 1‐undecene for just two hours had 2‐ to 30‐fold induction of irg‐1, irg2, and irg‐3 genes compared to the naive worms (Fig 4A). Transcripts for irg‐7, dod‐24, mul‐1, clec‐67, C17H12.8 were not upregulated while F01D5.5 was downregulated in 1‐undecene‐exposed worms (Fig 4A) suggesting that not all but one specific immune response pathway, regulated by ZIP‐2, is engaged upon 1‐undecene sensing. Next, we examined the expression of transcripts in zip‐2(tm4248) mutant and found that upregulation of irg‐1 and irg‐2 induced by 1‐undecene was significantly lower than in N2 worms (Fig 4B). Using an irg‐1p::GFP reporter, we confirmed increased reporter expression in 1‐undecene‐exposed worms compared to naive worms (Fig 4C). Next, we examined transcript levels in odr‐3 and lim‐4 mutants and found that the induction of irg‐1, irg‐2, and irg‐3 transcripts during 1‐undecene odor exposure was dependent on ODR‐3 and it also required functional AWB neurons (Fig 4D and E) confirming the involvement of 1‐undecene‐responsive AWB neurons in mounting the immune response.

Figure 4

1‐Undecene odor induces pathogen‐specific immune response in C. elegans

Real‐time PCR analysis of P. aeruginosa‐specific immune response genes in naive and 1‐undecene odor‐exposed N2 worms. n = 3. *P ≤ 0.05, **P ≤ 0.01 as determined by two‐tailed unpaired t‐test. Error bars indicate SEM. The negative values are arrived at by representing FC value less than 1 as (−1/FC).

Real‐time PCR analysis of irg‐1, irg‐2, and irg‐3 genes in N2 and zip‐2(tm4248) worms exposed to 1‐undecene odor upon respective naive worms. n = 3. ns (not significant) P > 0.05, *P ≤ 0.05, **P ≤ 0.01 as determined by two‐tailed unpaired t‐test. Error bars indicate SEM.

irg‐1p::GFP induction in naive worms and worms exposed to P. aeruginosa (6 h) and 1‐undecene odor (2 h). Scale bar = 500 µm.

Real‐time PCR analysis of irg‐1, irg‐2, and irg‐3 genes in N2, odr‐3(n2150), and odr‐3(n2046) worms exposed to 1‐undecene odor upon respective naive worms. n = 3. *P ≤ 0.05, **P ≤ 0.01 as determined by two‐tailed unpaired t‐test. Error bars indicate SEM.

Real‐time PCR analysis of irg‐1, irg‐2, and irg‐3 genes in N2, lim‐4(ky403), and lim‐4(yz12) worms exposed to 1‐undecene odor upon respective naive worms. n = 3. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001 as determined by two‐tailed unpaired t‐test. Error bars indicate SEM.

Kaplan–Meier survival curve on the full lawn of P. aeruginosa for N2 (naive) worms and N2 worms pre‐exposed to 1‐undecene odor. Survival assay was performed at 20°C.

Time required for 50% of worms to die (TD50) on lawn of P. aeruginosa. Each data point indicates replicates with ~ 100 worms each. n = 3 assays. **P ≤ 0.01 as determined by two‐tailed unpaired t‐test. Error bars indicate SEM.

Real‐time PCR analysis of irg‐1, irg‐2, and irg‐3 genes in N2 worms exposed to odor of undA mutant upon PA14. n = 3. *P ≤ 0.05, **P ≤ 0.01 as determined by two‐tailed unpaired t‐test. Error bars indicate SEM.

odr‐3 and lim‐4 mutants did not regulate the expression of other P. aeruginosa‐specific immune response genes suggesting the specificity of 1‐undecene response towards ZIP‐2‐regulated immune response (Appendix Fig S4A and B). To understand whether odor‐induced upregulation of ZIP‐2 pathway protects worms from infection, we pre‐exposed worms to 1‐undecene followed by chronic infection with P. aeruginosa. As shown in Fig 4F, worms pre‐exposed to 1‐undecene had significantly better survival than naive worms (also see Fig 4G). To further confirm that irg upregulation is induced by 1‐undecene odor emanating from P. aeruginosa lawns, we examined the levels of transcripts in worms exposed to PA14 odor or undA mutant odor. We found that the induction of irg‐1, irg‐2, and irg‐3 by PA14 odor was completely abrogated in worms exposed to undA odor (Fig 4H). Moreover, the survival of worms on undA was significantly lower than on wild‐type P. aeruginosa PA14 (Appendix Fig S4E and F) suggesting that 1‐undecene‐induced immune response contributes to the survival of worms during P. aeruginosa infection. These results suggested that 1‐undecene present in PA14 odor is necessary and sufficient to induce immune effector upregulation.

To confirm that 1‐undecene‐induced response was specific to P. aeruginosa, we also examined the expression of eight immune response genes induced 10‐ to 100‐fold in worms in response to Gram‐positive bacteria, Enterococcus faecalis and Staphylococcus aureus, or pathogenic yeast Cryptococcus neoformans (Dasgupta et al, 2020). We found that 1‐undecene odor exposure showed little or no induction in expression of transcripts of fmo‐2, acs‐2, lipl‐1, lipl‐3, cpr‐4, cpr‐5, asp‐14, and lys‐3 (Appendix Fig S4D). The inability of 1‐undecene to induce fmo‐2 could also be observed using a fmo‐2p::GFP reporter (Appendix Fig S4C). We also analyzed the induction of heat shock response or oxidative stress response using hsp‐16.2p::GFP and gst‐4p::GFP, respectively (Appendix Fig S4G and H) (Link et al, 1999; Link & Johnson, 2002). We examined two additional markers, ATF‐4 and HSP‐4 (Shen et al, 2001; Glover‐Cutter et al, 2013), to test the effect of 1‐undecene exposure on homeostasis and found no appreciable change (Appendix Fig S4I). Thus, we found that 1‐undecene odor exposure did not disrupt cellular homeostasis.

Based on our study, we propose a model for volatile‐based pattern recognition in C. elegans. We show that 1‐undecene, a volatile produced abundantly by P. aeruginosa, is sensed by AWB odor sensory neurons of C. elegans leading to activation of behavioral response and immune response specific to the pathogen leading to enhanced survival of the host (Fig 5).

Figure 5

Model for olfaction driven flight and fight response in C. elegans

A non‐heme oxidase UndA is required for the production of a 11‐carbon olefin, 1‐undecene, a volatile. This volatile induces calcium signaling in AWB odor sensory neurons leading to aversion response (FLIGHT) as well as upregulation of pathogen‐specific immune response genes (FIGHT) and resistance in C. elegans.

Discussion

Our study provides molecular and neuronal bases of the flight and fight responses of C. elegans upon exposure to pathogenic bacterium P. aeruginosa. We show that a specific volatile organic compound, 1‐undecene, produced by the pathogen P. aeruginosa is sensed as an aversive signal by C. elegans. We show that olfaction‐mediated modulation of behavior by 1‐undecene is dependent on neuronal signaling in the AWB odor sensory neurons. Not only does 1‐undecene initiate protective aversion responses in individual worms, but it also induces P. aeruginosa‐specific immune response and enhances survival during infection providing evidence that 1‐undecene is a pathogen‐associated molecular pattern (PAMP) for worms.

To qualify 1‐undecene as a molecular pattern, we first asked whether 1‐undecene is a Pseudomonas‐specific signal? The presence of 1‐undecene in P. aeruginosa headspace has been reported earlier and has been suggested as a Pseudomonas‐specific volatile (Goeminne et al, 2012; Rui et al, 2014). Previous studies have reported that all Pseudomonas species have undA gene (Rui et al, 2014). Moreover, we also could not detect 1‐undecene in E. coli OP50, Salmonella typhimurium, and Enterococcus faecalis headspace or find it in the published literature (Siripatrawan, 2008; Zhu et al, 2010; Worthy et al, 2018). This provided additional evidence that 1‐undecene is a signature for Pseudomonas species and perceived by worms as a molecular pattern. 1‐Undecene produced by Pseudomonas species is also known to affect fungus Phytophthora infestans and plants (Hunziker et al, 2015; Lo Cantore et al, 2015), suggesting that this volatile mediates inter‐kingdom interactions.

The most relevant evidence for 1‐undecene as a pathogen‐associated molecular pattern comes from the fact that exposure of worms to volatile alone induces upregulation of immune response genes specific to P. aeruginosa. Specifically, ZIP‐2 transcription factor‐regulated genes, irg‐1, irg‐2, and irg‐3, are induced in an AWB neuron‐dependent manner. ZIP‐2 transcription factor is required for upregulation of irg‐1 and irg‐2 in the intestine during 1‐undecene odor exposure providing evidence for a brain‐gut axis of immunity. Moreover, 1‐undecene volatile failed to induce each one of the 8 effectors specific to pathogenic Gram‐positive bacteria and yeast we tested. We also did not find evidence for induction of heat shock response or oxidative stress response. All this evidence suggests that 1‐undecene is a P. aeruginosa‐specific PAMP for C. elegans.

Chemosensation‐based perception of pathogens is significant for worms in light of the fact that worms do not have traditional pattern recognition receptors of Toll and Nod family of pattern recognition receptor, except TOL‐1 which has a limited role in C. elegans microbe interactions (Pradel et al, 2007; Tenor & Aballay, 2008; Brandt & Ringstad, 2015). However, worms have a large sensory repertoire of ~ 1,500 putative G protein‐coupled receptors in the genome (Robertson, 1998), which likely serve the nematode well in a complex sensory environment created by a diverse range of microbes present in rotting vegetation, the natural habitat of C. elegans. A number of stimuli sensed by C. elegans (Bargmann et al, 1993) are bacterial secondary metabolites suggesting that many of the present‐day sensory mechanisms may have evolved and been retained in this bacterivore to influence its ability to find food and to prevent infection.

At least two lines of evidence for perception of bacterial secondary metabolites by worms, allowing them to avoid bacteria, exist. Serrawettin W2, a surfactant produced by S. marcescens, is a non‐olfactory cue that contributes to the aversion response of C. elegans to this bacterium (Pradel et al, 2007). Pyochelin and phenazine‐1‐carboxamide produced by P. aeruginosa are examples of non‐volatile, water‐soluble molecules sensed by C. elegans leading to heterotrimeric G‐protein signaling in ASJ chemosensory neurons (Meisel et al, 2014). It is important to ask what is the contribution of olfaction versus taste in the perception of P. aeruginosa by C. elegans? Our study suggests that ODR‐3‐based olfaction contributes to aversion response in a large fraction of population of worms suggesting that 1‐undecene is a physiologically relevant stimulus for worms.

In nature, worms have to move from a piece of rotting fruit or vegetation to another at considerable distance. Under these circumstances, olfaction might allow worms to sense the aversive signals (or attractant) quicker than using the contact‐based perception of taste or texture. This ability can allow worms to make appropriate decisions without incurring contact‐mediated damage due to toxins. Our study shows that a unique volatile organic compound released by pathogen serves as a PAMP. C. elegans can use such cues to engage in flight‐or‐fight responses to increase its chances of survival. Additional studies of microbe–host interactions would reveal the extent of usage of olfaction in the animal kingdom for perception of pathogens.

Materials and Methods

Strains and growth media

All C. elegans strains were maintained as hermaphrodite at 20°C, on nematode growth media (NGM) plates seeded with E. coli OP50, as previously described (Brenner, 1974). The N2 (Bristol) was used as C. elegans wild‐type strain. All strains used in this study are listed in Appendix Table S1. All experiments done with C. elegans were conducted at 25°C unless otherwise mentioned.

Aversion response assay

All the aversion assays were performed on modified NGM agar plates called slow killing (SK) agar plates used for P. aeruginosa infection study. 3 ml of bacterial culture was grown at 37°C for 12 h. The assay plates were spot dried with 50 µl of overnight grown bacterial culture and incubated for 12 h at 37°C. Approximately 40–50 young adult worms were introduced in the center of the bacterial lawn and the plates were incubated at 25°C. The lawn occupancy of the worms was scored at different time intervals (4 h, 8 h, 12 h) on the basis of their position in the lawn, and the data were used to plot the lawn occupancy of worms with time. Each data point in the data set represents one assay plate with 40–50 worms each. The percent of lawn occupancy was calculated by the formula: (Number of worms inside the lawn/Total number of worms).

Chemotaxis assays

All chemotaxis assays were performed in 90‐mm buffered agar plate (1 mM CaCl2, 1 mM MgSO4, 25 mM KPO4 buffer and 2% agar). The chemicals used in this study were diluted with chloroform (solvent control). 2 µl of test or solvent was spotted on opposite sides of the assay plate (Fig S1D). Young adult worms were washed twice with S‐basal buffer and once with assay buffer (1 mM CaCl2, 1 mM MgSO4, 25 mM KPO4 buffer). Approximately 60–80 worms were placed in the center of the assay plate and 2 µl of 1 M sodium azide was spotted close to the test or solvent spot to immobilize worms once they have made their choice. The assay plates were incubated at 25°C, and the position of worms in the test zone, center zone, and control zone (Fig S1D) was noted at 2 h. The chemotaxis index was calculated by the formula: Chemotaxis Index (CI) = (Worms in test zone − Worms in control zone)/Total number of worms. Each data point in all the chemotaxis experiments represents one assay plate with 40–60 worms each.

For volatile‐mediated chemotaxis: In a two‐plate arrangement, 2 µl of test or solvent was spotted on two sides of the lower plate and approximately 40–60 young adult worms were washed and placed in the center of the upper plate and allowed to move to test area or control area on the basis of volatile cues (Fig S2A). Scoring was done after 2 h by counting the number of worms in the test area, and the chemotaxis index was calculated by the formula: Chemotaxis Index (CI) = (Worms in test area − Worms in control area)/Total.

Bacterial odor choice assay

Bacterial odor choice was done in a tripartite plate with SK agar in two sectors. 50 µl of overnight grown bacterial cultures were spotted in one sector at 0 h and in the neighboring sector at 16 h and incubated at 37°C for another 8 h (Fig 1F). Young adult worms were washed twice with S‐basal and once with assay buffer. 60–80 worms were placed in the third sector with buffered agar and incubated at 25°C. Worms attracted by the odor reached the barrier between the sectors and got immobilized there due to the presence of sodium azide. At 2 h, choice index was calculated by counting the number of worms in individual area facing the test side (young lawn) or control side (old lawn), and the choice index was calculated by the formula: Choice Index (CI) = (Worms toward test side − Worms toward control side)/Total.

C. elegans survival assay

The survival assay was modified from a previous study (Singh & Aballay, 2006). 50 µl of overnight grown bacterial culture was spread uniformly on 60 mm SK plates and incubated at 37°C for 12 h. 50 young adult worms were placed in the center of the lawn, incubated at 25°C, and were scored for survival after 24 h for every 4 h. For “half lawn” survival experiments, 50 µl of the bacterial culture was spread in the center (20 mm diameter) on a 60‐mm Petri dish. For odor pre‐exposure, worms were exposed to 1‐undecene (3 µl) for 2 h followed by 2 h of no odor, while naive worms were exposed to no odors for the entire duration of 2 h. Following this, the worms were exposed to full lawn of P. aeruginosa and scored for survival at 20°C. Time to death of 50% of the worms (TD50) was calculated from three or more Kaplan–Meier survival curves. The curves were analyzed using the log‐rank (Mantel–Cox) test. Each data point in all the survival experiment represents one assay with 90–100 worms each.

Quantification of intestinal lumen width

Young adult N2 worms were exposed to E. Coli OP50 lawn and P. aeruginosa lawn in the same bacterial lawn size setup as done in the aversion response assay and incubated at 25°C for 12 h. The worms were then randomly picked from inside or outside of the lawn; washed with M9 buffer containing 50 mM sodium azide, and mounted on 2% agar pads. The worms were visualized using Leica DMi8 inverted fluorescence microscope.

SPME‐GC‐MS/MS analysis of volatiles produced by P. aeruginosa

An overnight culture of P. aeruginosa PA14 or mutant was used to seed with 50 µl spots on a 60‐mm SK plate. The plates were then incubated for 22 h at 37°C. For the collection of volatiles, 2 P. aeruginosa PA14 or mutant spotted plates were sealed together using parafilm. The SPME fiber was inserted in between the two bacterial plates by puncturing a hole in parafilm between the plates with a needle (Fig S1A). A solid‐phase microextraction fiber (SPME; divinylbenzene/carboxen/polydimethyl siloxane, 50/30 µm; Supelco, Sigma‐Aldrich, Cat No. 57328‐U) was used for the collection of volatiles. The fiber was exposed to the bacterial volatile for 1 h at room temperature. Immediately after collection, the SPME fiber was inserted into the GC injection port for desorption of the bacterial volatiles. The analysis was performed using 6890C gas chromatography (Agilent) interfaced with a 7000C mass selective detector (GC‐MS). A capillary column, DB5‐MS capillary column (30 m × 0.25 mm I.D. and film thickness 0.25 m, Agilent, Palo Alto, CA, USA), was used for separation with ultra‐pure helium gas as the carrier gas at a constant flow rate of 3 ml/min. The injector was kept in split mode with a split ratio of 50:1. The column temperature program consisted of injection at 40°C, hold for 1 min, a temperature increase of 5°C/min to 170°C, followed by a temperature increase of 100°C/min to 270°C and hold for 2 min. The temperatures of the injector and MS source were maintained at 225°C and 265°C, respectively. Volatiles were identified by comparing the mass spectra obtained with the mass spectral library of the GCMS data system, NIST 11 (National Institute of Standards and Technology) mass spectral library.

Calcium imaging

Calcium imaging in individual olfactory neurons expressing GCaMP (Appendix Table S1) was performed in a custom‐designed microfluidic device (Chalasani et al, 2007). Transgenic worm expressing GCaMP family of genetically encoded calcium indicator in individual odor sensory neurons was trapped in the device, and their calcium response was assayed under different dilutions of 1‐undecene. The dilutions of the chemicals were made using M9 buffer (5 g NaCl, 3 g KH2PO4, 6 g Na2HPO4,1 ml 1 M MgSO4 per liter of water). GCaMP imaging was performed using a Zeiss inverted microscope using a Photometrics EMCCD camera. The imaging was performed by capturing stacks of TIFF files for 180 s at 10 frames per s using Metamorph software. In 180 s imaging session for each worm, the chemical stimulus was provided in the sequence‐10 s stimulus OFF, 120 s stimulus ON followed by 50 s stimulus OFF state. The images were analyzed using MATLAB scripts to plot the change in fluorescence to the baseline F o values. Bar diagrams were plotted as the average change in 10 s window after stimulus addition (time 11–20 s) and 10 s window after stimulus removal condition (time 131–140 s).

Worm tracking and mean square displacement analysis

We designed a simple imaging setup for recording the motion of worms in the presence or absence of 1‐undecene odor on agar plate. We took 15‐min (12,286 frames) long recording of worms (N = 5) to track their behavior. Using MATLAB Image processing toolbox, we converted recorded frames into set of binary images, and then, using Trackmate, an ImageJ plugin, we tracked their motion. To analyze the obtained tracks, we used ensemble‐averaged mean square displacement (MSD), 〈δr 2(t)〉 (Bewerunge et al, 2016) which helps in characterizing the dynamics of the worms. We then estimated the slope (μ) of double logarithm MSD curve. MSD is defined by〈δr 2(t)〉∝ t, where δr is displacement of worms in time t and the slope of double logarithm MSD curve is defined by the slope μ. For diffusion μ = 1, for sub‐diffusion μ < 1 and for super‐diffusion μ > 1. In the analysis of video obtained from data (Fig 2E), we found that μ for control condition was 0.71 and μ under 1‐undecene exposure was 0.92 (Fig S2B), which shows that the worms under 1‐undecene odor exposure covered more area as compared to control condition.

Gene expression analysis

Around 800–1,000 worms of L4 stage were exposed to the odor of 1‐undecene provided by inverting 60 mm NGM plate with four spots of 3 µl 1‐undecene used on the lid. Blank NGM plates were used as lids for control or naive worms. After 2 h of exposure at 25°C, worms were collected from plates by washing with M9 buffer and frozen in QIAzol lysis reagent at −80°C. Total RNA was extracted from the 1‐undecene‐exposed and naive worms using RNeasy plus universal mini kit (Qiagen, Cat. No. 73404), followed by DNase I (Thermo Scientific Cat. No. EN0525) treatment to remove genomic DNA. cDNA was synthesized using the iScript cDNA synthesis kit (Biorad, Cat. No. 170‐8891) and used in real‐time PCR for gene expression analysis using SYBR Green detection (Biorad Cat. No. 1725124) on quantStudio3 (Applied Biosystems) machine. All the Ct values were normalized to actin‐1. The comparative ∆Ct method was used to determine the fold change of each target gene. Primer sequences are available upon request.

For RNA isolation from the P. aeruginosa PA14, overnight culture was prepared from a fresh streaked plate. 500 µl of each culture was spot dried in SK plate. The plates were incubated at 37°C for 24 h for old lawn and 8 h for young lawn. The lawn from the plate was washed with PBS buffer, and 2 ml of 1 OD600 culture was taken for RNA isolation for each sample. RNA isolation was done by hot acid phenol–chloroform method, modified from (Singh et al, 1995).

Microscopy

Adult irg‐1p::GFP transgenic worms containing myo‐3p::mCherry co‐marker in the pharynx were exposed to 1‐undecene odor or control for 2 h as described in the previous section. For P. aeruginosa exposure, worms were allowed to feed on 12‐h old complete lawn of PA14 at 25°C for 4 h. Adult fmo‐2p::GFP transgenic worms were allowed to feed on 12‐h old complete lawn of E. coli OP50, E. faecalis or P. aeruginosa for 8 h and on E. coli OP50 with 1‐undecene exposure for 2 h. Adult hsp‐16.2p::GFP worms were exposed to lawn of E. coli OP50 (naive), or lawn of E. coli OP50 with 1‐undecene odor exposure for 2 h. For heat shock, worms placed on a lawn of E. coli OP50 were incubated at 37°C for 2 h followed by recovery for 4 h at 20°C. Adult gst‐4p::GFP worms were exposed to lawn of E. coli OP50 (naive), or lawn of E. coli OP50 with 1‐undecene odor exposure from lid for 2 h. For oxidative stress, worms were placed on E. coli OP50 lawn on an NGM plate with 20 mM paraquat and incubated for 5 h at 25°C.

In all conditions, worms were washed off with M9 buffer after treatment followed by suspending them in M9 buffer containing 50 mM sodium azide and mounting on 2% agar pads. The worms were visualized using an Olympus IX71 inverted fluorescence microscope.

Statistical analysis

All statistical analysis was done using GraphPad Prism. Statistical analysis was done either by two‐tailed unpaired t‐test or one‐way ANOVA, followed by Dunnett’s multiple comparison test (mentioned in individual figure legends). The Kaplan–Meier method was used to calculate survival fractions and the log‐rank test to compare survival curves. The significance annotations according to P values—ns (not significant) P > 0.05; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001. All the experiments were repeated at least three times unless otherwise indicated.

Author contributions

RV, SHC, and VS supervised the project. DP and VS designed the study and performed the analysis. DP performed aversion, survival, choice, chemotaxis, gene expression analysis experiments; DP and BR performed GC‐MS/MS experiments for volatile analysis under the supervision of RV; DP performed the GCaMP imaging under the supervision of SHC; DP and AMS performed Exploratory behavior experiment; DP, SHC and VS wrote the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Appendix

Click here for additional data file.

Movie EV1

Click here for additional data file.

Movie EV2

Click here for additional data file.

Movie EV3

Click here for additional data file.

Review Process File

Click here for additional data file.