Pharmacological or genetic targeting of Transient Receptor Potential (TRP) channels can disrupt the planarian escape response

In response to noxious stimuli, planarians cease their typical ciliary gliding and exhibit an oscillatory type of locomotion called scrunching. We have previously characterized the biomechanics of scrunching and shown that it is induced by specific stimuli, such as amputation, noxious heat, and extreme pH. Because these specific inducers are known to activate Transient Receptor Potential (TRP) channels in other systems, we hypothesized that TRP channels control scrunching. We found that chemicals known to activate TRPA1 (allyl isothiocyanate (AITC) and hydrogen peroxide) and TRPV (capsaicin and anandamide) in other systems induce scrunching in the planarian species Dugesia japonica and, except for anandamide, in Schmidtea mediterranea. To confirm that these responses were specific to either TRPA1 or TRPV, respectively, we tried to block scrunching using selective TRPA1 or TRPV antagonists and RNA interference (RNAi) mediated knockdown. Unexpectedly, co-treatment with a mammalian TRPA1 antagonist, HC-030031, enhanced AITC-induced scrunching by decreasing the latency time, suggesting an agonistic relationship in planarians. We further confirmed that TRPA1 in both planarian species is necessary for AITC-induced scrunching using RNAi. Conversely, while co-treatment of a mammalian TRPV antagonist, SB-366791, also enhanced capsaicin-induced reactions in D. japonica, combined knockdown of two previously identified D. japonica TRPV genes (DjTRPVa and DjTRPVb) did not inhibit capsaicin-induced scrunching. RNAi of DjTRPVa/DjTRPVb attenuated scrunching induced by the endocannabinoid and TRPV agonist, anandamide. Overall, our results show that although scrunching induction can involve different initial pathways for sensing stimuli, this behavior’s signature dynamical features are independent of the inducer, implying that scrunching is a stereotypical planarian escape behavior in response to various noxious stimuli that converge on a single downstream pathway. Understanding which aspects of nociception are conserved or not across different organisms can provide insight into the underlying regulatory mechanisms to better understand pain sensation.

Introduction Normal locomotion of freshwater planarians, termed gliding, is achieved through synchronous beating of cilia in a layer of secreted mucus [1][2][3]. Gliding planarians display a smooth motion without major body shape changes, except for turning movements of the anterior end. However, when exposed to certain noxious stimuli (e.g. low pH, high temperature, or amputation), planarians switch to a muscular-based escape gait that is characterized by oscillatory body length changes [4]. We termed this gait scrunching and showed that it has a characteristic set of 4 quantifiable parameters: 1. frequency of body length oscillations, 2. relative speed, 3. maximum amplitude, and 4. asymmetry of body elongation and contraction [4]. Moreover, scrunching is conserved among different planarian species, with each species demonstrating a characteristic frequency and speed. Although scrunching shares similarities with peristalsis, another muscle-based oscillatory gait that occurs when cilia beating is disrupted [2,[5][6][7], scrunching is cilia-independent, can be induced in animals performing peristalsis, and is distinguishable from peristalsis based on the 4 parameters listed above [4], demonstrating that scrunching and peristalsis are distinct gaits. Because scrunching is such a stereotypical response involving many steps of neuronal communication (sensation, processing, neuromuscular communication), scrunching in response to noxious heat has proven to be a useful and sensitive readout of neuronal function in planarian toxicological studies [8,9]. However, which molecular targets and neuronal circuits regulate scrunching remain an open question.
Recently, it has been shown using RNA interference (RNAi) that the Transient Receptor Potential (TRP) channel, TRP ankyrin 1 (TRPA1), is required for avoidance behaviors in Schmidtea mediterranea in response to noxious heat and the pungent ingredient in mustard oil, allyl isothiocyanate (AITC) [10]. The authors also showed that the noxious heat response was mediated by H 2 O 2 and/or reactive oxygen species which directly activate SmTRPA1, causing the planarian to avoid hot regions. Moreover, in response to physical injury (tail snips), Smed-TRPA1-knockdown worms scrunched with a significantly reduced amplitude [10]. Based on these results, the authors hypothesized that SmTRPA1 signaling, induced through H 2 O 2 upregulation at the site of wounding, may regulate amputation-induced-scrunching in S. mediterranea. Whether H 2 O 2 exposure alone induces scrunching or whether SmTRPA1 also plays a role in triggering scrunching in response to other stimuli is still unknown.
TRPA1 is a member of the TRP superfamily, comprised of widely conserved transmembrane, nonselective cation channels [11]. TRP channels mediate responses to almost all classes of external stimuli, including various nociceptive stimuli such as extreme temperatures, ultraviolet light, and specific chemical irritants, and as such mediate the initial steps of pain sensation [11][12][13]. TRPs are classified into sub-families depending on their main mode of activation (mechanical, thermal, chemical. . .), but are often polymodal, integrating different stimuli in the same channel [11,13,14].
In addition to TRPA1, TRPV channels are also good candidates for possibly regulating scrunching. Scrunching is activated by low pH and noxious heat [4], stimuli which are known to activate members of the TRP vanilloid (TRPV) sub-family, named after their sensitivity to vanilloid compounds such as capsaicin, in vertebrates and invertebrates [15][16][17][18][19][20]. TRPV channels are activated by a diverse range of stimuli and exhibit a high level of species-dependent functional differences [20,21]. For example, while human and rat TRPV1 are highly sensitive to capsaicin, rabbit and bird have greatly reduced sensitivities [21][22][23]. Historically it was thought that, like fruit flies and nematodes [24][25][26], most invertebrates were also insensitive to this chili pepper irritant [27]. However, medicinal leech was recently found to contain a capsaicin-sensitive TRPV channel [16]. Interestingly, although no TRPV homologs exist in the parasitic flatworm Schistosoma mansoni, it was shown that TRPA1 in this species mediates the behavioral response to capsaicin [28,29]. Previous analysis of a Dugesia japonica transcriptome has estimated that at least 25 TRP homologs may be present in this planarian species [30]. Thus far, partial sequences for seven TRP genes have been identified and their expression profiles characterized [31]. DjTRPMa channels have been shown to regulate thermotaxis behavior at lower temperatures (0-25˚C) [31], but the function of DjTRPA and DjTRPV channels has not yet been studied in this species.
Thus, based on previous literature in planarians and the known activators of scrunching, we hypothesized that planarian TRPA1 and TRPV channels control scrunching to specific stimuli. To test this hypothesis, we first assayed a variety of chemicals for their ability to induce scrunching in two freshwater planarian species, S. mediterranea and D. japonica. We focused on chemical compounds that have been shown to activate TRPA1 and/or TRPV, such as AITC, an agonist of planarian [10] and mammalian [32,33] TRPA1, and capsaicin, a specific agonist of mammalian TRPV1 [34,35].
We found that scrunching is a specific response to known modulators of mammalian TRPA1 or TRPV channels, including AITC and capsaicin. These findings were substantiated by knocking down either SmTRPA1, DjTRPAa or DjTRPVa/DjTRPVb using RNAi and evaluating the behaviors of these worms when exposed to a subset of the confirmed scrunching inducers. We found that planarian TRPA1, and partially TRPV, modulate scrunching in response to different triggers.
The observation that scrunching is a stereotypical response that is the same for different stimuli and sensing mechanisms suggests the existence of a single convergent pathway that regulates scrunching downstream of planarian TRP activation. TRP channels are involved in various chemical and physical sensing capacities across eukaryotes, from yeast to humans, yet exhibit a high level of diversity, both within the superfamily and across species [11,13]. By understanding how these channels are used in different species, such as planarians, we gain better insight into their regulatory mechanisms, with the potential to reveal elements important to control pain sensation.

Animal care
Asexual freshwater planarians of the species Dugesia japonica and Schmidtea mediterranea were used for all experiments. The animals were fed organic chicken or beef liver 1-2 times per week, cleaned twice per week, and starved for 5-7 days prior to experimentation. Planarians were stored in a temperature-controlled Panasonic incubator in the dark at 20˚C with D. japonica in dilute (0.5 g/L) Instant Ocean (IO) water (Spectrum Brands, Blacksburg, VA, USA) and S. mediterranea in 1X Montjüic Salts (MS) water [36].

Behavioral assays
Pharmacological perturbations. All chemicals used are listed in Table 1. Chemicals were stored according to supplier specifications. All stock solutions were made directly in IO or MS water or in 100% dimethyl sulfoxide (DMSO, Sigma-Aldrich, St. Louis, MO, USA), depending on chemical solubility. For chemicals prepared in DMSO, the final DMSO concentrations were kept �1%, which does not induce scrunching (S1 Fig). Specific conditions for each chemical experiment were determined empirically by qualitative observation of planarian scrunching behavior ( Table 1). The lowest exposure concentration tested which induced the most straight-line scrunches in wildtype worms was used for each experiment. The pH of all exposure solutions (except for hydrogen chloride) was checked and adjusted with NaOH to fall between 6.90-7.10, to ensure the observed scrunching behavior was not due to low pH conditions. Planarians were exposed to the chemicals either in a bathing solution or by pipetting a fixed volume directly onto a worm. Pipetting allowed for the usage of small volumes of locally higher chemical concentrations and was used in cases of poor chemical solubility or when baths failed to produce sufficiently long stretches of straight-line scrunching that could be used for quantitative analysis. Working solutions for chemicals were made fresh immediately prior to starting experiments. Planarians were individually placed into 100 x 15-mm or 60 x 15-mm petri dishes (Celltreat Scientific Products, Shirley, MA, USA) depending on whether experiments were conducted in baths or by pipetting, respectively. The behavior of each planarian was recorded, starting immediately after initial exposure to the chemical, for up to five minutes at 10 frames per second (fps) using a charge-coupled device camera (PointGrey Flea3 1.3MP Mono USB 3.0) with a 16-mm lens (Tamron M118FM16 Megapixel Fixed-focal Industrial Lens) attached to a ring stand.
Amputation experiments. Individual S. mediterranea planarians were placed into 100 x 15-mm petri dishes containing 15 mL of MS water. Using a razor blade, planarians were amputated just above the pharynx. The behavior of each planarian was recorded at 10 fps using the same setup as in the chemical assays. The number of scrunches of the resulting head piece was counted for each amputation with the scrunching sequence beginning after the first immediate contraction.
High temperature experiments. 60 mm x 15-mm petri dishes were filled with 5 mL of either IO or MS water. Individual planarians were placed in the dishes and induced to scrunch by pipetting 100 μL 65˚C IO/MS water as in [4]. Additionally, we tested the effects of a heated water bath using an automated set-up for screening individual planarians in a 48-well plate [8]. To induce scrunching, the plate was placed on a warmed peltier plate (TE Technology Inc., Traverse City, MI), whose temperature was computer controlled to heat the water in the wells. The temperature of the peltier was initially set to 65˚C for the first 30 seconds to quickly heat up the plate from room temperature and then gradually decreased to 43˚C to stabilize the aquatic temperature across the plate at around 32˚C for 4 minutes. The plate was imaged from above and the movies were analyzed using a custom, automated MATLAB (MathWorks, Natick, MA, USA) script to detect instances of scrunching, as previously described [8].
Scrunching quantification. Recordings of planarian behavioral responses to the noxious stimuli were processed using ImageJ (National Institutes of Health, Bethesda, MD, USA). The background-subtracted image sequences were cropped to capture the first set of at least four consecutive straight-line scrunches or oscillations. An ellipse was then fit to the sequence of binary images of the worm to track and quantify the major axis (length of the worm) over time. From these data, the parameters frequency (number of scrunching/oscillation cycles per second), maximum elongation (difference between longest and shortest elongations/contractions as a fraction of the longest), relative speed (product of maximum elongation and frequency), and fraction of time spent elongating were then quantified in MATLAB as in [4]. Unless stated otherwise, all values denote mean ± standard deviation (SD). Statistical significance for each scrunching parameter (or number of scrunches for amputation experiments) was calculated using a student's t-test comparing to either previously published values for amputation for wild-type animals or to the control RNAi population for RNAi experiments. Behavior scoring. Recordings of worms in chemical baths were scored by 2 blind reviewers. For every 15 s interval in the first 90 s of recording, worms were scored as either scrunching, exhibiting a non-scrunching reaction, or not reacting. Worms were scored as scrunching if they scrunched at least once in a given 15 s interval, based on the definitions set in [4]. Worms were scored as exhibiting a non-scrunching behavior if they performed other behaviors, such as head shaking, frequent turning or abnormally long body elongation. Planarians that glided unhindered throughout the 15 s interval were scored to have no reaction. Three experimental replicates were carried out for each condition, with N = 8 S. mediterranea and N = 10 D. japonica used per replicate. The mean of the scored responses from the two reviewers and across the experimental replicates for each 15 s interval are shown in the respective figures. To compare the timing of the initiation of scrunching in AITC or capsaicin alone or with the addition of HC-030031 or SB-366791, respectively, a Fisher's exact test was performed by comparing the number of planarians scrunching or not scrunching (no reaction or a nonscrunching reaction) in two early time periods (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30), and 31-45 s), using the averaged numbers from the two reviewers. A Fisher's exact test was also performed to compare the number of planarians reacting (scrunching or exhibiting a non-scrunching reaction) or not reacting in control versus DjTRPVab or DjTRPAa RNAi populations.

Mucus staining
Staining procedures were performed using fluorescently labeled Vicia villosa (VVA) lectins as described previously [4]. D. japonica planarians were individually placed in wells and induced to scrunch atop a glass coverslip using baths of 33 μM capsaicin or 50 μM AITC. The same staining procedure was followed for S. mediterranea planarians with scrunching being induced by a bath of 33 μM capsaicin or 50 μM AITC. Mucus trails were imaged in 4x under GFP fluorescence using a Nikon Eclipse Ci microscope (Nikon Corporation, Minato, Tokyo, Japan). Images were stitched together using Fiji [44] and the MosaicJ plugin [45].

Cilia imaging
To view cilia beating, D. japonica and S. mediterranea planarians were incubated for five minutes in baths of 100 μM anandamide before mounting between a glass slide and a 22 � 22' coverslip. Imaging procedures were performed as previously described in (4).

Partial cloning of Dugesia japonica TRP genes
Partial mRNA sequences for TRPA1 (DjTRPAa) and TRPV (DjTRPVa and DjTRPVb) homologs in D. japonica have been previously published [31]. Primers were designed using Primer3 [46] from these templates for DjTRPVa and DjTRPVb to generate 213 and 430 bp fragments, respectively ( Table 2). For DjTRPAa, using the published partial sequence as a starting point, we blasted against a D. japonica transcriptome (dd_Djap_v4) on PlanMine [47] to identify the full coding sequence (transcript dd_Djap_v4_9060_1_1). An 895 bp fragment was identified from this transcript and cloned using the primers in Table 2.

RNAi feedings and injections
Expression of SmTRPA1 and DjTRPAa were knocked down separately in their respective species. Expression of DjTRPVa and DjTRPVb were knocked down in combination and referred to as DjTRPVab RNAi. The respective TRP genes of interest were knocked down by injecting S. mediterranea or D. japonica worms on four consecutive days with in vitro transcribed dsRNA to a final concentration of at least 1 μg/μL as in [51]. Negative control populations of both species, denoted as control RNAi, were injected with unc22 dsRNA, a non-homologous C. elegans gene. Approximately 180 nL dsRNA were injected into each worm per day of injection using a standard dissection microscope and Pneumatic PicoPump Model PV 820 (World Precision Instruments, Sarasota, FL, USA). Needles were made by pulling 1-mm/0.58-mm OD/ID Kwik-Fill borosilicate glass capillaries through a two-stage program on a P-1000 micropipette puller (Sutter Instrument Company, Novato, CA, USA). On the fourth day of injection, after the fourth injection had been administered, worms were fed organic chicken liver mixed with at least 1 μg/μL dsRNA. Worms were then starved for six days prior to experiments.

qRT-PCR
RNA was extracted from ten worms for each RNAi population using TRIzol (Invitrogen, Carlsbad, CA, USA) then purified using an RNeasy Mini Kit (QIAGEN, Germantown, MD, USA) including a DNase treatment. cDNA was synthesized from each RNA pool using the SuperScript1 III First-Strand Synthesis System for RT-PCR (Invitrogen, Carlsbad, CA, USA), following the manufacturer's protocol and priming with random hexamers. Primers for qPCR were designed using Primer3 [46] and are listed in Table 2. DjGAPDH and SmedGAPDH were used as housekeeping genes for their respective species. qPCR was performed on an MJ Research PTC-200 thermocycler equipped with a Chromo4 Real-Time PCR Detector (Bio-Rad Laboratories, Hercules, CA, USA), using PerfeCTa1 SYBR1 Green FastMix1 (Quantabio, Beverly, MA, USA). Technical triplicates were run for all reactions within an experiment, and two biological replicates were performed. To analyze primer efficiency, standard curves were obtained using a 1:1:1:1 mix of all cDNA pools for each species, serially diluted. The efficiency for each primer pair was found to be between 87-116%. Analysis of relative expression for the genes targeted by RNAi was performed using the ΔΔCt method, where reported values are the mean of all replicates.  [10], we tested known chemical agonists of planarian and/or other species' TRPA1 and TRPV (Table 1) for their ability to trigger scrunching in D. japonica and S. mediterranea. The two oscillatory planarian gaits, scrunching and peristalsis, can be hard to distinguish qualitatively by eye. Thus, if oscillatory motion was observed, we distinguished peristalsis and scrunching by quantifying the 4 characteristic parameters (frequency, speed, maximum elongation, asymmetry of elongation/ contraction) and comparing with published reference values for these gaits [4]. Notably, asymmetry between contraction and elongation cycles is the most distinct feature of scrunching that is conserved among different planarian species [4]. We previously found that each planarian species exhibits a characteristic scrunching frequency and speed, with D. japonica scrunching at higher speeds and with almost double the frequency of S. mediterranea planarians [4]. Therefore, all comparisons are done with references in the same species.

TRPA1 and TRPV agonists induce scrunching in S. mediterranea and D. japonica
Under normal conditions, planarians glide, maintaining a constant body length over time (Fig 1A and 1B). When exposed to 50 μM of the TRPA1 activator, AITC, planarians scrunched showing oscillations of body length elongation and contraction ( Fig 1C) with quantitative parameters consistent with those previously determined for D. japonica and S. mediterranea using amputation (Table 3).
Previous studies in S. mediterranea demonstrated that TRPA1 is directly activated by H 2 O 2 [10]; therefore, we assayed whether H 2 O 2 exposure could induce scrunching. As expected from the results of AITC exposure, 40 mM H 2 O 2 elicited scrunching in both planarian species (Fig 1D and Table 3). Notably, while S. mediterranea scrunching parameters were statistically insignificant from those induced by amputation, D. japonica scrunched at slightly (but statistically significant) lower speeds in 40 mM H 2 O 2 compared to the reference values for amputation-induced scrunching in this species (Table 3). However, all other scrunching parameters were consistent with the previous reference values, demonstrating the overall characteristics of scrunching. This decrease in speed may be due to negative health effects of the exposure, since we found that D. japonica but not S. mediterranea disintegrated within a day following the 5 min H 2 O 2 exposure unless excessively washed post H 2 O 2 exposure. Even after washing in three separate 50 mL baths of IO water, 1/12 D. japonica planarians died within 24 hours. Additional range-finding tests were unable to determine a concentration of H 2 O 2 that induced scrunching without negative health effects in D. japonica. Lower concentrations (10 and 20 mM) only induced head wiggling within 5 min of exposure. It is currently unclear why D. japonica show such an increased sensitivity to H 2 O 2 . We have previously demonstrated that scrunching is induced by high heat and low pH [4], which are known activators of TRPV1 in other species [15][16][17][18][19][20]. Thus, we tested whether the classical TRPV1 agonist, capsaicin, and the endocannabinoid anandamide, known to directly activate TRPV1 in mammalian systems [35,38,[52][53][54], were able to induce scrunching (Fig 1E  and 1F). Although not all organisms are equally sensitive to these mammalian TRPV1 modulators, some invertebrates such as medicinal leech [16] and parasitic flatworms [28] have been shown to exhibit behavioral effects when exposed to capsaicin, suggesting that planarians may also be sensitive to TRPV1 modulators.
Indeed, both S. mediterranea and D. japonica scrunched with stereotypical parameters when exposed to 165 μM capsaicin ( Fig 1E and Table 3). Additionally, in both species, scrunching in capsaicin was often accompanied by vigorous head shaking and jerking (S1 Movie). In contrast, exposure to 100 μM anandamide elicited scrunching in D. japonica but not in S. mediterranea ( Fig 1F and Table 3). The slightly reduced maximum body elongation in D. japonica is likely because anandamide also induced other behaviors, such as increased head lifting or head wiggling, but all other parameters are consistent with scrunching. S. mediterranea worms displayed oscillatory locomotion (Fig 1F), but a quantitative analysis of the parameters shows that S. mediterranea performed peristalsis (Table 3), which we have previously demonstrated to be a distinct gait from scrunching [4]. Peristalsis is induced when cilia beating is disrupted, whereas scrunching is cilia-independent [2,[4][5][6][7]. Using cilia imaging, we confirmed that cilia beating was disrupted in S. mediterranea but not D. japonica exposed to 100 μM anandamide (S3 Fig). A possible explanation for this finding is that in addition to being a low potency agonist of TRPV1, anandamide also activates cannabinoid receptor 1 (CB-1) [35,38]. Although the cannabinoid receptor(s) have not yet been directly identified in (right) planarians during (B) gliding or (C-F) chemically induced scrunching, when exposed to (C) 50 μM AITC, (D) 40 mM H 2 O 2 , (E) 165 μM capsaicin, or (F) 100 μM anandamide. D. japonica planarians only scrunched temporarily in capsaicin, as shown in the representative plot in (E), where scrunching occurs from 2-9 seconds. Length was normalized by the average gliding length for all plots. (G-H) Example S. mediterranea (i) and D. japonica (ii) mucus trails stained with fluorescein-conjugated VVA lectin (see Methods) for worms exposed to (G) 50  planarians, pharmacological experiments with specific cannabinoid receptor agonists and antagonists in the planarian Dugesia gonocephala suggest the presence of functional cannabinoid receptor homologs in planarians [55,56]. Complicating matters, in other systems, crosstalk with the cannabinoid system can modulate the responsiveness of TRPV1 [38]. Furthermore, the efficiency of anandamide binding to TRPV1 appears to be species-specific [35]. Thus, these factors could interact to produce different manifestations of similar, yet distinct, oscillatory gaits in the two species, resulting in scrunching in D. japonica, but peristalsis in S. mediterranea. This dissimilarity in behavioral phenotypes, together with the sensitivity differences to H 2 O 2 exposure, emphasize that care needs to be taken when attempting to extrapolate findings from pharmacological studies from one planarian species to another. Similar pharmacological differences have also been found in parasitic flatworms as AITC was shown to activate TRPA in S. mansoni, but not in the closely related S. haematobium [28]. Finally, we also visualized the mucus trails of worms of both species exposed to either AITC or capsaicin (Fig 1G and 1H) and saw the characteristic profiles of triangular anchor points that we have previously demonstrated to be associated with scrunching [4].
In summary, we found that archetypal agonists of TRPA1 and TRPV channels induce scrunching in planarians, supporting our hypothesis that induction of this gait is mediated by TRPA1 and TRPV activation.

Increasing concentrations of AITC and the TRPA1 antagonist HC-030031 enhance scrunching
It has been previously shown that increasing the concentration of AITC decreases the latency to initiate the nocifensive response in the medicinal leech [40]. We observed this same trend in both planarian species, as at higher concentrations of AITC more planarians scrunched earlier (Fig 2A), with more pronounced differences seen in S. mediterranea. However, in all concentrations of AITC tested, even when planarians did not scrunch initially, they still reacted to the AITC as evidenced by vigorous head turning (shown as percent worms reacting in Fig 2A), thus demonstrating that the initial sensation of AITC does not appear to be affected by concentration within the range tested here. Interestingly, the scrunching parameters were also dependent on AITC concentration, with increasing concentrations causing significantly increased maximum elongation and speed for both D. japonica and S. mediterranea when compared to 50 μM AITC (S1 Table). Whereas for D. japonica, these values, as well as frequency in 100 μM AITC, were also significantly different from the scrunching parameters in response to amputation, the parameters for S. mediterranea exposed to the different concentrations of AITC were not significantly different from amputation-induced scrunching parameters, indicating these differences were within the range observed in this species (S1 Table).
A striking behavioral difference was observed between S. mediterranea and D. japonica when exposed to AITC. In all tested AITC concentrations, the majority (at least~80% in all tested concentrations) of scrunching D. japonica ceased scrunching by 90 seconds and began gliding, as seen by the decrease in both the percent worms scrunching and reacting in Fig 2A. This apparent desensitization was concentration-dependent; D. japonica planarians at higher AITC concentrations started and ceased scrunching earlier than those at lower AITC concentrations. S. mediterranea did not share this desensitization behavior and showed longer periods where all worms were scrunching (Fig 2A, compare 100 μM AITC between the two species). Consistent with this observed desensitization to prolonged scrunching activation in D. japonica, the continuous application of high concentrations of AITC completely desensitizes currents in the dorsal root ganglion neurons of mice [57]. AITC is thought to activate TRPA1 through covalent modification of conserved cysteine residues [58], causing seemingly irreversible activation on the time scale of the mice electrophysiological experiments (15-60 mins) [57]. Since acute or repeated exposure to lower concentrations of AITC were not shown to induce desensitization of mouse TRPA1 [57], it is possible that differences in sensitivity between the two planarian species, which could be due to differences in chemical uptake and/ or receptor sensitivities, could explain the lack of desensitization observed at the used concentrations in S. mediterranea.
HC-030031 is a specific TRPA1 antagonist that has been shown to block nocifensive responses to AITC in other systems, including rat and the medicinal leech [39,40]. Therefore, we tested whether HC-030031 could block or at least attenuate planarian scrunching. During initial tests with multiple concentrations of HC-030031, we unexpectedly found that 200 μM HC-030031 induced scrunching in 10/10 D. japonica planarians at some point within 2 minutes of exposure, whereas it did not have that effect on S. mediterranea (S4A Fig). At 100 μM HC-030031, neither planarian species scrunched, but D. japonica displayed a mild reaction including vigorous head turning, which was not observed in S. mediterranea. However, since scrunching was absent at this concentration in both species, 100 μM HC-030031 was used for further experiments.
When co-administered with 50, 75, or 100 μM AITC, 100 μM HC-030031 (Fig 2B) decreased the latency to induce scrunching in the majority of planarians at 50 and 75 μM AITC compared to AITC treatment alone (Fig 2A) in both planarian species (S2 and S3 Movies). This decreased latency was confirmed by Fisher's exact tests showing a statistically significant increase in the number of worms scrunching at 15-30 and/or 31-45 seconds in 100 μM HC-030031 compared to AITC alone (S2 Table). These data suggest a cooperative interaction between AITC and HC-030031, which mimicked the trend seen in increasing concentrations of AITC alone (Fig 2A). This effect was not as pronounced at 100 μM, especially in S. mediterranea where no significant differences were found, suggesting that the maximal activity may have already been reached with 100 μM AITC alone.
Together, these findings suggest that increasing concentrations of AITC or cooperative actions of AITC and HC-030031 enhance scrunching, further supporting the idea that TRPA1 is involved in mediating the scrunching response.

Genetic modulation of TRPA1 expression disrupts scrunching in response to AITC, H 2 O 2 , and amputation
Our chemical experiments suggest that TRPA1 activation in S. mediterranea and D. japonica induces scrunching. Sequence analysis of the putative proteins encoded by SmTRPA1 and DjTRPA1 further support that they would be sensitive to AITC as 2 of the 3 cysteines important for covalent modification by AITC [58] are conserved (S5 Fig). To confirm this, we knocked down SmTRPA1 and DjTRPAa using RNAi and evaluated how this affected scrunching in response to TRPA1 modulators. Gene knockdown was confirmed by qRT-PCR showing 73.0% knockdown in SmTRPA1 RNAi populations and 51.4% knockdown in DjTRPAa RNAi populations compared to expression in the species-specific control RNAi populations (S6A and S6B Fig).
When exposed to 100 μM AITC for 90 s, none of the SmTRPA1 RNAi or DjTRPAa RNAi planarians scrunched, while all control RNAi animals of each species scrunched under the same conditions (Fig 3A and 3B, S4 and S5 Movies). Similarly, scrunching in response to 200 μM HC-030031 alone or to 100 μM HC-030031 + 100 μM AITC was completely lost in DjTRPAa RNAi animals (S4B and S4C Fig), demonstrating that planarian TRPA1 is essential for AITC-induced scrunching and that HC-030031 activates TRPA1. Our results suggest a cooperative, rather than antagonistic, action of HC-030031 on planarian TRPA1. Different organisms have been shown to have different sensitivities to the antagonistic effects of HC-030031. For example, divergence of a single amino acid (N855 in human TRPA1) in frog and zebrafish TRPA1 is responsible for their insensitivity to the inhibitor [59]. Although the mechanism of human TRPA1 inhibition could not be resolved structurally [60], it has been suggested that HC-030031 causes a conformational change in human TRPA1 which disrupts ligand binding [59]. Thus, it is possible that in planarians HC-030031 may cause a different conformational change in TRPA1 to instead potentiate AITC activation.
Scrunching was also completely lost in both SmTRPA1 RNAi and DjTRPAa RNAi populations exposed to 40 mM H 2 O 2 for either 270 s or 60 s, respectively, during which times all control RNAi planarians of both species scrunched (Fig 3C and 3D). These data are consistent with previous reports of H 2 O 2 as a direct activator of S. mediterranea TRPA1 [10]. Together, these results show that planarian TRPA1 is essential to induce scrunching with either AITC or H 2 O 2.
One of the most robust but unspecific inducers of scrunching is amputation [4]. Arenas et al. found that when doing tail snips on filter paper, Smed-TRPA1 RNAi animals exhibited a decreased scrunching amplitude compared to control RNAi animals [10]. Because dry environments alone can induce scrunching (S7A Fig and [4]), we did not perform amputation experiments on filter paper, but in an aqueous environment instead. Under these experimental conditions, we found that knockdown of SmTRPA1 caused reduced scrunching compared to control RNAi animals after amputation, evidenced by fewer total scrunches (S6B Fig). Thus, SmTRPA1 appears to partially mediate scrunching in response to amputation. These results are consistent with the work of Arenas et al., who also found attenuated rather than completely abolished scrunching in Smed-TRPA1 RNAi animals after amputation [10].
In S. mediterranea, it has been shown that amputation leads to a burst of H 2 O 2 production at the wound site [61]. Thus, because H 2 O 2 directly activates SmTRPA1, it has been suggested that mechanical injury (such as amputation) indirectly activates SmTRPA1 through H 2 O 2  Table 3 and S1 Table). Plots are representative of the total number of worms tested, as indicated in the respective panels for each condition.
https://doi.org/10.1371/journal.pone.0226104.g003 TRP channels in planarian scrunching production [10]. While our results confirm that H 2 O 2 activation of TRPA1 induces scrunching, H 2 O 2 activation of TRPA1 is likely not the only mechanism mediating amputation-induced scrunching since scrunching is not completely abolished in amputated SmTRPA1 RNAi planarians. We were unable to perform these same experiments with the D. japonica RNAi populations as even in control RNAi animals, amputation only induces few scrunches robustly.
Together, our results confirm that TRPA1 in both S. mediterranea and D. japonica is necessary to induce scrunching in response to AITC and H 2 O 2 and is partially involved in amputation-induced scrunching in S. mediterranea .

TRPV antagonist SB-366791 enhances scrunching
As we did for TRPA1, we similarly dissected the role of TRPV in scrunching. Because anandamide did not elicit scrunching in S. mediterranea and because two D. japonica TRPV genes have previously been characterized [31], we carried out all further experiments in D. japonica only. When comparing the behavioral responses to different concentrations of capsaicin, we found that at all tested concentrations, the planarians initially reacted by vigorously turning and shaking their heads and then transitioned to a scrunching phenotype over time. Because of this transitional nature of the behavior, it was often difficult to confidently distinguish nonscrunching reactions from scrunching by eye alone. Increasing the concentration of capsaicin decreased the latency time to switch to a scrunching reaction, similarly to AITC ( Fig 4A). As with AITC (Fig 2A), after 45 s in 165 μM capsaicin, D. japonica scrunching behavior began to cease (Fig 4A). However, unlike in AITC where D. japonica began gliding again, D. japonica continued to react in capsaicin by maintaining a contracted body length and exhibiting minor oscillations (S6 Movie). Many nociceptors, including TRPA1 and TRPV, have been shown to become desensitized following prolonged activation. This desensitization is why certain TRP agonists, such as capsaicin, have been used therapeutically as analgesics [62]. In rat neuronal cell culture, it was shown that prolonged capsaicin exposure causes rat TRPV1 channels to be removed from the membrane through endocytosis and lysosomal degradation [63]. A similar desensitization to scrunching induction appears to be present in D. japonica, though the underlying mechanism remains to be determined.
We then tested SB-366791, a selective antagonist of human and rat TRPV1 [41,42]. Initial experiments using a range of different concentrations of SB-366791 showed that D. japonica began vigorous head turning at a concentration of 10 μM SB-366791 but did not scrunch. No abnormal behaviors were observed at 1 μM (S7 Movie). Similarly to AITC co-administration with HC-030031, co-administration of capsaicin with 10 μM SB-366791 decreased the scrunching latency compared to exposure to capsaicin alone (compare Fig 4A and 4B for the same capsaicin concentrations). Statistically significant differences were seen in the proportion of worms scrunching at 16-30 or 31-45 seconds in 33 and 82.5 μM capsaicin with or without 10 μM SB-366791 (S3 Table). However, unlike the trends seen with HC-030031, co-administration with SB-366791 decreased the number of worms which stopped scrunching over time, creating a more prolonged scrunching reaction compared to capsaicin alone. Together these results suggest that planarians are sensitive to a known agonist and antagonist of human and rat TRPV1, though the identity of this purported "capsaicin-receptor" remains to be verified. While SB-366791 enhances capsaicin-induced scrunching it does not have the same potentiation effects seen with HC-030031 and AITC.

DjTRPVab modulates scrunching behavior in response to anandamide
To assay their roles in mediating scrunching in response to the TRPV modulators, we knocked down both known DjTRPV genes (DjTRPVa and DjTRPVb) [31] in combination via RNAi (referred to as DjTRPVab RNAi). Gene knockdown was confirmed by qRT-PCR showing a 41.2% decrease in DjTRPVa and 83.3% decrease in DjTRPVb in the DjTRPVab RNAi population compared to expression in control RNAi planarians (S6C and S6D Fig). Because TRPA1 has been found to modulate sensitivity to capsaicin in the parasitic flatworm S. mansoni, which does not have any TRPV homologs [28,29,43], we also evaluated the reactions of DjTRPAa RNAi worms to the mammalian TRPV agonists. Neither DjTRPVab nor DjTRPAa RNAi significantly attenuated scrunching in response to 165 μM capsaicin (Fig 5A and S6  Movie). Although there was a slight decrease in the percentage of worms scrunching in the DjTRPVab and DjTRPAa RNAi populations compared to control RNAi, all DjTRPVab and DjTRPAa RNAi planarians either reacted or scrunched when exposed to 165 μM capsaicin. Sequence comparisons between TRPV1s which are sensitive (human, rat) or insensitive (chicken, rabbit) to capsaicin have revealed 3 residues (Y511, S512, and T550 using human TRPV1 numbering) which are important for capsaicin sensitivity [22,23]. None of these has not yet been identified in this species. Thus, none of these channels are solely responsible for capsaicin-induced scrunching.
Although mammalian TRPV1 was originally identified as the "capsaicin receptor", capsaicin-sensing ability and the responsible receptor varies dramatically across invertebrates. Several invertebrate species, including fruit flies and nematodes, are insensitive to capsaicin [25,26]. In Caenorhabditis elegans, capsaicin potentiates the thermal avoidance response, but this effect is not dependent on OSM-9, the purported C. elegans TRPV1 homolog, suggesting another unknown receptor is involved [25]. A similar situation appears to be present in D. japonica, where scrunching is not dependent on two previously identified TRPV homologs, DjTRPVa and DjTRPVb (Fig 5A). Our results also show that, unlike in S. mansoni, TRPA1 in both planarian species is not responsible for capsaicin sensing, suggesting evolutionary divergence. While our data show that DjTRPVa/b and DjTRPA are not essential for capsaicin sensing in planarians, it is possible that other planarian TRP channels may be involved. Transcriptomic analysis suggests D. japonica may have at least 25 TRPs [30], of which only 7 have been characterized so far [31]. What receptor is responsible for capsaicin sensing, whether it be another unidentified TRPV, a different TRP channel, or some unrelated protein, in freshwater planarians remains to be determined.
Next, we tested whether scrunching induced by anandamide could be affected by knockdown of either DjTRPVab or DjTRPAa. Even in control RNAi planarians, scrunching in response to 125 μM anandamide was complicated as non-scrunching behaviors, such as vigorous head turning or other body shape contortions (Fig 5B, S8 Movie), were also prevalent. Thus, we focused our comparisons on the number of worms reacting by either clearly scrunching or exhibiting these mixed behavioral phenotypes. Most (75%) control RNAi planarians began reacting within 30 seconds, whereas in DjTRPVab RNAi planarians, anandamideinduced scrunching was attenuated, evidenced by an increase in the latency time to induce scrunching ( Fig 5B, S8 Movie). This increased latency was confirmed by a Fisher's exact showing a significant decrease in the number of DjTRPVab RNAi planarians reacting during 16-30s compared to control RNAi planarians at the same time (p-value: 0.02). No significant differences were found in the behaviors of DjTRPAa RNAi planarians. These data suggest that DjTRPVa/b partially mediate anandamide-sensing in D. japonica, though other receptor(s) are likely also involved.
Anandamide and other cannabinoids have complicated relationships with TRP channels. Both endogenous and synthetic cannabinoids act through the canonical cannabinoid receptors, CB-1 and CB-2, but some have been found to activate TRPV and TRPA1 channels as well [54,64]. Additionally, because of the extensive crosstalk between the endocannabinoid system and TRPV1, leading to sensitization of TRPV1 to other endogenous ligands, it has been suggested that even when anandamide treatment mimics the physiological outcomes of TRPV agonists, the effects are not necessarily due to direct activation of TRPV1 [38]. Thus, it is unclear from our RNAi results whether anandamide's role in scrunching is due to direct activation of DjTRPVab or indirectly through its role as an endocannabinoid.
Finally, we assayed the scrunching response of all RNAi populations (SmTRPA1, DjTRPA1, and DjTRPVab) to noxious heat and low pH exposure, which are known to affect TRPV in other species [15,16,[18][19][20]. We observed scrunching in all populations (S7 Fig), indicating that none of these 3 genes are involved in the scrunching response to these stimuli.Strengthening this conclusion is the observation that, in addition to pipetting experiments, when scrunching was induced by heating the aquatic environment, scrunching was observed in all RNAi populations with no statistically significant differences, as determined by a Fisher's exact test (p>0.05). Scrunching was found in 24/34 DjTRPAa and 20/28 DjTRPVab RNAi planarians, similar to control RNAi worms (25/36). Consistent results were also found for S. mediterranea as similar proportions of animals scrunched in the control (9/22) or SmTRPA1 (7/21) RNAi populations. The finding that in this assay S. mediterranea planarians across RNAi populations scrunched much less than D. japonica planarians may be a consequence of the experimental setup being optimized for D. japonica [8]. While previous reports have shown that SmTRPA1 is involved in mediating heat avoidance behaviors via direct activation by H 2 O 2 [10], our data suggest that other channels may be involved in triggering scrunching in response to high temperatures.
Taken together, these data demonstrate that TRPA1 is required for scrunching in response to AITC and H 2 O 2 in both planarian species, whereas DjTRPVab partially regulates anandamide-induced scrunching in D. japonica. It remains to be determined which other receptor(s) are involved in regulating the anandamide response and are responsible for the other scrunching inducers, including capsaicin, low pH and noxious heat. Importantly, because scrunching in response to capsaicin, heat or pH were not affected by knockdown of either DjTRPA or DjTRPVab, these genes are likely not responsible for the general scrunching response but rather mediate sensing of specific stimuli.

Conclusions
Combining the results presented here with our previous studies of scrunching allowed us to partially decipher the molecular mechanisms responsible for sensing noxious stimuli in planarians. In this and our previous work, we found that planarian TRPA1 and TRPV channels, as well as the planarian Big Potassium ion channel SLO-1 which mediates ethanol-induced scrunching [65], are involved in inducing scrunching in response to specific stimuli. Using RNAi, we found that some inducers are specific to one of these pathways, such as AITC and H 2 O 2 to TRPA1, while others, such as anandamide and amputation, rely on potential epistatic or overlapping functions of other unidentified channels. Lastly, the responsible receptors mediating the response to some scrunching inducers, including capsaicin, low pH, and noxious heat remain elusive and it remains to be determined which receptor(s) are responsible for sensing these stimuli. Based on our results, it is likely that multiple receptors could have epistatic or overlapping functions. Further complicating matters, we found species differences with several of the scrunching inducers. For example, while anandamide induced scrunching which was partially mediated by DjTRPVa/b in D. japonica, it induced peristalsis in S. mediterranea. These observations and the differences in sensitivity to H 2 O 2 and HC-030031 between the two species demonstrate that although these two species are closely related, caution must be used when extrapolating the pharmacological effects of one species to another.
It is striking that despite the existence of multiple induction routes, the dynamic features of scrunching are independent of the inducer, and hence of the sensing pathway. This suggests some form of signal integration occurring downstream of these receptors, as illustrated graphically in Fig 6, which represents our current understanding of the molecular mechanisms of scrunching induction.
Signal integration could occur at the neuronal level, raising the question of which parts of the planarian nervous system are involved. Our previous results [4] have shown that tail pieces which lack a brain are still capable of scrunching in response to some stimuli, albeit much more rarely. This would suggest that the ventral nerve cords are sufficient for scrunching execution but that the brain plays an important role in achieving consistent induction.
Our pharmacological studies revealed that several agonist-antagonist pairs (AITC/HC-030031 and capsaicin/SB-366791) either both triggered scrunching and/or were unable to pharmacologically rescue the scrunching phenotype. These results were surprising given that in other systems, including invertebrates such as the medicinal leech and schistosomes, the antagonists work as expected to inhibit the action of the agonists [16,40,43]. In contrast, in the two planarian species studied here, we found that both a TRPA1 agonist and antagonist induced scrunching, with potentiating effects when co-exposed in both planarian species. Similarly, although the mammalian TRPV antagonist SB-366791 did not induce scrunching alone, it also potentiated capsaicin-induced scrunching in D. japonica. Only RNAi against the target genes allowed us to suppress scrunching in response to specific chemical inducers. One possible explanation for these findings is that the planarian sensory system is highly sensitive to any deviation from normal and that scrunching is a default downstream response to system perturbations or stress. However, the observed species differences demonstrate that scrunching is not always triggered, in agreement with our previous findings that ethanol, but not methanol, trigger scrunching [65]. Furthermore, knockdown of a single receptor, such as SLO-1 [65] or TRPA1, abolishes the scrunching response to specific inducers (ethanol and AITC/H 2 O 2 , respectively), without perturbing scrunching in response to other inducers. Together, these observations argue that scrunching is a specific response, whose regulation, despite the progress made in this work, remains poorly understood.
Understanding the molecular mechanisms controlling the scrunching gait, from initiation to execution will require systematic studies of these different aspects using chemical and/or molecular approaches as presented here. The observed complexity and myriad of pathways involved in scrunching initiation reported here may explain why scrunching is a sensitive readout of neurotoxicity [8] and gaining a deeper understanding of its regulation will allow for more mechanistic studies of potential toxicants in the future using the planarian system. Moreover, by understanding the extent that aspects of nociception are conserved (or not) across species will provide better informative context to understand species-specific sensitivity differences and provide insight into the important mechanisms regulating noxious stimuli and pain sensation.
Supporting information S1 Table. Maximum elongation and speed of scrunching are dependent on AITC concentration. Scrunching parameters for D. japonica and S. mediterranea exposed to 50, 75, or 100 μM AITC, denoted as mean ± standard deviation. For each concentration, planarians were observed to scrunch with the parameters listed within the first minute in the bath. � denotes p < 0.05 and �� denotes p < 0.01 significance level compared to 50 μM AITC given by a two-tailed t-test.^denotes p < 0.05 and^^denotes p < 0.01 significance level compared to amputation given by a two-tailed t-test. a Amputation data are previously published values [4], provided for reference. (PDF) S2 Table. P-values comparing scrunching rate in AITC treatment with or without HC-030031. A Fisher's exact test was used to compare the number of worms scrunching vs not scrunching (no reaction or non-scrunching reaction) at each listed time point in different concentrations of AITC alone or co-exposed with 100 μM HC-030031. � denotes p < 0.05 and �� denotes p < 0.01 significance level. (PDF) S3 Table. P-values comparing scrunching rate in capsaicin with or without SB-366791. A Fisher's exact test was used to compare the number of worms scrunching vs not scrunching (no reaction or non-scrunching reaction) at each listed time point in different concentrations of capsaicin alone or co-exposed with 10 μM SB-366791. � denotes p < 0.05 and �� denotes p < 0.01 significance level. (PDF)