C. elegans PEZO-1 is a mechanosensitive ion channel involved in food sensation

Mechanosensitive ion channels regulate several physiological processes including osmotic balance in bacteria (Kung et al., 2010; Martinac et al., 1987); turgor control in plants (Hamilton et al., 2015) and touch (Geffeney and Goodman, 2012; Yan et al., 2013; Ikeda et al., 2014; Maksimovic et al., 2014; Ranade et al., 2014; Woo et al., 2014; Chesler et al., 2016), pain (Murthy et al., 2018; Szczot et al., 2018), proprioception (Woo et al., 2015), hearing (Pan et al., 2018), lineage choice (Pathak et al., 2014), and blood pressure regulation in mammals (Retailleau et al., 2015; Wang et al., 2016; Rode et al., 2017; Zeng et al., 2018). These channels are ubiquitous, as they transduce mechanical stimuli into electrochemical signals in all kingdoms of life (Douguet and Honoré, 2019; Geffeney and Goodman, 2012; Kung et al., 2010). In 2010, the PIEZO1 and PIEZO2 channels were identified as essential components of distinct, mechanically activated cation channels in mammalian cells (Coste et al., 2010). Since then, many physiological roles have been assigned to these two ion channels (Parpaite and Coste, 2017).

Mammalian PIEZO channels have been associated with several hereditary pathophysiologies (Alper, 2017). Piezo1 gain-of-function (GOF) mutations display slow channel inactivation, leading to an increase in cation permeability and subsequent red blood cell dehydration (Albuisson et al., 2013; Bae et al., 2013; Ma et al., 2018; Zarychanski et al., 2012). For instance, the human Piezo1 hereditary mutation R2456H, located in the pore domain, decreases inactivation, while substitution by Lys slows inactivation even further (Bae et al., 2013). Piezo1 global knockouts (KOs) are embryonically lethal in mice (Li et al., 2014; Ranade et al., 2014), and cell-specific KOs result in animals with severe defects (Ma et al., 2018; Wu et al., 2017a). Intriguingly, both Piezo2 KO and GOF mutations are associated with joint contractures, skeletal abnormalities, and alterations in muscle tone (Chesler et al., 2016; Coste et al., 2013; Yamaguchi et al., 2019). GOF and loss-of-function (LOF) mutations are useful genetic tools for determining the contribution of PIEZO channels to mechanosensation in diverse physiological processes and in various animals.

The Caenorhabditis elegans genome encodes an orthologue of the PIEZO channel family, namely pezo-1 (wormbase.org release WS280). Recently, Bai et al. (2020) showed that pezo-1 is expressed in several tissues, including the pharynx. The worm’s pharynx is a pumping organ that rhythmically couples muscle contraction and relaxation in a swallowing motion to pass food down to the animal’s intestine (Keane and Avery, 2003). This swallowing motion stems from a constant low-frequency pumping, maintained by pharyngeal muscles, and bursts of high-frequency pumping from a dedicated pharyngeal nervous system (Avery and Horvitz, 1989; Lee et al., 2017; Raizen et al., 1995; Trojanowski et al., 2016). In mammals, the swallowing reflex is initiated when pressure receptors in the pharynx walls are stimulated by food or liquids, but the identity of the receptors that directly evoke this mechanical response remain to be identified (Tsujimura et al., 2019). Interestingly, the Drosophila melanogaster PIEZO orthologue is a mechanosensitive ion channel (Kim et al., 2012) required for feeding and is also important for avoiding food overconsumption (Min et al., 2021; Wang et al., 2020). To date, whether pezo-1 encodes for a mechanosensitive ion channel or regulates worm pharyngeal activity has yet to be determined.

Here, we found a strong and diverse expression of the pezo-1 gene in pharyngeal tissues by imaging a pezo-1::GFP transgenic reporter strain. By leveraging genetic dissection, electrophysiological measurements, and behavior analyses, we also established that PEZO-1 is required for proper low-frequency electrical activity and pumping behavior. Analyses of pezo-1 KO and GOF mutants demonstrated that decreasing or increasing PEZO-1 function up-regulates pharyngeal parameters. Likewise, mutants display distinct pharyngeal activities triggered by the neurotransmitter serotonin or with various buffer osmolarities. Using elongated bacteria as a food source, we demonstrated that pezo-1 KO decreases pharyngeal pumping frequency, whereas a GOF mutant features increased frequency. Finally, electrophysiological recordings of pezo-1–expressing cells from C. elegans embryo cultures and the Spodoptera frugiperda (Sf9) cell line demonstrate that pezo-1 encodes a mechanosensitive ion channel. Altogether, our results reveal that PEZO-1 is a mechanosensitive ion channel involved in a novel biological function, regulating pharyngeal pumping and food sensation.

Worms were propagated as previously described (Brenner, 1974). N2 (var. Bristol) is referred to as WT throughout the paper. The following strains were used: VVR3 unc119(ed3)III;decEx1(pRedFlpHgr)(C10C5.1[20789]::S0001_pR6K_Amp_2xTY1ce_EGFP_FRT_rpsl_neo_FRT_3xFlag)dFRT::unc-119-Nat, COP1553 (KO, 6,616-bp deletion) pezo-1 (knu508) IV, COP1524 (GOF: R2373K) pezo-1 (knu490) IV, LX960 lin-15B&lin-15A(n765) X; vsIs97 [tph-1p::DsRed2 + lin-15(+)], DA572 eat-4(ad572) III, and DA1051 avr-15(ad1051) V. Transgenic strain VVR3 was obtained by microinjecting a fosmid construct (from the TransgeneOme Project) into a unc-119(ed3) strain (InVivo Biosystems). COP1553 and COP1524 were obtained using the CRISPR-Cas9 method (InVivo Biosystems). Transgenic worm VVR3, expressing GFP under the control of Ppezo-1::GFP, was crossed with pezo-1 mutants COP1553 and COP1524 to obtain VVR69 and VVR70, respectively. LX960 was kindly provided by Dr. Kevin Collins (University of Miami, Miami, FL).

PCR was performed with worm lysates of N2 and COP1553. Phusion High-Fidelity DNA Polymerase (Thermo Fisher Scientific) was used to amplify pezo-1 DNA fragments. Primer sequences are as follows: F1, 5′-GAC​GAC​CAA​CTG​CCT​ACA​T-3′; F2, 5′-TGC​TGT​AAA​TTA​GCA​CTC​GGG​T-3′; and R1, 5′-TGT​ACC​GTA​ATC​CAG​AAA​CGC​A-3′.

Cultured worms were washed with M9 buffer (86 mM NaCl, 42 mM Na₂HPO₄, 22 mM KH₂PO₄, and 1 mM MgSO₄) and collected into 15-ml Falcon tubes. Trizol was added to pelleted worms, and RNA was isolated using the freeze-cracking method as previously described (Van Gilst et al., 2005). Isolated total RNA was purified using the RNAeasy kit (Qiagen). The SuperScript III One-Step RT-PCR system with Platinum Taq DNA Polymerase was used for RT-PCR, following the manufacturer's protocol (Invitrogen). Primers were designed based on the pezo-1 sequences comprising the deletion regions of the knu508 allele of the COP1553 mutant: F3, 5′-GCA​ACG​TCA​CCA​AGA​AGA​GCA​G-3′, and R2, 5′-GCA​TTC​AAT​AGT​CTC​GTT​GCT​G-3′.

Worms were individually selected and dropped into 15 µl of M9 buffer and then paralyzed on a glass slide with 2% agarose pads containing 150 mM 2,3-butanedione monoxime. Bright-field and fluorescence imaging were performed on a Zeiss 710 Confocal microscope using a 20× or 40× objective. Images were processed using Fiji ImageJ (Schindelin et al., 2012) to enhance contrast and convert to an appropriate format.

For all pharyngeal pumping assays, worms were synchronized by placing young adults onto fresh nematode growth media (NGM) plates seeded with OP50 (Escherichia coli strain) and left to lay eggs for 2 h at 20°C. Adults were removed, and the plates were incubated at 20°C for 3 d.

A serotonin aliquot (InVivo Biosystems) was diluted in M9 Buffer before experiments and discarded after 3 h. 42 synchronized worms were picked and transferred to 200 µl of M9 Buffer supplemented with 2, 5, 10, or 20 mM serotonin and incubated at 20°C for 30 min before loading onto a microfluidic chip (SC40, The ScreenChip System; InVivo Biosystems).

OP50 was grown in liquid Luria-Bertani (LB) medium under sterile conditions at 37°C and diluted to OD 1.0. Bacterial cultures were stored at 4°C for up to 1 wk.

The day before the experiment, OP50 colonies were picked from a fresh LB plate and incubated in 2 ml of LB overnight. The following day, 0.5 ml of this culture was used to inoculate 1.5 ml of LB medium and incubated until growth was exponential, which was verified by optical density (OD 0.5). Cephalexin (Alfa Aesar) was added to a final concentration of 60 µg/ml, and the culture was incubated for 2 h. Spaghetti-like OP50 were verified under a microscope and washed three times using 2 ml of M9 buffer, followed by centrifugation at 400 g to gently pellet the elongated bacteria.

Worms were loaded one by one into the microfluidic chip recording channel and left to adjust for 1 min before recording. All recordings were 2 min long. Records were analyzed using NemAnalysis software (InVivo Biosystems) with the brute force algorithm turned off. Parameters were adjusted for each record to include the maximum number of clearly identifiable pharyngeal pumps. Results were exported from the software in sheet form, and parameters were plotted and statistically analyzed using Matlab R2019a (MathWorks). Matlab analysis code is available at: https://github.com/JonathanMillet/Pharyngeal_Pumping_Analysis_Script_JM2021.git.

Young adults were allowed to lay eggs on NGM plates seeded with control or spaghetti-like bacteria for 2 h. Spaghetti-like bacteria were cultured as described above. Animals (10–20 worms) were removed from plates after 2 h, and the number of eggs laid was counted. After 3 d of incubation, animals that reached adulthood were counted in each trial, and results were compared across four trials.

A drop of fresh culture containing control or spaghetti-like bacteria with 2 µM DiI dye (CAS 41085-99-8; Sigma-Aldrich) was placed on an NGM agar plate. Young adults were fed bacteria with DiI for 30 min. Next, worms were transferred onto OP50-seeded NGM without dye for 5 min (Vidal-Gadea et al., 2012). Finally, animals were placed on a thin-layered 2,3-butanedione monoxime–agarose plate for imaging under a Nikon SMZ18 stereomicroscope. Food occupation in the digestive tract was detected by fluorescence.

C. elegans embryonic cells were generated as previously described (Strange et al., 2007). Worms were grown on 10-cm enriched peptone plates with NA22 E. coli. NA22 bacteria grow in very thick layers that provide an abundant food source for large quantities of worms. Synchronized gravid hermaphrodites were bleached to release eggs and washed with sterile egg buffer (118 mM NaCl, 48 mM KCl, 2 mM CaCl₂, 2 mM MgCl₂, and 25 mM HEPES, pH 7.3, 340 mOsm, adjusted with sucrose). The isolated eggs were separated from debris by centrifugation in a 30% sucrose solution. Chitinase (1 U/ml; Sigma-Aldrich) digestion was performed to remove eggshells. The embryo cells were dissociated by pipetting and filtered through a sterile 5-µm Durapore filter (Millipore). The cells were plated on glass coverslips coated with a peanut lectin solution (0.5 mg/ml; Sigma-Aldrich) and cultured in L15 medium (Gibco) supplemented with 50 U/ml penicillin, 50 µg/ml streptomycin, and 10% FBS (Invitrogen) for 72–96 h.

To express pezo-1 in Sf9 cells, we produced recombinant baculoviruses, according to the manufacturer’s instructions (Bac-to-Bac expression system; Invitrogen). To generate this baculovirus, we used a pFastBac construct (Epoch Life Science) containing an 8× histidine–maltose binding protein tag and a synthesized pezo-1 isoform G nucleotide sequence (one of the longest isoforms according to RNA sequencing; wormbase.org release WS280). For expression of pezo-1 R2373K, the construct contained an 8× histidine–maltose binding protein tag and a synthesized pezo-1 isoform G with the R2373K point mutation. We infected Sf9 cells with either pezo-1 baculovirus for 48 h. Infected cells were plated on glass coverslips coated with a peanut lectin solution (1 mg/ml; Sigma-Aldrich) for patch-clamp experiments.

Primary cultured embryo cells labeled with Ppezo-1::GFP from strains VVR3, VVR69, or VVR70 were recorded in the cell-attached configuration of the patch-clamp technique. Control and infected Sf9 insect cells were recorded in the whole-cell or inside-out patch-clamp configurations. For on-cell recordings, the bath solution contained 140 mM KCl, 6 mM NaCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM glucose, and 10 mM HEPES, pH 7.4; 340 mOsm, adjusted with sucrose. The pipette solution contained 140 mM NaCl, 6 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM glucose, and 10 mM HEPES, pH 7.3; 330 mOsm, adjusted with sucrose. Cells were mechanically stimulated with negative pressure applied through the patch pipette using a High-Speed Pressure Clamp (ALA Scientific) controlled with a MultiClamp 700B amplifier through Clampex (Molecular Devices). Cell-attached patches were probed using a square-pulse protocol consisting of −10-mmHg incremental pressure steps, each lasting 1 s, in 10-s intervals. Cells with giga-seals that did not withstand at least six consecutive steps of mechanical stimulation were excluded from analyses. I_steady was defined as the maximal current in the steady state. Deactivation was compared by determining the percentage of I_steady remaining 100 ms after removal of the mechanical stimuli.

For whole-cell recordings, the bath solution contained 140 mM NaCl, 6 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM glucose, and 10 mM HEPES, pH 7.4. The pipette solution contained 140 mM CsCl, 5 mM EGTA, 1 mM CaCl₂, 1 mM MgCl₂, and 10 mM HEPES, pH 7.2. For indentation assays, Sf9 cells were mechanically stimulated with a heat-polished blunt glass pipette (3–4 µm) driven by a piezo servo controller (E625; Physik Instrumente). The blunt pipette was mounted on a micromanipulator at an ∼45° angle and positioned 3–4 µm above the cells without indenting them. Displacement measurements were obtained with a square-pulse protocol consisting of 1-µm incremental indentation steps, each lasting 200 ms, with a 2-ms ramp in 10-s intervals. Recordings with leak currents >200 pA and with access resistance >10 MΩ, as well as cells with giga-seals that did not withstand at least five consecutive steps of mechanical stimulation, were excluded from analyses. For inside-out recordings, symmetrical conditions were established with a solution containing 140 mM CsCl, 5 mM EGTA, 1 mM CaCl₂, 1 mM MgCl₂, and 10 mM HEPES, pH 7.2, and mechanical stimulation was performed identically to on-cell recordings. Pipettes were made from borosilicate glass (Sutter Instruments) and were fire-polished before use until a resistance between 3 and 4 MΩ was reached. Currents were recorded at a constant voltage (−60 mV, unless otherwise noted), sampled at 20 kHz, and low-pass filtered at 2 kHz using a MultiClamp 700B amplifier and Clampex (Molecular Devices). Leak currents before mechanical stimulation were subtracted offline from the current traces.

Data and statistical analyses were performed using DataGraph 4.6.1, Matlab R2019a (MathWorks), and GraphPad Instat 3 software. Statistical methods and sample numbers are detailed in the corresponding figure legends. No technical replicates were included in the analyses.

Fig. S1 shows pezo-1::GFP expression at different developmental stages and in various tissues in C. elegans.Fig. S2 shows the molecular details of the pezo-1 KO strain (COP1553; pezo-1 (knu508) IV). Fig. S3 shows the pharyngeal pumping frequencies of WT, pezo-1 KO, and R2373K challenged with 2 mM serotonin or fed with control E. coli. Fig. S4 shows the differences between control and spaghetti-like E. coli and their effect in WT and pezo-1 mutants’ physiological traits (feeding, development, and pharyngeal pumping frequency). Fig. S5 shows the GFP intensity across the worm strains used to generate embryo cultures for patch-clamp experiments. Fig. S6 shows single-channel trace recordings and current–voltage relationships of WT cells expressing pezo-1::GFP in the cell-attached configuration. Fig. S7 shows representative single-channel trace recordings of pressure-evoked currents from pezo-1::GFP cells expressing pezo-1 WT and R2373K. Fig. S8 shows the current densities evoked with displacement of naive and pezo-1–infected Sf9 cells. Fig. S9 shows representative single-channel trace recordings of pressure-evoked currents from Sf9 cells expressing pezo-1 WT and R2373K.

To determine the expression of pezo-1 in C. elegans, we used a fluorescent translational reporter made by the TransgeneOme Project (Hasse et al., 2016). This fosmid construct contains pezo-1 native cis-regulatory elements, including introns, up to exon 17 and 3′ untranslated region (UTR) sequences linked in-frame to GFP (Fig. 1 A). The position of GFP with respect to the remainder of the gene creates an unnatural truncated version of the PEZO-1 protein. Hence, it likely expresses a nonfunctional protein that excludes 16 exons, which contain most of the pezo-1 sequence (including the pore domain). GFP signals are present in multiple cells at all developmental stages (Fig. S1, A and B). Furthermore, it does not appear to be mosaic, as similar expression patterns were observed in at least three independent transgenic lines. We imaged pezo-1::GFP worms at various focal planes to identify the different cells expressing GFP based on their morphological features (i.e., cell-body position, neurites extension and position along the body, and branching; Fig. 1, B–G). The strongest GFP signals that we identified came from the pharyngeal gland cells (Fig. 1 B, bright and fluorescence fields). These cells are composed of five cell bodies (two ventral g1s, one dorsal g1, and two ventral g2s) located inside the pharynx terminal bulb and three anterior cytoplasmic projections: two short projections that are superposed, ending in the metacorpus, and a long projection reaching the end of the pm3 muscle. These cells are proposed to be involved in digestion (Albertson and Thomson, 1976; Ohmachi et al., 1999), lubrication of the pharynx (Smit et al., 2008), generation and molting of the cuticle (Höflich et al., 2004; Singh and Sulston, 1978), and resistance to pathogenic bacteria (Höflich et al., 2004). Additionally, we visualized pezo-1::GFP in a series of cells surrounding the muscle of the corpus and the isthmus (Fig. 1 C; Albertson and Thomson, 1976). We also recognized the arcade cells as putative pezo-1–expressing cells, according to their morphology and location (Fig. 1, C and E). Arcade cells and the pharyngeal epithelium form the buccal cavity and connect the digestive tract to the outside (Altun and Hall, 2009). We also observed many other anterior cells labeled with GFP; however, we cannot currently confirm if they represent neurons and/or amphids.

By crossing pezo-1::GFP with a tph-1::DsRed marker–carrying strain, we were able to identify pezo-1 expression in the pharyngeal neurosecretory, motor, and sensory (propioceptive/mechanosensory) neurons (NSM_L/R; Fig. 1 G). Importantly, these serotoninergic neurons have been proposed to sense food in the lumen of the pharynx through their proprioceptive-like endings and trigger feeding-related behaviors (i.e., increased pharyngeal pumping, decreased locomotion, and increased egg laying; Albertson and Thomson, 1976; Avery et al., 1993). In addition to the pharyngeal cells, we observed expression of pezo-1 in the ventral nerve cord neurons, according to their morphology and location (Fig. 1 D), striated muscles (Fig. S1 C), coelomocytes (Fig. S1, D and E), spermatheca (Fig. S1, B and E), vulval muscles (Fig. S1 F), and various male neurons, including the ray neurons (Fig. S1 G). Our pezo-1::GFP expression patterns are consistent with those recently described in Hughes et al. (2021) (Preprint). Importantly, the expression pattern reported by our pezo-1 fosmid construct in NSM neurons matches what is reported in the C. elegans Neuronal Gene Expression Map & Network (CeNGEN; Taylor et al., 2021). The strong and varied pezo-1 expression in the pharynx, along with the function of the cells expressing it, led us to investigate the potential contribution of PEZO-1 to pharyngeal function.

To analyze the contribution of pezo-1 to pharyngeal pumping in C. elegans, we used the ScreenChip system (InVivo Biosystems), which can record electropharyngeograms (EPGs; Fig. 2 A; Raizen and Avery, 1994) by loading single, live worms into a microfluidic chip. Fig. 2, A and B, summarizes the pharynx anatomy, electrical properties measured during an EPG, and the neurons involved in pharyngeal function. For instance, the excitation event (E spike) precedes the pharyngeal contraction and is modulated by the MC pacemaker neuron (Fig. 2 B, top), whereas the repolarization event (R spike) leads to pharyngeal relaxation and correlates with the activity of the inhibitory M3 motor neurons (Fig. 2 B, middle). After three to four pumps, there is relaxation of the terminal bulb (isthmus peristalsis), which is modulated by the M4 motor neuron (Fig. 2 B, bottom; Avery and Horvitz, 1989). The main EPG events are regulated by the pharyngeal proprioceptive neuron, NSM. Importantly, the proprioceptive NSM neurons and the I3 interneuron express pezo-1 according to our data (Fig. 1 G) and CeNGEN (Taylor et al., 2021), respectively.

Analysis of the EPG records allows for the determination of various pharyngeal pumping parameters, including frequency, duration, and the time interval that separates two pumping events (hereafter referred to as the interpump interval). We used serotonin to increase pharyngeal activity, since in the absence of food or serotonin, pumping events are infrequent. Serotonin mimics food stimulation by activating the MC_L/R and M3_L/R neurons (Niacaris and Avery, 2003). First, we established a serotonin dose–response profile of the WT (N2) strain pharyngeal pumping parameters (Fig. 2, C–G). Serotonin increases pharyngeal pumping frequency in a dose-dependent manner, with concentrations >5 mM increasing the likelihood of reaching 5 Hz (Fig. 2 C). We averaged the EPG recordings at each serotonin concentration and found a clear difference in pump duration between 0 and 5 mM. Concentrations ≥5 mM evoked similar pump durations (∼100 ms; Fig. 2 D). Interestingly, analysis of the pump duration distribution profile under serotonin stimulation revealed that pharyngeal pump duration fits into two categories: fast (∼80 ms) and slow (100–120 ms; Fig. 2 E, gray rectangles).

We observed that the fast and slow categories displayed an inverse relationship with respect to serotonin concentration (Fig. 2 E, arrows). Unlike pump duration, we observed only a single category for interpump interval, ∼95–120 ms for serotonin concentrations of 5–20 mM (Fig. 2, F and G). Interestingly, we did not observe interpump intervals faster than 90 ms, regardless of the serotonin concentration. The interpump interval results support the idea that there is a minimum refractory period between two pumps. This set of analyses allowed us to establish a suitable model for evaluating the role of pezo-1 function in vivo.

To determine whether pezo-1 has a functional role in pharyngeal pumping, we engineered LOF and GOF mutants. A putative LOF mutant was obtained by deleting 6,616 bp from the pezo-1 locus (hereafter referred to as pezo-1 KO; Fig. 3 A, top; and Fig. S2, A–C). This CRISPR-KO transgenesis deleted 638 amino acid residues from PEZO-1 that, according to the cryo-EM structure of the PIEZO1 mouse orthologue (Ge et al., 2015; Guo and MacKinnon, 2017; Saotome et al., 2018; Zhao et al., 2018), eliminates 12 transmembrane segments, 7 extra- and intracellular loops, and the beam helix that runs parallel to the plasma membrane (Fig. S2 D). Previous works demonstrated that the substitution of R2456H (located at the pore helix) in the human Piezo1 gene orthologue increases cation permeability (GOF) and causes hemolytic anemia (Albuisson et al., 2013; Bae et al., 2013; Zarychanski et al., 2012). Moreover, a conservative substitution of Lys for Arg at position 2456 in the human PIEZO1 channel exhibits a pronounced slowed inactivation when compared with the WT or R2456H channels (Bae et al., 2013). Hence, we engineered a putative GOF mutant strain, obtained by substituting the conserved Arg 2373 with Lys (hereafter referred to as pezo-1 R2373K or GOF; Fig. 3 A, bottom). Parenthetically, the R2373K numbering position is based on isoform G, one of the longest isoforms according to RNA sequencing (wormbase.org release WS280). We also included two mutants known to alter pharyngeal function, eat-4(ad572) and avr-15(ad1051), in our analysis. EAT-4 is a glutamate-sodium symporter involved in postsynaptic glutamate reuptake. eat-4(ad572) affects the neurotransmission efficiency of all glutamatergic pharyngeal neurons (I2_L/R, I5, M3_L/R, M4, MI, and NSM_L/R; Lee et al., 1999). AVR-15 is a glutamate-gated chloride channel expressed in the pm4 and pm5 pharyngeal muscles (both synapsed by M3_L/R) and is involved in relaxation of the pharynx. Its mutant allele ad1051 lengthens pump duration by delaying relaxation of the pharynx in a similar fashion to laser ablation of M3_L/R neurons (Dent et al., 1997). With these strains, we sought to determine if altering PEZO-1 function would affect the worm’s pharyngeal phenotype.

At a 2-mM concentration of exogenous serotonin (to elicit pharyngeal activity), both pezo-1 KO and R2373K mutants displayed higher pumping frequencies than WT, albeit not statistically significantly (WT, 1.92 ± 0.11; KO, 2.21± 0.09; and GOF, 2.29 ± 0.1 Hz; mean ± SEM; Fig. 3 B), similar to avr-15(ad1051) (2.22 ± 0.09 Hz; mean ± SEM; Fig. 3 B). On the other hand, the eat-4(ad572) mutant displayed lower pumping frequency at this serotonin concentration. To further assess the altered pharyngeal function pezo-1 mutants, we analyzed the pump duration distributions from the EPG records. pezo-1 KO distribution is similar to the WT (Fig. 3 C, red versus black), whereas the R2373K mutant profile is reminiscent of avr-15(ad1051), as both mutant strains displayed significantly narrower distributions around 100-ms pump events (Fig. 3, C and D, blue and green versus black), when compared with the WT (significance was determined by a Z test). Moreover, the R2373K mutant lacked fast pump events, 50–80 ms (Fig. 3 C), similar to the WT features observed at high serotonin concentrations (≥5 mM; Fig. 2 E) and the eat-4(ad572) and the avr-15(ad1051) mutants at a 2-mM serotonin concentration (Fig. 3 D). Analysis of the distribution of interpump intervals revealed that pezo-1 KO and R2373K mutants, although different, both spend less time resting between pumps (95–120 ms) than the WT (≈140 ms; Fig. 3 E). This enhancement in function resembles the WT activity measured at 5–20-mM serotonin concentrations (Fig. 2, F and G) and could account for the slight increase in frequency shown in Fig. 3 B. The close resemblance between the pharyngeal pumping parameters of PEZO-1 GOF and the avr-15(ad1051) mutant suggests a potential link between PEZO-1 and pharyngeal relaxation.

Mechanical stimuli come in many forms, including stretching, bending, and osmotic forces (Cox et al., 2019). To further understand the functional role of pezo-1, we evaluated pharyngeal pumping parameters after challenging worm strains with varying osmolarities (in the absence of serotonin or food). The worm’s pharynx draws in liquid and suspended bacteria from the environment and then expels the liquid but traps bacteria (Avery and You, 2012). We adjusted a standard laboratory solution used for worm experiments (M9 buffer, 320 mOsm) to varying osmolarities (150, 260, and 320 mOsm). Parenthetically, M9 buffer elicits acute withdrawal behavior in the absence of food, and another buffer (M13) with lower osmolarity (∼280 mOsm) is commonly used to study molecules that elicit acute avoidance behavior (Hart et al, 2006; Jang and Bargmann, 2013; Caires et al, 2017, 2021; Geron et al, 2018). Because we measured the pharynx function in the absence of food for this set of experiments, we refer to the M9 buffer as a “high-osmolarity solution.” Low-osmolarity solutions would be equivalent to swallowing liquid containing few solutes (150 mOsm), whereas high osmolarities would represent a “gulp” of liquid with a large amount of solutes (320 mOsm). Of note, higher osmolarities were associated with smaller mean pumping frequencies for WT worms (Fig. 4 A). Our results indicate that a larger number of solutes in solution corresponds to an increased retention time in the pharynx before moving to the intestine of the WT worms. Notably, at 260 mOsm, pezo-1 mutants exhibited lower frequency than WT (albeit the KO was not statistically significant), and at 320 mOsm, both pezo-1 KO and GOF mutants displayed a significantly higher pumping frequency than WT worms (Fig. 4 A). In contrast, we did not measure significant differences between WT worms and the pezo-1 mutants at 150 mOsm. Akin to human Piezo2 KO or GOF mutations causing joint contractures (Chesler et al., 2016; Coste et al., 2013; McMillin et al., 2014), we demonstrated that lack of or enhanced PEZO-1 function modulated pharyngeal pumping frequencies similarly (at high osmolarities). Next, we further examined the EPG parameters at high osmolarity (320 mOsm). Analysis of the distribution of pump durations and length of the mean interpump intervals revealed that both pezo-1 mutants had more frequent fast pumps (80–120 ms, Fig. 4 B), and the KO spent less time resting between pumps, compared with the WT (Fig. 4 C). Interestingly, high osmolarity (320 mOsm) revealed resemblances between PEZO-1 GOF and avr-15(ad1051) versus PEZO-1 KO and eat-4(ad572) pharyngeal pumping parameters (frequency and duration, Fig. 4, D and E). Altogether, our results suggest that PEZO-1 is required for fine-tuning pharyngeal function in response to osmolarity changes.

To determine the impact that PEZO-1 function has on food intake, we recorded pharyngeal pumping of WT and pezo-1 strains in response to varying food stimuli. It has been hypothesized that food quality and feeding preferences displayed by worms are linked to bacterial size (Shtonda and Avery, 2006). To this end, we measured worm pharyngeal pumping while feeding them with conventional food used in the laboratory for maintenance (E. coli strain OP50). We found that feeding WT worms E. coli elicited lower pumping frequencies than 2 mM serotonin (Fig. S3; E. coli, 1.36 ± 0.09; and serotonin, 1.92 ± 0.11 Hz; mean ± SEM), whereas pezo-1 mutants displayed similar pumping frequencies with exogenous serotonin or E. coli (Fig. S3). Future experiments are needed to understand why exogenous serotonin stimulation and feeding result in varying pezo-1 influence on pharyngeal function. Additionally, we varied the dimensions of OP50 using cephalexin, an antibiotic that prevents the separation of budding bacteria and generates long spaghetti-like filaments of bacterium, as observed under a microscope and elsewhere (Fig. S4 A; Hou et al., 2020; Martinac et al., 1987). Specifically, cephalexin yields bacteria whose contour length is 5–10 times longer and 1.5 times stiffer (i.e., Young's modulus and the bacterial spring constant) than untreated bacteria, as determined by fluorescence imaging and atomic force microscopy (Hou et al., 2020). Hence, feeding worms with these two physically different types of food could help to elucidate the physiological role of PEZO-1 in detecting food with varying mechanical properties (i.e., small and soft versus large and rigid food). A similar method was previously described using the antibiotic aztreonam and was shown to affect pharyngeal pumping (Ben Arous et al., 2009; Gruninger et al., 2008). WT and pezo-1 mutants are able to ingest spaghetti-like bacteria and reached adulthood in 3 d, similar to worms fed with control bacteria (Fig. S4, B and C). Notably, feeding worms with control or spaghetti-like bacteria revealed distinctive pharyngeal traits between the pezo-1 mutants and the WT worms (Fig. S4 D). When fed with control E. coli, both pezo-1 mutants (KO and GOF) had higher mean frequencies, shorter mean pump durations, narrower pump duration distributions, and faster mean interpump intervals than the WT worms (Fig. 5, A–E). Conversely, feeding worms with spaghetti-like E. coli elicited opposing effects on the pharyngeal pumping parameters of the pezo-1 mutants. For instance, feeding with spaghetti-like E. coli decreased pezo-1 KO mean frequency, while the mean pump duration and distribution remained similar to WT worms (Fig. 5, A–C). Furthermore, this modified diet significantly increased the mean interpump interval of the KO in comparison to the GOF mutant (Fig. 5 D). Unlike the KO and WT strains, the R2373K pezo-1 mutant displayed high-frequency, shorter pumps (mean and distributions; Fig. 5, A–C) and reduced mean interpump interval durations (mean and distributions; Fig. 5, D and E). Altogether, our results indicate that PEZO-1 regulates the pharynx response to the physical parameters of food, such as the length and stiffness of ingested bacteria.

The PEZO-1 protein sequence shares 60–70% similarity with mammalian PIEZO channel orthologues. However, whether PEZO-1 responds to mechanical stimuli has not yet been established. To address this major question, we cultured C. elegans cells from three different strains endogenously expressing pezo-1 WT, KO, or the R2373K GOF mutation in the background of the VVR3 strain that expresses a nonfunctional pezo-1::GFP reporter (Fig. 1 A). pezo-1 WT, KO, and GOF strains expressed similar levels of GFP (Fig. S5). Embryonic pezo-1::GFP cells were patch-clamped using the cell-attached configuration, with application of constant negative pressure (−70 mmHg) at different voltages (Fig. 6, A–C). The normalized steady-state current (I/I_max)-versus-voltage relationship is characterized by a reversal potential of +9.06 mV (Fig. 6 D), indicating that PEZO-1 mediates a slight cation selective conductance, like the mouse and Drosophila orthologues (Coste, 2012). Importantly, most of the pezo-1 WT cells displayed multiple-channel opening events upon mechanical stimulation. Nevertheless, from the few traces that carry pressure-dependent single-channel opening events, we were able to determine that PEZO-1 displays an outward slope conductance of 46.4 pS and an inward slope conductance of 34.8 pS (Figs. 6, E and F; and S6). These current magnitudes are similar to the conductance of human PIEZO1 (37.1 pS) reported by others (Gottlieb et al., 2012).

Mechanical stimulation of embryonic pezo-1::GFP cells expressing WT and GOF PEZO-1, but not the KO, yielded several channel openings upon increasing negative pressure (Fig. 7, A and C). As would be expected with increased activity, it was more difficult to identify single-channel opening events in the traces coming from the GOF, as compared with the WT (Figs. 7 B and S7). Cells expressing WT PEZO-1 displayed mechanodependent currents with a half-pressure activation (P_1/2) corresponding to −59.1 ± 4.3 mmHg (mean ± SEM; Fig. 7, A and D). Alternatively, PEZO-1 R2373K displayed mechanodependent currents with a significantly lower P_1/2 than the WT channel (−39.2 ± 2.2 mmHg; Fig. 7, A and D), indicating that the GOF mutant requires less mechanical stimulation to open. Since we could not reach saturating stimulus, these P_1/2 values might be inaccurate; hence, future experiments are needed to unequivocally determine the differences in sensitivity between the WT and GOF. Notably, the R2373K mutation introduced a latency for activation that was not detected in the WT (Fig. 7 A, blue traces; and Fig. 7 E). The decrease in mechanical threshold, along with the slowed activation, were previously reported for the equivalent human PIEZO1 R2456K mutation in mammalian cell lines (Albuisson et al., 2013; Bae et al., 2013; Romero et al., 2019; Zarychanski et al., 2012). Future experiments are needed to understand the origin of these differences in activation.

Unlike pezo-1 WT, ∼50% of the mechanocurrents elicited from the pezo-1 R2373K–expressing cells remained active after removal of the mechanical stimulus (Fig. 7 A, blue traces; Fig. 7 F; and Fig. S7). This slow deactivation is also reminiscent of the human PIEZO1 R2456K GOF phenotype previously characterized by Bae et al. (2013). Overall, our results suggest that PEZO-1 is a mechanosensitive ion channel and that a conserved mutation in the pore domain elicits activation and deactivation changes similar to those of its human counterpart.

One caveat of our PEZO-1 electrophysiological characterization in C. elegans cultured cells is that we cannot identify (at this point) which type of cells we are measuring from. We are only confident that those cells express pezo-1 since they are labeled with the nonfunctional pezo-1::GFP reporter. These patch-clamp experiments were blind to genotype, yet we consistently found that pezo-1::GFP–labeled cells coming from KO worms had negligible or no currents under the same voltage and pressure regimes used for pezo-1 WT and GOF cell cultures (Fig. 7 C). Hence, to further validate that the pezo-1 gene encodes for a mechanosensitive ion channel, we heterologously expressed one of the longest isoforms of pezo-1 (isoform G; wormbase.org release WS280) in Sf9 cells and measured its function in the whole-cell patch-clamp configuration while stimulating with a piezo-electrically driven glass probe (Fig. 8 A). Similar to mammalian PIEZO channels, PEZO-1 mediates indentation-activated currents (Fig. 8 B). Uninfected Sf9 cells do not display mechanosensitive channel currents (Figs. 8 B and S8). Importantly PEZO-1 displayed the properties described for mammalian PIEZOs in other cell types (Coste et al., 2010; Wu et al., 2017b), including voltage-dependent inactivation (Fig. 8, C–F) and nonselective cation currents, as determined by the reversal potential (−1.15 mV; Fig. 8 G). Our results demonstrate that expressing pezo-1 in a naive system was sufficient to confer mechanosensitivity to Sf9 cells.

We also evaluated the PEZO-1 R2373K mutation in Sf9 cells and found that, like the human equivalent (Bae et al., 2013), the GOF elicits larger mechanically evoked currents that inactivate more slowly than the WT channel (Fig. 8, B–F; and Fig. S8). WT and GOF PEZO-1 channels are characterized by similar reversal potentials and unitary conductance, as determined by whole-cell and inside-out patches, respectively (Fig. 8, G and H; and Fig. S9). As previously reported with human PIEZO1 (Bae et al., 2013), we found that PEZO-1 (WT and GOF) features unique characteristics when activated with two different types of mechanical stimuli (negative pressure or displacement). For instance, with negative pressure, WT and mutant channels displayed steady-state currents, but only the GOF mutant exhibited a pronounced latency for activation and slowed deactivation. With displacement stimulation, it is possible to determine that PEZO-1 GOF inactivates more slowly than the WT. Overall, our results indicate that PIEZO orthologues are functionally conserved.

In 2010, Coste and collaborators reported that the C. elegans genome contained a single Piezo gene, pezo-1 (Coste et al., 2010). However, the functional role of pezo-1 remained elusive, even a decade after its discovery. Here, we showed that PEZO-1 is a mechanosensitive channel with a novel functional role in the worm pharynx by combining fluorescent reporters, genome editing, EPG recordings, behavior, and patch-clamp measurements. We found that pezo-1 is highly expressed in gland cells and the proprioceptive-like NSM neuron, among many other tissues. In addition to its expression, several lines of evidence suggested that PEZO-1 modulated several discrete but reliable features of pharyngeal function. Lack or augmentation of PEZO-1 function decreased interpump intervals when worms were challenged with 2 mM serotonin, while it also increased pumping frequency in high-osmolarity conditions or feeding with control bacteria. In the absence of functional PEZO-1, worms had reduced pharyngeal function (i.e., low frequency and long pump intervals) when fed with spaghetti-like bacteria. Finally, we demonstrated that the pezo-1 gene (WT or GOF) encodes a mechanosensitive ion channel by measuring its native function in C. elegans cells or with heterologous expression in insect cells. Altogether, our results established that PEZO-1 is important for pharyngeal function, regulation, and food sensation.

C. elegans feeding relies on the ability of the pharynx to contract and relax. The pharynx is a tube of electrically coupled muscle cells that pump continuously throughout the worm’s life (Mango, 2007). Several ion channels have been identified to be crucial for the pharyngeal muscle action potential, including acetylcholine receptors, T- and L-type Ca²⁺ channels, glycine receptors, and K⁺ channels (Avery and You, 2012). Although the pharyngeal muscle is capable of pumping (albeit at low frequencies) without nervous system input, higher pumping frequencies are controlled by pharyngeal motor neurons, namely MC_L/R and M3_L/R (Avery and You, 2012). Nevertheless, the role of the nervous system in the control of rhythmic pharyngeal pumping is not completely understood. It is known, however, that the pharynx responds to a variety of neuromodulators (Avery and Horvitz, 1989). We found that pezo-1 is expressed in proprioceptive/mechanosensory NSM_L/R neurons (which are important for the pharyngeal nervous system). Moreover, Taylor et al. (2021) reported expression of pezo-1 in the pharyngeal I3 interneuron. Unlike NSM_L/R and M3_L/R, the function of I3 has not been established (Avery, 1993; Avery and Thomas, 1997). Our results suggest that PEZO-1 is not essential for pharyngeal muscles, but instead fine-tunes the role of the nervous system in controlling pharynx function. This is reminiscent of the novel role of mammalian PIEZO1 and PIEZO2 in mediating neuronal sensing of blood pressure and the baroreceptor reflex (Zeng et al., 2018).

NSM_L/R and M3_L/R (NSM_L/R are pezo-1–expressing neurons), have been postulated to sense bacteria in the pharynx lumen, via their proprioceptive endings, and secrete serotonin in response to this mechanical stimulus (Avery, 1993; Avery and Thomas, 1997). Laser ablation of NSM_L/R in unc-29 mutants leads to subtle changes in pharyngeal pumping rate; however, this was done while simultaneously ablating other pharyngeal motor neurons (M1, M2_L/R, M3_L/R, M5, and MI; Avery, 1993). This approach could exert antagonistic effects on pumping rate, yielding a steady pharyngeal activity. Using the ScreenChip system allowed us to reveal the potential roles of extrapharyngeal neurons expressing pezo-1 (NSM_L/R). Our results determined that proper function of PEZO-1 lengthened interpump intervals with 2 mM serotonin, in the absence of food. They further demonstrated that PEZO-1 modulated the feeding behavior of worms confronted with food of various mechanical properties (control and spaghetti-like bacteria). This led us to hypothesize that PEZO-1 is involved in food sensation and modulates pharyngeal pumping rate. Hence, like the mammalian orthologue PIEZO2, PEZO-1 is expressed in proprioceptive endings and is involved in stretch reflexes (Chesler et al., 2016; Woo et al., 2015). Nevertheless, it remains to be determined if mammalian PIEZO channels play a role in food sensation and/or the swallowing reflex.

Humans sense various organoleptic food qualities, such as visual aspects (color and shape), odorants through smell, and texture and flavor through tasting. In nematodes, there is a lack of understanding of what is sensed as food. Worms are able to filter particles from fluid in a size-dependent manner (Fang-Yen et al., 2009; Kiyama et al., 2012), and feeding is facilitated by attractive odors or suppressed by repellents (e.g., diacetyl, isoamyl alcohol, quinine; Gruninger et al., 2008; Li et al., 2012). Others have demonstrated that worms prefer to feed from active (i.e., bacteria reproducing rapidly and emitting high levels of CO₂) rather than inactive bacteria (Yu et al., 2015). We determined that pezo-1 KO worms “choke” when presented with long and stiff spaghetti-like bacteria, whereas WT and GOF strains increase pharyngeal pumping when ingesting this elongated and rigid food. Therefore, we propose that the pharynx itself might be a sensory organ, as worms modify their pumping parameters when they sense solutions of different osmolarities or food with different textures and/or consistencies. We further hypothesize that worms can perceive changes in texture and adjust their pumping frequency through a mechanism requiring PEZO-1. Since pezo-1 is not essential for C. elegans when cultured in standard laboratory conditions (e.g., monoaxenically on E. coli OP50), we wonder if in its natural biotic environment this mechanosensitive ion channel plays a crucial role, as it does in humans and Drosophila. Given that worms grow in microbe-rich and heterogeneous environments (feeding from prokaryotes of the genera Acetobacter, Gluconobacter, and Enterobacter, for example; Schulenburg and Félix, 2017), they might encounter bacteria of different dimensions and stiffness that would make pezo-1 function more relevant to the worm’s ability to discriminate the food on which it grows best.

Why do pezo-1 LOF and GOF mutations cause similar behavioral phenotypes? Our data show that both pezo-1 mutants (KO and GOF) increase the pumping frequency of the pharynx in different settings: serotonin exposure (albeit not statistically significantly), high osmolarity, and ingestion of control bacteria. While it may seem counterintuitive at first, there are some scenarios in which too little or too much mechanosensation can be detrimental for animal behavior. In humans, PIEZO2 LOF (premature stop codon) and GOF (missense mutation I802F) alleles cause joint contractures, skeletal abnormalities, and alterations in muscle tone (Chesler et al., 2016; Coste et al., 2013; Yamaguchi et al., 2019). Only when feeding worms with spaghetti-like bacteria were we able to uncover differences in the pharyngeal parameters between the LOF and GOF mutants. Hence, we hypothesize that lacking the function of PEZO-1 significantly slows pharyngeal function when passing the lengthy and rigid bacteria from the pharynx to the gut.

Several requirements must be met for a channel to be considered mechanically gated (Arnadóttir and Chalfie, 2010). Accordingly, we found that pezo-1 is expressed in the proprioceptive NSM neuron; knocking out pezo-1 inhibits worm pharyngeal function when fed with elongated and stiff bacteria; engineering a single point mutation in the putative pore domain (R2373K) elicited similar inactivation, activation, and deactivation delays that are reminiscent of the gating behavior reported for the human PIEZO1 R2456K (Bae et al., 2013); and expression of pezo-1 (WT and GOF) confers mechanosensitivity to, otherwise naive, Sf9 cells. We propose that PEZO-1 is a mechanosensitive ion channel given that the time it takes to reach half of the steady-state currents ranges between 3.5 and 15 ms upon application of negative pressure. These are faster than activation times reported for the Drosophila phototransduction cascade, one of the most rapid second messenger cascades (Hardie, 2001). These combined efforts highlight the versatile functions of the PIEZO mechanosensitive channel family, as well as the strength of the C. elegans model organism to reveal physiological functions.

Our findings revealing PEZO-1 as a mechanosensitive ion channel that modulates pharyngeal function raise several important questions. How does pezo-1 modulate pumping behavior electrical activity? Does pezo-1 equally enhance or inhibit the function of the pharyngeal hypodermal, gland, and muscle cells, as well as neurons, expressing this channel? Could pezo-1 phenotypes be exacerbated if the gene function is nulled in a cell-specific manner? Does the slow deactivation and/or inactivation of the GOF mutant, determined at the patch-clamp level, account for the enhancement in pharyngeal function when worms are fed with bacteria? Does PEZO-1 require auxiliary subunits and/or the cytoskeleton for gating? Regardless of the answers, the plethora of physiologic roles that this eukaryotic family of mechanosensitive ion channels play is outstanding. More experimental insight will be needed to fully grasp the full implications of pezo-1 in the physiology of C. elegans.

Data supporting the findings of this paper are available from the corresponding author upon reasonable request. The source data underlying figures and supplementary figures are provided as a Source Data file, doi: https://doi.org/10.6084/m9.figshare.16992058.v3.

Jeanne M. Nerbonne served as editor.

We thank Dr. Julio F. Cordero-Morales, Dr. Andrés G. Vidal-Gadea, and Dr. Christopher E. Hopkins for critically reading the manuscript, as well as Dr. Rebeca Caires and graduate students Olufunke Falayi and Soumi Mazumdar for technical assistance. LX960 was provided by Dr. Kevin Collins (University of Miami).

C. elegans (N2, DA572, and DA1051) and E. coli strains (OP50 and NA22) were obtained from the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health Office of Research Infrastructure Programs (P40 OD010440). This work was supported by the American Heart Association (16SDG26700010 to V. Vásquez) and the National Institutes of Health (R01GM133845 to V. Vásquez), as well as the Neuroscience Institute, University of Tennessee Health Science Center (Research Associate Matching Salary Support to J. Lee).

The authors declare no competing financial interests.

Author contributions: Conceptualization, V. Vásquez; Formal Analysis, V. Vásquez, J.R.M. Millet, and L.O. Romero; Funding Acquisition, V. Vásquez; Investigation, J.R.M. Millet, L.O. Romero, J. Lee, and B. Bell; Methodology, V. Vásquez and J.R.M. Millet; Project Administration, V. Vásquez; Supervision, V. Vásquez; Writing-original draft, V. Vásquez and J.R.M. Millet; Writing-review and editing, V. Vásquez.