Deciphering mechanically activated ion channels at the single-channel level in dorsal root ganglion neurons

Somatosensory neurons convey mechanosensory stimuli such as touch, proprioception, deep pressure, and acute pain by the activation of mechanically activated (MA) ion channels that transduce force into biological signals (Chalfie, 2009; Delmas and Coste, 2013; Hao et al., 2015). The first evidence for MA currents in somatosensory neurons was described in rat dorsal root ganglion (DRG) neurons (McCarter et al., 1999), where mechanical indentation of the DRG soma with a blunt glass probe induced MA macroscopic currents in the whole-cell patch clamp mode. Subsequent studies have demonstrated that these MA currents have a latency within a few milliseconds, suggesting that it is indeed mediated by an ion channel, are non-selective cationic, and are blocked by generic MA channel inhibitors like gadolinium and amiloride (Coste et al., 2007; Drew et al., 2002; Hu and Lewin, 2006; McCarter and Levine, 2006).

Since the first description of the DRG MA current, the identity of the ion channel(s) that induce these currents had remained elusive until the discovery of the MA ion channel family PIEZOs (Coste et al., 2010). In DRG neurons, PIEZO2 accounts for 40% of the in vitro MA currents and mediates the firing of rapidly and slowly adapting Aβ mechanoreceptors and affects the threshold of Aδ fibers in ex vivo skin nerve preparations (Coste et al., 2012; Maksimovic et al., 2014; Murthy et al., 2018b; Woo et al., 2014). Additionally, PIEZO2 mediates in vivo calcium signals in trigeminal ganglion neurons and in vivo electrophysiological signals from spinal cord dorsal horn neurons in response to innocuous stimuli such as brush stroke and air puff (von Buchholtz et al, 2021; Chirila et al, 2022). Furthermore, through mechanosensory assays performed in Piezo2 knockout (KO) mice and patients with PIEZO2 deficiency disorder, we now know that PIEZO2 is required for touch sensation, proprioception, sensing bladder pressure, and mechanical allodynia, but does not affect deep pressure sensation and acute mechanical pain (Case et al., 2021; Chesler et al., 2016; Marshall et al., 2020; Ranade et al., 2014; Szczot et al., 2018; Woo et al., 2015). PIEZO1, on the other hand, is expressed in itch-specific DRG neurons and is required for the sensation of mechanical itch in mice (Hill et al., 2022). Beyond PIEZOs, other candidates have been proposed as putative noxious mechanosensors in DRG neurons. Whereas the role of some candidates, like TMEM63s, in mechanosensation remains to be determined, other candidates like Tentonin3 and TACAN have fallen short of conclusive evidence (Hong et al, 2016; Anderson et al, 2018; Murthy et al, 2018a; Beaulieu-Laroche et al, 2020; Niu et al., 2021; Parpaite et al., 2021; Rong et al., 2021). Therefore, there is a need to identify the MA ion channel that mediates non-PIEZO2-dependent physiology in somatosensory neurons.

Characterization of MA currents has been focused on membrane indentation-induced macroscopic current properties like inactivation kinetics (Coste et al., 2007; Drew et al., 2002; Hao and Delmas, 2010). Based on the decay constant determined by the mono-exponential fit of MA current inactivation kinetics (τ), DRG macroscopic currents are grouped into rapidly adapting (RA, τ < 10 ms), intermediately adapting (IA, τ > 10 < 30 ms), and slowly/ultraslowly adapting (SA, τ > 30 ms) (Hao and Delmas, 2010). Although this distribution has largely guided studies in determining the nature of the MA channel that induces these currents, it is not clear whether each group is mediated by a single MA channel or whether it is the combined activity of two different MA channels within the same group. Accordingly, a substantial percentage of DRG macroscopic currents have mixed decay properties and are best fit with two exponential fits (Parpaite et al., 2021). The majority of RA currents and a proportion of IA currents are PIEZO2 dependent (Murthy et al, 2018b; Ranade et al., 2014); however, the identity of the channel(s) that mediate the remaining MA currents is a mystery. Notably, fewer attempts have been made to characterize stretch-activated currents and single-channel properties of MA ion channels in DRG neurons. One study identified two distinct classes of MA ion channels, low and high threshold, with single-channel conductance of 58.3 pS and 15.0 pS, and half-maximal pressure values of 60.6 ± 1.2 and 83.1 ± 2.4 mmHg, respectively (Cho et al., 2002). How this data corresponds to the macroscopic current properties is poorly understood.

To gain further insight into the properties of the ion channel that mediate the macroscopic MA currents in DRG neurons, we sought to correlate macroscopic MA currents to single-channel data. By combining whole-cell and excised outside-out patch clamp recordings, in cultured DRG neurons from wild type (WT), Piezo2 expressing, and Piezo2 KO mice, we isolate distinct stretch-activated single-channel properties underlying the indentation-induced macroscopic responses. We identify PIEZO2-dependent stretch-activated currents in DRG neurons. Finally, our analysis predicts that in addition to PIEZO2, DRG neurons likely express three different MA channels.

All animal procedures were approved by the Institutional Animal Care and Use Committees of Scripps Research (California) and Oregon Health and Science University. HoxB8-Cre Piezo2 knockout mice have been previously described (Woo et al., 2015). HoxB8-Cre mice were crossed to Piezo2^fl/− mice containing one floxed allele and one null allele and were used as Piezo2 knockouts. C57BL/6 and Npy2r-GFP were used as is.

Mouse DRG neurons were isolated and cultured with the procedure as described (Coste et al., 2010). Briefly, once DRGs were extracted from the animal, they were incubated for 1 h at 37°C in a culture medium (Ham’s F12/DMEM with 1% penicillin-streptomycin) containing 1.25% collagenase (Invitrogen). This was followed by a half-hour incubation at 37°C in 4.5 ml culture media containing 500 ml of 1.25 U/ml papain. Cells were gently triturated using fire-polished Pasteur pipettes and centrifuged at 80 g for 10 min in a culture medium containing 15% BSA to separate neurons from debris. Neurons that were pelleted were plated on laminin (2 μg ml⁻¹)–coated poly-D-lysine coverslips. Growth medium supplemented with growth factors, 100 ng/ml nerve growth factor (NGF), 50 ng/ml GDNF, 50 ng/ml BDNF, 50 ng/ml NT-3, and 50 ng/ml NT-4 were added an hour after plating. Mechanically activated currents were recorded on days 1 to 3 after plating.

Wild type and PIEZO1-knockout human embryonic kidney 293T cells (HEK-P1KO, original HEK293T cell RRID: CVCL_0063) were used for all heterologous expression experiments and were negative for mycoplasma contamination. HEK-P1KO cells were generated in-house using CRISPR-Cas9 nuclease genome editing technique as described previously (Dubin et al., 2017; Lukacs et al., 2015). Cells were grown in Dulbecco’s modified Eagle medium (DMEM) containing 4.5 mg ml⁻¹ glucose, 10% fetal bovine serum, 50 U ml⁻¹ penicillin, and 50 µg ml⁻¹ streptomycin. Cells were plated onto 12-mm round glass poly-D-lysine coated coverslips placed in 24-well plates and transfected using lipofectamine 2000 (Invitrogen) according to the manufacturer’s instruction. All plasmids were transfected at a concentration of 600 ng ml⁻¹. Cells were recorded from 24 to 48 h after transfection.

Mechanically activated currents from DRG neurons or HEK cells were recorded in whole-cell patch clamp mode, cell-attached patch clamp mode, or excised outside-out patch clamp mode using an Axopatch 200B amplifier. Currents were sampled at 20 kHz and filtered at 2 kHz. For whole-cell and outside-out recording, electrodes had a resistance of 4 to 7 MΩ when filled with gluconate-based low-chloride intracellular solution: 125 mM K-gluconate, 7 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, 10 mM Hepes (pH with KOH), 1 mM tetra-K BAPTA (1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid), 4 mM Mg-ATP, and 0.5 Na-GTP. The extracellular bath solution was composed of 133 mM NaCl, 3 mM KCl, 2.5 mM CaCl₂, 1 mM MgCl₂, 10 mM Hepes (pH 7.3 with NaOH), and 10 mM glucose. For cell-attached experiments, pipettes were filled with a solution consisting of (in mM) 130 NaCl, 5 KCl, 10 Hepes, 1 CaCl₂, 1 MgCl₂, 10 TEA-Cl (pH 7.3 with NaOH), and the external solution used to zero the membrane potential consisted of (in mM) 140 KCl, 10 Hepes, 1 MgCl₂, 10 glucose (pH 7.3 with KOH). All experiments were done at room temperature.

For whole-cell MA currents, mechanical stimulation was achieved using a fire-polished glass pipette (tip diameter, 3–4 μm) positioned at an angle of 80° relative to the cell being recorded. Displacement of the probe toward the cell was driven by a Clampex-controlled piezoelectric crystal microstage (E625 LVPZT controller/amplifier; Physik Instrumente). The probe had a velocity of 1 μm.ms⁻¹ during the ramp phase of the command for forward movement and the stimulus was applied for a duration of 150 ms. For each cell, a series of mechanical steps in 1-μm increments were applied every 30 s.

For cell-attached recordings, membrane patches were stimulated with brief negative pressure pulses through the recording electrode using a Clampex controlled pressure clamp HSPC-1 device (ALA Scientific). Otherwise stated, stretch-activated channels were recorded at a holding potential of −80 mV with 500 ms or 1 s pressure steps from 0 to −100 mmHg (−10 mmHg increments). For excised patch recordings, after recording whole-cell currents, the recording electrode was slowly retraced to form an outside-out patch. The patch was stretched using the HSPC-1 device, but positive pressure 1-s steps were applied from 0 to +100 mmHg (+10 mmHg increments).

Current–pressure relationships were fitted with a Boltzmann equation of the form: $I(P)={1+exp[−(P–P50)/s)]}−1,$ where I is the peak of stretch-activated current at a given pressure, P is the applied patch pressure (in mmHg), P50 is the pressure value that evoked a current value which is 50% of Imax, and s reflects the current sensitivity to pressure.

Stretch-activated single-channel currents were recorded in the outside-out patch clamp configuration. Because single-channel amplitude is independent of the pressure intensity, after running a series of Δ+10 mmHg episodic pulses, the most optimal pressure stimulation was used to elicit responses that allowed single-channel amplitude measurements. Single-channel amplitude at −80 mV potential was measured from trace histograms of 2 to 29 repeated recordings. Histograms were fitted with Gaussian equations using Clampfit 11 software to measure the open state amplitude, which was then divided by membrane potential to determine chord conductance.

The experimenter was not blinded to any condition. Experimental data was verified by performing at least two independent experiments, from two or more animals or two or more transfection sets across different days, for any given condition. All data were plotted and statistical analysis was performed in GraphPad Prism 9.5.1 software, and P values < 0.05 were considered statistically significant. For single-channel amplitude graphs, individual measurements from each cell are plotted as scatter and mean as bar ± SEM. For single-channel conductance vs. inactivation τ graphs, the mean is plotted as a dot plot or bars ± SD. Mean single-channel amplitude from different cells was compared using Kruskal–Wallis ANOVA. When two clusters of single-channel amplitude from the same cell were compared, Mann–Whitney test was used. Biophysical properties of PIEZO1 and PIEZO2 were compared using Mann–Whitney test.

To unbiasedly classify the different single-channel conductance into distinct groups, based on mean conductance and standard deviation for each cell measurement, frequency distribution was plotted at 1, 2, 3, 4, or 5 pS bin centers (Fig. S1). Working our way from the largest to smallest bin center by using ordinary one-way ANOVA, we statistically determined the number of conductance groups for each condition. Distributing WT cell conductances into 5 pS bins resulted in six groups: 5, 10, 15, 20, 25, and 30. However, 10 pS vs. 15 pS and 25 pS vs. 30 pS were not statistically different (Fig. 1 F). Therefore, the non-significant groups were combined, which resulted in four different conductances that all cells could be grouped into: 5 pS (3.9 pS, n = 1), 15 pS (15.9 ± 0.75 pS, n = 6), 20 pS (20.0 ± 0.60 pS, n = 7), and 25 pS (26.7 ± 0.96 pS, n = 5; Fig. 1 F). A similar approach was taken for the Npy2r and HoxB8-Cre;Piezo2^f/− population. For the Npy2r population, although the two predominant conductances were at 16.27 ± 0.3 pS, n = 3 and 19.49 ± 0.3 pS, n = 4, the groups are not statistically significant (Fig. 3 F). For HoxB8-Cre;Piezo2^f/−, distributing the data into four groups, 15, 20, 25, and 30 pS, resulted in no significance between 15 pS vs. 20 pS, which led us to combine the two groups, revealing three significantly distinct conductance groups; 20 (19.8 ± 0.40 pS, n = 9), 25 (25.2 ± 0.6 pS, n = 3), and 30 (29.72 ± 1.2 pS, n = 2; Fig. 4 E).

Fig. S1 shows mean single-channel conductance of different cells from WT, Npy2r, and HoxB8-Cre;Piezo2^f/− mice, plotted based on frequency distribution at different bin centers (1–5 pS). This data was used to unbiasedly distribute the different conductance levels into statistically distinct groups. Data S1 provides parameter data for the main figures.

Whole-cell macroscopic currents are an ensemble of single-channel activity. Therefore, extracting membrane indentation-induced whole-cell macroscopic current properties, such as inactivation time kinetics and stretch-activated single-channel data from the same cell would inform us about the properties of the MA ion channels expressed in DRG neurons. Specifically, we wanted to determine whether the three major groups of DRG MA currents (RA, IA, and SA) were induced by three separate MA channels or whether two different MA ion channels could induce macroscopic currents with similar inactivation kinetics. We also reasoned that by acquiring single-channel conductance data, we might be able to predict how many MA ion channels are expressed in DRG neurons. To associate whole-cell MA macroscopic current properties to single-channel data, we combined two different patch clamp electrophysiology techniques (Fig. 1 A). First, in the whole-cell patch clamp mode, MA currents were recorded at −80 mV by indenting the soma with a blunt glass probe at 1 µm increment. We avoided indenting the soma at a higher threshold to prevent seal rupture. However, this was unavoidable in some cases, and such cells were discarded from analysis. Next, the recording electrode was slowly retracted to excise the membrane and form an outside-out patch. The recording electrode was connected to a high-speed pressure clamp (HSPC) system, which allowed us to apply positive pressure increments to induce stretch-activated currents from the excised patch (the time from whole-cell recording to stretch-activated recording after patch excision was within 30 s to 1 min; Fig. 1 B). We frequently observed stretch-activated single-channel events in excised patches in response to pressure stimulus within the range of 20–100 mmHg, allowing us to measure single-channel amplitude and conductance at −80 mV. Interestingly, macroscopic stretch-activated currents were rarely observed, suggesting that the number of channels in excised patches was generally low.

In cultured WT DRG neurons, we were able to successfully record membrane indentation-induced as well as stretch-activated responses from 61 cells across six mice. 54 of these cells had whole-cell MA currents, out of which 19 cells (31%) had a discernable stretch-activated channel response (Fig. 1 B). The low incidence of channel activity in excised patches is expected because of weak expression of endogenous MA channels, but yet, excised patches increase the frequency of observing stretch-activated currents due to a larger membrane patch compared with cell-attached patches (Lewis and Grandl, 2015). Furthermore, out of the seven cells that did not induce a whole-cell MA response, two cells had stretch-activated channel response. Because our whole-cell recordings were stopped prematurely, we think that the two cells with a stretch-activated channel response were from cells that expressed a high threshold MA channel. Or, these two cells represent MA channels in DRG neurons that are activated only by membrane stretch but not by membrane indentation.

We find that for whole-cell macroscopic currents, plotting the maximal current response vs. the inactivation time constant from a given cell gives an unbiased distribution of the current profile and highlights the heterogeneity in inactivation time constants. We observed a wide range in inactivation kinetics across different cells, and grouping cells into RA, IA, and SA based on a broad time constant range might not represent the spectrum of MA ion channel macroscopic currents. Therefore, we chose to portray the data as maximal current (Imax) vs. inactivation tau (τ) graphs (Fig. 1 C). We also find that ∼8% of all of the MA whole-cell currents are better fit with biexponential time constants (Table 1). Interestingly, stretch-activated single-channel analysis indicated that a fair amount of heterogeneity exists at the single-channel amplitude level, too (Fig. 1 D). Whereas most cells exhibited a single amplitude, some cells had two significantly different amplitudes, likely suggestive of two different MA channels in the same cell (Fig. 1 E). From the single-channel amplitudes, chord conductance was calculated and further analyzed to determine whether the cells can be grouped into classes of different conductance MA ion channels. To unbiasedly cluster the different conductance values into significant groups, we looked at the frequency distribution of the numerous conductance values from low (1 pS) to high (5 pS) bin centers (Fig. S1 and Fig. 1 F, left panel). Using statistics, we then distributed the cells into four distinct groups: 5 pS (3.9 pS, n = 1), 15 pS (15.9 ± 0.75 pS, n = 6), 20 pS (20.0 ± 0.60 pS, n = 7), and 25 pS (26.7 ± 0.96 pS, n = 5; Fig. 1 F, right panel). We were unable to draw a correlation between macroscopic currents properties and single-channel conductance (Fig. 1 G). This suggests that different MA ion channels in DRG neurons can induce macroscopic currents with similar inactivation kinetics and likely that all RAs or all IAs are not mediated by a single conductance channel.

We next tested if we could isolate PIEZO2-dependent stretch-activated single-channel currents in DRG neurons. It is debated in the field whether PIEZO2 is stretch-activated (Moroni et al., 2018; Szczot et al., 2021). A few studies have reported PIEZO2-dependent stretch-activated single-channel currents that can be altered by mutations in the PIEZO2 pore region (Coste et al., 2015; Geng et al., 2020; Verkest et al., 2022; Zhao et al., 2016), but robust macroscopic current properties of PIEZO2 have not been described or characterized. Therefore, we first determined whether PIEZO2 can be activated by membrane stretch when overexpressed in WT HEK and HEK-P1KO cells. In cell-attached patch clamp mode, we were able to record PIEZO2-induced stretch-activated currents in 40% of successful attempts; 15 out of 38 cells had a maximal response above that of vector-transfected cells (Fig. 2, A–C). Although the frequency of PIEZO2-induced responses was not as robust as PIEZO1, overall maximal responses were comparable, but statistically different from vector-transfected cells (Fig. 2 C). Interestingly, stretch-activated current properties of PIEZO2 are distinct from PIEZO1. PIEZO2-induced currents have slower inactivation constant (PIEZO2: 126 ± 30 ms, n = 8 vs. PIEZO1: 36 ± 5 ms, n = 5), larger steady-state current (PIEZO2: 63.6 ± 5.3%, n = 8 vs. PIEZO1: 20.7 ± 5.2%, n = 5), and higher pressure for half maximal activation; P₅₀ (PIEZO2: 50 ± 4 mmHg, n = 10 vs. PIEZO1: 25 ± 5.6 mmHg, n = 5; Fig. 2, D–F). Overall, our data strengthen the evidence to suggest that PIEZO2 can be activated by membrane stretch.

To identify which out of the five conductances in the stretch-activated single-channel data from WT DRG neurons belong to PIEZO2, and to determine whether this conductance could be correlated to RA macroscopic currents, we turned to a DRG subpopulation that is enriched with Piezo2. Based on single-cell RNA-seq data, we identified a subclass of neurons marked by the Neuropeptide Y receptor type 2 (Npy2r) as a unique expresser of Piezo2 compared with other neuronal subpopulations (Sharma et al., 2020; Zheng et al., 2019). Using a BAC transgenic mouse that marks Npy2r neurons with GFP, we performed our combined whole-cell and excised outside-out patch analysis to determine the macroscopic and single-channel profile within this population. We successfully collected whole-cell and stretch-activated responses from 20 cells (two mice), 19 of which had a MA whole-cell response and 7 cells had a stretch-activated channel response (∼37%), with 2 cells that exhibited macroscopic stretch-activated response (Fig. 3 A). From the MA whole-cell currents, we observed that the majority of them had an inactivation time constant faster than 20 ms but a number of cells had slower inactivation kinetics between the range of 20 and 40 ms (Fig. 3 B). 15% of the MA whole-cell currents had a biexponential decay (Table 1), and similar to WT DRG neurons, we observed some heterogeneity in the single-channel amplitude (Fig. 3 C). Frequency analysis of the conductance calculated from these amplitudes indicates that the cells could be distributed in three or two groups (Fig. S1; and Fig. 3, D and E). However, the statistical analysis failed to classify them into distinct groups, although we observe two predominant clusters, one at 15 pS (16.27 ± 0.3 pS, n = 3) and another at 20 pS (19.49 ± 0.3 pS, n = 4), conductance from all Npy2r population cells averaged at 18.11 ± 0.7 pS, n = 7 (Fig. 3 D, right panel). The 16.27 ± 0.3 pS value is similar to PIEZO2 unitary conductance in a heterologous expression system (Coste et al., 2015). However, since our recordings were done in excised patches with ionic compositions that were different from the previous studies, we remeasured PIEZO2 single-channel conductance under the same conditions used for the DRG neuron recordings.

Stretch-activated single-channel currents were recorded from excised outside-out patches pulled from HEK cells expressing PIEZO2. The ionic composition of the extracellular as well as intracellular solution matched those of the DRG recordings. Under these conditions, single-channel conductance of PIEZO2 was 16.18 ± 0.6 pS (n = 4), which closely matches one of the two clusters measured in the Npy2r population (16.27 ± 0.3 pS) but is also comparable with the overall average conductance of all Npy2r cells (18.11 ± 0.7; Fig. 3 F). Furthermore, the kinetics of the macroscopic stretch-activated response recorded from the Npy2r population (Fig. 3 A) was similar to PIEZO2-stretch currents recorded in cell-attached patches from HEK cells (Fig. 2 B). Together, results from the Npy2r population indicate that inactivation kinetics of whole-cell MA currents in the high Piezo2-expressing population is not restricted to a τ slower than 10 ms (as expected for RA currents), that single-channel conductance of MA channels expressed in this population can be grouped into the 20 pS class, and that a subpopulation of the conductance is likely mediated by PIEZO2.

To determine how the profile of stretch-activated single-channel data changes after Piezo2 deletion, we next performed our combined whole-cell and outside-out stretch analysis on DRG neurons extracted from the caudal region of Piezo2 KO mice (HoxB8-Cre; Piezo2^f/−), which lack PIEZO2-dependent most RA and some IA MA currents in in vitro DRG cultures (Fig. 4 A; Murthy et al, 2018b; Ranade et al., 2014; Woo et al., 2015). We collected whole-cell macroscopic responses from 59 neurons (four mice), out of which 33 neurons had MA currents (55%). The high percentage of non-responders matched previous data from these Piezo2 KO mice (Murthy et al, 2018b). Excised patches were pulled from all 60 neurons; 12 out of the 33 neurons that had a macroscopic MA response exhibited a stretch-activated single-channel response (∼36%), whereas 2 non-responder cells had a stretch-activated single-channel response.

Deletion of Piezo2 resulted in a loss of whole-cell macroscopic responses with an inactivation time constant of <20 ms, and compared with WT and Npy2r population, we saw the lowest percentage of MA currents with biexponential decay (3%; Fig. 4 B and Table 1). Interestingly, the single-channel analysis of the remaining cells with a MA macroscopic response revealed heterogeneity in single-channel amplitudes similar to the previous conditions (Fig. 4 C). While most cells had one single-channel amplitude, we did observe a cell that had two unitary amplitudes in the excised patch (Fig. 4 D). Frequency distribution and statistics on different groups led to the distribution of cells into three distinct classes of conductance values: 20 pS (19.8 ± 0.40 pS, n = 9), 25 pS (25.2 ± 0.6, n = 3), and 30 pS (29.72 ± 1.2 pS, n = 2; Fig. 4 E). It is interesting to note that the 5 pS and 15 pS conductance that we had observed in WT DRG neurons was not identified in the Piezo2 KO neurons. The 15 pS channel conductance can be attributed to PIEZO2, as deduced in Fig. 3. Together, this data further confirms that the 15 pS conductance seen in cells with RA and IA responses is largely PIEZO2-dependent (Fig. 4 F). Importantly, our single-channel analysis in the Piezo2 KO DRG neurons reveals that in addition to PIEZO2, there remain three other MA channels with a conductance of 20 pS, 25 pS, and 30 pS, which can induce whole-cell MA responses with inactivation time constants slower than 20 ms.

Evidence from patch clamp recordings in cultured DRG neurons and ex vivo skin nerve preperation recordings suggest that in addition to PIEZO2, other MA ion channels exist in somatosensory neurons, likely the sensor for acute mechanical pain observed at the behavioral level. Therefore, the molecular identity of the noxious pain sensor remains unknown. Single-cell RNA-seq data including patch-seq analysis from DRG neurons have instructed screening strategies to identify candidate genes that could explain non-PIEZO2-dependent mechanosensory modalities (Parpaite et al., 2021; Sharma et al., 2020; Zheng et al., 2019). Our study adds to this effort by providing additional insights into the nature of the MA ion channels expressed in DRG neurons. By obtaining stretch-activated single-channel data as well as membrane indentation-induced whole-cell inactivation kinetics for the same cell, we were able to decipher the number of MA channels expressed in DRG neurons. Importantly, our data suggest that although at the molecular level single-channel conductance can be clustered into three or four groups, no direct correlation can be made between the conductance and the whole-cell response. Therefore, the initial hypothesis that perhaps all IA currents are mediated by one MA ion channel, whereas all SA currents are mediated by another MA ion channel is challenged.

The stretch-activated single-channel analysis from WT DRG neurons highlight four different channel conductance (5, 15, 20, and 25 pS groups), likely suggestive of four different MA ion channels. Whereas the 15 pS channel is accounted for by PIEZO2 (this conductance is not observed after Piezo2 deletion), the presence of a 3.9 ± 0.63 pS (grouped as 5 pS in WT data, Fig. 1 F) channel is intriguing. This conductance was not observed in the Piezo2 KO DRG dataset. In WT DRG neurons, the 5 pS channel was observed only once; therefore, it is possible that the population of cells with this channel is very low and requires more sampling to be identified in the Piezo2 KO neurons. In the Piezo2 KO neurons, we observe an additional group of two cells with a conductance that clusters at 30 pS (29.7 ± 1.2, n = 2, Fig. 4 E). Although we observed cells with 30 pS conductance in WT neurons too, we couldn’t statistically differentiate them from the group of 25 pS channels (28.8 ± 0.9, n = 2, Fig. 1 F, left panel).

Among the known MA ion channel candidates that are expressed in DRG neurons, the 20 pS channel conductance could be attributed to PIEZO1, which has a similar conductance (Coste et al., 2015). It has recently been reported that PIEZO1 is expressed in itch-specific neurons marked by Nppb and is required for mechanical itch sensation (Hill et al., 2022). Interestingly, Npy2r expression overlaps with Nppb (Sharma et al., 2020). Therefore, it is possible that the MA currents from a population of the cells in the Npy2r dataset is mediated by Piezo1. In the Piezo2 KO neurons, there are a large number of cells (9 out of 14) that cluster around the 20 pS conductance value. Piezo2 deletion could lead to upregulation of Piezo1, which might result in a larger representation of Piezo1-like macroscopic and single-channel currents. However, this possibility is speculative and future studies involving detailed characterization of MA currents in DRG neurons lacking Piezo1 and Piezo2 will confirm this. The MA ion channels TMEM63A/B have a sub-picoamp single-channel amplitude in a heterologous expression system which could correspond to the 3 pS conductance, but further studies are needed to determine this conclusively (Murthy et al, 2018a). Finally, the experimental conditions under which our recordings were made are different from that of Cho et al. (2002), who reported two stretch-activated conductance in DRG neurons, 58.3 pS and 15 pS, therefore, we do not know how these two studies compare.

Activation of PIEZO2 by membrane stretch has been debated (Coste et al., 2015; Moroni et al., 2018; Szczot et al., 2021; Taberner et al., 2019). Here, we provide conclusive evidence to demonstrate that PIEZO2 is indeed stretch activated. PIEZO2-induced stretch currents have properties distinct from PIEZO1, such as smaller single-channel conductance, (previously described by Coste et al., 2015; Geng et al., 2020; Verkest et al., 2022), slower/non-inactivating currents, and higher pressure for half-maximal activation (P₅₀). Furthermore, the observation of stretch-activated currents with comparable properties in Piezo2-expressing DRG subpopulation bolsters our overexpression data. Future functional studies will determine what features in PIEZO2 molecular structure govern their unique channel properties.

Our data also suggest that in some cells there is more than one MA ion channel expressed since we observed two distinct channel conductances from the same patch, WT (cell 9 and cell 11, Fig. 1 E) and Piezo2 KO (cell 11, Fig. 4 D) DRG neurons. In all examples with two different amplitudes, the difference is significant and the single-channel amplitude values cluster at that specific current level. Therefore, it supports the hypothesis that a small percentage of DRG MA macroscopic currents is likely the combined activity from two different MA ion channels. Nonetheless, we cannot exclude variance in single-channel amplitude introduced from the methodology or modulation of conductance by splice variants, posttranslational modifications, or auxiliary subunits. For instance, it has been demonstrated that different splice forms of Piezo2 differ in their ion permeability, sensitivity to calcium modulation, and inactivation kinetics of MA whole-cell currents (Szczot et al., 2017). The ionic composition can also modulate PIEZO single-channel conductance; in solutions composed of physiological ionic concentrations single-channel conductance of PIEZO1 and PIEZO2 is half, compared with the conductance recorded in Na⁺- or K⁺-based solutions (Coste et al., 2010; Coste et al., 2015; Gottlieb et al., 2012). These observations together highlight the possibility that our single-channel data might be an overrepresentation and that two different conductances could belong to the same channel. Finally, we have to consider the possibility that excising the membrane patch could lead to modulation of PIEZO2 or other MA ion channel activity due to cortical cytoskeleton loss. Indeed, it has been shown that PIEZO1 stretch-activated currents from outside-out membrane patches excised from HEK cells require higher pressure for channel activation and display slower inactivation kinetics (Lewis and Grandl, 2015). Similar factors could affect the single-channel currents observed in our excised membrane patches from DRG neurons. For this reason, we refrained from correlating pressure thresholds or open probability of the single-channel currents to the whole-cell MA currents.

Our goal to combine MA whole-cell macroscopic currents with stretch-activated single-channel currents was to gain more insight into the properties of the underlying MA ion channels expressed in DRG neurons. However, it is important to note that whole-cell indentation of the membrane is biophysically different from membrane stretch. The possibility, therefore, remains that the channels activated by membrane stretch leading to the single-channel currents are different from the channels activated by membrane indentation that mediate the whole-cell macroscopic currents. Nonetheless, our data furthers our understanding of the nature of MA currents in DRG neurons and will inform future investigations on screening strategies in the quest for the noxious mechanical pain sensor.

Data for Figs. 1, 2, 3, and 4 are available in the published article, in Data S1, and are also available from the author upon reasonable request.

Crina M. Nimigean served as editor.

The author gratefully acknowledges Dr. Patapoutian (Howard Hughes Medical Institute, Dorris Neuroscience Center, Scripps Research, La Jolla, CA) for providing mice and insightful discussions and comments on the manuscript.

Funding was provided by the Silver Family Innovation Fund.

Author contribution: S.E. Murthy conceived the project, performed the experiments, analyzed the data, and wrote the manuscript.

Data for Figs. 1, 2, 3, and 4 are available in the published article, in Data S1, and are also available from the author upon reasonable request.