Phosphatidic acid is an endogenous negative regulator of PIEZO2 channels and mechanical sensitivity

Abstract

Mechanosensitive PIEZO2 ion channels play roles in touch, proprioception, and inflammatory pain. Currently, there are no small molecule inhibitors that selectively inhibit PIEZO2 over PIEZO1. The TMEM120A protein was shown to inhibit PIEZO2 while leaving PIEZO1 unaffected. Here we find that TMEM120A expression elevates cellular levels of phosphatidic acid and lysophosphatidic acid (LPA), aligning with its structural resemblance to lipid-modifying enzymes. Intracellular application of phosphatidic acid or LPA inhibits PIEZO2 but not PIEZO1 activity. Extended extracellular exposure to the non-hydrolyzable phosphatidic acid and LPA analog carbocyclic phosphatidic acid (ccPA) also inhibits PIEZO2. Optogenetic activation of phospholipase D (PLD), a signaling enzyme that generates phosphatidic acid, inhibits PIEZO2 but not PIEZO1. Conversely, inhibiting PLD leads to increased PIEZO2 activity and increased mechanical sensitivity in mice in behavioral experiments. These findings unveil lipid regulators that selectively target PIEZO2 over PIEZO1, and identify the PLD pathway as a regulator of PIEZO2 activity.

Fig. 1. Changes in cellular lipid content mediate inhibition of PIEZO2 by TMEM120A.

a LC-MS/MS experiments using N2A-Pz1-KO cells transiently transfected with vector (mock, black), TMEM120A (red), or TMEM120B (blue) as described in the methods section. Scatter plots and mean ± SEM of the relative saturated lipid intensities detected for n = 3 independent transfections for LPA (ANOVA, F(2,6) = 9.274, p = 0.0146), phosphatidic acid (PA) (ANOVA, F(2,6) = 1148, p = 1.77 × 10⁻⁸), DG (ANOVA, F(2,6) = 5.528, p = 0.0435), TG (ANOVA, F(2,6) = 44.500, p = 0.0002). Post-hoc Tukey test p-values are displayed on the plots. b Scheme of the Kennedy pathway for de novo TG synthesis. Lipids are depicted with short acyl chains because of spatial restrictions. c Structure of TMEM120A with magnified (black box) residues interacting with CoA generated using PyMOL from the publicly available pdb file 7F3T. d–h Whole-cell voltage-clamp experiments at −60 mV in N2A-Pz1-KO cells transiently transfected with PIEZO2-GFP, and tdTOMATO (black, n = 14 cells), TMEM120A-tdTom-WT (red, n = 15 cells), or TMEM120A-tdTom-W193A (blue, n = 17 cells). d Mechanically activated currents were evoked by indentations of increasing depth with a blunt glass probe. Current amplitudes are plotted (mean ± SEM) and statistical differences for the area under the curve (AUC) for 4.0–8.0 µm stimuli are calculated with the two-tailed Mann–Whitney test. e Scatter and mean ± SEM for the threshold of membrane indentation to elicit mechanically activated current (two-tailed Mann–Whitney). Mechanical threshold is only quantifiable for cells where membrane indentation elicited current, thus non-responding cells reduce the n for PIEZO2-GFP + tdTOMATO to n = 13 cells and for PIEZO2-GFP + TMEM120A-tdTom-W193a to n = 12 cells. f Percentage of responding cells. g Scatter and mean ± SEM for maximum currents (Kruskal–Wallis, χ² = 17.467, df = 2, p = 0.0001; two-tailed Mann–Whitney tests displayed). h Representative current traces.

Fig. 2. Phosphatidic acid inhibits PIEZO2 but not PIEZO1.

Whole-cell patch-clamp experiments at −60 mV in N2A-Pz1-KO cells transiently transfected with PIEZO2-GFP or PIEZO1-GFP with patch-pipette solution supplemented with the indicated lipids as described in the methods section. a–d PIEZO2-GFP transfected cells supplemented with 300 μM dioctanoyl-phosphatidic acid (PA, n = 14 cells) or lipid-free (n = 15 cells). a Current amplitudes are plotted (mean ± SEM) and statistical difference for the area under curve (AUC) for 4.0–8.0 µm stimuli calculated with the two-tailed Mann–Whitney test. b Membrane indentation depth threshold to elicit mechanically activated current (two-tailed t-test). Mechanical threshold is only quantifiable for cells where membrane indentation elicited current, thus non-responding cells reduce the n for dioctanoyl-PA treated cells to n = 6 cells. c Maximum current amplitudes (Mann–Whitney). d Representative current traces. e–h PIEZO1-GFP-transfected cells supplemented with 300 μM dioctanoyl-PA (n = 15 cells) or lipid-free (n = 15 cells). e Current amplitudes are plotted (mean ± SEM) and the statistical difference of AUC for 4.0–8.0 µm stimuli is calculated with the two-tailed Mann–Whitney test. f Membrane indentation depth threshold to elicit mechanically activated current (two-tailed Mann–Whitney). g Maximum current amplitudes (two-tailed Mann–Whitney). h Representative current traces. All bar graphs are plotted with scatter and mean ± SEM.

Fig. 3. LPA inhibits PIEZO2 but not PIEZO1.

Whole-cell patch-clamp experiments at −60 mV in N2A-Pz1-KO cells transiently transfected with PIEZO2-GFP or PIEZO1-GFP with patch-pipette solution supplemented with the indicated lipids as described in the methods section. a–d PIEZO2-GFP-expressing cells supplemented with 30 μM palmitoyl-LPA (n = 17 cells) or lipid-free (n = 15 cells). a Current amplitudes are plotted (mean ± SEM) and statistical difference of AUC for 4.0–8.0 µm stimuli calculated with the two-tailed Mann–Whitney test. b Membrane indentation depth threshold to elicit mechanically activated currents (two-tailed Mann–Whitney). Mechanical threshold is only quantifiable for cells where membrane indentation elicited current, thus non-responding cells reduce the n for lipid-free treated cells to n = 12 cells and for palmitoyl-LPA-treated cells to n = 8 cells. c Maximum current amplitudes (two-tailed Mann–Whitney). d Representative current traces. e–h PIEZO1-GFP expressing cells supplemented with 30 μM palmitoyl-LPA (n = 17 cells) or lipid-free (n = 15 cells). e Current amplitudes are plotted (mean ± SEM) and the statistical difference of AUC for 4.0–8.0 µm stimuli is calculated with the two-tailed Mann–Whitney test. f Membrane indentation depth threshold to elicit mechanically activated current (two-tailed Mann–Whitney). g Maximum current amplitudes (two-tailed Mann–Whitney). h Representative current traces. i–l PIEZO2-GFP expressing cells supplemented with 30 μM stearoyl-LPA (n = 12 cells), 30 μM oleoyl-LPA (n = 12 cells), 30 μM linoleoyl-LPA (n = 12 cells), or lipid-free (n = 15 cells). i Current amplitudes are plotted (mean ± SEM). Statistical comparison of AUC for 4.0–8.0 µm stimuli calculated with the Kruskal–Wallis test (p-value displayed on plot; χ² = 16.979, df = 3, p = 7.13 × 10⁻⁴). Post-hoc two-tailed Mann–Whitney tests: Lipid-Free/Stearoyl-LPA, p = 3.28 × 10⁻⁴; Lipid-Free/Oleoyl-LPA, p = 1.64 × 10⁻³; Lipid-Free/Linoleoyl-LPA, p = 9.28 × 10⁻⁴; Stearoyl-LPA/Oleoyl-LPA, p = 0.087; Stearoyl-LPA/Linoleoyl-LPA, p = 0.791; Oleoyl-LPA/Linoleoyl-LPA, p = 0.0972. j Membrane indentation depth threshold to elicit mechanically activated current (ANOVA, F(3,37) = 3.58, p = 0.022; post-hoc Tukey tests displayed). Mechanical threshold is only quantifiable for cells where membrane indentation elicited current, thus non-responding cells reduce the n for lipid-free treated cells to n = 14 cells, for stearoyl-LPA-treated cells to n = 9 cells, for oleoyl-LPA-treated cells to n = 11 cells, and for linoleoyl-LPA-treated cells to n = 7 cells. k Maximum current amplitudes (Kruskal–Wallis, χ² = 20.374, df = 3, p = 1.42 × 10^-4; post-hoc two-tailed Mann–Whitney tests displayed). l Representative current traces. All bar graphs are plotted with scatter and mean ± SEM.

Fig. 4. Extracellular application of ccPA inhibits PIEZO2 currents.

Whole-cell patch-clamp experiments were performed at −60 mV as described in the methods section. a–d N2A-Pz1-KO cells transiently transfected with PIEZO2-GFP mock-treated (black, n = 24 cells) or treated with extracellular palmitoyl-ccPA overnight (orange, n = 24 cells). a Current amplitudes are plotted (mean ± SEM) and statistical difference in the area under the curve (AUC) for 4.0–8.0 µm stimuli is calculated with the two-tailed Mann–Whitney test. b Membrane indentation depth threshold to elicit mechanically activated currents (two-tailed t-test). Mechanical threshold is only quantifiable for cells where membrane indentation elicited current, thus non-responding cells reduce the n for palmitoyl-ccPA-treated cells to n = 20 cells. c Maximum current amplitudes (two-tailed Mann–Whitney). d Representative current traces. e–l Isolated mouse DRG neurons mock-treated (black, n = 100 cells) or treated with extracellular palmitoyl-ccPA overnight (orange, n = 100 cells). e Proportion of rapidly adapting (RA), intermediately adapting (IA), slowly adapting (SA), and non-responders (NR). Chi-squared test. f Representative current traces. g Rapidly adapting maximum current density (two-tailed t-test). h Intermediate adapting maximum current density (two-tailed t-test). i Slowly adapting maximum current density (two-tailed Mann–Whitney). j Membrane indentation depth threshold to elicit rapidly adapting current (two-tailed t-test). k Membrane indentation depth threshold to elicit intermediately adapting current (two-tailed t-test). l Membrane indentation depth threshold to elicit slowly adapting current (two-tailed Mann–Whitney). All bar graphs are plotted with scatter and mean ± SEM.

Fig. 5. Optogenetic activation of PLD inhibits PIEZO2.

Whole-cell patch-clamp experiments at −60 mV in N2A-Pz1-KO cells transfected with PIEZO2-GFP and active Opto-PLD (orange, n = 8 cells) or inactive Opto-PLD (black; H170A, n = 6 cells) as described in the methods section. a Scheme of blue light activation of the Opto-PLD system (Created with BioRender.com) with representative images for mCherry fluorescence (60x) before and after blue light exposure. b PIEZO2-GFP currents with fixed, continuous membrane indentations after blue light exposure normalized to currents before blue light. c Representative current traces. Blue shaded area indicates blue light exposure. Downward deflections indicate mechanically activate PIEZO2 currents. d Normalized PIEZO2-GFP currents co-expressed with inactive Opto-PLD after blue light exposure (Repeated-Measures ANOVA, F = 0.365, df = 2,10, p = 0.702; paired two-tailed t-tests displayed). e Normalized PIEZO2-GFP currents co-expressed with active Opto-PLD after blue light exposure (Repeated-Measures ANOVA, F = 21.386, df = 2,14, p = 5.546 × 10⁻⁵; paired two-tailed t-tests displayed). f Normalized PIEZO2-GFP currents with inactive or active Opto-PLD after 9.5 min of blue light exposure (two-tailed t-test). Figure 5/panel a created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license. All bar graphs are plotted with scatter and mean ± SEM.

Fig. 6. Phospholipase D (PLD) negatively regulates PIEZO2 but not PIEZO1.

Whole-cell patch-clamp experiments at −60 mV in transiently transfected N2A-Pz1-KO cells as described in the methods section. a Scheme of PLD inhibition by FIPI. b–e N2A-Pz1-KO cells transiently transfected with PIEZO2-GFP and were treated with DMSO (black, n = 26 cells) or 500 nM FIPI (orange, n = 31 cells) for 30 min, see methods for details. b Current amplitudes are plotted (mean ± SEM) and the statistical difference of AUC for 4.0–8.0 µm stimuli is calculated with the two-tailed Mann–Whitney test. d Membrane indentation depth threshold to elicit mechanically activated current (two-tailed Mann–Whitney). Mechanical threshold is only quantifiable for cells where membrane indentation elicited current, thus non-responding cells reduce the n for DMSO-treated cells to n = 24 cells and for FIPI-treated cells to n = 30 cells. d Maximum current amplitudes (two-tailed Mann–Whitney). e Representative current traces. f–i N2A-Pz1-KO cells transiently transfected with PIEZO1-GFP treated with 500 nM FIPI (orange, n = 25 cells) for 30 min or the vehicle DMSO (blue, n = 25 cells) for 30 min. f Current amplitudes are plotted (mean ± SEM) and the statistical difference of AUC for 4.0–8.0 µm stimuli is calculated with the two-tailed Mann–Whitney test. g Membrane indentation depth threshold to elicit mechanically activated currents (two-tailed Mann–Whitney). Mechanical threshold is only quantifiable for cells where membrane indentation elicited current, thus non-responding cells reduce the n for FIPI-treated cells to n = 24 cells. h Maximum current amplitudes (two-tailed Mann–Whitney). i Representative current traces. j–m N2A-Pz1-KO cells transiently transfected with PIEZO2-GFP alone (black, n = 12 cells), or together with PLD1 (blue), or PLD2 (orange). j Current amplitudes are plotted (mean ± SEM) and the statistical difference of AUC for 4.0–8.0 µm stimuli calculated was with the two-tailed Mann–Whitney test. k Membrane indentation depth threshold to elicit mechanically activated current (Kruskal–Wallis, χ² = 3.298, df = 2, p = 0.0192; two-tailed Mann–Whitney tests displayed). The mechanical threshold is only quantifiable for cells where membrane indentation elicited current, thus non-responding cells reduce the n for PIEZO2-GFP alone cells to n = 13 cells, for PIEZO2-GFP + PLD1 cells to n = 13 cells, and for PIEZO2-GFP + PLD2 cells to n = 9 cells. l Maximum current amplitudes (Kruskal–Wallis, χ² = 7,906, df = 2, p = 0.0192; two-tailed Mann–Whitney tests displayed). m Representative current traces. All bar graphs are plotted with scatter and mean ± SEM.

Fig. 7. PLD modulates native rapidly adapting mechanically activated currents in peripheral sensory neurons.

a–d Whole-cell patch-clamp at −60 mV in hiPSC-sensory neurons treated with 500 nM FIPI (orange, n = 25 cells) or DMSO (black, n = 20 cells) for 30 min as described in the methods section for details. a Current densities are plotted (mean ± SEM) and the statistical difference of AUC for 4.0–8.0 µm stimuli is calculated with the two-tailed Mann–Whitney test. b Membrane indentation depth thresholds to elicit mechanically activated current (two-tailed Mann–Whitney). Mechanical threshold is only quantifiable for cells where membrane indentation elicited current, thus non-responding cells reduce the n for DMSO-treated cells to n = 18 cells. c Maximum current densities (two-tailed Mann–Whitney). d Representative current traces. e Whole-cell patch-clamp at −60 mV in isolated mouse DRG neurons treated with DMSO (black, n = 10 cells) or FIPI (orange, n = 10 cells) as described in the methods section. Representative current traces (top), and membrane indentation depth thresholds to elicit mechanically activated RA currents (bottom) (two-tailed t-test). f–g Whole-cell current-clamp in isolated mouse DRG neurons treated with DMSO (black, n = 10 cells) or FIPI (orange, n = 10 cells) for 30 min as described in the methods section. f Representative traces (top), and membrane indentation depth threshold to elicit an action potential (bottom) (two-tailed t-test). g Representative traces (top), and current injection threshold to elicit an action potential (bottom) (two-tailed t-test) in the same cells used in panels e and f. In three cells the seal was lost before current injections could be performed. h–j Mechanical testing using Von Frey filaments and thermal testing using the Hargreaves apparatus were performed on mice receiving an I.P. injection of 3 mg/kg FIPI (n = 14 mice) or vehicle (n = 14 mice) as described in the methods section. h Experimental design (created with BioRender.com). i von Frey testing 50% mechanical threshold (two-tailed Mann–Whitney). j Hargreaves paw withdrawal latency (two-tailed Mann–Whitney). Figure 7/panel h was created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license. All bar graphs are plotted with scatter and mean ± SEM.