Deciphering and disrupting PIEZO1-TMEM16F interplay in hereditary xerocytosis

Abstract

Cell-surface exposure of phosphatidylserine (PS) is essential for phagocytic clearance and blood clotting. Although a calcium-activated phospholipid scramblase (CaPLSase) has long been proposed to mediate PS exposure in red blood cells (RBCs), its identity, activation mechanism, and role in RBC biology and disease remain elusive. Here, we demonstrate that TMEM16F, the long-sought-after RBC CaPLSase, is activated by calcium influx through the mechanosensitive channel PIEZO1 in RBCs. PIEZO1-TMEM16F functional coupling is enhanced in RBCs from individuals with hereditary xerocytosis (HX), an RBC disorder caused by PIEZO1 gain-of-function channelopathy. Enhanced PIEZO1-TMEM16F coupling leads to an increased propensity to expose PS, which may serve as a key risk factor for HX clinical manifestations including anemia, splenomegaly, and postsplenectomy thrombosis. Spider toxin GsMTx-4 and antigout medication benzbromarone inhibit PIEZO1, preventing force-induced echinocytosis, hemolysis, and PS exposure in HX RBCs. Our study thus reveals an activation mechanism of TMEM16F CaPLSase and its pathophysiological function in HX, providing insights into potential treatment.

Graphical abstract

Figure 1.

TMEM16F is responsible for Ca²⁺–activated lipid scrambling in murine RBCs. (A) Schematic of the fluorescence lipid scrambling assay in RBCs. (B-C) Ca²⁺ (red)–induced phosphatidylserine (PS) exposure (AnV, green) in the RBCs from TMEM16F wild-type (WT; B) and TMEM16F knockout (KO; C) mice. Ten micromolar ionomycin was used to trigger Ca²⁺ influx and subsequent lipid scrambling. White boxes show an enlarged view. (D) Time-dependent lipid scrambling reported by AnV in the TMEM16F WT and KO murine RBCs induced by ionomycin. Error bars represent standard error of the mean (SEM). WT (n = 16 cells) and KO (n = 15 cells) from at least 3 biological replicates. (E) Schematic of PCLSF assay to simultaneously monitor TMEM16F lipid scramblase and ion channels activities (see “Methods” for details). (F) Representative images of lipid scrambling activity in WT and TMEM16F-KO RBCs under PCLSF. PS exposure was detected by membrane binding of the fluorescent AnV conjugates. (G) Representative current traces in WT and TMEM16F-KO RBCs elicited by a voltage step protocol from –100 mV to +160 mV with 20 mV increment. The currents were recorded ∼2.5 minutes after the whole-cell patches were established. (H) Representative time course of AnV signal on WT and TMEM16F-KO RBCs. (I) Fluorescence intensity of AnV signal at 3 minutes. Two-sided Student t test, ∗∗∗P < .001 (n = 5). (J) I-V relationship of the currents recorded in (G). The currents were normalized to cell capacitance. The results are presented as mean ± SEM (n = 6). (K) Current density at +160 mV. The results are presented as mean ± SEM. Two-sided Student t test, ∗∗∗P < .001 (n = 6).

Figure 2.

Ca²⁺ influx through PIEZO1 activates TMEM16F in murine RBCs. (A) Representative cell-attach patch-clamp recording (schematic on left) in TMEM16F WT and knockout (KO) murine RBCs with 0 and –50 mmHg suction. The membrane was depolarized by voltage steps from –100 mV to +160 mV with 20 mV increment. A total of 2.5 mM Ca²⁺ in the pipette and a holding voltage at –60 mV enable Ca²⁺ influx and subsequent activation of TMEM16F current. (B) I-V relationship of the current recorded in A. (C) Statistical analysis of current at +160 mV. The results are presented as mean ± SEM. Two-sided Student t test, ∗∗∗P < .001 (n = 12 for WT and 10 for KO). (D) Representative images of 10 μM Yoda1-induced Ca²⁺ increase (red) and lipid scrambling (AnV, green) in RBCs from TMEM16F WT (left) and KO (right) mice. White boxes show enlarged views. (E) Time-dependent lipid scrambling reported by AnV in the RBCs from TMEM16F WT and KO mice stimulated by Yoda1. Error bars represent SEM. At least 3 biological repeats were done for WT (n = 18) and KO (n = 19) groups. (F) Flow cytometry quantification of Ca²⁺–induced lipid scrambling (reported by AnV 488) in WT (left) and TMEM16F-KO (right) murine RBCs in response to different concentrations of Yoda1. The y-axis represents cell counts. (G) Quantification of the percentage of PS-positive RBCs in response to different concentrations of Yoda1. Error bars represent SEM. The statistical analysis was done with 2-way analysis of variance followed by Sidak multiple comparisons tests, ∗∗∗∗P < .0001, n = 6 to 9 repeats for WT and KO RBCs at each dose. (H) Cartoon illustration of PIEZO1-TMEM16F coupling in healthy RBCs. Ca²⁺ influx through PIEZO1 activates TMEM16F in the vicinity.

Figure 3.

PIEZO1-TMEM16F coupling is enhanced in the RBCs from patients with HX with GOF PIEZO1 mutations. (A) Spatial distribution of 3 PIEZO1-HX mutations in a PIEZO1 structure. (B) Sequence alignment of PIEZO1 of HDs and patients with HX. (C) Yoda1 dose-dependent increase of intracellular Ca²⁺ in HD and HX RBCs. The results were averaged from 4 HD samples and 3 HX samples. Measurement of each sample was repeated at least 3 times. (D) Total Ca²⁺ influx shown in panel C as quantified by area under curve (AUC). (E) Flow cytometry analysis of 2 μM Yoda1-induced PS exposure in RBC samples from HD and patients with HX. (F) Percentage of PS-positive cells in RBCs of HDs and patients with HX with and without Yoda1 treatment. (G) AnV mean fluorescence intensity (MFI) in RBCs of HDs and patients with HX with (right) and without (left) Yoda1 treatment. MFI was calculated as geometric mean in FCS Express software. (H) Representative images of 2 μM Yoda1-induced PS exposure reported by AnV in RBCs from an HD and HX001 patient. (I) The averaged percentage of PS-positive cells in HDs (n = 4) and patients with HX (n = 3) at different time points after Yoda1 induction. For panels D,F-G,I, each dot represents the average of at least 3 repeats for each blood sample (n = 3 and 4 for HD and HX, respectively). Unpaired 2-sided Student t test. ∗∗P < .01; ∗∗∗P < .001; ns, no significance. (J) Representative cell-attached patch-clamp recording (schematic on left) in HD and HX001 RBCs elicited by a –50 mmHg negative pressure. The membrane was depolarized by voltage steps from –100 mV to +160 mV with 20 mV increment. A total of 2.5 mM Ca²⁺ was included in the pipette and a holding voltage at –60 mV enables Ca²⁺ influx and subsequent activation of endogenous TMEM16 current. (K) I-V relationship of suction-induced currents recorded in panel J. (L) Statistical analysis of peak current amplitude at +160 mV. The results are presented as mean ± SEM (unpaired 2-sided Student t test, ∗P < .05, n = 6 and 7 for WT and HX, respectively). (M) Illustration of enhanced PIEZO1-TMEM16F coupling resulting in elevated PS exposure.

Figure 4.

PIEZO1 inhibitor GsMTx-4 breaks PIEZO1-TMEM16F coupling and prevents excessive PS exposure in HX RBCs. (A) GsMTx-4 dose-dependently inhibits Yoda1–induced Ca²⁺ influx in HX001 RBCs. Error bars represent ± SEM from 3 replicates. (B-C) Representative flow cytometry data of GsMTx-4 inhibition on Yoda1 (2 μM)-induced PS exposure in HX001 (B) and HX002 (C) RBCs. CF488-AnV was used as a fluorescence PS marker. Each concentration of GsMTx4 was repeated in 3 independent experiments. (D) Dose-response curves of GsMTx-4 inhibition on Ca²⁺ increase (solid circle) and PS exposure (open square) in the RBCs from HX001 (black) and HX002 (red) patients. All signals were normalized to 0 GsMTx-4 condition and fitted with the Hill equation (see “Methods”). (E) IC₅₀ of GsMTx-4 on Yoda1-induced Ca²⁺ influx and PS exposure in HX001 and HX002 RBCs. Unpaired 2-sided Student t test. ∗∗P < .01, n = 3 to 4 repeats.

Figure 5.

Benzbromarone (Benz) suppresses Yoda1-induced PS exposure by inhibition of PIEZO1. (A) Benz dose-dependently inhibits Yoda1-induced PS exposure in HX001 RBCs. The flow cytometry experiments were done using CF488-AnV as a PS marker. (B) Benz dose-dependently inhibits Yoda1–induced PS exposure in HX001 RBCs. The AnV signals were normalized to zero Benz condition. The signals were fitted with the Hill equation (see “Methods”). Three independent repeats were done for each Benz concentration. (C) Representative PCLSF images of TMEM16F-mediated PS exposure in the absence (top) and presence of 20 μM extracellular Benz (bottom). Images were acquired every 5 seconds after membrane break-in under whole-cell configuration. TMEM16F was activated by pipette Ca²⁺. (D) Time course of TMEM16F-mediated PS exposure under PCLSF with and without Benz. t_1/2 is the time for the AnV intensity to reach half maximum within the recorded time frame. The results are presented as mean ± SEM (n = 7 for each group). (E) Comparison of t_1/2 with and without Benz. The results are presented as mean ± SEM. Statistical analysis was done by unpaired 2-sided Student t test (ns means no significant, n = 7 for each group). (F) Benz dose-dependently inhibits Yoda1 (2 μM)-induced Ca²⁺ influx in HX001 RBCs. Error bars represent SEM from 4 independent repeats. (G) Dose-response curve of Benz inhibition on Yoda1-induced Ca²⁺ influx in HX001 RBCs. The signals were normalized to zero Benz condition. The Ca²⁺ signals were fitted with the Hill equation (see “Methods”). (H) Representative cell-attached patch-clamp recording of PIEZO1 current in the presence and absence (Control) of 20 μM Benz in the pipette solution. Human PIEZO1 was overexpressed in HEK293T cells. The current was elicited by pressure clamp from 0 to −60 mmHg with a holding potential at −80 mV. (I) PIEZO1 current-pressure relationship with and without extracellular Benz. Error bars represent SEM (n = 10 and 8 for control and Benz group, respectively). (J) PIEZO1 current amplitudes at –60 mmHg with and without extracellular Benz. Statistical analysis was done by unpaired 2-sided Student t test. (∗∗∗∗P < .0001, n = 10 and 8 for control and Benz groups, respectively).

Figure 6.

Inhibition of PIEZO1 prevents echinocytosis and hemolysis in HX001 RBCs. (A) Representative images of RBCs from a HD before and after centrifugation. (B-C) Representative images of the RBCs from HX001 with various concentrations of GsMTx-4 (B) and Benz (C) before and after centrifugation. (D-E) Effect of GsMTx-4 (D) and Benz (E) on spin-induced RBC echinocytosis (see “Methods” for details). (F) Representative photos of 2 μM Yoda1(Y1)-induced hemolysis in HX RBCs with and without 5 μM GsMTx-4. 0 Yoda1 (left) and H₂O (right) serve as negative and positive controls, respectively. (G) Representative images of 2 μM Yoda1-induced hemolysis with and without 20 μM Benz at different time points as indicated. (H) Time course of 2 μM Yoda1(Y1)-induced with and without 5 μM spider toxin GsMTx-4 in HX001 RBCs. Results were normalized to the water-induced fully lysed group. (I) Time course of 2 μM Yoda1 (Y1)-induced hemolysis with and without 20 μM Benz in HX001 RBCs. Results were normalized to the water-induced fully lysed group. (J) Yoda1-induced RBCs hemolysis percentage at 120 minutes with and without GsMTx-4 and Benz. Unpaired 2-sided Student t test, ∗∗∗∗P < .0001, n = 3 to 6 repeats for each condition. (K) Partially inhibiting PIEZO1 can effectively disrupt PIEZO1-TMEM16F coupling in HX RBCs and prevent HX-associated complications.