TRPM7 is an essential regulator for volume-sensitive outwardly rectifying anion channel

Animal cells can regulate their volume after swelling by the regulatory volume decrease (RVD) mechanism. In epithelial cells, RVD is attained through KCl release mediated via volume-sensitive outwardly rectifying Cl− channels (VSOR) and Ca2+-activated K+ channels. Swelling-induced activation of TRPM7 cation channels leads to Ca2+ influx, thereby stimulating the K+ channels. Here, we examined whether TRPM7 plays any role in VSOR activation. When TRPM7 was knocked down in human HeLa cells or knocked out in chicken DT40 cells, not only TRPM7 activity and RVD efficacy but also VSOR activity were suppressed. Heterologous expression of TRPM7 in TRPM7-deficient DT40 cells rescued both VSOR activity and RVD, accompanied by an increase in the expression of LRRC8A, a core molecule of VSOR. TRPM7 exerts the facilitating action on VSOR activity first by enhancing molecular expression of LRRC8A mRNA through the mediation of steady-state Ca2+ influx and second by stabilizing the plasmalemmal expression of LRRC8A protein through the interaction between LRRC8A and the C-terminal domain of TRPM7. Therefore, TRPM7 functions as an essential regulator of VSOR activity and LRRC8A expression.

A nimal cells regulate their cell volume even under hypotonic stress (due to decreased extracellular osmolarity or increased intracellular osmolarity) by displaying regulatory shrinking activity, termed the regulatory volume decrease (RVD), after passive osmotic cell swelling. The RVD response is attained through the release of water driven by channel-and transporter-mediated KCl efflux (see Reviews: 1,2 ). In epithelial cells, volume-regulatory KCl release was shown to primarily result from parallel activation of the volume-sensitive Cl − channels and Ca 2+ -activated K + channels [3][4][5] , the latter identified later as IK1 6 . Activation of these volume-regulatory channels is preceded by activation of a type of nonselective cation channels [7][8][9] , which was recently identified as TRPM7 in human epithelial cells and shown to be induced by membrane stretch associated with osmotic cell swelling 10 . Stretch-induced activation of Ca 2+ -permeable TRPM7 causatively leads to opening of Ca 2+activated K + channels by increasing the intracellular free Ca 2+ level 10 . While it is thus clear that there is a functional coupling between TRPM7 and Ca 2+ -activated K + channels, the possibility of functional coupling between TRPM7 and the volume-sensitive Cl − channel has not been studied.
The volume-sensitive Cl − channel was discovered in 1988 independently by two groups 3,11 and is called the volume-sensitive outwardly rectifying anion channel (VSOR) 12 , the volumeregulated anion channel 13 , or the volume-sensitive organic osmolyte-anion channel 14 with exhibiting common phenotypical properties. Recently, a leucine-rich repeat (LRR)-containing protein, LRRC8A, was identified as an essential core component of VSOR through unbiased genome-wide approaches 15,16 . It was also shown that functional VSOR activity requires that LRRC8A forms multimers with LRRC8C, 8D, or 8E [16][17][18] , creating functional heteromeric channels of various subunit configurations. However, there remains a possibility that an as-yet-undetermined porerelated component or regulatory subcomponent other than LRRC8 members is a prerequisite to VSOR activity (see Reviews: 19,20 ). Also, it is highly feasible that LRRC8A/C/D/E interacts with other protein(s), to form a fully functioning VSOR, because the LRR motif is known to mediate protein-protein interactions 21 . Thus, it is intriguing to examine whether TRPM7 exhibits any molecular interaction with LRRC8A, thereby facilitating the RVD process.
In the present study, we used electrophysiological and genetic approaches to investigate functional and molecular interactions between VSOR/LRRC8A and TRPM7 in human epithelial cellderived HeLa cells and chicken B cell-derived DT40 cells. We demonstrate that TRPM7 expression is essential for LRRC8A expression and VSOR activity and, thereby playing a crucial prerequisite role in the RVD process.

Results
TRPM7 regulates VSOR activity and LRRC8A expression in human epithelial HeLa cells. To examine the functional linkage between TRPM7 and VSOR activity, we first performed patchclamp measurements of VSOR Cl − channel activity in mocktransfected and TRPM7-siRNA-treated human epithelial HeLa cells, since HeLa cells endogenously express both Mg 2+ -sensitive TRPM7 channels 10 and DCPIB-sensitive VSOR channels 22 . Under Cs + -rich conditions that eliminate K + channel currents, hypotonic stimulation of mock-transfected control cells induces activation of anionic currents with biophysical properties characteristic of VSOR Cl − channel (see Reviews: [12][13][14]: instantaneous activation elicited by voltage pulses, followed by deactivation time course at high positive membrane potentials (Fig. 1a), and an outwardly rectifying current-voltage (I-V) relationship (Fig. 1b). The VSOR whole-cell currents were markedly suppressed in TRPM7-siRNA-treated HeLa cells (Fig. 1a, b). At +100 and −100 mV, the currents were suppressed to 50.0 ± 9.8 and 52.3 ± 7.6% (n = 5), respectively, broadly consistent with the decrease in TRPM7 mRNA expression, down to 45.3 ± 5.1% (n = 5), shown by RT-PCR ( Fig. 1c and Supplementary Fig. 3a) and observed in previous studies 10,23,24 . These results suggest that the endogenous VSOR activity in HeLa cells is functionally regulated by TRPM7 expression. Because LRRC8A has been described as an essential component of VSOR channels in human cells 15,16 , mouse cells 25 , and rat cells 26,27 , we sought to determine if there is a relationship between TRPM7 and LRRC8A. Using RT-PCR, we explored the possibility of the molecular linkage between TRPM7 and LRRC8A expression. Indeed, we found that expression of hLRRC8A mRNA was reduced to 59.4 ± 3.3% (n = 5) in HeLa cells treated with TRPM7-siRNA ( Fig. 1c and Supplementary Fig. 3a). The downregulation of LRRC8A expression in TRPM7-silenced HeLa cells supports the existence of a molecular relationship between TRPM7 and LRRC8A that, at least in part, be responsible for the observed functional coupling between TRPM7 and VSOR activity in HeLa cells.
In contrast to TRPM7, it appears that another mechanosensitive TRPM member, TRPM4 28 , is not involved in regulation of VSOR activity and LRRC8A expression, because siRNAmediated knockdown of endogenous TRPM4, as shown in Supplementary Fig. 1, failed to affect the VSOR whole-cell currents (a, b) and the endogenous expression of hLRRC8A (c) in HeLa cells.
To explore whether the regulatory role of TRPM7 is mediated by steady-state Ca 2+ influx through the plasmalemmal TRPM7 channel, effects of 2-day treatment with a TRPM7 blocker NS8593 29 or EGTA were observed in HeLa cells. Such long-time exposure to NS8593 (30 μM), as shown in Fig. 2 (NS8593), actually not only reduced TRPM7 currents (a, b) and the cytosolic free Ca 2+ level (c) to the extent similar to the Ca 2+ level reduced by TRPM7 knockdown (c: siRNA) but also largely suppressed expression of LRRC8A mRNA (d; Supplementary Fig. 3b) and whole-cell VSOR currents (e) again to the extent similar to the VSOR currents reduced by TRPM7 knockdown (e: siRNA). Treatment with extracellular EGTA (2 mM) for 2 days, as shown in Fig. 2 (EGTA), duplicated the suppressive effect of NS8593 (NS8593) on the cytosolic Ca 2+ level (c), LRRC8A mRNA expression (d; Supplementary Fig. 3b), and VSOR currents (e).
Thus, it appears that TRPM7-induced regulation of VSOR activity and LRRC8A expression is mediated by steady-state Ca 2+ influx via TRPM7 channels, which is involved in maintenance of the control level of cytosolic free Ca 2+ .
Functional expression of VSOR and TRPM7 in chicken DT40 cells. To further investigate the functional and molecular interactions between TRPM7 and VSOR, we next used the chicken B-cell line, DT40, commonly used to study gene disruption and recombination because of their high ratio of targeted random recombination 30 . First, we confirmed the endogenous activities of VSOR and TRPM7 in DT40 cells by patch-clamp experiments. As shown in Fig. 3a, the whole-cell Cl − current slowly developed after applying a hypotonic solution. The current responses to voltage step pulses showed instantaneous activation followed by slow inactivation kinetics at high positive membrane potentials ( Fig. 3a: B) as well as an outward-rectifying I-V relationship ( Fig. 3b: Hypo). In addition, the swelling-activated whole-cell current was blocked by a VSOR-specific blocker, DCPIB 31 , added to the bath solution ( Fig. 3a: C, b: +DCPIB). These electrophysiological and pharmacological properties are characteristic of VSOR channels and identical to those observed in human epithelial HeLa cells. On the other hand, endogenous TRPM7 activity in DT40 cells was revealed by removal of extracellular divalent cations even under isotonic conditions ( Fig. 4a: Iso), as reported previously in other cell types [32][33][34] , and additionally increased after application of hypotonic solution ( Fig. 4a: Hypo), as previously observed in HeLa cells 10 . The current-voltage relationship showed that osmotic cell swelling increased the currents in the entire range of voltages between +100 and −100 mV (Fig. 4b), as clearly seen at ±100 mV (Fig. 4c). Finally, the hypotonicity-augmented cationic currents were suppressed by application of extracellular Mg 2+ , and Mg 2+ -induced suppression of inward currents was more prominent than that of the outward currents (Fig. 5), as previously observed in HeLa cells 10 . Such voltage-dependent sensitivity of the currents in DT40 cells to extracellular Mg 2+ administration further confirmed that these are TRPM7-mediated currents. Therefore, endogenous VSOR and TRPM7 channel activities are present in DT40 cells.
Functional linkage between TRPM7 and VSOR upon osmotic cell swelling in DT40 cells. Next, we examined the possibility of a functional linkage between TRPM7 and VSOR activities in DT40 cells. Our experimental approach involved simultaneous measurements of cationic TRPM7 and anionic VSOR currents by clamping, in an alternating manner, the membrane potentials close to the equilibrium potentials for VSOR-conductive anions (Cl − plus aspartate − ) and TRPM7-conductive cations (Cs + plus Na + ), respectively. Since there is no reported data on the permeability ratio between Cl − and aspartate − (P asp /P Cl ) through the VSOR anion channel and that between Cs + and Na + (P Na /P Cs ) through the TRPM7 cation channel endogenously expressed in DT40 cells, we first determined these values by measuring the shifts of the reversal potential in response to replacement of extracellular Cl − and Cs + with aspartate − and Na + , respectively. We obtained a P asp /P Cl value for VSOR of 0.053 ± 0.004 (n = 11) and the P Na /P Cs value for TRPM7 of 0.76 ± 0.02 (n = 11). Based on these values, cationic TRPM7 currents and anionic VSOR currents were simultaneously measured at the equilibrium potentials for anions and cations, −30 and +30 mV, respectively, while at ramp-clamp, as shown in Fig. 6. The cationic current was predominantly activated within 2 min immediately after application of hypotonic solution ( Fig. 6a: filled circles) with shifting the reversal potential (E rev ) to the positive direction (open triangles) up to the mean peak (positive) value of 6.7 ± 0.6 mV (n = 7). On the other hand, thereafter, the anion current became predominantly activated (Fig. 6a: open circles) with shifting the E rev value to the negative direction (open triangles) down to the mean bottom (negative) value of −14.7 ± 0.7 mV (n = 7: P (=0.000000000047) <0.0001 compared to the above mean peak value by t-test). Here, as seen in Fig. 6a (I-V curves), the I-V relationship at the time point (i) corresponds almost solely to that of TRPM7 currents (see Fig. 4b). As time passed from time point (ii) to time point (iii), the VSOR-like I-V relationship (see Fig. 3b) became predominant (Fig. 6a: right panel). When peak current densities were plotted, the relationship of TRPM7-mediated cationic current measured at −30 mV and VSOR-mediated anionic current measured at +30 mV for each time point over 8 min after hypotonic stimulation in the same cell could be fitted well with a straight line of y = −5.6x − 2.0 (R 2 = 0.97), as shown in Fig. 6b, showing that there is indeed a functional correlation between VSOR and TRPM7 activity in DT40 cells.
Effects of TRPM7 expression on LRRC8A expression and RVD in DT40 cells. As shown earlier for HeLa cells (Fig. 1), downregulation of LRRC8A expression in the cells treated with TRPM7-siRNA suggests that a molecular relationship between TRPM7 and LRRC8A could be, at least in part, responsible for the observed functional coupling between TRPM7 and VSOR activity in HeLa cells. To pursue a more in-depth investigation into the molecular basis of this functional coupling, we used chicken DT40 cells which are more amenable to genetic manipulation. As shown in Fig. 7a, b and Supplementary Fig. 3c, when gallus TRPM7 (gTRPM7) was knocked out in DT40 cells (KO), LRRC8A mRNA expression was largely suppressed. On the other hand, when gTRPM7-deficient DT40 cells were complemented with human TRPM7 (hTRPM7), the mRNA expression level of gallus LRRC8A was prominently increased (Fig. 7a, b: +hM7 and Supplementary Fig. 3c: +hM7). As plotted in Fig. 6c, gTRPM7deficient DT40 cells complemented with wild-type (WT) hTRPM7 exhibited a functional correlation with a linear relationship, fitted with y = −8.6x -16.2 (R 2 = 0.83), between VSOR activity and TRPM7 activity, as was the case in control DT40 cells (Fig. 6b). These results indicate that both LRRC8A expression and VSOR activity are closely correlated with TRPM7 expression.
Since VSOR activity plays an essential role in cell volume regulation, RVD, after osmotic cell swelling by serving as the volume-regulatory Cl − efflux pathway (see Reviews: 12,20,35 ), we next examined the effects of knockout of gTRPM7 and complementary expression with hTRPM7 on RVD in DT40 cells. As shown in Fig. 7 (c, e), control DT40 cells responded to hypotonic stimulation (78% osmolarity) with RVD within 30 min after rapid a Representative I-V relationships of currents in response to ramp pulses (50 ms duration) from −100 to +100 mV before (control) and after 2-day treatment with 30 µM NS8593 (NS8593). b Mean peak current densities at +100 mV in the absence (black column) and presence of NS8593 (red column) (n = 6). * indicates a P (=0.000058) value of <0.0001 compared to the corresponding control (left column) by t-test. c Cytosolic Ca 2+ levels were determined by fura-2 ratio imaging in the cells before (control) and after treatment with 30 µM NS8593 for 2 days (NS8593), 2 mM EGTA for 2 days (EGTA), mock siRNA for 2-3 days (Mock), and TRPM7-siRNA for 2-3 days (siRNA) (n = 24). *P (=0.0000000014, 0.000000000067, and 0.0000000032 for NS8593, EGTA, and siRNA, respectively) <0.0001 compared to the corresponding control data by one-way ANOVA followed by the post hoc Tukey's test. d Representative PCR products of human TRPM7, LRRC8A, and GAPDH in the cells before (control) and after 2-day treatment with 30 µM NS8593 (NS8593) or 2 mM EGTA (EGTA) from three independent experiments. The expression of TRPM7 and LRRC8A mRNAs was greatly decreased in NS8593-or EGTA-treated cells, while that of GAPDH mRNA stayed almost constant. The nucleotide sequences of the PCR products obtained with TRPM7-and LRRC8A-specific primers were completely identical to the corresponding sequences in TRPM7 (human: NM_017672) and LRRC8A (human: NM_001127244), respectively. e Swelling-induced peak VSOR current densities recorded at +40 mV before (control) and after treatment with NS8593 (NS8593), EGTA (EGTA), negative control siRNA (Mock), and TRPM7-siRNA (siRNA) (n = 6-9). Each column represents the mean ± SEM (vertical bar) of n samples. *P (=0.038, 0.036, and 0.037 for NS8593, EGTA, and siRNA, respectively) <0.05 compared to corresponding control data by one-way ANOVA followed by the post hoc Tukey's test.
osmotic cell swelling (control); this RVD event was mostly abolished by a VSOR blocker DCPIB (+DCPIB). The RVD event was also largely inhibited in gTRPM7-deficient DT40 cells (KO), whereas it was significantly augmented in gTRPM7-deficient DT40 cells complemented with WT hTRPM7, as shown in Fig. 7 (d, e: +hM7). Taken together, these experiments show that TRPM7 expression upregulates the molecular expression of LRRC8A, thereby increasing VSOR activity and enhancing RVD efficacy.
Roles of the kinase domain and its enzyme activity of TRPM7 in VSOR regulation in DT40 cells. Since TRPM7 possesses both ion-conductive and enzymatic activities 36 , we next assessed whether its kinase domain or its kinase activity is involved in TRPM7-mediated VSOR regulation. We used gTRPM7-deficient DT40 cells (KO) and those complemented with (i) WT hTRPM7 (+WT), (ii) the Δ-kinase construct in which the entire α-kinase domain of hTRPM7 was deleted following amino acid 1569 37 (+Δ-kinase), and (iii) the K1648R construct in which the α-kinase was rendered inactive with a point mutation in the ATP binding site 37,38 (+K1648R). First, we confirmed the presence of expressed hTRPM7 channels in the plasma membrane of gTRPM7deficient DT40 cells by immunostaining using an antibody to the HA tag of hTRPM7 and viewing Alexa-488 fluorescence under a confocal laser microscope. Indeed, as shown in Fig. 8a, there is robust positive staining in the periphery region of gTRPM7deficient DT40 cells complemented with WT hTRPM7 (+WT). gTRPM7-deficient DT40 cells complemented with Δ-kinase Step pulses from −100 to +100 mV in 20-mV increments were applied at time points indicated by (i), (ii), and (iii). Bars above the recordings indicate the time during applications of hypotonic solution and DCPIB (10 µM). b Mean I-V relationships for isotonic (Iso) and hypotonicity-induced (Hypo) whole-cell currents before and during application of DCPIB (+DCPIB: n = 8). Each data point represents the mean ± SEM (vertical bar) of n samples. *P (=0.048, 0.028, 0.0499, 0.0054, 0.016, 0.0018, 0.00050, 0.00056, 0.0011, and 0.00031 at −100, −80, −60, −40, −20, +20, +40, +60, +80, and +100 mV, respectively) <0.05 compared to corresponding data in Hypo group by t-test. Fig. 4 Augmentation of whole-cell TRPM7 currents by hypotonic stimulation in DT40 cells. a Effects of hypotonic stimulation on whole-cell TRPM7 currents under ramp-clamp (every 5 s at 4 mV/ms) in the cells equilibrated with the ATP-free and low-Cl − intracellular (pipette) solution. Representative peak outward and inward currents recorded at +100 and −100 mV, respectively, before (Iso) and after (Hypo) application of hypotonic (86% osmolarity) solution. b Representative I-V relationships of currents in response to ramp pulses (50-ms duration) from −100 to +100 mV before (at i in a) and during (at ii in a) hypotonic stimulation. c Peak current densities at +100 (blank columns) and −100 mV (filled columns) in isotonic (Iso) and hypotonic (Hypo) conditions (n = 8). Each data point represents the mean ± SEM (vertical bar) of n samples. *P (=0.0029 and 0.00017 at +100 and −100 mV, respectively) <0.005 compared to corresponding data in Iso group by t-test.
COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-021-02127-9 ARTICLE COMMUNICATIONS BIOLOGY | (2021) 4:599 | https://doi.org/10.1038/s42003-021-02127-9 | www.nature.com/commsbio (+Δ-kinase) and K1648R (+K1648R) also showed similar staining patterns, although fewer +Δ-kinase cells were positively stained (Fig. 8b). By contrast, when gTRPM7 was knocked out, no Alexa-488 fluorescence could be seen in cells (Fig. 8a: KO). Next, western blot analysis using an antibody against the HA tag of hTRPM7 confirmed the hTRPM7 channel protein expression in DT40 cells. In fact, as shown in Fig. 8c and Supplementary  Fig. 4a, gTRPM7-deficinet DT40 cells complemented with WT TRPM7 or the K1648R mutant also showed bands of similar size (+WT, +K1648R), but gTRPM7-deficient DT40 cells Representative I-V relationships of currents in response to ramp pulses from −100 to +100 mV before (at i in a) and during application of Mg 2+ (at ii in a). c Peak current densities at +100 (blank columns) and -100 mV (filled columns) before (0Mg 2+ ) and after application of 1 mM Mg 2+ (1Mg 2+ ) (n = 13). Each data point represents the mean ± SEM (vertical bar) of n samples. *P (=0.048 and 0.021 at +100 and −100 mV, respectively) <0.05 compared to corresponding data in 0Mg 2+ group by t-test. cells. Currents were elicited by ramp-clamp (every 10 s at 4 mV/ms). The value of E rev was evaluated by analyzing the measured reversal potentials of the all current traces in response to ramp pulses. Right panels show representative I-V relationships of currents in response to ramp pulses from -100 to +100 mV applied at (i), at (ii), and at (iii) in the left panel. Insets represent I-V curves expanded in the y-axis direction. b, c Relationships between the mean peak current densities recorded at +30 mV (VSOR activity) and those at -30 mV (M7 activity) (n = 7) in WT DT40 cells (b) and in gTRPM7deficient DT40 cells complemented with hTRPM7 (c).
complemented with Δ-kinase were slightly smaller in size due to a C-terminal defect (+Δ-kinase). In contrast, no protein was detected in the cells in which endogenous gTRPM7 was knocked out (Fig. 8c: KO and Supplementary Fig. 4a).
Electrophysiological recordings of VSOR currents are shown in Fig. 8d, e and Supplementary Fig. 2a (+M7). In gTRPM7deficient cells, surprisingly, VSOR currents were almost abolished (KO). When these KO cells were stably transfected with WT hTRPM7, however, VSOR currents were rescued (+WT). In contrast, transfection with hTRPM4 failed to rescue VSOR activity in KO cells, as shown in Supplementary Fig. 2a  (+M4). In WT hTRPM7-expressing cells, the amplitude of whole-cell VSOR currents (270.4 ± 28.6 pA/pF; Fig. 8e: +WT) was much larger than that of VSOR currents recorded in control DT40 cells (see open circle in Fig. 3b at +100 mV; 72.8 ± 10.7 pA/pF, P (=0.000095) <0.0001 by t-test). In Δ-kinaseexpressing cells (+Δ-kinase), VSOR currents similar in magnitude to those in WT hTRPM7-expressing cells (+WT) were recorded in five of seventeen tested cells (29.4%), but could hardly be recorded in the other twelve cells (Fig. 8d: +Δ-kinase), as was the case of gTRPM7-deficient cells (KO). Thus, the averaged VSOR current in Δ-kinase-expressing cells was very much smaller than that in WT hTRPM7-expressing cells (Fig. 8e). In contrast, in K1648R-expressing cells, swelling-induced VSOR currents, similar in magnitude to those in WT hTRPM7-expressing cells, were amply recorded (Fig. 8d, e), although their α-kinase is inactive. The whole-cell TRPM7 currents recorded in these DT40 cells under hypotonic conditions are shown in Fig. 8f, g. In all K1648R-expressing cells (+K1648R), the TRPM7 current amplitude was not significantly different from that in WT hTRPM7-expressing cells (+WT). As expected, no TRPM7-like current was observed in KO cells (Fig. 8f, g). In K1648R-expressing cells, the TRPM7 current amplitude was not significantly different from that in WT hTRPM7-expressing cells (Fig. 8f, g). In Δ-kinase-expressing cells, as in the case of VSOR currents, almost no TRPM7 current was observed in most cells (Fig. 8f: +Δ-kinase), although five of seventeen tested cells (29.4%) exhibited TRPM7 currents comparable in magnitude to those in WT hTRPM7-expressing cells, and thus the averaged TRPM7 current in Δ-kinase-expressing cells was very much smaller than that in WT hTRPM7-expressing cells (Fig. 8g: +Δ-kinase). These results suggest that TRPM7-mediated regulation of VSOR activity is independent of kinase activity of TRPM7 but is dependent on the presence of its cytosolic αkinase domain, but not its enzyme activity.  -20). Each data point represents the mean ± SEM (vertical bar) of n samples. *P (=0.00035, 0.011, and 0.000070 for +DCPIB, KO, and +hM7, respectively) <0.05 compared to corresponding data in control data by one-way ANOVA followed by the post hoc Tukey's test.
Next, we examined possible spatial co-localization of TRPM7 and LRRC8A in HeLa cells transfected with TRPM7-EGFP and LRRC8A-mCherry. Only after a hypotonic challenge, as shown in Fig. 9 (a, b: LRRC8A-mCherry), LRRC8A was found to become localized to the periphery region of cells transfected with WT TRPM7 or the K1648R mutant, whereas such periphery localization of LRRC8A was not much observed in those transfected with the Δ-kinase mutant. Also, their merged images suggest that LRRC8A becomes apparently co-localized with WT TRPM7 or the K1648R mutant but not much with the Δ-kinase Fig. 8 Effects of gTRPM7 knockout and complementary expression with hTRPM7 or its mutant on the protein expression monitored by immunohistochemistry as well as on VSOR and TRPM7 currents recorded by whole-cell patch-clamp in DT40 cells. a Representative fluorescence micrographs of gTRPM7-KO DT40 cells immunostained with anti-HA antibody before (KO) and after complementary expression with WT hTRPM7 or its mutant (Alexa-488: green) as well as their bright field micrographs. Scale bar, 5 μm. b The percentage of TRPM7 localization to the plasma membrane is significantly lower in the C-terminal defects (+Δ-kinase: *P (=1.1 × 10 −233 ) <0.0001 compared to +WT data by χ 2 test). Observations and analyses were performed from 53-110 individual cells. c Western blot analyses of α-tubulin, HA-tagged hTRPM7 from KO, +WT, +Δ-kinase, and +K1648R DT40 cells. The data are representative of three independent experiments. d Representative whole-cell VSOR currents elicited by step pulses in isotonic conditions (Iso) and in hypotonic conditions (Hypo) in gTRPM7-knockout (KO) DT40 cells and those complemented with hTRPM7 (+WT), hTRPM7-Δ-kinase (+Δ-kinase), and hTRPM7-K1648R (+K1648R). e Peak VSOR current densities recorded at +100 mV (n = 8-26). Each column represents the mean ± SEM (vertical bar) of n samples. *P (=0.000000000069 and 0.0000000029 for KO and +Δ-kinase, respectively) <0.0001 compared to corresponding +WT data by one-way ANOVA followed by the post hoc Tukey's test. f Representative I-V relationships of whole-cell TRPM7 currents in KO, +WT, +Δ-kinase, and +K1648R DT40 cells exposed to hypotonic (86% osmolarity) solution. The I-V curves were recorded under ramp-clamp (50-ms duration applied every 5 s at 4 mV/ms) from −100 to +100 mV in the absence (i, iii, v, vii) and presence (ii, iv, vi, viii) of 1-mM Mg 2+ . g Peak current densities recorded at +100 mV (blank columns) and -100 mV (filled columns) in hypotonic conditions (n = 6-19). Each column represents the mean ± SEM (vertical bar) of n samples. *P (=0.0000029 and 0.0012 at -100 mV and 0.00000000020 and 0.0000015 at +100 mV for KO and +Δ-kinase, respectively) <0.005 compared to corresponding data in +WT cells by one-way ANOVA followed by the post hoc Tukey's test. mutant ( Fig. 9a: merged image). Then, a possible physical interaction of TRPM7 and LRRC8A was tested by coimmunoprecipitation of LRRC8A and TRPM7 extracted from osmotically swollen HEK293T cells co-transfected with LRRC8A-mChery and TRPM7-EGFP. As shown in Fig. 9c (WT; Supplementary Fig. 4b: WT) and Supplementary Fig. 2b (TRPM7-EGFP), LRRC8A was found to be co-precipitated with WT TRPM7 (IP). In contrast, such co-precipitated bands (IP) were not detected when co-transfected with LRRC8A-mCherry and TRPM4-EGFP (Supplementary Fig. 2b: TRPM4-EGFP). The K1648R mutant of hTRPM7 exhibited similar co-precipitation with LRRC8A (IP: K1648R), whereas co-precipitated bands were not much found with the Δ-kinase mutant (IP: Δ-kinase). These data strongly suggest that LRRC8A becomes localized at the plasma membrane and interacts with TRPM7 presumably via the C-terminal kinase domain of TRPM7 after osmotic cell swelling.

Discussion
VSOR and TRPM7 are ubiquitously expressed anion and cation channels, respectively, that have important roles in cell volume regulation, viability, proliferation, and death (see Reviews: 28,[39][40][41][42][43] ). Here, we provide molecular and electrophysiological evidence that by influencing LRRC8A expression and VSOR activity, TRPM7 plays a crucial prerequisite role in RVD. We also demonstrated, for the first time, that chicken B cell-derived DT40 cells endogenously express VSOR activity (Fig. 3) and its core molecule LRRC8A ( Fig. 7a and Supplementary Fig. 3c, b), and thereby can attain the RVD event after exposure to hypotonic solution (Fig. 7c,  e). The whole-cell VSOR currents activated upon osmotic cell swelling exhibited outward rectification, voltage-dependent inactivation kinetics, and limited but significant permeability to aspartate − (P asp /P Cl~0 .05). DT40 cells were also found to molecularly express TRPM7 (Fig. 7a and Supplementary Fig. 3c), as originally shown by Nadler et al. 32 . The whole-cell TRPM7 currents were observed to be activated by removal of extracellular divalent cations, augmented by osmotic cell swelling (Fig. 4), and blocked by Mg 2+ added to the extracellular solution (Fig. 5), as observed previously in other cell types 10,24,32,44 . For the first time in DT40 cells, here, we determined the P Na /P Cs value of 0.76, the value of which is slightly smaller than that observed in mouse Chinese hamster ovary CHO cells expressing TRPM7 36 . Finally, we show that the presence of TRPM7 kinase domain, but not its kinase activity, is required for VSOR activity (Fig. 8).
The functional coupling between VSOR and TRPM7 was shown through four lines of evidence: (i) gene-silencing knockdown of TRPM7 largely suppressed swelling-induced activation of VSOR currents in HeLa cells (Fig. 1). (ii) Pharmacological inhibition of TRPM7 activity and of TRPM7-mediated steadystate Ca 2+ influx largely reduced VSOR currents in HeLa cells (Fig. 2). (iii) In DT40 cells, TRPM7 knockout markedly abolished VSOR activity, whereas complementary expression with hTRPM7 in gTRPM7-deficient cells enhanced VSOR activity (Fig. 8d, e). (iv) The VSOR current amplitude is proportionally positively correlated to the TRPM7 current amplitude in WT DT40 cells and gTRPM7-deficient DT40 cells complemented with hTRPM7 (Fig. 6b, c). Our data also showed that the functional link between TRPM7 and VSOR is underpinned, in part, by the molecular interactions between TRPM7 and the VSOR core molecule LRRC8A based on the following observations: (i) TRPM7 gene silencing reduced expression not only of TRPM7 mRNA but also of LRRC8A mRNA in HeLa cells (Fig. 1c and Supplementary  Fig. 3a). (ii) TRPM7 knockout and its complementary expression reduced and increased, respectively, expression of LRRC8A mRNA in DT40 cells ( Fig. 7a and Supplementary Fig. 3c, b). (iii) Complementary expression with the Δ-kinase construct of TRPM7 very markedly suppressed not only TRPM7 currents but also VSOR currents in DT40 cells (Fig. 8d-g). (iv) Co-expression of TRPM7 Δ-kinase and LRRC8A suppressed plasmalemmal expression of LRRC8A in HeLa cells under hypotonic stimulation (Fig. 9a, b). (v) Co-expression of TRPM7 Δ-kinase and LRRC8A weakened the physical interaction between TRPM7 and LRRC8A ( Fig. 9c and Supplementary Fig. 4b). Based on these pieces of cumulative evidence, we here conclude that TRPM7 serves as an essential regulator of VSOR activity and LRRC8A expression via its C-terminal α-kinase domain. Since the K1648R mutant failed to affect VSOR activity in DT40 cells (Fig. 8d, e), the plasmalemmal LRRC8A expression in HeLa cells (Fig. 9a, b), and the physical interaction between TRPM7 and LRRC8A in HEK293T cells (Fig. 9c and Supplementary Fig. 4b), it is concluded that the enzyme activity of α-kinase domain is not involved in the regulatory role of TRPM7 on VSOR expression and activity.
Functional interplays between TRP cation channels and Ca 2+activated chloride channels were so far reported to be mediated by Ca 2+ influx via a number of TRP cation channels, such as TRPC2 45 , TRPC6 46 , and TRPV6 47 . Molecular or physical interaction between TRPs and TMEM16A/ANO1 was also found by Tominaga's group for TRPV4 48,49 and TRPV1 50 . On the other hand, overexpression of TRPV4 was found to induce downregulation of VSOR activity in HEK293 cells 51 . The present study adds to this growing theme of functional and molecular interactions between TRP cation channels and anion channels in regulating important aspects of cell physiology such as RVD.
Osmotic cell swelling has been shown to activate many types of TRP cation channels, including TRPV1 51,52 , TRPV2 53 75 have been shown to be involved in RVD facilitation. Using the genesilencing technique, we previously provided molecular evidence for an involvement of swelling-induced activation of TRPM7 in RVD in HeLa cells 10 . Here, molecular evidence for this involvement was supplemented by observations that TRPM7 gene knockout and complementary expression with hTRPM7 effectively abolished and augmented, respectively, the RVD event in DT40 cells (Fig. 7d, e).
In sum, we conclude that TRPM7 dually contributes to RVD not only by stimulating Ca 2+ -activated K + channels but also by augmenting volume-sensitive Cl − channels upon osmotic cell swelling. The former occurs when swelling-induced activation of TRPM7 causes Ca 2+ influx, which leads to Ca 2+ -induced Ca 2+ release from intracellular Ca 2+ stores 76 , thereby bringing about a sustained rise of intracellular-free Ca 2+ level and eventually activating IK1 Ca 2+ -activated K + channels (see Review: 43 ). The latter occurs when swelling-induced TRPM7 activation augments both functional activities of VSOR Cl − channels and molecular expression of the core component of VSOR, LRRC8A. We here showed, for the first time, that TRPM7 dually exerts an augmenting action on VSOR activity: first by molecularly increasing expression of LRRC8A mRNA through the mediation of steadystate Ca 2+ influx and second by stabilizing the plasmalemmal expression of LRRC8A protein through the molecular interaction with the C-terminal α-kinase domain in a manner independent of its enzyme activity.
Our findings have important implications because of the physiological and pathological roles of osmotic swelling, VSOR activity, and TRPM7 channels. Depending on cell and tissue types, these physiological roles could include immunological processes, insulin secretion, adipocyte insulin signaling, astrocyteneuron communication, etc. Dysfunction of RVD is known to be closely coupled to, and implicated in apoptotic, necrotic, and ischemic cell death processes under physiological and pathological conditions (see Reviews: 2,43,77 ). Therefore, the molecular and functional link between TRPM7 and VSOR could an efficient target for therapeutic development and clinical treatments.

Methods
Cell culture. Human cervical HeLa cells or human embryonic kidney 293T (HEK293T) cells were grown as a monolayer in modified Eagle's medium or Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 40 U/ml penicillin G, and 100 μg/ml streptomycin under 95% air-5% CO 2 atmosphere at 37°C. For electrophysiological experiments, the cells were detached from the plastic substrate and cultured in suspension with stirring for 15-300 min. For some experiments, HeLa cells were provided after culturing for 2 days in the presence of 30 μM NS8593 or 2 mM EGTA. Chicken B cell-derived DT40 cells were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum, 1% chicken serum, 100 U/ml penicillin, and 100 μg/ml streptomycin in a humidified 95% air, 5% CO 2 atmosphere at 37°C. DT40 cell line construction. Construction of a stable gTRPM7-deficient DT40 cell line was made as previously described 32,37 . Briefly, the targeting construct pLTRPC7 (TRPM7) -bsr was introduced into DT40 cells by targeting the second TRPM7 allele of the gTRPM7 heterozygotes. The gTRPM7-deficient cell line was obtained by culturing in a medium supplemented with 10 mM Mg 2+ and selecting with Blasticidin. Tetracycline-inducible and stable expression of hTRPM7-WT, hTRPM7-K1648R, and hTRPM7-Δ-kinase DT40 cells was made as reported previously 37 . Briefly, the pcDNA TM 4/TO vector was prepared to contain a sequence in which two tetracycline operator sequences (TetO 2 ) are inserted between the TATA box of the CMV promoter and the transcription initiation site. The hTRPM7-constructed pcDNA TM 4/TO vectors (Invitrogen) were transfected into a gTRPM7-deficient DT40 cell line and subjected to a drug selectivity test supplemented with blasticidin and zeocin.
Construction of plasmid. The full-length human TRPM7 cDNA 24 was deleted and replaced with cDNAs of the following two different mutants prepared by using the QuikChange Lightning Site-Directed Mutagenesis Kits (Agilent Technologies, CA, USA): the Δ-kinase mutant, in which the entire α-kinase domain following amino acid 1569 in TRPM7, was deleted 37 , and the K1648R mutant, the enzyme activity of which is inactive by a point mutation at the ATP binding site in the amino acid α-kinase domain 37,38 . Full-length human LRRC8A cDNA 78 , human WT TRPM7, TRPM7-Δ-kinase, TRPM7-K1648R, and human TRPM4 79 have been subcloned into pEGFP-N (Takara-Bio, Shiga, Japan) or pmCherry-N (Takara-Bio). The sequence was confirmed by sequencing the entire insertion area. The recombinant plasmids for heterologously expression of pIRES2-EGFP-TRPM7, and pIRES2-EGFP-TRPM4 were subcloned into pIRES2-EGFP vector (Clontech, Mountain View, CA), as previously described 24,79 . Briefly, the end of full-length human TRPM7 cDNA and human TRPM4 cDNA was cleaved with an appropriate restriction enzyme and inserted into the multicloning site of pIRES2-EGFP.
Electrophysiology. Cells were dissociated by mechanical agitation and lodged onto coverslips placed in tissue culture dishes. Membrane currents of these cells were recorded at room temperature (22-27°C) using the whole-cell mode of the patchclamp technique, with an Axopatch 200B (Axon Instruments/Molecular Devices, Union City, CA, USA) patch-clamp amplifier. For whole-cell recordings, patch electrodes prepared from borosilicate glass capillaries had an input resistance of 3-7 MΩ. Current signals were filtered at 5 kHz with a four-pole Bessel filter and digitized at 20 kHz. pCLAMP software (version 10.5.1.0; Axon Instruments/ Molecular Devices) was used for command pulse control, data acquisition, and analysis. Data were also analyzed using Origin software (OriginLab Corp., Northampton, MA, USA). For whole-cell recordings, the series resistance was compensated (to 70-80%) to minimize voltage errors, and the liquid junction potential emerged between the extracellular solution and the pipette solution was evaluated by using Clampex software (Axon Instruments/Molecular Devices) and compensated before each experiment.
For VSOR current measurements, the amplitude of whole-cell current was recorded by applying alternating pulses of ±40 mV (500-ms duration) from a holding potential of 0 mV every 15 s. To observe voltage dependence of the current profile and the inactivation kinetics at large positive potentials, step pulses (800-ms duration) were applied from a pre-potential of −100 mV to test potentials of −100 to +100 mV in 20-mV increments. The amplitude of the instantaneous current was measured at~5 ms after the start of the step pulse. Whole-cell recordings of VSOR currents were performed using the external solution containing (in mM) 110 CsCl, 5 MgSO 4 , and 10 HEPES (pH 7.4 adjusted with CsOH, and osmolality adjusted to 320 mosmol/kg H 2 O in isotonic solution or 275 mosmol/kg H 2 O in hypotonic solution with D-mannitol). The pipette solution contained (in mM) 110 CsCl, 2 MgSO 4 , 1 EGTA, 10 HEPES, 2 Na 2 ATP, and 0.3 GTP (pH 7.3 adjusted with CsOH, and osmolality adjusted to 300 mosmol/kg H 2 O with D-mannitol; ionic strength 130 mM). The liquid junction potential emerged between these solutions was evaluated to be +0.5 mV. For Fig. 2e, cell swelling was induced by adjusting the intracellular osmolality to 360 mosmol/kg H 2 O with D-mannitol and recorded the amplitude of whole-cell currents. To evaluate the P asp /P Cl value for VSOR currents in DT40 cells, CsCl was replaced with Cs-aspartate in the pipette solution.
For TRPM7 current recordings, ramp pulses were applied from −100 to +100 mV at a speed of 1 mV/ms. The intracellular (pipette) solution contained (in mM): 100 Cs-aspartate, 1 EGTA, 10 HEPES, and 0.5 CsCl (pH adjusted to 7.4 with CsOH, and osmolality adjusted to 300 mosmol/kg H 2 O with D-mannitol; ionic strength~110 mM). Since VSOR activation is known to be induced or facilitated by reduction of intracellular ionic strength [80][81][82][83][84][85] , we here eliminated ATP and most Cl − ions from the intracellular solution to prevent possible VSOR Cl − current contamination 2,12 . The bath solution contained (in mM): 100 Na-aspartate (or Cs-aspartate) and 10 HEPES (pH adjusted to 7.4 with NaOH (or CsOH), and osmolality adjusted to 320 mosmol/ kg H 2 O with D-mannitol). The liquid junction potential emerged between these solutions was evaluated to be +7.9 mV. To estimate the P Na /P Cs value for TRPM7 currents in DT40 cells, Cs-aspartate was replaced with Na-aspartate in the pipette solution.
Mean cell volume measurements. Mean cell volume was measured at room temperature by electronic sizing with a Coulter-type cell size analyzer (CDA-500; Sysmex, Hyogo, Japan). The mean volume of the cell population was calculated from the cell volume distribution measured after the machine was calibrated with latex beads of known volume. Isotonic or hypotonic solution (275 or 215 mosmol/kg H 2 O adjusted by D-mannitol) contained (in mM) 100 NaCl, 5 KCl, 0.4 MgCl 2 , 0.42 CaCl 2 , 10 D-glucose, and 5 HEPES (pH 7.4 adjusted by NaOH). Relative cell volumes in Fig. 7 are defined by the following equation: relative cell volume = V A /V Ctl , where V Ctl and V A are the mean cell volumes before and after hypotonic challenge, respectively.
Modification of gene expression. To reduce the expression of human TRPM7 or human TRPM4, siRNA-mediated knockdown was performed in HeLa cells at 70-80% confluence. For siRNA transfection, Lipofectamine 2000 (Invitrogen Life Technologies, Carlsbad, CA, USA) was employed according to the manufacturer's protocols. siRNAs for hTRPM7 or hTRPM4 were designed and synthesized by Sigma-Aldrich. The target sequence and knockdown efficiency of the siRNA used were those previously confirmed 10,87 . A negative control siRNA conjugated with Alexa Fluor 488 (AllStars; Qiagen, Hilden, Germany) was used for mock cells. Cells were used in experiments 48-72 h after siRNA transfection.
Gene expression of TRPM7-WT, TRPM7-K1648R, and TRPM7-Δ-kinase in the DT40 cell line was induced by administration of 1 µg/ml tetracycline for 24 h, and its efficiency has been described previously 37 . gTRPM7-deficient DT40 cells were transfected with a recombinant plasmid pIRES2-EGFP vector (Clontech, Mountain View, CA), pIRES2-EGFP-TRPM7, or pIRES2-EGFP-TRPM4, by lipofectamine 2000 reagent. Briefly, DNA and lipofectamine 2000 were mixed in Opti-MEM TM medium (Thermo Fisher Scientific) to form a complex. The formed complex was added to a 6-cm non-coated dish containing DT40 cells and allowed to stand in a CO 2 incubator at 37°C for 6 h. Then, the cells were transferred to and incubated in fresh medium for an additional 24-48 h. Transfection of LRRC8A-mCherry with pEGFP-N, TRPM4-EGFP, or TRPM7-EGFP into HEK293T or HeLa cells were also performed by using Lipofectamine 2000 for 36-48 h.
Intracellular Ca 2+ measurements. The intracellular Ca 2+ concentration was measured by ratio imaging of fura-2 fluorescence using SPARK10M plate reader (Tecan Group Ltd., Männedorf, Switzerland). After culturing 1 × 10 4 HeLa cells in 96-wells, they were treated under each condition, cultured for 48 h, and washed with PBS. Fura-2 acetoxymethyl (2.5 μM: Dojindo Laboratories, Kumamoto, Japan) in DMEM medium was added to each well of a 96-well plate, and the plate was incubated at 37°C for 30 min, and then washed twice with Tyrode solution containing (in mM) 140 NaCl, 5 KCl, 2 CaCl 2 , 1 MgCl 2 , 10 HEPES, and 10 D-glucose (pH adjusted to 7.4 with NaOH, and osmolality adjusted to 300 mosmol/kg H 2 O with D-mannitol). Plates were cultured at 37°C for 1 h to stabilize cells in Tyrode solution. By exciting fura-2 fluorescence at 340/380 nm, intracellular Ca 2+ fluorescence signals were measured using a 510-nm filter. Fluorescence intensity was calculated as the F340/F380 ratio and shown as changes in the intracellular free Ca 2+ concentration.
Statistics and reproducibility. All data are expressed as means ± SEM. We accumulated the data for each condition from at least three independent experiments. We evaluated statistical significance with Student's t test for comparisons between two mean values. A value of P < 0.05 was considered significant. For Figs. 2c, e, 7b, e, 8e, g and Supplementary Fig. 2a, statistical differences of the data were evaluated with one-way analysis of variance followed by a Tukey's test for multiple comparisons and considered to be significant at P < 0.05. For Figs. 8b and 9b, the χ 2 test was made to evaluate the association between two categorical variables with a value of P < 0.05 considered to be significant.