The TRPM6 Kinase Domain Determines the Mg·ATP Sensitivity of TRPM7/M6 Heteromeric Ion Channels*

Background: TRPM6 and TRPM7 combine ion channel and α-kinase functions. Results: ATP inhibits TRPM7 but not TRPM6 or heteromeric TRPM6/M7 channels. Disruption of phosphorylation activity of TRPM6 kinase re-establishes ATP sensitivity to heteromeric channels. Conclusion: TRPM6 uncouples heteromeric channels from cellular energy levels. Significance: The altered characteristics of TRPM6/M7 compared with homomeric TRPM7 may enhance Ca2+ and Mg2+ transport in tissues co-expressing both proteins. The transient receptor potential melastatin member 7 (TRPM7) and member 6 (TRPM6) are divalent cation channel kinases essential for magnesium (Mg2+) homeostasis in vertebrates. It remains unclear how TRPM6 affects divalent cation transport and whether this involves functional homomeric TRPM6 plasma membrane channels or heteromeric channel assemblies with TRPM7. We show that homomeric TRPM6 is highly sensitive to intracellular free Mg2+ and therefore unlikely to be active at physiological levels of [Mg2+]i. Co-expression of TRPM7 and TRPM6 produces heteromeric TRPM7/M6 channels with altered pharmacology and sensitivity to intracellular Mg·ATP compared with homomeric TRPM7. Strikingly, the activity of heteromeric TRPM7/M6 channels is independent of intracellular Mg·ATP concentrations, essentially uncoupling channel activity from cellular energy status. Disruption of TRPM6 kinase phosphorylation activity re-introduces Mg·ATP sensitivity to the heteromeric channel similar to that of TRPM7. Thus, TRPM6 modulates the functionality of TRPM7, and the TRPM6 kinase plays a critical role in tuning the phenotype of the TRPM7·M6 channel complex.

The transient receptor potential melastatin member 7 (TRPM7) and member 6 (TRPM6) are divalent cation channel kinases essential for magnesium (Mg 2؉ ) homeostasis in vertebrates. It remains unclear how TRPM6 affects divalent cation transport and whether this involves functional homomeric TRPM6 plasma membrane channels or heteromeric channel assemblies with TRPM7. We show that homomeric TRPM6 is highly sensitive to intracellular free Mg 2؉ and therefore unlikely to be active at physiological levels of [Mg 2؉ ] i . Co-expression of TRPM7 and TRPM6 produces heteromeric TRPM7/M6 channels with altered pharmacology and sensitivity to intracellular Mg⅐ATP compared with homomeric TRPM7. Strikingly, the activity of heteromeric TRPM7/M6 channels is independent of intracellular Mg⅐ATP concentrations, essentially uncoupling channel activity from cellular energy status. Disruption of TRPM6 kinase phosphorylation activity re-introduces Mg⅐ATP sensitivity to the heteromeric channel similar to that of TRPM7. Thus, TRPM6 modulates the functionality of TRPM7, and the TRPM6 kinase plays a critical role in tuning the phenotype of the TRPM7⅐M6 channel complex.
TRPM7 and TRPM6 belong to the melastatin-related subfamily of TRP 5 channels and represent unique bifunctional proteins with both ion channel and protein kinase function (1)(2)(3)(4). The two proteins are highly homologous to each other with Ͼ50% sequence identity. TRPM7-and TRPM6-mediated membrane currents exhibit a highly nonlinear current-voltage (I/V) relationship with pronounced outward rectification (3,5). The small inward currents at physiologically relevant negative membrane voltages are carried by divalent cations, including Mg 2ϩ and Ca 2ϩ , and the large outward currents at positive voltages are due to efflux of monovalent cations (5,6).
Considerable efforts have been directed toward understanding the physiological functions of TRPM7 (7). TRPM7 represents the only known ion channel that is essential for cell viability (3,8,9). Its channel function is inhibited by both intracellular free Mg 2ϩ and Mg⅐ATP (3,10), and both factors serve as the major physiological mechanisms controlling TRPM7-mediated divalent cation transport. Information is more limited for TRPM6, and some of it remains controversial. Conflicting results have been reported concerning TRPM6 expression in heterologous expression systems. Two independent groups were able to measure homomeric TRPM6 currents in HEK-293 cells (5,(11)(12)(13), whereas the other two show that TRPM6 is retained intracellularly when expressed alone (14 -17). Hence, it has been proposed that TRPM6 may function in association with TRPM7 by forming a heteromeric TRPM7⅐M6 channel complex (14 -17). Channel kinase heteromerization results in a slightly altered permeation and pharmacology profile, pH sensitivity, and single channel conductance (11,12). However, important questions about regulation and functionality of heteromeric channel kinases remain unresolved.
RT-PCR-Total RNA was isolated from four replicates of HEK-293 wild-type cells using RNeasy mini kit (Qiagen). SuperScript III first-strand synthesis system for RT-PCR (Invitrogen) was used following the manufacturer's procedure to synthesize cDNA from 1 g of total RNA primed with oligo(dT) primers. Gene-specific primers for TRPM6 (forward primer 5Ј-TGCCCTGGAACAAGCAATGTCAG-3Ј and reverse primer 5Ј-CTTTTCATCAGCACAGCCCAAACC-3Ј), TRPM7 (forward primer 5Ј-AGCATACAGAACAGAGCCCAACGG-3Ј and reverse primer 5Ј-TTCCAACAGTGCCATCATCCACC-3Ј), and GAPDH (forward primer 5Ј-GGAGCCAAAAGGGT-CATCATCTC-3Ј and reverse primer 5Ј-AGTGGGTGTCGC-TGTTGAAGTC-3Ј) were designed using MacVector and synthesized by Invitrogen. PCR was performed in reaction volumes of 50 l containing 1 l of dNTPs (10 mM), 2 l of each primer (10 pmol/l), 2 l of cDNA solution, 5 l of reaction buffer (10ϫ), 37 l of water, and 1 l of Pfu Ultra II fusion HS DNA polymerase (Stratagene) on a Thermal Cycler (Bio-Rad). Denaturation was carried out at 94°C for 20 s, annealing at 55°C for 30 s, and elongation at 72°C for 30 s for 35 cycles, followed by extension at 72°C for 3 min. PCR products were detected in 0.8% agarose gel containing 1ϫ SYBR Safe DNA Gel Stain (Invitrogen).
To test the effects of osmotic challenges on the channels, hypertonic solution was made by adding appropriate quantities of mannitol to the standard external solution, and hypotonicity was achieved by reducing the concentration of NaCl from 140 to ϳ90 mM. The control solution for hypotonic solution was made by adjusting the osmolarity to 310 mosM with the appropriate concentration of mannitol. The osmolarities of the solutions were verified by an osmometer (Wescor). Standard internal solution contained (in mM) the following: 140 cesium glutamate, 8 NaCl, 10 Hepes-CsOH, 10 EGTA-CsOH; pH 7.2, adjusted with CsOH or HCl. In some cases, the strong Mg 2ϩ chelator (EDTA) was employed either to entirely remove internal Mg 2ϩ or to attain more accuracy in calculating internal free Mg 2ϩ concentrations. The free Mg 2ϩ levels in pipette solutions were calculated with WebMaxC Standard (Chris Patton, Stanford University). The detailed compositions of the pipette solutions for current recordings are shown in Tables 1-5. All chemicals were from Sigma.
Electrophysiology-Patch clamp experiments were performed under a tight seal whole-cell configuration at room temperature (20 -25°C). High resolution whole-cell currents were recorded by EPC9 (HEKA, Lambrecht, Germany) and Patchmaster version 2.4 (HEKA, Lambrecht, Germany). Unless otherwise stated, all voltages were corrected for a liquid junction potential of 10 mV. Patch pipettes pulled from borosilicate glass had resistances of 1.5-2.5 megohms when filled with internal solutions. Voltage ramps of 50-ms duration spanning a voltage range of Ϫ100 to ϩ100 mV were applied to the cells from a holding potential of 0 mV at a rate of 0.5 Hz following the establishment of whole-cell configuration. The currents were filtered at 2.9 kHz and digitized at 10 kHz. Currents were analyzed with FitMaster version 2.11 (HEKA, Lambrecht, Germany) and Igor Pro (Wavemetrics, Portland). Inward currents were extracted at Ϫ80 mV and outward currents at ϩ80 mV. Data were normalized to cell size as current density (pA/picofarad). All data are given as means Ϯ S.E.

Differential Pharmacological Modulation of Homomeric and
Heteromeric Channel Kinases-Homomeric TRPM6 and TRPM7 have essentially identical I/V characteristics (3,5), which makes it challenging to functionally differentiate between these two proteins. Interestingly, 2-aminoethoxydiphenyl borate (2-APB) inhibits TRPM7 and facilitates TRPM6 (11). This prompted us to  (17). The whole-cell patch clamp technique was used with solutions designed to maximally activate the channels by supplementing intracellular solutions with 10 mM EDTA and using Mg 2ϩ -free external solutions. Within 30 s after whole-cell break-in, both TRPM7 (Fig. 1E) and TRPM7/M6 (Fig. 1, C and D) expressing HEK-293 cells (HEK-TRPM7; HEK-TRPM7/M6) developed large currents with the characteristic outwardly rectifying I/V relationship typical for heterologously overexpressed channels. However, cells overexpressing TRPM6 alone had almost identical current sizes and kinetics of current development compared with uninduced cells (Fig. 1F), consistent with previous biochemical analyses on these cells showing that the majority of overexpressed TRPM6 localizes intracellularly (17). These results are also consistent with the observation made in previous attempts to overexpress TRPM6, including various alternative vectors (14,16,17). Transient transfection of both TRPM6 and TRPM7 using the pcDNA3.1 vector in HEK-293 cells produced plasma membrane insertion of TRPM6 only in the presence of TRPM7 but not with any other TRPM channel (M1, M2, M4, and M5) (14). Similarly, cloning of TRPM6 and TRPM7 into the pcDNA5/ TO-FLAG plasmid confirmed the inability to overexpress TRPM6 alone in the chicken DT40 and HEK-293 cell systems (17). Moreover, an alternative method using cRNA of TRPM6 injected in Xenopus laevis showed that TRPM6 and TRPM7 only overexpress in tandem (14).
We therefore resorted to transient overexpression of TRPM6 in wild-type HEK-293 cells using pCINeo-IRES-GFP-TRPM6 (5), which currently is the only expression vector that results in surface expression of monomeric TRPM6 channels, even if this may not be representative of physiological circumstances in native cells. In agreement with previous reports using this vector (5,11), cells analyzed 30 h post-transfection developed outwardly rectifying currents that had the typical I/V characteristics of TRPM6 (Fig. 1, A and B) and were at least 12 times larger than endogenously measured TRPM7-like currents ( Fig. 1F) (3). Although it remains unclear as to why this particular vector allows the formation of overexpressed TRPM6 channels in the plasma membrane, it enabled us to assess the effect of 2-APB on TRPM6, TRPM7, and TRPM7/M6 using these four heterologous cell systems (transient TRPM6 and inducible TRPM7, TRPM6, and TRPM7/M6). External application of 200 M 2-APB significantly enhanced TRPM6 currents (Fig. 1, A and B) (11). Conversely, TRPM7 currents were markedly blocked by 200 M 2-APB (Fig. 1E). When TRPM6 and TRPM7 were cooverexpressed upon tetracycline induction, 200 M 2-APB was without significant effect on TRPM7/M6 currents (Fig. 1, C and D). Tetracycline induction of HEK-TRPM6 cells did not lead to currents above control levels; however, application of 200 M 2-APB revealed that TRPM6 indeed can be inserted into the plasma membrane, as currents were facilitated rather than inhibited by this compound (Fig. 1, F and G). These data indicate that the composition of the heteromeric channel kinase protein determines its overall sensitivity to 2-APB, resulting in an intermediate response phenotype that averages the facilitatory and inhibitory effects of 2-APB on the monomeric channels.
2-APB is an unspecific pharmacological tool interfering with the activity of a variety of ion channels (11, 18 -20). In contrast, waixenicin A is a highly potent and specific inhibitor of TRPM7 that does not affect heterologously expressed TRPM6 (21). We tested the compound's efficacy to inhibit overexpressed TRPM7/M6 currents. Perfusing cells with the standard internal solution containing 780 M free Mg 2ϩ revealed that fully activated TRPM7/M6 currents inactivated by about 30% within 400 s ( Fig. 2A). Application of 10 M waixenicin A decreased TRPM7/M6 currents by 72% during that time, and this effect was also evident in the inward currents assessed at Ϫ80 mV ( Fig. 2A). Repeating the experiment in the absence of intracellular Mg 2ϩ did not affect inhibition of heteromeric channels by waixenicin A (Fig. 2B), leading to an overall 40% compoundinduced inhibition. This is in strong contrast to currents carried by homomeric TRPM7, where intracellular Mg 2ϩ significantly alters the inhibitory efficacy of waixenicin A (21). In addition,   The free Mg 2ϩ is ϳ20 M, and the total complexed Mg⅐ATP under these conditions amounts to 0, 0.5, and 1.4 mM, respectively. Mg⅐ATP Insensitivity of TRPM6/TRPM7 Heteromeric Ion Channels FEBRUARY 21, 2014 • VOLUME 289 • NUMBER 8 TRPM7 currents are completely suppressed within this application time and concentration of waixenicin A. Thus, although TRPM7/M6 heteromeric channels are sensitive to waixenicin A inhibition, they are less so than TRPM7 alone, and the compound's inhibitory effect is uncoupled from intracellular Mg 2ϩ dependence. The differential behavior of waixenicin A on TRPM7, TRPM6, and TRPM7/M6 currents may provide an additional pharmacological tool when characterizing TRPM7like currents in native cells. Sensitivity of TRPM6 and TRPM7/M6 Channels to Hypoand Hyper-osmolarity-TRPM7 is sensitive to osmotic challenges in human kidney cells (22). Because TRPM6 is predominantly expressed in the gut and kidney where osmotic changes are common, we tested whether TRPM6 would also be osmosensitive. Considering that intracellular Mg 2ϩ availability has significant impacts on channel functionality, we delivered the osmotic changes in the absence and presence of intracellular Mg 2ϩ . In the presence of 10 mM EDTA, TRPM6 currents were not affected by hypotonic solutions (210 mosM, Fig. 2C), whereas hypertonic solutions slightly reduced the currents by 30% (510 mosM, Fig. 2C). However, an increase in series resistance as a result of cell shrinkage was observed when the cells were exposed to hypertonic solution, which may account for this small decrease in TRPM6 currents. In the presence of 20 M Mg 2ϩ (Fig. 2D), neither hypotonicity (210 mosM) nor hypertonicity (510 mosM) had an effect on TRPM6 compared with corresponding controls (isotonic). Collectively, TRPM6 is largely insensitive to osmotic gradients, which is in marked contrast to TRPM7.

Cesium glutamate NaCl Cesium-Hepes Cesium-EGTA MgCl 2 Na⅐ATP
We next examined the responses of the TRPM7/M6 heteromer to osmotic challenges. In the absence of intracellular Mg 2ϩ , TRPM7/M6 behaved similarly to TRPM6 in response to osmotic changes in that 210 mosM had no effect on currents, and 510 mosM induced a small reduction of currents by about 30% (Fig. 2E). However, under conditions where internal free Mg 2ϩ was adjusted to more physiological levels of 790 M, TRPM7/M6 regained sensitivity to osmotic gradients in a dosedependent manner (Fig. 2F). The fitted dose-response curve yielded an IC 50 of 460 mosM for osmolarity-mediated suppression of TRPM7/M6 (Fig. 2G, Hill ϭ 3.3), similar to the IC 50 of 430 mosM for TRPM7 (22). However, TRPM7/M6 currents were sensitive to osmolarity over a wider range of osmotic gradients compared with TRPM7, as indicated by their distinctly different Hill coefficients (3.3 versus 11, respectively; Fig. 2G). This suggests that Mg 2ϩ is required for the osmosensitivity of TRPM7/M6 and that heteromeric complex formation results in a more graded response to osmotic change without changing the midpoint of osmotic sensitivity. As will be shown below, this is consistent with the preserved sensitivity of heteromeric channels to intracellular Mg 2ϩ , as a significant factor in mediating osmosensitivity is caused by molecular crowding of inhibitory molecules due to changes in cell volume (22).
Inhibition of Homomeric and Heteromeric Channel Kinases by Intracellular Mg 2ϩ -Intracellular Mg 2ϩ regulates both TRPM6 and TRPM7 (3,5,8,10). The IC 50 for Mg 2ϩ -dependent TRPM7 inhibition in the HEK-293 overexpression system is 720 M in the absence of Mg⅐ATP (10). However, a complete Mg 2ϩ dose-response curve has not yet been determined for native TRPM7 currents in wild-type HEK-293. RT-PCR established that wild-type HEK-293 cells express TRPM7 but did not detect significant TRPM6 (Fig. 3C), suggesting that endogenous conductances are mediated exclusively by TRPM7. As shown in Fig. 3, A and B, native TRPM7-like currents were inhibited by intracellular free Mg 2ϩ with an IC 50 of 569 M, resembling that of heterologously expressed TRPM7 (Fig. 3F) (10). As for TRPM6, Voets et al. (5) reported an IC 50 of 510 M for Mg 2ϩ block as determined by means of flash photolysis of caged Mg 2ϩ . We re-assessed the Mg 2ϩ sensitivity of TRPM6 using our standard whole-cell patch clamp approach to allow comparison between data sets obtained from TRPM7, TRPM6, and TRPM7/M6. We used 10 mM EDTA as the internal Mg 2ϩ chelator, and MgCl 2 was calculated to achieve the desired free Mg 2ϩ concentrations. Without internal Mg 2ϩ , transiently overexpressed TRPM6 currents assessed in HEK-293 cells rapidly activated within 30 s after break-in (Figs. 1A and 3D). Just 10 M Mg 2ϩ significantly suppressed the current size of TRPM6, and 300 M Mg 2ϩ almost abolished the currents (Fig.  3D). The fitted dose-response curve yielded an IC 50 of 29 M with a Hill coefficient of 0.7 (Fig. 3F), which is almost 25-fold more sensitive compared with TRPM7 (IC 50 ϭ 720 M) (10). We next investigated the Mg 2ϩ block of TRPM7/M6 heteromers in the tetracycline-inducible system (17), and we determined the Mg 2ϩ dose-response curve with intracellular free Mg 2ϩ buffered with EGTA to various concentrations. As shown in the Fig. 3E, intracellular Mg 2ϩ dose-dependently blocked TRPM7/M6 currents. The IC 50 fitted from the doseresponse curve for TRPM7/M6 approximated 466 M with a Hill coefficient of 0.8 (Fig. 3F). The differences in IC 50 values between native and overexpressed TRPM7 as well as TRPM7/ M6 did not amount to statistical significance, although there might be a trend to slightly higher sensitivity of heteromeric channels to Mg 2ϩ compared with monomeric TRPM7. However, TRPM7⅐M6 complexes are dramatically less sensitive to Mg 2ϩ (ϳ16-fold decrease) than TRPM6 homomers.
Previous work has demonstrated that the Mg 2ϩ sensitivity of monomeric TRPM7 channels is affected by the kinase domain, as kinase-dead point mutations reduce and kinase deletion mutants increase Mg 2ϩ sensitivity (8). The following two approaches were utilized to eliminate the function of TRPM6 kinase: mutation of a critical amino acid for the phosphotransferase activity (K1804R, TRPM6-K1804R) and removal of the entire kinase domain (TRPM6-⌬kinase) in the overexpressed proteins (17). We first determined the Mg 2ϩ sensitivity of both homomeric TRPM6-K1804R and TRPM6-⌬kinase   channels. Both of these TRPM6 mutants were strongly suppressed by Mg 2ϩ similar to TRPM6-WT (Fig. 4, A and B). Current normalization revealed that TRPM6-K1804R is slightly less sensitive to Mg 2ϩ , although TRPM6-⌬kinase is more sensitive compared with TRPM6-WT (Fig. 4C). This clarifies that TRPM6-WT and its kinase mutants all remain highly sensitive to Mg 2ϩ block. Next, we analyzed heteromeric combinations of these subunits with TRPM7 and constructed dose-response curves for TRPM7/M6-KR and TRPM7/M6-⌬kinase. As shown in Fig. 4, D and E, TRPM7/M6-KR currents and TRPM7/M6-⌬kinase currents were dose-dependently suppressed by intracellular Mg 2ϩ but with significantly lower sensitivity compared with homomeric assemblies of TRPM6 with an apparent IC 50 of 654 and 638 M, respectively (Fig. 4F). Thus, loss of function of the TRPM6 kinase rendered TRPM7/M6-KR slightly less sensitive to Mg 2ϩ block than TRPM7/M6-WT (IC 50 ϭ 466 M), with a phenotype reminiscent of TRPM7 (10). Even though the slight rightward shifts in sensitivity of the kinase-deficient mutants were not statistically significant, they appear to reverse the slight leftward trend observed in the TRPM7/M6-WT combination, despite their strong Mg 2ϩ sensitivity when expressed as homomers.

TRPM6 Kinase Domain Confers Insensitivity of TRPM7/M6
Heteromeric Channels to Intracellular Mg⅐ATP-Mg⅐ATP is an important regulator involved in setting TRPM7 activity that synergizes with free Mg 2ϩ to suppress TRPM7 (10). With intracellular free Mg 2ϩ at 780 M, the IC 50 for Mg⅐ATP inhibition of TRPM7 currents is around 2 mM. In contrast, modulation of TRPM6 by intracellular ATP reportedly occurs independent of Mg 2ϩ with an IC 50 of 1.3 mM and requires an intact kinase domain (13). Both TRPM7 and TRPM6 kinases need the presence of some Mg 2ϩ to maintain phosphotransferase activity (3,8,23,24). We therefore adopted an experimental strategy aimed at preserving the kinase activity by including low levels of free Mg 2ϩ in the pipette solution. Based on the obtained IC 50 value for Mg 2ϩ -mediated TRPM6 suppression, internal free Mg 2ϩ was set to 25 and 20 M for testing Mg⅐ATP and Na⅐ATP, respectively. Surprisingly, and in contrast to a previous study (13), we found that under these conditions Mg⅐ATP (3 or 9 mM) was completely ineffective to suppress transiently expressed TRPM6 currents (Fig. 5A). Not only that, but the currents remained stable after reaching a plateau, suggesting that ATP prevents current run down and thus may play a role in potentiating channel function. Similar to Mg⅐ATP, inclusion of Na⅐ATP (3 and 9 mM) in the patch pipette did not block TRPM6 currents but rather resulted in removal of current inactivation (Fig. 5B). Because activation and inactivation processes may compete with each other depending on the presence of ATP and Mg 2ϩ , a greater spread of maximal current amplitudes across individually patched cells can be observed in Fig. 5, A and B. We conclude that Na⅐ATP and Mg⅐ATP have similar effects on TRPM6; they do not inhibit TRPM6 homomeric channels and remove TRPM6 inactivation. We have no simple explanation for the conflicting observations made here compared with those of a previous study that found a Mg 2ϩ -independent inhibition of TRPM6 with an IC 50 of 1.3 mM (13), which would indicate an even higher ATP sensitivity than that of TRPM7. As will be shown below, we not only see the absence of ATP-dependent inhibition in homomeric TRPM6, but this property of TRPM6 is also conferred on heteromeric TRPM7/M6 channels.
We next asked how ATP might remove the inactivation of TRPM6 and considered the possible mechanisms of Mg⅐ATP on TRPM6 currents, as Mg⅐ATP is the physiological form of ATP in the cell. ATP-mediated removal of TRPM6 inactivation could potentially be caused by ATP hydrolysis through the kinase domain, be secondary to the generation of ATP metabolites ADP or AMP, or be due to the physical occupation of ATP-binding sites. Two nonhydrolysable structural analogs of ATP, ATP␥s (0.3 mM) and AMP-PNP (3 mM), failed to prevent TRPM6 inactivation (Fig. 5E), indicating that ATP binding alone is not responsible for ATP-mediated sustained opening of TRPM6. The intracellular administration of the ATP metabolites AMP (3 mM) or ADP (3 mM) were similarly ineffective (Fig.  5E). Taken together, this suggests that ATP removes TRPM6 inactivation by a mechanism that is dependent upon ATP hydrolysis.
To investigate whether the kinase domain is involved in ATP-mediated regulation of TRPM6, we tested the effects of Mg⅐ATP on kinase activity-deficient TRPM6 mutants transiently expressed in wild-type HEK-293 cells. After whole-cell break-in TRPM6-K1804R currents developed in a similar manner as TRPM6-WT currents followed by almost complete inactivation over the time course of the experiment (Fig. 5C). Unlike its effects in TRPM6-WT, internal Mg⅐ATP of 3 and 9 mM did not stabilize TRPM6 activity (Fig. 5C). Similar results were obtained when investigating the behavior of TRPM6-⌬kinase currents, although the overall current density was smaller in this mutant (Fig. 5D). Thus, it appears that a functional kinase domain with intact phosphotransferase activity is responsible for ATP hydrolysis and maintenance of TRPM6 function.
Mg⅐ATP regulation of heteromeric TRPM7/M6 was examined by using pipette solutions with higher Mg 2ϩ levels (753 M) around the IC 50 for free Mg 2ϩ . Surprisingly, even in the presence of intracellular Mg 2ϩ , TRPM7/M6 currents were not affected by perfusion of cells with physiological (4 mM) and even supra-physiological (8 mM) intracellular Mg⅐ATP (Fig. 5F). This indicates that the formation of heteromeric TRPM7/M6 uncouples channel kinases from cellular energy metabolism. Because the kinase domain of TRPM6 is able to cross-phosphorylate TRPM7 but not vice versa (17), we hypothesized that regulation of TRPM7 by the TRPM6 kinase might be responsible for the functional characteristics of TRPM7/M6 and that the TRPM6 kinase domain could be involved in setting the Mg⅐ATP sensitivity of the heteromeric channel complex. We tested this in kinase-deficient TRPM6 mutant constructs TRPM6-⌬kinase and TRPM6-K1804R (17). Fig. 5H demonstrates that in contrast to TRPM7/M6-WT, TRPM7/M6-K1804R currents were now inhibited by Mg⅐ATP in a dose-dependent fashion. Furthermore, removal of the TRPM6 kinase domain (TRPM6-⌬kinase) made heteromeric channel currents even more sensitive to intracellular Mg⅐ATP (Fig. 5H). This indicates that TRPM6 kinase-mediated cross-phosphorylation of TRPM7 is mandatory to render TRPM7/M6 heteromers insensitive to cellular ATP fluctuations.

DISCUSSION
We have found that homomeric TRPM6 channels and heteromeric TRPM7/M6 channels exhibit a number of functional and pharmacological differences and are regulated very differ-ently by free Mg 2ϩ and Mg⅐ATP in comparison with homomeric TRPM7 (see Table 6). Particularly, the regulation by Mg⅐ATP requires a functional TRPM6 kinase domain.
Robust TRPM6 currents could reliably be measured in HEK-293 cells transiently transfected with human TRPM6 in pCINeo-IRES-GFP vector. These findings are in line with other reports that used the same TRPM6 construct (5,13,26,27). In contrast to the transient expression system, we failed to detect markedly elevated TRPM6 channel activity in the tetracyclineinducible stable TRPM6 HEK-293. Here, a previous study demonstrated that without sufficient TRPM7 subunits available, TRPM6 is largely retained intracellularly (28). Similarly, TRPM6 overexpression using the expression vectors pcDNA3.1 or pINC(SP1)/Hygro did not result in TRPM6 plasma membrane translocation and neither did cRNA injection into Xenopus oocytes based on the pOGII vector (29,30). We currently cannot explain why the only system that seems to drive translocation of TRPM6 to the plasma membrane without the help of TRPM7 is through the pCINeo-IRES-GFP vector. It would be interesting to see whether the noncoding sequence of the TRPM6-pCINeo-IRES-GFP expression construct influences assembly and trafficking of the TRPM6 channel subunits.  Table 3 for composition of pipette solutions) or 20 M (B; see Table 4 for composition of pipette solutions) and was 753 M for heteromeric channels of TRPM7 plus TRPM6 (F-H; see Table 5 for composition of pipette solutions). A, HEK-293 cells transiently overexpressing TRPM6 were dialyzed with pipette solutions containing the indicated concentrations of Mg⅐ATP. Plots represent n ϭ 14, 11, and 10 cells for 0, 3, and 9 mM Mg⅐ATP, respectively. B, Na⅐ATP rather than Mg⅐ATP was tested in TRPM6 WT. Plots were averaged from 10, 10, and 9 cells for 0, 3, and 9 mM Mg⅐ATP, respectively. C, TRPM6 K1804R mutant was transiently overexpressed. Plots represent n ϭ 7, 7, and 9 cells for 0, 3, and 9 mM Mg⅐ATP, respectively. D, effects of Mg⅐ATP on the TRPM6 ⌬kinase mutant were tested in HEK-293 cells transiently overexpressing the mutant channel. Internal solutions had the same composition as in A and C. Plots represent n ϭ 7, 8, and 9 cells for 0, 3, and 9 mM Mg⅐ATP, respectively. E, normalized to current peak percent inactivation of transiently expressed TRPM6 WT in HEK-293. Cells were perfused with either no addition of Mg⅐ATP (control, n ϭ 14, see A), 3 mM (n ϭ 11, A), or 9 mM Mg⅐ATP (n ϭ 10, A) and 3 mM AMP-PNP (n ϭ 13), 3 mM ADP (n ϭ 9), 3 mM AMP (n ϭ 12), or 300 M ATP␥S (n ϭ 9). Despite these conflicting findings, the highly efficient membrane expression of TRPM6 using the pCINeo-IRES-GFP vector for transient TRPM6 expression provided an opportunity to examine the modulation of homomeric TRPM6 by various physiological regulators, such as intracellular Mg 2ϩ and Mg⅐ ATP. We examined the Mg 2ϩ sensitivity of homomeric TRPM6 by perfusing cells with defined buffered intracellular Mg 2ϩ solutions and obtained an IC 50 of ϳ30 M. This is at variance with the IC 50 of 510 M reported previously (5). However, the latter IC 50 for Mg 2ϩ might be due to the different experimental approach using flash photolysis of Mg 2ϩ . Because we here found large differences in Mg 2ϩ sensitivity between TRPM7 and TRPM6 under identical experimental conditions, we are confident that the higher Mg 2ϩ sensitivity of TRPM6 is a genuine feature of the channel. Physiological levels of intracellular Mg 2ϩ are thought to range between 0.5 and 1 mM. This would suggest that homomeric TRPM6 channels, even if expressed natively in epithelial cells, might play a rather limited role as a physiologically functional ion channel.
In contrast to the strong Mg 2ϩ sensitivity of TRPM6, Mg⅐ATP did not suppress TRPM6 currents at all and rather prevented channel inactivation. This is in contrast to previous work reporting that both Na⅐ATP and Mg⅐ATP dose-dependently blocked TRPM6 currents in EDTA-buffered, Mg 2ϩ -free intracellular pipette solutions (13). This discrepancy might be due to absence of intracellular Mg 2ϩ in the latter study, as Mg 2ϩ not only acts as an inhibitory modulator of channel kinases, but also a minimal amount of Mg 2ϩ seems to be required for proper function of TRPM6 and TRPM7. For example, early evidence uncovered the synergistic regulation of TRPM7 channel activity by magnesium and Mg⅐ATP, where a threshold magnesium concentration of at least 200 M is needed before inhibition of TRPM7 by ATP is gained (31). Furthermore, intracellular Mg 2ϩ dictates the inhibitory effectiveness of halides as well as waixenicin A in both TRPM6 and TRPM7 (32,33), and free Mg 2ϩ is required for the phosphotransferase activity of the TRPM7 channel kinase (23). Hence, complete removal of Mg 2ϩ , or even Zn 2ϩ bound to the kinase domain by a strong chelator such as EDTA, might impair channel function and/or regulation.
Our data show that TRPM6 inactivation was completely removed by ATP, implying that ATP provides a positive regulatory mechanism to maintain TRPM6 function. ATP seems to remove TRPM6 inactivation through kinase-dependent ATP hydrolysis, as kinase activity-deficient TRPM6 mutants inactivated independently of the presence or absence of intracellular ATP. Autophosphorylation of TRPM6 by its kinase domain could be one mechanism to prevent current inactivation. Thus, washout of intracellular ATP during whole-cell patch clamp measurements would result in channel dephosphorylation and lead to channel inactivation, whereas the supply of ATP would enable the TRPM6 channel kinase to hydrolyze ATP and maintain the phosphorylation on TRPM6 channels for sustained channel activity. Indeed, massive autophosphorylation has been detected in the serine/threonine-rich region of TRPM6 following ATP stimulation (34). However, further experiments are required to fully understand the mechanisms underlying ATP-mediated effects, for instance, identification of the specific amino acid targets that are autophosphorylated by the kinase domain.
Unlike other TRP channels that form heteromultimers with their close homologs, including TRPC1/4/5 (35), TRPC3/6/7 (35), TRPV5/V6 (36), and TRP1/3 (37), TRPM7 and TRPM6 possess highly homologous Ser/Thr kinase domains at their C termini. It has previously been shown that the TRPM6 kinase is able to cross-phosphorylate TRPM7 but not vice versa (28). Consistent with this observation, we found that disruption of TRPM6 kinase activity attenuated Mg 2ϩ -mediated suppression and restored sensitivity to Mg⅐ATP inhibition in the channel complex of TRPM7 with kinase-defective TRPM6 mutants. Thus, the phenotype of this kinase mutant is reminiscent of homomeric TRPM7, suggesting that TRPM6 shapes the phenotype of TRPM7/M6 via cross-phosphorylation of TRPM7.
To date, the molecular mechanisms governing the assembly of TRPM7 and TRPM6 remain unknown. Based on the structure of TRPM proteins, TRPM7/M6 most likely functions as a heterotetrameric channel complex. Indeed, emerging evidence suggests that both TRPM7 and TRPM6 subunits contribute to the heteromeric channel pore (27,29,30), although the exact stoichiometric architecture of TRPM7/M6 remains to be determined (38).
A model where TRPM6 co-assembles with TRPM7 to form a functional Mg 2ϩ -influx channels is attractive due to the overlapping expression profile of TRPM7 and TRPM6 (24,28,29,40). TRPM7 is a ubiquitously expressed ion channel, whereas the tissue expression of TRPM6 is much more restricted, predominantly in the intestine and kidney. A growing number of examined cell lines and tissues have been shown to co-express TRPM7 and TRPM6 (25,29,39). Our electrophysiological results from the functional point of view support a refined model and provide insights into characteristics and physiological regulation of the TRPM7⅐M6 channel complex. First, the IC 50 value of 466 M for Mg 2ϩ suppression suggests that TRPM7/M6 may be functional within the range of physiological levels of Mg 2ϩ . Second, the complexed channel responds to osmotic challenges with similar sensitivity as TRPM7, but in a more graded fashion, therefore preserving its proposed Mg 2ϩ transport efficiency in tissues where large osmotic changes are common. Third, like TRPM6, the heteromeric channel complex is not regulated by cellular ATP, thus leaving channel kinase-mediated divalent ion influx unaffected by cellular energy metabolism. In comparison with homomeric TRPM7, whose activity is strongly down-regulated by physiological levels of intracellular Mg⅐ATP, heteromeric TRPM7/M6 may be more efficient in regard to the transport of Mg 2ϩ and other divalent ions. These features indicate that the TRPM7/M6  (21) Partial block No effect (21) Mg⅐ATP Insensitivity of TRPM6/TRPM7 Heteromeric Ion Channels FEBRUARY 21, 2014 • VOLUME 289 • NUMBER 8 channel may be more efficient to function as a divalent ion influx channel than either TRPM7 or TRPM6 alone. This may in turn affect cellular functions both physiologically and pathophysiologically, e.g. in the case of tumor cell growth.