The kinase activity of the channel-kinase protein TRPM7 regulates stability and localization of the TRPM7 channel in polarized epithelial cells

The channel-kinase transient receptor potential melastatin 7 (TRPM7) is a bifunctional protein with ion channel and kinase domains. The kinase activity of TRPM7 has been linked to the regulation of a broad range of cellular activities, but little is understood as to how the channel itself is regulated by its own kinase activity. Here, using several mammalian cell lines expressing WT TRPM7 or kinase-inactive variants, we discovered that compared with the cells expressing WT TRPM7, cells in which TRPM7's kinase activity was inactivated had faster degradation, elevated ubiquitination, and increased intracellular retention of the channel. Mutational analysis of TRPM7 autophosphorylation sites further revealed a role for Ser-1360 of TRPM7 as a key residue mediating both TRPM7 stability and intracellular trafficking. Additional trafficking roles were uncovered for Ser-1403 and Ser-1567, whose phosphorylation by TRPM7's kinase activity mediated the interaction of the channel with the signaling protein 14-3-3θ. In summary, our results point to a critical role for TRPM7's kinase activity in regulating proteasome-mediated turnover of the TRPM7 channel and controlling its cellular localization in polarized epithelial cells. Overall, these findings improve our understanding of the significance of TRPM7's kinase activity for functional regulation of its channel activity.

The channel-kinase transient receptor potential melastatin 7 (TRPM7) is a bifunctional protein with ion channel and kinase domains. The kinase activity of TRPM7 has been linked to the regulation of a broad range of cellular activities, but little is understood as to how the channel itself is regulated by its own kinase activity. Here, using several mammalian cell lines expressing WT TRPM7 or kinase-inactive variants, we discovered that compared with the cells expressing WT TRPM7, cells in which TRPM7's kinase activity was inactivated had faster degradation, elevated ubiquitination, and increased intracellular retention of the channel. Mutational analysis of TRPM7 autophosphorylation sites further revealed a role for Ser-1360 of TRPM7 as a key residue mediating both TRPM7 stability and intracellular trafficking. Additional trafficking roles were uncovered for Ser-1403 and Ser-1567, whose phosphorylation by TRPM7's kinase activity mediated the interaction of the channel with the signaling protein 14-3-3. In summary, our results point to a critical role for TRPM7's kinase activity in regulating proteasome-mediated turnover of the TRPM7 channel and controlling its cellular localization in polarized epithelial cells. Overall, these findings improve our understanding of the significance of TRPM7's kinase activity for functional regulation of its channel activity.
In zebrafish, additional roles for the channel have been uncovered for the proliferation of pigment cells and pancreatic epithelial cells and for the differentiation and function of sensory neurons and kidney (9 -14). At the cellular level, TRPM7 is implicated in cellular Mg 2ϩ homeostasis, cell survival, proliferation, as well as cell adhesion and motility (15)(16)(17)(18)(19)(20)(21). Dysfunction of TRPM7 channel activity has been linked to defects in platelet biogenesis in humans (22). At the physiological level, TRPM7 is linked to the regulation of vertebrate magnesium homeostasis primarily through its physical association with its homolog TRPM6. Trpm6 is found mutated in human hypomagnesemia with secondary hypocalcemia disease (23)(24)(25). TRPM7's impact on whole-body magnesium homeostasis was later confirmed in mice (26). Several studies have further linked the TRPM7 channel to pathological processes such as cell death during organ ischemia, tumor proliferation, and metastasis (16,27).
A unique aspect of both TRPM7 and TRPM6 is the presence of a functional ␣-kinase domain on the channels' C termini. TRPM7's kinase is not required for channel gating but is involved in the modulation of the channel by intracellular Mg 2ϩ -ATP (28). Autophosphorylation of TRPM7 occurs at multiple serine and threonine residues across the protein, many of which are clustered in a serine/threonine (S/T)-rich domain proximal to the kinase domain (29 -31). Phosphorylation of the S/T-rich domain has been proposed to facilitate substrate binding for TRPM7 kinase (29). Several in vitro substrates, including nonmuscle myosin heavy chain IIA, annexin I, phospholipase C␥2, histones, and elongation factor 2 kinase, have been identified (32)(33)(34)(35)(36). The kinase retains its functionality in vivo when separated from the channel by caspase-dependent proteolytic cleavage (37). Liberated TRPM7 kinase has been reported to localize to the nucleus (37,38), where it contributes to chromatin remodeling (37). Studies using TRPM7 kinase-inactive mutant mice have linked the kinase activity of TRPM7 to the sensing and the coordination of both cellular and systemic responses to magnesium deprivation in mice (39). More recent studies have suggested a requirement for TRPM7's kinase in the regulation of the murine mast cell degranulation, intraepithelial T cell gut homing, ameloblast differentiation, and storeoperated Ca 2ϩ entry in platelets (40 -43). To fully understand the wide range of cellular and physiological activities that TRPM7's kinase influences, it is essential to understand the mechanisms that regulate its kinase activity and the functional significance of autophosphorylation of the channel.
We recently employed MS to map TRPM7 autophosphorylation sites (30), which led us to propose a novel mechanism by which the kinase activity of TRPM7 can be regulated by autophosphorylation. Here we report that inactivation of TRPM7 kinase affects proteasome-mediated turnover of the channel kinase, its cellular localization in polarized epithelial cells, and its interaction with the phospho-binding protein 14-3-3, opening up new avenues for regulation of the channel kinase.

TRPM7 kinase activity affects channel turnover
Although numerous studies have explored the consequences of kinase inactivation or kinase removal on TRPM7 channel activity, the impact of the kinase on the stability of the channel, its cellular localization, and its interaction with other cellular proteins remain comparatively unexplored. In our previous investigations of TRPM7, we observed that protein expression of the channel is affected by mutations that inactivate the catalytic activity of its kinase (20,30). Indeed, transient expression of the kinase-inactive TRPM7-K1646R mutant in HEK-293T cells has lower protein expression levels than the WT protein (Fig. 1A). To elucidate the role of TRPM7 kinase activity in regulating channel expression, we employed tetracyclineinducible HEK-293 cells expressing WT HA-tagged TRPM7 (293-TRPM7-WT) or a kinase-inactive mutant (293-TRPM7-K1646A) to follow the kinetics of TRPM7 protein turnover (20). Following 24 h of tetracycline induction, the cell medium was replaced with tetracycline-free medium containing cycloheximide (CHX) to inhibit protein synthesis. TRPM7 protein levels were then analyzed at various time points over the following 24 h. The protein levels of TRPM7-WT were significantly more stable than that of the TRPM7-K1646A mutant, with protein levels of TRPM7-WT not appreciably decreasing over 24 h of CHX treatment (Fig. 1B). By contrast, the protein levels of TRPM7-K1646A rapidly declined over time, with an apparent half-life of 8.1 h (Fig. 1B), indicating that the kinase-inactive mutant is subject to increased protein turnover.
To assess potential mechanisms controlling TRPM7 protein turnover, we first examined whether TRPM7 is ubiquitinated, because our MS analysis from our previous study identified several lysine residues on mouse TRPM7 (Lys-198, Lys-481, Lys-707, Lys-1019, and Lys-1253) that were modified with ubiquitin (30). We attempted to detect ubiquitinated TRPM7 proteins by co-expressing TRPM7 with ubiquitin in mammalian cells. However, we were unsuccessful in resolving ubiquitinated TRPM7 by SDS-PAGE and Western blotting using an anti-ubiquitin antibody. As an alternative approach, we used an anti-FLAG-agarose to immunoprecipitate ubiquitinated proteins and then probed the immunopurified proteins with an anti-TRPM7 antibody. Only the kinase-inactive TRPM7-K1646A mutant, but not TRPM7-WT, co-immunoprecipitated with FLAG-ubiquitin (Fig. 1C). Interestingly, the immunoprecipitated TRPM7-K1646A proteins appeared as a single band migrating at 220 kDa. This single band could indicate that TRPM7 was co-immunoprecipitated with other ubiquitinated proteins or that TRPM7 is modified by mono-or multiubiquitination. Nevertheless, the observation that only TRPM7-K1646A, but not TRPM7-WT, is strongly associated with ubiquitination, gave evidence that the turnover of the kinase-inactive TRPM7 mutant is mediated by the ubiquitin-proteasome pathway.
To further test whether TRPM7 protein expression is regulated by the proteasome, TRPM7-expressing 293-TRPM7-WT and 293-TRPM7-K1646A cells were induced with tetracycline for 24 h and then treated with the proteasome inhibitor MG132 for 24 h in the tetracycline-free medium. Proteasome inhibition by MG132 had little effect on TRPM7-WT levels (0.98 Ϯ 0.18fold change) but caused significant protein accumulation (2.22 Ϯ 0.72-fold) of TRPM7-K1646A (Fig. 1D). By further testing two different kinase-inactive TRPM7 mutants, we confirmed that increased proteasome-mediated TRPM7 turnover is independent of the mechanism by which TRPM7 kinase is rendered catalytically inactive. TRPM7-WT and two kinaseinactive mutants, TRPM7-K1646R and TRPM7-G1618D, were transiently expressed in HEK-293T cells for 24 h and treated with MG132 for 24 h. Both kinase-inactive mutants had a significant amount of protein accumulation (TRPM7-K1646R, 3.19 Ϯ 0.73-fold and TRPM7-G1618D, 2.84 Ϯ 0.75fold) compared with TRPM7-WT (1.01 Ϯ 0.26-fold) (Fig. 1E). These results collectively support the previous finding that TRPM7-WT is a very long-lived protein, whereas the kinase-inactive mutant is more unstable and subject to protein turnover by the ubiquitin-proteasome pathway. These experimental findings also point to a role for TRPM7 kinase and autophosphorylation in controlling the protein stability and turnover of the channel.

TRPM7 kinase affects cellular localization of the channel in polarized epithelial cells
Our results thus far suggest a role for TRPM7's kinase in controlling protein levels. We next asked whether the kinase influences the cellular localization of the channel. TRPM7 mRNA is widely expressed in adult mice, with the highest expression in the kidney where the channel contributes to Mg 2ϩ reabsorption (1,2,23). Protein expression of the native channel, however, was too low for us to detect its cellular localization by immunocytochemistry. Instead, we analyzed the cellular distribution of murine TRPM7 heterologously expressed in opossum kidney (OK) cells, a model proximal tubule epithelial cell line (44). Using confocal microscopy, we found that TRPM7-WT readily localized to the basal-lateral side of OK cells ( Fig. 2A). A certain amount of TRPM7 was also located on the apical membrane, albeit at much lower levels, where it colocalized with the apical and microvilli marker NHERF1 (Fig. 2,  A and B).
In contrast, TRPM7-K1646R poorly localized to basallateral or apical membranes in OK cells and was instead retained intracellularly (Fig. 3A). The percentage of OK cells displaying peripheral membrane TRPM7 localization for the kinase-inactive mutant (18.12 Ϯ 7.00%) was significantly lower than that for TRPM7-WT (71.44 Ϯ 6.98%) (Fig. 3A), pointing to a role for autophosphorylation in controlling the cellular TRPM7 kinase regulates channel stability and localization localization of the channel. Using the proteasome inhibition assay, we also confirmed that the kinase-inactive TRPM7 in OK cells was more prone to proteasome-mediated degradation compared with the TRPM7-WT (Fig. S1). We further validated TRPM7 cellular localization pattern using another well-established kidney tubule cell line, Madin-Darby canine kidney (MDCK) cells (45). Kinase inactivation of TRPM7 similarly interfered with the channel's localization to the cell border when transiently expressed in MDCK cells: only 5.16 Ϯ 1.92% of TRPM7-K1646R-expressing cells exhibited localization of the channel to cell border compared with 46.95 Ϯ 4.80% of cells expressing TRPM7-WT (Fig. 3B). The discovery that the TRPM7-K1646R fails to efficiently localize to the cell periphery in both OK and MDCK cells provides compelling evidence that TRPM7 kinase activity plays a critical role for localization of the channel in polarized epithelial cells. Because TRPM7's cellular localization is influenced by its kinase activity, we were also motivated to examine the cellular localization of TRPM7's homolog TRPM6. When expressed alone, TRPM6 does not localize to the cell border in OK or MDCK cells. Instead, we found compared with the WT channel. GFP and actin served as transfection and loading controls, respectively. B, HEK-293 cells stably expressing HA-TRPM7-WT and HA-TRPM7-K1646A were induced with tetracycline for 24 h and replaced with fresh medium containing 10 g/ml CHX to arrest protein synthesis. Cell lysates were harvested at various time points after CHX treatment. Equal amounts of total proteins were analyzed by SDS-PAGE and Western blotting. The relative TRPM7 levels were calculated by normalizing TRPM7 protein levels against the vinculin control in each sample. Time courses for TRPM7-WT and TRPM7-K1646A protein turnover are shown on the right (n ϭ 3, means Ϯ S.D.). C, HEK-293 cells stably expressing HA-TRPM7-WT and HA-TRPM7-K1646A were induced with tetracycline (Tet) and co-transfected with FLAG-ubiquitin (FLAG-Ub) for 24 h. Cell lysates were immunoprecipitated with anti-FLAG-agarose. TRPM7 and ubiquitinated proteins in FLAG-IP and lysates were probed by an anti-TRPM7 and anti-FLAG antibody, respectively. D, HEK-293 cells stably expressing HA-TRPM7-WT or TRPM7-K1646A were induced with tetracycline for 24 h and then replaced with fresh medium containing DMSO or MG132 (1 M) for 24 h. Equal amounts of cell lysates were resolved by SDS-PAGE and Western blotting. The fold of TRPM7 protein level change was calculated by dividing relative TRPM7 protein levels (TRPM7/vinculin) in each MG132-treated sample against DMSO-treated controls. Replicates (n ϭ 4) from two independent experiments were analyzed (means Ϯ S.D.). *, p ϭ 0.0158. E, HEK-293T cells transiently expressing HA-TRPM7-WT, HA-TRPM7-K1646R, and HA-TRPM7-G1618D were treated with DMSO or MG132 (1 M) for 24 h. Equal amounts of cell lysates were resolved by SDS-PAGE and Western blotting. Protein quantification was performed as described above. Replicates (n ϭ 4 -6) from three independent experiments were analyzed (means Ϯ S.D.). ****, p ϭ 0.0001; ***, p ϭ 0.0008.

TRPM7 kinase regulates channel stability and localization
that TRPM6 was primarily located in intracellular compartments ( Fig. S2), similar to what was previously reported for TRPM6 expressed in HEK-293 cells (23,46).

TRPM7-S1360 controls proteasome-mediated TRPM7 turnover and localization in epithelial cells
Our previous analysis of TRPM7 autophosphorylation uncovered many sites that are frequently phosphorylated on TRPM7 ( Fig. 4A) (30). Because kinase inactivation resulted in a more rapid TRPM7 turnover, we next investigated whether any of the phosphorylation sites we had previously identified are involved in mediating proteasome-dependent degradation of the channel. We mutated phosphorylated TRPM7 residues Ser-1255, Ser-1360, Ser-1403, Ser-1502, Thr-1503, and Ser-1567 to alanine and tested whether these modifications affected the channel's sensitivity to proteasome-mediated degradation (Fig.  4B). In response to MG132 treatment, the protein level of TRPM7-S1360A significantly increased 2.10 Ϯ 0.67-fold, whereas the protein levels of other TRPM7 alanine substitution mutants were unaffected (Fig. 4C).
Previously, Ser-1360 was identified as one of the most phosphorylated residues on TRPM7 when the channel is constitu-tively expressed in HEK-293 cells at low levels (Table S3 in Ref.  30). To test whether phosphorylation of Ser-1360 interferes with the proteasome-mediated turnover of TRPM7, we introduced phosphomimetic substitutions at Ser-1360 (TRPM7-S1360D and TRPM7-S1360E). Both TRPM7-S1360D and TRPM7-S1360E mutants were significantly more resistant to proteasome-mediated degradation than the TRPM7-S1360A mutant (Fig. 4, D and E). Interestingly, our previous MS analysis also detected phosphorylation of Ser-1360 on the TRPM7-K1646R kinase-inactive mutant, suggesting that Ser-1360 can be phosphorylated by other cellular kinases. We also investigated whether phosphomimetic substitutions of Ser-1360 could prevent the kinase-inactive forms of TRPM7 from being targeted by proteasome-mediated degradation but found them unable to do so, indicating that other regulatory phosphorylation sites may also be involved in controlling TRPM7 protein stability (Fig. S3). These results suggest that Ser-1360 phosphorylation is an essential post-translational signal for the control of proteasome-mediated turnover of the channel.
Because Ser-1360 phosphorylation affects TRPM7 protein stability, we were motivated to test whether Ser-1360 also affects the cellular localization of the channel (Fig. 4F). When expressed in OK cells, TRPM7-S1360A did not as efficiently localize to the cell border (20.25 Ϯ 5.91%) compared with the WT channel (61.43 Ϯ 4.34%) (Fig. 4G). By contrast, phosphomimetic substitutions of TRPM7-S1360D (51.80 Ϯ 8.70%) and TRPM7-S1360E (50.97 Ϯ 6.33%) were similarly efficient in localizing to the cell periphery as TRPM7-WT (Fig. 4G). Comparable findings were also observed for WT TRPM7 and TRPM7-S1360A/D/E mutants expressed in MDCK cells (Fig.  S4). These results suggest a model in which that phosphorylation of TRPM7-S1360 allows the channel to evade proteasome-mediated degradation to allow for effective post-translational trafficking of the channel.

Identification of the phospho-binding protein 14-3-3 as a binding partner of TRPM7
To gain additional insight into the mechanisms by which TRPM7's kinase affects the stability and localization of the channel, we conducted a yeast two-hybrid screen of a rat brain cDNA library by using the C terminus of the mouse TRPM7 (amino acid 1120 -1863) as the bait (47). One of the positive clones identified by the screen was 14-3-3. 14-3-3 belongs to a family of highly conserved and ubiquitously expressed proteins involved in many cell signaling pathways (48). Seven 14-3-3 isoforms are encoded in mammalian genomes, and they exist as homo-and heterodimers with each monomer capable of binding specifically to distinctive phospho-S/T residues on target proteins (49,50). Binding of 14-3-3 proteins to their cognate protein ligands has been shown to regulate protein enzymatic activity, elicit protein conformational changes, influence the cellular localization proteins, and mediate cross-bridging with other proteins (51).

Binding of 14-3-3 to TRPM7 requires autophosphorylation at S1403
Most of the 14-3-3 binding partners bear variations of consensus recognition motifs RSX(pS/T)XP or RX(Y/F)X(pS/T)XP containing a phospho-S/T residue (53). We screened for potential 14-3-3 binding sites on TRPM7 by creating GFP-TRPM7-Cterm mutants bearing alanine substitutions at highly conserved serine residues within potential 14-3-3 binding motifs (Ser-1403, Ser-1567, and Ser-1588) (Fig. 6A) and performed GST-pulldown purification assays with the mutants. The interaction between GFP-TRPM7-Cterm and GST-14-3-3 was completely abolished in the S1403A mutant and partially reduced in S1567A and S1588A mutants (Fig. 6B). Our recent study of TRPM7 in vivo autophosphorylation found that both Ser-1403 and Ser-1567 are autophosphorylation sites in vivo, whereas Ser-1588 is not (30), thus eliminating Ser-1588 as a 14-3-3 binding site. There are many examples where dimerized 14-3-3 proteins bind to their targets using a stronger and a weaker binding site (53). Our data point to TRPM7 Ser-1403 as a strong binding site for 14-3-3 and Ser-1567 as a

TRPM7 kinase regulates channel stability and localization
mutants' inability to interact with 14-3-3, phosphomimetic substitutions at these sites would be predicted not to rescue their intracellular localization, because we previously showed that phosphomimetic substitutions do not reconstitute the phosphorylation-dependent interaction of GFP-TRPM7-Cterm proteins with 14-3-3 (Fig. 6C). On the other hand, if the localization defects of TRPM7-S1403A and TRPM7-S1567A were due to a lack of phosphorylation at these sites in a 14-3-3-binding-independent manner, then the phosphomimetic substitutions at the corresponding residues would be predicted Figure 6. Binding of 14-3-3 to TRPM7 requires autophosphorylation of the channel at Ser-1403. A, three highly conserved potential 14-3-3 binding motifs (underlining) were identified on TRPM7's C terminus. Two of the motifs contain TRPM7 autophosphorylation sites (red). B, GFP-TRPM7-Cterm-WT, S1403A, S1567A, and S1588A were expressed in HEK-293T cells and subjected to a GST-pulldown assay using GSH agarose bound with 20 g of GST-14-3-3 or GST proteins. GFP-TRPM7-Cterm was examined by Western blotting using an anti-GFP antibody. GST proteins were visualized by SDS-PAGE and Coomassie staining. Ser-1403 was identified as the major 14-3-3 binding site on GFP-TRPM7-Cterm. C, GFP-TRPM7-Cterm-WT and Ser-1403 mutants (S1403A, S1403D, and S1403E) expressed in HEK-293T cells were subjected to the same GST-pulldown assay. GST-14-3-3 failed to pulldown GFP-TRPM7-Cterm carrying mutations at Ser-1403, demonstrating a requirement of Ser-1403 phosphorylation for TRPM7 and 14-3-3 interaction.

TRPM7 kinase regulates channel stability and localization
to increase the channel's localization to the cell periphery. Indeed, we found that both TRPM7-S1403D (41.40 Ϯ 3.60%) and TRPM7-S1403E (50.26 Ϯ 8.35%) exhibited similarly reduced peripheral localization in OK cells as compared with the alanine mutant TRPM7-S1403A (45.09 Ϯ 2.738%) (Fig. 7). The inability of the Ser-1403 phosphomimetic substitutions to stimulate the channel's localization to the cell border correlates with the inability of the phosphomimetic substitutions to increase the in vitro interaction of the channel with 14-3-3 (Fig. 6C). In MDCK cells, similar localization patterns of the TRPM7-S1403 mutants were also observed (Fig. S9), supporting our hypothesis of the role of 14-3-3 in regulating TRPM7 cellular localization. We attempted to disrupt in vivo binding between 14-3-3 and TRPM7 by overexpressing difopein, a protein that scavenges 14-3-3 proteins in preventing them from binding to targets (59). However, overexpression of difopein caused excessive cellular death, which precluded analysis of difopein's effect on TRPM7 localization. Interestingly, we found that phosphomimetic substitution of glutamate at TRPM7-S1567 had a higher level of peripheral localization (53.28 Ϯ 8.93%) compared with the alanine mutant (35.89 Ϯ 6.28%). Oddly, the introduction of aspartate at TRPM7-S1567 produced a severe trafficking defect (1.5 Ϯ 1.52%), which is likely nonspecific (Fig. 7). Nevertheless, the ability of TRPM7-S1567E to localize to the cell border with an efficiency similar to TRPM7-WT suggests that phosphorylation of TRPM7-S1567 contributes to the regulation of TRPM7 trafficking through both 14-3-3-dependent and independent mechanisms. Likewise, we also observed that the TRPM7-S1502A/ T1503A mutant, which does not strongly interact with 14-3-3, had a significant reduction in peripheral membrane localization (51.23 Ϯ 3.8%) compared with the WT protein (63.82 Ϯ 6.17) (Fig.  S6D). These data suggest that multiple phosphorylation sites contribute to the trafficking of TRPM7 in vivo and that the 14-3-3 binding site TRPM7-S1403 plays a vital role in regulating phosphorylation-dependent post-translational trafficking of TRPM7. Collectively these results underscore an important role of TRPM7 kinase activity in regulating the cellular trafficking of the channel.

Discussion
The results of our investigation point to a significant role for TRPM7's kinase in regulating proteasome-mediated turnover of the channel and controlling its cellular localization in polarized epithelial cells. We observed that TRPM7-kinase inactivation led to faster protein turnover and a higher level of protein ubiquitination. Inhibition of the proteasome with MG132 raised the expression levels of kinase-inactive TRPM7 mutants but not the WT protein. These findings suggest that the kinase activity of TRPM7 mediates the post-translational processing of the channel by mediating protein turnover via the ubiquitinproteasome degradation pathway. We speculate that the ER-associated degradation pathway is at play in mediating the turnover of the TRPM7 kinase-inactive mutants. Our previous MS analysis of TRPM7 uncovered several ubiquitination sites on TRPM7: Lys-198 and Lys-481 for TRPM7-WT and Lys-481, Lys-707, Lys-1019, and Lys-1253 on TRPM7-K1646R (30). In particular, residue Lys-1019, which was exclusively identified on the kinase-inactive TRPM7, is located on the extracellular loop between the membrane-spanning S5 and S6 domains of the channel. Because the ubiquitin-proteasome system is located outside of the ER, ubiquitination of TRPM7-K1019 on the extracellular/ER luminal side is likely achieved after the protein is retrotranslocated into the cytoplasm, an essential step of ER-associated degradation processing for targeted substrates (60). We speculate that critical TRPM7 phosphorylation events occur in the ER during channel assembly. During the synthesis of oligomeric membrane proteins, monomers are first translated and then assembled into the protein's final oligomeric form (61). It was previously shown that TRPM7 kinase remains catalytically inactive until assembled into dimers (62). Dimer formation occurs through domain swapping between the "exchange segment" of one kinase monomer and the catalytic domain of the other monomer (62). Together, these data suggest at least one model for control of TRPM7 kinase activity. When first translated in the ER as a monomer, TRPM7's kinase is predicted to be catalytic inactive. Upon channel assembly, dimer formation occurs, and the TRPM7 kinase becomes active and capable of autophosphorylation. How TRPM7's kinase is regulated after channel assembly remains unknown.
In our previous study, MS analysis identified Ser-1360 as the most frequently phosphorylated residue on constitutively expressed TRPM7 (30). Our experiments point to phosphorylation of Ser-1360 as critical for controlling protein stability and cellular localization of the channel. The TRPM7-S1360A mutant phenotypically resembled the kinase-inactive TRPM7-K1646R mutant in the proteasome inhibition assay and its epithelial cell localization. Phosphomimetic substitutions of Ser-1360 effectively reversed the defects of the alanine mutants, stabilized TRPM7 protein levels, and allowed for efficient trafficking of the channel to the cell periphery. How phosphorylation of Ser-1360 influences channel turnover is not apparent. One possibility is that TRPM7-S1360 phosphorylation may interfere with the ubiquitination of TRPM7 by the preventing the channel from binding to proteasome-related proteins such as E3 ubiquitin ligases. Interestingly, our previous MS analysis detected Ser-1360 phosphorylation on the overexpressed TRPM7 kinase-inactive mutant, indicating that Ser-1360 is not necessarily just an autophosphorylation site but could also be targeted by other kinases. In probing for the functional significance of the TRPM7 kinase, TRPM7 kinase-inactive knockin mice have been generated (39,63). Studies utilizing TRPM7 kinase-inactive knockin mice have revealed diverse phenotypes (39 -43). In light of our discovery that TRPM7 stability and cellular localization are affected by the same kinase-inactivating mutation, extra care must be taken in teasing apart the mechanistic origins of phenotypes attributed to TRPM7's kinase. In this genetic background, it is also possible that other kinases, including TRPM6, may substitute for the loss of TRPM7 kinase catalytic activity.
When heterologously expressed in OK cells, TRPM7 is efficiently trafficked to basal-lateral membranes and less efficiently to the apical membrane, where ion exchange occurs between cells and the tubular fluid (44). Using a second kidney epithelial cell line, MDCK cells, we found a similar pattern of TRPM7 cellular localization. Our experiments establish polarized epithelial kidney cell lines as a useful system for investigating the mechanisms involved in regulating the cellular localiza-TRPM7 kinase regulates channel stability and localization tion of the channel. In both OK and MDCK cells, we found that mutations disrupting TRPM7 kinase catalytic activity led to intracellular retention of the channel, suggesting a role for TRPM7 autophosphorylation in regulating cellular localization of the channel. Also, we discovered that 14-3-3, a member of a large phospho-binding signal protein family, interacts with TRPM7 in a TRPM7 kinase-dependent manner. Our biochemical analysis identified Ser-1403 and Ser-1567 as a strong and a weak 14-3-3 binding site on TRPM7. Having more than one 14-3-3 binding site potentially allows regulation of the 14-3-3-binding partners by two separate phosphorylation events (64,65), but whether such a mechanism occurs for TRPM7 remains to be determined. Our study did not find evidence that 14-3-3 binding to TRPM7 affects protein stability or kinase activity but instead suggested a role for 14-3-3 in the regulation of the channel's cellular localization. Indeed, 14-3-3 proteins have been found to mediate both forward tracking to and endocytic recycling of membrane proteins from the plasma membrane (66,67). Thus, it is reasonable to speculate that the autophosphorylation-dependent interaction of 14-3-3 with TRPM7 affects the intracellular trafficking of the channel. Consistent with this hypothesis, in both OK and MDCK cells we found that alanine mutation at the primary 14-3-3 binding site on TRPM7 (TRPM7-S1403A) significantly interfered with localization of the channel to the cell border. Phosphomimetic substitutions of Ser-1403 (TRPM7-S1403D and -S1403E) did not alleviate defects in peripheral localization of the channel, which is in agreement with our biochemical assays showing that phosphomimetic substitutions at Ser-1403 are not effective in promoting the interaction between TRPM7 and 14-3-3. These results suggest that it is the disruption of phosphorylation-dependent 14-3-3 interaction, not a lack of phosphorylation activity at this residue per se, that contributes to the defect of localization of the TRPM7-S1403 mutant channels. More work is required to evaluate the hypothesis that autophosphorylation-induced 14-3-3 binding is part of the mechanism by which the kinase activity of TRPM7 regulates the trafficking and localization of the channel. However, it is worth noting that it has been reported that several 14-3-3 isoforms (14-3-3, 14-3-3⑀, and 14-3-3␤) are enriched in TRPM7-containing intracellular vesicles, further implicating 14-3-3 proteins in the regulation TRPM7 localization via vesicle trafficking (68).
Our investigation also revealed that there might be 14-3-3independent mechanisms at play in controlling TRPM7's cellular localization. For example, TRPM7 exhibited peripheral localization defects when Ser-1567 was mutated to either alanine or aspartate. However, when Ser-1567 was replaced with a phosphomimetic glutamate substitution, the channel localized to the cell border as efficiently as the WT channel. Ser-1567 itself is a frequently phosphorylated residue of TRPM7, as has been identified by our previous MS analysis (30), and therefore could mediate TRPM7 localization through both 14-3-3-dependent and independent means. We also found that mutation of TRPM7-S1502/T1503, which does not strongly bind to 14-3-3, also disrupted TRPM7 localization in the OK cells. Thus, additional work is needed to determine how the Ser-1502/Thr-1503 and Ser-1567 TRPM7 phosphorylation sites affect the trafficking of the channel in different cell types and under varied physiological and pathological conditions. TRPM7-mediated ion conductance appears to be involved in a broad array of cellular and developmental processes. Similarly, evidence points to multifaceted roles for TRPM7's kinase. Our results thus far hint that each of the identified autophosphorylation sites of TRPM7 represents a unique regulatory mechanism to control the activity of the channel kinase in response to diverse cellular signals. The discovery that the kinase activity of TRPM7 functionally regulates the channel protein stability and cellular localization provides us with our first glimpse as to why nature endowed an ion channel its own kinase function. The intrinsic kinase activity of TRPM7 constitutes a mechanism to allow the channel to respond nimbly to different cellular demands and cellular contexts. Also, the effect of kinase-inactivation on TRPM7 protein stability and localization may be cell type-specific, because of the differences in intracellular signaling pathways unique to each cell type. Still not understood is how the kinase itself is regulated in vivo and when and under what conditions phosphorylation of the critical residues we identified occur. A key goal going forward will be to determine signaling pathways that control TRPM7 kinase, which will be needed to fully elaborate the function and regulation of this unique bifunctional ion channel.

Cell culture and transfection
HEK-293T cells are from ATCC (CRL-3216 TM ). 293-TRPM7-WT and 293-TRPM7-K1646A cells have been described previously (20). HEK-293 cells were cultured under standard conditions in Dulbecco's modified Eagle's medium (Thermo Fisher) supplemented with 10% fetal bovine serum (FBS) (Atlanta Biologicals). For biochemical analysis, transient transfections were performed using the Turbofect transfection reagent (Thermo Fisher) according to the manufacturer's protocol. The OK cell line was a generous gift of Dr. Judith A. Cole (Department of Biological Sciences, University of Memphis), cultured in Dulbecco's modified Eagle's medium/F-12 medium supplemented with 5% FBS. The MDCK cell line was a generous gift of Dr. Kenneth Irvine (Waksman Institute of Microbiology, Rutgers University) and maintained in minimal essential media (Thermo Fisher) supplemented with 5% FBS. Transfections of OK cells and MDCK cells were performed with Lipofectamine 3000 (Thermo Fisher) according to the manufacturer's protocol.

Protein stability assay
Approximately 0.6 ϫ 10 6 293-TRPM7-WT and 293-TRPM7-K1646A cells were seeded onto 6-well plates coated with poly-L-lysine (Sigma-Aldrich). Two hours after seeding, the cells were treated with 5 g/ml tetracycline to induce protein expression for 24 h. To arrest protein synthesis, the cells were replaced with fresh medium containing 10 g/ml CHX. At various time points following CHX treatment, the cells were lysed in lysis buffer containing 50 mM Tris (pH 7.4), 150 mM NaCl, 1% IGEPAL CA-630 (Sigma-Aldrich), and the protease inhibitor mixture (Roche Life Sciences). The protein concentrations of cell lysates were measured using the Bradford assay (Thermo Fisher). Equal amounts of protein (40 -80 g/well) were loaded for each sample for SDS-PAGE and Western blotting. The rabbit polyclonal TRPM7 antibody (anti-TRPM7) was used to probe for HA-TRPM7 (20). Vinculin, which was used as loading control, was detected by Western blotting using a monoclonal anti-vinculin antibody (clone hVIN-1; Sigma-Aldrich). Vinculin was chosen as a loading control because it is a very stable protein and also because unlike other focal adhesion proteins (e.g. talin), vinculin's protein expression levels are not effected by overexpression of TRPM7 (20). The protein concentration in the cell lysates were quantified and normalized so that equal amounts of proteins were loaded for analysis by SDS-PAGE and Western blotting. To quantify protein levels, immunochemiluminescence signals of the Western blots with unsaturated bands were detected by X-ray film exposure using horseradish peroxidase secondary antibodies. Intensities of the protein bands on the scanned films were measured using the LI-COR Image Studio software (LI-COR Biotechnology). The relative levels of TRPM7 in each sample were calculated by normalizing TRPM7 protein intensity against vinculin, which was used as a loading control. To calculate protein half-lives, the relative protein levels of TRPM7-WT and TRPM7-K1646A were fitted to a one-phase exponential decay using GraphPad Prism 7 software (GraphPad Software, Inc.).

Detection of ubiquitination-associated TRPM7
Approximately 3 ϫ 10 6 293-TRPM7-WT and 293-TRPM7-K1646A cells were seeded onto 10-cm culture plates. After attachment overnight, the cells were treated with 5 g/ml tetracycline to induce TRPM7 expression and transfected with 5 g of FLAG-tagged ubiquitin (pcDNA6 -FLAG-Ub). After 24 h of protein expression, the cells were then treated with 20 M of the proteasome inhibitor MG132 (Sigma-Aldrich) for 8 h and lysed in 1 ml of lysis buffer containing 10 mM of deubiquitinase inhibitor N-ethylmaleimide (Sigma-Aldrich). To detect ubiquitination-associated TRPM7, ubiquitinated proteins from cell lysates were immunoprecipitated using anti-FLAG M2 agarose affinity gel (Sigma-Aldrich), washed three times with lysis buffer, and eluted to 50 l of SDS sample buffer. Cell lysates and immunoprecipitated samples were resolved by SDS-PAGE and Western blotting following standard protocols. TRPM7 was detected by the rabbit anti-TRPM7 antibody, and ubiquitinated proteins were recognized by an anti-FLAG antibody (Sigma-Aldrich).

Proteasome inhibition assay with MG132
Approximately 0.7 ϫ 10 6 293-TRPM7-WT and 293-TRPM7-K1646A cells were seeded on 6-well plates coated with Poly-L-lysine and induced with g/ml tetracycline 2 h after seeding. After 24 h of expression, cell culture medium was replaced with fresh medium containing DMSO or 1 M MG132 to inhibit proteasome activity. The cells were harvested 24 h after MG132 treatments in 200 l of lysis buffer containing protease inhibitor mixture (Roche Life Sciences). For transient transfections, HEK-293T cells (0.3 ϫ 10 6 cells/well) or OK cells (0.2 ϫ 10 6 cells/well) were seeded onto 12-well plates with poly-L-lysine coating. The cells were transfected with HA-TRPM7-WT and mutants (0.5 g for HEK-293T cells and 0.8 g for OK cells) the next day at 70% confluence. After 24 h of expression, cell culture medium was replaced with fresh medium containing DMSO or 1 M MG132. After MG132 treatment (24 h for HEK-293T cells and 9 h for OK cells), the cells were lysed in 100 l of lysis buffer containing protease inhibitor mixture (Roche Life Sciences). Protein concentrations of the cell lysates were measured using Bradford assays (Thermo Fisher). Relative TRPM7 protein levels in each sample were calculated by normalizing TRPM7 protein levels against vinculin. The fold of protein level change induced by MG132 treatment was measured by dividing the relative TRPM7 protein levels in the MG132-treated group with that in DMSO-treated controls. Statistical analysis (Student's t test) was performed using GraphPad Prism 7 software (GraphPad Software, Inc.).

Immunocytochemistry
OK or MDCK cells (0.1 ϫ 10 6 cells) were seeded on glass coverslips in 24-well plates and transfected the next day at 70% confluence. To confirm TRPM7 apical localization in OK cells, 1 g of HA-TRPM7-WT and 0.5 g of FLAG-NHERF1 were co-expressed in OK cells for 24 h. To compare the cellular localization between TRPM7-WT and mutant HA-TRPM7 (0.5 g for OK cells and 0.8 g for MDCK cells), the proteins were allowed to express for 48 h. For immunochemical staining, the cells were fixed with 4% paraformaldehyde at room tempera-TRPM7 kinase regulates channel stability and localization ture for 30 min, permeabilized in PBS with 0.1% Triton X-100 for 20 min, and blocked with 5% FBS//PBS at 30°C for 1 h. A rabbit monoclonal anti-HA antibody (C29F4; Cell Signaling Technology) was used to detect the HA-tagged TRPM7, and a mouse anti-FLAG antibody (Sigma-Aldrich) was used to identify FLAG-NHERF1. Alexa Fluor TM 488 and Alexa Fluor TM 568 goat antibodies raised against rat and mouse were used as secondary antibodies (Thermo Fisher). F-actin was detected with Alexa Fluor TM 568 phalloidin (Thermo Fisher). Images were obtained at the Rutgers RWJMS CORE confocal facility using a Yokogawa CSUX1-5000 microscope under 63ϫ magnification using 488-and 561-nm excitation wavelengths. To quantify levels of TRPM7 peripheral localization, the cells were examined under an inverted fluorescent microscope (40ϫ magnification for OK cells and 100ϫ magnification for MDCK cells). On each slide, two to three nonoverlapping sections each containing 100 -150 TRPM7 positive-staining cells were counted. The level of peripheral localization was measured as the percentage of cells exhibiting TRPM7 localization to cell periphery against the total number of TRPM7-expressing cells. Each data point represents one section of the slides. Statistical analysis (Student's t test) was performed using GraphPad Prism 7 software (GraphPad Software, Inc.).

GST-pulldown assay
GST-14-3-3 fusion proteins were expressed in transformed BL21-DE3 cells (Stratagene, CA). Bacterial cells were lysed by sonication in ice-cold PBS containing 1% Triton X-100 and protease inhibitor phenylmethylsulfonyl fluoride (Sigma-Aldrich). The bacterial cell lysates were then incubated with GSH agarose (Sigma-Aldrich) overnight at 4°C with rotation. The agarose beads were washed with PBS with 1% Triton X-100. Concentrations of the GST proteins were measured by Coomassie stain on the SDS-PAGE gel using serial diluted BSA proteins as controls. To expressed TRPM7 proteins, 10 g of HA-TRPM7 or GFP-TRPM7-Cterm plasmids were transiently transfected into HEK-293T cells plated in a 10-cm dish. After 24 h, the cells were lysed in 1 ml of mild lysis buffer containing protease inhibitor mixture (Roche Life Sciences) and phosphatase inhibitor mixture (EMD Millipore), and spun down at 14,000 ϫ g for 10 min at 4°C. For GST-pulldown assays, the cell lysate supernatants were incubated with GSHagarose bound with 20 g of GST-14-3-3 or GST proteins overnight at 4°C with rotation. The bound proteins were washed with 1 ml of PBS with 0.1% Triton X-100 three times by rotation and eluted into 50 l of SDS sample buffer. Lysates input (20 l) and pulldown samples were separated on SDS-PAGE gels and analyzed by immunoblotting. The rat monoclonal anti-HA (3F10) and a mouse monoclonal anti-GFP antibody (sc-9996; Santa Cruz Biotechnology, CA) were used to probe HA-TRPM7 and GFP-TRPM7-Cterm proteins, respectively. GST-14-3-3 and GST proteins were visualized by SDS-PAGE and Coomassie staining. For experiments where protein dephosphorylation was conducted, a -phosphatase (New England Biolabs) was added to the lysis buffer in the absence of phosphatase inhibitor.

Immunoprecipitation
Approximately 1.0 ϫ 10 6 HEK-293T cells were seeded onto 60-mm dishes. 2.5 g of HA-TRPM7-WT or HA-TRPM7-K1646R plasmids were co-transfected with 1 g of FLAG-14-3-3 the next day. After 48 h of expression, cell lysates were collected in 1 ml of mild lysis buffer containing protease inhibitor mixture (Roche Life Sciences) and phosphatase inhibitor mixture (EMD Millipore), immunoprecipitated by 20 l of HA-agarose (Sigma-Aldrich) overnight at 4°C. Immunoprecipitated (IP) samples were washed three times with lysis buffer and eluted into 50 l of SDS sample buffer. Lysate input (20 l) and IP samples were separated on SDS-PAGE gels and analyzed by immunoblotting. The rabbit polyclonal anti-TRPM7 antibody was used to detect HA-TRPM7, and a mouse anti-FLAG antibody was used to detect FLAG-14-3-3. To detect potential 14-3-3 binding motifs on TRPM7, HA-TRPM7 was immunopurified from HEK-293T cells by HA-agarose (Sigma-Aldrich) and resolved by SDS-PAGE. The in vivo 14-3-3 binding motifs were detected with a mouse monoclonal phosphoserine-14-3-3 binding Motif (4E2) antibody (9606; Cell Signaling Technology). The total amounts of TRPM7 proteins in the IP samples were detected by a rat anti-HA antibody (3F10; Roche Life Science).

In vitro kinase assay
HA-TRPM7 and GFP-TRPM7-Cterm proteins were transiently expressed in HEK-293T cells and immunoprecipitated using HA-agarose (Sigma-Aldrich) and streptavidin-agarose (to target the streptavidin-binding protein epitope in the GFP tag) (Stratagene). In vitro kinase assays were performed in a kinase buffer containing 50 mM MOPS (pH 7.2), 100 mM NaCl, 2.5 mM MnCl 2 , and 0.5 mM ATP. The kinase reactions were performed at 30°C for 20 min in the presence of 4 Ci of [␥-32 P]ATP with 5 g of myelin basic protein as a substrate. Where indicated, 5 g of GSH-eluted GST or GST-14-3-3 proteins were added into the kinase reaction. The reactions were stopped by the addition of SDS sample buffer, and proteins in the reaction mix were resolved by SDS-PAGE. Proteins were detected by Coomassie Blue staining, and the gels were dried entirely using a gel dryer (Bio-Rad). Incorporation of [␥-32 P]ATP into substrates was analyzed by autoradiography using Cyclone Plus phosphorimaging device (PerkinElmer Life Sciences).