Importance of the renal ion channel TRPM6 in the circadian secretion of renin to raise blood pressure

Blood pressure has a daily pattern, with higher values in the active period. Its elevation at the onset of the active period substantially increases the risk of fatal cardiovascular events. Renin secretion stimulated by renal sympathetic neurons is considered essential to this process; however, its regulatory mechanism remains largely unknown. Here, we show the importance of transient receptor potential melastatin-related 6 (TRPM6), a Mg2+-permeable cation channel, in augmenting renin secretion in the active period. TRPM6 expression is significantly reduced in the distal convoluted tubule of hypotensive Cnnm2-deficient mice. We generate kidney-specific Trpm6-deficient mice and observe a decrease in blood pressure and a disappearance of its circadian variation. Consistently, renin secretion is not augmented in the active period. Furthermore, renin secretion after pharmacological activation of β-adrenoreceptor, the target of neuronal stimulation, is abrogated, and the receptor expression is decreased in renin-secreting cells. These results indicate crucial roles of TRPM6 in the circadian regulation of blood pressure.

A pproximately one billion people worldwide are estimated to have hypertension, which significantly increases the risk of various diseases, including potentially fatal diseases, such as ischemic heart disease and stroke. These hypertension-related fatal incidents are known to occur frequently in the early morning, the beginning of the active period when blood pressure rises sharply 1 . Renin secretion triggered by renal sympathetic neurons is considered to play a crucial role in this blood pressure elevation, but its regulatory mechanism is not well understood 2 .
It is widely accepted that some dietary minerals, particularly sodium and potassium, play important roles in the control of blood pressure. Magnesium is a major essential element involved in a variety of biological activities. Epidemiological studies have shown a significant inverse relationship between dietary magnesium levels and the risk of hypertension [3][4][5][6] . Moreover, urinary magnesium excretion, which is assumed to be in approximate equilibrium with intestinal magnesium absorption, is inversely correlated with the risk of hypertension 7 . These findings suggest a close association between organismal magnesium homeostasis and blood pressure regulation, highlighting the importance of Mg 2+ channels and/or transporters in the regulation of blood pressure.
A large quantity of magnesium is constantly being reabsorbed in the kidney. Most magnesium in the glomerular filtrate is reabsorbed in the thick ascending limb of Henle's loop, but the final step of reabsorption occurs at the distal convoluted tubule (DCT) 8 . The latter process is tightly regulated to adjust the amount of reabsorption to a level appropriate for maintaining magnesium homeostasis. Genomic analyses of congenital diseases with symptoms of hypomagnesemia have revealed several key molecules involved in this process. Trpm6, encoding transient receptor potential melastatin-related 6 (TRPM6), is mutated in patients with recessive hypomagnesemia with secondary hypocalcemia 9,10 . TRPM6 forms a Mg 2+ -permeable ion channel localized to the apical membrane of DCT cells and mediates Mg 2+ intake from the tubular lumen 11 . Another key molecule is cyclin M2 (CNNM2), a Mg 2+ transporter localized to the basolateral membrane of DCT cells 12,13 . Cnnm2 mutations cause familial dominant hypomagnesemia, which is characterized by defects in magnesium reabsorption in the kidney 12 . Cnnm2-deficient mice show similar symptoms of hypomagnesemia with magnesium wasting in the kidney 14 . The precise molecular role of CNNM2 is still controversial, but several lines of evidence suggest that it mediates Mg 2+ efflux from DCT cells through the basolateral membrane 15 , thereby contributing to magnesium reabsorption. Further analyses of Cnnm2-deficient mice have revealed that these mice also have significantly lower blood pressure than control mice 14 . However, the mechanistic details and the involvement of other Mg 2+ channels/transporters in this blood pressure regulation remain unknown.
The renal DCT is also known to play critical roles in blood pressure regulation, because it significantly reabsorbs sodium and regulates body fluid volume 16 . A number of cation channels/ transporters, such as the Na-Cl co-transporter (NCC), epithelial sodium channel (ENaC), and Na + /H + exchanger 2 (NHE2), are expressed at the DCT apical membrane and some of these proteins mediate Na + intake into DCT cells from the tubular lumen. Intracellular Na + is then extruded from the cells through the basolateral membrane by Na + /K + -ATPase to accomplish its reabsorption. Apart from a direct involvement in regulating body fluid volume, the DCT is located adjacent to the macula densa, which controls secretion of renin, a hormone with the principal role in orchestrating blood pressure regulation 17,18 . By sensing urinary Cl − levels, macula densa cells stimulate nearby juxtaglomerular (JG) cells to secrete renin.
In this work, we perform transcriptome analyses with kidneys of hypotensive Cnnm2-deficient mice. We find a significant decrease of Trpm6, prompting us to generate kidney-specific Trpm6-deficient mice. Unexpectedly, analyses of this mouse strain reveal a phenotype qualitatively different from that of Cnnm2-deficient: the loss of circadian variation of blood pressure. We also find that this phenotype is caused by impaired secretion of renin, a blood pressure-raising hormone of which blood level is normally elevated during the active phase.

Results
Downregulation of TRPM6 expression in Cnnm2-deficient mice. To explore the molecular mechanism of blood pressure reduction in Cnnm2-deficient mice, we performed DNA microarray analyses to investigate gene expression changes in the kidney of Cnnm2 fl/fl ;Six2-Cre mice, lacking both Cnnm2 alleles in the kidney 14 . Among the differentially expressed genes (Table 1), we chose to analyze Pvalb and Trpm6, both of which are expressed in the DCT 9,19 . Quantitative PCR (qPCR) analyses revealed that Trpm6 expression was similarly reduced in the kidneys of both Cnnm2 fl/fl ;Six2-Cre and Cnnm2 +/Δ mice (Supplementary Fig. 1a). Since both Cnnm2 +/Δ and Cnnm2 fl/fl ;Six2-Cre mice were hypotensive 14 , we analyzed TRPM6 expression in more detail. Immunoblotting analyses of kidney lysates confirmed a significant decrease of TRPM6 at the protein level by Cnnm2deficiency, while the levels of NCC and phosphorylated NCC (active form) were not markedly affected ( Supplementary Fig. 1b, CNNM2 expression was confirmed by successive immunoprecipitation and immunoblotting analyses). The decrease of TRPM6 in the kidneys of Cnnm2 fl/fl ;Six2-Cre mice was also confirmed by immunofluorescence analyses (Supplementary Fig. 1c).
We also examined TRPM6 expression in mice lacking CNNM4, which also show symptoms of hypomagnesemia 20 but have elevated blood pressure 14 . Immunoblotting and immunofluorescence analyses indicated that TRPM6 expression is augmented in DCT cells of Cnnm4-deficient mice (Supplementary Fig. 1d, e), suggesting that TRPM6 expression is closely associated with blood pressure values, but it is not directly linked to blood magnesium levels.
Regulation of TRPM6 expression by intracellular Mg 2+ in DCT cells. To characterize the molecular mechanism of TRPM6 downregulation in Cnnm2-deficient mice, we used a DCT-derived cell line MDCT 21 , which expresses dominantly Cnnm2 and moderately Cnnm3 and Cnnm4 among Cnnm family genes (Cnnm1-Cnnm4, Supplementary Fig. 1f). As CNNM2 and CNNM4 have strong Mg 2+ efflux activity while CNNM3 has no Mg 2+ efflux activity 15 , we performed small-interfering RNA (siRNA)-mediated knockdown of either Cnnm2, Cnnm4, or both. Cnnm2 knockdown augmented intracellular Mg 2+ levels ([Mg 2+ ] i ), and simultaneous knockdown of Cnnm4 led to slight elevation of average [Mg 2+ ] i value, which was not statistically significant ( Supplementary Fig. 1g). We next examined Trpm6 expression by qPCR and found a significant decrease after Cnnm2 knockdown and a much larger decrease after double knockdown of Cnnm2 and Cnnm4. Therefore, Trpm6 expression appears to be negatively regulated by intracellular Mg 2+ levels, which is consistent with previous reports 22,23 . To directly examine the importance of Mg 2+ , we cultured MDCT cells in media with various concentrations of Mg 2+ . As shown in Supplementary Fig. 1h, decreasing the extracellular Mg 2+ concentration led to the reduction in intracellular Mg 2+ level and the increase of Trpm6 expression, suggesting a suppressive role of Mg 2+ . In the case of double knockdown (Cnnm2/Cnnm4) MDCT cells, Trpm6 expression was suppressed with decreasing Mg 2+ concentrations down to 0.2 mM, but further reductions canceled the suppressive effect; both intracellular Mg 2+ and Trpm6 levels were not significantly different from control siRNA-transfected cells in both 0.1 and 0.02 mM conditions. These results indicate a Mg 2+ -dependent regulatory mechanism of Trpm6 expression, which explains the Trpm6 downregulation in the kidneys of Cnnm2-deficient mice.
Conditional disruption of Trpm6 in the kidney. Next, we aimed to generate Trpm6-deficient mice to further investigate the importance TRPM6 in the regulation of blood pressure. We used an embryonic stem (ES) cell clone carrying a recombinant Trpm6 allele (Trpm6 − ) that contains a splice acceptor sequence, forcing premature mRNA splicing and resulting in an mRNA truncated after exon 6 ( Supplementary Fig. 2a). Chimeric heterozygous mice were obtained by blastocyst injection of the ES cells and recombination was confirmed by PCR ( Supplementary Fig. 2b). We intercrossed these Trpm6 +/− mice, but no Trpm6 −/− offspring were obtained ( Supplementary Fig. 2c). This result is consistent with previous reports showing embryonic lethality in Trpm6-deficient mice 24,25 .
Therefore, we performed a kidney-specific conditional disruption of Trpm6 using Six2-Cre, which enables efficient and specific expression of GFP-fused Cre recombinase in the embryonic kidney (ref. 26 and Supplementary Fig. 3). Trpm6 fl/fl ;Six2-Cre mice (kidney-specific Trpm6-deficient mice) were born at the expected Mendelian ratio ( Supplementary Fig. 2c). To determine the expression of TRPM6, we performed immunoblotting analyses using an anti-TRPM6 antibody. TRPM6 protein levels were greatly reduced in the kidneys of Trpm6 fl/fl ;Six2-Cre mice, but no significant changes were observed in other organs (Fig. 1a). Immunofluorescence staining analyses also showed no expression of TRPM6 in NCC-positive DCT cells (Fig. 1b). We observed no significant TRPM6-positive signal in more than 100 such clusters of NCC-positive cells, suggesting that TRPM6 expression is almost completely suppressed in DCT cells of Trpm6 fl/fl ;Six2-Cre mice. These results confirm the successful generation of kidneyspecific Trpm6-deficient mice.
We first examined the effect of kidney-specific Trpm6deficiency on magnesium homeostasis. We maintained mice in metabolic cages and collected urine and serum samples, which were then subjected to colorimetric quantitation of magnesium levels using Xylidyl Blue-I. Serum magnesium levels were significantly lower in Trpm6 fl/fl ;Six2-Cre mice compared to those in control Trpm6 +/+ ;Six2-Cre mice, whereas urinary magnesium levels were higher (Fig. 1c). We also performed inductively coupled plasma-emission spectroscopy (ICP-ES) analyses to measure several major metal elements in the serum and found that only magnesium levels were significantly affected in Trpm6 fl/ fl ;Six2-Cre mice (Fig. 1d). These results clearly indicate that kidney-specific Trpm6-deficient mice have hypomagnesemia with renal magnesium wasting, as observed in kidney-specific Cnnm2deficient mice 14 .
Lack of circadian variation of blood pressure and renin activity. Next, we subjected Trpm6 fl/fl ;Six2-Cre mice to blood pressure measurements using radiotelemetry. Systolic and diastolic blood pressure values were both significantly lower in Trpm6 fl/fl ;Six2-Cre mice than control mice ( Fig. 2a and Supplementary Fig. 4). It is noteworthy that the usual circadian variation of blood pressure, with high values during the dark period and low values during the light period in mice, was not observed in Trpm6 fl/fl ;Six2-Cre mice. To examine whether the circadian rhythm proceeds normally, we used radiotelemetry to measure the locomotive activity of mice and confirmed a very clear circadian variation in Trpm6 fl/fl ;Six2-Cre mice (Fig. 2b). We also examined blood levels of vasopressin (AVP) and mRNA levels of two clock genes Bmal1 and Per2 in the brain and kidney; all are known to show typical circadian variation 27,28 . We analyzed at six different time points (at 0, 4, 8 Signature genes (p < 0.1; two-tailed Student's t-test (paired)) with mean fold-changes > +2.0 (more than twice) or < − 2.0 (less than half) are listed.
zeitgeber times (ZT) for the light period and at 12, 16, 20 ZT for the dark period). Control mice show clear circadian patterns of their expression as previously reported 28 , and again, we found no significant abnormalities in Trpm6 fl/fl ;Six2-Cre mice ( Fig. 2c and Supplementary Fig. 5). We also checked the expression level of Trpm7, which encodes an Mg 2+ -permeable channel closely related to TRPM6, in the kidney. The expression patterns were very similar in both Trpm6 +/+ ;Six2-Cre and Trpm6 fl/fl ;Six2-Cre mice, and did not show clear circadian patterns ( Supplementary  Fig. 5). Collectively, we conclude that the circadian rhythm proceeds normally in Trpm6 fl/fl ;Six2-Cre mice. It has been considered that blood pressure elevation during the active period (the dark period for mice and the light period for humans) in both humans and mice occurs in response to an increase in renin secretion into the blood 2 . We measured renin activity in blood samples collected at the six different time points described above. A clear increase in blood renin activity was observed during the dark periods in samples from control mice, but no significant increase was observed in samples from those of Trpm6 fl/fl ;Six2-Cre mice (Fig. 2d). We also measured the blood level of noradrenaline, the release of which is known to be stimulated by the activation of renin-angiotensin system 29,30 . As expected, their circadian variation also disappeared in Trpm6 fl/fl ;Six2-Cre mice (Fig. 2e). Meanwhile, the protein level of blood angiotensinogen (AGT), another rate limiting molecule for activating renin-angiotensin system in mice 31 , was unaffected by Trpm6 deficiency ( Supplementary  Fig. 6). The results suggest that the impaired circadian variation of renin is important for the loss of circadian variation of blood pressure by Trpm6-deficiency.

Proximity of TRPM6-expressing cells to renin-secreting cells.
Renin is expressed in and secreted from JG cells in the kidney 17,18 . JG cells are in direct contact with the macula densa, which is located adjacent to the DCT and which stimulates renin secretion from JG cells in response to the reduction of urinary Cl − levels 18 . Immunofluorescence staining for renin and neuronal nitric oxide synthase (nNOS), a macula densa marker, showed no abnormalities in either the pattern or intensity of positive signals observed near the glomeruli in Trpm6 fl/fl ;Six2-Cre mice (Fig. 3a). Therefore, cells directly involved in renin secretion and its regulation do not seem to be markedly affected by Trpm6 deficiency. As TRPM6 is co-expressed with NCC in the DCT ( Supplementary Fig. 1c), we next examined the spatial relationship between TRPM6/NCCexpressing cells (DCT) and the macula densa or JG cells. Confocal images of sagittal DCT sections showed almost a complete overlap of TRPM6 and NCC signals throughout the DCT (Fig. 3b). We then stained for nNOS and NCC, and found that the areas positive for nNOS and NCC are connected, without any signal overlap (Fig. 3c). In addition, double staining for renin and TRPM6/NCC revealed that some TRPM6/NCC-expressing cells are in proximity to renin-expressing JG cells (Fig. 3d, e), suggesting that TRPM6/ NCC-expressing cells, as well as the macula densa, may have a significant impact on renin secretion.
Impaired renin secretion by Trpm6 deficiency. In general, the autonomic nervous system plays a crucial role in orchestrating the progression of the circadian rhythm. Renin secretion from JG cells during the active period is thought to be stimulated by noradrenalin secreted from renal sympathetic neurons 32 . To directly investigate the effect of such stimulation on renin . c-e Whole blood was collected from 2-to 4-month-old mice of the indicated genotypes at indicated time periods, and plasma AVP (c, n = 3 mice for each genotype), renin activity (d, n = 5 at ZT0 and ZT4, n = 6 for Trpm6 +/+ ; Six2-Cre and n = 4 for Trpm6 fl/fl ; Six2-Cre at ZT8, n = 3 at ZT12, ZT16, and ZT20), and noradrenaline (e, n = 4 at ZT0, ZT4, and ZT8, and n = 3 at ZT12, ZT16, and ZT20) were measured. Data are shown as means ± SEM. The p-values were determined by two-way ANOVA with Holm-Sidak post hoc tests. *p < 0.05, **p < 0.01, ***p < 0.001. secretion, we prepared kidney slices and stimulated them with isoproterenol, a β-adrenoreceptor (AR) agonist. Results showed a very clear decrease in renin-positive signal remaining in the slice after isoproterenol stimulation for 30 min (Fig. 4a). Quantitative analyses showed that the percentage of glomeruli surounded by renin-positive signal decreased from approximately 80% to 30% after isoproterenol stimulation (Fig. 4b), indicating the occurrence of massive renin secretion. However, we did not observe a significant effect of isoproterenol stimulation on renin signal in kidney slices from Trpm6 fl/fl ;Six2-Cre mice. To confirm the result, we also measured the blood renin activity after isoproterenol administration. The results show that significant elevation of blood renin activity was only observed in control mice (Fig. 4c), while the increase of heart rate was commonly observed in both strains ( Supplementary Fig. 7). These results imply that reninsecreting JG cells cannot respond to stimulation from renal sympathetic neurons, consistent with the disappearance of the circadian increase in blood pressure and renin activity in Trpm6 fl/ fl ;Six2-Cre mice. JG cells can also be stimulated to secrete renin in response to prostaglandins, among which prostaglandin E 2 (PGE 2 ) is considered most significant since it is physiologically secreted by macula densa cells when urinary Cl − levels decrease 17,18 . Moreover, it is also known that PGE 2 treatment of kidney slices can stimulate renin secretion efficiently 33 . Stimulation of JG cells with PGE 2 and isoproterenol commonly activates cAMP production, which ultimately causes renin secretion. Therefore, we also tested the effect of PGE 2 and the membrane-permeable cAMP analog, dibutylyl-cAMP (db-cAMP). PGE 2 and db-cAMP successfully stimulated renin secretion in kidney slices from Trpm6 fl/fl ;Six2-Cre mice and control mice (Fig. 4a, b). This indicates that Trpm6 deficiency does not abrogate the general machinery involved in sensing external cues, relaying intracellular signals, and executing renin secretion, but it specifically affects the response to β-AR.
Decreased β-AR expression in JG cells by Trpm6 deficiency. Impaired β-AR receptor response (Fig. 4) in Trpm6 fl/fl ;Six2-Cre mice led us to investigate the status of the β-AR receptor expression, and thus we performed immunofluorescence of β1-AR, the major β-AR subtype expressed in JG cells 34 . As reported previously 34 , we found the band-like staining pattern in renal sections of both control and Trpm6 fl/fl ;Six2-Cre mice (Fig. 5a, b). It is known that β1-AR is strongly expressed in both blood vessel and JG cells 34 , and consistently, the β1-AR-positive cells at the end of the band-like staining in control mouse kidney were also positive for renin staining (shown with an arrowhead). In contrast, staining of Trpm6 fl/fl ;Six2-Cre mouse kidney showed a smaller number of β1-AR/renin double-positive cells, suggesting that β1-AR expression was decreased in JG cells. To corroborate this finding, we isolated JG cells by using Percoll density gradient centrifugation method according to the previous report 35 (Fig. 5c). Immunoblotting analyses revealed that renin-positive JG cells are enriched in band 3 fraction, which is consistent with their report 35 , and β1-AR expression in this band 3 fraction was specifically much lower in Trpm6 fl/fl ;Six2-Cre mice. Taken together, we found that Trpm6 deficiency decreases the expression level of β1-AR in JG cells, which is consistent with the impaired renin secretion.
The effect of artificial intervention on circadian blood pressure variation. In the final set of experiments, we tested the effect of various interventions on circadian blood pressure variation. To confirm the importance of renin, we administered a clinically available renin inhibitor aliskiren 36 to wild-type mice (Fig. 6a). Like Trpm6 ablation, aliskiren administration by surgically implanting an osmotic minipump significantly suppressed the circadian blood pressure variation. Next, we performed renal denervation operations, since the activation of β-AR on JG cells through the sympathetic nervous system is the major pathway for stimulating renin secretion in a circadian manner 32 (Fig. 6b). Again, we found that the intervention suppressed the blood pressure variation. These results are in agreement with our model that defects in β-AR response and renin secretion are responsible for the loss of blood pressure variation in Trpm6 fl/fl ;Six2-Cre mice.
We also tested the effect of elevating dietary magnesium content. For this purpose, wild-type mice fed a high-Mg diet containing 0.6% (normal diet: 0.3% magnesium) for 1 month. As expected, those mice showed elevated blood Mg level (Supplementary Fig. 8a). We then checked the expression level of TRPM6 at DCT; it decreased drastically by feeding high-Mg diet ( Supplementary Fig. 8b, c), and the extent of decreased TRPM6 expression was larger than renal Cnnm2 ablation. When we measured the blood pressure in these mice, we found that mice fed a high-Mg diet showed significantly smaller blood pressure variation (Fig. 6c) as Trpm6-deficient mice, which is consistent with our notion that renal TRPM6 is important for circadian blood pressure variation.

Discussion
In this study, we demonstrated the importance of TRPM6 in the regulation of blood pressure. In particular, we show almost complete disappearance of the circadian variation of blood pressure in kidney-specific Trpm6-deficient mice. We also found that secretion of renin, a key hormone orchestrating blood pressure, was severely abrogated in Trpm6-deficient mice, which is consistent with the observed blood pressure phenotype. Several reports have shown that TRPM7, another Mg 2+ -permeable channel closely similar to TRPM6, also affects blood pressure 37,38 . TRPM7 prevents the increase of blood pressure upon Angiotensin II administration by suppressing the response of vascular endothelial cells to Angiotensin II 37 . To the contrary, it was recently reported that TRPM7 also facilitates the elevation of blood pressure upon leptin administration by mediating Mg 2+ influx into the carotid body cells 38 . Thus, the role of TRPM7 in blood pressure regulation is ambiguous. However, since the expression level of TRPM7 in the kidney was not affected by renal Trpm6 ablation ( Supplementary Fig. 5), it appears that TRPM7 is not directly involved in the phenotypes observed in this study.
As observed in Cnnm2-deficient mice, blood pressure values were reduced in Trpm6-deficient mice (Fig. 2) 14 . However, it should be noted that there was a clear difference between these mice. Cnnm2 disruption generally reduced blood pressure, but circadian variation was still clearly observable. In contrast, this variation disappeared in Trpm6-deficient mice. Such a qualitative difference is also consistent with our observation that the circadian variation of renin was preserved in Cnnm2-deficient mice ( Supplementary Fig. 9). It should be noted that both CNNM2 and TRPM6 are thought to be important for renal Mg 2+ reabsorption by mediating the vectorial Mg 2+ transport through the DCT, and indeed, mice deficient in either gene exhibited identical renal Mg 2+ -wasting phenotype ( Fig. 1 and ref. 39  The percentages of glomeruli with adjacent renin-positive cells are shown as means ± SEM (in both genotype, n = 6 (mock), 6 (Isoproterenol), 4 (PGE 2 ), and 2 (db-cAMP)). The p-values were determined by two-way ANOVA with Holm-Sidak post hoc tests. ***p < 0.001. c 2-to 3-month-old mice of the indicated genotypes were intraperitoneally administered with 10 mg/kg isoproterenol, and plasma collected the day before and 30 min after drug administration were subjected to renin activity measurement (n = 4 for Trpm6 +/+ ; Six2-Cre, and n = 8 for Trpm6 fl/fl ; Six2-Cre). Data are shown as means ± SEM. The p-values were determined by two-way ANOVA with Holm-Sidak post hoc tests. * p < 0.05. difference of blood pressure phenotype between Cnnm2-and Trpm6-deficient mice suggests that Mg 2+ transport in the DCT, the common function of the two molecules, is not directly involved in the circadian variation of blood pressure and renin secretion, suggesting that the CNNM2-independent role of TRPM6 underlies these intriguing phenotypes. Several mechanisms work to stimulate renin secretion from JG cells to maintain blood pressure. Among them, the input from renal sympathetic neurons is considered to play the principal role in augmenting blood pressure during the active period of circadian variation 32 . These neurons secrete noradrenalin, which binds to and activates β-AR on JG cells to stimulate cAMP production, ultimately leading to renin secretion 17,32 . In our experiments using kidney slices obtained from Trpm6-deficient mice, stimulation with isoproterenol, a β-AR agonist, showed no effect on renin secretion, while stimulation with PGE 2 or db-cAMP efficiently increased renin secretion (Fig. 4). PGE 2 is secreted by the macula densa in response to low levels of urinary Cl − and it activates EP2 and EP4 receptors on JG cells, which, like β-AR activation, ultimately stimulates cAMP production 40 . Therefore, these results imply that β-AR signaling is selectively abrogated in JG cells of Trpm6-deficient mice, without affecting the downstream cAMP-mediated signaling pathway. Consistently, immunofluorescence of mouse kidney sections and immunoblotting of isolated JG cells revealed that β-AR expression was significantly decreased in the JG cells of Trpm6-deficient mice (Fig. 5). Such reduced β-AR expression should make JG cells unable to secrete renin in response to sympathetic nerve stimulation during the active period, thereby eliminating circadian variation of blood renin level and blood pressure. It should be noted that there is a study using β1/β2-AR knockout mice, which showed decreased circadian blood pressure variation as Trpm6deficient mice 41 . They ascribed it to the decreased locomotor activity, but it remains possible that reduced circadian variation of renin activity contributes to decreased circadian variation of blood pressure. Alternatively, the unusual background of their mice (mixed background of C57BL6/J, 129, and FVB; we used pure C57BL6/J background) might mask the effect of β-AR on blood pressure variation; genetic backgrounds are critical for circadian blood pressure variation 42 .
How the loss of TRPM6 in DCT cells leads to the decreased β-AR expression in JG cells has remained elusive. Experiments using other types of cells have shown that β-AR expression is regulated by various molecules known to mediate intercellular signaling: Addition of glucocorticoids to the culture medium of vas deferens-derived cell line DDT1 MF-2 increases β-AR expression 43 . Also, it was reported that nitric oxide (NO) suppressed β-AR degradation in cardiac cells through Snitrosylation of GRK2 44 . Among these molecules, it is noteworthy that NO is secreted from renal macula densa cells, stimulating JG cells in a paracrine manner to increase renin secretion 45 . While the expression of nNOS, the NO producing enzyme in the kidney, is restricted to macula densa cells, they are in direct contact with DCT cells, the next segment of the renal tubule 16 . Thus, it is possible that Trpm6-deficiency in DCT cells leads to the decreased β-AR expression in JG cells by affecting the production or release of NO from the adjacent macula densa cells.
It is known that JG cells also communicate with other cells by forming gap junctions with surrounding cells, such as mesangial cells, and receive Ca 2+ to suppress renin secretion 17  are still closely located (ref. 46 and Fig. 3) and it is possible that these cells use gap junctional communications through intervening cells, such as mesangial cells; all three cell types reportedly express Cx37 47,48 , a connexin isoform that oligomerizes and constitutes gap junctions. Further analysis may identify the molecule and the route responsible for the intercellular signaling between DCT and JG cells, which is important to unveil the overall mechanism of disrupted circadian blood pressure variation by Trpm6-deficiency.
In Trpm6-deficient mice, the circadian variation of blood levels of not just renin but also noradrenaline disappeared (Fig. 2e). Activation of the renin-angiotensin system stimulates the release of noradrenaline to the bloodstream 29,30 , and thus, it seems reasonable that the effect of Trpm6 ablation on blood renin and noradrenaline levels was similar. Since noradrenaline also acts on the β-AR of JG cells to promote renin secretion, the two molecules are assumed to constitute a positive feedback loop, stimulating each other's secretion. However, it should be noted that the results of isoproterenol administration experiments explicitly show the defect in renin secretion (Fig. 4), and thus it seems to be the primary cause of the loss of the circadian variation of both renin and noradrenalin. Thus far, a variety of effective drugs have been developed and used to treat hypertensive patients. However, a significant number of patients show resistance to the conventional treatments that use a combination of major antihypertensive drugs of different functional mechanisms 49 . Surgical denervation of renal sympathetic neurons using endovascular catheters has been conducted on a trial basis to treat such resistant hypertension 50 . Its practical effectiveness at improving blood pressure control has been controversial 51,52 , but recently published studies have shown a significant reduction in blood pressure compared with sham controls 53,54 . There are also mixed results on the effect of renal denervation on circadian variation of blood pressure. Some reports showed that the extent of blood pressure decrease by renal denervation was larger during the active periods 53 , but others reported no change in circadian variation of blood pressure 55,56 . In the analysis using laboratory animals, our renal denervation operations against mice suppressed the circadian variation of blood pressure (Fig. 6). Similar operations were also performed against rats, but even control wild-type rats show almost no circadian blood pressure variation in the first place 57,58 , making it difficult to evaluate the effect of the operation. More studies are required to make clear the impact of renal denervation on circadian blood pressure variation, which would also help us to understand the more detailed mechanism of blood pressure dysregulation by Trpm6-deficiency.

Methods
Animals. Cnnm2 fl/fl ;Six2-Cre, Cnnm2 +/Δ , Cnnm4 Δ/Δ , and control mice were generated in our previous studies 14,20,59 . Trpm6-deficient mouse strains were generated from an ES cell clone carrying the recombinant allele of Trpm6, Trpm6 − (ID: EPD0741_2_G10, purchased from Knockout Mouse Project [KOMP]). Germline chimeras from this ES cell were bred with wild-type C57BL6/J mice to obtain Trpm6 +/− mice. Trpm6 +/− mice were then bred with FlpO mice 59 to obtain mice with the Trpm6-floxed allele, Trpm6 fl , and were subsequently crossed with Six2-Cre mice to obtain Trpm6 fl/fl ;Six2-Cre mice. Genotyping PCR was performed using the primers listed in Supplementary Table 1. Mice were maintained in individually ventilated cages in a specific pathogen-free, temperature controlled (20-26°C) facility with a 12-h light/dark cycle. Age-and sex-matched male and female mice (2-4-month-old) were used in each independent experiment. All mouse experiments were conducted in accordance with the Guidelines of Proper Conduct of Animal Experiments (issued by the Science Council of Japan) after the receipt of approvals from the Animal Care and Use Committee of the Research Institute for Microbial Diseases, Osaka University, Japan.
Expression constructs. Human Trpm6 complementary DNA (cDNA) was purchased from Dharmacon (IMAGE: 100062461). The cDNA fragment was inserted into pCMV-Tag2B (Agilent) for expression in mammalian cells.
Validation of anti-TRPM6 antibody. The specificity of the house made anti-TRPM6 antibody used for immunofluorescence analyses (Figs. 1b, 3b, and 3e and Supplementary Figs. 1c, 1e, and 8c) was validated by successive immunoprecipitation and immunoblotting analyses as follows ( Supplementary Fig. 10) 61 . To reduce the background noise level due to the antibody itself, the anti-TRPM6 antibody was covalently linked to Protein A agarose beads (Thermo Fisher) by incubating with 20 mM dimethylpimelimidate (Sigma) for 30 min at room temperature. Then, the beads were mixed with lysates of cells or mouse organs. Lysates were prepared by solubilizing cells or tissues in lysis buffer (20 mM Tris-HCl [pH 7.5], 0.5% Triton X-100, 150 mM NaCl, 2 mM EDTA, supplemented with cOmplete Protease Inhibitor Cocktail [Roche]). The beads were rotated overnight at 4°C, washed five times with lysis buffer, and the resultant immunoprecipitates were subjected to immunoblotting analyses. To reduce the background noise level due to the precipitated antibody, the Clean-Blot IP Detection Kit was used, which selectively reacts to native IgGs (primary antibodies) but not to denatured IgGs as secondary antibodies. RNAi knockdown. Duplex siRNAs against mouse Cnnm2 and Cnnm4 were purchased from Sigma and Invitrogen, respectively. The target sequences were as follows: Cnnm2-siRNA, 5′-CCATTGTGCAGCGAGTGAA-3′ and Cnnm4-siRNA, 5′-GGTGGACGAG ACCACAACTCTT-3′. A non-targeting siRNA sequence was used as a control (negative control LO GC, Invitrogen).
Elemental quantitation. Elemental quantification was done as follows 20 . Urine was collected by housing each mouse in metabolic cages (CLEA Japan) after 1 week acclimation. For serum samples, whole blood was collected from the abdominal aorta and incubated for 30 min at room temperature. Serum was then collected by centrifugation. Each element was quantified either by Xylidyl Blue-I (Mg, Wako) or ICP-ES (Mg, Ca, Na, and K; ICPS-8100; Shimadzu).
Hormone quantitation. Whole blood was collected from the abdominal aorta in EDTA-containing tubes and plasma was isolated by centrifugation. Plasma renin activity was measured using a fluorescence renin assay kit (Abcam) and a microplate reader (SH-9000 controlled by SF61 software; both from Corona electric). Plasma AVP and noradrenaline levels were quantitated using Arg 8 -Vasopressin ELISA kit (Enzo Life Sciences) and Norepinephrine ELISA Kit (LDN), respectively, and the abovementioned microplate reader.
PCR analyses. For cDNA synthesis 63 , RNA from mouse tissues or MDCT cells was prepared with RNAiso Plus (Takara), and cDNA was synthesized from 2 μg RNA with revertra ace (Toyobo). Semi-quantitative PCR analyses of mouse CNNMs 64 were performed with a thermal cycler (T100; Bio-Rad) using Ex-taq (Takara). As a positive control, we also included samples using plasmid vectors containing the cDNA region of each mouse Cnnm as templates. Real-time qPCR experiments 62 were performed with MiniOpticon (Bio-Rad) using the iQ SYBR green Supermix (Bio-Rad) as a reagent, and the data were obtained under the control of Opticon Monitor software (Bio-Rad). The quality of the final PCR product was confirmed by agarose gel electrophoresis. No obvious non-specifically amplified DNAs were observed. The primer sets used are listed in Supplementary Table 1.
where R is the ratio of the signal intensity with 330 − 350 nm excitation (F 1 ) to that with 370 − 390 nm excitation (F 2 ) (R = F 1 /F 2 ), R max is the maximum value of R, R min is the minimum value of R, Q is the ratio of the signal intensity with 370 − 390 nm excitation when R min was obtained to the signal intensity with 370 − 390 nm excitation when R max was obtained (F 2min /F 2max ), and we used 1.5 mM as the K d value 20 . The value of R min , R max , F 2min , and F 2max were obtained after each experiment. The cells were incubated in serum-free DMEM supplemented with 6 μM 4-Bromo-A23187 (Wako) and 10 mM EDTA, and the fluorescent images were acquired to obtain R min and F 2min values. Subsequently, the media was replaced with serum-free DMEM supplemented with 6 μM 4-Bromo-A23187 and 50 mM MgCl 2 , and the fluorescent images were acquired to obtain R max and F 2max values. For [Mg 2+ ] i measurement of cells cultured under various extracellular Mg 2+ conditions, loading of Mag-fura2, washing, and data acquisition was performed by bathing the cells with magnesium/calcium-free DMEM (Thermo Fisher) supplemented with 1.8 mM CaCl 2 and the indicated concentrations of MgCl 2 .
Immunofluorescence analyses of tissue sections. Whole kidneys or kidney slices from 2-month-old mice were embedded in OCT compound (Sakura Finetechnical), frozen in liquid nitrogen, and then cut into 20-or 30-µm sections using a cryostat (Leica). Sections were mounted on glass slides, air-dried, and fixed with ice-cold 2% PFA in PBS for 10 min. For subsequent immunostaining 62,64 , the sections were blocked twice in PBSMT (2% skim milk, 0.3% Triton X-100 in PBS) for 1 h at 4°C. Then, the specimens were incubated with primary antibody diluted in PBSMT for overnight at 4°C. After 7 washes with PBSMT, the specimens were incubated with appropriate fluorophore-conjugated secondary antibodies for overnight at 4°C. Nuclei were also stained with 4',6-diamidino-2-phenylindole (DAPI). After seven washes with PBSMT, the sections were observed with a confocal scanning laser microscope (FLUOVIEW FV1000; Olympus) under the control of FV10-ASW software (Olympus).
Telemetric measurements of blood pressure, heart rate, and locomotor activity. Telemetric measurements of blood pressure were performed as follows 14 . Blood pressure transducers (TA11PA-C10, Data Sciences International) were surgically inserted into the left common carotid artery of 2-to 3-month-old mice under anesthesia (by intraperitoneal administration of 0.3 mg/kg medetomidine, 4 mg/kg midazolam, and 5 mg/kg butorphanol). After 3 weeks of recovery, systolic and diastolic blood pressure values were recorded with a 2-min scheduled sampling every 60 min by using Dataquest ART software (Data Sciences International). Records were obtained for at least five consecutive days, and the averages of measurements taken at the same time each day were determined for each mouse and used for further statistical analyses, except for Supplementary Fig. 4. Based on Nakamori et al. 68 , renal denervation was performed 2 weeks before inserting blood pressure transducers as follows. Kidneys were exposed by abdominal incision and connective tissue around the renal vessels was dissected. Then, the renal vessels were immersed in 95% ethanol for 2 min and thereafter, in PBS for 2 min.
Sham operation was performed using the equivalent procedure without ethanol immersion. Aliskiren was administered 25 mg/kg/day by subcutaneously embedding an osmotic pump (1004, ALZET), and recording was started from the day after pump introduction. For magnesium supplementation, mice were fed a magnesium-enriched diet containing 0.6% magnesium (CLEA Japan) for 1 month before blood pressure measurement.
Heart rate was measured with the same radiotelemetry system, except that the recording was performed with a 30-s scheduled sampling every minute, and a single measurement was used for further analysis.
Locomotor activities were assessed using the same radiotelemetry system, by monitoring the changes in signal strength from the transducer due to the animal's locomotion. In this case, transducers were surgically inserted into the subcutaneous space in the right flank. Recordings were performed with the same time intervals as described above.
Isolation of JG cells. JG cells were isolated according to the method reported by Della Bruna and colleagues 35 , with several minor modifications. Kidneys from four mice were minced and incubated in buffer 1 (130 mM NaCl, 5 mM KC1, 2 mM CaCl 2 , 10 mM glucose, 20 mM sucrose, 10 mM Tris-HCl [pH 7.4]) supplemented with 0.25% trypsin and 0.1 % collagenase at 37°C for 2 h. After filtration with a cell strainer (40 μm pore size, Falcon), cells were washed with buffer 1 and mixed with 20× volume of 30% isoosmotic Percoll solution. Centrifugation at 27,000 × g for 25 min yielded three bands, and each band was collected and used for further analyses.
Statistical analyses. All statistical analyses were performed using Prism software (GraphPad) after data analyses with Microsoft Excel 11, Metamorph (Molecular Device), or Dataquest ART (Data science international). Circadian pattern was determined by cosinor analyses, which fits the data to the following formula y = BaseLine + Amplitude × cos ([x -PhaseShift] × π/12). All statistical data are presented as means ± SEM and p-values were obtained by either Student's two-tailed ttest, one-or two-way ANOVA with Holm-Sidak post hoc test, or chi-square test, as described in the figure legends.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.