The kidney plays a key role in the maintenance of Mg2+ homeostasis. Specifically, the distal convoluted tubule (DCT) is instrumental in the fine-tuning of renal Mg2+ handling. In recent years, hereditary Mg2+ transport disorders have helped to identify important players in DCT Mg2+ homeostasis. Nevertheless, several proteins involved in DCT-mediated Mg2+ reabsorption remain to be discovered, and a full expression profile of this complex nephron segment may facilitate the discovery of new Mg2+-related genes. Here, we report Mg2+-sensitive expression of the DCT transcriptome. To this end, transgenic mice expressing enhanced green fluorescent protein under a DCT-specific parvalbumin promoter were subjected to Mg2+-deficient or Mg2+-enriched diets. Subsequently, the Complex Object Parametric Analyzer and Sorter allowed, for the first time, isolation of enhanced green fluorescent protein-positive DCT cells. RNA extracts thereof were analyzed by DNA microarrays comparing high versus low Mg2+ to identify Mg2+ regulatory genes. Based on statistical significance and a fold change of at least 2, 46 genes showed differential expression. Several known magnesiotropic genes, such as transient receptor potential cation channel, subfamily M, member 6 (Trpm6), and Parvalbumin, were upregulated under low dietary Mg2+. Moreover, new genes were identified that are potentially involved in renal Mg2+ handling. To confirm that the selected candidate genes were regulated by dietary Mg2+ availability, the expression levels of solute carrier family 41, member 3 (Slc41a3), pterin-4 α-carbinolamine dehydratase/dimerization cofactor of hepatocyte nuclear factor-1α (Pcbd1), TBC1 domain family, member 4 (Tbc1d4), and uromodulin (Umod) were determined by RT-PCR analysis. Indeed, all four genes show significant upregulation in the DCT of mice fed a Mg2+-deficient diet. By elucidating the Mg2+-sensitive DCT transcriptome, new candidate genes in renal Mg2+ handling have been identified.
- magnesium homeostasis
- gene expression microarray
- Complex Object Parametric Analyzer and Sorter
- distal convoluted tubule
hypomagnesemia is a common clinical manifestation associated with type 2 diabetes, hypertension, osteoporosis, tetany, seizures, depression, and the use of a variety of drugs (6, 21). Clinical studies addressing the Mg2+ status of critically ill patients showed that a substantial number of patients were hypomagnesemic. Moreover, low blood Mg2+ levels are associated with a poor clinical outcome (23, 40). The regulation of blood Mg2+ levels is an equilibrium between intestinal Mg2+ absorption and renal Mg2+ reabsorption. In the kidney, Mg2+ reabsorption in the proximal tubule and thick ascending limb of Henle (TAL) is a passive paracellular process, whereas in the distal convoluted tubule (DCT), Mg2+ reabsorption is a highly regulated and transcellular mechanism. This latter segment facilitates the fine-tuning of renal Mg2+ uptake, since no reabsorption takes places beyond the DCT.
Over the last decade, the elucidation of the molecular origin of human genetic diseases has helped to identify new proteins involved in Mg2+ transport in the DCT (6). In patients with hypomagnesemia and secondary hypocalcemia, mutations have been identified in transient receptor potential (TRP) subfamily M, member 6 (TRPM6), the gatekeeper of Mg2+ entry in DCT cells (39, 47). The abundance of TRPM6 at the plasma membrane is regulated by EGF, and, as a result, mutations in the pro-EGF gene are causative for hypomagnesemia (16). Mg2+ transport via TRPM6 is a passive process that depends on the membrane potential across the luminal membrane (14). Therefore, several players involved in the maintenance of this membrane potential have been identified by gene-linkage analyses in patients with hereditary hypomagnesemia. For instance, a mutation in KCNA1, which encodes the voltage-gated K+ channel Kv1.1, impairs the luminal extrusion of K+ necessary to maintain a substantial luminal membrane potential (14). At the basolateral membrane, the K+ channel Kir4.1 acts an important player in the K+ recycling that seems necessary to maintain high Na+-K+-ATPase activity (1). Indeed, mutations in KCNJ10, which codes for Kir4.1, and mutations in FYXD2, which codes for the γ-subunit of Na+-K+-ATPase, have been linked with hypomagnesemia (1, 4). Moreover, mutations in the transcription factor hepatocyte nuclear factor (HNF)-1β (HNF1B), which regulates the transcription of FXYD2, cause a similar phenotype (27). Recently, mutations in the DCT-expressed cyclin M2 (CNNM2) gene have been linked with hypomagnesemia (43). It was postulated that CNNM2 may act as an intracellular Mg2+ sensor regulating Mg2+ transport (7). Although gene linkage studies have gained insights into Mg2+ reabsorption in the DCT over the last years, several factors remain unidentified. For instance, at present, the Mg2+ extrusion mechanism of DCT cells is still unknown.
The aim of the present study was, therefore, to elucidate the Mg2+-sensitive transcriptome of the DCT cell. To this end, primary DCT cells were isolated from mouse kidneys, and a comprehensive expression profile specific for DCT cells was established using gene expression microarrays. Furthermore, by comparing DCT expression profiles of mice fed high and low Mg2+-containing diets, the Mg2+ sensitivity of the DCT transcriptome was determined. Here, we describe the identification of new candidate genes potentially involved in Mg2+ reabsorption in the DCT.
MATERIALS AND METHODS
All the experimental procedures are in compliance with the animal ethics board of Radboud University (Nijmegen, The Netherlands). Transgenic C57Bl/6 mice expressing enhanced green fluorescent protein (eGFP) under the parvalbumin (PV) promotor were kindly provided by Dr. Hannah Monyer (University of Heidelberg, Heidelberg, Germany) (29). The genotype was determined under ultraviolet light by checking the fluorescent emission of muscular PV expression in mouse legs. Littermates were housed in a temperature- and light-controlled room with standard pellet chow (SSNIFF Spezialdiäten, Soest, Germany) and deionized drinking water available ad libitum until the start of the experiment. PV-eGFP-positive mice were selected for experiments at the age of 4–6 wk. During the experiments, mice were fed low [0.02% (wt/wt)] or high [0.48% (wt/wt)] Mg2+-containing diets for 15 days (SSNIFF Spezialdiäten). During the last 48 h of the experiment, mice were housed in metabolic cages for urine and fecal collection (24-h adaptation, 24-h sampling). Blood samples were taken at the start of the experiment and just before euthanization.
Isolation of the DCT using Complex Object Parametric Analyzer and Sorter sorting.
PV-eGFP-positive tubules were isolated as previously described (26). In brief, mice aged 4–6 wk were anesthetized and perfused transcardially with ice-cold Krebs buffer [containing (in mM) 145 NaCl, 5 KCl, 1 NaH2PO4, 2.5 CaCl2, 1.8 MgSO4, 10 glucose, and 10 HEPES/NaOH; pH 7.4]. Directly after perfusion of the mice, whole kidneys were harvested, and eGFP-positive kidney material was manually selected under a microscope. This part of the procedure took up to 30–45 min. Subsequently, fragments of the kidney were digested in Krebs buffer containing 1 mg/ml collagenase (Worthington, Lakewood, NJ) and 2,000 U/ml hyaluronidase (Sigma, Houten, The Netherlands) for 15 min at 37°C. Subsequently, kidney tubules sized between 40 and 100 μm were collected by filtration. Oversized material was digested again in two additional cycles of 10-min digestion and filtration. Tubules collected from the three digestions were ice cooled and sorted by the Complex Object Parameric Analyzer and Sorter (COPAS; Union Biometrica, Holliston, MA) at a rate of 2,000–4,000 tubules/h. Sorted tubules were directly collected in 1% (vol/vol) β-mercaptoethanol containing RLT buffer supplied by the RNeasy RNA extraction kit (Qiagen, Venlo, The Netherlands). In total, the sorting procedures of tubules from one mouse took several hours. Per mouse, 4,000 eGFP-positive fluorescent tubules were collected, and, as a control, an additional 4,000 tubules were sorted from the same kidney sample without selection for eGFP-positive cells. This control sample contained both eGFP-positive and eGFP-negative cells of the same size as the selected sample. Four thousand ubules were pooled on a Qiagen Rneasy microcolumn for RNA extraction according to the manufacturer's protocol.
For microarray experiments, RNA from two male and two female mice per group was taken. RNA from low- and high-Mg2+ groups was matched in pairs based on the sex of the mice. Double-round RNA amplifications and labeling were performed as previously described (37) on an automated system (Caliper Life Sciences) with 10–50 ng total RNA from each sample. Briefly, total RNA was amplified twice from ∼50,000 cells by cDNA synthesis with oligo(dT) double-anchored primers followed by in vitro transcription using a T7 RNA polymerase kit (Ambion, Bleiswijk, The Netherlands). During the second round of transcription, 5-(3-aminoallyl)-UTP was incorporated into the single-stranded cRNA. Cy3 and Cy5 N-hydroxysuccinimide esters (Amersham Biosciences, Roosendaal, The Netherlands) were coupled to 2 μg cRNA. RNA quality was monitored after each successive step using the equipment described above. Mouse Whole Genome Gene Expression Microarrays (V2, Agilent Technologies), representing 39,429 Mus musculus 60-mer probes in a 4 × 44-K layout, were used for hybridizations with 1 μg of each alternatively labeled cRNA on a HS4800PRO system supplemented with QuadChambers (Tecan Benelux) according to van de Peppel et al. (45). After hybridization, slides were scanned using the Agilent G2565BA DNA Microarray Scanner.
After automated data extraction using Imagene 8.0 software (BioDiscovery), Lowess normalization was performed on mean spot intensities followed by dye bias correction based on a within-set estimate, as previously described (25, 50). Data were analyzed using MAANOVA (49). In a fixed effect analysis, sample, array, and dye effects were modeled. P values were determined by a permutation F2-test, in which residuals were shuffled 5,000 times globally. Genes with P values of <0.05 after family-wise error correction were considered significantly changed. Additionally, a twofold change cutoff was applied.
Mg2+ and Ca2+ analysis.
Serum and urine total Mg2+ and Ca2+ concentrations were determined using a colorimetric assay kit according to the manufacturer's protocol (Roche Diagnostics, Woerden, The Netherlands). Urine volume was measured to calculate 24-h Mg2+ excretion.
Quantitative real-time PCR.
The obtained mRNA from 4,000 tubules was reverse transribed using Moloney murine leukemia virus reverse transcriptase (Invitrogen, Bleiswijk, The Netherlands) for 1 h at 37°C. cDNA was subsequently used to measure PV, Na+-Cl− cotransporter (Ncc), Trpm6, Kcna1, Kcnj10, Egf, Fxyd2, Hnf1b, Cnnm2, solute carrier family 41, member 3 (Slc41a3), pterin-4 α-carbinolamine dehydratase/dimerization cofactor of HNF-1α (Pcbd1), TBC1 domain family, member 4 (Tbc1d4), and uromodulin (Umod) mRNA levels. Gene expression levels were determined by quantitative real-time PCR on a Bio-Rad analyzer and normalized for Gapdh expression. Primer sequences are shown in Supplemental Table S2.1
Immunohistochemistry was performed as previously described (13). In brief, costaining for UMOD with thiazide-sensitive NCC was performed on 5-μm sections of fixed frozen mouse kidney samples. Sections were incubated for 16 h at 4°C with the following primary antibodies: guinea pig anti-UMOD (1:750, BioTrend, Köln, Germany) and rabbit anti-NCC [1:100 (32)]. For detection, kidney sections were incubated with Alexa fluor-conjugated secondary antibodies. Images were taken with an AxioCam camera and AxioVision software (Zeiss, Sliedrecht, The Netherlands).
For RT-PCR experiments, statistical significance was determined using an unpaired Student's t-test. In all experiments, data are expressed as means ± SE. Differences with P values of <0.05 were regarded as statistically significant.
In our study, transgenic mice expressed eGFP downstream of a PV promoter were subjected to Mg2+ diets. Within the kidney, PV is exclusively expressed in the DCT (3). Combining PV-eGFP mice with the COPAS tissue sorting system allowed the specific isolation of DCT cells (26). Here, PV-eGFP mice were subdivided into 2 groups consisting of 12 mice each. Each group was subjected to a high [0.48% (wt/wt)] or low [0.02% (wt/wt)] Mg2+-containing diet for 15 days. Subsequently, DCT tubules were isolated using COPAS, and total RNA was extracted (Fig. 1A). Four of twelve pairs of RNA extracts were processed for and subjected to pairwise (high vs. low) microarray analysis. The eight remaining RNA extracts per group were used to validate the results of the microarray experiments by RT-PCR.
As a control for purity, transcriptional expression levels of the DCT marker genes Trpm6 and Ncc were determined by RT-PCR (Fig. 1, B and C). Indeed, expression of thiazide-sensitive Ncc and Trpm6 were enriched by 200- and 8-fold, respectively, compared with total kidney material of the same mice. Moreover, expression of the protein markers for proximal tubules [aquaporin 1 (Aqp1)] and the collecting duct (Aqp2) was reduced. A slight overlap was detected with the TAL and connecting tubules when samples were tested for the marker genes Na+-K+-2Cl cotransporter 2 (Nkcc2) and Trpv5, respectively (Fig. 1B). These results indicate that the COPAS-sorted tubules contained primarily DCT transcripts.
Effects of Mg2+ diets.
Before mRNA transcripts were subjected to microarray hybridization, the effects of Mg2+ diets on plasma and urinary Mg2+ levels were assessed (Fig. 2, A–C). In line with our previous experiments, the serum Mg2+ concentration dropped significantly in mice fed the Mg2+-deficient diet (Fig. 2A) (15). The Mg2+-enriched diet resulted in slight, albeit significant, hypermagnesemia (Fig. 2A). Urinary Mg2+ excretion showed the same pattern: high excretion in mice fed the Mg2+-enriched diet and low excretion in mice fed the Mg2+-deficient diet (Fig. 2B). Serum Ca2+ concentrations did not change significantly (Fig. 2C). In contrast, the urinary Ca2+ concentration was significantly increased in mice fed a high-Mg2+ diet (Fig. 2D). Urinary and serum Na+ and K+ values were not significantly altered by dietary Mg2+ availability (Fig. 2, E–H).
DCT transcriptome analysis.
The DCT transcriptome was first analyzed for abundance of individual mRNAs. This was achieved by analyzing the average log2 intensity of the two-channel microarray data [the so-called “A value”, where A = ½log2(RG)] after normalization between the two channels. Although the A value is influenced by the amplification of the transcripts and the position of microarray feature sequences, it gives a qualitative view of the gene expression level in DCT tubules. Of the most abundant transcripts, more than one-third were directly linked to mitochondrial function (see Supplemental Table S1). The DCT cells are the cells with the highest mitochondrial density in the mammalian kidney (9). Moreover, DCT marker genes have moderate to high A values. Ncc gave an average expression signal of 11.9, and Trpm6 had a value of 11.7. Several genes implicated in DCT-mediated Mg2+ transport were highly expressed. Egf and Fxyd2 were among the 50 most expressed DCT genes, with expression values of 16.7 and 16.5. Interestingly, marker proteins of the TAL showed low expression levels. For instance, Nkcc2 had a value of 8.6 and Claudin16 had an expression level of 6.9. Taken together, these results confirm the purity of the COPAS-selected DCT cells.
Mg2+-sensitive gene expression.
To identify the DCT genes involved in Mg2+ homeostasis, Mg2+-sensitive differential expression of the transcriptome was determined by pair-wise comparison of high versus low Mg2+ samples. As a first approach, we examined the expression of the genes that are associated with hypomagnesemia in genetic disease (6). Interestingly, only Trpm6 and Egf were shown to be significantly upregulated in mice fed the low-Mg2+ diet compared with mice fed with the high-Mg2+ diet (Table 1). To assess the reproducibility of our microarray data, the microarray results were validated by RT-PCR on a separate set of DCT mRNA transcripts (Fig. 3, A–H). Indeed, this analysis confirmed that dietary Mg2+ regulates Trpm6 expression. Trpm6 expression in low-Mg2+ diet-fed mice was 2.3-fold enriched compared with mice fed the high-Mg2+ diet (Fig. 3A). Additionally, only Egf and Cnnm2 were differentially regulated (Fig. 3, B and F). RT-PCR analyses showed enrichment in low-Mg2+ diet-fed mice of 1.5- and 2.1-fold, respectively, relatively to high-Mg2+ diet-fed mice. The enrichment of Cnnm2 was not significant in the microarray data (P = 0.68) but was highly significant in the RT-PCR analysis (P < 0.002). Kcna1 transcripts, which encode Kv1.1, were not detectable in the pure DCT material (Fig. 3C), confirming the low signal in the microarray (A value: 5.7). No major differences between microarray and RT-PCR data were observed, indicating high reproducibility between the RT-PCR and microarray analyses.
Next, we analyzed the microarray Mg2+ sensitivity of all potential mammalian Mg2+ transporters that have been described in the literature (34). Compared with mice fed a Mg2+-enriched diet, only Trpm6 and Slc41a3 mRNA transcripts showed a significant enrichment in mice fed a Mg2+-deficient diet (Table 2). Slc41a3 has been previously linked to Mg2+ transport, although the exact function of this gene was never examined (34). Cnnm1, Cnnm3, Cnnm4, Slc41a1, and Slc41a2 and all other transporters of the NaCl-inducible protein (Nipa), MagT, membrane Mg2+ transporter (MMgT), Mrs, and putative ZDHHC-type palmitoyltransferase 6-like (Zdhhc) families are not Mg2+ sensitive and presented relatively low A values, suggesting low DCT abundance.
New candidate genes.
The main objective of this study was to identify new players in renal Mg2+ handling. Therefore, we subsequently selected differentially expressed genes with statistical significance (P < 0.05) and a minimal fold change in expression of two. As a result of this stringent analysis, 46 gene candidates were identified (Table 3). Among the genes that were enriched in Mg2+-deficient mice, previously mentioned Trpm6 and Slc41a3 were detected. Additionally, 33 genes were identified that were enriched in the low-Mg2+ group (compared with the high-Mg2+ group). The gene with the highest fold change was sex-determining region Y-box 9 (Sox9), which encodes a transcription factor involved in renal development. Other high-ranking genes were the DCT marker Pv, glycoprotein Umod, and complement factor Cd55. Furthermore, we identified direct and indirect regulators of Na+-K+-ATPase, Tbc1d4 and Pcbd1. Notable was the presence of several genes that are mainly known for their role in Ca2+ homeostasis, such as calbindin28K (Calb1) and the vitamin D receptor (Vdr). Only 11 genes were significantly enriched with a minimal fold change of two in the high-Mg2+ group compared with the low-Mg2+ group. None of these genes were previously linked to Mg2+ homeostasis. The most profound differential expression was measured with the PGE receptor 3 (subtype EP3) gene (Ptger3), which encodes a prostaglandin receptor.
To confirm the results of the microarray analysis, mRNA transcript levels of all major candidate genes were determined using RT-PCR (Fig. 4, A–D). Indeed, Slc41a3 mRNA expression was 1.8-fold enriched in Mg2+-deficient mice, similar to the values observed on the microarray (Fig. 4A). Likewise, mRNA expression levels were determined for Pcbd1, Tbc1d4, and Umod, since all three genes are potentially involved in Mg2+ handling (Fig. 4, B–D). Pcbd1 is a regulator of HNF-1β function, Tbc1d4 has been shown to modulate Na+-K+-ATPase activity, and Umod effects transcellular Na+ transport (2, 28, 31). The fold enrichment values for these three genes estimated by RT-PCR corresponded with the values obtained by the microarray analysis. Since Umod is mainly described as a TAL gene, we further examined Umod DCT expression with immunohistological staining and RT-PCR. Interestingly, Umod was 3.5 times more expressed in DCT cells compared with total kidney material (Fig. 5A). UMOD colocalized with NCC in early DCT cells (Fig. 5C). Moreover, only the DCT fraction of Umod expression, but not the total kidney sample mainly containing TAL Umod expression, was shown to be Mg2+ sensitive (Fig. 5B).
To determine whether the differentially regulated genes belong to the same functional pathways or cellular processes, Gene Oncology (GO) term enrichment analysis was performed by gene set enrichment analysis (Table 4). In DCT samples from Mg2+-deficient mice, EGF signaling was upregulated, as evidenced by the differentially regulated GO term “EGF pathway.” EGF activates a signaling pathway leading to increased TRPM6 plasma membrane expression and, hence, more Mg2+ reabsorption (44). Interestingly, “bone mineralization” was also among the enriched GO terms in mice fed a low-Mg2+-diet. Mg2+ plays an important role in this latter process (38). The most evidently enriched functional pathways were observed in high-Mg2+ diet-fed mice. Among the upregulated functions, many GO terms indicating mitochondrial activity were evidenced.
In this study, the Mg2+-sensitive transcriptome of the DCT was elucidated. By taking advantage of the COPAS sorting system, we were the first to isolate primary DCT material with high purity for gene expression analysis. As a result of our approach, new promising candidate players in renal Mg2+ homeostasis were identified. Among the candidates, 1) SLC41A3 has been proposed as a potential Na+/Mg2+-exchanging mechanism, 2) PCBD1 might influence HNF-1β activity, 3) TBC1D4 is a possible regulator of ion channel membrane availability, and 4) UMOD could regulate Mg2+ transport by decreasing the urinary flow rate.
We isolated DCT cells from mouse kidneys combining the COPAS sorting system with PV-eGFP mice. Until now, it has been difficult to obtain high-quality expression profiles of the isolated DCT fraction because of the quantitative limits of the microdissection technique. Therefore, previously reported expression analyses resulted in a gene expression profile of the combined DCT and CNT (Connecting Tubule) fraction (33). The COPAS system provides the only high-efficiency and high-quality alternative for microdissection. The purity of the COPAS-sorted DCT cells was significantly higher than in previous reports, as demonstrated by the impressive fold enrichments of 200 and 8 for Ncc and Trpm6 (26). The high abundance of mitochondrial genes in our microarray data set further showed sample purity. In the kidney, DCT cells contain the highest amount of mitochondria (9). Inevitably, COPAS-sorted DCT cells can contain adjacent TAL and CNT cells. Although the TAL marker Nkcc2 was not significantly enriched compared with total kidney samples, transcripts isolated from COPAS-sorted cells may contain genes expressed in the TAL. More importantly, Trpv5 expression was upregulated in COPAS-sorted DCT cells, which can be explained by the gradual transition of DCT1, via DCT2, to the CNT region, causing a partial overlap of Pv and Trpv5 expression. Since the aim of our study was to identify Mg2+-sensitive gene transcription, the potential limit of sample purity will not affect the results of the microarray analysis. Altogether, this approach allowed the identification of magnesiotropic DCT genes.
Slc41a3 is probably the most interesting candidate among the newly identified Mg2+-sensitive DCT genes. SLC41A3 is part of a family of three putative Mg2+ transporters that were first described by Quamme et al. (34) Using the oocyte voltage-clamp model, it has been shown that all three SLC41 proteins transport Mg2+ within its physiological range, although these findings have never been repeated in mammalian cell models (19, 34). SLC41A3 has never been investigated in detail, but lessons from the other SLC41 family members have demonstrated that it consists of 11 transmembrane domains (24). A recent report (20) claimed that SLC41A1 could act as a Na+/Mg2+ exchanger. Likewise, the highly homologous SLC41A3 protein could have a similar function. Since SLC41A1 and SLC41A2 are not differentially regulated in the DCT, it seems possible that SLC41A3 is the DCT-specific Mg2+ extrusion mechanism. Further research characterizing the function of SLC41A3 is needed to confirm this hypothesis.
Interestingly, the expression of Hnf1b and Fxyd2 was not differentially regulated in this study. Nevertheless, we identified in Pcbd1 a potential modulator of HNF-1β/FXYD2 activity. First described in 1991, PCBD1 is a protein with dual activity (28). PCBD1 can stimulate HNF-1α and HNF-1β transcription and has a role in tetrahydrobiopterin regeneration (42). HNF-1 transcriptional activation by PCBD1 does not depend on its enzymatic activity (17). Its role in HNF-1β activation seems of special interest with respect to Mg2+ homeostasis. HNF-1β is a transcription factor that stimulates FXYD2 expression, encoding the γ-subunit of Na+-K+-ATPase (11). PCBD1, also known as dimerization cofactor of HNF-1 (DCOH), stabilizes the HNF-1 dimer formation necessary for Fxyd2 transcription. Although PCBD1 activation has never been linked to FXYD2 directly, PCBD1 could indirectly stimulate Fxyd2 transcription by enhancing HNF-1β function. Given that Pcbd1 is transcriptionally upregulated in Mg2+-deficient conditions and that such regulation is absent for Fxyd2 and Hnf1b, PCBD1 might be the trigger in FXYD2-mediated fine tuning of Mg2+ reabsorption in the DCT.
During analyses of the microarray data, another modulator of Na+-K+-ATPase activity caught attention. Tbc1d4, previously known as AS160, was significantly upregulated in low-Mg2+ diet-fed mice. TBC1D4 has been described as a Rab-GTPase-activating kinase involved in the regulation of Na+-K+-ATPase plasma membrane availability (2). In this ability, it might directly influence the transmembrane potential that is necessary for Mg2+ reabsorption. Interestingly, the function of TBC1D4 is not limited to the regulation of Na+-K+-ATPase. Originally described as a regulator of glucose transporter 4 (GLUT4), TBC1D4 has recently been linked to the regulation of channels including AQP2 and the epithelial Na2+ channel (18, 22, 30). Thus, it might be relevant to consider TBC1D4 as a modulator of TRPM6 plasma membrane trafficking. This notion is further supported by the fact that a phosphatidylinositol 3-kinase (PI3K)/Akt/Rac1-mediated pathway activated upon EGF stimulation regulates TRPM6 plasma membrane abundance and that TBC1D4 is an Akt substrate that can also be stimulated by EGF (12, 44). GO term enrichment analysis of our microarray results showed that the EGF pathway was upregulated in low-Mg2+ diet-fed mice.
Although previously implicated in renal ion homeostasis, our finding that Umod expression is highly Mg2+ sensitive is unprecedented (31). Umod, also described as Tamm-Horshfall glycoprotein, is the most abundant protein in urine, but its exact function remains to be determined (8). It has been proposed that after excretion in the pro-urine, Umod forms an anionic gel-like structure that retards the flow of positively charged ions (46). Increasing Umod expression would then decrease the urinary flow rate and thereby enhance the uptake of cations, such as Mg2+. Although often used as a marker for the TAL, Umod expression has been described to be not exclusively restricted to the TAL but also to be profound in the early DCT (33, 41). Studies using gene expression microarrays of microdissected DCT tubules were inconclusive. Whereas Pradervand and colleagues (33) showed clear enrichment of Umod mRNA expression, another study (5) detected barely any Umod mRNA transcripts in the DCT. Immunohistological stainings confirmed the colocalization of UMOD with the DCT marker NCC, although it cannot be excluded that TAL-cleaved UMOD reached the DCT by urinary flow and stuck to the luminal membranes. Interestingly, RT-PCR analysis showed that only DCT expression of Umod was Mg2+ sensitive, in contrast to Umod expression in total kidney samples. This suggests different functional roles of UMOD in DCT and TAL segments of the kidney.
Genes linked to hypomagnesemia.
Of all genes that have been linked to hypomagnesemia in human genetic diseases, only a few are transcriptionally regulated (6). Trpm6, Egf, and Cnnm2 were modulated, indicating that Mg2+ transport is partially regulated at the transcriptional level. Major regulatory pathways are executed on the protein level. The EGF signaling pathway is one of the best-described regulatory mechanisms of TRPM6-mediated Mg2+ transport in the DCT (16, 44). Although its mean effectors, such as PI3K, Akt, and Rac1, are not transcriptionally affected by alterations in dietary Mg2+ availability, EGF is among the enriched GO terms. Futhermore, Kcna1, which codes for the K+ channel Kv1.1, could not be detected in mouse DCT material by RT-PCR. Kcna1 is part of a larger family of voltage-gated K+ channels that are known to form heteromeric complexes. Although Kv1.1 has been implicated in Mg2+ handling in patients, its function might be compensated for by other K+ channels in the PV-eGFP mouse model (14). In this context, it is interesting to note that ROMK is 1.7-fold upregulated in low-Mg2+ diet-fed mice. Although this result does not reach statistical significance (P = 0.06), it suggests that ROMK might play a role in the maintenance of the luminal membrane potential necessary for Mg2+ reabsorption in the DCT.
Inevitably, although changing only dietary Mg2+ availability, our approach also led to the detection of many Ca2+ homeostasis-related genes, such as the vitamin D receptor, klotho, and calbindin-D28K (35). Activation of the Ca2+-sensing receptor (CaSR) is hypothesized to explain this phenomenon (10). The CaSR is not only sensitive to Ca2+ but is also activated upon Mg2+ binding. As a result, changes in the availability of Mg2+ activate Ca2+-regulatory pathways at the local and whole body levels. Hypomagnesemia is, therefore, often associated with a disturbed Ca2+ balance in patients (39, 47). In a similar fashion, dietary Mg2+ availability influenced Ca2+ homeostasis in our mice, as evidenced by the altered serum and urinary Ca2+ levels in the present and previous studies (15). Therefore, differential gene expression induced by dietary Mg2+ should be analyzed with care. Although serum Ca2+ and Mg2+ measurements showed that the alterations in Mg2+ balance are more profound, it has to be taken into account that differential gene expression might be the effect of Ca2+ alterations. The interlink between Ca2+ and Mg2+ homeostasis was further evidenced in the GO term analysis of the microarray data. Although analyzed in renal samples, bone mineralization was among the enriched GO terms. This might be explained by the fact that many genes involved in renal mineral homeostasis play similar roles in bone formation (6, 38, 44). In addition, GO term analysis identified the upregulation of mitochondrial genes in mice fed the Mg2+-rich diet. This could be a response to an increased intracellular Mg2+ concentration in these mice. Intracellular Mg2+ is an important ATP-binding factor. Therefore, high Mg2+ concentrations in the cell could result in a low unbound ATP availability. Mitochondrial activation will lead to increased ATP production and thereby compensate for the Mg2+ binding.
In conclusion, gene expression microarray analysis on primary DCT cells allowed the identification of ample Mg2+-sensitive genes of the mouse genome. Particularly, SLC41A3, PCBD1, TBC1D4, and Umod are candidates that require further investigation. We suggest that patients suffering from hereditary hypomagnesemia of unknown cause be screened for mutations in these candidate genes. The COPAS sorting system provides an excellent mechanism to obtain primary DCT cells. Establishing COPAS-sorted DCT monolayers have been proven to be an accurate method to study thiazide-sensitive Na+ transport (26, 36). By the use of stable Mg2+ isotopes, such as 25Mg2+ and 26Mg2+, this primary DCT culture could provide a unique model to examine DCT-specific Mg2+ transport. Furthermore, crossbreeding of PV-eGFP mice with the available knockout mouse models, including TRPM6 and NCC knockout mice, further expands the possibilities of the established procedure. This will, in particular, further substantiate the physiological importance of the studied target molecule (6, 48).
This work was supported by The Netherlands Organization for Scientific Research Grants ZonMw 9120.8026 and NWO ALW 818.02.001, Vici Grant NWO-ZonMw 016.130.668 (to J. Hoenderop), and European Union Seventh Framework Programme Grant FP7/2007-2013 (agreement no. 305608).
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: J.H.d.B., F.C.H., R.J.B., and J.G.H. conception and design of research; J.H.d.B., M.J.G.K., M.L., F.v.Z., and H.M. performed experiments; J.H.d.B. and M.J.G.K. analyzed data; J.H.d.B., F.C.H., R.J.B., and J.G.H. interpreted results of experiments; J.H.d.B., R.J.B., and J.G.H. prepared figures; J.H.d.B. drafted manuscript; J.H.d.B., F.C.H., R.J.B., and J.G.H. edited and revised manuscript; J.H.d.B., R.J.B., and J.G.H. approved final version of manuscript.
The authors thank AnneMiete van der Kemp, Judy Lin, Maikel School, Anke Lameris, Ellen van Loon, and Fariza Bouallala for technical support. Furthermore, Dr. Hannah Monyer (University of Heidelberg) kindly donated the PV-eGFP mice. The authors are grateful to Dr. Nicolas Markadieu and Dr. Titia Woudenberg-Vrenken for optimizing the COPAS sorting system for PV-eGFP tubule sorting and to Dr. Sjoerd Verkaart for critical reading of the manuscript.
↵1 Supplemental Material for this article is available at the American Journal of Physiology-Renal Physiology website.
- Copyright © 2013 the American Physiological Society