Magnesium (Mg2+) is a crucial micronutrient present in foods such as nuts, grains, seeds and vegetables. The recommended daily intake of Mg2+ is 420 mg for men and 320 mg for women(1). Mg2+ is the second most abundant cation intracellularly where it is involved in many biological processes, including hundreds of enzymatic reactions, cell signalling and DNA/RNA synthesis(Reference de Baaij, Hoenderop and Bindels2). Mg2+ homeostasis therefore has to be tightly maintained. In the gastrointestinal tract, the majority of Mg2+ absorption occurs paracellularly in the small intestine, while in the colon and cecum, Mg2+ is absorbed via a transcellular pathway(Reference Schuchardt and Hahn3). Normally, 30–50 % of dietary Mg2+ is absorbed, but this can be increased to 80 % when intake is low(Reference Graham, Caesar and Burgen4). In the kidney, Mg2+ is filtered into the pro-urine, from which 95–99 % is reabsorbed in different segments of the nephron(Reference Houillier5). In total, 10–25 % of Mg2+ reabsorption occurs in the proximal tubule and 50–70 % in the thick ascending limb of Henle's loop (TAL), both via a paracellular pathway. The remaining 5–10 % is reabsorbed transcellularly in the distal convoluted tubule (DCT), where the final urinary Mg2+ concentration is determined as this is the last segment where Mg2+ can be reabsorbed.
When the intestinal and renal regulatory mechanisms aimed at maintaining sufficient Mg2+ levels are disturbed, hypomagnesemia (serum Mg2+ <0⋅7 mm) can develop. Consistent with the widespread functions of Mg2+, severe symptoms such as epilepsy, fatigue, muscle cramps and cardiac arrhythmias have been observed as a consequence of hypomagnesemia(Reference de Baaij, Hoenderop and Bindels2). Various factors can cause hypomagnesemia. Low dietary Mg2+ intake is a growing problem, caused by unhealthy diets and a decreased Mg2+ content in soil and foods(Reference Cazzola, Della Porta and Manoni6). Other causes of hypomagnesemia are alcoholism, mutations in genes involved in Mg2+ regulation or certain drug treatments(Reference de Baaij, Hoenderop and Bindels2). Genetic and drug-induced Mg2+ deficiencies in particular have contributed greatly to our understanding of Mg2+ homeostasis. Since the number of hypomagnesemia-causing mutations and drugs is steadily growing, the underlying mechanisms are becoming more elaborate and complex. Studying how these mechanisms are connected can help with finding the most important contributors to Mg2+ maintenance. Therefore, the aim of this review is to provide an up-to-date overview of the molecular causes of genetic and drug-induced Mg2+ deficiencies and identify the main pathways of Mg2+ regulation shared between them.
Genetic causes of hypomagnesemia
The development of novel sequencing techniques has resulted in the identification of various genes that play a role in Mg2+ homeostasis (Table 1). In the following segment, the genetic Mg2+ deficiencies are subdivided into groups that share a common mechanism.
TRPM6/7, transient receptor potential melastatin type 6/7; EGF(R), epidermal growth factor (receptor); TAL, thick ascending limb of Henle's loop; CaSR, calcium-sensing receptor; DCT, distal convoluted tubule; NCC, Na+, Cl− co-transporter; Kir5⋅1, K+ inwardly rectifying channel 5⋅1; ClC-Kb, Cl− channel Kb; HNF1β, hepatocyte nuclear factor 1β; PCBD1, pterin-4α-carbinolamine dehydratase; POLG1, DNA polymerase subunit gamma; FAM111A, family with sequence similarity 111 member A; PLVAP, plasmalemma vesicle-associated protein.
Luminal Mg2+ uptake in the intestine and distal convoluted tubule
Transient receptor potential melastatin type 6 (TRPM6) is a divalent cation transporter with a high affinity for Mg2+ that is specifically expressed in the luminal/apical membranes of the colon and DCT(Reference Voets, Nilius and Hoefs7). Mutations in the TRPM6 gene cause severe hypomagnesemia with secondary hypocalcemia, which occurs because of hypomagnesemia-induced parathyroid failure(Reference Walder, Landau and Meyer8,Reference Schlingmann, Weber and Peters9) . There is some debate on whether hypomagnesemia is caused by intestinal Mg2+ malabsorption or renal Mg2+ wasting. Kidney-specific TRPM6 knockout mice display normal Mg2+ levels(Reference Chubanov, Ferioli and Wisnowsky10). However, in patients with TRPM6 mutations, intestinal malabsorption, as well as increased urinary Mg2+ excretion, is observed, suggesting both intestinal and renal mechanisms are involved(Reference Schlingmann, Sassen and Weber11). TRPM6 is able to form a heteromeric complex with its family member TRPM7, which is essential for TRPM6 to facilitate Mg2+ transport(Reference Chubanov, Ferioli and Wisnowsky10,Reference Li, Jiang and Yue12) . In contrast to TRPM6, TRPM7 is ubiquitously expressed and is able to transport Mg2+ and other divalent cations by itself(Reference Monteilh-Zoller, Hermosura and Nadler13). Consequently, TRPM7 deficiency usually decreases intracellular Mg2+ content(Reference Chubanov, Ferioli and Wisnowsky10,Reference Deason-Towne, Perraud and Schmitz14) . In TRPM6-expressing colon cells, however, TRPM7 down-regulation actually increased Mg2+ influx(Reference Luongo, Pietropaolo and Gautier15). This may be explained by the fact that in these cells, down-regulation of TRPM7 increases the relative abundance of TRPM6/7 heteromers compared to TRPM7 monomers. TRPM6/7 heteromers show higher Mg2+ currents than TRPM7 monomers and are less sensitive to inhibition by the intracellular Mg2+ concentration(Reference Ferioli, Zierler and Zaisserer16). Thus, TRPM7 by itself provides an adaptable Mg2+ transport throughout the body, while TRPM6/7 complexes facilitate a high and constitutive Mg2+ uptake in the intestine and DCT that is needed to maintain sufficient Mg2+ levels. Because of the ubiquitous, TRPM6-independent role of TRPM7 in the maintenance of the cellular cation balance, it remains to be seen whether TRPM7 mutations can cause a phenotype similar to TRPM6 mutations.
Epidermal growth factor (EGF) is an essential regulator of TRPM6 activity(Reference Thebault, Alexander and Tiel Groenestege17). EGF and EGF receptor (EGFR) mutations cause hypomagnesemia along with either mental retardation (EGF) or severe epithelial inflammation (EGFR)(Reference Groenestege, Thebault and van der Wijst18,Reference Campbell, Morton and Takeichi19) . Within the kidney, EGF is predominantly expressed in the DCT(Reference Groenestege, Thebault and van der Wijst18). In human embryonic kidney 293 cells transiently transfected with TRPM6, EGF dose-dependently increased TRPM6 activity(Reference Groenestege, Thebault and van der Wijst18). This effect is mediated by EGFR-activated Src family kinases, which in turn activate the PI3K/Akt pathway(Reference Thebault, Alexander and Tiel Groenestege17). Upon activation of this pathway, the downstream GTPase Rac1 increases the surface expression of TRPM6 (Fig. 1a). Recent studies have demonstrated that the EGFR directly binds to TRPM7 in the vasculature to regulate Mg2+ uptake(Reference Zou, Rios and Neves20). Whether this mechanism also contributes to Mg2+ uptake in the DCT and colon is unknown, though this is unlikely considering the localisations of TRPM6/7 and EGFR in these tissues are apical and basolateral, respectively. As EGF signalling is classically studied in cell differentiation and organ development(Reference Normanno, De Luca and Bianco21), the biological function of this regulatory pathway is unclear. If EGF would act as an Mg2+-regulating hormone, its release should depend on Mg2+ availability similar to parathyroid hormone for Ca2+ homeostasis(Reference Conigrave22). In mice fed an Mg2+-deficient diet, EGF up-regulation has been reported specifically in DCT cells(Reference de Baaij, Groot Koerkamp and Lavrijsen23), indicating Mg2+-dependent EGF transcription may occur locally in the DCT. Alternatively, EGF-induced Mg2+ uptake may contribute to the growth factor function of EGF, as cell growth requires high intracellular Mg2+ concentrations for transcription and translation. Several studies indeed report up-regulation of EGF and TRPM7 in cancer to facilitate rapid Mg2+ uptake and cell growth(Reference Sahni, Tamura and Sweet24–Reference Rybarczyk, Gautier and Hague27).
Paracellular Mg2+ reabsorption in the thick ascending limb of Henle's loop
The TAL is responsible for the majority of Mg2+ reabsorption via a passive paracellular pathway(Reference Di Stefano, Roinel and de Rouffignac28). This paracellular transport is enabled by cation-selective pores in tight junction complexes formed by claudin-16 and claudin-19, encoded by CLDN16 and CLDN19 (Reference Hou, Renigunta and Konrad29). By disrupting this complex, mutations in CLDN16 and CLDN19 cause familial hypomagnesemia, hypercalciuria and nephrocalcinosis with ocular abnormalities in the case of CLDN19 (Reference Simon, Lu and Choate30,Reference Konrad, Schaller and Seelow31) . Claudin-16 also interacts with claudin-14, which impairs the cation permeability of the tight junction complex and thus serves as a negative regulator of the claudin-16/claudin-19 channel (Fig. 1b)(Reference Gong, Renigunta and Himmerkus32).
The calcium-sensing receptor (CaSR) is an important regulator of claudin-mediated paracellular reabsorption. In the kidney, the CaSR is highly abundant on the basolateral membrane of the TAL(Reference Chattopadhyay, Baum and Bai33,Reference Riccardi, Hall and Chattopadhyay34) . The protein promotes Ca2+ excretion when Ca2+ concentrations in the blood are high(Reference Alfadda, Saleh and Houillier35), by inhibiting claudin-16 and promoting claudin-14 expression, thereby limiting paracellular transport in the TAL(Reference Gong, Renigunta and Himmerkus32,Reference Toka, Al-Romaih and Koshy36) . Moreover, CaSR activation inhibits the renal outer medullary potassium channel and the Na+, K+ and 2Cl− co-transporter (NKCC2) on the apical membrane of the TAL(Reference Wang, Lu and Hebert37). Renal outer medullary potassium channel and NKCC2 play an important role in generating a lumen-positive voltage which allows the paracellular transport of cations(Reference Gamba and Friedman38). By inhibiting renal outer medullary potassium channel and NKCC2 and activating claudin-14, gain-of-function mutations in CASR lead to autosomal-dominant hypocalcaemia and hypomagnesemia(Reference Pearce, Williamson and Kifor39,Reference Watanabe, Fukumoto and Chang40) . Not all autosomal-dominant hypocalcaemia patients develop hypomagnesemia, however, as this is dependent on the level of activity of the mutant CaSR(Reference Kinoshita, Hori and Taguchi41). This indicates that relatively mild wasting of Mg2+ in the TAL will not cause Mg2+ deficiency, probably because this can be compensated by an increased Mg2+ reabsorption in the DCT. Indeed, inhibition of ion transport in the TAL leads to the proliferation of the DCT and an increased expression of TRPM6(Reference Kaissling, Bachmann and Kriz42,Reference van Angelen, van der Kemp and Hoenderop43) .
Structural integrity of the distal convoluted tubule
Mutations in SLC12A3, encoding the thiazide-sensitive Na+ and Cl− co-transporter (NCC), cause Gitelman syndrome, one of the most common inherited renal disorders(Reference Knoers and Levtchenko44). Although NCC transports Na+ and Cl−, hypomagnesemia and hypokalaemia are the main symptoms of Gitelman syndrome. In a recent review, the link between NCC activity and Mg2+ transport was extensively discussed(Reference Franken, Adella and Bindels45). The most commonly accepted hypothesis is that NCC deficiency causes atrophy of the DCT segment (Fig. 1c)(Reference Loffing, Vallon and Loffing-Cueni46). This is in line with a reduction in the expression of the DCT marker parvalbumin and TRPM6 upon inactivation of NCC in mice(Reference Loffing, Vallon and Loffing-Cueni46,Reference Nijenhuis, Vallon and van der Kemp47) . The hypothesis of DCT atrophy is supported by other studies showing that the DCT has a high degree of plasticity and can remodel depending on the situation(Reference Kaissling, Bachmann and Kriz42,Reference Loffing, Le Hir and Kaissling48–Reference Grimm, Coleman and Delpire50) . Recently, the transcription factor AP-2β and its downstream target potassium channel tetramerisation domain containing 1 (KCTD1) were identified as regulators of DCT development(Reference Marneros51). Similar to the DCT atrophy observed in NCC knockout mice, impaired development of the DCT as a consequence of KCTD1 deficiency also leads to hypomagnesemia(Reference Marneros52). Clearly, proper development of DCT structure and function under the influence of genes such as SLC12A3 and KCTD1 is crucial for Mg2+ homeostasis. Although no hypomagnesemia-causing variants are known in the KCTD1 gene, it could be an interesting candidate to consider in the analysis of unsolved cases.
Na+-dependent extrusion of Mg2+ driven by the Na+/K+-ATPase
Although the exact molecular identity of the basolateral Mg2+ extrusion protein in the DCT is under debate, it is generally accepted that Mg2+ extrusion is Na+-dependent(Reference Romani53). This notion is further supported by the identification of hypomagnesemia-causing mutations in two genes encoding subunits of the Na+/K+-ATPase, ATP1A1 and FXYD2 (Reference Schlingmann, Bandulik and Mammen54,Reference de Baaij, Dorresteijn and Hennekam55) . The Na+/K+-ATPase consists of an α-, a β- and a regulatory FXYD/γ-subunit and crucially maintains favourable electrochemical gradients in all cells of the body by exchanging three Na+ ions for two K+ ions using ATP(Reference Jorgensen, Hakansson and Karlish56). Interestingly, hypomagnesemia is associated with seizures and intellectual disability in the case of ATP1A1 (Reference Schlingmann, Bandulik and Mammen54). However, it is unclear whether this is a secondary effect of hypomagnesemia, or indicates an intrinsic function of ATP1A1 in the brain. Based on animal models, the most abundant expression of ATP1A1 and FXYD2 proteins within the kidney is in the DCT(Reference El Mernissi and Doucet57,Reference Wetzel and Sweadner58) . Consequently, two mechanisms can be proposed for the hypomagnesemia in patients with ATP1A1 and FXYD2 mutations. First, the Na+/K+-ATPase may generate a favourable electrochemical gradient for transcellular Mg2+ transport in the DCT. As such, its function is essential for TRPM6-mediated Mg2+ entry into the cell(Reference Schlingmann, Bandulik and Mammen54). Secondly, the basolateral Na+ gradients may be essential for the function of the Na+/Mg2+-exchanger, which is driven by extracellular Na+ (Fig. 1d)(Reference Romani53). The relative contribution of both pathways to the development of hypomagnesemia is unclear. Interestingly, ATP1A1 and FXYD2 patients have a relatively mild Na+-wasting phenotype and often do not display hypokalaemia or metabolic alkalosis(Reference Schlingmann, Bandulik and Mammen54,Reference de Baaij, Dorresteijn and Hennekam55) . If Na+ reabsorption in the DCT is decreased, downstream segments will compensate for this by reabsorbing Na+ at the expense of K+ and H+ excretion, resulting in hypokalaemia and metabolic alkalosis. Since these phenotypes are not present, it seems that sufficient Na+/K+-ATPase function remains to maintain NCC-mediated Na+ reabsorption in the DCT. This suggests that Mg2+ reabsorption is more sensitive to disturbed Na+/K+-ATPase function than Na+.
In line with reduced Na+/K+-ATPase function, mutations in KCNJ10 cause EAST/SeSAME syndrome, characterised by epilepsy, ataxia, sensorineural deafness and a tubulopathy(Reference Bockenhauer, Feather and Stanescu59,Reference Scholl, Choi and Liu60) . Patients suffer from a Gitelman syndrome-like electrolyte phenotype including hypomagnesemia, hypokalaemia and metabolic alkalosis. KCNJ10 encodes the K+ inwardly rectifying channel 4⋅1 (Kir4⋅1), which together with its binding partner Kir5⋅1 (encoded by KCNJ16) forms the major K+ channel in the basolateral membrane of the TAL and DCT(Reference Lourdel, Paulais and Cluzeaud61,Reference Zhang, Wang and Su62) . These Kir4⋅1/Kir5⋅1 channels provide the driving force for Na+/K+-ATPase by recycling K+ at the basolateral membrane (Fig. 1d). Mutations in Kir4⋅1 will therefore limit the functionality of Na+/K+-ATPase, explaining the hypomagnesemia(Reference Bandulik, Schmidt and Bockenhauer63). Moreover, uncoupling of this ‘pump-leak mechanism’ at the basolateral membrane will result in plasma membrane depolarisation. As a result, the Cl− extrusion via the kidney-specific basolateral Cl− channel Kb will be decreased, leading to an increased intracellular Cl− concentration(Reference Zhang, Wang and Zhang64). As Cl− inhibits the NCC-activating with-no-lysine kinases(Reference Piala, Moon and Akella65,Reference Bazua-Valenti, Chavez-Canales and Rojas-Vega66) , NCC-mediated Na+ reabsorption will be decreased. The downstream compensatory mechanism of Na+ reabsorption at the expense of K+ and H+ thus explains the hypokalaemia and metabolic alkalosis. In accordance with this mechanism, mutations in CLCNKB, which encodes for Cl− channel Kb, are associated with a similar phenotype of Na+ and K+ wasting and metabolic alkalosis(Reference Jeck, Konrad and Peters67). Hypomagnesemia can also be observed in these patients, which is most likely due to the link between impaired NCC functioning and Mg2+ reabsorption.
Expression of FXYD2 is activated by the transcription factor hepatocyte nuclear factor 1β (HNF1β) and its coactivator pterin-4α-carbinolamine dehydratase (PCBD1)(Reference Adalat, Woolf and Johnstone68,Reference Ferre, de Baaij and Ferreira69) . Mutations in both HNF1B and PCBD1 are associated with hypomagnesemia(Reference Adalat, Woolf and Johnstone68–Reference van der Made, Hoorn and de la Faille70). These mutations abolish the HNF1β/PCBD1-induced transcription of FXYD2, confirming that FXYD2 plays an important role in Mg2+ homeostasis(Reference Ferre, de Baaij and Ferreira69,Reference Ferre, Veenstra and Bouwmeester71) . Importantly, HNF1β also regulates KCNJ16 transcription(Reference Kompatscher, de Baaij and Aboudehen72). Down-regulation of HNF1β indeed reduces the expression of Kir5⋅1 as well as Kir4⋅1 and NCC(Reference Kompatscher, de Baaij and Aboudehen72). This indicates that HNF1B mutations affect the Na+/K+-ATPase directly via FXYD2, but also less directly via Kir4⋅1/Kir5⋅1. Furthermore, it should be noted that HNF1β has many other target genes. Therefore, it cannot be excluded that additional targets play a role in disturbed Mg2+ homeostasis associated with HNF1B mutations. In line with this broader function, HNF1B mutations typically lead to widespread abnormal renal development of which hypomagnesemia is only one of the manifestations.
Several mutations in mitochondrial DNA or in genes encoding mitochondrial proteins have been associated with hypomagnesemia(Reference Harvey and Barnett73–Reference Belostotsky, Ben-Shalom and Rinat76). Although the mechanisms of these disorders have never been examined, the essential role of the Na+/K+-ATPase in Mg2+ transport may partially explain the phenotype. Since the energy demand of Na+/K+-ATPase is particularly high in the DCT, optimal ATP production by the mitochondria is essential to fuel its activity. Indeed, mitochondrial density is very high in the DCT(Reference Dorup77). However, as mitochondrial mutations often lead to variable phenotypes due to differences in affected tissues and heteroplasmy levels, it is particularly challenging to unravel how Mg2+ homeostasis is affected in these patients. Therefore, it should be noted that additional pathways may be involved.
Others
Not all genes in which hypomagnesemia-causing mutations have been found can already be placed in major Mg2+ regulatory pathways as described earlier. Mutations in the cyclin M2 gene CNNM2 cause hypomagnesemia, seizures, intellectual disability and obesity(Reference Stuiver, Lainez and Will78–Reference Franken, Muller and Mignot81). Therefore, CNNM2 likely plays a role in brain development and metabolism as well as Mg2+ balance(Reference Arjona, de Baaij and Schlingmann79,Reference Franken, Muller and Mignot81) . There is a debate on whether CNNM2 and its family members CNNM1, CNNM3 and CNNM4 are the putative Na+/Mg2+-exchangers responsible for Mg2+ extrusion(Reference Funato, Furutani and Kurachi82,Reference Arjona and de Baaij83) . The idea is supported by the fact that CNNM2 is highly expressed in the DCT, localises to the basolateral membrane, and stimulates Mg2+ efflux and Na+ influx when overexpressed(Reference Stuiver, Lainez and Will78,Reference Funato, Furutani and Kurachi82,Reference Hirata, Funato and Takano84) . Conversely, CNNM2-induced Mg2+ transport may be bi-directional, is not always shown to be Na+ dependent and is abolished by TRPM7 inhibition, indicating CNNM2 itself is not a transporter(Reference Stuiver, Lainez and Will78,Reference Arjona, de Baaij and Schlingmann79) . An explanation that seems to fit all observations is that CNNM2 has a regulatory rather than a transporting function, though its exact function and the molecular identity of the Na+/Mg2+-exchanger remains to be determined.
Other genes with an unsolved function in Mg2+ homeostasis include FAM111A, in which specific mutations cause hypomagnesemia along with bone and eye abnormalities and hypoparathyroidism(Reference Isojima, Doi and Mitsui85,Reference Nikkel, Ahmed and Smith86) . Its known functions in antiviral restriction and DNA replication cannot be clearly linked to Mg2+, so efforts to unravel this are ongoing. Mutations in GATA3, encoding a transcription factor, typically lead to hypoparathyroidism, deafness and renal anomalies, but hypomagnesemia has also been reported(Reference Al-Shibli, Al Attrach and Willems87). It is unknown whether this Mg2+ deficiency is a sporadically occurring secondary effect of the syndrome or if GATA3 regulates the expression of genes that are of interest for Mg2+ transport. Lastly, Mg2+ deficiency has also been observed in a patient with protein-losing enteropathy as a consequence of mutations in the plasmalemma vesicle-associated protein gene, which is expressed in the endothelium(Reference Broekaert, Becker and Gottschalk88). The hypomagnesemia may occur as a consequence of malabsorption due to the enteropathy, but renal abnormalities are also observed, so it is as of yet unknown how and where this endothelial protein affects Mg2+. Additional research is needed to decipher the molecular mechanism underlying hypomagnesemia in these syndromes.
Drug-induced hypomagnesemia
Hypomagnesemia is a side effect of various drugs (Table 2). Interestingly, hypomagnesemia-causing drugs and hereditary causes of hypomagnesemia share common pathways, confirming the essential role of these pathways for Mg2+ (re)absorption in the intestine and kidney. In this section of the review, hypomagnesemia-inducing drugs are described in the context of the pathways described earlier, which provides various additional insights into these pathways.
TRPM6/7, transient receptor potential melastatin type 6/7; CsA, cyclosporine A; TAL, thick ascending limb of Henle's loop; DCT, distal convoluted tubule.
* Incidences of hypomagnesemia are often not studied in large populations. The frequencies are based on the provided references.
Luminal Mg2+ uptake in the intestine and distal convoluted tubule
Somatic gain-of-function mutations in EGFR are associated with various cancers, which has led to the development of EGFR inhibitors for the treatment of lung, head and neck, colorectal and pancreas cancer(Reference Chong and Janne89). The treatment options are either monoclonal antibodies, which bind EGFR extracellularly, or tyrosine kinase inhibitors (TKI), which bind intracellularly and prevent phosphorylation of the receptor(Reference Imai and Takaoka90). In line with the role of EGF signalling in TRPM6 activation, hypomagnesemia is a common side effect of treatment with the EGFR antibody cetuximab, occurring in 20–30 % of the patients(Reference Groenestege, Thebault and van der Wijst18,Reference Fakih, Wilding and Lombardo91,Reference Tejpar, Piessevaux and Claes92) . Cetuximab prevents the EGF-induced activation of TRPM6 in kidney and colon cells, indicating the hypomagnesemia can be attributed to a decrease in both renal and intestinal (re)absorption(Reference Groenestege, Thebault and van der Wijst18,Reference Pietropaolo, Pugliese and Armuzzi93) . Hypomagnesemia is not reported in patients treated with EGFR TKI(Reference Izzedine and Perazella94). However, EGFR TKI are less specific than EGFR antibodies and their anti-tumour effects often do not solely rely on the inhibition of EGFR, but also complementary pathways(Reference Imai and Takaoka90). The specific inhibition of EGF signalling may therefore not be sufficient to induce hypomagnesemia in TKI treatment. Modest decreases in Mg2+ levels have been observed in animal models, however, indicating hypomagnesemia should not be completely ruled out in patients treated with EGFR TKI(Reference Dimke, van der Wijst and Alexander95,Reference Mak, Kramer and Chmielinska96) .
Another class of drugs that causes hypomagnesemia through interference with TRPM6 function are calcineurin inhibitors such as cyclosporine A (CsA) and tacrolimus, which are immunosuppressants used in transplant recipients. Hypomagnesemia is almost always reported in cohorts of calcineurin inhibitor-treated patients, with incidences ranging from 10 to 50 % for CsA and from 50 up to 98 % for tacrolimus(Reference June, Thompson and Kennedy97–Reference Navaneethan, Sankarasubbaiyan and Gross101). CsA and tacrolimus were found to reduce the expression of TRPM6(Reference Ikari, Okude and Sawada102,Reference Gratreak, Swanson and Lazelle103) . This effect may be mediated by the transcription factor c-Fos, as c-Fos is down-regulated by CsA and inhibition of c-Fos decreases both TRPM6 expression and Mg2+ uptake(Reference Ikari, Okude and Sawada102). Moreover, renal EGF production is decreased in CsA-treated patients, indicating TRPM6 down-regulation may also occur through lowered EGF signalling in the DCT(Reference Ledeganck, De Winter and Van den Driessche104).
Proton pump inhibitors (PPI) are used for gastric-acid-related diseases such as peptic ulcers and are among the most-used drugs worldwide(Reference Malfertheiner, Kandulski and Venerito105). Hypomagnesemia is a side effect that occurs in 5–15 % of PPI users and can become very severe, particularly after long-term treatment(Reference Epstein, McGrath and Law106–Reference Kieboom, Kiefte-de Jong and Eijgelsheim108). Oral supplementation of Mg2+ is often insufficient to fully restore normal Mg2+ levels in patients with PPI-induced hypomagnesemia(Reference Toh, Ong and Wilson109). Moreover, a reduced Mg2+ excretion is observed in these patients, indicating the kidney still functions normally and partially counteracts the hypomagnesemia by increasing Mg2+ reabsorption(Reference William, Nelson and Hayman110). Therefore, it is generally accepted that PPI reduce the intestinal absorption of Mg2+. Since a mild decrease in Mg2+ uptake in the intestine alone can usually be compensated, the risk of PPI-induced hypomagnesemia is strongest when additional factors contributing to Mg2+ depletion are present, such as the use of calcineurin inhibitors and diuretics, single nucleotide polymorphisms in TRPM6 or increasing age(Reference Kieboom, Kiefte-de Jong and Eijgelsheim108,Reference Zipursky, Macdonald and Hollands111–Reference Danziger, William and Scott114) . PPI increase the pH of the gastrointestinal tract by inhibiting the release of gastric acid, which may explain the malabsorption since Mg2+ is more soluble and TRPM6 and TRPM7 show higher activity at lower pH values(Reference Li, Jiang and Yue12). In a series of N-of-1 trials, lowering the luminal pH by supplementation of the naturally occurring polysaccharide inulin indeed significantly improved the PPI-induced hypomagnesemia(Reference Hess, de Baaij and Broekman115). Since the decrease in pH after inulin treatment occurs because of fermentation by colonic bacteria(Reference Petry, Egli and Chassard116), the microbiota might play an important role in Mg2+ (mal)absorption. Indeed, PPI alter the gut microbiota of patients(Reference Tsuda, Suda and Morita117–Reference Takagi, Naito and Inoue119) and a PPI-induced decrease in microbiota diversity is associated with hypomagnesemia in mice(Reference Gommers, Ederveen and van der Wijst120). These data highlight the importance of the microbiota in generating a favourable environment for Mg2+ absorption in the intestine (Fig. 1a).
Paracellular Mg2+ reabsorption in the thick ascending limb of Henle's loop
Loop diuretics are a class of drugs that lower the blood pressure by inhibiting Na+ reabsorption in the TAL, thereby promoting diuresis. The use of loop diuretics is associated with increased Mg2+ excretion(Reference Duarte121–Reference Leier, Dei Cas and Metra124). Loop diuretics inhibit NKCC2 and therefore interfere with the lumen-positive voltage required for paracellular cation uptake in the TAL(Reference Gamba and Friedman38). This inhibition particularly affects the paracellular transport of divalent cations such as Mg2+(Reference Quamme125). Despite the decreased Mg2+ reabsorption in the TAL, some studies report no association between loop diuretics and hypomagnesemia while others only observe it in patients already prone to Mg2+ depletion due to heart failure or PPI use(Reference Kieboom, Kiefte-de Jong and Eijgelsheim108,Reference Wester123,Reference Leier, Dei Cas and Metra124,Reference Kieboom, Zietse and Ikram126) . This indicates some additional risk factor has to be present before Mg2+ deficiency develops. Similar to the Mg2+ levels in patients with CASR mutations, whether hypomagnesemia develops likely also depends on the capacity of the DCT to compensate for the Mg2+ loss in the TAL(Reference van Angelen, van der Kemp and Hoenderop43).
Additionally, hypomagnesemia is observed in about 30 % of patients treated with aminoglycosides, a class of antibiotics used for tuberculosis and other bacterial infections(Reference Zaloga, Chernow and Pock127,Reference von Vigier, Truttmann and Zindler-Schmocker128) . Moreover, the mTOR inhibitor sirolimus (rapamycin), which is used for transplant recipients as an alternative to calcineurin inhibitors, leads to hypomagnesemia in about 10 % of cases(Reference Andoh, Burdmann and Fransechini129,Reference Van Laecke, Van Biesen and Verbeke130) . Aminoglycosides and sirolimus decrease the expression of NKCC2, suggesting mechanistic similarities to loop diuretic-induced Mg2+ loss(Reference Sassen, Kim and Kwon131,Reference da Silva, de Braganca and Shimizu132) . In addition, aminoglycosides are able to activate CaSR(Reference Quinn, Ye and Diaz133), which inhibits paracellular uptake of Ca2+ and Mg2+. By affecting these two pathways simultaneously, it may be expected that aminoglycosides cause a rather severe wasting of Mg2+ in the TAL, which would explain the relatively high incidence compared to the loop diuretic- and sirolimus-induced hypomagnesemia.
Structural integrity of the distal convoluted tubule
Thiazide diuretics are among the most commonly used antihypertensive drugs and long-term use frequently leads to increased Mg2+ excretion, which is associated with a 2- to 3-fold increased risk of developing hypomagnesemia(Reference Kieboom, Zietse and Ikram126,Reference Hollifield134,Reference Ryan135) . Thiazides promote diuresis by inhibiting NCC. Therefore, it can be expected that thiazide-induced Mg2+ loss shares many mechanistic similarities to the hypomagnesemia observed in Gitelman syndrome. Indeed, the DCT atrophy associated with Gitelman syndrome is also seen in rats treated with thiazide(Reference Nijenhuis, Hoenderop and Loffing136,Reference Loffing, Loffing-Cueni and Hegyi137) . However, other studies contradict this finding and report no deleterious effects on the DCT length in thiazide-treated mice(Reference Nijenhuis, Vallon and van der Kemp47). This inconsistency may be explained by interspecies variation and/or differences in dosages. Interestingly, TRPM6 down-regulation still occurred in these thiazide-treated mice, indicating NCC regulates TRPM6 not just through DCT development but also via a more direct mechanism(Reference Franken, Adella and Bindels45). If a direct link between NCC and TRPM6 indeed exists, it still remains to be determined whether this mechanism or structural changes to the DCT underlie thiazide-induced hypomagnesemia, or whether it is a combination of the two.
Various drugs have nephrotoxic effects, which can lead to damage to the DCT and consequently disturbances in Mg2+ reabsorption. The majority of patients treated with the chemotherapeutic cisplatin, for example, develop hypomagnesemia if no measures are taken to prevent this(Reference Lam and Adelstein138–Reference Markmann, Rothman and Reichman140). Cisplatin accumulates in the kidney and causes nephrotoxicity that mainly affects tubular structures such as the DCT, resulting in disturbed reabsorption (Fig. 1c)(Reference Pabla and Dong141). Because of this widespread effect on the tubules, other electrolytes are also disturbed upon cisplatin treatment(Reference Oronsky, Caroen and Oronsky142). Similarly, the hypomagnesemia-inducing anti-fungal agent amphotericin B(Reference Barton, Pahl and Vaziri143,Reference Marcus and Garty144) , as well as calcineurin inhibitors and aminoglycosides, can also cause nephrotoxicity(Reference Harbarth, Pestotnik and Lloyd145–Reference Mingeot-Leclercq and Tulkens147). In the case of calcineurin inhibitors and aminoglycosides, nephrotoxicity could thus be an additional explanation for the hypomagnesemia in addition to their respective effects on TRPM6 and NKCC2/CaSR. It should be noted that even though the DCT is prone to structural remodelling, the nephrotoxicity-induced changes affect multiple nephron segments, indicating the hypomagnesemia is not solely due to DCT damage.
Na+-dependent extrusion of Mg2+ driven by the Na+/K+-ATPase
Hypomagnesemia is observed in 10–20 % of patients treated with digoxin, which increases renal Mg2+ excretion(Reference Whang, Oei and Watanabe148–Reference Abu-Amer, Priel and Karlish150). Digoxin inhibits the Na+/K+-ATPase and is used to treat arrhythmias and heart failure, since an increase in Na+ in the cell leads to more intracellular Ca2+ and an increased contraction force. The digoxin-induced hypomagnesemia is in line with the importance of the Na+/K+-ATPase in basolateral Mg2+ extrusion in the kidney. Despite the ubiquitous function of the Na+/K+-ATPase in electrolyte transport, Mg2+ depletion upon digoxin treatment seems to be more frequent than disturbances of other electrolytes(Reference Young, Goh and McKillop149,Reference Abu-Amer, Priel and Karlish150) , again suggesting that Mg2+ homeostasis is relatively sensitive to decreased functioning of the Na+/K+-ATPase.
Others
β-adrenergic agonists activate the sympathetic nervous system and are frequently used in asthma patients to relax the airway muscles. Hypomagnesemia is observed in about 30 % of asthmatic patients and the use of β-adrenergic agonists contributes to this(Reference Khilnani, Parchani and Toshniwal151–Reference Das, Haldar and Ghosh153). In this case, the mechanism is unrelated to the main pathways of Mg2+ uptake. β-adrenergic agonists activate lipolysis, which results in an increased production of NEFA from triacylglycerols(Reference Kuppusamy and Das154–Reference Hom, Forrest and Bach156). NEFA are able to bind Mg2+ and thereby decrease free Mg2+ in the blood, which could be interpreted as hypomagnesemia(Reference Kurstjens, de Baaij and Overmars-Bos157). Indeed, the decrease in serum Mg2+ after treatment with β-adrenergic agonists coincides with an increase in NEFA levels(Reference Bremme, Eneroth and Nordstrom158). This interaction between Mg2+ and NEFA is an important additional determinant of Mg2+ homeostasis that should not only be considered in β-adrenergic agonist treatment. For example, high levels of triacylglycerols, which correlates to high levels of NEFA, is also associated with hypomagnesemia in patients with type 2 diabetes(Reference Kurstjens, de Baaij and Bouras159). Since NEFA affect free Mg2+ concentrations rather than the total Mg2+ content, it should be determined whether this has the same clinical effects as actual Mg2+ depletion.
Conclusions
Studies on hereditary causes of hypomagnesemia as well as hypomagnesemia-inducing drugs have been instrumental in the elucidation of the mechanisms involved in intestinal and renal Mg2+ (re)absorption (Fig. 1). Using this knowledge, novel treatment strategies and therapeutics can be developed to target these mechanisms and tackle the increasing problem of Mg2+ deficiency. For example, it has now become clear that serum Mg2+ levels can be increased not only by sufficient Mg2+ intake but also by healthy diets that sustain the gut microbiota and limit NEFA levels. A promising application of this is the use of dietary fibres that stimulate the microbiota and improve serum Mg2+ concentrations(Reference Hess, de Baaij and Broekman115). Development of additional treatment strategies is essential for patients with rare hereditary causes of hypomagnesemia and large patient groups dependent on hypomagnesemia-causing drugs.
Acknowledgements
We thank Heidi Schigt for her support with the preparation of the figure and prof. dr. Ailsa Welch for the kind invitation to speak at the Royal Society of Nutrition.
Financial Support
This work was financially supported by the IMAGEN project which is co-funded by the PPP Allowance made available by Health~Holland, Top Sector Life Sciences & Health, to stimulate public–private partnerships (IMplementation of Advancements in GENetic Kidney Disease, LSHM20009) and the Dutch Kidney Foundation (20OP + 018). Additionally, we received support from ZonMW under the frame of EJPRD, the European Joint Programme on Rare Diseases (EJPRD2019-40). This project has received funding from the European Union's Horizon 2020 research and innovation programme under the EJP RD COFUND-EJP No 825575 and the Netherlands Organization for Scientific Research (NWO Veni 016⋅186⋅012. Vici 016⋅130⋅668).
Conflict of Interest
None.
Authorship
The authors had sole responsibility for all aspects of preparation of this paper.