Substitution p.A350V in Na+/Mg2+ Exchanger SLC41A1, Potentially Associated with Parkinson's Disease, Is a Gain-of-Function Mutation

Parkinson's disease (PD) is a complex multifactorial ailment predetermined by the interplay of various environmental and genetic factors. Systemic and intracellular magnesium (Mg) deficiency has long been suspected to contribute to the development and progress of PD and other neurodegenerative diseases. However, the molecular background is unknown. Interestingly, gene SLC41A1 located in the novel PD locus PARK16 has recently been identified as being a Na+/Mg2+ exchanger (NME, Mg2+ efflux system), a key component of cellular magnesium homeostasis. Here, we demonstrate that the substitution p.A350V potentially associated with PD is a gain-of-function mutation that enhances a core function of SLC41A1, namely Na+-dependent Mg2+ efflux by 69±10% under our experimental conditions (10-minute incubation in high-Na+ (145 mM) and completely Mg2+-free medium). The increased efflux capacity is accompanied by an insensitivity of mutant NME to cAMP stimulation suggesting disturbed hormonal regulation and leads to a reduced proliferation rate in p.A350V compared with wt cells. We hypothesize that enhanced Mg2+-efflux conducted by SLC41A1 variant p.A350V might result, in the long-term, in chronic intracellular Mg2+-deficiency, a condition that is found in various brain regions of PD patients and that exacerbates processes triggering neuronal damage.


Introduction
The crucial role of magnesium (Mg) in normal cellular physiology has been described in many reports. Thus, unsurprisingly, Mg 2+ deficiency and/or changed intracellular Mg homeostasis (IMH) has been associated with a multitude of serious ailments among them neurodegenerative, neurological, and psychiatric disorders such as Alzheimer's disease (AD) [1], stroke [2], aggressive behavior [3], increased stress sensitivity [4], and hyperactivity [5]. In particular, several studies have suggested an association between a disturbed IMH and the incidence of Parkinsoǹs disease (PD) [6,7,8], a chronic, progressive, neurodegenerative disorder of the motor system mainly characterized by the degeneration of neurons in the substantia nigra pars compacta and the formation of Lewy bodies [9,10]. PD is estimated to affect ,1% of people over 60 years of age or ,0.3% of the entire population in industrialized countries [11,12]. Oyanagi and colleagues [13,14] have shown that continuous low Mg intake for two generations induces exclusive loss of dopaminergic neurons in rats. In agreement with previous data, the results of a study conducted among the population of Guam [15] and of a more recent case control study in Sweden [6] demonstrate that low Mg intake is linked to an increased risk of idiopathic PD. By means of phosphorus magnetic resonance spectroscopy ( 31 P-MRS), Barbiroli and colleagues [16] have demonstrated a significantly increased content of inorganic phosphate accompanied by a decreased concentration of free cytosolic Mg 2+ ([Mg 2+ ] i ) in the occipital lobes of PD patients compared with healthy subjects. On the other hand, Mg has been shown to decrease negative interactions between environmental (herbicides) and molecular factors (asynuclein) that are known to be involved in PD pathophysiology [17]. Moreover, Hashimoto and colleagues [18] have reported the significant preventive effects of Mg against 1-methyl-4-phenylpyridinium (MPP+) toxicity [19] to dopaminergic neurons.
Although these studies have demonstrated a beneficial effect of sufficient Mg intake for PD prevention and/or the deceleration of PD progression in patients, and despite an obvious link between IMH and PD pathophysiology, the underlying mechanism(s) has (have) remained elusive until now. However, the recent discovery of SLC41A1 as a Na + /Mg 2+ exchanger (NME; Mg 2+ -efflux system) [20] and its localization within the newly identified PD locus PARK16 [21,22,23,24] makes this protein an interesting candidate to explain the involvement of disturbed intracellular Mg 2+ homeostasis in PD pathophysiology.
Human SLC41A1 has been mapped to chromosome 1q31-32 and encodes a protein consisting of 513 amino acids having a molecular mass of 56 kDa [25]. Its 5-kb transcript has been detected in most of the tested tissues in humans and mice (notably in heart, muscle, testis, thyroid gland, and kidney) [25,26]. SLC41A1 has been characterized as an integral protein that is located in the cytoplasmic membrane [27] and that possesses 10 (strongly preferred computer-predicted model) or 11 transmembrane domains [28] with the N-terminus being oriented intracellularly [27,28]. SLC41A1 has also been demonstrated to form hetero-oligomeric complexes. However, the identities of its binding partners and their relevance for the normal NME function of SLC41A1 in vivo remain uncertain [27].
The evidence for NME being involved in PD etiology has further been strengthened by the identification of PD-specific SLC41A1 variants (c.436A.G resulting in p.K146E; c.1440A.G resulting in p.P480P; and c.552+50G.A) in the Chinese population [24] and of the variant of SLC41A1 carrying the amino acid substitution p.A350V (c.1049C.T) in one PD patient of Caucasian origin [23]. Noteworthy is also the fact that SLC41A3, also a member of the SLC41 family, when knockedout in mice displayed abnormal locomotor coordination (www. knockoutmouse.org; [29]).
Furthermore, the null mutation c.698G.T resulting in skipping of exon 6 of SLC41A1 (an in-frame deletion of a transmembrane helix) has been associated with a nephronophthisis-like phenotype (NPHP), therefore, suggesting that the disturbed renal Mg 2+ homeostasis may lead to tubular defects that result in a phenotype similar to NPHP [30]. Also, SLC41A1 has been found to be overexpressed in preeclamptic placental samples with an approximately five times higher frequency than in normoevolutive placental samples [31].
In this study, we have mainly examined the effect of the substitution p.A350V potentially related to PD on the performance of the Na + /Mg 2+ exchanger SLC41A1. Our findings show that the substitution p.A350V in SLC41A1 is a gain-of-function mutation leading to increased Mg 2+ extrusion from the cell.

Cell survival assay
HEK293 cells inducibly over-expressing wt or the p.A350V variant were grown to approximately 80% confluency, rinsed twice with PBS (PAN Biotech) and provided with fresh culture medium. Cell viability was determined, with a TC10 automated cell counter (BioRad), at 0 h and 24 h from the beginning of the induction.
Protein detection in tet-inducible HEK293 cell lines HEK293 cells, over-expressing flag-or HA-strep-tagged wt or p.A350V, and the respective uninduced controls were lysed with RIPA buffer for 30 min. Centrifugation was performed to pellet unsolubilized material (14000 rpm, 30 min, 4uC). The total protein concentration was determined with the Bradford protein assay (Biorad). For the flag-tagged wt variant, samples containing 10 mg total protein and, for the HA-strep-tagged wild-type variant and the p.A350V mutant, 30 mg total protein were run on a 10% SDS-polyacrylamide gel and transferred to a polyvinylidene difluoride (PVDF) membrane. Immunoblotting was performed in TBS-TWEEN plus 2.5% dry milk with antibodies against the flagtag (HRP-conjugated anti-flag M2, Sigma-Aldrich) and the streptag (Qiagen). Anti-mouse IgG linked to horseradish peroxidase (HRP; Cell Signaling Technology) was used as the secondary antibody for the anti-strep antibody. The antibody against RPL19 (Abnova), together with the anti-mouse secondary antibody stated above, was used to detect the loading control. Proteins were visualized by use of the SuperSignal TM West Dura system (Pierce). Image J software (http://rsb.info.nih.gov/ij/) was used for the densitometric analyses.

Membrane protein enrichment
The ProteoExtract TM native membrane protein extraction kit (Calbiochem) was used to extract and enrich membrane proteins from HEK293 cell lines over-expressing wt or p.A350V according to the manufacturer's protocol. Proteins (25 mg) were separated on a 10% SDS-polyacrylamide gel. Immunoblotting was performed as previously described. The soluble protein RPL19 was used to control the specificity of the separation between soluble and membrane proteins.

Subcellular fractionation of proteins
For fractionation of proteins according to their subcellular localization, we used the Qproteome cell compartment kit (Qiagen). Wt or p.A350V cells were induced or left untreated. Cells (4610 6 ) were processed according to the manufacturer's protocol. The obtained fractions (cytosolic, membrane, nuclear, and cytoskeletal) were electroseparated on an 8.5% SDSpolyacrylamide gel, and SLC41A1 variants were immunodetected as described previously. As a control for the specificity of the fractionation, parallel blots were run and probed with antibodies against RPL19 (cytosolic fraction), PMCA4 (membrane fraction; Sigma-Aldrich), or Lamin A (nuclear fraction; Sigma-Aldrich). Mouse secondary antibody conjugated to HRP (Cell Signaling Technology) was used for RPL19 and PMCA4, and an HRP-coupled rabbit antibody was used for Lamin A (Cell Signaling Technology).

Determination of the phosphorylation status of wt and p.A350V variant
The PhosphoProtein Purification Kit (Qiagen) was used according to the manufacturer's instructions. 1.5610 7 cells of the stably transfected HEK293 cell lines expressing strep-or flagtagged wt or strep-tagged variant p.A350V (HA-Strep-tagged) were used as starting material. 2.5 mg of total protein was used for the affinity purification of phosphorylated proteins. Flow through (unphosphorylated proteins) and elution fractions (phosphorylated proteins) were precipitated with 8% (weight/volume) trichloroacetic acid and washed once with acetone. The pellets were dissolved in 0.1 M Tris.HCl buffer containing 2 M urea. Proteins were separated on a 10% SDS-polyacrylamide gel, transferred to a PVDF membrane and immunostained with a primary anti-strep antibody and a secondary HRP-coupled mouse antibody. For the flag-tagged cell line M2 antibody was used. To detect the phosphorylated form of Akt, the phospho-akt (Ser473) primary antibody (Cell Signaling Technology) and the secondary mouse antibody were used. Images were acquired with the BioRad ChemiDoc TM MP System (BioRad).

Blue native electrophoresis and Western blot analysis of SLC41A1-and SLC41A1 p.A350V-protein complexes
Samples containing 10 and 20 mg of strep-affinity purified proteins (IBA & Qiagen) were loaded onto a native 5-18% polyacrylamide gradient gel, and blue native electrophoresis was performed according to Schä gger and Jagow [33]. Electroseparated proteins were transferred to a PVDF membrane. Wt and p.A350V protein complexes were immuno-detected as described previously. NativeMark TM unstained protein standard (Invitrogen) was used as size marker.

Confocal Microscopy
Specimens were prepared according to Kolisek and coworkers [27] except that, for fixation and permeabilization, we used methanol-acetone, and blocking was performed with 10% goat serum. For detection of strep-tagged wt and p.A350V, we used a primary anti-strep antibody (diluted 1:500; Qiagen). The Alexa Fluor-647-conjugated WGA (Invitrogen) was used as a cell membrane marker. Processed samples were mounted with Fluoroshield-DAPI (abcam). Confocal images were taken with a Confocal Laser Scanning Microscope LSM 510 META (Carl Zeiss) equipped with a 63x oil-immersion objective. For the excitation of Alexa-488, Alexa-647, and DAPI, an argon-ion laser (488 nm), helium-neon laser (647 nm), and blue diode laser (405 nm) were used respectively. Image J software (http://rsb.info. nih.gov/ij/) was used for the image merging and correction, as well as to quantify the percentage of co-localization.

Determination of free intracellular Mg 2+ by mag-fura 2 FF-Spectrofluorometry
The -tet and +tet wt and p.A350V cells were rinsed twice with ice-cold, completely divalent-free, Dulbecco's phosphate-buffered saline ( or CMF-HBS+ supplemented with 10 mM Mg 2+ . Differentiation of NME from other transport components was performed by means of the NME inhibitor imipramine (250 mM) [20] or the NME stimulator dB-cAMP (100 mM) [20,34].
Measurements were performed at 37uC in 3 ml cuvettes containing 2 ml cell suspension (cytocrit: 10%) while being stirred in a spectrofluorometer LS50-B (PerkinElmer) [20,27]. [Mg 2+ ] i values were calculated from the 340 to 380 nm ratio according to the formula of Grynkiewicz et al. [35] and as described in Kolisek et al. [20,27]. SLC41A1-dependent Mg 2+ extrusion from induced SLC41A1 (wt) and p.A350V cells was determined from the [Mg 2+ ] i changes observed in Mg 2+ -loaded cells during recovery in CMF-HBS+ solution and calculated by subtracting the respective values of uninduced cells.
If not otherwise stated, data are presented as means 6 SE. All statistical calculations were performed with Sigma-Stat (Jandel Scientific). Significance was determined by Student's t-test or Mann-Whitney rank sum test as appropriate; P#0.05 was considered to be significant.

Impedance-based measurement of cell adhesion and proliferation
The xCELLigence system (RTCA-SP, ACEA Biosciences Inc.) was used according to the manufacturer's instructions for the continuous real-time monitoring of cell adhesion and proliferation by cell-electrode impedance [36] displayed as the dimensionless Cell Index (CI). By using the RTCA Analyser, electrical impedance changes were measured across interdigitated microelectrodes integrated on the bottom of a specialised 96-well plate (E-Plate 96) and sent to the RTCA Control Subunit. The latter used the RTCA Software (version 2.0) for CI calculation from the frequency-dependent electrode resistances and real-time display of data.
Background impedance of E-Plate 96 wells was determined with 50 ml culture medium only or culture medium containing respective concentrations of tetracycline, imipramine, or dB-cAMP. Subsequently, per well, 5610 5 wt or p.A350V cells were plated in a final volume of 100 ml culture medium and half of the samples were induced with tetracycline. Then, localized on the RTCA SP Station the E-Plate 96 was placed into the CO 2incubator, and the CI was monitored every 15 minutes over a period of 48 hours. After about 24 h in culture, cells were either treated with medium or with compounds known to inhibit (imipramine, 250 mM) or activate (dB-cAMP, 100 mM) SLC41A1-dependent Mg 2+ efflux.

Results
Characterization of tet-induced over-expression of SLC41A1 wt and p.A350V variants in HEK293 cells Functional examination of both variants was performed in the newly generated HEK293 cell lines with tet-regulated expression of stably transfected HA-strep-SLC41A1 (HEK293-(HA-strep-SLC41A1); referred to further only as wt cells) and HA-strep-SLC41A1-p.A350V (HEK293-(HA-strep-SLC41A1-p.A350V); referred to further only as p.A350V cells). Thus, the basic characteristics of each cell line were acquired before the physiological characterization of the potentially PD-associated variant of SLC41A1, p.A350V, was performed.
Previously, we had observed that longer tet-induction (20-24 hours) of flag-SLC41A1 in HEK293 cells (clone 17) led to increased death rates of the cells [20]. Therefore, we examined whether a 24-h over-expression of SLC41A1 wt (further only wt) and SLC41A1 p.A350V (further only p.A350V) variants would be tolerated, or whether it would also result in increased death rates. After tet-induction (+tet), we determined 93% to 100% viability of the cells over-expressing either wt or p.A350V. Viability of uninduced (-tet) wt or -tet p.A350V cells ranged after 24 h between 95% and 100%, and therefore, we concluded that a 24-h tet-induced over-expression of the wt or p.A350V variant had no significant effect on cell viability. Weaker expression of wt and p.A350V was also demonstrated by the finding that we had to load a 3-fold greater amount of the protein onto the gel to be able to obtain a SLC41A1 signal equally strong as that in clone 17 (Fig. 1C) [20].
Next, we quantified the transcriptional activity of both transgenic SLC41A1 variants after 24 h of tet-induction and intet cells. A significantly higher amount of ,11.25-fold (+tet wt ddCt mean 123.7267.42/-tet wt ddCt mean 11.060.48; P = .89e-09) of the wt SLC41A1 transcript and a significantly higher amount of ,14.25 fold (+tet c.1049C.T ddCt mean 144.9634.62/-tet c.1049C.T ddCt mean 10.1760.79; P = 1.84e-08) of the SLC41A1-c.1049C.T transcript was detected, when compared with -tet cells (Fig. 1A and 1B). Therefore, we concluded that both wt and p.A350V cells produced similar amounts of transcript after 24 h of tet-induction. We also wished to know whether the overexpression of wt SLC41A1 or c.1049C.T mutant could influence the expression of SLC41A2 and/or SLC41A3 in +tet and -tet wt and p.A350V cells, respectively. Indeed, we did not detect any significant influence of wt or p.A350V over-expression on the expression of SLC41A2 and SLC41A3 (data not shown). This also confirmed the specificity of the hSLC41A1fw and hSLC41A1rev primers.
Leaky expression of the gene of interest can often complicate functional studies [27,37]. Therefore, we tested whether wt and p.A350V over-expressing cell lines exhibited considerable levels of leaky expression of the wt and p.A350V variants. As a control, we used clone 17 of HEK293-(flag-SLC41A1) cells as previously characterized by Kolisek et al. [20,27]. Western blot analysis revealed bands specific for the wt and p.A350V variants (both ,56 kDa) almost exclusively in +tet cells (Fig. 1C). These data confirmed that both tested cell lines exhibited a negligible leaky expression of the transgenic SLC41A1 variants and, therefore, were suitable for downstream experimentation. Next, we determined whether wt and p.A350V cells expressed comparable amounts of SLC41A1 protein. With densitometric analyses performed on three blots with equal amounts of protein isolates from wt and p.A350V cells, we calculated that the density of p.A350V-specific bands was ,1% lower compared with that of wt-specific bands. Therefore, we concluded that both wt and p.A350V cells produced nearly identical amounts of the respective SLC41A1 variants after 24 h of tet-induction. These data further underlined the suitability of both cell lines for further functional experimentation.

SLC41A1 p.A350V localizes within the cytoplasmic membrane
Next, by performing Western blot analysis of the soluble protein-and membrane-protein-enriched cellular fractions (SF, MF) and of subcellular protein fractions (cytosolic, membrane, nuclear, and cytoskeletal), we examined whether the potentially PD-associated variant p.A350V of SLC41A1 was properly targeted into the cytoplasmic membrane, as demonstrated for wt SLC41A1 [20,27]. Figure 1D shows a Western blot analysis of MF and SF isolated from -tet and +tet wt and p.A350V cells. The ,56 kDa bands corresponding to wt and p.A350V were predominantly detected in the MF, with markedly lower abundance in SF of +tet cells. Flag-hSLC41A1 was used as a positive control [27]. Cytosolic protein RPL19 was used to control the specificity of the membrane fraction enrichment and, as expected, was detected only in the soluble protein fraction. We also performed subcellular protein fractionation with HEK293 cells over-expressing wt or p.A350V (Fig. 1E). Probing of the electroseparated fractions with an antibody against the strepepitope resulted in the almost exclusive detection of both the wt and the p.A350V variants in the fraction enriched in plasma membrane proteins. The specificity of the fractionation was controlled by running and probing parallel blots with antibodies against PMCA4 (membrane fraction; positive control; Fig. 1E), against RPL19 (cytosolic fraction, data not shown), or against Lamin A (nuclear fraction; data not shown). These results were in accordance with our confocal microscopy data, which revealed that both wt and p.A350V SLC41A1 variants were predominantly localized in the plasma membrane. This was shown by colocalization of the green fluorescent signal of immunolabeled SLC41A1 variants (anti-strep:GAM-Alexa-488), with the red fluorescent signal of wheat germ agglutinin conjugated to Alexa-647 (Fig. 2). Colocalization correlation analysis between Alexa-488-and Alexa-647-specific signals in wt and p.A350V variants revealed a 94.561.8% (N = 6) and 9262.3% (N = 7) overlap of the green and red pixels, respectively. In contrast, no wt-or p.A350Vspecific fluorescence was seen in -tet cells ( Figure S1). Taken together, these data demonstrated the plasma membrane localization of the potentially PD-related p.A350V variant of SLC41A1. Thus, we conclude that the mutation p.A350V does not affect the intracellular localization of SLC41A1.  [27]. Therefore, we wondered whether the mutation p.A350V could affect the complex-forming characteristics of SLC41A1. Figure 3 shows Western blot analysis performed on strep-affinity purified native protein isolations separated with blue native electrophoresis [27,33]. Wt and p.A350V variants showed identical separation patterns (Fig. 3), both forming two identical complexes with molecular masses between 242 and 480 kDa. This leads us to the assumption that the mutation p.A350V has no obvious effect on the complex-forming abilities of the SLC41A1 protein.
Amino acid substitution p.A350V in human SLC41A1 induces increased Mg 2+ efflux As in our previous study [20], -tet and +tet wt and p.A350V cells were Mg 2+ -loaded by a 20-min pre-incubation in solutions containing 10 [20]. The results for +tet wt and +tet p.A350V cells are summarized in figure 4 showing a significantly stronger Mg 2+ extrusion of 81.264.7 mM/10 min (Np. A350V = 113) in p.A350V cells compared with 48.267.0 mM/10 min (N wt = 82) in the wt cells used as control. These data clearly demonstrate an increased efflux capacity of p.A350V cells.
However, if the measurements were performed in solutions containing 5 or 10 mM Mg 2+ , thereby lowering the inside-out Mg 2+ gradient, no Mg 2+ extrusion occurred from Mg 2+ -loaded +tet wt and +tet p.A350V cells. Instead, as can be seen in figure (Fig. 5A) known to inhibit the NME function of SLC41A1 [20].
In our previous study [20], we have demonstrated that phosphorylation, postulated to be a mechanism for the activation of Mg 2+ extrusion [20,34], regulates NME activity of SLC41A1 and that wt is being detectable in the phosphoprotein-specific fraction (P). To this end we performed Western blot analysis on fractionated protein lysates of induced +tet flag-wt, strep-wt, and strep-p.A350V. Figure 6 demonstrates that the ,56 kDa bands corresponding to wt and p.A350V were detected in the P fractions. The specificity of the fractionation was controlled with an antibody exclusively recognizing phosphorylated Akt.
We also tested whether the SLC41A1-dependent Mg 2+ efflux from +tet wt and p.A350V cells could be further stimulated by the application of dibutyryl-cAMP (dB-cAMP), a membrane-permeant cAMP analog that activates the holoenzyme complex of protein kinase A (PKA) [20]. No additional effects on Mg 2+ extrusion from p.A350V cells were observed after the application of 100 mM dB-cAMP (N p.A350V = 15; Fig. 5B). However, as shown in figure 5B, dB-cAMP increased Mg 2+ release by 77620% (N wt = 15) in wt controls.   figure 5C. Compared with +tet wt cells, the normalized CI (nCI) of +tet p.A350V cells was reduced (2.0360.03 vs. 2.2660.08; P,0.001) after 48 h in culture (Fig. 5C). Application (24 h after seeding) of imipramine and of dB-cAMP in parallel to its effects on NME activity decreases and increases the 48-h nCI of +tet wt and p.A350V cells compared with control values (Fig. 5C). After imipramine application, the 48-h nCI amounted to 0.1860.17 in wt and to 0.2660.40 in p.A350V cells, thus showing a strong reduction in both groups. However, the dB-cAMP-induced increase of the 48-h nCI was much stronger in +tet wt cells compared with +tet p.A350V cells (2.8160.08 vs. 2.5560.08; P,0.001).
RAB7L1, and SLC41A1 have been identified within the newly revealed PD-susceptibility locus PARK16 at chromosome 1q32 [21,22,23,24]. SLC41A1 has been shown by our group to be a cell-membrane-localized Mg 2+ carrier, conducting the exchange of intracellular Mg 2+ for extracellular Na + (NME) [20,27]. NME has been shown to be functionally active in nearly all investigated cells and tissues including neurons [34,40,41,42,43]. It has also been shown to be responsible for the maintenance of an optimal [Mg 2+ ] i for a variety of processes critical for the cell such as bioenergetics [44], the regulation of cellular pH, volume, and the total ion balance, [45,46], and cell proliferation and differentiation [47,48].
Recently, in one PD patient, Tucci and colleagues [23] have found a coding variant of SLC41A1, carrying substitution p.A350V. Here, we have investigated if this potentially PDassociated mutation affects the molecular and/or functional properties of SLC41A1. Our experiments have revealed no changes regarding the cellular localization, phosphorylation status, or complex-forming ability of the p.A350V variant when compared with the wt protein. However, we have demonstrated that +tet p.A350V cells are able to perform Mg 2+ -efflux more efficiently than +tet wt cells. Under our experimental conditions, short-term, 10-min Mg 2+ release is increased by 69610% (P,0.001) after the induction of p.A350V over-expression compared with cells over-expressing wt. Moreover, as a consequence of an enhanced NME activity, we found a reduced proliferation rate in p.A350V compared to wt cells. As the growth experiments were performed for long periods (48 h) and with cells incubated in complete culture media containing 1.2 mM Mg 2+ , the effects of the p.A350V mutation seem to be of relevance also under physiological conditions.
In both cell lines, Mg 2+ extrusion is blocked by .90% after imipramine application, clearly showing that it results from SLC41A1-mediated NME activity [42]. Imipramine, which besides quinidine, is the current pharmacological choice for NME inhibition [20,49], is known to act on the extracellular Na +binding site of the NME, and competition between these two compounds slows Mg 2+ efflux. In agreement with this, the   inhibitory effect of imipramine is in the order of that of sodium withdrawal, which amounts to 91% in our previous study with SLC41A1-over-expressing HEK293 cells [20]. Another characteristic feature of NME is its activation by PKAdependent phosphorylation [20,34]. Elevation of the intracellular cAMP concentration specifically stimulates Na + -dependent efflux either directly via an increased affinity of the transporter for intracellular Mg 2+ [34,50] or by Mg 2+ mobilization from intracellular organelles, e.g. from mitochondria [51]. Interestingly, the application of dB-cAMP (a cell-membrane-permeant cAMPanalog) increased (77620%) Mg 2+ efflux only in +tet wt cells, whereas no effect has been seen in p.A350V cells. This insensitivity of mutant NME to cAMP stimulation might be of pathophysiological importance as under normal conditions various hormones or mediators, e.g., adrenergic substances, prostaglandin E2, and angiotensin II, use this pathway to induce a transient [Mg 2+ ] i decrease that directly or indirectly influences cellular transport mechanisms and physiological functions [49,51,52]. For example, in this study, the growth-promoting effect of dB-cAMP is reduced in p.A350V-over-expressing cells compared with wt cells. The inability further to increase p.A350V-related NME function via cAMP also suggests maximum or near-maximum activation of the transporter and is in accordance with the observation of enhanced Mg 2+ efflux in mutants compared with wt cells. The [Mg 2+ ] i , a main determinant of NME activity [53], is similar (0.4460.01 mM) between wt and p.A350V cells. Therefore, one can speculate that the p.A350V mutation augments the affinity of the transport protein for intracellular Mg 2+ , changes the Mg 2+carrier-complex formation, or dysregulates transporter gating to facilitate the Mg 2+ transport rate.
In the long-term, the increased activity of the NME might contribute to the development of intracellular Mg 2+ deficiency [43], if not compensated for by Mg 2+ influx. Mg deprivation, whether by gene defects such as p.A350, toxins (rotenone, MPTP), or restricted Mg intake, induces and/or exacerbates processes such as oxidative stress accompanied by an increase in NO and free radicals [54,55], dysfunction of mitochondria and the endoplasmic reticulum [56,57], impairment of Ca 2+ homeostasis [58], iron accumulation [59], alterations in the autophagy-lysosome pathways, protein mishandling, and inflammatory responses [58], all of which are known to trigger neuronal damage in neurodegenerative diseases including PD [56,60,61]. In accordance, Oyanagi and colleagues [14] have been able to induce the severe loss of dopaminergic neurons in rats fed for one year with an Mgrestricted diet containing only one-fifth of the normal Mg content. Furthermore, a lower concentration of Mg in various brain regions and in the cerebrospinal fluid of PD patients has been found [16,62,63]. On the other hand, a high extracellular [Mg] of $1.2 mM has been demonstrated to protect dopaminergic neurons of the substantia nigra from MPP+ toxicity [18] and, because of its Ca 2+ -antagonizing effects, to reduce neuroinflammation [64]. Moreover, spontaneous and Fe 2+ -induced accelerated aggregation of a-synuclein can be inhibited by 0.8 mM Mg 2+ [65]. In this study, by using an extracellular [Mg] of 5 and 10 mM, we have been able to block SLC41A1-related Mg 2+ efflux in both wt and p.A350V cells. Increasing the extracellular Mg 2+ concentration will reduce the driving force for an electroneutral Mg 2+ efflux and suggests that the exchanger switches to the reverse mode, thereby performing Mg 2+ uptake [27,42]. Thus, Mg supplementation might be useful for preventing a loss of intracellular Mg 2+ , a loss that is detrimental to neurons.
Under physiological conditions, an adequate Mg intake should be seen as an important positive environmental factor protecting neurons against accelerated ageing caused by slowly acting deleterious environmental factors (e.g. toxins) and/or genetic risk factors. With regard to the latter, Mg 2+ is an essential co-factor in almost all enzymatic systems involved in DNA processing and in nucleotide excision repair, base excision repair, and mismatch repair [66]. Furthermore, PD-relevant genes with a recessive pattern of inheritance (parkin/PARK2, PINK1, and DJ-1/PARK7) are all related to mitochondria dysfunction and oxidative stress making it possible that defects in these genes and disturbances of IMH intervene in these pathways to induce nigral mitochondrial cytopathy [67].

Conclusions
Magnesium deficiency (both systemic and intracellular) has long been suspected to be involved in various human disease complexes such as metabolic syndrome and neurodegeneration in general including PD. An obvious molecular link between disturbed IMH and PD is however missing. In this study, we have examined the functional properties of a recently identified potentially PDassociated coding variant of the NME SLC41A1, p.A350V. We have demonstrated that p.A350V is able to perform Mg 2+extrusion more efficiently than wt NME and shows insensitivity to cAMP stimulation and have found a reduced proliferation rate in p.A350V compared with wt cells. Our data therefore indicate that the rare conservative substitution p.A350V is a gain-of-function mutation leading to an increased Mg 2+ efflux capacity with likely a long-term consequence in systemic deterioration, particularly under conditions of low extracellular Mg 2+ concentration. By extrapolation, our data are in agreement with the findings of epidemiological and case-control studies and suggest that the chronic loss of Mg 2+ from brain tissue and, thus, latent intracellular hypomagnesemia, contributes to neurodegeneration. Na + /Mg 2+ exchange in neurons and SLC41A1 per se might therefore represent a PD-relevant therapeutic target, with Mg 2+ supplementation of PD patients possibly being beneficial. Figure S1 Confocal immunolocalization of HA-strep-SLC41A1 (wt) and HA-strep-SLC41A1-p.A350V in -tet (24 h) cells. Streptagged wt and p.A350V were immunolabeled with primary mouse anti-strep and secondary GAM Alexa-488 antibodies (green signal). Plasma membranes were fluorescently contrasted with wheat germ agglutinin (WGA) conjugated to Alexa-647 (red signal). Nuclei were stained with DAPI (blue signal). Scale bar indicates 10 mm. (DOC) Figure 6. Immunodetection of phosphorylated recombinant flag-SLC41A1 (wt), HA-strep-SLC41A1 (wt), and HA-strep-SLC41A1-p.A350V with the PhosphoProtein purification kit (Qiagen). Total (T), flow-through (U; containing unphosphorylated proteins) and elution (P; containing phosphorylated proteins) fractions were probed with antibodies against strep-or flag-tag. A signal specific for phosphorylated wt or p.A350V was detected in all three cell lines. The specificity of the fractionation was controlled with an antibody against phosphorylated Akt. doi:10.1371/journal.pone.0071096.g006