Cellular Mg²⁺ levels play a fundamental role in many critical processes, including regulation of cellular ionic composition, enzymatic activity, cell volume and initiation of protein synthesis (Raftos et al., 1999; Rubin, 2005; Vidair and Rubin, 2005). Small variations in cellular or serum Mg content or Mg²⁺ have been associated with various pathological conditions such as diabetes, hypertension, sickle cell anemia and cancer (Sartori et al., 1992; Resnick et al., 1993a; Picado et al., 1994; Barbagallo et al., 1996; Dewitte et al., 2004; Maltezos et al., 2004; Zehtabchi et al., 2004). Mammalian cells maintain cellular Mg²⁺ levels within a narrow range by the combined action of ion transporters and cellular Mg²⁺ buffering capacity. ATP, 2,3-bisphosphoglycerate, cellular phosphoproteins (including phosphorylation-regulated protein kinases and phosphatases themselves) and other nucleic acids (including polynucleic acids) constitute the major cellular Mg²⁺ buffers (Laing et al., 1994; Raftos et al., 1999; Chiu and Dickerson, 2000; Waas and Dalby, 2003). Cellular Mg²⁺ transport encompasses both influx of extracellular Mg²⁺, attributed to TRPM6/7 channels and (more controversially) to CNNM and MagT1 transporters (Nadler et al., 2001; Runnels et al., 2001; Monteilh-Zoller et al., 2003; Schmitz et al., 2003; Voets et al., 2004; Goytain and Quamme, 2008; Giménez-Mascarell et al., 2019), and efflux of intracellular Mg²⁺, mediated by both Na⁺-dependent and Na⁺-independent mechanisms (Féray and Garay, 1986; Ferreira et al., 2004). Na⁺-dependent Mg²⁺ efflux (Na⁺/Mg²⁺ exchange) has been functionally described in various cell types including human RBC (Féray and Garay, 1986; Murphy et al., 1991; Günther and Vormann, 1992; Romani et al., 1993; Wolf et al., 1997; Touyz and Schiffrin, 1999a; Touyz et al., 2001; Standley and Standley, 2002; Ferreira et al., 2004). We and others have shown that Mg²⁺ transport across the RBC plasma membrane is regulated by insulin, norepinephrine, angiotensin II, and vasopressin through activation of Na⁺/Mg²⁺ exchange (Romani et al., 1993; Touyz and Schiffrin, 1999b; Cefaratti and Romani, 2003; Ferreira et al., 2004; He et al., 2005; Rubin, 2005). The SLC41A1 gene has been proposed to encode the erythroid Na⁺/Mg²⁺ exchanger polypeptide (Schweigel et al., 2000; Sahni et al., 2007; Kolisek et al., 2008; Kolisek et al., 2012), a hypothesis strengthened by the demonstration that SLC41A3 mediates mitochondrial Na⁺/Mg²⁺ exchange (Mastrototaro et al., 2016). However, more a more recent report has failed to detect extracellular Na⁺-dependence of SLC41A1-mediated Mg²⁺ transport (Arjona et al., 2019).

Type 2 diabetes (T2D) affects nearly 500 million people worldwide, with a rapidly rising annual toll exceeding four million deaths (Palomino-Schätzlein et al., 2020). The multiple pathophysiological effects of hyperglycemia are not completely explained by glucose-induced changes in cellular energy stores or in red cell (Morabito et al., 2020) and mononuclear cell oxidative stress (Fortuño et al., 2006). In particular, the relationship between divalent cation metabolism and the pathogenesis and complications of hyperglycemic, often insulin-resistant conditions such as T2D is not fully understood. The hypomagnesemia observed in up to half of people with T2D (Gommers et al., 2016), as well as in many children with insulin-dependent diabetes (Rohn et al., 1993), has been traditionally attributed to urinary magnesium wasting secondary to the osmotic diuresis accompanying diabetic glycosuria (Gommers et al., 2016). In contrast, elevated glucose concentrations have been shown to lower RBC Mg content through as yet undefined mechanisms (Paolisso et al., 1986; Resnick et al., 1993b; John, 1997; Rodríguez-Morán et al., 2011; Chan et al., 2015). Serum glucose concentrations are associated with low Mg²⁺ levels in both serum and RBC (Resnick et al., 1993b; Guerrero-Romero and Rodríguez-Morán, 2002; Rodríguez-Morán and Guerrero-Romero, 2003). The mechanisms by which glucose induces reduction in cellular Mg content and contributes to development of diabetic complication are not completely understood. Elevated serum glucose levels lead to glycation of circulating and of membrane-bound proteins. Indeed, glycated HbA_1c is an excellent clinical marker of glycemic status over time (John, 1997; Pani et al., 2008). HbA_1c in T2D subjects is higher than in control subjects and negatively correlates with Mg levels in plasma, platelets, mononuclear cells, and muscle cells (Allegra et al., 1997).

We now present evidence that glucose modulates erythrocyte Mg levels by stimulating Mg²⁺ efflux through activation of Na⁺/Mg²⁺ exchange in ex vivo RBC from humans with type 2 diabetes (T2D) and otherwise healthy individuals, as well as in a rodent model of T2D. We also show that RBC from T2D subjects have lower cellular Mg content and increased baseline Na⁺/Mg²⁺ exchange activity as compared to cells from normal control subjects. Treatment of RBC from T2D subjects with extracellular N-glycosidase increased cellular Mg content and decreased Na⁺/Mg²⁺ exchange activity to non-diabetic control levels. Thus, our results suggest that RBC protein glycation adducts and protein N-glycans directly or indirectly modulate Na⁺/Mg²⁺ exchanger activity, contributing to dysregulation of Mg²⁺ homeostasis in T2D.

Materials: A23187, bovine serum albumin (BSA, fraction V), choline chloride (Choline Cl), PP2 and SU6656 were purchased from Calbiochem (San Diego, CA). Wortmannin (WT) was from Alexis Corp. (San Diego, CA). Dimethyl sulfoxide (DMSO) was from Fisher Scientific (Pittsburgh, PA). Acationox was from Baxter Scientific Products (McGaw Park, IL). All other reagents were purchased from Sigma Aldrich (St. Louis, MO).

Isolation of Red Cells: Human blood samples were obtained between 7:00 a.m. and 10:00 a.m. after overnight fast (>6 h) per approved clinical protocol and processed within 3 h of receipt, as previously described (Ferreira et al., 2004; Rivera et al., 2005). Briefly, freshly isolated blood was passed through cotton and eluted with choline wash solution (CWS-Mg free, containing (in mM) 150 choline chloride, 10 Tris MOPS pH 7.4 at 4°C) to remove the buffy coat (>85% of white blood cells and platelets). The red cells were washed 4× at 4°C with CWS-Mg free kept on ice until use. Hematological parameters were measured by ADVIA hemoanalyzer (Bayer, Terrytown, NY). Glycated hemoglobin (HbA_1c) was measured by Hitachi 917 autoanalyzer (Manheim/Boehringer). Total intracellular contents of Na, K, and Mg were measured by atomic absorption spectrophotometry (Perkin Elmer 800). To avoid changes in mean cellular volumes (MCV), mouse blood was prepared as specified for human RBC, but in solutions adjusted to the osmolarity of normal mouse plasma (330 mOsm).

Cellular Mg ²⁺ efflux measurements: Cells were loaded with Mg²⁺ as we previously described (Ferreira et al., 2004; Rivera et al., 2005). Cells at 10% hematocrit were incubated with 6 µM Ca²⁺/Mg²⁺ ionophore A23187 in Mg²⁺-loading solution (MLS) containing (in mM) 140 KCl, 12 MgCl₂, 10 D-glucose, and 10 Tris MOPS (pH 7.4) for 30 min at 37°C. To obtain a range of different intracellular Mg concentrations in Figure 5B, MLS solutions contained 0–16 mM MgCl₂ concentrations between 0 and 16 mM, with corresponding KCl concentrations between 138 and 158 mM. A23187 was subsequently removed with four washes of 25 volumes of 0.1% BSA in MLS at 37°C at 15 min intervals. Mg²⁺-loaded cells at 3% hematocrit were incubated at 37°C for periods of 5 or 45 min in NaCl or choline choline chloride flux media containing (in mM) 140 NaCl or 140 choline Cl, 10 D-glucose, 10 Tris MOPS (pH 7.4), 20 sucrose, 0.1 ouabain, and 0.01 bumetanide. Mg²⁺ efflux was calculated from the slope of linear regression analyses of Mg content (expressed as calculated concentration) in supernatant vs. time (5 and 45 min in triplicate determinations). Na⁺/Mg²⁺ exchange activity was calculated as the difference between Mg²⁺ efflux in NaCl and in choline Cl flux media. Flux values were corrected for changes in mean corpuscular volume (MCV) and expressed as mmol/10¹³ cells × h (flux units, FU), as previously described (Ferreira et al., 2004; Rivera et al., 2005). Mouse RBC assay osmolytes were adjusted to reflect the normal mouse plasma osmolarity of ∼330 mOsm, to minimize cell volume changes and hemolysis during Mg²⁺ loading (Rivera et al., 2005).

Glucose-induced Mg ²⁺ efflux: RBC were incubated up to 24 h at 37°C in isotonic saline solution containing (in mM): 140 NaCl, 0–100 D-glucose, 10 Tris MOPS (pH 7.4), 0–20 sucrose, 0.1 ouabain, and 0.01 bumetanide, with the indicated concentrations of D-glucose or sorbitol. Incubated RBC were centrifuged 5 min at 2,500 rpm at 37°C and suspended in Mg²⁺-free choline wash solution as previously described (Acosta et al., 2000). Previous investigation (Viskupicova et al., 2015) revealed that 1 h exposure of RBC to >45 mM glucose at 37°C does not increase hemolysis, eryptosis or GSSG/GSH ratio (Viskupicova et al., 2015). Moreover, lipid peroxidation, superoxide production and intracellular (Ca²⁺) remain unchanged under these conditions. We observed no hemolysis after 24 h incubation of RBC in the presence of 50 mM glucose. We noted trace hemolysis after 24 h incubation in the presence of 100 mM glucose, but no further hemolysis was evident in RBC after resuspension in flux medium. Aliquots of 50% suspension were taken for measurements of Mg²⁺ efflux and total cellular contents of Mg, K, and Na as described above. We also studied the effects of Peptidyl-N-Glycosidase F (PNGase F), an amidase that cleaves the GlcNAc-asparagine linkage of high mannose and complex oligosaccharides of N-linked glycoproteins (Maley et al., 1986). Intact RBC were incubated with 0.1 U/ml PNGase F (New England BioLabs, Ipswich, MA) for 1.5 h at 37°C in isotonic saline solution, then centrifuged 5 min at 2,500 rpm at 37°C and resuspended in Mg²⁺-free choline wash solution as described (Tarentino and Plummer, 1994). Timed aliquots of 50% suspension were taken for measurement of total RBC Mg, K, and Na.

Study subjects: The study protocol was approved by the Institutional Review Board of the Brigham and Women’s Hospital (Protocol #: 2003P001861). Normal (n = 33) and T2D individuals (n = 30) were recruited at the Brigham and Women’s Hospital. Informed written consent was obtained from all subjects before participation. T2D was diagnosed according to accepted guidelines (National Diabetes Data Group: Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance, 1979). Diabetic participants were not being treated with antihypertensive or any other medication except for medication related to their diabetes. Not all subjects were studied in all analyses, and subjects were not matched for clinical characteristics. Subjects varied in age between 22 and 62, with body mass index (BMI) <31 kg/m² for women or <33 kg/m² for men. Diabetic and control groups did not differ significantly in age.

db/db Mice: All mouse studies were conducted under protocols approved by the Institutional Animal Care and Use Committee of Brigham and Women’s Hospital. db/db mice (The Jackson Laboratory; Bar Harbor, Maine; Catalog #000642) and heterozygote controls (Catalog #000662) were purchased at 6 weeks of age, fed ad libitum with rodent chow (LabDiet #5053; Richmond, IN) and water, then sacrificed and venisected by intracardiac puncture at 25 weeks of age. db/db mice are homozygous for a spontaneous mutation of the leptin receptor (Lepr ^db ) leading to unrestrained weight gain and eventual development of obesity-associated diabetes. db/db but not db/+ mice exhibit hyperglycemia as early as 8 weeks of age and have been used to model human T2D (Shafrir and Sima, 2001; Joost and Schürmann, 2014). Although not measured in these experimental mouse groups, blood glucose levels in mice of the same age (25 weeks) from the same supplier and maintained in the same animal facility had blood glucose levels of 159 ± 21 mg/dl (db/+) and 769 ± 41 mg/dl (db/db) (Guo et al., 2008).

Statistical Analyses: The data are reported as means ± standard error of the mean (SEM), with statistical significance (p value as indicated) determined by non-parametric analysis t-test unless otherwise stated. The kinetic parameters of Na⁺/Mg²⁺ exchange activation were analyzed with GraphPad Prism version 8.0.0 for Windows (GraphPad Software, San Diego, California United States). Pearson and Spearman coefficients of correlation were estimated using SPSS 10.0 software (SPSS, Chicago, IL).

We hypothesized that glucose modulates intracellular ionic composition of RBC, and that dysregulated ion metabolism characterizes the pathophysiology of T2D. We report our observation that HbA_1c levels correlate directly with Na⁺/Mg²⁺ exchange activity and inversely with intracellular Mg levels in RBC from healthy subjects. The dysregulated erythroid Mg²⁺ homeostasis in T2D reflects enhanced glucose-regulated Na⁺/Mg²⁺ exchange activity. These findings suggest that hemoglobin glycation state predicts intracellular Mg levels controlled by regulation of Mg²⁺ efflux pathways in RBC, and complement previous observations of the inverse association between RBC Mg levels and fasting blood glucose (Resnick et al., 1993b; Barbagallo et al., 1996; Barbagallo et al., 2001). Na⁺/Mg²⁺ exchange thus joins the K_ATP channel, the L-type Ca²⁺ channel and GLUT4 of the pancreatic beta cell, K channel Kir4.1 of retina, and (via insulin) TRPM6 and the thiazide receptor NCC of kidney (Gommers et al., 2016) as a Mg²⁺-regulated ion transport pathway.

The regulation by glucose of Na⁺/Mg²⁺ exchange activity could in principle reflect the hyperglycemia-associated hypomagnesemia of diabetes, which has been attributed to diabetic glycosuria. Indeed, acute glucose infusion can promote magnesuria, and the SGLT2 knockout mouse exhibits hypermagnesemia. However, SGLT2 inhibitors used to treat T2DM do not promote magnesuria, but rather had been observed to lead to hypermagnesemia (Tang et al., 2016).

We also report pharmacological evidence that glucose-stimulated activation of Na⁺/Mg²⁺ exchange requires Src family tyrosine kinase activity. Inhibition of Src family tyrosine kinase activity was sufficient to attenuate the increased Na⁺/Mg²⁺ exchange activity or T2D RBC. Src tyrosine kinase functions in part as downstream signaling molecule for receptors without intrinsic kinase activity (Parsons and Parsons, 1997). While the mechanism by which RBC Src tyrosine kinase regulates Na⁺/Mg²⁺ exchange remains to be determined, Src kinase and caveolin are known together to contribute to activation of the receptor for advanced glycation end-products (RAGE) in vascular tissues (Reddy et al., 2006). RAGE mediates inflammatory and white blood cell migration-associated signals associated with vascular complications of hyperglycemia (Igarashi et al., 1999; Schmidt, 2015; Hudson and Lippman, 2018). Thus, the Src family of tyrosine kinases may be potential targets for development of new therapeutic strategies to ameliorate glucose-mediated complications associated with T2D.

Nonenzymatic glycation affects not only hemoglobin, but extends to proteins of the RBC surface (Miller et al., 1980). Red cell glycation status has been correlated with cellular survival and with rheology that is modifiable by control of glycemic status (Peterson et al., 1977). Hyperglycemia has also been associated with decreased red cell membrane fluidity (Watala, 1988). However, a direct relationship between red cell surface glycation status and transport protein function remains to be established. Nonetheless, exposure to extracellular glucose modulates activities of the Na⁺ pump (Garner et al., 1990; Umudum et al., 2002; Nandhini and Anuradha, 2003), Ca²⁺ pump (González Flecha et al., 1993), Na⁺/H⁺ exchanger (Williams and Howard, 1994), GLUT1 (Nandhini and Anuradha, 2003), as well as the non-erythroid Na⁺/Glucose cotransporter of epithelia (Han et al., 2005). We have extended these studies to demonstrate that glucose regulates erythroid Mg²⁺ levels by modulating Na⁺/Mg²⁺ exchange activity. Consistent with a role for deregulated cation metabolism in the pathogenesis of T2D, we found reduced cellular Mg²⁺ levels and increased Na⁺/Mg²⁺ exchanger activity in ex vivo RBC from T2D subjects as compared to cells from otherwise healthy subjects. ex vivo RBC from db/db mice showed Mg²⁺ transport abnormalities resembling those of RBC from T2D subjects. Exposure of RBC to extracellular glucose abolished apparent cooperativity in the extracellular Na⁺ -dependence of Na⁺/Mg²⁺ exchange activity, accompanied by reduction in K_1/2 for extracellular Na⁺ from 83 to 29 mM. This change may reflect a noncovalent effect of glucose or a direct or indirect effect of non-enzymatic glycation at or near the putative extracellular Na⁺ binding site of the Na⁺/Mg²⁺ exchanger, or of residues in one or more of the exchanger’s hypothesized regulatory proteins.

Covalent reaction of the open form of glucose with proteins leads to protein glycation (Rabbani and Thornalley, 2012) occurs predominantly on N-terminal α-amino groups and ε-amino groups of Lys residues, as well as on side chains of Arg and Cys residues. Early glycation proceeds sequentially through glycosylamine formation and dehydration to Schiff bases in a process requiring hours. Subsequent Amadori rearrangement to fructosamine or non-Amadori rearrangement to α-oxo-aldehydes, both of which can then more slowly degrade over weeks in vitro (Valencia et al., 2004) to multiple advanced glycation end-products (AGEs). Protein susceptibility to glycation and the gradually increasing burden of protein glycation in circulation and in cells likely contributes to pathogenesis of hyperglycemic disorders (McAvan et al., 2020).

SLC41A1 has been often (Sponder et al., 2013), if controversially (Sahni and Scharenberg, 2013), modeled as 10 transmembrane spans with intracellular N- and C-termini. In this model, SLC41A1 has four ecto-Lys residues in two putative extracellular loops and multiple ecto-Arg residues in each of the five predicted extracellular loops, any of which might serve as glycation targets. As glucose-induced activation of Na⁺/Mg²⁺ exchange occurs within 1.5 h, the observed effects of extracellular glucose on RBC Na⁺/Mg²⁺ exchange activity do not likely reflect the action of AGE formation of binding to their receptors (RAGEs). The time- and concentration-dependence of Na⁺/Mg²⁺ exchange activation by extracellular glucose, is consistent with those previously reported by Watala et al. for red cell membrane glycation (Watala, 1988). Also consistent with our results, impaired Mg²⁺ uptake and homeostasis characterizes the streptozotocin-induced rat model of type 1 diabetes (T1D) (Cefaratti and Romani, 2003). Hepatic plasmalemmal vesicles from streptozotocin-treated rats, exhibited >2-fold higher rates of Na⁺-dependent Mg²⁺ efflux than those from untreated rats (Cefaratti et al., 2004; Fagan et al., 2004). These data together demonstrate that hyperglycemia, as observed in T2D and in a model of T1D, induces changes in Mg²⁺ transport.

We found that RBC pre-treatment with extracellular PNGase F restores Na⁺/Mg²⁺ exchange activity to normal levels, reversing its elevation by high glucose levels. PNGase F exposure at 37°C for 1 h has been used to partially remove surface N-linked glycans from intact T regulatory cells without compromising cellular function (Cabral et al., 2017). We showed that PNGase F treatment of RBC from T2D subjects increased glucose-mediated intracellular Mg levels and decreased Na⁺/Mg²⁺ exchanger activity, suggesting that enzymatic removal of RBC surface protein-linked N-glycans can regulate Mg²⁺ homeostasis in T2D RBC. The 10-transmembrane span model of SLC41A1 predicts a single consensus site for ecto-N-linked glycosylation, but neither mutational nor enzymatic evidence for this site has yet been presented. Thus, the functionally important RBC surface polypeptide substrate(s) for extracellular PNGase regulation of erythroid Na⁺/Mg²⁺ exchange remain to be defined.

Enhanced glycation of RBC and endothelial surface proteins can enhance vascular dysfunction, suggesting that attenuation of protein glycation might delay development of vascular complications in T2D. Our results suggest that RBC surface glycation and N-glycan state of the Na⁺/Mg²⁺ exchange protein or of one or more interacting regulatory protein(s) can regulate intracellular Mg²⁺ and Mg²⁺ homeostasis in T2D RBC. Additional work is needed to identify the glycated and glycosylated proteins responsible for Na⁺/Mg²⁺ exchange activation in RBC of T2D subjects. Also remaining to be investigated is the possible effect on Na⁺/Mg²⁺ exchange of the recently reported deposition of amylin aggregates in or on T2D red cells (Verma et al., 2020). As regulation of RBC Mg²⁺ transport may have parallels in endothelial and vascular smooth muscle cells, we speculate that identification and modulation of Mg-regulatory glycation and N-deglycosylation targets may have therapeutic potential in treatment of the vascular complications of T2D.