Effects of magnesium on prostacyclin synthesis and intracellular free calcium concentration in vascular cells

ARTICLE

Auteur(s) : Kazuo Satake, Jong-Dae Lee, Hiromasa Shimizu, Hiroyasu Uzui, Yasuhiko Mitsuke, Hong Yue, Takanori Ueda

First Department of Internal Medicine, Fukui University, 23 Shimoaizuki, Matsuoka-cho, Fukui 910-1193, Japan

Introduction

Several types of cultured cells, including both endothelial and smooth muscle cells of the vessel wall, can synthesize prostacyclin (PGI₂), a potent vasodilator which inhibits platelet function [1, 2]. Prostaglandin (PG) synthesis requires Ca²⁺ since PG production is stimulated by an increase in cytosolic free Ca²⁺ concentration ([Ca²⁺]_i) and is suppressed by a decrease in [Ca²⁺]_i [3, 4]. Agonists such as histamine, norepinephrine that increase [Ca²⁺]_i promote PGI₂ synthesis in vascular endothelial cells or smooth muscle cells [5, 6].

Magnesium ion (Mg²⁺) modulates the influx of extracellular Ca²⁺ and the release of intracellular Ca²⁺ stores in smooth muscle cells [7, 8]. An elevation or decrease of the extracellular Mg²⁺ concentration ([Mg²⁺]_e) will respectively rapidly inhibit or potentiate vascular contractile tension. However, few investigations of the influence of [Mg²⁺]_e on [Ca²⁺]_i in vascular endothelial cells have been done [9, 10].

Watson et al. showed that [Mg²⁺]_e amplifies the basal release of PGI₂ by endothelial cells [11]. However, the mechanisms by which Mg²⁺ promotes vascular endothelial cells to produce PGI₂ are not well understood. Furthermore, it is unknown whether [Mg²⁺]_e amplifies PGI₂ synthesis in endothelial cells under agonist-stimulated conditions. Since [Mg²⁺]_e suppresses [Ca²⁺]_i in vascular smooth muscle cells, it is also unknown whether Mg²⁺ amplifies PGI₂ synthesis in vascular smooth muscle cells under basal and agonist-stimulated conditions.

The present study explored the influence of [Mg²⁺]_e on the functional properties of endothelial and smooth muscle cells by increasing [Mg²⁺]_e in the incubation media from 0.3 to 3 mM. The concentration of 0.3 mM [Mg²⁺]_e corresponds to the lowest physiological level of this ion which can be measured in human serum, the 1 mM [Mg²⁺]_e to the physiological level, and the 3.0 mM [Mg²⁺]_e to the level reached by therapeutic supplementation of Mg²⁺ in patients with arrhythmia or preeclampsia. We explored the effects of [Mg²⁺]_e on basal [Ca²⁺]_i and agonist-stimulated [Ca²⁺]_i transients, on basal PGI₂ production and agonist-stimulated PGI₂ production in cultured vascular endothelial cells and smooth muscle cells. We further compared the effect of [Mg²⁺]_e on PGI₂ production in smooth muscle cells with that of L-type Ca²⁺ channel blockers, nifedipine and diltiazem.

Materials and Methods

Materials

Norepinephrine, histamine, and indomethacin were purchased from Sigma Chemical Co. (ST. Louis, MO). Sodium arachidonic acid was purchased from Nu-Check Prep. Inc. (Elysian, MN). ³H-6-keto PGF1α was obtained from New England Nuclear (Boston, MA). Medium 199 (M-199) and Dulbecco's modified Eagle's medium (DMEM) were purchased from GIBCO (Grand Island, NY). Both 6-keto-PGF1α and anti-6-keto PGF1α antibody were kindly supplied by Ono Pharmaceutical Co. (Osaka, Japan). Fura-2/acetoxymethyl ester (fura-2AM) was obtained from Nacalai Tesque (Kyoto, Japan). Diltiazem HCl and nifedipine were kindly supplied by Tanabe Pharmaceutical Co. (Tokyo, Japan) and Bayer Pharmaceutical Co. (Osaka, Japan) respectively.

Cell isolation and culture

Endothelial cells were prepared from the human umbilical vein by enzymatic dissociation using techniques similar to those described previously from our laboratory [12]. Human umbilical vein endothelial cells (HUVEC) were cultured in M-199 supplemented with 50 µg/ml endothelial cell growth factor (Collaborative Research, Inc., Waltham, MA), 20% fetal calf serum (FCS), 1% glutamine, 100 U/ml porcine intestinal heparin (Sigma) and 1% penicillin-streptomycin. Cultures were maintained at 37 °C in 95% O2/5% CO2 and used at passages 2-3.

Vascular smooth muscle cells (VSMC) were isolated from the thoracic aorta of Sprague-Dawley rats (200-250 g) by the modified explant method of Ross [13] and were cultured in DMEM, supplemented with 10% FCS, and 1% penicillin-streptomycin, maintained at 37 °C in 95% O2/5% CO2, and used at passages 4-8. Cultured cells were identified as smooth muscle cells by their typical hill-and-valley morphology and by immunocytochemistry using HHF-35, a monoclonal antibody that recognizes muscle-specific actin [14].

PGI₂ Determination

Confluent monolayers of HUVEC in 12-well culture dishes were washed three times with 1 ml of M-199, and then incubated for 20 min at 37 °C with Hepes buffer A (5.5 mM dextrose, 137 mM NaCl, 5 mM KCl, 10 mM Hepes, 1.8 mM CaCl₂, pH 7.35) containing various concentrations of MgSO₄ (0.3-3 mM) [11] in the presence or absence of 10^–5 M histamine, or 2 × 10^–5 M sodium arachidonate. We used MgSO₄ in the present study, since Watson et al. showed that MgSO₄ amplifies the basal release of PGI₂ by endothelial cells as well as MgCl₂ [11]. Confluent VSMC in 6-well culture dishes were washed three times with 1 ml of DMEM, and then incubated for 20 min at 37 °C with Hepes buffer A containing various concentrations of MgSO₄ in the presence or absence of 10^–6 M norepinephrine or 2 × 10^–5 M sodium arachidonate. In the experiments using L-type Ca²⁺ channel blockers, either 10^–6 M diltiazem or 10^–7 M nifedipine was added to Hepes buffer A containing 1 mM MgSO₄. After incubation, the cell-free supernatant from each well was taken for the analysis of the PGI₂ concentration. After aspiration, the cells were trypsinized with saline-trypsin-versen (0.05% trypsin and 0.02% EDTA in phosphate-buffer saline [PBS]), harvested and counted in a hemocytometer using phase-microscopy. PGI₂ was quantitated by standard radioimmunoassay of its hydration product (6-keto-PGF_1α) using phosphate-gelatin buffer (50 mM KH₂PO₄, 0.14 M NaCl, 0.1% gelatin, pH 7.4), a 1:2000 dilution of antibody, and 10000-dpm of ³H-6-keto PGF1α. After overnight incubation at 4 °C, charcoal-coated dextran was used to separate bound from unbound radioactivity. This assay can detect as little as 200 pg/ml of 6-keto-PGF_1α.

Measurement of intracellular free Ca²⁺ concentration

The confluent HUVEC and VSMC in the 100 mm cultured dishes were trypsinized with saline-trypsin-versen to disperse. After centrifugation, cells were resuspended in physiological buffer (145 mM NaCl, 10 mM glucose, 5 mM KCl, 1 mM MgCl₂, 10 mM Hepes, 2 mM CaCl₂, pH 7.45) containing 10% FCS, 2 µM fura-2AM, and incubated for 30 minutes at 37 °C. The cells were washed once with PBS and maintained for 20 minutes at room temperature. The loaded cells were washed, centrifuged, resuspended at 1 × 10⁶ cells/ml in Hepes buffer B (118 mM NaCl, 4.7 mM KCl, 2.5 mM CaCl₂, 1.2 mM KH₂PO₄, 5 mM Hepes, 10 mM glucose and 1 mg/ml bovine serum albumin [BSA], pH7.4) containing various MgSO₄ concentrations (0.3-3 mM). Where Ca²⁺ -free conditions were required, the fura-2AM-loaded cells were resuspended in fresh buffer B without CaCl₂, but containing 0.1 mM EGTA and various MgSO₄ concentrations (0.3-3 mM). The suspensions were placed in quartz cuvettes containing a stir bar and then fluorescent measurements were made at 37 °C with two-wave spectrofluorophotometer, RF-5000 (Shimazu Co. Ltd., Kyoto, Japan). The excitation wavelengths were 340 and 380 nm, with emission at 500 nm. The time taken to switch between the 340 and 380 nm excitation wavelengths was 0.8 sec. [Ca²⁺]_i was calculated every 1.6 sec from the ratio (R) of the 340 nm/380 nm fluorescent values using equation of Grynkiewicz et al. [15]:

[Ca²⁺]_i = Kd × (R – R_min)/(R_max – R) × β

A Kd is the affinity of fura-2 for Ca²⁺ (224 nM at 37 °C) and β is the ratio of the fluorescence values obtained at 380 nm in the absence and presence of saturating [Ca²⁺]_i The maximum fluorescence intensity ratio (Rmax) was obtained by permeabilizing the cells with 10% Triton-X, which exposed the dye to the Ca²⁺ concentration of the assay buffer. The minimum fluorescence intensity ratio (Rmin) was obtained by chelating Ca²⁺ with 10 mM EGTA.

Statistical analysis

Data are expressed as means ± standard errors. Differences were tested for statistical significance by one way ANOVA followed by an unpaired Student's t test. A level of P < 0.05 were considered to be statistically significant.

Results

Effects of extracellular Mg²⁺ on basal and agonist-stimulated PGI₂ production by HUVEC

When cultured HUVEC monolayers were incubated in Hepes buffer A with various concentrations of MgSO₄, [Mg²⁺]e dose-dependently increased basal release of PGI₂ by HUVEC (figure 1). [Mg²⁺]_e also increased 10^–5 M histamine-stimulated PGI₂ production by HUVEC.

Effects of extracellular Mg²⁺ on basal and agonist-stimulated [Ca²⁺]i in HUVEC.

Lowering or raising [Mg²⁺]_e did not affect basal [Ca²⁺]_i in HUVEC. However, when HUVEC were treated with 10^–5 M histamine, increased [Mg²⁺]_e suppressed the [Ca²⁺]_i transients (figure 2A). To assess the involvement of extracellular Ca²⁺ concentration in the effect of [Mg²⁺]_e on histamine-stimulated [Ca²⁺]_i transients, similar experiments were performed in Ca²⁺ -free Hepes buffer containing 0.1 mM EGTA. Also in Ca²⁺ -free buffer, increasing extracellular Mg²⁺ slightly but significantly reduced histamine-stimulated [Ca²⁺]_i transients, although it did not affect basal [Ca²⁺]_i (figure 2B). These results indicate that the effects of [Mg²⁺]_e on the histamine-stimulated [Ca²⁺]_i transients in HUVEC are caused by both the suppression of Ca²⁺ influx and its release from intracellular stores.

Effects of extracellular Mg²⁺ on PGI₂ production and [Ca²⁺]_i in VSMC

When cultured VSMC were incubated with various MgSO₄ concentrations, the basal release of PGI₂ was increased in a Mg²⁺ concentration dependent manner. [Mg²⁺]_e also increased 10^–6 M norepinephrine-stimulated PGI₂ production in a concentration-dependent manner (figure 3).

In contrast to HUVEC, increased [Mg²⁺]_e dose-dependently reduced basal [Ca²⁺]_i in VSMC (0.3 mM [Mg²⁺]_e: 207.8 ± 12.8 nM; 1 mM [Mg²⁺]_e: 107.3 ± 2.8 nM; 3 mM [Mg²⁺]_e: 89.3 ± 2.5 nM). [Mg²⁺]_e significantly suppressed agonist-stimulated [Ca²⁺]_i transients in VSMC (0.3 mM [Mg²⁺]_e:451.9 ± 12.0 nM; 1 mM [Mg²⁺]_e: 298.0 ± 8.5 nM; 3 mM [Mg²⁺]_e:182.6 ± 8.7 nM).

Effects of Ca²⁺ channel blockers on [Ca²⁺]_i and PGI₂ production in VSMC

Since Mg²⁺ is called a nature's physiologic Ca²⁺ channel blocker, we compared the effects of Mg²⁺ with those of synthetic Ca²⁺ channel blockers on PGI₂ production by VSMC. As shown in figure 4A, when 10^–6 M diltiazem or 10^–7 M nifedipine were added to the VSMC in Hepes buffer B containing 1 mM MgSO₄, agonist-stimulated [Ca²⁺]_i transients were suppressed. This suppression was same degree as that for [Ca²⁺]_i transients in Hepes buffer B containing 3 mM MgSO₄ in the absence of Ca²⁺ channel blockers. However, as shown in figure 4B, neither Ca²⁺ channel blocker affected agonist-stimulated PGI₂ production by VSMC, while high [Mg²⁺]_e (3 mM) increased agonist-stimulated PGI₂ production. These results suggest that the Mg²⁺ effect on PGI₂ production is not caused by regulation of [Ca²⁺]_i.

Effects of extracellular Mg²⁺ on sodium arachidonate-stimulated PGI₂ production by HUVEC and VSMC

Arachidonic acid liberated from the cell membrane by phospholipase A2 (PLA2), a key enzyme in PGI₂ production, is converted to PGI₂ by two other key enzymes, cyclooxygenase and PGI₂ synthetase. In order to probe the effect of Mg²⁺ on activities of cyclooxgenase and PGI2 synthetase, 2 × 10^–5 M exogenous sodium arachidonate was added to HUVEC or VSMC in Hepes buffer A with various concentrations of MgSO₄. As shown in figure 5, [Mg²⁺]_e dose-dependently increased the sodium arachidonate-induced PGI2 production by HUVEC and VSMC.

Discussion

[Mg²⁺]_e influences the tone and reactivity of veins and arteries, and even small changes in [Mg²⁺]_e exert significant effects on vascular smooth muscle contractility [16, 17, 18]. In vivo and in vitro studies have demonstrated that elevation of [Mg²⁺]_e inhibits spontaneous tone of arteries, dose-dependently dilates vessels and attenuates agonists-stimulated contraction [16, 19]. It has been reported that [Mg²⁺]_e dose-dependently reduced basal and agonists (vasopressin, angiotensin II)-stimulated [Ca²⁺]_i responses in cultured VSMC [7, 8, 20]. The present study also demonstrated that [Mg²⁺]_e dose-dependently reduced basal and agonist (norepinephrine)-stimulated [Ca²⁺]_i responses in cultured VSMC. On the other hand, few investigations have been done on the influence of [Mg²⁺]_e on [Ca²⁺]_i under basal and agonist-stimulated conditions in endothelial cells. An elevation of [Ca²⁺]_i is known to be closely coupled to important functional properties of vascular endothelial cells, such as synthesis/release of bioactive substances in response to a number of agonists or physical stimuli [21]. This elevated [Ca²⁺]_i in endothelial cells results from an influx of extracellular Ca²⁺ through ionic channels and release of intracellular Ca²⁺ stores [22]. Our study demonstrated that [Mg²⁺]_e suppressed the histamine-induced [Ca²⁺]_i responses in HUVEC, by suppressing both an influx of extracellular Ca²⁺ and release from intracellular Ca²⁺ stores. These findings suggest that changes of [Mg²⁺]_e may affect important functions of vascular endothelial cells.

Watson et al. [11] showed that Mg²⁺ directly amplifies the basal release of PGI2 by HUVEC. Nadler et al. reported that MgSO₄ infusion produced a significant increase in the renal excretion of the stable PGI₂ metabolite in humans [23]. However, the mechanisms by which Mg²⁺ promotes PGI₂ production are not well understood. In this study, we have shown for the first time that extracellular Mg²⁺ dose-dependently promoted basal PGI₂ production by not only HUVEC but also VSMC, and agonist-stimulated PGI₂ production by both cells.

Studies of the molecular structure and function of PLA₂, a key enzyme in PGI₂ production, have shown a calcium requirement for PG synthesis [4, 24]. Ca²⁺ is needed for cytosolic PLA₂ to translocate from the cytoplasmic region to the cell membrane where the PLA₂ stimulates the release of arachidonic acid [3]. Histamine-induced PGI₂ production by HUVEC is associated with a concentration-dependent discharge of the internal Ca²⁺ stores and a subsequent influx of extracellular Ca²⁺, thereby activating PLA₂ [25, 26]. In our study, Mg²⁺ promoted the histamine- or norepinephrine-stimulated PGI₂ production by HUVEC or VSMC, even though Mg²⁺ suppressed the agonist-stimulated [Ca²⁺]_i responses in both cells. However, while synthetic Ca²⁺ channel blockers suppressed agonist-stimulated [Ca²⁺]_i responses, as did Mg²⁺, they did not produce an effect similar to Mg²⁺ on PGI₂ production. The mechanism by which Mg²⁺ promotes PGI₂ production is thus likely to be independent of [Ca²⁺]_i changes.

Mg²⁺, which is called nature's physiologic Ca²⁺ channel blocker, has the effect of activating various enzymes, being different from synthetic Ca²⁺ channel blockers. For example, Mg²⁺ is a cofactor in the activation of phosphorylases, adenylate cyclase, and ATPase [16, 27]. Cyclooxgenase and PGI₂ synthetase convert the endogenous arachidonic acid to PGI₂. In this study, extracellular Mg²⁺ amplified exogenous sodium arachidonate-stimulated PGI₂ production by HUVEC and VSMC. Our results suggest that Mg²⁺ stimulate not PLA₂ activity but cyclooxygenase and/or PGI₂ synthetase activity.

Conclusion

[Mg²⁺]_e suppressed the [Ca²⁺]_i rise induced by vasoactive agents not only in VSMC but also in vascular endothelial cells. Furthermore, [Mg²⁺]_e promoted either basal or vasoactive agents-stimulated PGI₂ production by vascular endothelial cells and smooth muscle cells. Synthetic Ca²⁺ channel blockers did not have this effect. Maintaining a physiological serum concentration of Mg²⁺ is clinically important for the regulation of vascular tone. Indeed, Mg²⁺ deficiency is associated with numerous cardiovascular disorders, which are associated with changes in vascular tone, including hypertension, myocardial infarction, and vasospastic angina [28, 29]. Our results suggest that extracellular Mg²⁺ regulates vascular tone by antagonizing the effects of Ca²⁺ in smooth muscle cells and modulating the synthesis of biologically active substances such as PGI₂ by the vascular endothelial cells and smooth muscle cells. Further study is required to clarify the actions of intracellular Mg²⁺ and its effects on several enzyme activities in these cells.

Acknowledgement

We would like to express our appreciation to Dr. Tatsuro Tomokage of Tomokage Hospital for generously providing the specimen of human umbilical cord, and to Ono Pharmaceutical Co. for the kind gifts of 6-keto PGF_1α and anti-6-keto PGF_1α. This work was supported in part by a Grant-in-Aid for Scientific Research (No.05770455) from the Ministry of Education, Science and Culture, Japan.

References

1. Baenziger NL, Becherer PR, Majerus PW. Characterization of prostacyclin synthesis in cultured human arterial smooth muscle cells, venous endothelial cells and skin fibroblasts. Cell 1979; 16: 967-74.

2. Wen FQ, Watanebe K, Yoshida M. Lipopolysaccharide and cytokines enhance bradykinin-stimulated production of PGI₂ by human pulmonary artery smooth muscle cells. Cell Biol Int 1997; 21: 321-7.

3. Whorton AR, Willis CE, Kent RS, Young SL. The role of calcium in the regulation of prostacyclin synthesis by porcine aortic endothelial cells. Lipids 1984; 19: 17-24.

4. Hassid A, Oudinet JP. Relationship between intracellular calcium and prostaglandin synthesis in cultured vascular smooth muscle cells. Prostaglandins 1986; 32: 457-78.

5. Baenziger NL, Fogerty FJ, Mertz LF, Chernuta LF. Regulation of histamine-mediated prostacyclin synthesis in cultured human vascular endothelial cells. Cell 1981; 24: 915-23.

6. Nebigil C, Malik KU. Prostaglandin synthesis elicited by adrenergic stimuli mediated via alpha-2C and alpha-1A adrenergic receptors in cultured smooth muscle cells of rabbit aorta. J Pharmachol Exp Ther 1992; 260: 849-58.

7. Zhang A, Cheng TP-O, Altura BM. Magnesium regulates intracellular free ionized calcium concentration and cell geometry in vascular smooth muscle cells. Biochim Biophys Acta 1992; 113: 425-9.

8. Yoshimura M, Ohshima T, Matsuura H, Ishida T, Kanbe M, Kajiyama J. Extracellular Mg²⁺ inhibits capacitative Ca²⁺ entry in vascular smooth muscle cells. Circulation 1997; 95: 2567-72.

9. Zhang A, Cheng TP-O, Altura BT, Altura BM. Mg²⁺ and caffeine-induced intracellular Ca²⁺ release in human vascular endothelial cells. Br J Pharmacol 1993; 109: 291-2.

10. Yoshimura M, Ohshima T, Matsuura H, Inoue T, Kanbe M, Kajiyama J. Differential effects of extracellular Mg²⁺ on thrombin-induced and capacitative Ca²⁺ entry in human coronary arterial endothelial cells. Arterioscler Thromb Vasc Biol 1997; 17: 3356-61.

11. Watson KV, Moldow CF, Ogburn PL, Jacob HS. Magnesium sulfate: rationale for its use in preeclampsia. Proc Natl Acad Sci USA 1986; 83: 1075-8.

12. Shimizu H, Takayama H, Lee J-D, Satake K, Taniguchi T, Yamamura H, Nakamura T. Effect of vanadate on prostacyclin and endothelin-1 production and protein-tyrosine phosphorylation in human endothelial cells. Thromb.Haemostasis 1994; 72: 973-8.

13. Ross R. The smooth muscle cell. II. Growth of smooth muscle in culture and formation of elastic fibers. J Cell Biol 1971; 50: 172-86.

14. Tsukada T, Tippens D, Gordon D Ross R, Gown A. HHF-35,A muscle-actin-specific monoclonal antibody: immunohisto-chemical and biochemical characterization. Am J Pathol 1987; 12: 651-60.

15. Grynkiewicz G, Poenie M, Tsien Y. A new generation of Ca²⁺ indicators with greatly improved fluorescence properties. J Biol Chem 1985; 260: 3440-50.

16. Altura BM, Altura BT. New perspectives on the role of magnesium in the pathophysiology of the cardiovascular system. Experimental aspects. Magnesium 1985; 4: 245-71.

17. Gold ME, Buga GM, Wood KS, Byrns RE, Chaudhuri G, Ignarro LJ. Antagonistic modulatory roles of magnesium and calcium on release of endothelium-derived relaxing factor and smooth muscle tone. Circ Res 1990; 66 (2): 355-66.

18. Farago M, Szabo C, Dora E, Horvath I, Kovach AG. Contractile and endothelium-dependent dilatory responses of cerebral arteries at various extracellular magnesium concentrations. J Cereb Blood Flow Metab 1991; 11: 161-4.

19. Noguera MA, D'o con MP. Modulatory role of magnesium on the contractile response of rat aorta to several agonists in normal and calcium-free medium. J. Pharm. Pharmacol 1993; 45: 697-700.

20. Touyz RM, Laurant P, Schiffrin EL. Effects of magnesium on calcium responses to vasopressin in vascular smooth muscle cells of spontaneously hypertensive rats. J Pharmacol Exp Ther 1998; 284: 998-1005.

21. Furchgott RF, Vanhoutte PM. Endothelium-derived relaxing and contracting factors. FASEB J 1989; 3 (9): 2007-18.

22. Schilling WP, Elliott SJ. Ca²⁺ signaling mechanisms of vascular endothelial cells and their role in oxidant- induced endothelial cell dysfunction. Am J Physiol 1992; 262: H1617-H1630.

23. Nadler JL, Goodson S, Rude RK. Evidence that prostacyclin mediates the vascular action of magnesium in humans. Hypertension 1987; 9: 379-83.

24. Clark JD, Lin LL, Kriz RW, Ramesha CS, Sultzman LA, Lin AY, Milona N, Knopf JL. A novel arachidonic acid-selective cytosolic PLA₂ contains a Ca²⁺ dependent translocation domain with homology to PKC and GAP. Cell 1991; 65: 1043-51.

25. Jacob R, Merritt J, Hallam TJ, Rink TJ. Repetitive spikes in cytoplasmic calcium evoked by histamine in human endothelial cells. Nature 1988; 335: 40-5.

26. Hallam TJ, Jacob R, Merritt JE. Influx of bivalent cations can be independent of receptor stimulation in human endothelial cells. Biochem J 1989; 259: 125-9.

27. Grubbs RD, Maguire ME. Regulation of magnesium but not calcium transport by phorbol ester. J Biol Chem 1986; 261: 12550-4.

28. Iseri LT, French JH, Calif I. Magnesium: nature's physiologic calcium blocker. Am Heart J 1984; 108: 188-94.

29. Satake K, Lee J-D, Shimizu H, Ueda T, Nakamura T. Relationship between severity of magnesium deficiency and frequency of angina attacks in patients with variant angina. J Am Coll Cardiol 1996; 28: 897-902.