Main

There has been keen interest in the use of magnesium as a potential neuroprotective intervention in perinatal medicine. Magnesium plays a critical role in the functioning of excitatory neurotransmitters, mostly at the postsynaptic level. It is a noncompetitive, voltage-dependent channel blocker acting on a modulatory site within the N-methyl-d-aspartate (NMDA) channel (1). This function is thought to allow magnesium to potentially block distal manifestations of the excitotoxic pathway. This is highly relevant inasmuch as excitatory neurotransmitters, such as glutamate, function as neurotoxins when in high concentrations in the brain extracellular fluid (ECF) (2). The latter occurs during hypoxia and/or ischemia due to excessive release of neurotransmitters as well as blockade of reuptake mechanisms (3). The high concentrations of excitatory transmitters stimulate surface receptors of neurons excessively and result in an influx of calcium, which in turn triggers a cascade of intracellular events culminating in tissue injury. There is specific interest in the neuroprotective effects of magnesium during the perinatal period. This is reflected by a broad experience with regard to the use of magnesium during pregnancy, including serum magnesium levels, knowledge of magnesium pharmacokinetics during pregnancy, the ready passage of magnesium across the placenta, and the tolerance of the fetus to elevated levels (4–8). The latter observations allow initiation of magnesium therapy before birth.

We have previously used newborn swine and brain microdialysis to demonstrate that increases in plasma concentration of magnesium result in entry into the brain ECF (9). Elevations in [Mg]ecf, in turn, occur in an environment whereby interactions can readily occur at the neuronal cell surface. Although these observations support the prevalent focus of neuroprotection by magnesium, they also suggests the possibility that increases in [Mg]ecf could raise [Mg]i. There has been very little work on alterations in [Mg]i in the perinatal period, but given magnesium's physiologic role as a calcium antagonist (10), there could be important intracellular effects of increases in [Mg]i. Thus, the purpose of this study was to use brain microdialysis and P-31 magnetic resonance spectroscopy (MRS) to examine the effects of increases in plasma magnesium concentration on brain [Mg]ecf, [Mg]i, and brain phosphorylated metabolites and pHi before, during, and after HI in a newborn swine model.

METHODS

Experimental protocol.

The study protocol was approved by the Institutional Animal Care and Research Advisory Committee of the University of Texas Southwestern Medical Center at Dallas. Animals were initially sedated with ketamine (20 mg/kg i.m.), followed by local skin infiltration of the neck with 1% Xylocaine, rapid tracheotomy, and initiation of mechanical ventilation using a 30:70 oxygen:nitrous oxide mixture. Analgesia was provided with nalbuphine (0.15 mg/kg i.m.) and muscle relaxation with tubocurarine Cl (0.1 mg/kg i.v.). Rectal temperature was maintained at 38–39°C using a heating pad. Catheters were placed in the external jugular vein and carotid artery. A 2-mm inflatable vascular occluder (In Vivo Metric, Healdsburg, CA, U.S.A.) was placed around the left common carotid artery and the tubing to control inflation/deflation was exteriorized. After local scalp infiltration with 1% Xylocaine and administration of sodium pentothal (20 mg/kg i.v.), the scalp was retracted, and a burr hole was made through which a loop-type microdialysis probe was inserted to a depth of 1 cm into the cerebral cortex and situated 1 cm anterior and lateral to the coronal and sagittal cranial sutures, respectively, for measurements of brain [Mg]ecf. A 1 cm depth was used to insure that microdialysis samples were obtained from primarily gray matter of the cerebral cortex. A radiofrequency surface coil was then situated over the cranial vault for NMR data collection, and the animals were placed in a 4.7 T magnet. The microdialysis probes were positioned within the inner diameter of the surface coil (4 × 3 cm), on each side of the sagittal suture, with each probe 0.5–0.75 cm from the coil. Based upon prior work with phantoms, it is estimated that the majority of the NMR signal is derived from the first centimeter of cortex underneath the coil in the frontal, parietal, and occipital regions. Animals were allowed to stabilize for 3 1/2 hours before obtaining baseline values of HR, MAP, [Mg]plasma, brain [Mg]ecf, and brain intracellular energy metabolites over a 30-min period. Anesthesia was continued with 70% nitrous oxide along with intermittent doses of nalbuphine for analgesia.

Five miniswine (age 6 ± 3 d, wt 1.4 ± 0.4 kg, mean ± SD) were administered MgSO4 and denoted MgHI. The target [Mg]plasma range was 4–6 mM based upon preliminary studies, which indicated that this was the highest concentration of Mg that could be tolerated without detrimental physiologic side-effects (9). Magnesium sulfate (30 mg of anhydrous powder per milliliter of H2O; Sigma Chemical, St. Louis, MO, U.S.A.) was infused intravenously at a dose of 275 mg/kg body weight for 30 min, followed by 100 mg/kg for an additional 30 min in an effort to maintain [Mg]plasma at constant levels. Five miniswine (age 5 ± 3 d, wt 1.1 ± 0.3 kg) were administered Na2SO4 and denoted NaHI. Sodium sulfate (30 mg of anhydrous powder per milliliter of H2O; Sigma Chemical) was administered in the NaHI group at a dose of 275 mg/kg followed by 100 mg/kg. HI was induced using the method of LeBlanc et al.(11). Ischemia was begun at 30 min of either MgSO4 or Na2SO4 infusion by hemorrhagic hypotension to decrease MAP by one third, followed by induction of HI at 45 min of the infusions by inflating the vascular occluder and reducing the inspired oxygen to 10% for a 15-min interval. At the completion of the infusion, HI was terminated by changing the inspired gas concentration to 21%, deflating the vascular occluder, and reinfusing the withdrawn blood. The animals were studied an additional 4 h after the completion of HI. HR and MAP were monitored continuously and recordings were made every 30 min. Arterial pH and blood gases (partial pressures of O2 and CO2) were sampled every 30 min, and oxygen contents were determined at control, at the completion of HI, and at study completion. [Mg]plasma was measured every 15 min from the start of the infusion protocol for both groups. NMR measurements were obtained over successive 10-min intervals, and changes in [Mg]ecf were monitored by collecting consecutive 10-min samples using microdialysis as described below. At the end of the protocol, animals were killed using sodium pentobarbital (200 mg/kg, i.v.).

Microdialysis measurements.

Loop-type microdialysis probes were constructed of commercially available hollow-fiber dialysis tubing (internal diameter = 150 μm, molecular weight cut-off = 17 kD; Spectrum Medical Industries Inc., Los Angeles, CA, U.S.A.) as described previously (9). After determination of the dead space from the probe to the effluent collection opening, probes were infused with artificial cerebrospinal fluid (154 mM NaCl, 1.3 mM CaCl2, pH 7.2) at a rate of 5 μL/min and calibrated by dialysis against a standard solution with a known concentration of MgSO4 (154 mM NaCl, 1.3 mM CaCl2, and 1.0 mM MgSO4). Recovery factors were determined from the ratio of total magnesium concentrations of the effluent and standard solutions, and were used to calculate the apparent [Mg]ecf from microdialysis fluid. Microdialysis data are presented at the midpoint of 10-min collection periods and corrected for real time by using the dead space determinations for each probe (i.e. 6–8 μL or 1.2–1.6 min). Throughout the experiment, brain [Mg]ecf microdialysis samples of 50 μL volume were collected continuously every 10 min.

Magnesium concentration determinations.

Total Mg was determined on 45 μL of microdialysis fluid and 10 μL of blood via atomic absorption spectroscopy at 285.2 nm (Spectra AA, Varian, Palo Alto, CA, U.S.A.) using an air/acetylene flame. Lanthanum oxide (0.5% wt/vol; Sigma Chemical) was added to reduce the potential for interferences from the incomplete dissociation of compounds in blood or microdialysis fluid, as recommended in the Varian analytical methods manual. Mg standard curves were prepared from stock MgSO4 solutions.

NMR measurements.

NMR measurements were made at 4.7 T using a double-tuned radiofrequency surface coil and protocol for interleaved 1H and 31P NMR data collection similar to that described previously (12). The collected 1H NMR data were not used for this study. NMR data collection commenced 20 min before starting the MgSO4 infusion and continued for 4 h after infusion. A 90° excitation pulse was applied at the 31P frequency and its resulting free induction decay (FID) was acquired over 256 ms using a 5000-Hz sweep width and 2048 data points per FID; each spectrum was the sum of 200 FID (10 min per spectrum). The 31P FID were processed by applying a baseline correction for DC offsets, left shifting to remove the first three data points, right shifting to re-establish phase integrity, apodization using an exponential filter corresponding to 10 Hz line broadening, zero filling once to 4096 data points, Fourier transformation, and baseline straightening via a spline interpolation routine. The chemical shift scale was calibrated with respect to PCr at 0.0 parts per million (ppm) in a control spectrum and the chemical shift locations of 31P NMR peaks were determined by fitting individual peaks using a Lorentzian peak shape.

Data analysis.

[Mg]i was calculated using the following formula (13):

where KD is the apparent dissociation constant of MgATP, ςα−β is the observed frequency difference between the α-ATP and β-ATP 31P NMR peaks, and ςATP and ςMgATP are the frequency differences between these two peaks in the absence and presence of saturating amounts of Mg, respectively. The KD at a pH of 7.2, 37°C, and an ionic strength 0.15 used was 35 μM. KD was adjusted specifically for the pHi, using the formula:

The endpoints, ςATP and ςMgATP, were adjusted for specific pH values using the formulas:

Intracellular brain pH (pHi) was calculated from the chemical shift difference of inorganic phosphate (Pi) as described previously (14) using the following equation:

where K is the apparent dissociation constant for Pi, ςHP and ςp equal the chemical shift at the acid and basic endpoints of the titration, respectively, and p equals the chemical shift difference in ppm between the Pi resonance frequency at different stages in the protocol and the PCr resonance at control (0 p.p.m.). Values of pK = 6.683, ςHP = 3.153, and ςp = 5.728 were selected for calculations of pHpi, as described previously (14).

Changes in phosphorylated metabolites were examined by expressing the peak area of nucleoside triphosphate (NTP), Pi, and PCr during and after infusions of MgSO4 or Na2SO4 as a percentage of values before infusions (control, set at 100%). NTP represents the average of the α, β, and γ NTP peaks. The average NTP concentration was used because preliminary analysis confirmed that the changes in the average of the γ, α, and β peaks are similar during ischemia to the change in the β-NTP alone. Use of the average reduces the variability associated with decreasing signal to noise of the NMR measurement during ischemia. The SD of phosphorylated metabolites during control was determined by expressing replicate values as a percentage of the first value and calculating the SD of the difference. All systemic data (HR, MAP, pH, blood gases, and hematocrit) as well as %PCr, %Pi, %NTP, pHi, [Mg]plasma, [Mg]i, and [Mg]ecf for each group were analyzed by two-way repeated measures ANOVA. Group interactions were further evaluated and localized by pair-wise comparisons (Student-Newman-Keuls). To facilitate and simplify the analysis, data collected in the postinfusion interval were grouped into 1-h blocks. Regression analysis was used to explore the relationship between [Mg]i and the cerebral energy state. A p value of < 0.05 was considered significant. All data are expressed as mean ± SD.

RESULTS

Arterial pH, blood gases, and oxygen contents were the same for both groups at control (Table 1). During HI, both groups demonstrated similar changes in pH but lower Po2 values for the MgHI group. In spite of this, oxygen contents decreased to a similar extent in each group, indicating that MgHI and NaHI were comparable in the degree of induced hypoxia. There was a small decrease in Pco2 in each group during HI, and this was different from control only for NaHI group. With the exception of a persistent acidosis, both groups demonstrated recovery to control values by 45 min after HI. MAP was similar between groups before HI (Fig. 1, upper panel). Both MgHI and NaHI had similar decreases in MAP during HI, and prompt increases in MAP to values similar to control shortly after resolution of HI. Patterns of changes in HR differed between MgHI and NaHI groups (Fig. 1, lower panel). NaHI animals had increases in HR above control values during HI, whereas MgHI animals had a decrease in HR leading to group differences at the end of HI.

Table 1 Arterial pH, blood gases, and O2 content during control, HI, and 45 min after HI
Full size table
Figure 1
figure 1

Changes in MAP (upper panel) and HR (lower panel) are plotted for MgHI (▴) and NaHI (•) groups. The times of control measurements (C), Na2SO4 and MgSO4 infusions (Na and Mg, respectively), ischemia (I), hypoxia-ischemia (HI), and the period after HI (Post-HI) are indicated on the horizontal bars above the graphs. Values different from control are indicated by asterisks and differences between groups by crosses (p < 0.05).

Full size image

Control values of [Mg]plasma were similar between groups, but differed both during and after HI (Fig. 2, bottom panel). During infusions of MgSO4, [Mg]plasma was elevated above baseline, reached a peak value of 9.1 ± 1.5 mM at the end of HI, and decreased after completion of the infusion but remained significantly elevated throughout the remainder of the experiment. During the Na2SO4 infusion, [Mg]plasma rose slightly by the end of HI (0.65 ± 0.36 to 1.14 ± 0.64 mM), but was not statistically different. Values of [Mg]ecf at baseline were similar between groups (Fig. 2, middle panel). Infusion of MgSO4 was associated with increases in [Mg]ecf, which peaked at 74% above baseline values during HI (0.23 ± 0.11 to 0.40 ± 0.14mM). In contrast, infusion of Na2SO4 did not change [Mg]ecf during or after HI. Group differences for [Mg]ecf extended from 15 min of the respective infusions, through HI, and for 50 min after resolution of HI. Calculated [Mg]i values were similar between groups at control, and did not change with infusions of either MgSO4 or Na2SO4 (Fig. 2, top panel). With the onset of HI, significant elevations of [Mg]i occurred in both groups. At 15 and 25 min after HI, group differences were present, with higher values of [Mg]i in the MgHI group (p < 0.05).

Figure 2
figure 2

Magnesium concentrations of MgHI (▴) and NaHI (•) groups are plotted for fluid compartments of plasma ([Mg]plasma, lower panel), brain extracellular fluid ([Mg]ecf, middle panel) and brain intracellular fluid ([Mg]i, upper panel). The times of control measurements (C), Na2SO4 and MgSO4 infusions (Na and Mg, respectively), ischemia (I), hypoxia-ischemia (HI), and the period after HI (Post-HI) are indicated on the horizontal bars above the graphs. Values different from control are indicated by asterisks and differences between groups by crosses (p < 0.05).

Full size image

Relative concentrations of phosphorylated metabolites (NTP, Pi, PCr) and brain pHi at control, during infusions of MgSO4 or Na2SO4, and after hemorrhagic hypotension but before HI are listed in Table 2. Groups were similar in all variables until HI. The decreases in PCr and pHi at 45 min of the infusion reflects the effects of hemorrhagic hypotension initiated at 30 min of the infusion. During HI, infusion of MgSO4 was associated with greater changes in NTP concentration compared with animals infused with Na2SO4. Specifically, the MgHI group had a maximal reduction in NTP to 48 ± 6% of control compared with a decrease to only 70 ± 3% in the NaHI group, and the maximum decrease occurred 5 min after HI. Differences between groups in NTP concentration extended from the last 5 min of HI until 25 min after HI (Fig. 3, upper panel). Accordingly, there was a greater change in Pi for the MgHI compared with NaHI groups (248 ± 49%versus 144 ± 20% of control at 5 min after HI, respectively, p < 0.05; data not shown). The larger changes in phosphorylated metabolites in the MgHI group were accompanied by greater brain acidosis. The pHi values for MgHI and NaHI groups were similar at control (6.99 ± 0.03 and 7.01 ± 0.02, respectively), but was lower at 15 min after HI in the MgHI group (6.01 ± 0.07) compared with the NaHI group (6.51 ± 0.14, p < 0.05). Group differences in pHi extended until 35 min after HI (Fig. 3, bottom panel). Values of PCr decreased to a similar extent for each group (14 ± 6%versus 13 ± 5% of control for MgHI and NaHI, respectively, at 5 min before completion of HI) and recovered to baseline levels by 25 min after HI (data not shown).

Table 2 NMR measurements at control, after 25 min of infusions of either MgSO4 or Na2SO4, and after 45 min of infusions before HI
Full size table
Figure 3
figure 3

Changes in brain NTP (upper panel) and pHi (lower panel) are plotted for MgHI (▴) and NaHI (•) groups. Values for NTP are expressed as a percentage of control, with the latter set at 100%. The times of control measurements (C), Na2SO4 and MgSO4 infusions (Na and Mg, respectively), ischemia (I), hypoxia-ischemia (HI), and the period after HI (Post-HI) are indicated on the horizontal bars above the graphs. Values different from control are indicated by asterisks and differences between groups by crosses (p < 0.05).

Full size image

The relationship between [Mg]i and either brain NTP concentration or pHi during control and HI are plotted in Figure 4. For each group there was a strong and nearly identical inverse correlation between the reduction in NTP concentration relative to control and the increase in [Mg]i described by the following equations:Y = −0.006 X + 0.78, r2 = 0.73, p < 0.001 for MgHI, and Y = −0.006 X + 0.82, r2 = 0.59, p < 0.001 for NaHI (where Y = [Mg]i and X = percentage NTP). Similarly, there was a strong and nearly identical inverse correlation between the extent of brain acidosis and the increase in [Mg]i described by the following equations:Y = −0.32 X + 2.44, r2 = 0.85, p < 0.001 for MgHI, and Y = −0.29 X + 2.23, r2 = 0.85, p < 0.001 for NaHI (where Y = [Mg]i and X = pHi).

Figure 4
figure 4

Scatter plots for intracellular Mg concentration ([Mg]i) as a function of NTP concentration (top panel) and pHi (lower panel). Data are plotted for both the MgHI (▴) and NaHI (•) groups using values obtained at control, at 45 min of the protocol (during infusion of either MgSO4 or Na2SO4, and the start of HI), at 55 min (during the infusions and 10 min of HI), and at 65 min (5 min after the termination of HI and completion of the infusions). These time points were chosen to capture the most prominent changes in brain energy state and acidosis. Regression lines for the MgHI and NaHI groups are represented by the broken and solid lines, respectively. Equations describing the regression lines are included in the text.

Full size image

DISCUSSION

This investigation examined the effects of induced changes in plasma magnesium concentrations during and after HI on [Mg]ecf, [Mg]i, and on brain phosphorylated metabolites and pH using brain microdialysis and 31P NMR in newborn swine. The principal findings of this investigation were 1) acute elevation in [Mg]plasma was readily associated with increases in brain [Mg]ecf; 2) increases in [Mg]i occur during HI in both MgSO4 and Na2SO4 groups, but the increment was greater for animals with an acute elevation in [Mg]plasma and brain [Mg]ecf; and 3) HI with hypermagnesemia was associated with greater alterations in NTP concentration and pHi than HI with a normal plasma [Mg]. Newborn miniswine were used because we have previously examined the relationship between systemic and brain magnesium concentrations in this model. Furthermore, newborn miniswine have been well characterized in this laboratory, including brain energy reserves and utilization rates (12), agonal glycolytic rate (15), and lactate/acid clearance (16). All of the latter processes reveal maturational changes over the first month and support newborn miniswine as a good model for neonatal events. However, newborn miniswine are probably more mature than newborn rodent models based upon energy utilization rates (12), magnitude of cerebral flow (CBF) (17), and ratio of PCr to NTP (12).

We have previously used newborn miniswine to demonstrate that acute increases in [Mg]plasma to 4–6 mM under normoxic/normotensive conditions results in an approximate doubling of brain [Mg]ecf(9). The present investigation adds to these observations by examining [Mg]plasma and brain [Mg]ecf during HI. Using the same dose and time of infusion of Mg as in our prior study (9), MgHI attained a plasma [Mg] of 6.4 ± 1.1 mM at 30 min of the infusion (before HI) that was similar to values in our prior report (5.7 ± 1.1 mM). Between 30 and 60 min of the infusion (MgSO4 or Na2SO4) HI was associated with a further increase in [Mg]plasma to 9.1 ± 1.5 mM, higher than the value of 5.4 ± 1.1 mM achieved under normoxic/normotensive conditions. The reason for the higher [Mg]plasma during HI is unclear but may reflect redistribution of cardiac output and an altered volume of distribution (not measured). In spite of differences in [Mg]plasma during normoxia and HI, the increase in [Mg]ecf was similar. This suggests a limitation of Mg entry into the brain during HI relative to normoxia. This could be secondary to decreased brain Mg delivery and would be consistent with the pronounced ischemic changes noted in brain energy state even though plasma levels were raised. Saturation of transport mechanisms for Mg entry into the brain is another possibility.

Systemic physiologic variables before and during HI were well matched among the groups in spite of different plasma [Mg]. An important difference between groups was the pattern of HR change with reductions in HR during HI in Mg-infused animals and increases in HR in Na2SO4 infused. Reduction in HR with Mg supplementation has been reported previously in newborn swine under normoxic conditions from this laboratory (9) and others (18). The reduction in HR is part of global hemodynamic changes associated with hypermagnesemia, the most consistent of which is a dose-related reduction in systemic vascular resistance (19, 20). The blood pressure response in turn depends on the cardiac output, which will reflect a dose-dependent negative inotropic effect of Mg. These effects may contribute to the observation that MgHI animals had greater reductions in NTP and brain acidosis compared with NaHI. This occurred in the presence of similar changes in MAP and systemic oxygenation (indicated by O2 content). Given the vasodilatory effect of Mg on cerebral vessels (21), different cerebral flow (CBF) during and after HI among the two groups is a potential etiology for the changes in NTP and pHi. Reduction in CBF during HI would be expected to be associated with progressive vasodilation. Additional vasodilatory stimuli such as hypermagnesemia may promote greater reductions in CBF at the same blood pressure compared with animals with a normal [Mg].

Calculated values of [Mg]i before infusions and HI were identical in each group (0.22 mM), and were lower than other values for brain using either ion-selective electrodes [0.46 mM (22)], or Mg indicators such as Mag-Fura-2-acetoxymethyl ester [0.48 mM (23)]. In part, this may reflect the specific algorithm used to derive [Mg]i using P31-MRS data, as previously demonstrated in this laboratory (13). Irrespective of the absolute value of [Mg]i, both groups had significant increases in [Mg]i during or immediately after HI that was higher in the MgSO4 group (164% above control) compared with the Na2SO4 group (109% above control).

Based upon the results of this study, the most likely reason for the observed increase in [Mg]i and differences between groups in [Mg]i are the changes in brain phosphorylated metabolites and pHi. Thus, the metabolic pattern of increased [Mg]i, brain acidosis, reduced NTP and PCr, and increased inorganic phosphorus is consistent with utilization of NTP and release of intracellular bound magnesium. This pattern of change has been noted in adults with stroke (24) and in guinea pig brain in vitro during severe hypoxia and hypoglycemia (25). The strong, inverse correlations between [Mg]i and both the reduction in NTP concentration and the extent of brain acidosis for both groups are supportive of this concept (Fig. 4). Furthermore, the greater change in [Mg]i for the MgHI compared with the NaHI group most likely reflects greater alterations in the brain energy state during HI.

There are other possibilities for the observed increase in [Mg]i during HI. First, magnesium could be transported from the ECF into cells. Although regulation of [Mg]i is not completely understood (26), there are data to support a reversible, ATP-independent, Na+-Mg+2 antiport in isolated leech neurons (22). In the latter, increasing the [Mg]ecf increased [Mg]i and decreased both [Na]i and membrane depolarization; a reduction in [Mg]ecf had the opposite effect. In contrast to magnesium efflux, use of amiloride to inhibit the Na+-Mg+2 antiport suggested that most of the magnesium influx occurs through passive pathways and not reversal of the Na+-Mg+2 antiport (22). Transport/influx of magnesium could be secondary to the elevated [Mg]ecf in MgHI animals, however, this is not consistent with the unchanged [Mg]ecf in the NaHI group. A second possibility is neuron depolarization. In cultured dorsal root ganglion neurons of newborn rats, depolarization is associated with a 3-fold increase in [Mg]i(23, 27). The increase in [Mg]i occurred in the absence of [Mg]ecf, and the increase disappeared in the absence of extracellular calcium. These observations suggest that depolarization causes an influx of calcium, which in turn releases Mg from intracellular organelles such as mitochondria, and endoplasmic reticulum (27). A final possibility for the increase in [Mg]i is chemical stimulation. Vasopressin (28), bradykinin (29), and cholecystokinin (30) can increase [Mg]i in vascular and nonvascular tissue outside of the CNS. Glutamate can induce increases in [Mg]i in cultured rat cortical neurons (31). In contrast to depolarization, the mechanism for glutamate-stimulated increases in [Mg]i involves Mg entry through NMDA-activated ion channels (32). Based on prior study of HI in neonatal animal models (3, 33), it is expected that brain glutamate would be increased for both MgHI and NaHI groups.

The clinical implications of an increased brain [Mg]i during and immediately after HI are unclear. Raising the [Mg]i may contribute to the putative neuroprotection mechanisms associated with magnesium. For example, magnesium is a physiologic antagonist to calcium (10). It has been speculated that the calcium-mediated rise in [Mg]i with neuronal depolarization may serve as an important negative feedback to suppress neuronal excitation (27). Furthermore, an increase in [Mg]i after depolarization may be critical for maintaining regulation of ionic channels in the face of elevated intracellular calcium (27). Increased intracellular calcium associated with HI is a critical focal point of the excitotoxic pathway to brain injury, and increases in [Mg]i may counter these effects in a parallel fashion to depolarization. However, if the increased [Mg]i merely reflects the extent of brain energy failure, then neuroprotection provided by increases of [Mg]i is doubtful. Furthermore, if physiologic effects mediated by [Mg]i (e.g. enzyme activation, ionic channel regulation, maintenance of protein conformation) are dependent on a constant [Mg]i, then benefit may not be expected from Mg supplementation therapy. Interestingly, there are a number of studies that report a reduced [Mg]i associated with head trauma, and that Mg supplementation obviated the decrease in [Mg]i and attenuated the extent of neurologic injury (34, 35).

The results of this investigation demonstrate that Mg supplementation has extremely complex effects on the brain. Ultimately, the only way to ascertain whether Mg supplementation provides neuroprotection is to perform parallel experiments with markers of brain injury as outcome variables, including both histologic and neurobehavioral assessments. In view of the complexity of performing brain microdialysis to measure [Mg]ecf and simultaneous brain biochemical assessments using 31P NMR during and after HI, this investigation focused on immediate perturbations of these variables. Given these limitations, the results of this investigation suggest that the presence or absence of neuroprotection associated with Mg supplementation will probably reflect a net function of Mg effects in the brain ECF, intracellular compartment, and on the determinants of brain phosphorylated metabolites and pH. Increasing [Mg]ecf is expected to be beneficial because there is blockade of NMDA receptor activity and limitation of calcium influx, in addition to possible inhibition of lipid peroxidation (36). In contrast, for the reasons discussed above, increasing [Mg]i may have beneficial or detrimental effects on the brain. The effect of Mg on the brain energy state during acute HI is expected to be detrimental. Greater energy failure with HI (primary energy failure) is correlated with subsequent secondary energy failure, and ultimately histologic evidence of brain injury (37).

These divergent biochemical and physiologic effects of Mg therapy may provide insights into the discrepant reports regarding magnesium as a potential neuroprotective therapy in both animals (38–45) and humans (46–48). Magnesium administration either before or after HI, ischemia, or excitotoxic exposure in perinatal and adult rats ameliorated the extent of neurologic injury (38–43). In contrast, systemic magnesium administration after HI in newborn piglets did not prevent the development of delayed energy failure, but much lower [Mg]plasma were achieved compared with the present study (44). Furthermore, systemic magnesium administration after asphyxia in fetal sheep did not reduce the extent of cytotoxic edema, did not improve the recovery of the EEG, and did not alter the histologic outcome (45). Retrospective epidemiologic human data, however, suggests that in utero exposure of the fetus to magnesium is associated with less cerebral palsy in children of very low birth weight (46, 47). The latter observations have been questioned because MgSO4 has not been found to reduce the incidence of cranial ultrasound abnormalities that are strong predictors of cerebral palsy in preterm infants (48). Uncertainties over the neuroprotective role of magnesium therapy could be clarified by further study of magnesium supplementation on brain damage assessed several days after HI using the newborn swine model presented here.