Main

Studies in term infants with neonatal encephalopathy (NE) suggest that, although antenatal or genetic factors might predispose some infants to perinatal brain injury (1,2), events in the immediate perinatal period are important in neonatal brain injury (3). A consistent pattern of energy derangement has been observed by several groups in infants with NE due to intrapartum asphyxia using phosphorus (31P) and proton (1H) magnetic resonance (MRS), spectroscopy (4–7). 31P and 1H MR spectra were often normal within the first few hours after resuscitation, but after 8–24 h there was a progressive decline in PCr/Pi and rise in brain lactate despite adequate oxygenation and circulation in the infant. The nadir of the energetic disturbance was seen after 12–24 h and the magnitude of the fall in PCr/Pi and rise in brain lactate correlated with the subsequent neurodevelopmental abnormality (8,9). This sequence of events was termed “secondary energy failure” to distinguish it from the “primary” decline in PCr and ATP and rise in lactate seen during the insult in experimental animal models of hypoxia-ischemia (HI) (10,11). The biphasic pattern of energy failure during and after HI has been the cornerstone behind the realization that neuroprotective strategies might interrupt the cascade of irreversible injury if administered within hours of a perinatal insult (12), and the concept has been an important catalyst in the development of neuroprotection trials with moderate hypothermia (13–15).

Brain pHi is alkaline during secondary energy failure. This phenomenon was first described in an experimental model in the early 1980s, where, despite the marked intracellular acidosis during HI, brain pHi normalized within 15 min of reperfusion and subsequently increased above the control value (16). The Na+/H+ transporter is responsible for brain pHi and cell volume regulation under physiologic conditions by extruding protons from and taking sodium up into cells (17). Excessive stimulation of the Na+/H+ transporter after transient HI is thought to lead to deleterious effects in the cell, one of which is an alkaline pHi. Amiloride is an inhibitor of the Na+/H+ transporter and has been shown in some studies to ameliorate reperfusion injury (18).

The precise relationship between brain pHi and injury severity has been studied in infants with NE during the period of secondary energy failure and during the first year after birth (19). The extent of the brain alkalosis in the first 2 wk after birth was associated with the severity of brain injury on MR imaging, neurodevelopmental outcome at 1 y, and brain lactate/Cr. Brain alkalosis persisted for weeks in infants with a severe neurodevelopmental outcome and was associated with the development of cerebral atrophy. These observations suggest that brain pHi may be a marker or mechanism of brain injury severity. Indeed, mild acidosis (both extra- and intracellular) during reperfusion after transient HI protects the brain (20,21); in a tissue culture model of cerebral ischemia, the prevention of rebound alkalosis using amiloride after in vitro HI delayed the onset of injury for as long as the cells remained acidotic (22).

Interestingly, in some term infants with NE two components of the Pi peak were observed (19); two Pi components have also been observed in preterm infants at term (23) and in experimental models studied at high magnetic field strength of 7 T (10). These data suggest that there may be a degree of pHi heterogeneity in vivo.

The pHi of the brain slices is generally more alkaline than extracellular pH, in contrast to in vivo observations where brain pHi is more acid than extracellular pH (24,25). Compared with intact brain, slices have a lower ATP and PCr (26) with an increased production of lactate (27). The precise mechanisms leading to these metabolic differences between the brain slice and intact brain are unclear, but we considered that reduced high-energy phosphates and intracellular alkalosis suggests that brain slices may be undergoing a process similar to the secondary energy failure seen after transient HI in newborn infants (4–7).

We therefore studied the energetics of immature rat brain slices over time using MRS techniques specifically to assess whether the brain slice model develops intracellular alkalosis with a progressive decline in energetics, analogous to secondary energy failure. We hypothesized that any such change in energetics would be ameliorated by hypothermia or with amiloride, strengthening the analogy to secondary energy failure.

MATERIALS AND METHODS

As studies of electrocortical activity have suggested that the 14-d rat pup is the most suitable age for comparative studies in the term human infant (28), the 14-d brain slice model was used in the majority of experiments. Any gestational age effect on energy decline was checked in the 7-d model; however, as no difference was observed, amiloride studies were performed only in the 14-d model.

For the baseline studies, KHB containing NaCl (124 mM), KCl (5 mM), KH2PO4 (1.2 mM), MgSO4 (1.2 mM), CaCl2 (1.2 mM), NaHCO3 (26 mM), and D-glucose (10 mM), equilibrated with O2/CO2, 95%/5% at 37°C and at a pH of 7.4–7.5 was used. Because the presence of bicarbonate and CO2 have been reported to interfere with the changes in pHi due to amiloride (29), a bicarbonate-free buffer, HEPES, was therefore used for the amiloride experiments. The HEPES buffer contained 20 mM N-(2-hydroxyethyl)piperazine-N′-(2-ethanesulfonic acid)/20 mM 4-(2-hydroxyethyl)piperazine-1-(2-ethanesulfonic acid sodium salt, NaCl (124 mM), KCl (3 mM), KH2PO4 (1.2 mM), MgSO4 (1.2 mM), CaCl2 (2.0 mM), and D-glucose (10 mM). The HEPES medium was gassed with 100% O2 throughout the study and maintained at 37°C and at a pH of 7.4–7.5.

Brain slices from 14-d-old rat pups were perfused in a 500 MHz MR spectroscopy system for 8 h under the following conditions: i) at 37°C in KHB; ii) at 32°C in KHB; iii) at 37°C in HEPES buffer, and iv) with 1 mM amiloride at 37°C in HEPES buffer. Brain slices from 7-d-old rat pups were perfused i) at 37°C in KHB and ii) at 32°C in KHB. Interleaved 31P and 1H MR spectra were acquired in all studies in KHB. 31P MR spectra only were acquired in the amiloride studies because of overlap of 1H signals from HEPES buffer with the 1H metabolite resonances of interest.

All studies used were done in accordance with the Home Office guidelines and were consistent with the Animals (Scientific Procedures) Act, 1986 (UK).

Preparation and maintenance of brain slices under control conditions at 37°C in KHB.

Brain slices, 350 μm thick, were prepared from both cerebral hemispheres of 7- (n = 3) and 14-d (n = 3) Wistar rats of either sex using a McIlwain tissue chopper (Mickle Laboratory Engineering, Mill Works, Gomshall, Surrey, UK). For each experiment, two to five brains were used. Immediately after slices had been prepared they were placed in a flask containing 100 mL of KHB equilibrated with O2/CO2, 95%/5% at 37°C at a pH of 7.4–7.5. Slices were gently mixed and washed five times before decanting into a standard 15-mm diameter NMR tube containing 10 mL of the gassed medium at 37°C. The NMR tube was fitted with a polytetrafluorethylene insert carrying a custom-made superfusion apparatus (30).

The perfusion apparatus included two peristaltic pumps fitted with quick-release pump heads to make up a complete nonrecirculating perfusion circuit. The superfusion rate was adjusted to 25–30 mL/min to ensure gentle movement of the slice tissue and sufficient perturbation to prevent tissue clumping. The temperature of the slices was controlled at 37°C throughout the experiment by using the variable temperature control unit of the spectrometer and the heated water jacket for the perfusate. A third peristaltic pump controlled the rate of water flow through the water jacket; the temperature of the water in the water jacket was adjustable by using the water bath thermometer. The temperature of the KHB bathing the slices was monitored by a fiber-optic thermometer (Luxtron 3100, Luxtron, Santa Clara, CA). The pH of the KHB was monitored and maintained at around 7.5 throughout the study.

Initially, the slices were superfused at 37°C in KHB outside the bore of the magnet for 30–45 min. The apparatus was then placed within the bore of the magnet and one set of 31P and 1H NMR spectra in normal phosphate KHB was recorded. Pi was then omitted from the KHB and replaced with an equimolar amount of KCl to monitor the intracellular Pi concentration and the chemical shift of Pi, which may be used as an indicator of pHi using a standard calibration curve (31).

Stabilization was defined as 1.5 h after slicing and approximately 30 min after omission of Pi from the buffer. Slices were then perfused for a further 6.5 h, 8 h from the time of slicing.

Preparation and maintenance of brain slices under hypothermic conditions.

Brain slices, 350 μm thick, were prepared from both cerebral hemispheres of 7- (n = 5) and 14-d (n = 4) Wistar rats of either sex. The temperature of KHB was maintained at 32°C throughout the study. All other factors were identical to the preparation and maintenance of brain slices under control conditions.

Preparation and maintenance of brain slices with HEPES and amiloride.

Brain slices were prepared from 14-d (n = 5) Wistar rats of either sex and placed in a flask containing the nonbicarbonate buffer, HEPES. The HEPES medium was gassed with 100% O2 throughout the study and maintained at 37°C and at a pH of 7.4–7.5. In two experiments, 1 mM amiloride hydrochloride (Sigma Chemical, St. Louis, MO), initially dissolved in 50% DMSO, was added to the HEPES medium. Again, the HEPES medium was gassed with 100% O2 throughout the study and maintained at 37°C and at a pH of 7.4. For all these experiments, all other factors were identical to the preparation and maintenance of brain slices under control conditions.

MR methods.

1H and 31P MR spectra were obtained at 500 MHz (JEOL, Tokyo, Japan) ECP + 500 MHz NMR spectrometer) using an 11.7 T superconducting magnet (Oxford Instruments, Oxford, UK) and a dedicated Doty (Doty Scientific Inc., Columbia, SC) 15-mm high-resolution solution state probe with 1H decouple/multinuclear observe.

Magnetic field inhomogeneities were reduced by shimming on the water signal. The proton 90° pulse length was measured. Interleaved 31P and 1H MR spectra were then acquired for the duration of the study in KHB buffer and 31P MR spectra were acquired for studies using HEPES buffer. 31P MR spectra were acquired using pulse-collect techniques. A pulse angle of 45° was used with a repetition time of 1.8 s. For each spectrum 256 transients were averaged. T1 values of individual metabolites were assumed to be constant throughout the study. 1H MR spectra were acquired using a spin echo technique. The water was suppressed using presaturation techniques combined with T2 relaxation during evolution of the spin echo sequence. An echo time (TE) of 270 ms was used as a compromise between adequate water suppression, separation of lipid/lactate resonances, and metabolite signal loss due to T2 effects. The repetition time was 2.6 s and 256 transients were averaged for each spectrum.

31P MR spectra were processed using linear prediction singular value decomposition-based backward linear projection method (order of operation was 16; 128 points were sampled and 8 points were reconstructed) to reduce the contribution from the broad underlying resonance. A single exponential line broadening function of 20 Hz was then applied before Fourier transformation. Peak positions and integrals were measured using the system software (Delta 3.1, JEOL). On the basis of published chemical shift values, resonances were assigned PME, Pi, phosphodiesters (PDE), PCr, and NTP. The NTP peak area was measured from the sum of the composite peak areas of α, β, and γ NTP and therefore included any contribution from nucleoside diphosphates such as ADP.

Brain slice pHi was measured from the 31P MR spectrum. The chemical shift difference (δ) of the Pi resonance(s) was referenced to the PCr peak, and the pHi value(s) were calculated at both temperatures using the Henderson-Hasselbach equation

where δ is the Pi shift relative to PCr (31). The mean pHi values were also adjusted for temperature of study (32).

1H MR spectra were processed using a single exponential line broadening function of 2 Hz before Fourier transformation. An interactive baseline correction was applied before the peak positions and integrals were measured using the system software (Delta 3.1, JEOL). On the basis of published chemical shift values, resonances were assigned to choline-containing compounds (Cho), creatine plus phosphocreatine (Cr), N-acetyl aspartate (NAA), and lactate (9).

Statistical analysis.

All data in this study are presented as mean ± SD. Statistical analyses of the data were performed using SPSS, version 10.1 for Windows (SPSS Inc., Chicago, IL). Parametric data were analyzed using t tests. In one study, the sample size was two; because these data were obtained in an unbiased way from an underlying population that was normally distributed, it was legitimate to use t tests for comparisons involving this sample.

The stabilization point when the initial brain slice spectral parameters were recorded was 1.5 h after slicing. The stabilization baseline was assigned the value of 100 and was constructed by averaging metabolite ratios acquired between 1.5 and 2.5 h after slicing. To demonstrate any change in metabolite ratios over time, subsequent ratios acquired at 5 h were compared with this averaged baseline value using confidence intervals. The changes in the NTP/PME ratio over time for the three groups (37°C, 32°C, and 1 mM amiloride at 37°C) were compared using a multivariate repeated measures model with a Bonferroni correction. Two tests of within-subjects contrasts were also performed.

RESULTS

None of the data sets were skewed and therefore parametric analytic tests were used. The assumption of equal variances was not always valid when using the t test and, in these cases, an unequal variance modification was used.

Metabolite ratios values from 7- and 14-d models acquired after stabilization in KHB at 37°C and 32°C are shown in Table 1. Representative 31P and 1H MR spectra from the 7- and 14-d model at 37°C after stabilization in KHB are shown in Figure 1. Both the 7- and 14-d models demonstrated a high degree of metabolic integrity; mean PCr/Pi was >1.5 and lactate/NAA was <0.5 in both models. NAA/Cr increased with increasing maturity (p = 0.064 at 37°C and p < 0.001 at 32°C). PCr/Pi and NTP/PME were higher when the temperature was 32°C compared with 37°C in both the 7- and 14-d models, but this only reached significance for NTP/PME in the 7-d old model (p < 0.01).

Table 1 Metabolite ratios and pHi values at stabilization (1.5 hours after slicing) in the 7- and 14-day brain slice models at 37°C and 32°C and in the 14-day model in different buffers at 37°C
Full size table
Figure 1
figure 1

Representative 1H and 31P MR spectra at stabilization in the neonatal rat brain slice model at 37°C. The slices were perfused with KHB with no external phosphate added. (A,C) 1H MR spectra from 7- and 14-d brain slice model, respectively. The peak area ratio lactate/NAA was <0.5 in both models. NAA was more prominent in the 14-d model, reflecting an increase with maturation. (B,D) 31P MR spectra from 7- and 14-d brain slice model, respectively. In both spectra, there were two components to the Pi peak. The PME peak was the most prominent resonance in the 7-d model whereas the PCr peak was most prominent in the 14-d spectrum. Cr, creatine plus phosphocreatine; Glx, glutamine/glutamate; NAA, N-acetylaspartate.

Full size image

Metabolites ratios from the 14-d model acquired after stabilization in HEPES alone and HEPES with amiloride are also shown in Table 1. There were no significant differences in PCr/Pi and NTP/PME in HEPES buffer compared with the stabilization values in KHB. After stabilization in HEPES with amiloride, NTP/PME was significantly increased but there was no significant difference in PCr/Pi compared with values in HEPES alone and in KHB.

Brain slice pHi.

pH values from 7 and 14-d models acquired after stabilization in KHB at 37°C and 32°C and from the 14-d model in HEPES and HEPES with amiloride shown in Table 1. In KHB buffer with Pi omitted there were two conponents of the Pi peak; the peaks were assigned pHi (1) and pHi (2) (most alkaline and least alkaline, respectively). Brain pHi values were generally similar with increasing maturity at both 37°C and 32°C, although pHi (2) was significantly more alkaline in the 7-d-old model compared with the 14-d-old model at 32°C (p < 0.01). There was a trend toward a more alkaline pHi at 32°C compared with 37°C in both the 7- and 14-d models, which reached significance for pHi (2) in the 14-d old model (p < 0.05) and for pHi (1) in the 7-d model (p < 0.01).

In HEPES alone, a single Pi peak at 7.28 was detected in the 14-d old model at 37°C. In HEPES with amiloride a doublet Pi structure was detected, and pHi (2) was significantly less alkaline than the pHi in HEPES alone (p = 0.001).

Changes in brain slice energetics over time at 37°C and 32°C in 7- and 14-d models perfused with KHB with Pi omitted.

Taking the stabilized baseline value at 100 for each metabolite ratio, the metabolite ratios at 5 h at 37°C and 32°C in the 14-d models are summarized in Table 2.

Table 2 Brain slice energetics, taking the stabilized baseline as 100, and pH values at 5 hours the 14-day brain slice models perfused with KHB with Pi omitted at 37°C and 32°C and in HEPES with and without amiloride at 37°C
Full size table

At 37°C, PCr/Pi was significantly reduced at 5 h in both the 7- and 14-d models compared with baseline (Fig. 2, A and B). The value in the 14-d model at 5 h was not significantly different from PCr/Pi in the 7-d model (t = 2.61; df = 4; p = 0.06). Lactate/NAA was significantly increased at 5 h in the 14-d model (Fig. 3) but was not significantly different from baseline in the 7-d model. NTP/PME was not significantly different from baseline at 5 h in both 7- and 14-d models (Table 2). pH values were unchanged at 5 h compared with baseline values in both the 14-d (Fig. 4) and 7-d models.

Figure 2
figure 2

Changes in PCr/Pi over 8 h at 37°C (open circles; n = 3) and 32°C (closed circles; n = 4) in (A) the 7-d and (B) the 14-d neonatal rat pup brain slices perfused with KHB with no external phosphate added. A stabilization baseline value, assigned to 100, was achieved averaging metabolite ratios acquired between 1.5 and 2.5 h after slicing. After 5 h, PCr/Pi at 37°C lay outside the 99% confidence interval (p < 0.01) of baseline value, but at 32°C was not significantly different from baseline in both ages of model.

Full size image
Figure 3
figure 3

Changes in lactate/NAA over 8 h at 37°C (open circles; n = 3) and 32°C (closed circles; n = 4) in 14-d neonatal rat pup brain slices perfused with KHB with no external phosphate added. After 5 h, lactate/NAA was outside the 99.9% confidence interval (p < 0.001) of baseline value at 37°C but was not significantly different from baseline at 32°C.

Full size image
Figure 4
figure 4

Representative pHi values from the two Pi peak components of the 14-d brain slice model at 37°C over 8 h. There was clear pHi heterogeneity, as shown by pH (1) (open circles) and pH (2) (closed circles). No change in these pHi values occurred throughout the study.

Full size image

Effect of amiloride on changes in brain slice energetics over time at 37°C.

In the 14-d model, taking the stabilized baseline as 100, at 5 h the PCr/Pi was 68 in HEPES alone and 59 ± 11 in HEPES with amiloride; these values lay outside the 99% confidence interval of the stabilized baseline values.

Effect of hypothermia or amiloride on NTP/PME in the 14-d model.

Considering the individual metabolite ratios in the slices prepared from 14-d neonatal rat pups in KHB buffer with Pi omitted, at 5 h NTP/PME was higher at 32°C (1.81 ± 0.21) than at 37°C (0.93 ± 0.23, t = 4.45, df = 5, p = 0.01).

Comparing the NTP/PME ratio in KHB with Pi omitted and HEPES with amiloride, at 5 h the NTP/PME ratio was 0.93 ± 0.23 in KHB and 1.88 ± 0.23 in HEPES with amiloride (t = 3.51, df = 3, p = 0.03) (Figs. 5 and 6). The preservation of NTP/PME with amiloride (1.88 ± 0.23) at 5 h did not differ from that which occurred with hypothermia (1.81 ± 0.21, t = 0.30, df = 4, p > 0.1) (Fig. 6).

Figure 5
figure 5

A representative 31P MR spectrum of 14-d neonatal rat brain slices in oxygenated HEPES buffer with 1 mM amiloride HCl 5 h after slice preparation. Mean NTP/PME was 1.88.

Full size image
Figure 6
figure 6

NTP/PME levels over 8 h in 14-d slices perfused with oxygenated HEPES buffer alone at 37°C (open circles; n = 3), KHB with no external phosphate added at 32°C (closed circles; n = 4), and with HEPES buffer with 1 mM amiloride added at 37°C (open triangles; n = 2). The preservation in NTP/PME was similar in the amiloride-exposed slices and the slices exposed to hypothermic conditions.

Full size image

Repeated measure analysis using all data obtained over the 8 h demonstrated similar results. The three groups differed significantly (F = 62.67, df = 2,24, p < 0.0001). The two tests of within-subject contrasts demonstrated that the 37°C and 32°C groups differed (F = 100.65, df = 1,12, p < 0.0001) but that the 32°C and the 1 mM amiloride at 37°C groups did not differ significantly (F = 0.76, df = 1,12, p = 0.4).

DISCUSSION

Neonatal rat brain slices maintained for a number of hours showed changes in bioenergetics with time; under control conditions at 37°C neonatal brain slices were alkaline and underwent a gradual decline in PCr/Pi and rise in lactate/NAA over 8 h. Moderate hypothermia at 32°C delayed the decline in PCr/Pi and rise in lactate/NAA. Brain slice pHi acidified on exposure to amiloride, both at baseline and at 5 h. NTP/PME was preserved at lower temperatures and in slices exposed to amiloride, both at baseline and at 5 h.

Our findings differ from those from a previous study of neonatal brain slice energetics over a prolonged period (33). 31P and 1H metabolite concentrations were unchanged for >10 h in a small number of calibration and comparison studies using a perfusate free of inorganic phosphate; there is no obvious explanation for this difference.

Validity of brain slice model.

There are some drawbacks to the brain slice model. Firstly, it is inevitable that some slices were lost during 8 h of perfusion and occasional slices were seen in the waste fluid. To correct for the decline in tissue volume we assigned the metabolite ratio at 1.5 h after slicing as 100% and compared subsequent ratios. Secondly, KHB and HEPES buffers were gassed with 95% O2/5% CO2 and 100% O2, respectively, which may have generated suboptimal conditions due to the associated increased production of oxygen free radicals (34). Thirdly, amiloride was dissolved in DMSO and brain slices were exposed to a concentration of 0.01% DMSO over 8 h. Such a small concentration is unlikely to affect brain slice energetics, however, as DMSO is a free radical scavenger that in higher concentrations appears to preserve ATP after ischemia-induced damage in the liver (35), further studies should include controls exposed to DMSO alone to differentiate mechanisms of amiloride/DMSO treatment. Finally, the interrelationship of PCr, creatine kinase equilibrium, and pH is complex (36), so a comparison of PCr/Pi at different pH values needs caution. It was not possible to measure PCr/Pi within different pH compartments from a given spectrum. However, because the pH was constant over time at 32°C and 37°C in both the 7- and 14-d models, we argue that it is valid to interpret a change in PCr/Pi during a time course study.

Interpretation of the changes in bioenergetics under control conditions as secondary energy failure.

We speculate that primary energy failure occurs during decapitation and slicing of the brain slices. This phenomenon, “primary energy failure,” associated with brain slice preparation was demonstrated more than 50 y ago; after depletion of PCr and ATP during slicing, sliced brain tissue was able to resynthesize high-energy phosphates to around two-thirds of in vivo levels when placed in saline with glucose and oxygen as substrates (26). These experiments demonstrated that stable levels of PCr and ATP could be maintained for at least 2 h after preparation and that, during this time, the tissue was responsive to the metabolic demands of electrical pulses (37).

We speculate that the secondary decline in the bioenergetic status of brain slices that we have observed in our model over 8 h under control conditions is “secondary energy failure.” Secondary energy failure is a concept rather than a mechanism; although there are close associations between secondary energy failure and histologic damage (38), the precise mechanisms leading to secondary energy failure are unclear. It is likely that neurotoxic cascades are activated by both HI associated with the slice preparation and trauma to the cut edges of the brain tissue. We demonstrated that PCr/Pi declined and lac/NAA increased progressively over 6–8 h under control conditions despite adequate substrate and oxygen supply. These findings are similar to those that define the period of secondary energy failure in term infants after perinatal HI. The reduction in PCr/Pi and increase in lactate concentrations are thought to result from mitochondrial damage leading to impaired oxidative phosphorylation. The consequent rise in the intramitochondrial NADH/NAD+ ratio inhibits pyruvate dehydrogenase activity, resulting in less pyruvate entering the Krebs cycle and an increased conversion of pyruvate to lactate in the cytosol. The rate of lactate production is further increased by the alkaline pHi and its effect on the pKa of phosphofructokinase. The reduction in NAA reflects declining levels of neuronal viability and density.

The decline in energetics of the brain slice model is unlikely to be due to chronic ischemia, as we ensured constant perfusion and steady and gentle movement of slices, slices were of the optimal slice thickness to allow adequate perfusion of glucose and oxygen (<400 μm thick), and the brain slice pHi was alkaline.

Although the metabolic parameters during the evolution of secondary energy failure are similar in vivo and in the brain slice model, there are a number of factors that limit the strength of a brain slice/in vivo comparison. The reduced oxygen consumption of brain slices (40% of in vivo brain) (39,40) may be due to the absence of spontaneous neural activity in the preparation rather than a defect in the mitochondrial electron transport chain and a decreased oxygen uptake. In addition, the severity of the HI and traumatic insult in the brain slice model is profound; coupled with the lack of circulating trophic factors, the decline in energetics may be accelerated in the slice model.

Comparison of the effects of amiloride and hypothermia on brain slice energetics and pHi.

NTP/PME was preserved over time in brain slices exposed to both amiloride and hypothermia. PCr/Pi levels were unchanged compared with baseline in those exposed to hypothermic conditions, whereas PCr/Pi levels were significantly lower at 5 h on exposure to amiloride. Brain slice pHi at baseline and at 5 h was alkaline during hypothermia and acidic during exposure to amiloride. We were unable to measure relevant 1H MR spectra during exposure to amiloride because of spectral overlap with resonances from the buffer, so a comparison of changes in lactate/NAA was not possible.

The bioenergetic effects of hypothermia and amiloride differed in terms of pHi and PCr/Pi, yet both led to a preservation of NTP/PME over time. The mechanisms leading to preservation of energetics may therefore differ, particularly as hypothermia was associated with an alkalosis in brain slice pHi whereas exposure to amiloride was associated with a relative acidosis.

Hypothermia.

We demonstrated that the decline in PCr/Pi and rise in lactate/NAA were slower at 32°C compared with 37°C in both the 7 and 14-d brain slice models. A previous study in neonatal rat brain slices demonstrated that mild hypothermic conditions improved beta-ATP recovery after in vitro ischemia, but metabolites were measured for only about 2 h after the second insult so the effect on secondary energy failure could not be assessed (41).

In our study, the ameliorating effects of moderate hypothermia on the secondary decline in PCr/Pi and rise in lactate/NAA in brain slices were similar to the effects of moderate hypothermia seen in experimental models of transient HI, where 12 h of cooling in the neonatal piglet reduced the delayed impairment in cerebral energy metabolism and reduced histologic injury (42,43). A previous neonatal rat brain slice study demonstrated that both mild hypothermic and hypercarbic conditions applied after in vitro ischemia (following recovery from slicing) delayed the normalization of brain slice pHi. When these postischemic perturbations were applied in isolation, there was an improvement in recovery of beta-ATP (41). The mechanisms leading to more favorable slice energetics in this experimental model may thus also be associated with a sustained mildly acidotic pHi during the postischemic phase.

The protective effects of hypothermia may include a reduction in the delayed increase in extracellular glutamate (44), a decreased production of nitric oxide (45), a reduction in the number of depolarizations in injured but viable tissue (46), and a reduction in the number of cells undergoing apoptosis (47). In addition, a 5°C decrease in brain slice temperature may offer neuroprotection by enhancing the maintenance of brain ATP concentrations. A linear relationship exists between brain energy utilization rate and brain temperature; specifically, a 1°C reduction in brain temperature from normothermic values of 37°C leads to a 5.3% reduction in the brain energy utilization rate (48). Therefore, a 26.5% reduction in cerebral energy utilization that accompanied a 5°C decrease in brain temperature in our brain slice studies may explain the observed preservation in PCr/Pi and NTP/PME. A reduction of ATP is the hallmark of cerebral energy failure and is critical for initiating the cascade of cellular events such as loss of ionic gradients (49), release of excitatory neurotransmitters (44), and elevated intracellular calcium (50).

Amiloride.

The presence of amiloride in the HEPES buffer acidified brain slice pHi; this was likely to be due to inhibition of the activated Na+/H+ transporter. The excessive activation of the Na+/H+ transporter which is triggered by the intracellular acidosis of HI restores physiologic pHi by extruding H+. Although exposure of brain slices to amiloride had no effect on the decline in PCr/Pi at 5 h compared with HEPES alone or KHB at 37°C, NTP/PME was higher at 5 h in brain slices exposed to amiloride at 37°C compared with those slices exposed to normal HEPES buffer with no added amiloride and those in KHB. The preservation of NTP/PME was similar to that in those slices maintained under hypothermic conditions.

One of the mechanisms of neuroprotection by amiloride may be related to the change in pHi. An alkaline pHi may exacerbate injury during the evolution of secondary energy failure by the following mechanisms: i) Exacerbation of excitotoxic neuronal injury due to increased N-methyl-d-aspartate activation at alkaline pH (51,52) and activation of phospholipases and proteases, which have an alkaline pKa (53). Inhibition of Na+/H+ exchange has been shown to block phospholipase activation in the ischemic/reperfused rat cerebral cortex as a consequence of its stabilizing effect on acidotic pHi and preventing alkalosis (54). ii) Alteration of the delicate balance of pro- and anti-apoptotic triggers in mitochondria. Alkalinization has been shown to: a) activate Bax (a pro-apoptotic protein), inducing its mitochondrial translocation (55); b) inhibit the import of ADP into mitochondria, which impairs the synthesis of ATP (56); and c) transform the mitochondrial permeability transition pore into a high-conductance state, a critical step leading to cell commitment to death (57). Alkalinization has been described as a critical trigger in the apoptotic process (55).

Further mechanisms of neuroprotection by amiloride may relate to a reduction in energy consumption and calcium influx into the cell. For example, as well as influencing pHi, the excessive action of the Na+/H+ transporter after HI results in an increase in intracellular Na+ concentration and subsequent activation of the Na+/K+ ATPase, with a consecutive increase in energy consumption. The high intracellular Na+ level contributes to activate the sarcolemmal Na+/Ca++ antiporter, which leads to raised intracellular Ca++ (17,58). A variety of deleterious effects result from calcium overload, including mitochondrial accumulation of calcium, activation of proteolytic enzymes, free radical production, and lipolysis. Amiloride inhibits the Na+/H+ transporter and the Na+/Ca++ antiporter, potentially leading to an attenuation of the harmful consequences of excessive Na+/H+ transporter activity. In addition, the favorable brain slice energetics seen with amiloride exposure in our study may be due to a direct effect on NMDA receptors, either in addition to or independent of its effect on the Na+/H+ exchanger (55–59).

CONCLUSION

We have demonstrated that neonatal brain slices are alkaline and undergo a gradual decline in PCr/Pi and rise in lactate/NAA over 8 h under control conditions at 37°C. Our interpretation is that the brain slice model replicates secondary energy failure after a primary insult during decapitation and slicing of the tissue and that the resuscitation and reperfusion phases occurred when slices were placed in oxygenated KHB. Moderate hypothermia delayed the decline in PCr/Pi and rise in lactate/NAA, an effect typical of secondary energy failure. Up-regulation of the Na+/H+ transporter was implicated in the production of brain slice alkalosis, as amiloride, an inhibiter of the Na+/H+ transporter, acidified brain slice pHi. A preservation of NTP/PME was observed at lower temperatures and in slices exposed to amiloride. This preliminary observation in neonatal rat brain slices suggests that a delay in the pHi recovery or a more acidic pHi after HI was neuroprotective and suggests that studies examining the neuroprotective effects of amiloride after HI should be assessed in experimental models in vivo.