Corresponding author C. D. Benham: Department of Neuroscience Research, SmithKline Beecham Pharmaceuticals, New Frontiers Science Park, Third Avenue, Harlow, Essex CM19 5AW, UK. Email: firstname.lastname@example.org
1The effects of energy substrate removal and metabolic pathway block have been examined on neuronal and glial survival in organotypic slice cultures of rat hippocampus.
2Slice cultures resisted 24 h of exogenous energy substrate deprivation. Application of 0.5 mM α-cyano-4-hydroxycinnamate (4-CIN) for 24 h resulted in specific damage to neuronal cell layers, which could be reversed by co-application of 5 mM lactate.
3Addition of 10 mM 2-deoxyglucose in the absence of exogenous energy supply produced widespread cell death throughout the slice. This was partly reversed by co-application of 5 mM lactate.
4These effects of metabolic blockade on cell survival were qualitatively similar to the effects on population spikes recorded in the CA1 cell layer following 60 min application of these agents.
5The data suggest that monocarboxylate trafficking from glia to neurons is an essential route for supply of energy substrates to neurons particularly when exogenous energy supply is restricted.
Glucose is the preferred energy substrate for the adult brain. The blood-brain barrier (BBB) restricts the uptake of other plentiful potential substrates such as monocarboxylates (reviewed by Cremer, 1982), although recent evidence suggests that the permeability of the BBB to lactate is higher than previously estimated (Hassel & Brathe, 2000). In healthy individuals, plasma glucose is critically regulated to ensure adequate supplies to the CNS and other tissues. While the BBB restricts access to energy supplies other than glucose, the concept has developed that astroglia and neurons compose a functional energy supply unit, with most glucose utilisation occurring in the astroglia. Glia then export lactate as an energy source that can be utilised by neurons. Thus, once within the compartment defined by the blood-brain barrier the major energy currency for use by neurons is postulated to be lactate. This process has been termed the lactate shuttle and is particularly important during periods of energy demand (Tsacoupolos & Magistretti, 1996).
When normal physiological maintenance of blood glucose fails, as in patients with insulin-dependent diabetes mellitus who suffer an insulin overdose, severe hypoglycaemia leads to loss of conciousness and, rarely, coma and death. In addition, there is now increasing evidence that, repeated severe hypoglycaemic episodes, although apparently resolving with full recovery, do correlate with long term cognitive decline and loss of hippocampal volume as determined by magnetic resonance imaging (Perros et al. 1997). In addition to acute interruptions in glucose supply to the whole brain, at the cellular level impairment in glucose transport with ageing may result in slow neurodegeneration (Mark et al. 1997). Thus there is considerable interest in understanding the cellular mechanisms essential for energy utilisation in the adult CNS.
A number of functional studies have been performed on acute preparations of cortical and hippocampal slices to substantiate the lactate shuttle hypothesis. Acute cortical slice experiments have shown that lactate can replace glucose to maintain synaptic activity (Schurr et al. 1988) and that lactate supplies the increased energy demands that result from activation of excitatory glutamatergic pathways (Schurr et al. 1999). Further acute experiments have demonstrated a role for lactate in pathophysiological protocols. Lactate, rather than glucose, fuels the recovery of function from hypoxia (Schurr et al. 1997a) and glia are the source of lactate (Schurr et al. 1997b). Izumi et al. (1997) used metabolic pathway inhibitors to provide further support for an important role of lactate and other monocarboxylates in supporting synaptic transmission in conditions of glucose deprivation.
Substrate utilisation by the brain during development shows important differences. Suckling neonates are more able to utilise circulating monocarboxylates, such as lactate, in part due to upregulation of expression of monocarboxylate transporters in cerebrovascular endothelia immediately postnatal (Pellerin et al. 1998). However, relatively little is known of the functional expression of substrate utilisation pathways in the brain parenchyma during development and their physiological and pathophysiological importance. In infants, hypoglycaemia occurs as a result of a number of conditions of which the most common is persistent hyperinsulinaemic hypoglycaemia of infancy (PHHI), with an incidence of about 1:2500 newborns in Caucasian populations (reviewed by Aynsley-Green et al. 1998). Post mortem analysis of human infants and animal studies show characteristic patterns of neuronal cell death that differ from that resulting from hypoxia or ischaemia. For example, in the hippocampus, the dentate gyrus is much more sensitive than the CA1 and CA3 pyramidal cell layers to hypoglycaemic insults (Auer & Siesjo, 1988).
We have used organotypic hippocampal slices to examine the role of energy substrates on neuronal survival after 24 h of hypoglycaemia. Our aim was to investigate the contributions of the lactate shuttle to energy supply in the various neuronal cell layers of the hippocampus that differentially express monocarboxylate transporters (MCTs). In addition, we wanted to examine the importance of the lactate shuttle in a neonatal culture system, which might differ from adult tissue. The slice cultures are resistant to damage by short periods of glucose deprivation unless lactate transport is pharmacologically blocked, when pyramidal and granule cell death results. Our results suggest that the provision of lactate by glia as a neuronal energy substrate is an important facet of energy utilisation.
Organotypic hippocampal slice cultures were prepared as described by Stoppini et al. (1991). Neonatal Wistar rats (8-10 days old) were killed by decapitation, the hippocampus was rapidly dissected, and transverse sections (400 μm) were cut on a McIlwain tissue chopper and immediately placed into ice-cold Geys balanced salt solution (GBSS; Gibco, UK) supplemented with 5 mg ml−1 glucose. Slices were plated onto sterile, semiporous tissue culture inserts (Millicell-CM, Millipore, UK, four slices per well) and maintained in culture at 37°C in a 5 % CO2 atmosphere. The support medium consisted of 50 % minimum essential medium (MEM, Gibco, UK), 25 % Hanks’ balanced salt solution (HBSS; Gibco, UK) and 25 % heat-inactivated horse serum (ICN, UK) supplemented with 5 mg ml−1 glucose and 1 mM glutamine (Sigma). The medium was changed every 3 days and cultures were used for experiments after 14 days in vitro.
Slice cultures were then transferred into serum free medium (SF: 100 % MEM, 1 mM glutamine, 5 mg ml−1 glucose) containing 5 μg ml−1 propidium iodide (PI; Molecular Probes, USA) for 20 min and then imaged under a fluoroescence microscope fitted with a rhodamine filter. Any cultures exhibiting fluorescence at this stage were discarded from further study. Following this, cultures were placed into SF medium with 5 μg ml−1 PI and containing either no substrate (GFM: glucose-free medium); energy substrate (30 mM glucose, 5 mM lactate); metabolic inhibitor (10 mM 2-deoxyglucose, 0.25-5 mM 4-CIN, 10-70 μM quercetin); or a combination of metabolic inhibitor and substrate. After 24 h incubation under appropriate experimental conditions, cultures were imaged for PI fluorescence and then fixed in 4 % paraformaldehyde and stained with thionine. PI fluorescence was excited between 515 and 560 nm using a mercury vapour lamp and observed using a standard Leica inverted microscope fitted with a rhodamine filter. Images were captured using a monochrome CCD video camera and frame grabber, stored on either magnetic optical discs or on compact disc, and were later analysed using NIH Image 1.55 software. Quantification of cell damage was determined as previously described by Pringle et al. (1996) and results are expressed as the percentage area of each hippocampal subfield in which fluorescence was observed. Results are expressed as means ±s.e.m. Significance was determined by one way ANOVA followed by Dunnett's post-test.
Stocks (100 mM) of lactate, glucose, pyruvate and 2-deoxyglucose (Sigma) were all made up in GF medium to give either a 1 M or 100 mM stock so that the substrate was added to the medium to give no more than 1:10 dilution. α-Cyano-4-hydroxycinnamate (4-CIN) (Sigma, UK) was initially dissolved in heated methanol to give a 100 mM stock solution. This was then further diluted to either 1:5 or 1:10 in GF medium to give a stock for each experiment and then added to the medium in the wells at no more than 1:4 dilution. The use of methanol was controlled for by using methanol alone at the relevant dilutions in the medium and after the cultures were exposed to it for 24 h. Quercetin (Sigma, UK) was dissolved in 1 M NaOH to give a 100 mM stock and then further diluted in GF medium to give a 1 mM stock and a 10 mM stock. Quercetin was then added to the culture medium at no more than a 1:10 dilution. The pH of all solutions following changes in composition remained in the range pH 7.2 to 7.4.
To study the effects of lactate and metabolic inhibitors on synaptic function in organotypic hippocampal slice cultures, we recorded field potentials from the CA1 region in response to afferent stimulation of the Schaffer collateral commissural/associational pathway. The membrane of the Millicell culture insert on which the cultures were grown was carefully cut away from its support ring and placed in a Peltier device-controlled microscope stage incubator (Medical Systems Corp., USA) and cultures were perfused continuously with artificial cerebro-spinal fluid (ACSF; mM): NaCl 117.8, glucose 10, MgSO4 1, CaCl2 2.5, KCl 3.3, KH2PO4 1.26, NaH2PO4 26, pH 7.4, 306 mosmol l−1 maintained at 34 ± 0.5°C and oxygenated by bubbling through with 95 % O2-5 % CO2. A stimulating electrode consisting of a pair of 0.28 mm twisted polyamide-insulated steel wire (Plastics One, USA) was placed into stratum lucidum at the transition between CA3 and CA1 to stimulate the afferent fibres into the CA1 region. Glass microrecording electrodes were pulled on a standard electrode puller to a tip resistance of 5 MΩ when filled with 2.5 M NaCl. Recording electrodes were placed under visual guidance into the stratum pyramidale of the CA1 region. Single or paired 0.1 ms monopolar stimuli 0-20 V were delivered to the slice cultures at an interval of no less than 10 s. Field potential population responses were digitised (Axon Instruments, USA), recorded, stored and analysed using software developed in-house (A2D, E. Stockley, Southampton University). Population spike amplitudes and field EPSP rise times were measured in response to increasing afferent stimulus intensity (stimulus-response relationship) and a measure of recurrent inhibition was also determined by delivering paired stimuli (15 ms apart). Cultures that did not display paired-pulse inhibition or which had population spikes of less than 2 mV at a stimulus of 20 V were discarded from the study.
To study the effects of substrate modulation or metabolic inhibition on synaptic function, paired stimuli of equal intensity were delivered to the slice cultures at an intensity which was set at just below the intensity required to produce a maximal response as determined from the stimulus-response curve (typically 10 V). Experiments were started at t = -15 min to determine and record a stable baseline to be established before starting the experimental procedure at t= 0. At this point, the perfusion was switched from the control glucose-containing ACSF to the appropriate experimental condition. Responses at the appropriate stimulus intensity were obtained every 5 min throughout the course of the experiment. At the end of the insult period, the perfusion was switched back to glucose-containing ACSF and the recovery of field potentials was monitored for a further 30 min.
Our initial experiments focused on determining the sensitivity of the slice cultures to interruption of supply of energy generating substrate. Glucose deprivation for 24 h was not toxic (Fig. 1). This observation suggests that there are significant energy stores in these cultures that are utilised in the absence of exogenous substrates. As neurons do not store significant energy substrates this implies that energy must be supplied by astrocytes and other satellite cells present in the culture. We investigated the role of monocarboxylate (principally lactate and pyruvate) transport by blocking MCTs with the selective inhibitor 4-CIN. 4-CIN treatment (0.5 mM) for 24 h resulted in a significant level of damage in CA1, CA3 and DG regions of the cultures. 4-CIN treatment in the presence of exogenous glucose had no effect indicating that this concentration of 4-CIN was not acting as a non-specific cytotoxin, nor was it primarily acting as a mitochondrial pyruvate transport blocker (Halestrap & Denton, 1974; Halestrap, 1975) (Fig. 1). As 4-CIN is a reversible, competitive, substrate site directed inhibitor of MCT we next checked that exogenous lactate could compete with the inhibitor. Co-application of 5 mM lactate with 0.5 mM 4-CIN resulted in less damage than 4-CIN alone, as expected if 4-CIN can be competitively displaced. Damage in CA3 was reduced almost to the unmeasurable level seen in the absence of 4-CIN (Fig. 1). Lactate replacement alone did not cause any damage (data not shown), and thus increasing exogenous lactate is not toxic per se.
We next looked at the effects of varying concentrations of 4-CIN on the amount of damage at 24 h in the absence of glucose (Fig. 2). The data show a concentration-dependent increase in damage at 24 h, with the CA1 and dentate granule cell layers showing greater sensitivity to 4-CIN than the CA3 region. Damage became maximal at 2.5-5 mM. This shows that 0.5 mM 4-CIN has a submaximal effect which supports the explanation that partial reversal of the 4-CIN effect by lactate was due to substrate competition. Attempts to produce similar damage with quercetin, 5-fold more potent than 4-CIN on MCT2 (Broer et al. 1999) were confounded by quercetin auto-fluorescence and anti-oxidant effects in the same concentration range (Gao et al. 2000) that might be protective. Quercetin (70 μM) induced modest damage in CA1 (11.7 ± 4.8 %; n= 24, P > 0.05, relative to control).
The resistance of these cultures to glucose deprivation suggests they have significant metabolisable energy stores. We tested the role of glycolysis in supplying this energy by blocking glycolysis with 10 mM 2-deoxyglucose in the absence of glucose. This treatment resulted in maximal damage in all neuronal regions indicating an important role for glycolysis in maintaining neuronal viability in the absence of simple substrates. Addition of exogenous lactate should circumvent the glycolytic blockade by 2-deoxyglucose by supplying substrate downstream of the block. Addition of 5 mM lactate with 2-deoxyglucose did protect significantly compared with 2-deoxyglucose alone and this protective effect of 2-deoxyglucose was prevented by addition of 4-CIN (Fig. 3).
The ability to image the whole slice culture enabled us to further examine the regional specificity of the various insults. Control serum free cultures and cultures exposed to 0.5 mM 4-CIN in the presence of glucose showed no significant PI signal in any region of the slice (Fig. 4a). Images taken from slice cultures exposed to 10 mM 2-deoxyglucose in glucose-free medium for 24 h show that a PI signal was evident in all regions of the culture and not limited to the neuronal cell layers (Fig. 4B). In contrast, the damage evoked by 0.5 mM 4-CIN incubation was confined to the CA1, CA3 and dentate granule neuronal cell layers (Fig. 4C).
The substrate dependence of synaptic function has been explored in acute slice preparations (Izumi et al. 1997) but not in these organotypic cultures, so we were interested to see the acute effects of substrate manipulation on synaptic function. We recorded population spikes in the CA1 region following Schaffer collateral pathway stimulation in slice cultures exposed for 60 min to substrate changes and/or inhibitor (Fig. 5). As expected from the cell viability measures, glucose removal for 1 h had little effect, and replacement with 5 mM lactate completely prevented the trend to run-down of population spike amplitude seen in the absence of glucose. Both 2-deoxyglucose application in the absence of glucose and 4-CIN application in the absence of glucose caused a depression in population spike amplitude that reached greater than 50 % inhibition after 60 min. As with the cell viability measures at 24 h, co-application of lactate was able to partially reverse the effects of 4-CIN (Fig. 5).
The results presented here can be accommodated by the basic model of metabolic substrate supply interactions in neuroglia proposed by Tsacopoulos & Magistretti (1996). This is outlined in Fig. 6 and emphasises the major role of lactate derived from glial glycogen and glucose as a neuronal energy source. Although, the earliest experimental support was obtained using the honey bee retina as a model where lactate is not the exchange currency (Tsacopoulos et al. 1988), subsequently, supporting data have been generated in mammalian CNS tissue (Schurr et al. 1988; Izumi et al. 1997). Our data clearly show that inhibition of the monocarboxylate transporter is toxic to neurons in the absence of exogenous energy substrates when energy must be supplied from stores. This toxicity can be prevented either by lactate or glucose addition (Fig. 1). This is consistent with the proposal that firstly, energy derived from energy stores is supplied to neurons via monocarboxylate transport as shown in pathway (3) in Fig. 6 and that this pathway may be the most significant route of energy transfer from these stores. Secondly, exogenous glucose is directly available as a substrate to glia and possibly neurons. The fact that glucose can rescue from MCT block might be explained two ways. Glucose could be directly taken up by neurons. Alternatively, it could be transported into glia and processed into lactate in sufficient concentration to overcome MCT block. As Fig. 2 shows that 0.5 mM 4-CIN is submaximal, any lactate production by glia in addition to glycogen store-derived lactate should immediately start to reverse the block. Finally, glucose transfer from glia to neurons (2) is not significant as otherwise MCT block, in the absence of added energy substrates, would not be toxic.
Exogenous lactate rescues neurons from 4-CIN toxicity (Fig. 1). This could be due to straight competition on the transporter or it could mean that 4-CIN is more potent on the glial MCT exporting lactate and is less effective at preventing exogenous uptake into neurons through the neuronal MCT. If the latter were the case, then exogenous lactate would effectively bypass the site of blockade (Fig. 6). This would require expression of two different MCT isoforms in the appropriate locations. Of the eight known MCTs, protein distribution patterns are published for only MCT1 and MCT2 both of which are expressed in the CNS (reviewed by Halestrap & Price, 1999). In addition MCT7 and MCT8 mRNAs are relatively abundant in the human brain (Price et al. 1998). Localisation data and the kinetic properties of the transporters are consistent with MCT1 being the major glial MCT. MCT2 or another isoform may be the predominant form on neurons (Halestrap & Price, 1999). In situ hybridisation data show strong expression of MCT2 mRNA in CA1 and DG cell body layers in the hippocampus (Pellerin et al. 1998). This would suit flux of lactate predominantly from glia to neurons using the higher affinity for lactate of MCT2 (Broer et al. 1998, 1999). 4-CIN blocked MCT2 about 20-fold more potently than MCT1 when the rat transporters were expressed in Xenopus oocytes. IC50 values were 0.024 mM for MCT2 and 0.42 mM for MCT1 with 0.05 mM lactate (Broer et al. 1999). Thus, in our experiments, relief of 4-CIN block of MCT1 by 5 mM lactate will be greater, but taking into account the 100-fold lower substrate concentrations used in the oocyte experiments, MCT2 block is also likely to be relieved. Thus, simple substrate competition is the most parsimonious explanation of this data rather than effects on the glial MCT. Further dissection of the roles of different MCT isoforms will depend on more specific pharmacological blockade or gene knockout experiments.
2-Deoxyglucose would be expected to be taken up both by glia and neurons and thence to block glycolysis. This might be expected to be equally damaging to both neurons and glia (Fig. 6). Our observation of non-specific damage in the presence of 2-deoxyglucose is thus consistent with this model. This result also further emphasises the specific nature of the damage induced by MCT inhibition. In the presence of 4-CIN, lactate supply is cut off to neurons but the glia are largely unaffected. This is consistent with the energy supply in the absence of exogenous substrate originating from glycogen stores in glia which would be unaffected by this manoeuvre.
The reduced sensitivity of synaptic transmission to removal of exogenous substrate in these slice cultures is in contrast to the sensitivity of acute adult hippocampal slices (Schurr et al. 1988). This might reflect larger energy stores in the slices as a consequence of the time in tissue culture or the lack of maturity of the tissue, or a more flexible use of other metabolic substrates in these cultures. The dependence of neuronal viability on MCT transport, suggests that a larger glycogen store in glia or more generous glial:neuronal ratio is the most parsimonious explanation. Glial fibrillary acidic protein (GFAP) stained cultures show a high number of astrocytic processes within the culture and astrocytes can be seen to envelop the entire culture suggesting that they are present in increased numbers compared with acute slices (Pringle et al. 1997).
The intact brains of neonates, unlike adults, are able to utilise the rich circulating supply of lactate, pyruvate and other substrates at birth and whilst suckling (Cremer, 1982; Vicario et al. 1991; Nehlig & de Vasconcelos, 1993). This transient adaptation to monocarboxylate substrate supply in blood is in part achieved by a transient increase in expression of MCT1 in brain capillary endothelial cells (Pellerin et al. 1998). Our data suggest that the subsequent handling of substrates is very similar to that of adult brain parenchyma, with the lactate shuttle between glia and neurons playing a prominent role. The comparability of metabolic state of these slice cultures to the in situ properties of neonatal hippocampus has been poorly explored. Other properties examined show morphological and functional maturation at the same rate as in situ (reviewed by Gahwiler et al. 1997). Thus these cultures probably best represent the developmental status in situ at P15-25.
This new work highlights the role of the MCT family of transporters as key proteins in the maintenance of energy supply to neurons. Given the long term deleterious effects of metabolic stress on neurons and the likely association with neurodegenerative conditions (Mark et al. 1997), examination of this family of proteins might be a productive area to search for linkage with genetic diseases. The genes for MCT1 and MCT2 are localised on chromosome 1p13.2 (Garcia et al. 1994) and 12q13 (Lin et al. 1998). No specific disease linkages to these regions or to those of other known MCTs have been reported to date in the On-Line Mendelian Inheritance in Man (OMIM) database (http:www.ncbi.nlm.nih.govomin). Disruption at some point in this pathway may be an important source of metabolic stress leading to neuronal loss in neurodegenerative diseases.