Monocarboxylate transporters in the central nervous system: distribution, regulation and function

Authors


Address correspondence and reprint requests to Luc Pellerin, Institut de Physiologie, 7 rue du Bugnon, 1005 Lausanne, Switzerland.
E-mail: Luc.Pellerin@unil.ch

Abstract

Monocarboxylate transporters (MCTs) are proton-linked membrane carriers involved in the transport of monocarboxylates such as lactate, pyruvate, as well as ketone bodies. They belong to a larger family of transporters composed of 14 members in mammals based on sequence homologies. MCTs are found in various tissues including the brain where three isoforms, MCT1, MCT2 and MCT4, have been described. Each of these isoforms exhibits a distinct regional and cellular distribution in rodent brain. At the cellular level, MCT1 is expressed by endothelial cells of microvessels, by ependymocytes as well as by astrocytes. MCT4 expression appears to be specific for astrocytes. By contrast, the predominant neuronal monocarboxylate transporter is MCT2. Interestingly, part of MCT2 immunoreactivity is located at postsynaptic sites, suggesting a particular role of monocarboxylates and their transporters in synaptic transmission. In addition to variation in expression during development and upon nutritional modifications, new data indicate that MCT expression is regulated at the translational level by neurotransmitters. Understanding how transport of monocarboxylates is regulated could be of particular importance not only for neuroenergetics but also for areas such as functional brain imaging, regulation of food intake and glucose homeostasis, or for central nervous system disorders such as ischaemia and neurodegenerative diseases.

Abbreviations used
CHC

α-cyano-4-hydroxycinnamate

CNS

central nervous system

DBDS

4,4′-dibenzamidostilbene-2,2′-disulphonate

DIDS

4,4′-diisothiocyanostilbene-2,2′-disulphonate

GluR

glutamate receptor

MAP2

microtubule-associated protein 2

MCT

monocarboxylate transporter

pCMBS

p-chloromercuribenzene sulphonate

PSD

postsynaptic density

TAT1

T-Type amino-acid transporter 1

UTR

untranslated region

VMH

ventromedial hypothalamus

Monocarboxylates such as lactate, pyruvate and ketone bodies appear to play an important but incompletely understood role in brain energy metabolism. This is particularly true for lactate, which has long been considered a waste product to be disposed of via the circulation, when not assimilated to a potentially toxic compound for the brain, and certainly a sign of tissue suffering. It is not surprising that in these conditions, little attention was paid to the presence of specific transporters for these substances on cells within the central nervous system (CNS). Fortunately, our view about the importance of monocarboxylates for sustaining brain functions has changed in the last few years (see Pellerin 2003), bringing new attention on monocarboxylates and their transporters. Rapid progress has been made on the identification and the distribution of different transporter subtypes in the brain, and regulation of their expression has just started to be explored. This is an exciting time because we have just begun to unravel the putative functions of monocarboxylates and their transporters in areas as diverse as synaptic transmission and neuronal survival. Some of these different aspects pertaining to monocarboxylate transporters and their presence in the CNS will be reviewed here.

Importance of monocarboxylate utilization by the brain

Monocarboxylates, together with other non-glucose substrates, have been known for some time to represent substantial energy substrates for the developing brain (Nehlig and Pereira de Vasconcelos 1993; Erecinska et al. 2004). For example, lactate is elevated in the newborn blood immediately after delivery, and constitutes an important brain energy substrate in the first few hours after birth (Dombrowski et al. 1989). In addition, acetoacetate and β-hydroxybutyrate, two ketone bodies formed by the hepatic oxidation of fatty acids contained in the maternal milk, represent significant energy substrates for the developing brain during the preweaning period (Hawkins et al. 1971; Cremer 1982). Significant monocarboxylate utilization by the brain was also reported in different pathological states such as diabetes, prolonged starvation or under hypoglycaemia during adulthood (Gjedde and Crone 1975; Fernandes et al. 1982; Hawkins et al. 1986). Moreover, under normal physiological conditions, it was shown that elevated circulating lactate levels, occurring for example during intense exercise, can provide an energy source readily used by the brain (Dalsgaard et al. 2003; see Nybo and Secher 2004 for a comprehensive review on this topic). Such a role for blood-borne lactate was revealed both by the measure of arteriovenous differences (Ide et al. 1999, 2000), as well as by the reduction in cerebral glucose utilization detected with fluorodeoxyglucose coupled to positron emission tomography (Smith et al. 2003). These data are supported by a number of NMR studies showing that lactate is readily oxidized within the brain once it has passed the blood–brain barrier (Bouzier et al. 2000; Hassel and Brathe 2000). In addition, evidence has been provided that lactate can sustain neuronal activity (Schurr et al. 1988; Izumi et al. 1997; Takata et al. 2001; Sakurai et al. 2002; Cater et al. 2003), even if some discrepancies on the extent of its efficacy have been reported (see Bouzier-Sore et al. 2002 for a critical review). For example, replacement of extracellular glucose with sodium lactate or sodium pyruvate sustained synaptic transmission after transient glucose depletion (Sakurai et al. 2002). Lactate was even demonstrated to have a neuroprotective effect under excitotoxic or after ischaemic conditions (Schurr et al. 1999, 2001; Cater et al. 2001, 2003; Ros et al. 2001; Bliss et al. 2004).

At the cellular level, several reports have demonstrated that neurones can efficiently use monocarboxylates as energy substrates (reviewed in Bouzier-Sore et al. 2002; Pellerin 2003). Thus, lactate can be oxidized in a variety of neuronal preparations such as brain slices (Ide et al. 1969; Fernandez and Medina 1986), cultured telencephalic neurones (Vicario et al. 1991; Tabernero et al. 1996; McKenna et al. 2001), aggregated neuronal cultures (Honegger et al. 2002), sympathetic ganglia (Larrabee 1983, 1992, 1995, 1996), as well as synaptosomes (McKenna et al. 1993, 1994, 1998; Waniewski and Martin 2004). It was shown that, despite the fact that lactate transport is more rapid in astrocytes than in synaptosomes (but of lower affinity), lactate is metabolized to CO2 more rapidly in synaptosomes than in astrocytes (Waniewski and Martin 2004). Moreover, even if cultured neurones were shown to spontaneously produce and release some lactate (Walz and Mukerji 1988a), they exhibited a preferential use of lactate over glucose as substrate for oxidative metabolism (Bouzier-Sore et al. 2003; Itoh et al. 2003). By contrast, astrocytes produce large amounts of lactate (Walz and Mukerji 1988a,b) and they most likely represent the major source of lactate among brain cells. Under certain conditions, it was shown that astrocytes can use some lactate as an energy substrate, but to a much lesser extent than neurones and oligodendrocytes (Bouzier et al. 1998; Itoh et al. 2003; Waniewski and Martin 2004), and they can utilize lactate as a precursor for gluconeogenesis and glycogen synthesis (Dringen et al. 1993a). These pathways, however, are probably of minor significance as compared to the prominent lactate production. Lactate release by astrocytes was found to be stimulated upon exposure to glutamate (Pellerin and Magistretti 1994; Akaoka et al. 2001). It was also shown that glycogenolysis activated for example by neurotransmitters such as noradrenaline, vasoactive intestinal peptide and adenosine would enhance lactate release (Magistretti et al. 1993). Moreover, it was shown that glycogen mobilization and the release of a monocarboxylate, most likely lactate, is essential to sustain action potential propagation in axons (Brown et al. 2003). The importance of astrocytic glycogen as a provider of lactate and its role in brain energy metabolism has received renewed attention recently (reviewed in Brown 2004).

These data highlight the fact that lactate and other monocarboxylates are used as energy substrates by brain cells, and in particular by neurones. The conditions under which lactate may become a significant energy substrate and the specific contribution of glucose and lactate metabolized by each cell type remain unanswered questions (Pellerin and Magistretti 2003). Nevertheless, under certain circumstances, lactate as well as other monocarboxylates from the systemic circulation are taken up by the brain. In addition, some parenchymal cells, such as astrocytes, produce and release large amounts of lactate within the brain. Thus, blood-borne monocarboxylates must cross the blood–brain barrier to reach the brain and enter brain cells to be used as substrates, whereas lactate produced within the brain could be exchanged between brain cells. Because monocarboxylates are hydrophilic substances and do not cross membranes easily, specific transport systems must exist to carry them across the blood–brain barrier and within each cell type, or to transfer them from one cell type to the other when monocarboxylates are produced locally.

The monocarboxylate transporter family

Identification, cloning and sequencing of the different monocarboxylate transporters (MCTs) was extensively reviewed previously (Halestrap and Price 1999; Enerson and Drewes 2003) and only some aspects will be considered here. The monocarboxylate transporter family (also termed the SLC16 gene family) is composed of 14 members based on sequence homologies, identified as MCT1–9, MCT11–14 and T-Type amino-acid transporter-1 (TAT1) (Halestrap and Meredith 2004). However, it was recently demonstrated that MCT8 is a thyroid hormone transporter instead of a monocarboxylate transporter (Friesema et al. 2003). Indeed, only the first four MCTs (MCT1–4) were shown so far to cotransport monocarboxylates and protons by a symport mechanism with an equimolar (1 : 1) stoichiometry (Carpenter and Halestrap 1994; Garcia et al. 1994, 1995; Takanaga et al. 1995; Tamai et al. 1995; Bröer et al. 1997, 1998, 1999; Yoon et al. 1997; Lin et al. 1998; Dimmer et al. 2000; Manning Fox et al. 2000). By determining the effects of pH on the kinetics of lactate flux, transport was shown to follow an ordered, sequential mechanism (De Bruijne et al. 1983, 1985). Transport involves first binding of a proton, followed by the anionic form of lactate. Translocation of the monocarboxylate and the proton occurs next across the membrane, followed by the release of the monocarboxylic anion from the transporter on the other membrane side. This process is freely reversible and the rate-limiting step for net monocarboxylic acid flux is the return of the unloaded carrier on the initial side. Transport can be enhanced both by a decrease of pH on the same side where the monocarboxylate is applied, and by an increase of pH on the opposite side. Moreover, lactate flux can be accelerated if the substrate concentration is increased on the opposite side (Garcia et al. 1994; Juel 1996; Dimmer et al. 2000).

MCT1–4 were shown to transport several endogenously produced monocarboxylates (lactate, pyruvate and ketone bodies) as well as less common monocarboxylates such as acetate, propionate and butyrate (Tamai et al. 1995; Carpenter and Halestrap 1994; Broër et al. 1997, 1998, 1999; Dimmer et al. 2000; Grollman et al. 2000). MCT1 transports many short-chain monocarboxylates with Km values decreasing as the chain length increases. Km values of ∼3.5 mm for lactate and ∼1.0 mm for pyruvate were reported (Broër et al. 1998). MCT1 is stereospecific for lactate, with a Km for the l-isomer being lower than for the d-isomer (Carpenter and Halestrap 1994; Takanaga et al. 1995; Broër et al. 1998). Stereospecificity was not observed for other monocarboxylates such as 2-chloropropionate or β-hydroxybutyrate. MCT1 is reversibly inhibited by some 4,4′-substituted stilbene-2,2′-disulphonate compounds, such as 4,4′-diisothiocyanostilbene-2,2′-disulphonate (DIDS) or 4,4′-dibenzamidostilbene-2,2′-disulphonate (DBDS). Irreversible inhibition is caused by various thiol and amino reagents, including the potent organomercurial thiol reagent p-chloromercuribenzene sulphonate (pCMBS). Several aromatic monocarboxylate derivatives act also as competitive inhibitors of MCT1 transport. They include 2-oxo-4-methylpentanoate, phenylpyruvate and derivatives of α-cyanocinnamate such as α-cyano-4-hydroxycinnamate (CHC) (Poole and Halestrap 1993; Halestrap and Meredith 2004; references therein). Caution should be exerted, however, when drawing conclusions from the use of such inhibitors, as they have also been shown to block the mitochondrial pyruvate carrier with an even higher affinity (Halestrap and Denton 1975). MCT2 displays the lowest Km values, ∼0.7 and 0.08 mm for lactate and pyruvate, respectively (Broër et al. 1999). This observation is in line with a role for MCT2 in tissues (liver or kidney) or cells (neurones) exhibiting important monocarboxylate uptake and/or consumption. Like MCT1, MCT2 is stereospecific for lactate, with a higher affinity for the l-isomer (Broër et al. 1999). MCT2 is sensitive to the same inhibitors as MCT1, including CHC, DBDS and DIDS, but is not inhibited by pCMBS (Garcia et al. 1995; Bonen et al. 2000). Despite its lower affinity, d-lactate has been used as a competitive inhibitor of neuronal l-lactate uptake, presumably via MCT2 (Ros et al. 2001), although a clear validation of its specificity and efficacy is still awaited. MCT3 exhibits a Km value for lactate in the same range as MCT1 (Grollman et al. 2000). MCT3 is sensitive to a diethyl pyrocarbonate inhibition but insensitive to CHC and pCMBS (Grollman et al. 2000). The isoform MCT4 has been identified as a major lactate transporter of white (glycolytic) muscle cells (Wilson et al. 1998; Pilegaard et al. 1999) and astrocytes (Bergersen et al. 2001, 2002; Rafiki et al. 2003; Pellerin et al. 2005). Such a distribution is consistent with its characteristic as a low-affinity, lactate-preferring transporter, which is adapted to the release of lactate from glycolytic cells (Dimmer et al. 2000). Moreover, the very low affinity of MCT4 for pyruvate (Km, 150 mm) prevents the loss of pyruvate from the cell and thus permits high rates of glycolysis and cytosolic ATP production (Manning Fox et al. 2000). The low affinity of MCT4 for other substrates such as ketone bodies is also consistent with its presence in glycolytic tissues that do not oxidize or produce these metabolites (Manning Fox et al. 2000). Little inhibition by DIDS or CHC is observed for MCT4 at concentrations giving more than 50% inhibition in the case of MCT1 (Halestrap and Meredith 2004).

Distribution of MCT1, MCT2 and MCT4 in the rodent central nervous system

The first four MCTs (MCT1–4) can be distinguished not only by their kinetic properties but also by their tissue distribution and their cellular localization. MCT1 was found to be present in almost all tissues including muscle, kidney, liver and heart (see Fig. 1), generally with specific cellular expression within each tissue (Koehler-Stec et al. 1998; reviewed in Halestrap and Price 1999). Western blot analysis revealed that MCT1 encodes a protein with an apparent molecular mass of ∼45 kDa (Poole and Halestrap 1992). By contrast, MCT2, which exhibits an apparent molecular weight of ∼40–43 kDa (Bergersen et al. 2001; Pierre et al. 2003), is expressed in fewer tissues, like the liver, kidney or testis (Fig. 1; Garcia et al. 1995; Jackson et al. 1997; Koehler-Stec et al. 1998; Boussouar et al. 2003). MCT3, with a molecular weight of about 43 kDa is exclusively expressed in the retinal pigment epithelium (Philp et al. 1998, 2001), whereas MCT4 (∼43 kDa; Bergersen et al. 1999) is strongly expressed in skeletal muscle (Pilegaard et al. 1999; reviewed in Juel and Halestrap 1999).

Figure 1.

Western blot analyses of MCT1 and MCT2 expression in different mouse tissues. The anti-MCT1 antibody labelled a single band at around 43 kDa in muscle, brain, kidney, liver and heart homogenates. Lung and stomach homogenates displayed only a faintly labelled band. The anti-MCT2 antibody recognized a band of about 43 kDa in fewer tissues than MCT1: it was found only in brain, kidney and liver homogenates.

In the CNS, monocarboxylate carriers were first functionally identified on endothelial cells of the blood–brain barrier (Cremer et al. 1976; Cornford et al. 1982; Cornford and Cornford 1986; Terasaki et al. 1991). They were shown also to be present on parenchymal cells in vivo (Kuhr et al. 1988) as well as in hippocampal slices (Assaf et al. 1990), whereas cultured neurones and astrocytes were found to exhibit functionally distinct monocarboxylate carriers (Tildon and Roeder 1988; Dringen et al. 1993b; Nedergaard and Goldman 1993; Tildon et al. 1993, 1994). In the last few years, expression of MCT1, MCT2 and MCT4 has been studied in the rodent brain, and their distribution at the regional and cellular levels has been characterized as described below. A summary of their distribution is provided in Table 1. However, before going into a detailed description of their regional and cellular distribution, a general note of caution is necessary. The distribution of each MCT in the CNS is based on either the presence of mRNA (by in situ hybridization), protein expression (by immunohistochemical techniques), or both. A number of discrepancies have been observed between different studies and/or techniques, and several reasons can explain these differences. Among them, the specificity of each antibody used to identify the MCTs is of utmost importance and it turned out that it was not the same for all antibodies that have been used (see Bergersen et al. 2001 for a discussion). Species differences is another factor to take into account. Finally, discrepancies between mRNA and protein expression are not uncommon and may indicate regulation at the translational level. In the description below, we have attempted to outline conflicting data and offer some possible explanations.

Table 1.  Regional distribution and cellular localization of monocarboxylate transporters (MCTs) in the adult rodent central nervous system
MCT isoformRegional distributionCellular localization
In vivoIn vitro
MCT1Widespread. Neuropil labelling with a few entirely labelled cellsEndothelial cells forming blood vessels, ependymocytes, astrocytes. Some neurons in ratAstrocytes, weak in a few neurons
MCT2Widespread but strong expression in cortex, hippocampus, cerebellum. Neuropil labelling and negative somata except Purkinje cellsNeurons, some astrocytes in rat (white matter)Neurons. Astrocytes also from rat
MCT4Widespread but strong cellular expression in cortex, hippocampus, striatum, cerebellumAstrocytesNot determined

MCT1

MCT1 expression, both at the mRNA and protein levels, is homogenously distributed throughout the whole rodent brain. The mRNA coding for MCT1 was found to be abundant in the cortex, the hippocampus and the cerebellum of young (15-day-old) and adult rodent brain; it partly colocalized with an endothelial cell marker at postnatal day 15 but was detected only in parenchymal cells in the adult brain (Pellerin et al. 1998a; Vannucci and Simpson 2003). These parenchymal cells were identified as astrocytes using reverse transcriptase–polymerase chain reaction method (Mac and Nalecz 2003).

In young and adult rodent brain, MCT1 protein was abundantly expressed in endothelial cells forming microvessels and ependymocytes lining the four brain ventricles (Gerhart et al. 1998; Hanu et al. 2000; Pierre et al. 2000; Baud et al. 2003). MCT1-positive glial end-feet surrounding capillaries and glial-like processes were also visible using light and electron microscopy (Gerhart et al. 1997; Leino et al. 1999). Initially, it had been observed that fewer parenchymal cells contained MCT1 protein as compared to mRNA, suggesting possible translational control (Gerhart et al. 1997; Koehler-Stec et al. 1998; Leino et al. 1999; Pierre et al. 2000). Confocal microscopy using double labellings with glial markers such as glial fibrillary acidic protein (GFAP) or S100β confirmed association of MCT1 immunoreactivity with astrocytic processes, both in vitro (see Figs 2a1 and a2) and in vivo (Hanu et al. 2000; Pierre et al. 2000; Ainscow et al. 2002; Debernardi et al. 2003). MCT1 immunoreactivity was present in the cytoplasm of cultured astrocytes, both in the soma and processes, and was also clearly associated with their plasma membrane (Figs 2a1 and a2). In the case of neurones, contrasting results were obtained by separate groups using different antibodies in two species. Immunoreactivity for MCT1 in neurones was never observed in the adult mouse brain (Pierre et al. 2000), whereas a faint expression was found in cultured mouse cortical neurones, although it remained undetectable using western blotting analysis on the same culture extracts (Debernardi et al. 2003). By contrast, MCT1 immunoreactivity was visible in a few scattered neurones in the rat brain (Leino et al. 1999) and also in cultured hypothalamic neurones that were heavily labelled (Ainscow et al. 2002; Kang et al. 2004).

Figure 2.

MCT1 and MCT2 expression in cultured astrocytes and neurones, respectively, from mouse cortex (modified from Debernardi et al. 2003 with permission). (a1, a2) Double immunolabelling of primary cultures of mouse cortical astrocytes after 14 days in vitro with antibodies directed against MCT1 (green) and S100β (red). MCT1 immunoreactivity was visible in all S100β-positive cells (a1, a2). MCT1 immunoreactivity was distributed in the cytoplasm and was also associated with the plasma membrane (arrows). (b1, b2) Double immunolabellings of primary cultures of mouse cortical neurones after 7 days in vitro with antibodies against MCT2 (green) and microtubule-associated protein 2 (MAP2) (red). MCT2 immunoreactivity was found in virtually all MAP2-positive cells. A strong punctiform immunoreactivity was visible in the cytoplasm of somata and dendrites, as well as on their plasma membrane (arrows). MCT1 and MCT2 were revealed with a FITC-conjugated secondary antibody, S100β and MAP2 with a Texas Red-conjugated secondary antibody. Observation with immunofluorescence viewed with appropriate filters.

MCT2

Northern blot and in situ hybridization studies showed that MCT2 mRNA is abundant in the cortex, the hippocampus and the cerebellum of the mouse brain (Bröer et al. 1997; Koehler-Stec et al. 1998; Pellerin et al. 1998a; Debernardi et al. 2003; Vannucci and Simpson 2003). At the cellular level, the mRNA coding for MCT2 had a predominantly neuronal pattern of expression in young as well as in adult rodent brain (Pellerin et al. 1998a; Vannucci and Simpson 2003). The use of mouse cultured cells confirmed that MCT2 mRNA was absent from astrocytes, whereas it was strongly expressed in neurones (Bröer et al. 1997; Debernardi et al. 2003). Expression of MCT2 mRNA in capillaries remains controversial, as it was detected in endothelial cells in certain (Koehler-Stec et al. 1998; Pellerin et al. 1998a; Mac and Nalecz 2003) but not all studies (Vannucci and Simpson 2003).

Expression of the MCT2 protein follows the distribution pattern of its mRNA, as it is strongly expressed in the cortex, the hippocampus and the cerebellum of the rodent brain (Bergersen et al. 2001, 2002; Pierre et al. 2002; Rafiki et al. 2003). Moreover, MCT2 is the major neuronal tranporter in the rodent brain. Cortical neurones in culture exhibited MCT2 immunoreactivity at the surface and in the cytoplasm of their cell bodies and neurites (Figs 2b1 and b2). In vivo, MCT2 immunoreactivity was visible as a strong, punctiform labelling surrounding immunonegative neuronal somata (Fig. 3a). Double labelling experiments using the neuronal marker microtubule-associated protein 2 (MAP2) showed a colocalization between MCT2 and MAP2-positive processes, both in vitro and in vivo (Figs 2b1 and b2 and Fig. 3b). Moreover, intense MCT2 immunoreactivity was found in cerebellar Purkinje cell bodies and their processes, as well as on mossy fibers in the cerebellum (Figs 3c and d). Apart from dendritic arborization and in addition to mossy fibers in the cerebellum, MCT2 appears to be present on many axonal projections, supporting a putative role in lactate transfer into axons to fuel action potential propagation (Brown et al. 2003). It was particularly visible on sensory fibers in the brainstem as well as on facial and trigeminal nerves (Figs 3e and f). Except in glia limitans, MCT2 protein expression was never detected in glial-like cells of the mouse brain, whereas it could be visible in some endothelial cells forming brain capillaries (Pierre et al. 2000, 2002). In the rat brain, however, contradictory results have been obtained between groups using different antibody sources. The commercial antibody (Chemicon, Temecula, CA, USA) raised against the last 15 amino acids of the C-terminal portion of MCT2 showed that MCT2 immunoreactivity was restricted to astrocytes (Gerhart et al. 1998; Hanu et al. 2000). This was in contrast to results obtained with two antibodies developed in two separate laboratories and raised also against the same C-terminal 15 amino acids of MCT2. Studies with these antibodies revealed that MCT2 was predominantly expressed in neurones throughout the whole brain and particularly in the hippocampus as well as in the cerebellum (Pierre et al. 2000, 2002; Bergersen et al. 2001, 2002; Rafiki et al. 2003), even if it could be detected in some astrocytic-like cells or identified astrocytes (Rafiki et al. 2003; Baud et al. 2003). In that case, MCT2-positive astrocytes were located generally in the white matter, and MCT2 was also found in astrocytic end-feet specializations on blood vessels. Western blotting revealed that the Chemicon antibody gave rise to a relatively weak band at the appropriate molecular weight for the MCT2 protein, but also to some extra bands including a low-molecular weight band in the brain and liver protein extracts from rat (Bergersen et al. 2001). By contrast, other MCT2 antibodies, including an antibody raised against a GST fusion protein containing the last 53 amino acids of MCT2, revealed a single band at the appropriate molecular weight (Bergersen et al. 2001). Reasons for such discrepancies are not known with certainty but one can not exclude that the different antibodies recognize different parts of the C-terminal amino-acid sequence. Moreover, the phosphorylation status of the MCT2 C-terminal sequence (which includes several serine and one threonine residue) may differ between glial and neuronal cells, and differentially affect antibody binding.

Figure 3.

Neuronal MCT2 expression in the mouse brain (modified from Pierre et al. 2002 with permission). (a) The mouse cortex is heavily labelled for MCT2. Labelling occurs in the neuropil around immunonegative somata. (b) Several MCT2-positive puncta (green) are associated with neuronal processes as they colocalize with microtubule-associated protein 2 (MAP2) (red) using double immunofluorescent labelling (overlay of the two antigens in yellow). (c and d) Proteinase K pretreatment of mouse cerebellum sections (see Pierre et al. 2002 for more details) led to an intense MCT2 immunoreactivity of Purkinje neurones. MCT2 immunolabelling is visible in the cytoplasm of their soma and dendritic processes. Moreover, immunoreactivity for MCT2 is visible in fibers (probably mossy fibers) going through the granular cell layer (c, arrows). (e and f) Strong immunoreactivity for MCT2 was observed in sensory fibers of the brainstem, such as the facial (e) and the trigeminal nerves (f). (a, c, e and f) Immunoperoxidase viewed with light microscopy. (b) MCT2 was revealed with a FITC-conjugated secondary antibody, MAP2 with a Texas Red-conjugated secondary antibody; double optical projection made with confocal microscopy. (d) MCT2 was revealed with a FITC-conjugated secondary antibody; observation with immunofluorescence microscopy, using appropriate filters.

Accumulating evidence suggests that part of MCT2 expression is associated with synapses. First, observation in developing neurones in cultures following double immunofluorescent labellings with the presynaptic marker synaptophysin and MCT2 revealed a parallel distribution (Debernardi et al. 2003). More specifically, although no colocalization of MCT2 immunoreactivity with presynaptic elements could be observed, they were found to be closely apposed, suggesting a postsynaptic localization for MCT2. In the mouse brain cortex, a similar observation was made, and a colocalization with PSD95, the major component of the postsynaptic density (PSD), was seen occasionally (Pierre et al. 2002). Moreover, MCT2 immunoreactivity is present in the postsynaptic membrane of parallel fibre-Purkinje cell synapses in the rat cerebellum (Bergersen et al. 2002) and colocalizes with postsynaptic δ2-glutamate receptors as revealed by electron microscopy (Bergersen et al. 2001). Recently, MCT2 was shown to be located at the PSD of distinct glutamatergic synapses in the hippocampus as well as in the cerebellum of the rat and mouse brain, respectively (Bergersen et al. 2005).

Interestingly, based on expressed sequence tag (EST) database and northern blotting, very little if any MCT2 could be detected in the human brain as well as in human liver (Price et al. 1998), in contrast to the strong expression observed in these two organs in rodents. This observation highlights the importance of species differences for MCT expression. Considering the purported role of MCT2 for lactate metabolism in rodent neurones, the question of the identity of the transporter in human for this important neuronal energy substrate remains open.

MCT4

MCT4 expression has been extensively studied in the skeletal muscle, where it is known to be responsible for the export of lactate from glycolytic muscle fibers (Juel and Halestrap 1999; Bonen et al. 2000; Dimmer et al. 2000). In rodent brain, however, few studies have been conducted but they all indicate that MCT4 is expressed exclusively in astrocytes. Immunoreactivity for MCT4 was found in astrocytic processes in the adult rat cerebellum where both Bergmann glia in the molecular layer and astrocytes in the granular layer were labelled (Bergersen et al. 2001, 2002; Rafiki et al. 2003). Likewise the labelling in the hippocampus and in the corpus callosum was restricted to astrocytes (Rafiki et al. 2003). MCT4 was present in many astrocytes throughout the whole mouse brain, following a pattern of distribution similar to that obtained in the rat brain with intensely labelled astrocytes visible in the cortex, the striatum and the hippocampus for example (Fig. 4).

Figure 4.

Expression of MCT4 by astrocytes in various rat brain regions (taken from Pellerin et al. 2005 with permission). Immunolabellings carried out with an antibody against MCT4 show immunopositive astrocytes in the cortex (top left panel and inset at higher magnification), hippocampus (bottom left panel), paraventricular nucleus (PVN; top right panel), and capsula internalis (bottom right panel). Immunoperoxidase immunoreactivity viewed by light microscopy.

Regulation of monocarboxylate transporter expression in the brain

The brain has the ability to adjust its supply of monocarboxylates to meet specific energy requirements, under physiological (Cremer 1982; Dombrowski et al. 1989; Vicario et al. 1991) as well as pathological conditions (Hawkins et al. 1986; Schurr et al. 1999, 2001). A remarkable example takes place during the early postnatal development. It was determined that monocarboxylate uptake increases sevenfold during that period in parallel with a similar enhancement of their utilization by the brain (Cremer et al. 1976; Cornford et al. 1982; Edmond et al. 1985; Cornford and Cornford 1986; Nehlig et al. 1991; Nehlig and Pereira de Vasconcelos 1993). It is expected that the expression of MCTs both on endothelial cells forming blood vessels as well as on parenchymal cells would be regulated in the developing brain, as the availability of these major energy substrates vary considerably during peri- and postnatal development. Indeed, there is evidence suggesting that MCT expression effectively varies during that period. In rat brain ependymocytes, MCT1 and MCT2 protein expression was strong at late embryonic stages, decreased around birth and increased again during the early postnatal period (Baud et al. 2003). In endothelial cells constituting blood vessels, both MCT1 mRNA and protein expression were found to be 25-fold higher in 17-day-old rats when compared to adults (Leino et al. 1999). Concerning parenchymal cells, MCT1 and MCT2 mRNA levels increased postnatally to reach their highest levels around postnatal day 15. Their expression decreased after weaning but remained elevated at adulthood (Pellerin et al. 1998a; Vannucci and Simpson 2003). It was also shown that mature astrocytes expressed increasing levels of MCT1 in the postnatal rat brain, and that there was a transient increase of MCT2 expression (Baud et al. 2003). MCT4 expression by astrocytes was found to be sparse in the early postnatal period but was enhanced strongly at postnatal day 14 in rat hippocampus and cerebellum (Rafiki et al. 2003). The presence of MCT1 on glia limitans was detected starting at embryonic day 16 (Baud et al. 2003) and remained throughout adulthood in both rat and mouse (Gerhart et al. 1997; Pierre et al. 2000). These changes in expression of various MCTs on different cell types concur with the formation of the blood–brain barrier and the gradual switch from a prominent use of blood-borne monocarboxylates to glucose utilization by the brain. In addition, the observation of sustained mRNA and protein expression for several MCTs on parenchymal cells after weaning age reinforces the view of intercellular exchanges of lactate taking place within the adult brain.

If both lactate and ketone bodies are known to play a crucial role as energy substrates for the perinatal brain, there is now strong evidence that they can represent also appreciable energy substrates for the adult brain under specific circumstances. In these conditions, it is likely that modifications in the expression of the various monocarboxylate transporters might take place to accompany changes in energy substrate utilization. In adult rat, a ketogenic diet was found to induce brain MCT1 expression (Leino et al. 2001). In addition, evidence has been provided for modifications of MCT expression in pathological situations. MCT1 expression became patchy at the surface of CA1 pyramidal cells 21 days post-ischaemia in rats (Tseng et al. 2003), whereas it was enhanced in endothelial cells, other neurones, astrocytes and adjacent ependymal linings following the ischaemic insult. The promoter of the MCT1 gene has been recently cloned and studied (Cuff and Shirazi-Beechey 2002) but novel information about how it controls MCT1 expression has not emerged yet. No splice variants have been identified for MCT1, but the large increase in MCT1 mRNA expression occurring in the newborn animal, accompanied by an increase in protein levels, would suggest the possibility of a transcriptional regulation. The 3′-untranslated region (UTR) of MCT1 is very long (1.6 kb) and may play a role in translational regulation either by looping back to interact with the UTR or by binding to regulatory factors or binding proteins making the mRNA unavailable for translation (Miyamoto et al. 1996; Halestrap and Price 1999). Up to date, however, specific modulators of MCT1 expression in each brain cell type remain to be identified.

Recently, both MCT1 and MCT2 were shown to be expressed by mouse cultured cortical neurones. In these cultured cells, the level of MCT2 expression was correlated with synaptogenesis, suggesting that its expression could be regulated by factors in relation with synaptic activity (Debernardi et al. 2003). MCT2 expression in neurones appears to be under the control of neuroactive substances. Thus, it was shown to be enhanced by noradrenaline, in cultured cortical neurones, via the activation of a cAMP-mediated pathway. Moreover, regulation of MCT2 expression occurred at the translational level (Pierre et al. 2003). Discrepancies in the distribution of MCT2 mRNA and protein in different organs including the brain had previously led to the suggestion that MCT2 expression could be regulated at the level of translation (Jackson et al. 1997; Lin et al. 1998; Halestrap and Price 1999). In addition, the presence of multiple mRNA transcripts for MCT2 (Koehler-Stec et al. 1998; Pellerin et al. 1998a) raises the possibility that tissue-specific, post-transcriptional regulation of MCT2 expression may occur through alternative splicing within the 5′- or 3′-UTRs, leading to differences in translation efficiency (Lin et al. 1998).

Concerning MCT4, eight different mRNAs have been detected that are thought to arise from alternative splicing (Enerson and Drewes 1999), suggesting the possibility of transcriptional regulation, although at the protein level it was detected so far only as a single molecular mass in different organs (Dubouchaud et al. 2000; Zhao et al. 2001). Although MCT4 expression was shown to vary during developmental stages, there is actually no example of regulation of its expression in the adult brain under either physiological or pathological conditions.

Association with other proteins and putative implications

Extensive colocalization between MCTs and specific proteins has been reported previously and in some cases shown to be due to tight protein–protein interactions (Kirk et al. 2000; Philp et al. 2003). Thus, MCT1 was shown to be specifically linked to GP-70, a 70 kDa glycoprotein belonging to the immunoglobulin superfamily related to basigin (Poole and Halestrap 1997). MCT1 and MCT4 were shown to colocalize with the CD147/Basigin glycoprotein [also called extracellular matrix metalloproteinase inducer (EMMPRIN), HT7 or OX-47 in the rat] in pancreatic cells (Zhao et al. 2001), and both coimmunoprecipitated with CD147 in isolated heart cells (Kirk et al. 2000). It was reported that targeting of MCT1 and MCT4 to the plasma membrane was facilitated by their association with CD147/Basigin (Kirk et al. 2000; Philp et al. 2003). MCT2 did not interact with CD147, although its expression at the cell surface also requires an ancillary protein, not yet identified (Kirk et al. 2000).

In the CNS, colocalization has been observed between MCTs and synaptic proteins or membrane components. Targeting of MCT1 to specific cellular membrane locations was shown for MCT1 in the choroid plexus (Leino et al. 1999), but there is actually no data reporting that it may be due to an association with CD147 or an other chaperone molecule. MCT2 was shown to colocalize in PSD's with PSD95 (Pierre et al. 2002), with δ2-glutamate receptors of parallel fibre–Purkinje cell synapses (Bergersen et al. 2001), and with glutamate receptor (GluR)2/3 subunits of AMPA receptors (Bergersen et al. 2005). MCT2 was expressed also on vesicular structures similar to those containing GluR2/3 receptor subunits. However, there is no evidence at the moment for a direct interaction between MCT2 and any of these PSD components.

Functional significance and perspectives

Although the role of monocarboxylates, and particularly lactate, as essential energy substrates for the brain in development is now well-established, the same role in the adult brain is still a subject of debate (for review, see Gladden 2004). However, there is now strong evidence that extracellular lactate can be used by neurones (Waagepetersen et al. 1998; Schurr et al. 1988, 1999; Bouzier et al. 2000; Qu et al. 2000), and even represents a preferential oxidative energy substrate over glucose for neurones in culture (Bouzier-Sore et al. 2003; Itoh et al. 2003). Significant expression of MCT1 on cerebral vessels during adulthood suggests that monocarboxylates could cross the blood barrier and become substantial energy substrates for the brain. This would be particularly important in circumstances where glucose availability is reduced, for example during hypoglycaemia, or when blood lactate levels are particularly elevated, for example following exercise. Importance of monocarboxylate use as energy substrates is recognized when glucose, the most important energy fuel, is reduced in some altered metabolic situations such as diabetes, starvation (Pollay and Stevens 1980; Nehlig 1997) or in pathological conditions such as ischaemia (Inao et al. 1988; Frerichs et al. 1990; Schurr et al. 1999, 2001). Enhancing their uptake and utilization by the brain via modulation of MCT expression on both blood vessels and parenchymal cells could become a valuable neuroprotective approach (Sapolsky 2003). A proof of principle has been provided already, as it was shown that transfection of MCT2 in cultured neurones offers neuroprotection against excitotoxicity (Bliss et al. 2004).

The hypothalamus, and particularly the ventromedial hypothalamus (VMH), is suggested to play a central role in food intake control. It contains glucose-sensing neurones that can trigger hypoglycaemic counterregulation (Biggers et al. 1989). Such neurones alter their activity in response to changes in glucose levels. In vitro studies have shown that the by-product of glucose metabolism, lactate, can influence the behaviour of glucose-sensing neurones from various brainstem and hypothalamic nuclei including the VMH (Xang et al. 1999; Himmi et al. 2001; Mobbs et al. 2001; Song and Routh 2005). In parallel, lactate perfusion in the VMH was shown to suppress counterregulatory responses induce by hypoglycaemia (Borg et al. 2003). The expression of the monocarboxylate transporter MCT1 by VMH neurones (Ainscow et al. 2002) is consistent with the concept that the VMH acts as a fuel sensor rather than strictly as a glucose sensor. Such ideas have given rise to novel hypotheses concerning mechanisms controlling energy homeostasis for the entire body that take into account the role of both circulating and locally produced monocarboxylates on brain regulation of energy resource allocation (Peters et al. 2004).

A new concept in neuroenergetics has emerged in the last few years whereby astrocytes, which have a high glycolytic capacity, enhance their lactate production with increased neuronal activity (Pellerin and Magistretti 2004). Coupled with the capacity of neurones to use lactate as an energy substrate, these observations have given rise to the astrocyte-neurone lactate shuttle hypothesis, which has received experimental support recently (Kasischke et al. 2004). The discovery of several monocarboxylate transporter isoforms that are expressed by different cell types in the CNS has opened up several new perspectives in our understanding of exchanges between the different cell types within the brain. Considering the specific kinetic characteristics of each MCT, with MCT2 displaying the lowest Km making it particularly suitable for uptake, the observed distribution of MCT1 and MCT4 (astrocytic) as well as of MCT2 (the predominant neuronal transporter of monocarboxylates) is consistent with the proposed concept of lactate transfer between astrocytes and neurones (Pellerin et al. 1998b; reviewed in Bouzier-Sore et al. 2002). MCT1 and MCT4 would be essential to ensure lactate release in the extracellular space, allowing cells to maintain their high glycolytic rate that would cease if lactate was accumulating inside the cell. Lactate produced and released by astrocytes, accumulating to form an extracellular pool, then becomes available to serve as an additional energy substrate for active neurones via its uptake through MCT2 transporters that are widely expressed on neurones both on dendrites and axons as described previously (Bergersen et al. 2001; Pierre et al. 2002; Vannucci and Simpson 2003). Such a scenario would be consistent with many observations showing that lactate, together with glucose, sustains neuronal activity including action potential propagation (Schurr et al. 1988; Takata et al. 2001; Sakurai et al. 2002; Brown et al. 2003).

Observations that MCT2 is expressed within the PSD area of several glutamatergic, but not GABAergic, synapses in the cortex, the hippocampus and the cerebellum (Bergersen et al. 2001, 2005; Pierre et al. 2002) raise the question whether this synaptic localization is necessary to ensure an adequate supply of energy substrates to postsynaptic terminals upon excitation. Furthermore, it leads to the possibility that expression of MCT2 could be regulated in order to adjust supply of monocarboxylates in register with demands. Consistent with this idea, the neurotransmitter noradrenaline was shown to enhance MCT2 expression in cultured neurones (Pierre et al. 2003). In addition, the parallel distribution of MCT2 with AMPA receptor GluR2/3 subunits in the same vesicles within spines bearing glutamatergic synapses suggests that MCT2 might participate to the same translocation processes that AMPA receptors undergo at these synapses (Bergersen et al. 2005). If it was demonstrated to be the case, this would provide an elegant mechanism to further tie changes in activity linked to synaptic plasticity with energy metabolism.

Progress in our understanding of the roles of monocarboxylate transporters in the CNS might provide new insights in different areas of neuroscience. Apart from the obvious implications in neuroenergetics, changes in expression of MCTs could alter the metabolic responses of a brain region upon activation, leading to modified brain imaging signals that are based on metabolic parameters (Bonvento et al. 2002). A specific role of monocarboxylate transporters in fuel sensing and possible control of food intake as well as body mass is also emerging. Moreover, a variety of pathological states are characterized by metabolic perturbations of the CNS (e.g. Alzheimer's disease) and it is likely that MCTs could be involved either as contributing factors or as part of adaptive mechanisms. Thus, the study of monocarboxylate transporters in the CNS, just like in other tissues (Halestrap and Meredith 2004), might be of prime importance to fully appreciate how energy metabolism represents a critical factor for brain function.

Ancillary