Monocarboxylates such as pyruvate, lactate, and the ketone bodies (acetoacetate and β-hydroxybutyrate) play essential roles in carbohydrate, fat, and amino acid metabolism and must be rapidly transported across the plasma membrane of cells (1, 2). Transport is mediated by proton-linked monocarboxylate transporters (MCTs), four of which (MCTs1–4) have been characterized in detail and will be the major focus of this review. They are part of a family of transporter proteins, known as the MCT or SLC16 solute carrier family, that share characteristic sequence motifs. All family members are predicted to have 12-transmembrane helices (TMs) with intracellular C- and N-termini and a large cytosolic loop between TMs 6 and 7. As in other major facilitator superfamily members, the TM regions are more conserved than the loops and C-terminus [see (2–4)]. The MCT family has a total of 14 members whose predicted phylogeny is shown in Fig. 1 but only MCTs1–4 have been confirmed to function as proton-linked MCTs. MCT10 was identified as an aromatic amino acid transporter originally called T-type amino acid transporter 1 (TAT1) (5) and MCT8 shown to be an important thyroid hormone transporter (6). Transport mediated by MCT8 and MCT10 is not proton linked. The function of the other eight MCTs is unknown (2, 4). In addition to their normal metabolic roles, which are summarized in Fig. 2, MCTs may also be important for the transport of some drugs across the plasma membrane (4). There are also two distinct sodium-linked monocarboxylate transporters that are members of the SLC5 solute carrier family. These play a key role in endothelial monocarboxylate transport in the gut and kidney and will not be discussed here but have been well reviewed elsewhere (7).
This article will briefly review what is known of the structure and properties of the well-characterized members of the MCT family: MCT1 (SLC16A1), MCT2 (SLC16A7), MCT3 (SLC16A8), MCT4 (SLC16A3), MCT8 (SLC16A2), and MCT10 (SLC16A10). A subsequent article (8) will review their tissue specific metabolic roles and regulation. At the outset, the reader should note that the nomenclature of the MCT family is confusing, as in many cases, the MCT and SLC16 numbering do not coincide. This discrepancy arose because the MCTs were named in order of their characterization at the functional level, while the SLC16 numbers were generated as the cDNA sequences became available. The reader is referred to a previous review (3) for additional information on how these nomenclature issues arose fuller account of the current nomenclature of all members of the SLC16/MCT family with details of the chromosomal location and gene structure of the human MCTs is in (2). The accompanying article will review the metabolic roles and regulation of members of the MCT family (8).
CHARACTERIZATION OF THE DIFFERENT MCT FAMILY MEMBERS
The existence of a proton-linked monocarboxylate transporter was first demonstrated in this laboratory by showing that transport of L-lactate and pyruvate into human red blood cells was specifically inhibited by α-cyano-4-hydroxycinnamate (CHC) (9). The substrate and inhibitor specificity and detailed kinetics of this transporter were extensively characterized in both the author's and Deuticke's laboratories [see (1, 10)]. Subsequently, the molecular identity of the protein responsible was established by protein purification and sequencing in this laboratory (11) and independently at the cDNA level by Christine Kim Garcia in the laboratory of Goldstein and Brown. These authors named the transporter MCT1 (12). MCT1 is found in the great majority of tissues in all species studied, with no evidence for splice variants within the coding region (2, 4).
Characterization of the properties of MCT1 was initially performed in human and rat erythrocytes using radioactive techniques [see (1, 10)] and confirmed in a mouse tumor cell line by monitoring changes in intracellular pH using the fluorescent pH indicator 2′-7′-bis-(carboxyethyl)-5-6-carboxy-fluorescein (BCECF) (13). Subsequently, MCT1 was expressed in Xenopus laevis oocytes that exhibit no significant endogenous MCT activity and its activity again determined either radioactively or by monitoring transport-mediated changes in intracellular pH with BCECF or pH microelectrodes (14, 15). All these techniques confirmed that MCT1 demonstrates Michaelis Menten kinetics with a broad specificity for short-chain monocarboxylates including those substituted on the 2 and 3 positions with small groups such as halides, hydroxyl, and carbonyl groups as illustrated in Table 1. In addition, the transport of unsubstituted short-chain fatty acids, such as acetate, propionate, and butyrate, is strongly facilitated by MCT1, but these substrates can also enter cells rapidly by free diffusion of the undissociated acid [see (1, 10)]. Natural occurring substrates for MCT1 include L-lactate, pyruvate, β-hydroxybutyrate, and acetoacetate (1, 2) and Km values for these substrates are within the range found physiologically (1, 8). More hydrophobic ketoacids derived by transamination of amino acids may also be transported by MCT1; these include phenylpyruvate (from phenylalanine), α-ketoisocaproate (from leucine), α-ketoisovalerate (from valine), and α-keto-β-methylvalerate (from isoleucine). However, the hydrophic side chain of these substrates impairs the release of the bound substrate following transport resulting in very slow rates of net transport of these substrates by MCT1; indeed, they act as potent competitive inhibitors of the transport of other monocarboxylates (13). Interestingly, the transport of lactate is relatively stereoselective, with D-lactate being a poor substrate compared to L-lactate, whereas such stereoselectivity is not demonstrated for 2-chloropropionate or β-hydroxybutyrate. By far, the predominant role of MCT1 is to facilitate unidirectional proton-linked transport of L-lactate across the plasma membrane. This may represent either influx or efflux of lactic acid depending of the prevailing intracellular and extracellular substrate concentrations and the pH gradient across the plasma membrane. Net rates of transport of any monocarboxylate will be determined by the difference between influx and efflux, and at thermodynamic equilibrium, the concentration ratio of monocarboxylate inside the cell to outside the cell is equal to the ratio of [H+]out to [H+]in. However, MCT1 can also exchange one monocarboxylate for another without net movement of protons [see (1–3)].
Table 1. Km Values of different MCT isoforms for a range of monocarboxylates
MCT1 tumor cells
Data are for Km values (mM) of endogenous MCT1 in tumor cells or for MCT1, MCT2, and MCT4 expressed in Xenopus oocytes as indicated and are taken from ref. 14 where further details may be found. The L-lactate Km for MCT3 was measured following expression in Yeast and taken from ref. 16.
In addition to inhibition by competing monocarboxylates such as those described above (Substrate Specificity section), numerous nonphysiological competitive inhibitors of MCT1 have been described. These include CHC analogs and stilbene disulfonates such as 4,4′-di-isothiocyanostilbene-2,2′-disulfonate (DIDS) and 4,4′-dibenzamidostilbene-2,2′-disulfonate (DBDS) [see (1, 10)]. These agents have been used in some published experiments as specific MCT1 inhibitors without recognizing their ability to act as significant powerful inhibitors of other transporters. Thus, CHC is at least two orders of magnitude more potent as an inhibitor of the mitochondrial pyruvate transporter than of MCT1, whereas DIDS and DBDS inhibit the chloride/bicarbonate exchanger AE1 more powerfully than MCT1. The noncompetitive inhibitor phloretin and the arginine and cysteine reactive reagents phenylglyoxal and p-chloromercuribenzenesulfonate (pCMBS) [see(1, 10)] are equally nonspecific.
A new class of specific and high-affinity inhibitors of MCT1 have recently been developed by AstraZeneca with Ki values in the nM region (17, 18). We have confirmed the potent inhibition of MCT1-mediated L-lactate transport into rat erythrocytes by one of these inhibitors, AR-C155858, and determined a Ki of about 2 nM (19). This very high-affinity enabled us to determine the number of molecules of MCT1 per erythrocyte (80,000) and the turnover number (kcat) of the transporter (12 s−1 at 6°C). We also characterized the inhibition by AR-C155858 of different MCT isoforms expressed in Xenopus oocytes and found it to be active against MCT1 and MCT2 but not MCT4. Comparison of the time dependence of inhibition following extracellular application or intracellular injection of AR-C155858 suggested that it binds to MCT1 from the cytosolic side. The use of chimeric transporters combining different domains of MCT1 and MCT4 revealed that the binding site for the inhibitor is contained within the C-terminal half of MCT1, and involves TM domains 7–10 (19). Subsequent studies on the inhibition of MCT2 by AR-C155858 revealed that inhibition of this isoform only occurred when MCT2 is associated with the ancillary protein basigin and not when associated with embigin (see Some MCTs Require an Ancillary Protein for Plasma Membrane Expression and Activity section), whereas the choice of ancillary protein has no effect on the inhibition of MCT1 (19).
Molecular Mechanism of MCT1
The molecular mechanism by which MCT1 transports monocarboxylates has been extensively studied by analysis of transport kinetics, the use of inhibitors and more recently through site-directed mutagenesis. Detailed kinetic analysis of monocarboxylate transport in erythrocytes revealed that MCT1 operates through an ordered mechanism [see (1, 10)], although a recent analysis of published data of lactate transport into sarcolemmal vesicles and oocytes mediated by MCT1 and MCT4 concluded that it is not possible to discriminate between an ordered and a random binding mechanism (20). However, it is surprising that these authors did not refer to the more rigorous erythrocyte studies described above that strongly support an ordered mechanism. The ordered mechanism predicts that when transporting lactate into the cell, MCT1 has a substrate binding site open to the extracellular matrix which binds a proton first followed by the lactate anion. The protein then undergoes a conformational change to a new “closed” conformation that exposes both the proton and lactate to the opposite surface of the membrane where they are released, lactate first and then the proton. For net transport of lactic acid, the rate-limiting step is the return of MCT1 without bound substrate to the open conformation. For this reason, exchange of one monocarboxylate inside the cell with another outside is considerably faster than net transport of a monocarboxylate across the membrane (1, 10). We have recently developed a structural model for MCT1 based on site-directed mutagenesis and homology modeling that can account for this mechanism (21, 22). This is described in Structure and Translocation Cycle of MCTs section below.
Even before the identification of MCT1, characterization of monocarboxylate transport into isolated rat liver cells (23) and heart cells (24) led us to propose that there might be several MCT isoforms. This was confirmed when Christine Kim Garcia in the laboratory of Goldstein and Brown cloned a second isoform which transported lactate and pyruvate when expressed in insect Sf9 cells (25). They named this isoform MCT2 and reported that it had a higher affinity for substrates than MCT1. This was subsequently confirmed when rat MCT2 was characterized following expression in Xenopus oocytes (26). Km values for pyruvate and L-lactate were found to be about 0.1 and 0.74 mM, respectively, compared to values of about 1 and 3.5 mM for MCT1 (see Table 1). MCT2 was also reported to be more sensitive than MCT1 to inhibition by a range of inhibitors including CHC, DBDS, and DIDS but insensitive to the organomercurial reagent pCMBS (25, 26). Subsequent work in this laboratory has shown that pCMBS actually inhibits MCT1 by binding to the ancillary protein, basigin (see Some MCTs Require an Ancillary Protein for Plasma Membrane Expression and Activity section) and that MCT2 is insensitive because it usually associates with embigin rather than basigin (27).
The sequence of MCT2 is far less conserved across species than that of MCT1 or MCT4 and there also appear to be considerable species differences in the tissue expression profile of this isoform (2). Both Northern blot analysis and inspection of the human expressed sequence tag (EST) database suggest relatively little expression of MCT2 in human tissues. However, in mouse, rat, and the hamster, both Northern and Western blot analyses and immunofluorescence microscopy show the protein to be expressed in liver, kidney, brain, and sperm tails, whereas in hamster, there is also evidence for its presence in skeletal muscle and heart (3, 25, 28). Where MCT2 is expressed together with MCT1, its exact location within the tissue is different, suggesting a unique functional role related to its high substrate affinity [see (2, 4)]. This may be especially important in the brain where MCT2 expression is largely confined to the postsynaptic density of the neurons and may facilitate the uptake of lactate for oxidation as a respiratory fuel (29, 30).
MCT3 was originally identified by Philp and coworkers (16) as a developmentally expressed protein in the chick retinal pigment epithelium but subsequently identified as a member of the MCT family and named MCT3. Its transport activity was confirmed by expression in yeast where it was reported to demonstrate a Km for L-lactate of about 6 mM and to be insensitive to CHC, phloretin, and pCMBS (31). MCT3 has yet to be characterized in Xenopus oocytes and detailed knowledge of its substrate and inhibitor specificity is lacking. Expression of MCT3 is confined to the retinal pigment epithelium and choroid plexus epithelia (32), where it is located on the basal membrane in contrast to MCT1 which is found on the apical membrane (33, 34). It is thought to play an important role in facilitating the transport of glycolytically derived lactic acid out of the retina.
MCT4 was identified in this laboratory during a search of the EST database for novel members of the MCT family (35) and was originally called MCT3 based on its sequence homology with chick MCT3. It was renamed MCT4 when a distinct mammalian MCT3 was identified in mammalian retinal pigment epithelium [see (2)]. Northern and Western blotting and EST database analyses showed MCT4 to be widely expressed and especially so in glycolytic tissues such as white skeletal muscle fibers, astrocytes, white blood cells, chondrocytes, and some mammalian cell lines [see (2–4)]. In the rat, MCT4 is expressed in the neonatal heart, which is more glycolytic in its energy metabolism than the adult heart where MCT4 is absent but MCT1 abundant (36, 37). This led us to propose that the properties of MCT4 might be especially appropriate for export of lactic acid derived from glycolysis and characterization of its properties by expression in Xenopus oocytes suggest this to be the case (15, 38). MCT4 exhibits a lower affinity for most substrates and inhibitors than MCT1, with Km and Ki values some 5–10-fold higher. Thus, Km values for L-lactate and pyruvate were determined to be 28 and 150 mM, respectively (see Table 1), and little inhibition by DIDS or CHC was observed at concentrations giving >50% inhibition of MCT1. The high Km for pyruvate may be especially significant as this avoids loss of pyruvate from the cell which, were it to occur, would prevent removal of the reduced form of nicotinamide adenine dinucleotide (NADH) produced in glycolysis by reduction of pyruvate to lactate (see ref. 2,8).
MCT8 (SLC16A2) was first cloned in 1994 by Lafreniere et al. and called XPCT because it contains a long N-terminus “PEST” domain, a proline/glutamate-rich region thought to mediate rapid proteolytic degradation (39). Following the identification of the closely related protein TAT1 (MCT10) as an aromatic amino acid transporter (see MCT10 section below), collaborative studies with the laboratory of Theo Visser enabled us to identify MCT8 as an active and specific iodothyronine (thyroid hormone) transporter. Expression of MCT8 in Xenopus oocytes revealed that both T4 and T3 were transported with high affinity (Km = 2–5 μM) whilst aromatic amino acids were not transported (40). Transport is Na independent, and there is no evidence for proton dependence. MCT8 is expressed in most tissues, including liver, kidney, heart, skeletal muscle, brain, pituitary, and thyroid and probably occurs naturally as a homodimer [see (41)]. Its importance in thyroid hormone transport has been confirmed by the identification of individuals with mutations in the MCT8 gene that exhibit major psychomotor retardation (42).
MCT10 (SLC16A10), also known as TAT1, was discovered by expression cloning in Xenopus oocytes and shown to transport aromatic amino acids (phenylalanine, tyrosine, and tryptophan) in a sodium and proton independent manner with Km values of about 1 mM (5). The original studies failed to show transport of thyroid hormone by MCT10 but subsequently such activity was demonstrated [see (41)]. The expression MCT10 is species dependent but it is usually strongly expressed in the intestine, kidney, liver, skeletal muscle, heart, and placenta [see (4)].
STRUCTURE AND TRANSLOCATION CYCLE OF MCTs
Hydrophobicity plots predict that MCTs should contain 12 TMs with intracellular C- and N-termini and a large intracellular loop between TMs 7 and 8. Indeed, proteolytic cleavage and labeling studies of MCT1 were entirely consistent with this topology (43). Sequence analysis suggests that MCTs are unlikely to be glycosylated (43) and no experimental evidence has been reported that would argue otherwise. Thus, bands on Western blots are not broad and cannot be shifted by deglycosylation (44). Using molecular modeling, we have generated a probable three dimensional structure of MCT1 in the “closed” conformation (with substrate binding site exposed to cytosol) based on the published structure of the E Coli glycerol phosphate transporter GlpT (1PW4) (21). This model, which is shown in Fig. 3, is consistent with results of extensive site-directed mutagenesis studies of MCT1 (reviewed in (21, 22). To model MCT1 in the “open” conformation (with the substrate binding site exposed to the extracellular medium), we used site-directed mutagenesis to identify key lysine residues involved in the binding of DIDS which inhibits MCT1 activity by competing with the monocarboxylate for its extracellular binding site (22). In our model, the 6-helix N-terminal domain and the similar 6-helix C-terminal domain of MCT1 are linked by a 30-residue loop of unknown, or no, structure. Interconversion between the two conformations is predicted to occur by a modest reorganization of the interface between the two domains (22) as originally suggested by Bröer et al. (45). Evidence has been provided for similar domain rearrangements in the E Coli transporter Lac-Permease during its catalytic cycle [see (46)].
The model structures we have proposed are consistent with critical roles for key residues identified by site-directed mutagenesis to be important for MCT1 function and suggest the following translocation cycle (22). In the “open” conformation, MCT1 has an essential lysine (K38) in a hydrophobic pocket at the bottom of a substrate-binding channel. The hydrophobic environment ensures that this is normally uncharged, but when it accepts a proton it provides a binding site for the monocarboxylate anion. This induces a domain rearrangement to form the closed conformation with the lactate and proton being passed through the channel. This is likely to involve aspartate and arginine residues (D302/R306) that are present as an ion pair in the channel and known to be essential for activity from site-directed mutagenesis (21, 45). As lactic acid bound to K38 is transferred to D302/R306, the K38 is deprotonated again, and MCT1 relaxes back to the closed state, opening the D302/R306 site to the intracellular medium. This allows the bound lactic acid to diffuse into the intracellular medium, and the transport cycle to repeat. Close to the D302/R306 ion pair is phenylalanine 360 that has been identified to play an important role in determining which substrates are transported by MCT1 (21). When phenylalanine 360 is mutated to a cysteine, MCT1 can transport the larger monocarboxylate, mevalonate (12).
Measurement of the Km values of chimeric transporters combining different domains of MCT1 and MCT4 revealed that both the C- and N-terminal halves of the molecule influence transport kinetics consistent with our proposed translocation mechanism that requires Lys38 in TM1 in addition to Asp302 and Arg306 in TM8 (19). Similar studies with MCT1/MCT2 chimaeras suggested that the TM domains, and especially TM7-TM12, are the major determinants of L-lactate affinity with the C-terminus and intracellular loop between TMs 6 and 7 having little effect (19).
SOME MCTS REQUIRE AN ANCILLARY PROTEIN FOR PLASMA MEMBRANE EXPRESSION AND ACTIVITY
Our studies on the inhibition of MCT1 in rat erythrocytes by DIDS showed its binding to be reversible initially but subsequently to become irreversible as one of its two isothiocyano groups covalently modifies a lysine on MCT1 [see (1)]. However, Western blotting with antibodies against MCT1 revealed that by using its second isothiocyano group DIDS can also crosslink MCT1 to another protein of about 70 kDa which we identified as embigen (also known as gp-70) (44). Embigin is not expressed in many tissues but basigin (also known as CD147, OX-47, EMMPRIN, or HT7) is a closely related protein that is widely expressed. Both basigin and embigin have a single transmembrane domain containing a conserved glutamate residue, a short intracellular C-terminus, and a large glycosylated extracellular domain with two or three immunoglobulin domains depending on the splice variant (47). We showed that when attempts were made to overexpress MCT1 or MCT4 in a variety of cell lines the proteins failed to reach the plasma membrane but accumulated in the perinuclear region characteristic of the Golgi apparatus. However, when coexpressed with basigin both MCT1 and MCT4 were correctly targeted to the plasma membrane, suggesting that basigin acts as an essential chaperone for both MCTs (48). The use of a CD2/basigin chimera revealed that it is the transmembrane domain and/or intracellular tail of basigin rather than the extracellular domain that is required for this role (48).
Immunofluorescence microscopy revealed that MCT1 and MCT4 colocalize with basigin in a variety of different cells and tissues implying that the proteins remain associated (48–50), and this continued association was confirmed by several techniques. First, MCT1 and basigin could be coimmunoprecipitated from solubilized plasma membranes (48). Second, inducing aggregation of basigin to a cap at one end of a cell with a cross-linking antibody induced caused MCT1 to move to the same cap (48). Third, constructs of MCT1 containing either the cyan- or yellow-variants of green fluorescent protein (CFP and YFP) at the C- or N-terminus were coexpressed at the plasma membrane with basigin containing either YFP or CFP on the C-terminus. Using a confocal microscope, we observed fluorescence resonance energy transfer (FRET) between the CFP and YFP which can only occur when the two proteins are < 100 Å apart (51). We were also able to demonstrate FRET between MCT1-CFP and MCT1-YFP which implied that MCT1 can form dimers. Taken together with the antibody capping studies, these data suggest that MCT1 is likely to exist as an MCT1-basigin dimer. Subsequent experiments using site-directed mutagenesis and molecular modeling led us to propose that the TM of the ancillary protein lies adjacent to TMs 3 and 6 of the MCT (21, 22, 27). We were able to model this interaction (34) as shown in Fig. 3.
When available, basigin is the preferred binding partner for MCT1 as confirmed by expression studies in Xenopus oocytes (19). However, in the absence of basigin, MCT1 will interact with embigin as it does in rat red blood cells (44). By contrast, overexpression of MCT2 in the plasma membrane of mammalian cells requires coexpression of embigin rather than basigin (27) and exogenous expression of MCT2 in Xenopus oocytes is enhanced by coexpression of exogenous embigin and not basigin (19). These data imply that embigin rather than basigin is the preferred binding partner for MCT2. However, MCT2 can interact with basigin if no embigin is available (as in oocytes), and we have shown that the C-terminus of MCT2 is essential for this interaction with basigin which is not the case when MCT2 associates with embigin (19). The C-terminus of MCT1 had no effect on its interaction with either embigin or basigin, and we have been unable to detect any change in the kinetic properties of MCT1 when it is associated with embigin rather than basigin (19). However, inhibition of MCT1 by organomercurial reagents such as pCMBS does require its association with basigin rather than embigin (27), accounting for the earlier observation that MCT2, which usually associates with embigin, was insensitive to pCMBs (25). Indeed, we showed that organomercurial agents attack a reactive extracellular cysteine residue in basigin that is lacking in embigin (19, 27) and the resulting conformation change in the basigin appears to weaken its interaction the MCT leading to inhibition of transporter activity. Very recent studies in our laboratory have revealed that inhibition of MCT2 by AR-C155858 was greatly reduced when MCT2 was expressed in Xenopus oocytes with its preferred partner embigin, rather than with endogenous basigin, whereas for MCT1 the choice of ancillary protein was without effect (19). Use of MCT1/MCT2 chimaeric transporters and MCT1 and MCT2 with their C-terminal tails removed showed that this effect of embigin to modulate MCT2 sensitivity to AR-C155858 involves interactions of embigin with both the intracellular C-terminus and TMs 3 and 6 of MCT2 (19).
Considerable progress has been made in characterizing the properties of different members of the MCT family, and there is an increasing awareness that MCTs play critical metabolic roles in a wide range of normal and pathological conditions [see (2, 4, 8)]. However, there remain eight members of the MCT family whose function has yet to be identified. Our working model of the molecular mechanism of MCT1 suggests key residues required for proton translocation are not present in MCT 5, MCT6, MCT9, MCT12, and MCT14. Thus, only MCT 7, MCT11, and MCT13 are likely to be proton-linked transporters. Our own unpublished studies have failed to demonstrate proton-linked transport of a wide range of monocarboxylates by MCT7 when expressed in Xenopus oocytes. Clues as to the function of these orphan members of the MCT family may emerge as disease states are identified that are associated with their mutation.
The recent development by AstraZeneca of a class of potent MCT1 inhibitors that act as powerful immunosuppressive agents by inhibiting T-lymphocyte proliferation (17) provides a valuable tool for studying the metabolic roles of MCT1 as well as illustrating how MCTs may be promising pharmacological targets. Clearly, drugs that are specific and potent inhibitors of the other MCT isoforms would also be desirable, both for probing their roles in normal metabolism and as potential therapeutic agents. Development of such drugs would be greatly enhanced if three-dimensional structures of the different MCT isoforms were available and we are currently working on this.
The author thanks the numerous colleagues who have worked on MCTs in his laboratory over many years and the many funding bodies who have supported the research. Constraints on the maximum size of the bibliography means that it has not been possible to cite directly all authors of the work reviewed.