Cellular and Subcellular Expression of the Monocarboxylate Transporter MCT1 in Rat Heart
A High-Resolution Immunogold Analysis
An antibody to the C-terminus of the monocarboxylate transporter MCT1 was used to study the precise cellular and subcellular distribution of this transporter in rat heart. Postembedding immunogold procedures revealed that the labeling in the heart was restricted to cardiomyocytes and concentrated along the plasma membrane, including the transverse tubules. Gold particles occurred with highest densities in intercalated disks, where they avoided desmosomes and gap junctions. Labeling was also associated with plasmalemmal invaginations having ultrastructural features typical of caveolae. Internal membrane compartments were unlabeled. Quantitative analyses following postembedding labeling showed that the distribution of gold particles across the plasma membrane was nearly symmetrical, indicating that the C-terminus of the transporter is situated very close to the cell membrane. In preembedding immunogold experiments, the gold particles were localized at the external aspect of the plasma membrane, suggesting that the C-terminus is extracellular. From the present data, it can be concluded that even under basal conditions the majority of the MCT1 molecules in heart is present in the myocyte plasma membrane, implying that there is a constitutive functional expression of this transporter. It follows that the increased transmembrane flux of lactate during exercise or in pathological conditions such as ischemia must be a result of altered substrate gradients rather than of translocation of MCT1 molecules to the plasma membrane.
It has been known for some time that lactate serves as an energy substrate in heart, but it is only recently that this monocarboxylate has come to be appreciated as a major energy source for cardiomyocytes.  The role of lactate is particularly prominent during physical exercise, when lactate oxidation may account for >50% of the oxygen consumption in heart muscle.  Ischemia, on the other hand, is associated with a rapid efflux of lactate from the affected cells. 
Cardiomyocytes contain a proton-lactate cotransport system that is capable of both inward and outward transport, depending on the substrate gradients across the sarcolemma. [1,4,5] Much of the proton-lactate cotransport is likely to be mediated through the monocarboxylate carrier MCT1, which was recently cloned by Garcia et al.  It transports lactate as well as pyruvate and has been shown by Western blotting [7,8] and light microscopic immunocytochemistry [6,7] to be present in the heart.
A high cellular content of a transporter does not necessarily reflect a high functional activity. Recent data on the glucose transporter GLUT-4 serve to illustrate this point. In basal conditions, the majority of GLUT-4 in cardiac and skeletal muscle cells is localized to internal membranes, physically separated from the extracellular compartment. [9-11] However, the internal pools of GLUT-4 can be translocated by insulin to the plasma membrane and transverse tubules, [9,10,12] thus becoming functionally expressed.
The aim of the present study was to resolve the cellular and subcellular expression of MCT1 in rat heart so as to identify the membrane domains that could be responsible for lactate transport. Sensitive high-resolution immunogold procedures based on freeze-substituted tissue specimens [13-15] were used for this purpose.
Materials and Methods
The polyclonal MCT1 antibody was produced by immunizing rabbits with a synthetic peptide (CPQQNSSGDPAEEESPV) corresponding to the C-terminus of the molecule.  The antibody was affinity-purified. A search in the Swiss-Prot sequence database revealed no membrane proteins with significant homologies to the peptide sequence used for immunization. For comparison, the gap junction protein connexin43 [16,17] was visualized by a commercially available polyclonal rabbit antibody (Zymed).
Proteins were isolated as previously described. [18,19] Protein samples of different tissues were separated on 12% SDS-polyacrylamide gels (150 V, 1 hour). Proteins were then transferred to Immobilon polyvinylidene difluoride membranes (100 V, 90 minutes). The membranes were incubated with the MCT1 antibody (0.1 micro g/mL, 2 hours), followed by donkey anti-rabbit IgG conjugated to horseradish peroxidase (Amersham, 0.1 micro g/mL, 1 hour). MCT1 was detected using an enhanced chemiluminescence method (Hyperfilm-ECL, Amersham). Molecular weight markers were included.
Preparation of Animals and Tissue
Male Wistar rats (250 to 300 g, Mollegaard, Ejby, Denmark) were used. They had been allowed free access to food and drinking water. When they were killed for study, they were deeply anesthetized by an intraperitoneal injection of a mixture of midazolam, fentanyl citrate, and fluanisone (3.8 mg, 0.24 mg, and 7.5 mg/kg body wt respectively). They were then perfused through the abdominal aorta with 2% dextran (molecular weight, 70 000) in 0.1 mol/L sodium phosphate buffer (PB) (pH 7.4, 4 degrees C, 15 seconds), followed by a mixture of 0.5% glutaraldehyde and 4% formaldehyde (freshly depolymerized from paraformaldehyde) in the same buffer (room temperature, 50 mL/min for 20 minutes). The heart was left in situ overnight. All specimens were dissected from the anterior left ventricular wall.
Postembedding Immunogold Cytochemistry
The tissue samples were subjected to freeze substitution [13,14] as described by Hjelle et al.  In brief, the specimens were cryoprotected by immersion in graded concentrations of glycerol (10%, 20%, and 30%) in PB and plunged rapidly into liquid propane cooled to -170 degrees C by liquid nitrogen in a cryofixation unit (KF 80, Reichert). The samples were then immersed in 0.5% uranyl acetate dissolved in anhydrous methanol (-90 degrees C) in a cryosubstitution unit (AFS, Reichert). The temperature was raised in steps of 4 degrees C/h from -90 degrees C to -45 degrees C. The samples were washed with anhydrous methanol and infiltrated with Lowicryl HM20 resin at -45 degrees C, with a progressive increase in the ratio of resin to methanol. Polymerization was carried out with UV light (360 nm) for 48 hours. Ultrathin sections (70 to 80 nm) were cut with a Reichert ultramicrotome, mounted on nickel grids or gold-coated grids, and processed for immunogold cytochemistry as described by Matsubara et al.  Briefly, the sections were treated with a saturated solution of NaOH in absolute ethanol (2 to 3 seconds), rinsed in PB, and incubated sequentially in the following solutions (at room temperature): (1) 0.1% sodium borohydride and 50 mmol/L glycine in Tris buffer (5 mmol/L) containing 0.1% Triton X-100 and 0.05 mol/L NaCl (TBNT, 10 minutes), (2) 2% human serum albumin (HSA) in TBNT (10 minutes), (3) primary antibody against MCT1 (2 micro g/mL) in TBNT containing 2% HSA (2 hours), (4) 2% HSA in TBNT (10 minutes), and (5) gold-conjugated secondary antibody (15-nm particles, GAR15, Amersham) diluted 1:20 in TBNT containing 2% HSA and polyethylene glycol (0.5 mg/mL, 2 hours). Finally, the sections were counterstained and examined in a Philips CM10 transmission electron microscope.
Postembedding immunogold double labeling was carried out as previously described,  using formaldehyde vapor  to avoid interference between the two sequential incubations. MCT1 and connexin43 were then visualized by 30-nm and 10-nm gold particles, respectively. Reversal of the antibody sequence did not change the labeling pattern.
Preembedding Immunogold Cytochemistry
Vibratome sections (50 micro m thick) were rinsed in PB and incubated sequentially in the following solutions (at room temperature, if not stated otherwise): (1) 2% HSA in TBNT (1 hour), (2) primary antibody against MCT1 (2 micro g/mL) in TBNT containing 2% HSA (46 hours at 4 degrees C and 2 hours at room temperature), (3) 2% HSA in TBNT (6x10 minutes and 3x30 minutes), (4) secondary Fab fragments coupled to <1-nm gold particles (Aurion Fab 2 GAR GP-Ultra Small) diluted 1:60 in TBNT containing 2% HSA (overnight), (5) 2% HSA in TBNT (3x10 minutes), (6) PB (3x10 minutes), (7) 2.0% glutaraldehyde in PB (10 minutes), (8) PB (3x10 minutes), (9) 0.5% osmium tetroxide in PB (20 minutes, 4 degrees C), and (10) PB (3x10 minutes). For silver intensification, the sections were incubated in Aurion R-Gent Silver for 25 minutes (in a dark chamber, room temperature) and rinsed in deionized water (5x20 seconds, 2x5 minutes) and PB (3x20 seconds). Finally, the sections were dehydrated in a series of ethanols and embedded in an epoxy resin (Durcupan ACM, Fluka; 30 minutes at 56 degrees C). The resin was left at room temperature overnight and then polymerized for 48 hours at 56 degrees C. Ultrathin sections were mounted on nickel grids and counterstained.
The following controls were performed in postembedding immunogold cytochemistry: (1) adsorption controls, made by preincubating the affinity-purified antibodies (2 micro g/mL) with excess of immunizing peptide (50 ng/mL) or a heterologous peptide (corresponding to amino acids 679 to 697 of protein kinase C gamma, at the same concentration), and (2) incubation with rabbit IgG (2 micro g/mL) or buffer instead of the primary antibody.
The distribution of gold particles was recorded (1) along plasma membranes, expressed as linear density (number of particles per micro m of membrane), and (2) across the membranes, expressed as the distance between the membrane and the individual gold particles. For the linear density measurements (Table 1), particles were included only if they were situated within 28 nm of the midline of the membrane, corresponding to the experimentally defined maximum distance between an epitope-bearing surface and the center of the corresponding 15-nm gold particles.  Since the membranes that were analyzed are not exactly perpendicular to the plane of section (and the sections are labeled on both sides), some particles bound to the epitopes in the membrane inevitably end up distal to 28 nm in the electron micrographs (see Figure 6).
Electron micrographs were taken after a randomization procedure. In the prints (altogether some 270 in number, usually at x116 250 magnification), all fragments of plasma membrane were accepted for analysis, provided they were longer than [nearly =]0.25 micro m and appeared transversely cut. Altogether, 489 independent fragments were collected, with the majority being between 0.5 and 1.5 micro m in length. Each membrane fragment was digitized (carefully following the center line of the membrane), as were the centers of any associated gold particles. It should be noted that caveola-like membrane invaginations (see below) with associated particles were excluded in this process. A computer program (H.K. Ruud and T.W. Blackstad, unpublished data, 1997) was used to calculate the length of each fragment and the particle-to-membrane distance.
The cardiomyocyte membrane fragments were classified in three main categories: those facing a neighboring myocyte (outside of intercalated disks, Figure 2 and Figure 3), those facing a capillary (Figure 2), and those engaged in intercalated disks, excluding desmosomes and gap junctions. For simplicity, these categories were termed "myo-myo," "myo-cap," and "disk," respectively. In the first two categories, a variable sparse amount of basement membrane and connective tissue material (and in myo-myo, rarely a fibroblast) covered the sarcolemma. The plasma membrane labeling of fibroblasts and endothelial cells was also analyzed.
Each of the line density figures (Table 1) is the average of three sets of results obtained in three grids, respectively, from two animals. In each set, the same grids were used for all three membrane categories. Although the absolute particle counts varied between the grids (reflecting interexperimental and interanimal variation), the relative labeling intensity was the same. Thus, the ratios of the values for myo-myo, myo-cap, and disk remained unchanged between sets and animals (not shown). The pattern of labeling in the sections used for quantitative analysis was reproduced qualitatively in four additional experiments.
For recording of particles in intercalated disks, a modified cruder sampling procedure was used because of the tortuous course and orientation of the plasma membranes there. Particles were counted when measured manually to be <28 nm from the nearest plasma membrane. The membranes of the disk are closely apposed. The particle number used for the calculation of the line density was therefore obtained by first counting the total number of particles associated with the disk membrane pair and then dividing this number by two.
For statistical evaluation of line densities, groups of 5 to 10 plasma membrane fragments were pooled to form a number of larger, nearly equal spans. For each of these, a line density was calculated (ratio of total number of particles to total length).
The MCT1 polyclonal antibody yielded a single band on Western blot corresponding to [nearly =]43 kD (Figure 1), which is consistent with the molecular mass reported for MCT1. [6,8] Strong immunoreactivity was observed in heart, diaphragm, and brain, followed by red gastrocnemius; only a faint reaction occurred in white gastrocnemius. The blank lane in Figure 1 corresponded to a negative synaptosome fraction from brain.
Cellular Distribution of MCT1
Gold particles signaling MCT1 were concentrated in the cardiomyocytes as opposed to other cell types (Figure 2, Figure 3, Figure 4, and Figure 5). The linear density of particles in plasma membranes of endothelial cells (Figure 2) and fibroblasts (Figure 3B) was about one tenth that in cardiomyocyte membranes (Table 1). No significant difference was found between the luminal and abluminal membranes of the endothelial cells (Table 1). The scattered gold particles over mitochondria (Figure 2) reflect unspecific labeling, since it remained after preadsorption with the peptide used for immunization (Figure 3D).
Distribution of MCT1 Along the Cardiomyocyte Plasma Membrane
Each of the three categories of cardiomyocyte plasmalemma (myo-myo, myo-cap, and disk) showed conspicuous immunogold labeling (Figure 2, Figure 3, Figure 4, and Figure 5). Calculation of the mean line density of gold particles (Table 1) revealed the highest value along disk membranes (III), an intermediate value in the myo-myo category (II), and the lowest value in the myo-cap category (I). As indicated, desmosomes and gap junctions, and caveola-like structures, were not part of the basis for these calculations.
Many gold particles were associated with membrane invaginations of [nearly =]50 to 90 nm in diameter (Figure 2) and displayed ultrastructural features typical of caveolae. [23-25] From the data presented in the Table 1, it can be calculated that the fraction of gold particles that was associated with caveolae was 13.6% for myo-myo and 9.0% for myo-cap. However, this difference primarily reflects the [nearly =]40% higher number of caveolae per micrometer membrane in myo-myo than in myo-cap (Figure 2, Table 1). The proportion of labeled caveolae and the mean number of gold particles per immunopositive caveola were similar in the two membrane domains (Table 1). The disk membrane did not display any caveola-like structures.
The dense MCT1 immunolabeling in intercalated disks often appeared as clusters of particles (Figure 4 and Figure 5). Gold particles were found along the entire disk membrane but avoided desmosomes and areas of close membrane apposition that resembled gap junctions (Figure 4). No particles were recorded at such contacts along the 152 micro m of disk membrane that was included in the quantitative analysis shown in the Table 1 (the particle indicated by a thin arrow in Figure 4A was a rare exception). An unequivocal identification of gap junctions was made possible by use of antibodies selective for connexin43, the building block of gap junctions in the heart.  Double-labeled preparations, with different gold particle sizes representing MCT1 and connexin43, confirmed the spatial segregation of these two proteins (Figure 5).
MCT1 Associated With Transverse Tubules (T Tubules)
A high density of immunogold particles was associated with T tubules (Figure 3A). Although most of these particles could be attributed to the tubule membranes, it could not be excluded that some particles represented epitopes in the adjacent sarcoplasmic reticulum. This is because the two membrane systems are so tightly apposed that the spatial resolution of the immunogold technique becomes a limiting factor. For this reason, we refrained from a quantitative analysis. Qualitatively, the labeling appeared to be particularly dense at sites where T tubules were in proximity to mitochondria (Figure 3A).
Distribution of Gold Particles Across the Plasma Membrane
After postembedding labeling (which only reveals epitopes exposed at the surface of the section), the gold particles were concentrated within 50 nm of the plasmalemma (Figure 6A) and appeared to be symmetrically distributed across the membrane (Figure 6B). Particles up to 52 nm are shown in Figure 6B, but the symmetrical distribution continues as far as 100 nm and beyond. The average position of the particles <28 nm from the center line was -0.6 nm (intracellular negative), ie, very close to the midpoint of the membrane (the 95% confidence interval for the mean being -1.6 to 0.4 nm). Thus, since with postembedding labeling it could not be resolved whether the epitope (C-terminus) is on the extracellular or intracellular side, preembedding immunogold labeling was performed. With this approach, the immunoglobulins have less rotational freedom, implying that the particles will end up on the side of the membrane where the epitope is expressed. The disposition of silver-intensified particles suggested an extracellular localization of the epitope (Figure 3C).
The cloning of the first monocarboxylate transporter MCT1 is a recent event.  It was shown by light microscopic immunocytochemistry and Western blotting that this transporter was expressed in the heart.  The present study was undertaken to reveal the cellular and subcellular localization of MCT1 by use of quantitative immunocytochemistry at the electron microscopic level. To our knowledge, this is the first study of its kind.
Our data suggest that in the heart the cardiomyocytes are the primary site of MCT1 expression and that this transporter is differentially expressed along the cardiomyocyte plasma membrane. A localization of MCT1 to the intercalated disk was noted by Garcia et al  using immunofluorescence, and the present quantitative analysis establishes that a high density of MCT1 immunoreactivity occurs at this site. Moreover, our data indicate that MCT1 occupies membrane domains separate from gap junctions or desmosomes. Recent studies have shown inositol tris-phosphate receptors to be concentrated in the fasciae adherentes of the disk  and that the Kv 1.5 K sup + channel protein is also abundant in the disk region.  Together with these studies, the present findings strengthen the notion that the separate subdomains of the cardiomyocyte plasma membrane differ both biochemically and functionally.
The membranes engaged in the intercalated disks constitute much of the total cell surface ([nearly =]90% according to Page and McCallister ). As will be outlined below, there are two possible pathways for lactate transfer through the narrow cleft of the intercalated disks.
Exchange between myocytes and capillaries would require diffusion along the tortuous space of the disk. Currents associated with the action potential cause rapid accumulation of K sup + in the cleft,  but lactate fluxes are slower, and there is wide access to the cleft along the entire circumference of the cells, allowing for transfer to the capillaries. Hence, in a well-perfused myocardium, the presence of MCT1 at the disk and in other plasma membrane domains should ensure a high lactate transport capacity, which would be especially beneficial during physical exercise (allowing for a net uptake).
The second pathway might involve transfer of lactate between contiguous heart muscle cells. Especially during ischemic conditions, when there is reduced or no diffusional exchange with blood, the MCT1 in the disk could be assumed to mediate an intercellular transfer of lactate. This would require that the transporter operate in lactate efflux mode in one cell and in the reverse mode in the neighboring cell. Our data do not provide evidence that the transporters are organized in register in the same manner as the connexin molecules, but efficient transfer of lactate between cells may occur provided lactate is distributed throughout the intercellular space of the disk. Another requirement would be that the transporters operate close to their equilibrium so that even small changes in proton and lactate gradients may switch the transporter from one mode to the other. If the above requirements are met, the location of MCT1 to the disk would provide a mechanism by which protons and lactate ions can escape from an ischemic region.
Our finding that no or very modest amounts of MCT1 are present in endothelial cells raises the question of how lactate is transferred across the capillary wall. It has been suggested that lactate transfer occurs by simple diffusion along paracellular pathways.  However, the possibility cannot be ruled out that the endothelial cells express a lactate transporter different from MCT1. It should be noted that the heart was recently found to contain a second monocarboxylate transporter (MCT2), but no evidence of an endothelial location was provided. 
To our knowledge, GLUT-4 is the only transporter of organic molecules in heart that has been subjected to a quantitative immunogold analysis similar to the one at hand.  Under basal conditions, most of the GLUT-4 immunoreactivity (>99%) was localized in internal membrane compartments, including tubulovesicular elements near the sarcolemma. Only after stimulation with insulin was a significant fraction of GLUT-4 expressed in the plasma membrane. An important question is whether a similar recruitable pool also exists for MCT1.
Our data suggest that this is not the case. In sharp contrast to GLUT-4, MCT1 under basal conditions is primarily present in the plasma membrane, indicating that there is a constitutive functional expression of this transporter. Gold particles signaling MCT1 were rarely associated with internal membrane compartments. However, during the present study, the possibility had to be considered that tubulovesicular elements close to the plasma membrane (other than caveolae; see below) had failed to be recognized as such and that these were in fact a repository of MCT1. It speaks against this possibility that the gold particles were symmetrically distributed across the plasma membrane after postembedding labeling. Thus, there was no peak or raised plateau of labeling subjacent to the sarcolemma, and the particle density reached a stable minimum value within 50 nm of the sarcolemma.
The finding of gold particles associated with caveola-like membrane invaginations warrants discussion. One fourth to one fifth of these invaginations were labeled (Table 1), but this is an underestimate of the proportion of MCT1-containing caveolae. This is because some of the caveolae will be completely embedded in the section and thus inaccessible for postembedding labeling. It should also be recalled that the immunocytochemical procedure has a limited sensitivity. On the other hand, membrane patches with the molecular composition typical of caveolae may "collapse" into the plasma membrane,  and the possibility should be considered that some of the particles along the plasmalemma are associated with such patches. In regard to the issue of translocation (see above), it should be emphasized that the caveolae in cardiomyocytes are believed to occur predominantly in an open state (or cycle rapidly between open and closed states),  implying that the MCT1 that they contain should be continually exposed to the interstitial fluid. The caveolae are known to act as sites of entry for ions and small molecules through a process called potocytosis.  For example, caveolae in cardiomyocytes and other cells express high concentrations of a calcium pump,  and they also contain an inositol 1,4,5-tris-phosphate-regulated Ca2+ channel.  MCT1 should now be added to the list of caveola-associated proteins. As to T tubules, quantification was precluded for technical reasons. However, the present immunolabeling pattern indicated that MCT1 occurs in high concentrations in T-tubule membranes in proximity to mitochondria. This would facilitate a rapid delivery of lactate to this organelle and ensure its usefulness as an energy substrate.
Concerning the organization of the MCT1 molecule in the plasma membrane, the symmetrical distribution of the postembedding labeling with a mean (and mode) coinciding almost precisely with the membrane midline suggested that the C-terminus of the molecule is close to the external or internal membrane surface. To distinguish between these two possibilities, we took advantage of the preembedding immunogold technique. With this technique, the membranes will act as physical constraints for the disposition and growth of the silver-intensified gold particles, thus restricting the labeling to the membrane surface (cytoplasmic or external) at which the epitope is localized.  The data suggest that the C-terminus is extracellular. Brain sections were processed in parallel with the heart sections but incubated with an aquaporin-4 antibody instead of the MCT1 antibody. The labeling in these sections was restricted to the internal aspect of the plasma membrane.  This attests to the capacity of our preembedding immunogold procedure to differentiate between intracellular and extracellular epitopes.
The topology of MCT1 was recently investigated by biochemical techniques. Based on the observation that carboxypeptidase Y cleaves rat erythrocyte MCT1 when the membranes are leaky but not when they are intact, it was concluded that the C-terminus was likely to be intracellular.  The reason for this discrepancy between the two studies remains to be resolved.
This study was supported by the Norwegian Research Council and the Natural Sciences and Engineering Research Council of Canada. The authors are indebted to Karen Marie Gujord, Bjorg Riber, Gunnar Lothe, and Carina Knudsen for excellent technical assistance. We thank Dr A.P. Halestrap, Bristol University, for the gift of MCT1 antibody.
Received October 18, 1996; accepted December 31, 1996.
- © 1997 American Heart Association, Inc.
- Poole RC,
- Halestrap AP
- Vandenberg JI,
- Metcalfe JC,
- Grace AA
- Wang X,
- Levi AJ,
- Halestrap AP
- Wang X,
- Poole RC,
- Halestrap AP,
- Levi AJ
- Garcia CK,
- Brown MS,
- Pathak RK,
- Goldstein JL
- Slot JW,
- Geuze HJ,
- Gigengack S,
- James DE,
- Lienhard GE
- Rodnick KJ,
- Slot JW,
- Studelska DR,
- Hanpeter DE,
- Robinson LJ,
- Geuze HJ,
- James DE
- Holman GD,
- Leggio LL,
- Cushman SW
- Marette A,
- Burdett E,
- Douen A,
- Vranic M,
- Klip A
- van Lookeren Campagne M,
- Oestreicher AB,
- van der Krift TP,
- Gispen WH,
- Verkleij AJ
- Matsubara A,
- Laake JH,
- Davanger S,
- Usami S-i, Ottersen OP
- Han XX,
- Handberg A,
- Petersen LN,
- Ploug T,
- Galbo H
- McCullagh KJA,
- Poole RC,
- Halestrap AP,
- O'Brien M,
- Bonen A
- Severs NJ
- Anderson RGW
- Fujimoto T
- Kijima Y,
- Saito A,
- Jetton TL,
- Magnuson MA,
- Fleischer S
- Kordylewski L,
- Goings GE,
- Page E
- Nielsen S,
- Nagelhus EA,
- Amiry-Moghaddam M,
- Bourque C,
- Agre P,
- Ottersen OP
- Poole RC,
- Sansom CE,
- Halestrap AP