Few Amino Acid Exchanges Expand the Substrate Spectrum of Monocarboxylate Transporter 10*

Abstract

Monocarboxylate transporters (MCTs) belong to the SLC16 family within the major facilitator superfamily of transmembrane transporters. MCT8 is a thyroid hormone transporter mutated in the Allan-Herndon-Dudley syndrome, a severe psychomotor retardation syndrome. MCT10 is closely related to MCT8 and is known as T-type amino acid transporter. Both transporters mediate T₃ transport, but although MCT8 also transports rT₃ and T₄, these compounds are not efficiently transported by MCT10, which, in contrast, transports aromatic amino acids. Based on the 58% amino acid identity within the transmembrane regions among MCT8 and MCT10, we reasoned that substrate specificity may be primarily determined by a small number of amino acid differences between MCT8 and MCT10 along the substrate translocation channel. Inspecting the homology model of MCT8 and a structure-guided alignment between both proteins, we selected 8 amino acid positions and prepared chimeric MCT10 proteins with selected amino acids changed to the corresponding amino acids in MCT8. The MCT10 mutant harboring 8 amino acid substitutions was stably expressed in Madin-Darby canine kidney 1 cells and found to exhibit T₄ transport activity. We then successively reduced the number of amino acid substitutions and eventually identified a minimal set of 2–3 amino acid exchanges which were sufficient to allow T₄ transport. The resulting MCT10 chimeras exhibited K_M values for T₄ similar to MCT8 but transported T₄ at a slower rate. The acquisition of T₄ transport by MCT10 was associated with complete loss of the capacity to transport Phe, when Tyr184 was mutated to Phe.

Monocarboxylate transporter (MCT)8 (SLC16A2) (1) is a transmembrane transport protein of the major facilitator superfamily (MFS) and mediates import and export of iodothyronines across the plasma membrane (1). Mutations in MCT8 lead to the Allan-Herndon-Dudley syndrome (OMIM 300523), a severe X-linked psychomotor retardation syndrome (2–5). This finding has renewed interest in the transporter molecules that mediate thyroid hormone uptake across cellular membranes (6). The first thyroid hormone transporter identified was organic anion-transporting polypeptide (Oatp)3 (7) followed by Oatp1 (8) both belonging to organic anion transporting polypeptide subfamily of MFS. A closely related transporter is Oatp14 (Slco1c1), which selectively transports T₄ and is expressed along the blood-brain barrier (9). L-type amino acid transporters were also found to transport thyroid hormones (10, 11). The T-type amino acid transporter (TAT1), also known as MCT10 (SLC16A10), is a very good T₃, but less active T₄, transporter (12). Finally, the liver-specific sodium-taurocholate transporting polypeptide (Slc10a1) was also found to transport thyroid hormones (8) but belongs to an entirely different class of transporters.

Several gene families have evolved the ability to bind or transport thyroid hormones. We are interested in the question whether they converged on one way to recognize the ligand or whether there are fundamentally different ways to transport thyroid hormones across the membrane (13). There is no experimental structure for any of these thyroid hormone transporters. Homology models have been generated for Oatp14, MCT8, L-type amino acid transporters (Lat)1, and Lat2 (14–17). We have been trying to delineate the ligand-protein interactions in MCT8 and identified several amino acids which are involved in transport activity. We first focused on charged amino acids within the transmembrane region and suggested a role in transport for Arg445 and Asp498 (15). Both amino acids are mutated in patients (18, 19), and their potential interaction has been subsequently confirmed by others (20). We then noted that, in analogy to binding the ligand in the T₃ receptor, thyroid hormone may fit between His and Arg pairs spaced about 15 Å apart in MCT8 (21). Binding of thyroid hormones to their various binding proteins is mediated mainly by hydrophobic pockets and few polar interactions (22). A role for His192 in T₃ uptake by MCT8 was independently confirmed (23), and a His likely plays a role in thyroid hormone binding in deiodinases (24). Taken together, these results suggest that our homology model allows to make reasonably good predictions about possible roles of individual amino acids in MCT8.

MFS transporters exhibit a central substrate translocation channel formed by the surface of 2 6-helix bundles which are comprised of transmembrane helix (TMH)1–TMH6 and TMH7–TMH12 (25). In order to more precisely locate amino acids participating in substrate transport, we chose to compare MCT8 and MCT10 and relate differences in amino acids along the proposed substrate translocation channel with substrate specificity of the 2 proteins. MCT8 and MCT10 share 58% amino acid identity within the transmembrane regions (12). Both transporters mediate T₃ uptake, but MCT8 is a much better T₄ transporter, whereas MCT10 (SLC16A10) is not good at transporting T₄ but is capable of aromatic amino acid uptake (26). Based on the homology model of MCT8, a set of 8 amino acid positions along the substrate translocation channel of MCT10 was chosen and all amino acids were exchanged to the corresponding amino acids present in MCT8 yielding a chimeric MCT10x8 protein. Because this mutant protein was capable of T₄ uptake with similar characteristics as MCT8, we tried to identify the minimal set of amino acid positions required to allow chimeric MCT10 uptake of T₄. We show here that exchange of a minimal set of 3 amino acids in MCT10 is sufficient to make MCT10x3 an active T₄ transporter, and 2 proteins with double mutations achieved T₄ uptake, albeit at lower rates. Aromatic amino acid uptake by MCT10 mutants capable of T₄ transport was generally reduced and entirely abrogated in those MCT10 chimeras that contained the Y184F exchange.

Materials and Methods

Cell culture

Stably transfected Madin-Darby canine kidney (MDCK1) cells were cultured in DMEM/F12 (1:1) (GIBCO) + 10% fetal calf serum (GIBCO) + 1% penicillin (5000 U/mL)/streptomycin (5000 μg/mL). Two days before the experiments, 50 000–70 000 cells per well were seeded into 24-well plates. One day before the experiment, a medium change to DMEM/F12 without serum was performed in order to prevent competition with amino acids or iodothyronines present in serum.

Site-directed mutagenesis

Mutations for creating MCT10^MCT8 chimeras were introduced in human MCT10 using QuikChange Lightning Site-Directed Mutagenesis kit (Agilent Technologies) with primers given in Table 1. Plasmids were stably transfected into MDCK1 cells as described previously (27).

Table 1.

Oligonucleotide Sequences Used to Introduce Amino Acid Exchanges in MCT10

Q88H	Q88Hfwd 5′-GTGTTCGGCATCCATAACGCTTGCGGGGT-3′
	Q88Hrev 5′-ACCCCGCAAGCGTTATGGATGCCGAACAC-3′
A90S	A90Sfwd 5′-GGTGTTCGGCATCCAGAACTCTTGCGGGGTGCTCTTCGTGTCC-3′
	A90Srev 5′-GGACACGAAGAGCACCCCGCAAGAGTTCTGGATGCCGAACACC-3′
V93I	V93Ifwd 5′-CCAGAACGCTTGCGGGATCCTCTTCGTGTCCATGCTGG-3′
	V93Irev 5′-CCAGCATGGACACGAAGAGGATCCCGCAAGCGTTCTGG-3′
S121A	S121Afwd 5′-CATGGGTAGGTTCTCTCGCCATGGGGATGATTTTC-3′
	S121Arev 5′-GAAAATCATCCCCATGGCGAGAGAACCTACCCATG-3′
Y184F	Y184Ffwd 5′-GGCTGCTCCTTTGCATTCCAGCCTTCATTGG-3′
	Y184Frev 5′-CCAATGAAGGCTGGAATGCAAAGGAGCAGCC-3′
T207S	T207Sfwd 5′-GGTGAATGGCATTGTCTCTGCTGGCAGCAGTGTCTTCAC-3′
	T207Srev 5′-GTGAAGACACTGCTGCCAGCAGAGACAATGCCATTCACC-3′
H317Y	H317Yfwd 5′-TGTGCCTTATGTTCACTTGATGAAATATGTAAATGAAAGATTTCAAGATG-3′
	H317Yrev5′-CATCTTGAAATCTTTCATTTACATATTTCATCAAGTGAACATAAGGCACA-3′
E325I	E325Ifwd 5′-GATGAAACATGTAAATGAAAGATTTCAAGATATAAAAAATAAAGAGGTTGTTCTCATGTGCATT-3′
	E325Irev 5′-AATGCACATGAGAACAACCTCTTTATTTTTTATATCTTGAAATCTTTCATTTACATGTTTCATC-3′

Surface biotinylation and Western blotting

Surface biotinylation was used for the detection and purification of cell surface proteins as previously described (21). Briefly, equal amounts of 10 μL of biotinylated protein were separated on 10% SDS gels, transferred onto nitrocellulose membranes, and probed with a Human influenza hemagglutinin (HA) tag antibody (Abcam). An antibody directed against β-actin (Rockland) was used for loading control.

Radioactive thyroid hormone uptake experiments

¹²⁵I-T₃ and ¹²⁵I-T₄ (PerkinElmer), respectively, was purified from iodide ions by adsorption chromatography as described previously (15) and finally resuspended in uptake buffer (125mM NaCl, 5mM KCl, 1.3mM CaCl₂, 1.2mM MgCl₂, 25mM HEPES, and 5.6mM glucose; pH 7.4). Iodothyronines were first dissolved in DMSO (1000-fold final concentration) and then in uptake buffer.

For competitive inhibition studies, cells were exposed to 10nM ¹²⁵I-T₃ tracer in uptake buffer for 3 or 10 minutes in the presence or not of 10μM iodothyronine or 1mM tryptophane (Trp). Buffer was removed, cells were washed with ice-cold PBS and lysed in 40mM NaOH. Radioactivity of the lysate was measured in a γ-counter (Wizard; PerkinElmer). For time-course assays, stably transfected cell lines were incubated with 10nM ¹²⁵I-T₃ or 1nM¹²⁵I-T₄ (added with 10nM nonradioactive T₄), respectively, for 1–30 minutes at 37°C. Cells were washed, lysed, and radioactivity associated with the cells was determined. For the determination of Michaelis constants K_M, MDCK1 cells were incubated for 3–10 minutes with T₄ at concentrations ranging from 500nM to 12μM containing ¹²⁵I-T₄ as tracer (15). A Michaelis-Menten transport mechanism was assumed for calculations (GraphPad 4.0; GraphPad, Inc).

Each experiment was performed on at least 2 different days with different batches of cells in triplicate. Radioactivity associated with empty vector-transfected cells was subtracted as background.

Liquid chromatography–tandem mass spectrometry (LC-MS²)-based uptake assays

For efflux-assays MDCK1 cells were first loaded with 10nM T₃ or T₄ for 30 minutes. Cells were briefly washed with PBS and uptake buffer without thyroid hormone was added for 5 minutes. After incubation, the cell lysate and supernatant was prepared for LC-MS² measurement.

For the determination of apparent Michaelis constants, MDCK1 cells were incubated for 5 minutes with rT₃ at concentrations ranging from 500nM to 12μM. After incubation, cells were washed once with ice cold PBS and prepared for LC-MS² measurement.

For substrate mix endpoint assays, MDCK1 cells were incubated for 30 minutes in a 100nM substrate mix (of each T₄, T₃, rT₃, 3,5-diiodo-L-thyronine [3,5-T₂], and 3′,5′-diiodo-L-thyronine [3′,5′-T₂]) dissolved in uptake buffer. After incubation, cells were washed once with ice cold PBS and prepared for LC-MS² measurement.

Sample preparation consisted of a fast alkaline lysis of the cells, followed by internal standard (IS) addition and liquid-liquid extraction (28). In short, cells cultured in a 24-well plate format were lysed with 100-μL 0.1N NaOH and 100-μL homogenization buffer (250mM sucrose, 20mM HEPES, and 1mM EDTA at pH 7.4) for 3 minutes on ice; 20-μL glacial acetic acid and IS were added. The liquid was transferred to a tube. Wells were washed with 10-μL glacial acetic acid in 90-μL homogenization buffer. After transfer to the tube, the liquid was incubated for 30 minutes at 37°C. This was followed by 2 1-mL isopropanol-methyl tert-butyl ether (30:70 vol/vol) extractions. The combined extracts were evaporated. The pellet was resuspended in 100-μL 50% methanol with 0.1% formic acid. Detection of iodothyronines was performed by LC-MS² (28) with the modification that 40 μL of sample extract was injected. Quantification was based on a standard curve, with correction for extraction efficiency using the IS. Each experiment was performed on at least 2 different days with different batches of cells in triplicate. Uptake associated with empty vector-transfected cells was subtracted as background.

Uptake measurements in Xenopus oocytes

In vitro transcription of 5-μg plasmid DNA were performed using the mMessage mMachine T7 kit (Life Technologies) following the manufacturer's protocol. Xenopus laevis oocyte preparation and cRNA injection were accomplished by EcoCyte Bioscience. Oocytes were injected with 30-ng cRNA. Injected oocytes were kept in Barth's buffer (88mM NaCl, 1mM KCl, 2.4mM NaHCO₃, 0.82mM MgSO₄, 0.66mM NaCO₃, 0.75mM CaCl₂, and 50mM HEPES in dH₂O [pH 7.4] + 10 mg/L gentamycin) for additional 3 days before performing uptake measurements at room temperature. Radioactive compounds were dissolved in Barth's buffer with the following concentrations: 10nM ¹²⁵I-T₃ and 0.5mM Phe plus 4nM ³H-Phe. Oocytes were incubated with tracer (for 3–20 min for T₃ time course and 20 min for Phe endpoint assay, respectively), washed with ice cold PBS and lysed with 40mM NaOH. Absorbed radioactivity of each oocyte was measured using γ- or β-counters (LB2111 Multi Crystal, Berthold Technologies, Wizard; PerkinElmer). H₂O injected oocytes served as negative controls.

Results

Amino acids that determine differences in substrate specificity between MCT10 and MCT8

Previously, we have systematically investigated the molecular determinants of MCT8 substrates by competing uptake of the canonical ¹²⁵I-T₃ substrate with a 1000-fold (for thyroid hormones) excess of unlabeled test compounds (15). This analysis demonstrated that efficient transport by MCT8 requires an L-amino acid moiety and one iodine atom in each ring (3- and 3′ positions) of the thyronine phenolether. Here, we present a similar analysis for MCT10 (Figure 1). The data show that T₃, rT₃, and 3,3′-diiodo-L-thyronine (3,3′-T₂) inhibit MCT10 and are thus potential substrates. T₄, in contrast, was a rather weak inhibitor of MCT10 and may not be a substrate of MCT10 at all. 3-T₁ inhibited MCT10, whereas it is not a known MCT8 substrate or inhibitor. Inhibition of MCT10-mediated T₃ uptake by 3,5-T₂ suggests that MCT10 may be the elusive 3,5-T₂ transporter. Consistent with its known activity as a Trp transporter, MCT10-mediated T₃ uptake was inhibited by 1mM Trp (Figure 1).

Figure 1. Substrate spectrum of MCT10.

Competition experiments were performed by coincubating 10nM ¹²⁵I-T₃ with 1μM various iodothyronines or 1mM Trp in MDCK1 cells stably expressing MCT10. Cell associated radioactivity was measured after 3 minutes. ¹²⁵I-T₃ uptake of cells transfected with pcDNA3 was considered as background and subtracted. Uptake without test compound was set to 0% inhibition. ***, P < .001; NS, not significant vs Dimethyl Sulfoxide.

We wanted to identify those amino acids which prevent MCT10 from transporting T₄. An amino acid sequence alignment between MCT10 and MCT8 revealed 49% identity and 63% similarity (Supplemental Figure 1). Based on our MCT8 homology model (15), we marked those amino acids which lined the presumed substrate translocation channel. Among those amino acid positions that differ between MCT10 and MCT8, we selected 8 positions and prepared an MCT10^MCT8 chimera with the corresponding amino acids mutated to those found in MCT8: Q88H, A90S, V93I, S121A, Y184F, T207S, H317Y, and E325I. For brevity, this mutant is designated MCT10x8. The positions of mutated amino acids are indicated in bold type in the schematic representation of MCT10 (Figure 2A). During the course of this study, we prepared a number of mutant MCT10 proteins, which are summarized in Figure 2B. Compared with MCT10x8, MCT10x6 does not carry the A90S and the V93I exchanges, because these amino acids are bigger in MCT8 than in MCT10, and it is unlikely that bigger amino acids would help accommodate a bigger substrate. We prepared 4 MCT10x4 chimeras, which are permutations of 4 couples of amino acid exchanges from MCT10x8. Only MCT10x4.1 and MCT10x4.2 were informative. MCT10x3 is MCT10x4.2 less the Q88H exchange, which we knew from previous work not to be essential. Finally, 3 MCT10x2 chimeras, representing permutations of the exchanges in MCT10x3, and the respective 3 single mutants were prepared.

Figure 2. Amino acid substitutions in MCT10.

A, Schematic representation of MCT10 based on the model of MCT8 (15). The 8 target amino acids of MCT10x8 are represented in boldface. Known interactions in MCT8 (salt bridge, His-Arg clamp) are indicated for orientation. B, Overview of all amino acid exchanges of the relevant MCT10^MCT8 chimeras. The colors correspond to A.

Biochemical characterization of MCT10^MCT8 chimeric proteins

We then asked whether T₃ uptake activity is preserved in the chimeric MCT10^MCT8 proteins. We performed ¹²⁵I-T₃ uptake experiments in stably transfected MDCK1 cells selected for similar MCT10 expression. In an endpoint assay, ¹²⁵I-T₃ uptake was similar among all wild-type and chimeric proteins (Figure 3A). We then injected equal amounts of in vitro transcribed transporter cRNA into Xenopus oocytes. Chimeras MCT10x8 and MCT10x4.1 exhibited similar T₃ transport activity as MCT8 (Figure 3B).

Figure 3. Biochemical characterization of MCT10^MCT8 chimeric proteins.

A, Uptake of 10nM ¹²⁵I-T₃ after 30 minutes in stably transfected MDCK1 cells expressing various MCT10^MCT8 chimeras and the wild-type MCT8 and MCT10, respectively. Uptake of cells transfected with pcDNA3 was considered as background and subtracted. B, T₃ time-course experiment in Xenopus oocytes. Oocytes injected with the appropriate cRNA were exposed to 10nM ¹²⁵I-T₃ for 3–20 minutes, mock-injected oocytes served as background control. T₃ uptake over time is not changed in MCT10x4.1 and MCT10x8 compared with MCT10. C, Uptake of 10nM ¹²⁵I-T₄ after 30 minutes in stably transfected MDCK1 cells expressing various MCT10^MCT8 chimeras and the wild-type MCT8 and MCT10, respectively. Uptake into mock-transfected cells was subtracted. MCT10x8 and MCT10x6 gained the ability to transport T₄. D, Determination of K_M values for T₄ of MCT10^MCT8 chimeras in relation to MCT8. Uptake into mock-transfected cells was subtracted as background. Measurements were performed with increasing concentrations of T₄ for 3 minutes. Uptake of MCT8 at highest concentration (12.5μM) was set to 100%. E, Eadie-Hofstee plot of data from D, revealing similar K_M values for MCT10x8 and MCT10x6 compared with MCT8. F, Western blotting for anti-HA tag with 10-μL biotinylated and affinity-purified plasma membrane protein fraction of MDCK1 stably expressing MCT10^MCT8 chimeras MCT10 and MCT8 wild type. β-Actin served as loading control. All chimeras reached the cell surface and were expressed in a similar manner. *, P < .5; **, P < .01; ***, P < .001; NS, not significant.

Subsequently we tested whether any of the MCT10^MCT8 chimeras acquired the ability to take up ¹²⁵I-T₄ in stably transfected MDCK1 cells. MCT8 served as the positive control. MCT10x8 and MCT10x6 showed T₄ uptake rates similar to MCT8, whereas MCT10x4.1 and MCT10x2.1 exhibited no significant activity above MCT10 negative controls (Figure 3C). Kinetic analysis yielded similar Michaelis constants for T₄ of 3.0 ± 0.2μM for MCT8, 3.6 ± 0.2μM for MCT10x8, and 4.1 ± 0.4μM for MCT10x6 (Figure 3, D and E). To assess protein expression levels and translocation to the plasma membrane, we biotinylated membrane proteins with the cell-impermeable Sulfo-NHS-SS-Biotin, purified these from cell lysates and performed Western blot experiments with anti-HA tag antibody directed against an N-terminal HA-tag cloned into all expression constructs (Figure 3F) (27). Surface expression levels of the MCT10^MCT8 chimeras analyzed were not reduced in comparison with wild-type MCT10 in MDCK1 cells. Accordingly, up to 8 amino acid exchanges in MCT10 chimeras did not disturb protein translocation to the cell surface.

We then asked whether MCT10^MCT8 chimeric proteins capable of T₄ uptake were also able to release T₄. First, we preloaded MDCK1 cells with T₃ and assessed its accumulation by LC-MS² analysis, exchanged the incubation buffer and then followed T₃ release into buffer. T₃ release was more similar in MCT10^MCT8 cells and MCT8 than in MCT10 cells (Figure 4A). We then preloaded cells with T₄ and again followed its release into the buffer (Figure 4B). As shown in the previous experiment, MCT10 was not able to take up T₄. Again, MCT10x6 and MCT10x8 exhibited T₄ uptake and release very similar to MCT8, whereas MCT10x4.1 was incapable of T₄ accumulation or release.

Figure 4. Efflux of substrates by MCT10 chimeras compared with MCT10 and MCT8.

MDCK1 cells expressing various MCT10^MCT8 chimeras and the wild-type MCT8 and MCT10, respectively, were incubated for 30 minutes with 10nM substrate (A) T₃, or (B) T₄ After substrate-loading, cells were washed briefly with PBS and incubated for 5 minutes with uptake buffer w/o substrate. Substrates from cell lysate and supernatant were extracted and measured by LC-MS². The background of empty-vector transfected cells was subtracted. MCT10x6 and MCT10x8 mirror the uptake and efflux of MCT8, whereas MCT10x4.1 behaved like MCT10. *, P < .5; **, P < .01; ***, P < .001; NS, not significant.

The substrate spectra of MCT10^MCT8 chimeric proteins

To obtain a quick overview whether the amino acid exchanges changed the substrate spectra of MCT10^MCT8 chimeric proteins in comparison with MCT10, a mixture of several iodothyronines at 100nM concentration was prepared, and stably transfected MDCK1 cells expressing MCT10^MCT8 proteins were exposed to the mixture. After 30 minutes, cells were washed and iodothyronines associated with the cells or release into the supernatant were extracted and analyzed by LC-MS². We now included additional MCT10^MCT8 chimeric proteins, MCT10x4.2 (which is another out of 4 permutations of MCT10x4 chimeras with MCT10x4.3 and MCT10x4.4 incapable of T₄ uptake, data not shown) and MCT10x3 (derived from MCT10x4.2 lacking Q88H, because this change does not influence T₄ uptake as shown previously) (21). As demonstrated in above experiments with ¹²⁵I-T₃ or ¹²⁵I-T₄ as substrates, T₃ was taken up by all MCT10^MCT8 chimeras with MCT10x6 and MCT10x8 resembling the efficiency of MCT8, whereas the other chimeras behaved more like MCT10 (Figure 5A). T₄ was not a substrate for MCT10 and MCT10x4.1, whereas MCT10x3, MCT10x4.2, MCT10x6, and MCT10x8 were capable of T₄ transport (Figure 5B). These amino acid exchanges further changed the substrate spectra of MCT10^MCT8 chimeras: uptake of rT₃ was very weak in MCT10 but approached MCT8 levels in MCT10x6 and was only slightly lower in MCT10x8 and MCT10x3. Uptake of rT₃ was intermediate in MCT10x4.2 and still lower in MCT8x4.1 (Figure 5C).

Figure 5. Substrate spectra of MCT10^MCT8 chimeras compared with MCT10 and MCT8.

MDCK1 cells expressing various MCT10^MCT8 chimeras and the wild-type MCT8 and MCT10, respectively, were incubated for 30 minutes with 100nM each of the indicated substrates. Substrates from cell lysate was extracted and measured by LC-MS². The background of empty-vector transfected cells was subtracted. MCT10x3, MCT10x4.2, MCT10x6, and MCT10x8 mirror the substrate spectrum of MCT8, whereas MCT10x4.1 shows the same substrate specificity as MCT10. **, P < .01; ***, P < .001.

The superiority of direct uptake measurements over competition experiments was evident in the case of 3,5-T₂ as substrate. In accordance with the competition experiments in Figure 1, 3,5-T₂ is an MCT10 substrate, but MCT10x3, MCT10x6, and MCT10x8 as well as MCT8 accumulated more 3,5-T₂ than MCT10. Note, however, the different scale as compared with T₃ and T₄ uptake (Figure 5D). The pattern of uptake of 3′,5′-T₂ by MCT10^MCT8-expressing cells resembled the uptake of rT₃ (Figure 5E). MCT10x4.2 appeared to differentiate between substrates carrying 2 iodine atoms in positions 3,5 and 3′,5′, respectively.

Because the LC-MS² measurements confirmed our data of the competitive inhibition studies (Figure 1), except for rT₃, we examined rT₃ uptake more deeply and determined the Michaelis constants (K_M) for rT₃ for one chimera more resembling MCT8 (MCT10x8) and one resembling MCT10 (MCT10x4.1) as well as the 2 wild-type transporters (Figure 6). Kinetic analysis yielded an almost identical K_M value of 6.2 ± 0.8μM for MCT8 and 6.0 ± 0.4μM for MCT10x8. The Michaelis constant of 7.6 ± 2.3μM for MCT10 was in a similar range, but its maximum velocity (v_max) was 80% lower compared with MCT8. Interestingly, the MCT10x4.1 chimera showed an elevated K_M of 21.5 ± 8.6μM, whereas its v_max was increased to 285% compared with the MCT10. These findings confirmed the results of the previous experiment with the substrate mixture (Figure 5C).

Figure 6. Determination of rT₃ K_M values.

A, MDCK1 cells stably transfected with MCT10x4.1, MCT10x8, MCT10, and MCT8 were incubated with increasing concentrations of rT₃ for 5 minutes. rT₃ from cell lysate was extracted and measured by LC-MS². The background of empty-vector transfected cells was subtracted as background. B, Eadie-Hofstee plots of data from A, revealing very similar K_M values for MCT8 and MCT10x8. *, P < .05; **, P < .01; ***, P < .001; NS, not significant.

Two combinations of 2 amino acid exchanges are sufficient to turn MCT10 into a T₄ transporter

Our experiments showed that MCT10x3 and MCT10x4.2 were capable of T₄ uptake and had 2 amino acid exchanges in common (Y184F and T207S). The corresponding mutant was designated MCT10x2.1, albeit it was a weak T₄ transporter. In contrast, the additional presence of S121A in MCT10x3 allowed T₄ uptake at a K_M of 3.8 ± 0.5μM. We therefore combined S121A each with Y184F and T207S substitutions creating chimeras MCT10x2.2 and MCT10x2.3, respectively. In addition, single amino acid replacements of the critical amino acids were prepared and the MCT10^MCT8 chimeric proteins stably expressed in MDCK1 cells (Figure 7A). All these mutants mediated T₃ uptake in the LC-MS² assay (not shown). Both 2-fold chimeras, MCT10x2.2 and MCT10x2.3, were capable of T₄ uptake, whereas the chimeras with only a single amino acid exchange all behaved like MCT10 and did not mediate T₄ uptake (Figure 7B). K_M determinations revealed values of 2.7 ± 0.6μM and 3.4 ± 0.8μM for the MCT10x2.2 and MCT10x2.3 chimeras, respectively, which is in the range of MCT8 or even better (Figure 7, C and D). Increased incubation times of 10 minutes instead of 3 minutes were necessary to perform these measurements, because these mutants mediated slower T₄ uptake than MCT8 (Figure 7E). Although v_max determinations of recombinantly expressed proteins are inherently difficult to interpret in a cellular system, it is obvious that the MCT10x2 and MCT10x3 chimeras mediate slower T₄ uptake than MCT8, because of lower v_max and not because of higher K_M.

Figure 7. Biochemical characterization of additional MCT10^MCT8 chimeric proteins.

All experiments were performed in stably transfected MDCK1 cells expressing various MCT10^MCT8 chimeras and the wild-type MCT8 and MCT10, respectively. Uptake into mock-transfected cells was subtracted as background. A, Western blotting for anti-HA tag with 10-μL biotinylated and affinity-purified plasma membrane protein fraction of MDCK1 cells. β-Actin served as loading control. All chimeras reached the cell surface and were expressed at a similar level. B, Uptake of 10nM ¹²⁵I-T₄ after 30 minutes. MCT10x4.2, MCT10x3.1, MCT10x2.2, and MCT10x2.3 gained the ability to transport T₄, whereas the chimeras with a single mutation behaved like MCT10. C, Determination of K_M values of T₄ transporting MCT10^MCT8 chimeras in relation to MCT8. Measurements were performed with increasing concentrations of T₄ for 10 minutes (MCT8 3 min). Uptake of MCT8 at the highest concentration (12.5μM) was set to 100%. D, Eadie-Hofstee plot of data from C, revealing similar K_M but decreased v_max values compared with MCT8. E, T₄ time-course experiment. MDCK1 cells were exposed to 10nM ¹²⁵I-T₄ for 1–30 minutes. MCT10x2.2 and MCT10x2.3 show a slower uptake of T₄ compared with MCT8.

Effects of structure guided mutagenesis on aromatic amino acid uptake

Do MCT10^MCT8 chimeras that have acquired T₄ as a new substrate retain the ability to transport aromatic amino acids? To answer this question, we performed experiments in which we competed 10nM ¹²⁵I-T₃ with 1mM Trp during uptake experiments with MDCK1 cells transfected with all our MCT10^MCT8 chimeras (Figure 8A). MCT10 served as a positive control, showing strong inhibition, whereas the T₃ uptake of the negative control MCT8 remained almost unaffected by Trp coincubation. A careful analysis revealed that all MCT10^MCT8 chimeras which contained a single amino acid exchange, Y184F, did not show any significant competition of T₃ uptake by Trp.

Figure 8. Effect of amino acid exchanges on amino acid uptake in MDCK1 cells and Xenopus oocytes.

A, Competition uptake of ¹²⁵I-T₃ (10nM) and Trp (1mM) after 10 minutes in MCT10 chimeras compared with wild-type MCT8 and MCT10. All chimeras including the Y184F mutation show no significant inhibition. B, ³H-Phe uptake activity of MCT10 chimera expressed in Xenopus oocytes after 20 minutes. Only the oocytes injected with MCT10-cRNA, which did not include a Tyr184 mutation, showed a clear accumulation of ³H-Phe. *, P < .5; **, P < .01; ***, P < .001; NS, not significant.

In order to formally address aromatic amino acid uptake in these MCT10^MCT8 chimeras, we performed direct ³H-Phe uptake experiments in Xenopus oocytes injected with transporter cRNA. The cells were incubated with 0.5mM ³H-Phe over a period of 20 minutes and MCT8-expressing and mock-injected oocytes served as negative controls, whereas MCT10-injected cells served as positive control. Besides MCT10, only the single mutants MCT10^S121A and MCT10^T207S and the chimeras MCT10x2.3 and MCT10x4.1 showed significant Phe uptake activity (Figure 8B). Phe uptake was incompatible with the Y184F amino acid exchange. These findings identify the critical role of Tyr184 in the transport process of aromatic amino acids by MCT10.

Discussion

A structure-guided approach greatly facilitated this study

The highly homologous proteins MCT10 and MCT8 are both efficient T₃ transporters, yet differ in their substrate spectra. MCT10 is incapable of T₄ transport, whereas MCT8 cannot transport aromatic amino acids. MCT8 is an efficient T₄ transporter but does not transport amino acids. We therefore reasoned that differences in amino acids lining the substrate translocation channel should determine substrate specificity. Given the high homology between MCT10 and MCT8, we hypothesized that exchanging amino acids that occupy the same position in both proteins is unlikely to disrupt protein structure. This tenet held, because all MCT10^MCT8 proteins expressed at similar levels and translocated to the plasma membrane. This study was greatly facilitated by inspection of our MCT8 homology model (15). Amino acid alignment suggests that both MCT8 and MCT10 share the critical amino acids (numbered according to MCT8) Arg445, Asp498, and His415 identified by us and others (15, 20, 21, 23), and both proteins share very similar structures. The MCT8 model helped to select amino acids likely to contact the substrate, because they are predicted to line the interface of the 2 6-helix bundles of which MCT8 and MCT10 are composed. We then prepared a series of MCT10^MCT8 chimeric proteins harboring up to 8 amino acid exchanges (MCT10x8) corresponding to amino acids found in MCT8. Starting from MCT10x8, which exhibited efficient T₄ uptake in stably transfected MDCK1 cells, we were able to identify the MCT10x3 mutant which is only slightly slower than MCT8 in T₄ uptake, and arrived at 2 MCT10x2 mutants which still can transport T₄, but at lower rates. Exchanging only 2 pairs of amino acids, ie, S121A+Y184F and S121A+T207S were enough to bestow MCT10 with the capacity to transport T₄. It is intriguing to see that, in the first instance, only 2 oxygen atoms are effectively removed from the protein, and in the second instance an oxygen and a methyl group are removed from MCT10. Although our initial working hypothesis was to identify those amino acids that directly interact with the 5′-iodine of T₄, it now appears as if these mutations simply create space along the interface of the 2 domains of the transporter. All these amino acid exchanges appeared unlikely choices at the outset of our study. This finding underlines the usefulness of our approach, because 3 out of 515 amino acids means that 0.6% of all MCT10 amino acids were exchanged. If these positions had been known from the beginning, but suitable amino acids had to be found by systematic testing, 20³ = 8000 combinations had to be assessed. The model thus has greatly facilitated our analysis of MCT-mediated iodothyronine transport and allowed to conduct this study with only about 15 mutant proteins. A similar study comparing MCT1 and MCT4 sensitivity towards an inhibitor has been recently performed by Halestrap and coworkers (29). Their homology model of MCT1 is also based on the structure of GlpT (30).

Substrate spectra of MCT10^MCT8 proteins

Although transport of T₄ was our primary goal to achieve, we were clearly interested in how the mutations affected the substrate spectrum of MCT10. Because of the slow uptake of the MCT10x2 chimeric proteins, we studied MCT10x3, which is a more effective T₄ transporter and contains all exchanges present in the MCT10x2 mutants. We employed LC-MS² analyses available in our laboratory to study a number of iodothyronines as substrates. T₃ uptake was not affected by any of the mutants described in this work. Reverse T₃ is a very weak MCT10 substrate. MCT10x8, MCT10x6, and, to a slightly lesser extent MCT10x3, accepted rT₃ as a substrate, but even MCT10x4.1 and MCT10x4.2 displayed rT₃ uptake above MCT10, albeit at high K_M and low velocity. It is interesting that MCT10x8, MCT10x6, and MCT10x3 accepted both 3,5-T₂ and 3′,5′-T₂, whereas MCT10x4.2 effectively enhanced 3′,5′-T₂ uptake, but not 3,5-T₂ uptake, over MCT10.

MCT10 chimeras MCT10x8, MCT10x6, MCT10x4.2, and MCT10x3 all are able to accommodate 2 iodines in the phenolic ring, whereas the MCT10x4.1 chimera apparently does not tolerate the lack of 2 phenolic ring iodines. To delineate the mechanism behind this observation will, ultimately, require an investigation that takes dynamic movements during substrate translocation of the protein and the ligand into consideration. The creation of space, however, to expand the substrate spectrum, has been observed with experiments on Lat2, which acts as a 3,3′-T₂ and (to a lesser degree) T₃ transporter, but does not transport rT₃ and T₄ (31). In a subsequent publication, the authors hypothesized that the substrate specificity in Lat2 is accomplished due to steric hindrance in the binding pocket of the transporter. Indeed, by shortening the side chains of specific amino acids, 3,3′-T₂ transport was increased, whereas side chain enlargements led to a decreased 3,3′-T₂ uptake (17). It should be noted that there may be a fundamental difference between LAT (and related transporters) and MCT family members in that LATs are known to bind the amino acid moiety of the substrate via main chain interactions (as shown in the crystal structures of AdiC and ApcT, see Ref. 17), whereas no such interactions have been described for MCTs. In this respect, the narrow substrate specificity of MCT8 has not yet been elucidated (15).

Ala121 occupies a critical position at the end of the substrate channel

The S121A mutation in MCT10 is common to MCT10x3 and both active MCT10x2 combinations, and is required to expand the substrate specificity of MCT10 to include T₄. The Ser in this position is highly conserved among the MCT family, with MCT8 being the only member which harbors an Ala (Ala224) in this position. At first glance, it seems very remarkable that Ala has such a strong effect on MCT10 substrate specificity, because its side chain is rather nonpolar, and it probably may not interact directly with the substrate but rather interacts with the surface of the domain comprised of TMH7–TMH12. Because of its prominent position in our MCT10 model at the end of the substrate translocation channel (inside open conformation) the change from a Ser to an Ala has consequences for the structure of the transporter. The loss of 1 oxygen atom may create a little more space and allow increased flexibility of the transporter during domain movements and facilitated accommodation the bulkier substrate, T₄.

The critical role of the amino acid at this position is underlined by 2 pathogenic mutations that have been repeatedly found in MCT8, A224T (32, 33), and A224V (2, 34, 35). Both amino acid substitutions introduce small functional groups, ie, hydroxyl and methyl groups, and 2 substituents at the Cβ atom thus reducing flexibility of the side chain. Our findings support our previous hypothesis that Ala224 might be vital in the switching process between the inside-open and the outside-open conformations (36).

Thr207 might have an impact on the substrate pocket size

Thr207 is located in TMH5 at the interface to TMH8. The hydroxyl side chain points towards the substrate channel. The biochemical properties of the exchanged amino acids are very similar, both containing a hydroxyl group, so the difference between both amino acids is the removal of the methyl group at the Cβ position again creating more space at the interface of the 2 domains. Again, the amino acid replacement appears to create space for domain movements rather than specifically probe the properties of the substrate. To our knowledge, a pathogenic mutation at the corresponding position in MCT8 (Ser310) is not known.

Tyr184 is vital for the transport of aromatic amino acids by MCT10

MCT10 is the only member of the MCT family which transports aromatic amino acids and therefore is also known as T-type amino acid transporter. T₃ uptake by MCT10 can be competed with aromatic amino acids (37). Our data give a first hint and demonstrate that Tyr184 is absolutely necessary to allow transport of aromatic amino acids in MCT10. Trp does not interfere with T₃ uptake in mutants containing the Y184F substitution. In addition, the simple removal of the phenolic oxygen at the side chain of amino acid 184 is sufficient to abrogate Phe uptake. The hydroxyl group of Tyr184 thus might directly interact with the amino acid substrate, whereas it interferes with T₄ uptake in MCT10. We are not aware of pathogenic mutations affecting Phe287, located at the corresponding position in MCT8.

In conclusion, we expected to identify amino acid side chains which specifically probe the 5′-iodine in T₄, but identified very minor changes (removal of hydroxyl groups in Y184F and S121A plus the removal of a methyl group in T207S) which appear to create space for easier interdomain movements. This space may be required to accommodate the bulkier T₄ and rT₃ substrates. Clearly, we cannot exclude that other not yet identified amino acids specifically interact with the substrate. Moreover, it remains so far enigmatic, why the hydroxyl of Tyr184 is essential for aromatic amino acid transport. Definite knowledge about transporter/substrate interactions will need to await analysis of crystal structures of MCT8 or MCT10 in different conformations in order to account for the dynamic changes during substrate transport. Until then, our homology model of MCT8 and its comparison with MCT10 may serve as useful approximations.

Funding Statement

This work was supported by Deutsche Forschungsgemeinschaft Grants SCHW914/3-1 and SCHW914/4-1 (to U.S.) and KO922/17-1 (to J.K.) and by grants from Universitätsklinikum Bonn (U.S.).