Membrane-anchored substrate binding proteins are deployed in secondary TAXI transporters

2.1 Deorphanization of TAXIPm-P

We selected the TAXIPm transporter from the Gram-negative pathogen Proteus mirabilis (Schaffer and Pearson 2015) based on the comparably high heterologous expression levels of the substrate binding protein TAXIPm-P and the membrane domain TAXIPm-QM in Escherichia coli. The latter is composed of a fusion of the small and large membrane domains Q and M, respectively, which is common for TAXI transport systems (Kelly and Thomas 2001). As the SBPs in these type of carriers are essential for transport and determine the substrate specificity (Mulligan et al. 2009), our deorphanization strategy initially focused on TAXIPm-P.

TAXIPm-P was produced without its 24 N-terminal amino acids as a soluble protein in the cytoplasm of E. coli and purified to homogeneity (Figure 1A and B). We used differential scanning fluorimetry (DSF) to screen a library with 31 compounds including C₄-, C₅-dicarboxylates, amino acids, and other known ligands of TRAP transporters (Mulligan et al. 2011; Vetting et al. 2015) (Supplementary Figure 1). Apo-TAXIPm-P showed a slightly biphasic melting curve with the main melting point at 56 °C and a minor shoulder near 47 °C (Figure 1C). We observed reproducible shifts in the T _m in the presence of α-hydroxyglutarate and α-ketoglutarate, but not in the presence of glutamate and glutarate (Figure 1C). This suggests that TAXIPm-P is highly selective as all four compounds are structurally-related and only differ in their substituents at C₂ (Figure 1D). While the thermal stabilization (ΔT _m ) of 5.3 ± 0.0 °C and 3.9 ± 0.1 °C for TAXIPm-P in the presence of α-hydroxyglutarate or α-ketoglutarate, respectively, appears modest, it correlates well with the substrate-induced stabilization observed previously for other substrate binding proteins of TRAP transporters (Gautom et al. 2021; Vetting et al. 2015). To explore the degree of ligand-induced stabilization in more detail we titrated α-ketoglutarate and followed protein unfolding (Supplementary Figure 2). While thermal stabilization initially increased up to a concentration of 12.5 µM with a ΔT _m of 9.3 ± 0.1 °C, the degree of stabilization reduced to a ΔT _m of only 4.0 ± 0.1 °C at 25 µM. At α-ketoglutarate concentrations of 50 µM and above no melting curve could be observed anymore, suggesting destabilization of the protein. The degree of thermal stabilization of TAXIPm-P thus depends on the substrate concentration used.

Figure 1:

Deorphanization of TAXIPm-P. (A) Size-exclusion chromatography of IMAC-purified TAXIPm-P. (B) Coomassie-stained SDS-PAGE analysis of the SEC-peak fraction. (C) Differential scanning fluorimetry of TAXIPm-P in the presence of buffer (designated “no addition”) or 40 µM α-hydroxyglutarate, α-ketoglutarate, glutamate, or glutarate. The peak in the normalized first derivative of the curve indicates the melting temperature. Curves represent average values of normalized data with standard errors of triplicates. For clarity only upward-facing error bars are shown. (D) Structures of the compounds tested in panel C. The change in melting temperature ± the standard error with respect to the ligand-free TAXIPm-P are indicated next to the compound name.

2.2 Substrate specificity of TAXIPm-P

Both α-hydroxyglutarate and α-ketoglutarate are dicarboxylates. These compounds occur in three charge states, neutral, monoanionic, and dianionic, depending on the pH value. To determine which species of α-ketoglutarate binds to TAXIPm-P we performed isothermal calorimetry (ITC) at pH 5.5, 6.0, and 7.5 (Figure 2A). Binding of α-ketoglutarate proved exothermic and equilibrium dissociation constants of 16.6, 15.4, and 8.4 µM were determined at pH 5.5, 6.0, and 7.5, respectively. Based on the pK _a values of 1.9 and 4.4 for α-ketoglutarate, the neutral form is an unlikely substrate given its negligible concentration even at the lowest pH value used (Figure 2B). The modest change in K _D determined for binding at pH 5.5 and 7.5 does not reflect the 70-fold decrease in concentration of the monoanionic species, suggesting that neither monoanionic α-ketoglutarate is the exclusive substrate of TAXIPm-P. This is further supported by the molar ratios at which binding is saturated; these barely vary indicating that the concentration of the species that is bound does not change in the pH range tested. Taken together, these data suggest that TAXIPm-P either exclusively binds dianionic α-ketoglutarate, or that it is indiscriminate towards both the monoanionic and dianionic species.

Figure 2:

pH-dependence in binding affinity of α-ketoglutarate to TAXIPm-P examined by ITC. (A) Top panels show the background-corrected raw heat exchange data associated with α-ketoglutarate binding to TAXIPm-P at pH 5.5 (left panel), pH 6.0 (middle panel), and pH 7.5 (right panel). Lower panels show the binding isotherms obtained from the integration of the heat pulses, normalized per mole of injection, and fitted by a one-site binding model. Individual K_D values are indicated. (B) Percentage of abundance of the different α-ketoglutarate protonation states as a function of pH.

2.3 Deorphanization of TAXIPm-PQM

Both DSF and ITC demonstrated that TAXIPm-P binds α-ketoglutarate. To verify whether this compound is indeed transported by the TAXIPm-QM membrane domain, we established a transport assay in E. coli BW25113(ΔkgtP), which lacks the endogenous α-ketoglutarate permease KgtP (Seol and Shatkin 1992). TAXIPm-PQM was expressed using its native operon structure. The TAXIPm-P signal sequence was substituted with the one of SiaP from Haemophilus influenzae to improve expression levels of the complex and TAXIPm-QM was decorated with a carboxy-terminal GFP. We assessed the folding quality of the membrane domain by comparing the GFP intensity of a DDM-solubilized whole cell lysate before and after ultracentrifugation (Figure 3A) (Marino et al. 2017). As both samples show equal intensities, we conclude that TAXIPm-QM is well-folded under these conditions.

Figure 3:

Cell-based transport assay for TAXIPm-PQM. (A) Assessment of the TAXIPm-PQM folding state upon expression in E. coli BW25113(ΔkgtP). In gel GFP fluorescence analysis of DDM-solubilized whole cell lysates before (minus) ultracentrifugation and the supernatant following ultracentrifugation (plus). (B) Time-dependent uptake of 2 µM α-ketoglutarate in E. coli BW25113 (ΔkgtP)/pEXC3sfGH-TAXIPm-PQM in the presence or absence of 10 µM CCCP. Curves represent average values with standard errors of triplicates. Control samples express TAXIMh-PQM from Marinobacter hydrocarbonoclasticus in order to impose a similar degree of physiological stress from protein overexpression. (C) Kinetics of α-ketoglutarate uptake in E. coli BW25113 (ΔkgtP)/pEXC3sfGH-TAXIPm-PQM. Datapoints represent average values with standard errors of duplicates. The data were fitted to the Michaelis–Menten equation.

We observed robust TAXIPm-PQM-dependent α-ketoglutarate uptake in pre-energized cells, in agreement with the α-ketoglutarate binding to TAXIPm-P (Figure 3B). The transport was completely abolished upon the addition of the protonophore carbonyl cyanide m-chlorophenylhydrazone (CCCP). This preliminary analysis suggests that TAXIPm-QM is driven by the proton-motive force, though great care should be taken in dissecting driving forces in intact cells (Holzhuter and Geertsma 2022). We further determined an apparent K_M for the complete TAXIPm-PQM transport system of 24 µM (Figure 3C), which approximates the K _D determined for TAXIPm-P.

2.4 Characterization of mature TAXIPm-P

Determination of the forces driving solute uptake is best done in vitro. To this end we first purified TAXIPm-QM (Supplementary Figure 3A). Decylmaltoside-solubilized TAXIPm-QM eluted in two peaks during SEC, which may resemble different oligomeric states. Reconstitution into detergent-solubilized liposomes (Mulligan et al. 2009) was inefficient. Only a small fraction of the protein was associated with the liposomes and the majority of that could not be resolubilized from the proteoliposomes using detergent (Supplementary Figure 3B), suggestive of protein aggregation. In contrast, TAXIPm-QM was efficiently reconstituted using a milder approach relying on detergent-doped liposomes and less abrupt detergent removal (Geertsma et al. 2008) (Supplementary Figure 3B). Using proteoliposomes prepared by the latter method, we attempted to measure in vitro α-ketoglutarate uptake. However, despite the high reconstitution efficiency we reproducibly failed to measure any substrate transport in the presence of external soluble TAXIPm-P at a concentration of 5 µM while applying inwardly-directed proton-and sodium-gradients (Supplementary Figure 3C). Similar conditions were previously used successfully for the TRAP transport systems SiaPQM from H. influenzae and V. cholerae (Mulligan et al. 2009, 2012) and appeared to recapitulate all relevant aspects allowing TAXIPm-PQM mediated α-ketoglutarate uptake in intact cells (Figure 3B).

Ultimately, we identified potential differences between the substrate binding proteins used in our in vivo and in vitro transport assays. Substrate binding proteins in Gram-negative bacteria are mostly preceded by a signal sequence that guides the pre-protein into the secretory pathway. Following transport by the Sec translocon this signal sequence is cleaved by Signal Peptidase I and the mature protein remains in the periplasm as a soluble protein. Our implementation of the SiaP signal sequence preceding TAXIPm-P in the cell-based transport assays (Figure 3B) had been imperfect, resulting in a reduction of the cleavage efficiency according to a SignalP prediction (Supplementary Figure 4AB) (Teufel et al. 2022), and potentially leading to a fraction of TAXIPm-P still anchored to the cytoplasmic membrane. Full-length wildtype TAXIPm-P contains an N-terminal stretch of positively charged amino acids followed by a hydrophobic core of approximately 17 amino acids. Analysis by SignalP suggested that the N-terminus does not contain any conventional signal sequence (Supplementary Figure 4C). Our hypothesis was therefore that TAXIPm-P matures to a membrane-anchored periplasmic binding protein, and not a soluble one. To confirm this we expressed TAXIPm-P with and without the first N-terminal 24 amino acids. Analysis by Western blot revealed a higher apparent mass of the full-length protein (Figure 4A). Intact protein mass analysis of purified full-length TAXIPm-P demonstrated that only the N-terminal methionine was removed from the protein (Supplementary Figure 5). Together, this suggests that the N-terminus of TAXIPm-P is relevant for both exposing the substrate binding domain to the periplasm and anchoring it to the cytoplasmic membrane. The difference in the transport activities observed in the in vitro system (Supplementary Figure 3C) and the cell-based system (Figure 3B) might therefore result from the absence of the membrane anchor on TAXIPm-P in our initial in vitro setup. To pursue this hypothesis, we co-reconstituted the TAXIPm-PQM system using the full-length TAXIPm-P and tested for membrane-localization and transport.

Figure 4:

In vitro transport assay for TAXIPm-PQM. (A) Western blot analysis of whole cell lysate from E. coli MC1061 expressing TAXIPm-P as soluble protein (SOL) or as full-length protein (FL). Proteins were detected using an anti-His antibody. (B) Size-exclusion chromatography of full-length TAXIPm-P, initially purified in the presence of 0.005% (w/v) Triton X-100. Purple and pink traces indicate the sample run in the absence of detergent, and subsequent analysis following addition of 0.2% (w/v) DM, respectively. (C) Assessment of the reconstitution efficiency of TAXIPm-PQM. Coomassie-stained SDS-PAGE of DDM-solubilized proteoliposomes before (minus) ultracentrifugation and the supernatant following ultracentrifugation (plus). Black and white arrows indicate TAXIPm-QM and full-length TAXIPm-P, respectively. (D) Time-dependent uptake of 10 µM [¹⁴C]-α-ketoglutarate into TAXIPm-PQM proteoliposomes at 20 °C. H⁺, Na⁺, and α-ketoglutarate refer to inwardly-directed proton- (out: pH 6.0; in: pH 7.5), Na⁺- (out: 50 mM Na⁺; in: 50 mM K⁺; symmetrical pH 7.5 unless otherwise stated), and α-ketoglutarate-gradients (out: 10 µM; in: 0 µM; symmetrical pH 7.5 unless otherwise stated). (E) Uptake of 10 µM [¹⁴C]-α-ketoglutarate into TAXIPm-PQM proteoliposomes at 20 °C driven by a proton-gradient. Arrow indicates the timepoint at which 1 mM [¹²C]-α-ketoglutarate was added to the sample indicated by the orange trace. Datapoints in (D) and (E) represent average values with standard errors of triplicates.

2.5 Co-reconstitution of TAXIPm-PQM

A membrane-anchored substrate binding protein can only diffuse in a two-dimensional space. Given similar expression levels its effective concentration is therefore expected to greatly exceed that of a soluble version. The observed lack of α-ketoglutarate uptake when using soluble TAXIPm-P may therefore be explained by a low binding affinity of the membrane domain TAXIPm-QM for TAXIPm-P.

To determine whether a membrane-tethered TAXIPm-P facilitates uptake in the TAXIPm-PQM reconstituted system, we established a co-reconstitution procedure. Full-length TAXIPm-P was efficiently purified using DDM and Triton X-100 (Supplementary Figure 6). As we scrambled the orientation of TAXIPm-QM in our proteoliposomes by means of sonication, we aimed to co-reconstitute TAXIPm-P in a right-side-out orientation as this would allow exclusive functional analysis of the transport complexes in this respective orientation. To achieve unidirectional reconstitution, we opted to avoid the use of detergent as much as possible. Purified TAXIPm-P solubilized in Triton X-100 was therefore submitted to size-exclusion chromatography in the absence of detergent. This resulted in the formation of large aggregates of TAXIPm-P that eluted in the void volume of the column (Figure 4B). As addition of detergent did dissolve these clusters again, we assume that the overall folding state of the protein was not affected by this treatment. For co-reconstitution we added detergent-depleted TAXIPm-P to proteoliposomes already containing TAXIPm-QM. Using this approach we observed high reconstitution efficiencies for both proteins (Figure 4C).

2.6 In vitro characterization of TAXIPm-PQM

Using proteoliposomes containing TAXIPm-QM and membrane-anchored TAXIPm-P we observed robust uptake of α-ketoglutarate driven by a combined inwardly-directed proton- and Na⁺-gradient (Figure 4D). This confirms the absolute requirement of a membrane-tethered substrate binding protein for transport. Further dissection of the driving forces revealed that the proton-gradient alone suffices for transport. Since the rate of transport and the peak α-ketoglutarate accumulation level measured in the presence of a combined Na⁺- and proton-gradient are comparable to those observed in the presence of a proton-gradient alone, we postulate that TAXIPm-PQM is a proton-dependent symporter.

The directionality of TAXIPm-PQM was subsequently determined using a substrate counterflow experiment (Mulligan et al. 2009). Following an initial accumulation of radiolabeled α-ketoglutarate, we added a 100-fold excess of unlabeled substrate (Figure 4E). For conventional secondary transporters this results in an immediate exchange of the internalized substrate for the unlabeled external compound, thereby reducing the radioactivity in the lumen of the proteoliposomes significantly (Mulligan et al. 2009). Here we do not observe such an immediate strong reduction, suggesting that TAXIPm-PQM operates as a strict unidirectional uptake system.

4.1 Cloning

Wild type genes of TAXIPm-P and TAXIPm-QM from P. mirabilis (ATCC 29,906) were amplified by PCR from genomic DNA, cloned into pINIT_cat by FX cloning (Geertsma and Dutzler 2011), and sequence verified prior to subcloning. The construction of pEXC3sfGH was based on the expression vector pES7SiaPQM (Mulligan et al. 2009). Following elimination of two endogenous SapI sites in the vector backbone, the sequence coding for the suicide cassette, a 3C protease site, superfolder GFP, and a decaHis tag was transferred from pBXC3sfGH to the vector backbone using Infusion cloning (Hamilton et al. 2007). The pEXC3sfGH harbors a pSC101 origin of replication and a bla marker, and allows expression of C-terminally superfolder-GFP- and decaHis-tagged protein from the lac promoter.

4.2 TAXIPm-P expression and purification

E. coli MC1061 transformed with the pBXNH3- or pBXC3H-vector containing the sequence coding for soluble or membrane-anchored TAXIPM-P was cultivated in TB medium supplemented with ampicillin at 37 °C. At an OD₆₀₀ between 1.0 and 1.5 the temperature was shifted to 25 °C and 1 h later cells were induced by the addition of 0.01% (w/v) arabinose for 16 h. For purification of soluble TAXIPm-P cell pellets were stored at −20 °C until use. Cell pellets were resuspended in 20 mM Hepes, pH 7.5, 150 mM NaCl, 10% (w/v) glycerol, supplemented with 1 mg/mL lysozyme, trace amounts of DNAse I, 2 mM MgSO₄, and 15 mM imidazole, and incubated for 1 h at 4 °C. Cells were disrupted using a pressure cell homogenizer from Stanstedt. PMSF was added to 1 mM, and the lysate was cleared by ultracentrifugation (30 min at 186,000 g). All steps of protein purification were carried out at 4 °C. Protein was purified by immobilized metal affinity chromatography (IMAC). Protein was eluted from the Ni-NTA resin using buffer supplemented with 300 mM imidazole, and subsequently treated with HRV 3C protease and dialyzed to 20 mM Hepes, pH 7.5, 150 mM NaCl, 10% (w/v) glycerol. Contaminants were removed by reverse-IMAC and following concentration the protein was subjected to size-exclusion chromatography (SEC) on a Superdex 200 10/300 GL Increase column using 20 mM Hepes pH 7.5 and 150 mM NaCl as running buffer. Protein-containing fractions were pooled, supplemented with 10% (w/v) glycerol, adjusted to a concentration of 10 mg/mL, flash frozen in liquid nitrogen and stored at −80 °C until further usage.

For purification of membrane-anchored TAXIPm-P, cells were disrupted immediately following harvest. Unbroken cells were removed by low-speed centrifugation (30 min at 15,000 g) and membrane vesicles were pelleted by ultracentrifugation (1 h, 186,000 g), resuspended in 50 mM KPi, pH 7.5, 150 mM NaCl, 10% (w/v) glycerol to a concentration of 500 mg wet weight/mL, flash-frozen in liquid nitrogen, and stored at −80 °C until use. Membrane vesicles were solubilized for at least 1 h by ten-fold dilution in 20 mM Hepes, pH 7.5, 150 mM NaCl, 10% (w/v) glycerol, and supplemented with 1.6% (w/v) dodecylmaltoside (DDM). Following removal of insoluble material by ultracentrifugation (30 min at 186,000 g), the solubilized membrane protein was purified by IMAC and SEC, essentially as detailed above but using buffers supplemented with 0.05% (w/v) Triton X-100. The peak fraction was submitted to another round of SEC using a running buffer devoid of detergent to assure reversible aggregation. Peak fractions were pooled, supplemented with 10% (w/v) glycerol, flash frozen in liquid nitrogen and stored at −80 °C until further usage.

4.3 TAXIPm-QM expression and purification

E. coli MC1061 transformed with the pBXC3sfGH-vector containing the sequence coding for TAXIPm-QM was cultivated in TB medium supplemented with ampicillin at 37 °C. At an OD₆₀₀ between 1.0 and 1.5 the temperature was shifted to 25 °C and 1 h later cells were induced by the addition of 0.01% (w/v) arabinose for 16 h. Cells were disrupted immediately following harvest. Unbroken cells were removed by low-speed centrifugation (30 min at 15,000 g) and membrane vesicles were pelleted by ultracentrifugation (1 h, 186,000 g), resuspended in 50 mM KPi, pH 7.5, 150 mM NaCl, 10% (w/v) glycerol to a concentration of 500 mg wet weight/mL, flash-frozen in liquid nitrogen, and stored at −80 °C until use. Membrane vesicles were solubilized for at least 1 h by ten-fold dilution in 20 mM Hepes, pH 7.5, 150 mM NaCl, 10% (w/v) glycerol, and supplemented with 1.6% (w/v) decylmaltoside (DM). Following removal of insoluble material by ultracentrifugation (30 min at 186,000 g), the solubilized membrane protein was purified by IMAC, essentially as detailed above but using buffers supplemented with 0.2% (w/v) DM, and eluted from the column by a 2 h incubation with HRV 3C protease. The protein was subjected to SEC on a Superdex 200 10/300 GL Increase column using 20 mM Hepes pH 7.5, 150 mM NaCl, and 0.2% (w/v) DM as running buffer. Protein-containing fractions were pooled, supplemented with 10% (w/v) glycerol, flash frozen in liquid nitrogen and stored at −80 °C until further usage.

4.4 Reconstitution into liposomes

Proteoliposomes were essentially prepared as described previously (Chang et al. 2019; Geertsma et al. 2008). In brief, purified membrane proteins were mixed with Triton X-100-destabilized liposomes composed of L-α-phosphatidylcholine from soy bean at a weight-to-weight ratio of 1:50 (protein:lipid) and detergent was subsequently removed by step-wise addition of Biobeads (BioRad). The resulting proteoliposomes were collected by ultracentrifugation at 200,000 g for 30 min at 15 °C and resuspended in 50 mM KPi or NaPi, pH 7.5, supplemented with 2 mM MgSO₄ to a lipid concentration of 20 mg/mL. To randomize the orientation of the reconstituted protein the proteoliposomes were sonicated in six cycles (10 s on/50 s off). Proteoliposomes were subjected to five freeze/thaw cycles and stored in liquid nitrogen until use. Reconstitution efficiency was determined as described (Holzhuter and Geertsma 2022). For co-reconstitution of membrane-anchored TAXIPm-P, TAXIPm-QM-containing proteoliposomes were thawed and extruded 11-fold through 400 nm pore size polycarbonate filters. Afterwards temporary aggregated TAXIPm-P holding its membrane-anchor was mixed with the proteoliposomes to achieve a TAXIPm-QM:TAXIPm-P ratio of 1:0.1 (weight-to-weight). The sample was incubated for 15 min at room temperature while rocking. Proteoliposomes were subsequently collected by ultracentrifugation (30 min at 200,000 g), resuspended in 50 mM KPi, pH 7.5, supplemented with 2 mM MgSO₄, to a lipid concentration of 100 mg/mL.

4.5 Differential scanning fluorimetry (DSF)

For DSF analysis, 10 µM soluble TAXIPm-P was mixed with potential ligands at a final concentration of 40 µM. Sypro Orange, purchased at a concentration of 5000Χ, was diluted 50-fold to reach a 100Χ substock. The 100Χ substock was subsequently diluted 20-fold in 20 mM Hepes, pH 7.5, 150 mM NaCl, 10% (w/v) glycerol containing TAXIPm-P and a potential ligand to reach a final Sypro Orange concentration of 5Χ. Samples were subjected to thermal melting in an RT-PCR cycler (Rotor-Gene Q, Qiagen). Protein was incubated from 25 °C up to 95 °C in steps of 1 °C with 1 min per incubation step, and Sypro Orange fluorescence was recorded (excitation: 470 nm, emission: 555 nm).

4.6 Isothermal titration calorimetry (ITC)

IMAC-purified soluble TAXIPm-P was thawed, subjected to SEC in 50 mM KPi, pH 5.5, 6.0, or 7.5, and peak fractions were concentrated to 30 µM, degassed, pre-warmed to 18 °C, and transferred to the reaction cells of the ITC device (MicroCal VP-ITC, Malvern). The ligand was dissolved in the same buffer as the respective protein preparation, adjusted to 600 µM and filled in the syringe of the ITC device. Once the reaction cell reached 20 °C, the ligand was titrated to the protein solution in 10 µL steps, resulting in a gradual increase of the ligand concentration from 6.4 µM to 95.5 µM. Data was evaluated using the software NITPIC for automatic baseline correction and subtraction of dilution heat, the software Sedphat to set the binding parameters to heteroassociation assuming a 1:1 binding stoichiometry and fitting the data, and the software GUSSI for plotting the data (Brautigam et al. 2016).

4.7 Radiolabeled transport assay on whole cells

Radioactive transport assays of α-ketoglutarate by TAXIPm-PQM were performed using E. coli BW25113 (ΔkgtP) transformed with pEXC3sfGH containing a sequence coding for TAXIPm-PQM in which the first twenty amino acids of TAXIPm-P were replaced with the first 21 amino acids of SiaP from H. influenzae (Supplementary Figure 4). Cells were grown overnight in LB medium, supplemented with 100 μg/mL ampicillin. The next day, 10 mL LB/amp were inoculated with 1% of the pre-culture. The cells were grown in tubes with a gas-permeable lid at 37 °C until an OD₆₀₀ of 0.6–0.8 was reached, followed by induction with 1 mM IPTG for 2 h at 37 °C. All subsequent steps were performed at 4 °C. Cells were harvested for 10 min at 2500 g, washed twice using 50 mM KP_i, pH 7.5, supplemented with 2 mM MgSO₄, and adjusted to an OD₆₀₀ of 100. The transport assay was performed at 30 °C in 50 mM KP_i, pH 6.0 or 7.5, supplemented with 2 mM MgSO₄, 0.1% (w/v) glucose. If needed, 10 µM carbonyl cyanide m-chlorophenylhydrazone (CCCP) was added to dissipate the proton-gradient. Following thermal equilibration of the buffer for 1 min, cells were added to a final OD of 3 and pre-energized for 2 min. Samples were continuously stirred. The transport reaction was started by the addition of 2 µM [¹⁴C]-α-ketoglutarate and stopped at dedicated timepoints by transferring 100 µL into 2 mL ice-cold 50 mM KP_i, pH 7.5, supplemented with 2 mM MgSO₄, and immediate filtrated over an 0.45 µm pore size nitrocellulose acetate filter. Subsequently, the filters were washed with another 2 mL ice-cold buffer and dissolved in 4 mL scintillation liquid overnight. The next day, the radioactivity was determined using a scintillation counter. TAXIPm-QM expression levels and folding quality were determined based on the C-terminal GFP-fusion as described (Marino et al. 2017).

4.8 Radiolabeled transport assay on proteoliposomes

Transport assays on TAXIPm-PQM proteoliposomes were performed at 20 °C in 50 mM KP_i, pH 6.0, or 50 mM NaPi, pH 6.0, supplemented with 2 mM MgSO₄ and 10 µM [¹⁴C]-α-ketoglutarate. A 25 µL aliquot of proteoliposomes in 50 mM KPi, pH 7.5, supplemented with 2 mM MgSO₄, at a lipid concentration of 100 mg/mL and a TAXIPm-QM concentration of 2 mg/mL were diluted in 575 µL of the respective outside buffer to a final TAXIPm-QM concentration of 83 μg/mL. Samples were continuously stirred. At dedicated timepoints the transport reaction was stopped by transferring 100 µL into 2 mL ice-cold 50 mM KPi, pH 7.5, supplemented with 2 mM MgSO₄, and immediate filtrated over an 0.45 µm pore size nitrocellulose acetate filter. Subsequently, the filters were washed with another 2 mL ice-cold buffer and dissolved in 4 mL scintillation liquid overnight. The next day, the radioactivity was determined using a scintillation counter.