Characterization of substrates and inhibitors of the human heterodimeric transporter 4F2hc-LAT1 using purified protein and the scintillation proximity radioligand binding assay

Amino acids have diverse and essential roles in many cellular functions such as in protein synthesis, metabolism and as precursors of different hormones. Translocation of amino acids and derivatives thereof across biological membranes is mediated by amino acid transporters. 4F2hc-LAT1 is a heterodimeric amino acid transporter that is composed of two subunits belonging to the SLC3 (4F2hc) and SLC7 (LAT1) solute carrier families. The ancillary protein 4F2hc is responsible for the correct trafficking and regulation of the transporter LAT1. Preclinical studies have identified 4F2hc-LAT1 as a valid anticancer target due to its importance in tumor progression. The scintillation proximity assay (SPA) is a valuable radioligand binding assay that allows the identification and characterization of ligands of membrane proteins. Here, we present a SPA ligand binding study using purified recombinant human 4F2hc-LAT1 protein and the radioligand [3H]L-leucine as tracer. Binding affinities of different 4F2hc-LAT1 substrates and inhibitors determined by SPA are comparable with previously reported K m and IC 50 values from 4F2hc-LAT1 cell-based uptake assays. In summary, the SPA is a valuable method for the identification and characterization of ligands of membrane transporters including inhibitors. In contrast to cell-based assays, where the potential interference with other proteins such as endogenous transporters persists, the SPA uses purified protein making target engagement and characterization of ligands highly reliable.


Introduction
Amino acids are the building blocks of proteins and play critical roles in the human body, e.g., as nutrients, metabolites, precursors of hormones and signaling molecules (Wu, 2009). Specific and controlled transport of amino acids through biological membranes is of fundamental physiological importance and mediated by amino acid transporters (AATs) that are embedded in lipid bilayers (Christensen, 1990;McGivan and Pastor-Anglada, 1994). Consequently, the absence, overexpression or dysfunction of AATs can lead to human diseases (Bröer and Palacin, 2011). Currently, eleven solute carrier (SLC) families containing AATs were reported (Kandasamy et al., 2018). Based on their substrate specificity and mechanism of transport, AATs are classified into different systems (Christensen, 1990;McGivan and Pastor-Anglada, 1994).
The methylotrophic yeast Pichia pastoris is a well-established cellular factory for the production of recombinant mammalian membrane proteins (Byrne, 2015;Looser et al., 2015;Pochini and Galluccio, 2022). Over the years, different human LATs and HATs were successfully overexpressed in P. pastoris for functional and structural studies (Costa et al., 2013;Meury et al., 2014;Rosell et al., 2014;Jeckelmann and Fotiadis, 2019;Kantipudi et al., 2020;Jeckelmann et al., 2022). For example, cell-based transport assays using P. pastoris overexpressing 4F2hc-LAT1 were established and used for the characterization of the specificity for substrates and inhibitors, and their kinetic parameters (Kantipudi et al., 2020;. Recently, a protocol for the production in P. pastoris and purification of human 4F2hc-LAT1 was reported that allows the isolation of milligram amounts of pure, correctly assembled, stable and properly folded heterodimer . For discovering and characterizing membrane transporterspecific ligands, direct, rapid and robust binding assays are of great advantage. Such a binding assay represents the scintillation proximity radioligand binding assay (SPA) (Quick and Javitch, 2007;Harder and Fotiadis, 2012). In the SPA, purified, detergentsolubilized target transporter is bound to scintillation beads and the radiolabeled substrate is added. Substrate binding to the transporter will induce photon emission from the SPA beads, because of the close proximity of the protein-bound radioligand to the scintillant. The luminescence signal is measured and reflects substrate binding. Such an assay has the advantage of being free of endogenous transporters and other proteins, e.g., as present in cellbased assays, thus delivering a clean target protein-ligand interaction-specific information.
Here, we present an extensive ligand-binding study for human 4F2hc-LAT1. The recombinant HAT was overexpressed in P. pastoris, purified and characterized using the SPA. Binding specificities and affinities of selected substrates and inhibitors were successfully determined for human 4F2hc-LAT1. The obtained results were compared and validated with previously published data, e.g., from cell-based transport assays, and are discussed, establishing the SPA as an excellent, robust ligandbinding assay for HATs. transformation into electro-competent P. pastoris strain KM71H cells by electroporation was described previously (Kantipudi et al., 2020). This construct contains recombinant human 4F2hc-LAT1 bearing N-terminal His-(4F2hc) and Strep-(LAT1) tags. The screening and selection protocol of 4F2hc-LAT1 protein expression clones was detailed in . Largescale overexpression in P. pastoris, membrane isolation and protein purification of human 4F2hc-LAT1 using the detergent glycodiosgenin (GDN, Anatrace) were performed as described in detail in .

Scintillation proximity radioligand binding assay experiments
Purified, GDN-solubilized human 4F2hc-LAT1 was attached via the His-tag to polyvinyltolune (PVT) copper His-tag SPA beads (Perkin Elmer) and the protein-bound [ 3 H]L-leucine quantified using a scintillation counter (a detailed description of SPA with 4F2hc-LAT1 is given in the methodological publication by ). Briefly, experiments were conducted in 96-well plates and per well a reaction volume of 50 µL containing 250 µg PVT-SPA-beads, 1.75 µg 4F2hc-LAT1 protein, 2.5 μM L-leucine spiked with 0.5 µCi [ 3 H]L-leucine (ARC/Anawa, 100 Ci/ mmol, 1 mCi/ml) plus a condition-specific substance. All components were dissolved in SPA-Buffer (100 mM BTP pH 8, 150 mM NaCl, 10% (v/v) glycerol, 0.5% (w/v) GDN). For some condition-specific substances (i.e., the thyroid hormones), 20 min sonication in a water bath at 4°C was necessary to dissolve them at 10× concentration in SPA-Buffer. Per reaction, 5 µL of a 10× stock solution of the condition-specific substance was placed in a well and diluted with 45 µL of a master mix containing the remaining components. For determining the substrate specificities To avoid minor unspecific binding of radioactive L-leucine to the beads, the master mix was created in a specific procedure: SPA-beads were first mixed with the unlabeled L-leucine and shaken for 2 h at 4°C before adding the remaining components (protein last). For background subtraction in the bar graph experiments, a noprotein control was included using protein-free master mix (i.e., containing buffer instead of protein solution). The plate was shortly mixed by shaking and incubated at 4°C for~18 h before the signals were counted using a microplate scintillation counter (Trilux Microbeta, Perkin Elmer). [ 3 H]L-leucine protein (full signal) and no-protein (background) SPA samples had typically~3600 and Determination of the human 4F2hc-LAT1 binding specificity for selected amino acids by SPA. All twenty proteogenic L-amino acids and D-leucine were used as competitors. A competitor concentration of 250 µM was used, which corresponds to ten times the K m of 4F2hc-LAT1 for L-leucine (Kantipudi et al., 2020). Residual binding of the radioligand [ 3 H]L-leucine in the presence of competitors was normalized with respect to control samples without competitors (Ctrl). Means with SD from normalized data of three independent experiments (each at least in triplicate) are displayed. The numbers above the bars represent the mean values in %.
Frontiers in Physiology frontiersin.org 03 ~800 CPM, respectively. Thus,~20% of the full signal was background. Three experiments with protein from at least two different purifications were performed, each at least in triplicates.

Data analysis, curve fitting and statistics
Data were analyzed with Prism 6 (GraphPad Software). In each of these experiments, the net binding signals were averaged and the K D value of homologous (L-leucine) and K i values of heterologous (L-histidine, BCH, JPH203, T3 and T4) L-leucine binding competition experiments were determined by fitting the respective sigmoidal model curve to these data. Each experiment was done with sample number of at least three (triplicate). Data of three independent experiments were merged by adding the normalized data to a common data set, which was used for curve fitting. For data analysis of the bar graphs, the signal of the noprotein control was subtracted from the transporter signal to obtain the net binding signal.

Results and discussion
An extensive ligand binding study was performed to evaluate the potential of the SPA on HATs using human 4F2hc-LAT1 as a paradigm. To this aim, recombinant heterodimer was overexpressed in the methylotrophic yeast P. pastoris and purified. Previous work from our laboratory demonstrated that recombinant human 4F2hc-LAT1 expressed and isolated from P. pastoris is correctly assembled, fully functional (Kantipudi et al., 2020;, pure and stable in detergent Frontiers in Physiology frontiersin.org 05 . As radioligand for SPA experiments, [ 3 H]Lleucine was chosen based on our recent results from uptake studies using P. pastoris cells expressing human 4F2hc-LAT1 (Kantipudi et al., 2020;. These previous functional studies identified L-leucine as a high affinity substrate. In a first set of experiments, the specificity of 4F2hc-LAT1 for the twenty proteinogenic amino acids and D-leucine was explored using purified protein and the SPA. Results from [ 3 H]L-leucine binding inhibition to 4F2hc-LAT1 on SPA beads ( Figure 1) indicated a pattern comparable to our previously published [ 3 H] L-leucine uptake inhibition study (Kantipudi et al., 2020), validating the here presented SPA approach. For example, L-leucine and L-histidine inhibited strongly the binding of [ 3 H]L-leucine to 4F2hc-LAT1. The aliphatic amino acids L-isoleucine and L-valine as well as the aromatic ones, i.e., L-phenylalanine, L-tyrosine and L-tryptophan, indicated with residual radioligand binding of 19-29% comparable and significant inhibitions of [ 3 H]L-leucine binding. Situated at the upper end of this range, L-glutamine indicated with 29% residual [ 3 H]L-leucine binding an inhibition comparable to L-phenylalanine. Considering that 4F2hc-LAT1 is involved in the exchange of L-glutamine from inside with L-leucine from outside for intracellular L-leucine mediated mTOR activation (Nicklin et al., 2009), this significant observation makes sense. In stark contrast and again similar to uptake inhibition, the imino acid L-proline did not show any radioligand binding inhibition (Kantipudi et al., 2020;. The other Frontiers in Physiology frontiersin.org 06 proteinogenic amino acids reduced the residual SPA signal to values between 34 and 90% indicating relatively low substrate specificities. Finally, and similar to results from previous uptake experiments (Kantipudi et al., 2020), the D-form of leucine showed about 50% inhibition indicating stereoselective binding of leucine to human 4F2hc-LAT1.
Next, and also for direct comparison with our previously published [ 3 H]L-leucine uptake inhibition data using yeast cells expressing human 4F2hc-LAT1 (Kantipudi et al., 2020;, we determined the K D and K i values of selected substrates and inhibitors, i.e., of L-leucine, L-histidine, BCH, JPH203, triiodothyronine (T3) and thyroxine (T4) using the SPA (Figure 2). For the two best amino acid substrates, we obtained K D and K i values of 26 µM (L-leucine) and 28 µM (L-histidine), respectively. These numbers are in good agreement with the previously obtained K m and IC 50 values of 25 µM (Kantipudi et al., 2020) and 22 µM (L-leucine) , and an IC 50 value of 23 µM (L-histidine) (Kantipudi et al., 2020) using our P. pastoris cell uptake assay. It should be noted that the K i value of L-histidine ( Figure 2B) is also in good agreement with the external K m value of L-histidine determined using proteoliposomes reconstituted with recombinant human LAT1 (Napolitano et al., 2015). Considering that SLC7 family transporters have external and internal K m s for their substrates (Meier et al., 2002;Napolitano et al., 2015;Bartoccioni et al., 2019;Errasti-Murugarren et al., 2019), e.g., the external and internal K m values of 24.6 µM and 2.8 mM for L-histidine as reported in the previously mentioned study using proteoliposomes (Napolitano et al., 2015), our ligand-binding parameters suggest that the external side of 4F2hc-LAT1 is accessible to the here studied ligands when the protein is solubilized and purified with the detergent GDN. For the thyroid hormones T3 and T4, K i values of 2.4 µM and 10 µM were obtained by SPA. These binding constants were comparable or identical to the previously obtained IC 50 values of 1.3 µM (T3) and 10 µM (T4) from [ 3 H]Lleucine uptake inhibition experiments . From the inhibitors, the well-known system L and the human 4F2hc-LAT1 specific inhibitors BCH and JPH203, delivered K i values of 81 µM and 79 nM, respectively. Again, the obtained binding constants were comparable with the IC 50 values of 72 µM (BCH) and 197 nM (JPH203) from [ 3 H]L-leucine uptake inhibition experiments (Kantipudi et al., 2020;.
In summary, K D and K i values from the selected substrates and inhibitors L-leucine, L-histidine, BCH, JPH203, T3 and T4 were in good agreement with the kinetic parameters from the same molecules using a radioligand cell-based 4F2hc-LAT1 uptake assay (Kantipudi et al., 2020;. Thus, the here presented comparative study validated the SPA for new 4F2hc-LAT1 ligand binding experiments. The amino acid substrates L-leucine and L-histidine showed the highest affinities towards human 4F2hc-LAT1 (Figures 2A,B). Therefore, we decided to explore the possibility of identifying new, non-proteinogenic amino acids and derivatives of L-leucine and L-histidine with higher affinities towards human 4F2hc-LAT1.
To this end, SPA competition binding experiments with [ 3 H] L-leucine versus L-leucine and selected non-proteinogenic amino acid were performed (Figure 3). The selected non-proteinogenic amino acid derivatives of L-leucine differ in the number of methyl groups at the γ-carbon atom or in the aliphatic chain lengths with methyl groups at different positions. L-leucine (with two methyl groups at γ-position) inhibited strongly the binding of [ 3 H]Lleucine with only~1% residual radioligand binding (RRB) (Figure 3). The non-proteinogenic amino acid L-neopentyl glycine (with three γ-methyl groups) and L-norvaline (with one γ-methyl group) showed a weaker inhibition with 32-43% RRB (Figure 3). This indicates that with addition or deletion of one methyl group at the γ-position of L-leucine the binding affinity towards 4F2hc-LAT1 decreases drastically. L-valine, with two methyl groups at the β-position instead of γ-position, showed with 27% RRB less competition than L-leucine (Figure 3). Altering the methyl groups on β-position as in the nonproteinogenic amino acids L-tert-leucine (three β-methyl groups) or L-2-aminobutyric acid (one β-methyl group) showed a further decrease of competition potential with 37-45% RRB (Figure 3). The amino acid L-isoleucine, which is an isostereomer of L-leucine is with 28% RRB in a similar range as L-valine ( Figure 3). Therefore, the position of the methyl group in the side chains of L-isoleucine and L-leucine, i.e., β-versus γposition, has a significant effect on the binding to 4F2hc-LAT1. Further increasing the aliphatic chain length by one methyl group with the non-proteinogenic amino acid L-norleucine decreased the competition potential to 62% RRB (Figure 3).
A second, comparable competition SPA experiment was performed to investigate the high-affinity substrate L-histidine and selected non-proteinogenic amino acid derivatives (Figure 4). The amino acid L-histidine inhibits [ 3 H]L-leucine binding to onlỹ 1% RRB (Figure 4). Addition of a methyl group at nitrogen atom either at position 1 (1-methyl-L-histidine) or at position 3 (3methyl-L-histidine) on the imidazole ring of L-histidine decreases radioligand binding to 4F2hc-LAT1 drastically, resulting in 59-64% RRB binding (Figure 4).
Binding studies with methylated forms of L-leucine and L-histidine amino acids such as N-methyl-L-leucine, L-leucinemethyl ester, N-methyl-L-histidine and L-histidine-methyl ester clearly show that there is no significant binding to 4F2hc-LAT1 (Figures 3, 4). These results indicate that a free amino and carboxy group at the α-carbon of L-leucine and L-histidine are essential for binding to human 4F2hc-LAT1.

Conclusion
The scintillation proximity radioligand binding assay is a powerful method to determine the specificity and affinity of ligands towards a specific target protein. Here, we successfully determined binding specificities and affinities of selected substrates and inhibitors for the human heterodimeric 4F2hc-LAT1 transporter by applying the SPA. As radioligand for the assay, the high affinity amino acid [ 3 H]L-leucine was used together with purified 4F2hc-LAT1 protein that was expressed in the methylotrophic yeast P. pastoris. The obtained ligand inhibition pattern (Figure 1), and K D and K i values ( Figure 2) were compared with previously determined data and IC 50 values of the same substrates and inhibitors using an uptake assay with human 4F2hc-LAT1 overexpressing P. pastoris cells (Kantipudi et al., 2020;. Protein-ligand Frontiers in Physiology frontiersin.org 07 parameters from both assays were similar, validating the SPA for future applications with human 4F2hc-LAT1 and enabling the focus on ligand binding. Beyond validation and using this SPA experimental set-up, derivatives of the high-affinity 4F2hc-LAT1 substrates L-leucine and L-histidine were tested (Figures 3, 4). Thus, the SPA also represents an excellent ligand-binding assay to screen for potential new substrates and inhibitors of 4F2hc-LAT1 and possibly other HATs.

Data availability statement
The raw data supporting the conclusion of this article will be made available by the authors, without undue reservation.

Author contributions
DF conceived and designed the study; SK and DH performed the experiments and collected the data; SK, DH, and DF analyzed the data; SK, DH, and DF wrote the manuscript. DF obtained the funding. SK, DH, and DF read, and approved the submitted version.

Funding
This research was funded by the University of Bern and Swiss National Centre of Competence in Research (NCCR) TransCure grant number 185544 from the Swiss National Science Foundation (SNSF).

Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.