ECD of PepT1 interacts with TM1 to facilitate substrate transport

Mammalian peptide transporters, PepT1 and PepT2, mediate uptake of a wide variety of di- and tri-peptides and are essential for the absorption of dietary peptides in the digestive tract and the recovery of peptides in renal filtrate. PepT also mediates absorption of many drugs and prodrugs to enhance their bioavailability. PepT has 12 transmembrane (TM) helices that fold into two domains, the N-terminal domain (NTD, TM1-6) and C-terminal domain (CTD, TM7-12), and a large extracellular domain (ECD) of ∼200 amino acids between TM9 and TM10. It is known that peptide transport involves large motions of the N- and C-domains, but the role of ECD remains unclear. Here we report the structure of PepT1 from Equus caballus (horse) determined by cryo-electron microscopy. The structure shows that ECD interacts with TM1 and bridges the N- and C-domains. Deletion of the ECD or mutations to the TM1-ECD interface both impair the transport activity. These results demonstrate a role of ECD in structure and function of PepT1 and enhance our understanding of the mechanism of transport in PepT1.


Introduction
Mammalian peptide transporters, PepT1 and PepT2, are members of the solute carrier (SLC) transporter family 15 1 . PepT1 is mainly found on intestinal brush border membranes and mediates uptake of small peptides 2,3 . PepT2 is found in epithelial cells in kidney and retrieves peptides from the glomerulus filtrate 1,4,5 . Due to their broad substrate spectrum, PepT1 and PepT2 also transport a large variety of small-molecule drugs such as antibiotics and antiviral drugs, and prodrugs with amino acids as part of the scaffold 6 . The concentrative uptake of peptides or small-molecule drugs by PepT is achieved by co-transport of protons and is thermodynamically driven by both the pH gradient and the membrane potentials 1 .
The structures show that the transition between the inward-and outward-facing conformations comes from rigid-body motions of the NTD and CTD. This is often referred to as the rock-switch model of alternating access, and is shared by other transporters of the major facilitator superfamily (MFS) fold 22 .
Different from bacterial homologs, mammalian PepT1 and PepT2 have an extracellular domain (ECD) composed of ~200 residues located between TM9 and TM10. The molecular weight of ECD is approaching to that the NTD or CTD. The ECD has a well-defined structure composed mainly of β-strands folded into two immunoglobin-like subdomains 7,8,23 . The structure of human PepT2 captured in the inward-facing conformation (PDB ID: 7PMY) shows that the ECD is in close proximity to the NTD, but no specific interactions were identified 7 . A previous mutational study showed that the PepT2 function is not perturbed by deletion or swapping of its ECD with that of PepT1 23 , however, the role of ECD in PepT1 was not assessed because PepT1 with its ECD deleted or swapped with that of PepT2 do not have sufficient level of expression in the Xenopus oocytes system 23 .
In this study, we determined the structure of PepT1 from horse (Equus caballus) in the inward-facing conformation. The structure shows that its ECD interacts with the NTD, which prompted us to examine the function of ECD in PepT1. We measured the transport activities of PepT1 in human embryonic kidney (HEK) cells, and found that deletion of the ECD or mutations to the interface of ECD and NTD significantly reduce the rate of substrate transport. These results led us to conclude that interactions between ECD and NTD are functionally relevant and may facilitate transition of PepT1 from the outward-facing to the inward-facing conformation.

Proton-coupled peptide transport by horse PepT1
To examine transport activity of horse PepT1, we expressed it in HEK293 cells and measured uptake of peptide and proton. As shown in Fig. 1a, cells expressing horse PepT1 accumulate a labeled dipeptide ( 3 H-Ala-Ala) while cells transfected with a control vector do not accumulate significant amount of the peptide (Fig. 1a). We also measured co-transport of protons with peptide substrates by first loading the cells with a pH sensitive dye and monitoring intracellular pH changes. Proton uptake occurs in the presence of the dipeptide substrate (Ala-Ala, 1mM), and lower the external pH, higher the amount of proton uptake ( Fig. 1b & 1c, and Extended Data Fig. 1). As a further test that the observed proton uptake is induced by peptide transport, we measured pH changes in the presence of two additional di-peptides, Glu-Glu, and Gly-Sar, and found that although all three dipetides are transported, Ala-Ala inducing the largest fluorescence increase (Extended Data Fig. 1d). The expression level of PepT1 between different batches of transfection was estimated by Western blot (Methods and Extended Data Fig. 2) and the amount dipeptide or proton uptake was normalized accordingly. These results indicate that horse PepT1 is a proton coupled peptide symporter similar to human PepT1 24 .

Cryo-EM structure of horse PepT1 in nanodisc
We expressed and purified the horse PepT1 from Sf9 (Spodoptera frugiperda) cells and reconstituted the protein into lipid nanodiscs for structure determination by cryo-electron microscopy ( Fig. 2 and Methods). The elution volume of horse PepT1 from a size-exclusion chromatography column is consistent with it being a monomer (Fig. 2a). PepTs of human and rat origins were also shown to exist as a monomer 7,8 . The amino acid sequence of horse PepT1 is 84% identical and 96% similar to that of human PepT1. The soluble ECD served as a fiducial marker for particle alignment during 3D reconstruction, and we were able to obtain a cryo-EM map with an overall resolution of ~3.6 Å (Fig 2b, and Extended Data Fig. 3). The map for the TM domain, which is mainly composed of α-helices, is of sufficient quality for de novo model building.
Although the map for the ECD, which is mainly composed of β-strands and not constrained by nanodisc, has modest resolution, the two subdomains and individual β-strands are readily recognizable. The overall structure of horse PepT1 is shown in Fig. 2c, and the fitting of individual TMs and ECD to the density map is shown in Extended Data Fig. 4.

Comparison of horse PepT1 to existing PepT structures
The horse PepT1 is captured in the inward-facing conformation, with the extracellular sides of its NTD and CTD making contacts and leaving the substrate binding site solvent accessible from the intracellular side. The overall structure of horse PepT1 aligns well with an inward-facing human PepT2 structure (PDB ID: 7PMY) 7 with an overall root-mean-square deviation (RMSD) of 2.5 Å (Extended Data Fig 5a).
The NTD or CTD of horse PepT1 aligns well to the NTD or CTD of mammalian PepTs of PepT1 and PepT2 by less than 30° rotation, the ECD in rat PepT2 seems to be an outlier (~79° rotation). The large change of ECD in rat PepT2 is likely caused by the presence of a nanobody used to facilitate structure determination 8 . These results suggest that the linkage between ECD and CTD is not rigid and allows flexibility for independent movement of the two domains.
The substrate binding pocket of horse PepT1 is similar to that of human PepT2 7 , and composed of conserved positive residues, Arg35 and Lys141, from the NTD, and negative residues, Asp299 and Glu595, from the CTD (Fig. 2d). The substrate is clamped in a cavity of opposite electric charges with its N-terminus facing negative charged surface in the CTD and C-terminus facing the positively charged surface in the NTD (Fig. 2e).
We mutated residues that line the substrate binding site in horse PepT1, and measured their transport activity ( Fig. 2f and Extended Data Fig. 2). Significant reduction in transport activity was observed for all the mutations, and this result is consistent with similar experiments in previous reports 25, 26 . These results highlight the importance of these conserved charged residues in the recognition of N-and C-terminal backbone groups on substrates. The conserved interactions with the backbone of substrates afford a conserved binding mode that is less sensitive to the side chain identity, which serves as a structural basis of the extreme substrate promiscuity of PepT1.

Extracellular gates in PepT1
We next examined interactions between the extracellular sides of NTD and CTD, which serve as a gate to limit access of the substrate binding site to the external side. NTD and CTD make contact in three places. Asn51 in TM2 interacts with the Arg304 in TM7 (Extended Data Fig 6a).
His58 in TM2 interacts with Asn630 in TM11, and the latter also interact with Asp299 in the substrate binding pocket (Extended Data Fig 6b). Arg186 at the end of TM5 interacts with Gln322 and Asp324 in the loop connecting TM7 and TM8 (Extended Data Fig 6c). Some of these interactions are known to be important for peptide transport, for example, His58 was implicated in mediating proton coupled transport in human PepT1 15 . Similar interactions are also observed in the structure of human PepT2 in the inward-facing conformation 7 .
The structure of horse PepT1 reveals interactions between ECD and NTD that was not observed in the structure of human PepT2 inward-facing conformation. Lys483 on the ECD interacts with the backbone carbonyl oxygen atoms of Leu45 and Phe46 at the C-terminal end of TM1 (Fig. 3a). The side chain of Lys483 is well-resolved in the density map. In addition, the positively charged Lys483 is further stabilized by a cation-π interaction with Phe45 on TM1 and by the helical dipole moment of TM1. Such ECD-NTD interactions do not exist in human PepT2 as the equivalent residue to the Lys in PepT1 is an Asn of no net charge on the side chain (Extended Data Fig. 7). Structural analysis on the extracellular side of TM1 of human PepT2 7 shows that a histidine (His76) on the TM1-TM2 loop caps the end of TM1 (Fig. 3b), while equivalent residue in PepT1 is a glycine (Gly47).
We then examined functional impact of the interactions between ECD and NTD. We first constructed a mutant horse PepT1 with its ECD deleted (ΔECD). ΔECD can be expressed in HEK cells and mediates proton and peptide transport, however, the rate of transport is significantly lower than these of the wild type ( Fig. 1a and Fig. 3c-d). We then mutated Lys483 to Ala, and found that Lys483Ala has significantly reduced transport activity (Fig. 3c-d). As a control, we made an alanine mutation to Glu482 which is close to Lys483 but does not interact with TM1.
Glu482Ala has a slightly reduced rate of transport compared to that of the wild type, but is substantially faster than that of Lys483Ala or the ΔECD. These results indicate that the interactions between ECD and NTD facilitate peptide transport. Further studies are required to understand how the interactions change the energy landscape of substrate transport in PepT1.

Discussion
The horse PepT1 is the first mammalian PepT1 captured in the inward-facing conformation.
The structure identified interactions between ECD and NTD, and further studies showed that the interaction has a significant functional impact. K483, which is a key residue on ECD involved in the interaction, is highly conserved in PepT1 but not in mammalian PepT2 (Extended Data Fig.   7). This may explain why the interaction was not observed in the structure of human PepT2 captured in the inward-facing state 7 .
The ECD-NTD interactions in PepT1 likely contribute to different transport kinetics of PepT1 and PepT2. It is well documented that the PepT1 has higher transport capacity, i.e., higher rate of transport, than PepT2 27 . The stoichiometry of the coupled protons is also different in PepT1 and PepT2 24,28 . The PepT1 was reported to require less proton to complete a transport cycle 28 .
Since the substrate binding pocket of PepT1 and PepT2 are highly conserved 7 , differences in the transport kinetics may arise from the extra ECD-NTD interactions unique to PepT1. Consistent with this notion, removal of ECD in the human PepT2 resulted in no significant changes in substrate transport 23 , while our studies showed that deletion of ECD in PepT1 reduces the rate of transport. Further analysis will help pinpoint the contribution of ECD to the function of PepT1.

Cloning, expression, and purification of horse PepT1
The PepT1 gene from horse (Equus caballus, UniProt ID: F6SG69) was codon-optimized and cloned into a pFastBac dual vector for production of baculovirus by the Bac-to-Bac method (Invitrogen). Sf9 cells (Invitrogen/Thermo Fisher) at a density of ~3´10 6 cells/ml were infected with baculovirus and grown at 27 °C for 60-70 hour before harvesting. Cell pellets were frozen in liquid nitrogen and stored in −80 ℃ fridge before purification.

Nanodisc reconstitution
MSP1D1 was expressed and purified following as described before 29

Model building and refinement
The initial model of horse PepT1 was generated by AlphaFold2 41 . The TMD and ECD were individually docked into the cryo-EM density map in Chimera 42 . The map was sharpened in a convolutional neural network-based algorithm, DeepEMhancer 43 , with the highRes model. Processed map has significantly better resolvability of side chains and was used for model refinement. The docked model was manually adjusted in COOT 44 , and subjected to real-space refinement with secondary structure and geometry restraints in Phenix 45 . The EMRinger Score 46 was calculated. All structure figures were prepared in ChimeraX 47 .

Expression of horse PepT1 in HEK cells
The cDNA of horse PepT1 was cloned into a pEG BacMam vector with a C-terminal GFP tag. Mutations to horse PepT1 were generated using the QuikChange method (Stratagene) and the entire cDNA was sequenced to verify the mutation. The primers information is provided in Extended Data Table 2.
The HEK 293S cells in FreeStyle 293 media (Invitrogen/Thermo Fisher) supplemented with 2% fetal bovine serum (FBS; Sigma) were maintained at 37 ℃ with 8% CO2 in suspension culture at 100 rpm. Cells were plated one day before transfection to reach >90% confluency.
Transfection of horse PepT1 plasmid or empty plasmid was performed with 293fectin transfection reagent (Invitrogen/Thermo Fisher) as per manufacturer's instruction. Transfected cells were incubated at 37 ℃ with 8% CO2 for 2 days. Cells on the plate were washed in PBS before scraping.
Cell membranes were solubilized in the lysis buffer plus 1% LMNG and Protease Inhibitor Cocktail (Roche) for 1 h at 4°C. Insoluble fractions were pelleted by centrifugation and supernatants were run in SDS-PAGE. Bands of target proteins were visualized by western blotting with mouse anti-GFP (Invitrogen/Thermo Fisher) antibodies as primary antibodies and IRDye-800CW anti-mouse IgG (Licor) as secondary antibody. Images were taken on an Odyssey infrared scanner (Licor).

Peptide uptake assay
The HEK 293S cells were plated on 6-well plates and transfected with the pEG BacMam plasmids with horse PepT1 as described above. Transfected cells were harvested after 2 days and transferred to 1.5 mL Eppendorf tubes. Cells were resuspended in transport buffer (20 mM HEPES, pH6.9, 140 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl2, 1.0 mM MgCl2) shortly before assays. The ligand Ala-Ala were filtered through 0.45 µm nitrocellulose filters (Millipore). Cells were lysed in 0.1M NaOH and 1% SDS for 10 min. The radioactivity retained on the filters was determined by liquid scintillation counting. Counts per minute were converted to pmol by comparing to a standard curve plotted with known amounts of 3 H-Ala-Ala. One-way ANOVA followed by the Dunnett's post hoc test was performed. All statistical analyses were performed in GraphPad Prism 8.2.1.

Proton transport assay
HEK293S cells from suspension culture were adjusted to a density of 0. Fluorescence readings at equilibrium (550 -600 s) were averaged to represent intrasellar pH changes. Two-way ANOVA was used to compare the effect of outside pH. Two-tailed Student's t-test was performed for comparison of mutants to the WT.

Acknowledgments
This work was supported by grants from NIH (DK122784, HL086392, and GM098878 to M.Z.), Cancer Prevention and Research Institute of Texas (R1223 to M.Z.). We acknowledge the use of Princeton's Imaging and Analysis Center, which is partially supported by the Princeton Center for Complex Materials, and the National Science Foundation (NSF)-MRSEC program (DMR-1420541).

Data Availability
The atomic coordinate file of horse PepT1 in nanodisc has been deposited in the PDB