Molecular mechanisms of drug transport through peptide transporters
<p>Proton-coupled oligopeptide transporters (POTs) — members of the major facilitator superfamily (MFS) with homology across the tree of life — couple the intestinal uptake of short peptides to the downhill symport of protons in an alternating access mechanism. They are of great pharmaceutical...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
2024
|
| Subjects: |
| _version_ | 1817931100899508224 |
|---|---|
| author | Lichtinger, SM |
| author2 | Biggin, PC |
| author_facet | Biggin, PC Lichtinger, SM |
| author_sort | Lichtinger, SM |
| collection | OXFORD |
| description | <p>Proton-coupled oligopeptide transporters (POTs) — members of the major facilitator superfamily (MFS) with homology across the tree of life — couple the intestinal uptake of short peptides to the downhill symport of protons in an alternating access mechanism. They are of great pharmaceutical interest owing to their substrate promiscuity and role in the oral bioavailability of several classes of approved drugs, including β-lactam antibiotics. The molecular mechanism by which they couple proton transfer, substrate binding and large-scale conformational changes have remained elusive, however, and structure-guided or computer-aided drug design for POT-mediated uptake has not yet been accomplished.</p>
<br>
<p>Recent cryo-EM structures of mammalian POTs provide a unique opportunity to investigate them
using molecular dynamics (MD) simulations, yet MD is beset by sampling problems due to limited accessible time scales. Several classes of enhanced sampling algorithms are available, however they frequently suffer from a bias towards the simulation starting state, a phenomenon known as hysteresis. We therefore first devised a strategy to address this shortcoming which we term MEMENTO (Morphing Endstates by Modelling Ensembles with iNdependent TOpologies), deriving history-free paths for umbrella sampling by using template-based modelling to fix unphysical intermediates from coordinate interpolation. We have developed this into an open-source python package, and validated the approach on four test systems of increasing complexity.</p>
<br>
<p>We then applied our method — together with free-energy and constant-pH simulation techniques —
to derive a sequence of protonation steps that drive alternating access in POTs, validating our predictions in cell-based transport assays. In particular, we demonstrate the function of an extracellular gate that is only conserved in mammalian POTs, while showing that universally conserved glutamate residues act as intracellular gate-opening triggers. Sequential trans-membrane movement of protons via these residues explains the uni-directional transport exhibited by POTs. We also examined the role played by the presence of substrate in the binding pocket, and suggest that it both exerts a subtle conformational bias towards inwards-facing protein conformations and enables proton translocation by perturbing the pKa value of a key glutamate side chain via a salt-bridge swap mechanism. These results are corroborated by a series of new ligand-bound cryo-EM structures, where we showed using further MD simulations that the ability to engage a conserved arginine residue that triggers the salt-bridge swap distinguishes between good and bad substrates, not substrate affinity to one protein conformation alone.</p>
<br>
<p>We lastly benchmarked docking, absolute binding free energy (ABFE) calculations and machine
learning models trained using ligand-only features on a retrospective dataset of 184 transport inhibition constants at PepT2. We show how the prediction of accurate protein-ligand binding poses constitutes a crucial limitation for the structure-based methods, and that the ligand-based machine learning models, while giving much better correlations to experimental data, fall short in their ability to capture pharmaceutically important activity cliffs. We have also started work on using these methods prospectively to design prodrugs for transport by POTs, though our efforts thus far have been hampered by solubility problems.</p>
<br>
<p>In summary, this thesis covers ground from the development of novel computational tools to their
application to the POT transport mechanism, and reaching further towards <em>in-silico</em> design of prodrugs for increased bioavailability. It therefore advances the fields of enhanced sampling of conformational changes and membrane transporter biology, with the potential of benefitting their applications to pharmacology in the future.</p> |
| first_indexed | 2024-12-09T03:16:39Z |
| format | Thesis |
| id | oxford-uuid:5bd3474a-b08f-4899-b583-e5fdbe92b50e |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-12-09T03:16:39Z |
| publishDate | 2024 |
| record_format | dspace |
| spelling | oxford-uuid:5bd3474a-b08f-4899-b583-e5fdbe92b50e2024-10-22T14:38:02ZMolecular mechanisms of drug transport through peptide transportersThesishttp://purl.org/coar/resource_type/c_db06uuid:5bd3474a-b08f-4899-b583-e5fdbe92b50eComputational chemistryBiochemistryBiophysicsEnglishHyrax Deposit2024Lichtinger, SMBiggin, PCNewstead, S<p>Proton-coupled oligopeptide transporters (POTs) — members of the major facilitator superfamily (MFS) with homology across the tree of life — couple the intestinal uptake of short peptides to the downhill symport of protons in an alternating access mechanism. They are of great pharmaceutical interest owing to their substrate promiscuity and role in the oral bioavailability of several classes of approved drugs, including β-lactam antibiotics. The molecular mechanism by which they couple proton transfer, substrate binding and large-scale conformational changes have remained elusive, however, and structure-guided or computer-aided drug design for POT-mediated uptake has not yet been accomplished.</p> <br> <p>Recent cryo-EM structures of mammalian POTs provide a unique opportunity to investigate them using molecular dynamics (MD) simulations, yet MD is beset by sampling problems due to limited accessible time scales. Several classes of enhanced sampling algorithms are available, however they frequently suffer from a bias towards the simulation starting state, a phenomenon known as hysteresis. We therefore first devised a strategy to address this shortcoming which we term MEMENTO (Morphing Endstates by Modelling Ensembles with iNdependent TOpologies), deriving history-free paths for umbrella sampling by using template-based modelling to fix unphysical intermediates from coordinate interpolation. We have developed this into an open-source python package, and validated the approach on four test systems of increasing complexity.</p> <br> <p>We then applied our method — together with free-energy and constant-pH simulation techniques — to derive a sequence of protonation steps that drive alternating access in POTs, validating our predictions in cell-based transport assays. In particular, we demonstrate the function of an extracellular gate that is only conserved in mammalian POTs, while showing that universally conserved glutamate residues act as intracellular gate-opening triggers. Sequential trans-membrane movement of protons via these residues explains the uni-directional transport exhibited by POTs. We also examined the role played by the presence of substrate in the binding pocket, and suggest that it both exerts a subtle conformational bias towards inwards-facing protein conformations and enables proton translocation by perturbing the pKa value of a key glutamate side chain via a salt-bridge swap mechanism. These results are corroborated by a series of new ligand-bound cryo-EM structures, where we showed using further MD simulations that the ability to engage a conserved arginine residue that triggers the salt-bridge swap distinguishes between good and bad substrates, not substrate affinity to one protein conformation alone.</p> <br> <p>We lastly benchmarked docking, absolute binding free energy (ABFE) calculations and machine learning models trained using ligand-only features on a retrospective dataset of 184 transport inhibition constants at PepT2. We show how the prediction of accurate protein-ligand binding poses constitutes a crucial limitation for the structure-based methods, and that the ligand-based machine learning models, while giving much better correlations to experimental data, fall short in their ability to capture pharmaceutically important activity cliffs. We have also started work on using these methods prospectively to design prodrugs for transport by POTs, though our efforts thus far have been hampered by solubility problems.</p> <br> <p>In summary, this thesis covers ground from the development of novel computational tools to their application to the POT transport mechanism, and reaching further towards <em>in-silico</em> design of prodrugs for increased bioavailability. It therefore advances the fields of enhanced sampling of conformational changes and membrane transporter biology, with the potential of benefitting their applications to pharmacology in the future.</p> |
| spellingShingle | Computational chemistry Biochemistry Biophysics Lichtinger, SM Molecular mechanisms of drug transport through peptide transporters |
| title | Molecular mechanisms of drug transport through peptide transporters |
| title_full | Molecular mechanisms of drug transport through peptide transporters |
| title_fullStr | Molecular mechanisms of drug transport through peptide transporters |
| title_full_unstemmed | Molecular mechanisms of drug transport through peptide transporters |
| title_short | Molecular mechanisms of drug transport through peptide transporters |
| title_sort | molecular mechanisms of drug transport through peptide transporters |
| topic | Computational chemistry Biochemistry Biophysics |
| work_keys_str_mv | AT lichtingersm molecularmechanismsofdrugtransportthroughpeptidetransporters |