Computational studies of ligand binding to TRIM PHD-bromodomains
<p>Modulating the function of epigenetic reader domains such as bromodomains (BRDs) and plant homeodomains (PHD), which bind acetylated lysine and methylated lysine respectively, has the potential to alter disease-associated epigenetic states. TRIM24 and TRIM33 are members of the tripartite mo...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
2023
|
| Subjects: |
| _version_ | 1811139384772132864 |
|---|---|
| author | Lee, B |
| author2 | Duarte Gonzalez, F |
| author_facet | Duarte Gonzalez, F Lee, B |
| author_sort | Lee, B |
| collection | OXFORD |
| description | <p>Modulating the function of epigenetic reader domains such as bromodomains (BRDs) and plant homeodomains (PHD), which bind acetylated lysine and methylated lysine respectively, has the potential to alter disease-associated epigenetic states. TRIM24 and TRIM33 are members of the tripartite motif (TRIM) protein family that contain a C-terminal tandem PHD finger-bromodomain. Few ligands are available to target the TRIM BRDs, especially the TRIM33 BRD isoforms (TRIM33α/β); developing new ligands to inhibit these BRDs would be useful for probing their function and assessing their feasibility as therapeutic targets. The work described in this thesis aimed to employ computational techniques to enhance the understanding of ligand recognition by TRIM PHD-BRDs, and to identify new ligands targeting the TRIM BRDs.</p>
<p>Work commenced by investigating the interactions of TRIM PHD-BRDs with peptides bearing post-translationally modified amino acids, using classical molecular dynamics (cMD) and grand canonical Monte Carlo (GCMC) simulations (Chapter 3). The non- canonical TRIM33α BRD was found to recognise acetylated lysine 18 (K18Ac) primarily through hydrophobic contacts rather than through hydrogen bonding to N1039, a conserved asparagine in the BRD pocket. This finding guided the subsequent docking of ligands into the TRIM33α BRD (Chapter 4). The work also indicated that the TRIM PHD fingers recognises trimethylated lysine 9 (K9Me3) via a cation–π interaction with a solvent-exposed tryptophan. This tryptophan is in close proximity to the BRD, suggesting that bivalent ligands targeting both the PHD finger and BRD could be viable.</p>
<p>Chapter 4 examined models of TRIM33α and TRIM33β bound to small molecule ligands using cMD and enhanced sampling simulations. This revealed that the ligands engage the TRIM33α BRD through a salt bridge to a glutamate residue (E981) and hydrophobic interactions with the pocket. They also indicated that the binding mode of ligands to TRIM33β is more complicated than anticipated: while the proposed binding mode had the ligand interacting with both N1039 and E981, this pose was rarely observed during enhanced sampling simulations. These findings could account for the difficulties with correlating the experimental binding affinities to the computational models of TRIM33β–ligand complexes.</p>
<p>Given the challenges with the structure-based approaches, machine learning (ML) models were explored to screen compound libraries for novel TRIM BRD ligands (Chapter 5). Proteochemometric models, which employ both ligand and protein descriptors, were developed for predicting ligand bioactivity and selectivity across BRDs. These models produced an activity prediction, and a calibrated probability value for prioritising compounds for experimental validation. The models achieved sensitivity and specificity > 0.8 in cross-validation and on the test set, and were applied for virtual screening of a compound library for new TRIM BRD ligands. Experimental validation of the TRIM24 ligand predictions is now underway.</p>
<p>In summary, the work in this thesis resulted in new insights into ligand recognition by the TRIM PHD-BRDs, and generated new hypotheses regarding the challenges in designing small molecule ligands for the TRIM33 BRDs. It also provides a ML model to facilitate the search for new TRIM BRD ligands.</p> |
| first_indexed | 2024-09-25T04:05:14Z |
| format | Thesis |
| id | oxford-uuid:765852cf-a4f1-4682-9311-fdd556c709eb |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-09-25T04:05:14Z |
| publishDate | 2023 |
| record_format | dspace |
| spelling | oxford-uuid:765852cf-a4f1-4682-9311-fdd556c709eb2024-05-20T09:43:02ZComputational studies of ligand binding to TRIM PHD-bromodomainsThesishttp://purl.org/coar/resource_type/c_db06uuid:765852cf-a4f1-4682-9311-fdd556c709ebBiochemistryComputational chemistryEnglishHyrax Deposit2023Lee, BDuarte Gonzalez, FConway, S<p>Modulating the function of epigenetic reader domains such as bromodomains (BRDs) and plant homeodomains (PHD), which bind acetylated lysine and methylated lysine respectively, has the potential to alter disease-associated epigenetic states. TRIM24 and TRIM33 are members of the tripartite motif (TRIM) protein family that contain a C-terminal tandem PHD finger-bromodomain. Few ligands are available to target the TRIM BRDs, especially the TRIM33 BRD isoforms (TRIM33α/β); developing new ligands to inhibit these BRDs would be useful for probing their function and assessing their feasibility as therapeutic targets. The work described in this thesis aimed to employ computational techniques to enhance the understanding of ligand recognition by TRIM PHD-BRDs, and to identify new ligands targeting the TRIM BRDs.</p> <p>Work commenced by investigating the interactions of TRIM PHD-BRDs with peptides bearing post-translationally modified amino acids, using classical molecular dynamics (cMD) and grand canonical Monte Carlo (GCMC) simulations (Chapter 3). The non- canonical TRIM33α BRD was found to recognise acetylated lysine 18 (K18Ac) primarily through hydrophobic contacts rather than through hydrogen bonding to N1039, a conserved asparagine in the BRD pocket. This finding guided the subsequent docking of ligands into the TRIM33α BRD (Chapter 4). The work also indicated that the TRIM PHD fingers recognises trimethylated lysine 9 (K9Me3) via a cation–π interaction with a solvent-exposed tryptophan. This tryptophan is in close proximity to the BRD, suggesting that bivalent ligands targeting both the PHD finger and BRD could be viable.</p> <p>Chapter 4 examined models of TRIM33α and TRIM33β bound to small molecule ligands using cMD and enhanced sampling simulations. This revealed that the ligands engage the TRIM33α BRD through a salt bridge to a glutamate residue (E981) and hydrophobic interactions with the pocket. They also indicated that the binding mode of ligands to TRIM33β is more complicated than anticipated: while the proposed binding mode had the ligand interacting with both N1039 and E981, this pose was rarely observed during enhanced sampling simulations. These findings could account for the difficulties with correlating the experimental binding affinities to the computational models of TRIM33β–ligand complexes.</p> <p>Given the challenges with the structure-based approaches, machine learning (ML) models were explored to screen compound libraries for novel TRIM BRD ligands (Chapter 5). Proteochemometric models, which employ both ligand and protein descriptors, were developed for predicting ligand bioactivity and selectivity across BRDs. These models produced an activity prediction, and a calibrated probability value for prioritising compounds for experimental validation. The models achieved sensitivity and specificity > 0.8 in cross-validation and on the test set, and were applied for virtual screening of a compound library for new TRIM BRD ligands. Experimental validation of the TRIM24 ligand predictions is now underway.</p> <p>In summary, the work in this thesis resulted in new insights into ligand recognition by the TRIM PHD-BRDs, and generated new hypotheses regarding the challenges in designing small molecule ligands for the TRIM33 BRDs. It also provides a ML model to facilitate the search for new TRIM BRD ligands.</p> |
| spellingShingle | Biochemistry Computational chemistry Lee, B Computational studies of ligand binding to TRIM PHD-bromodomains |
| title | Computational studies of ligand binding to TRIM PHD-bromodomains |
| title_full | Computational studies of ligand binding to TRIM PHD-bromodomains |
| title_fullStr | Computational studies of ligand binding to TRIM PHD-bromodomains |
| title_full_unstemmed | Computational studies of ligand binding to TRIM PHD-bromodomains |
| title_short | Computational studies of ligand binding to TRIM PHD-bromodomains |
| title_sort | computational studies of ligand binding to trim phd bromodomains |
| topic | Biochemistry Computational chemistry |
| work_keys_str_mv | AT leeb computationalstudiesofligandbindingtotrimphdbromodomains |