Periodic Reporting for period 3 - SPACE4AMPS (Chemical Space for Antimicrobials on a Peptide Basis)
Periodo di rendicontazione: 2024-01-01 al 2025-06-30
Multidrug resistant (MDR) bacteria represent a global public health threat against which new drugs are urgently needed, however research towards new antibiotics is almost non-existent. Solving the antibiotics crisis is critical for society because MDR bacteria are becoming one of the main causes of death in hospitals.
Antimicrobial peptides (AMPs), mostly derived from naturally occurring linear or cyclic peptides, can contribute to solving the problem, however AMPS with optimal activity and toxicity profiles are difficult to identify. The objective of SPACE4AMPS is to develop new computational tools to explore chemical space in search for novel AMPs with optimal properties, and to synthesize and test these new AMPs to identify new antibiotics for clinical development. The impact of SPACE4AMPS is not only to discover new antibiotics, but also to develop and test new methods for exploring chemical space that can make drug discovery faster and easier.
Antimicrobial peptides (AMPs), mostly derived from naturally occurring linear or cyclic peptides, can contribute to solving the problem, however AMPS with optimal activity and toxicity profiles are difficult to identify. The objective of SPACE4AMPS is to develop new computational tools to explore chemical space in search for novel AMPs with optimal properties, and to synthesize and test these new AMPs to identify new antibiotics for clinical development. The impact of SPACE4AMPS is not only to discover new antibiotics, but also to develop and test new methods for exploring chemical space that can make drug discovery faster and easier.
We completed the initial computational study aiming at classifying the natural products (NPs) chemical space and how antibiotics and in particular AMPs occupy this space. This led us to develop a new molecular fingerprints encoding the chemical structure of NPs and discover that the NP chemical space can be charted by structural types and biological origin.(1)
In view of exploiting machine learning (ML) in our approach to new antimicrobials, we have performed a discovery study for non-toxic antimicrobial peptides (AMPs). In this study, we found that recurrent neural networks (RNN) allowed to exploit public databases on bioactive peptide to design new non-toxic AMPs (2). We later applied the same approach to identify peptides active against cancer cells. In this context, we validated a genetic algorithm method which is a central computational tool for SPACE4AMP.(3) We have further validated this tool for further antimicrobial compounds and are currently exploiting this genetic algorithm to design new macrocyclic NP analogs as antimicrobials. Since SPACE4AMPS aims at NP analogs, we investigated the activity profile of the clinical NP polymyxin B, which we use as one of our reference compounds, and discovered a previously unknown effect of pH on its activity.(4) In the course of investigating bicyclic peptides, a type of peptides in which the chain form two connected loops, as antimicrobials, we accidentally discovered a highly active linear AMP of only 11 amino acids which exhibits very strong activity even when some of its amino acids have the non-natural D-chirality (5, 6). Work with these simple peptides led us to also think about the synthetic process used, solid-phase peptide synthesis, and develop an improved, non-toxic reagent for deprotection, a key step in the synthesis (7).
References:
(1) Capecchi, A.; Reymond, J.-L. Classifying Natural Products from Plants, Fungi or Bacteria Using the COCONUT Database and Machine Learning. Journal of Cheminformatics 2021, 13 (1), 82. https://doi.org/10.1186/s13321-021-00559-3(si apre in una nuova finestra).
(2) Capecchi, A.; Cai, X.; Personne, H.; Köhler, T.; Delden, C. van; Reymond, J.-L. Machine Learning Designs Non-Hemolytic Antimicrobial Peptides. Chem. Sci. 2021, 12 (26), 9221–9232. https://doi.org/10.1039/D1SC01713F(si apre in una nuova finestra).
(3) Zakharova, E.; Orsi, M.; Capecchi, A.; Reymond, J.-L. Machine Learning Guided Discovery of Non-Hemolytic Membrane Disruptive Anticancer Peptides. ChemMedChem 2022, e202200291. https://doi.org/10.1002/cmdc.202200291(si apre in una nuova finestra).
(4) Cai, X.; Javor, S.; Gan, B. H.; Köhler, T.; Reymond, J.-L. The Antibacterial Activity of Peptide Dendrimers and Polymyxin B Increases Sharply above PH 7.4. Chem. Commun. 2021, 57 (46), 5654–5657. https://doi.org/10.1039/D1CC01838H(si apre in una nuova finestra).
(5) Baeriswyl, S.; Personne, H.; Bonaventura, I. D.; Köhler, T.; Delden, C. van; Stocker, A.; Javor, S.; Reymond, J.-L. A Mixed Chirality α-Helix in a Stapled Bicyclic and a Linear Antimicrobial Peptide Revealed by X-Ray Crystallography. RSC Chem. Biol. 2021, 2, 1608–1617. https://doi.org/10.1039/D1CB00124H(si apre in una nuova finestra).
(6) Personne, H.; Paschoud, T.; Fulgencio, S.; Baeriswyl, S.; Köhler, T.; van Delden, C.; Stocker, A.; Javor, S.; Reymond, J.-L. To Fold or Not to Fold: Diastereomeric Optimization of an α-Helical Antimicrobial Peptide. J. Med. Chem. 2023, 66, 7570–7583. https://doi.org/10.1021/acs.jmedchem.3c00460(si apre in una nuova finestra).
(7) Personne, H.; Siriwardena, T. N.; Javor, S.; Reymond, J.-L. Dipropylamine for 9-Fluorenylmethyloxycarbonyl (Fmoc) Deprotection with Reduced Aspartimide Formation in Solid-Phase Peptide Synthesis. ACS Omega 2023, 8 (5), 5050–5056.
In view of exploiting machine learning (ML) in our approach to new antimicrobials, we have performed a discovery study for non-toxic antimicrobial peptides (AMPs). In this study, we found that recurrent neural networks (RNN) allowed to exploit public databases on bioactive peptide to design new non-toxic AMPs (2). We later applied the same approach to identify peptides active against cancer cells. In this context, we validated a genetic algorithm method which is a central computational tool for SPACE4AMP.(3) We have further validated this tool for further antimicrobial compounds and are currently exploiting this genetic algorithm to design new macrocyclic NP analogs as antimicrobials. Since SPACE4AMPS aims at NP analogs, we investigated the activity profile of the clinical NP polymyxin B, which we use as one of our reference compounds, and discovered a previously unknown effect of pH on its activity.(4) In the course of investigating bicyclic peptides, a type of peptides in which the chain form two connected loops, as antimicrobials, we accidentally discovered a highly active linear AMP of only 11 amino acids which exhibits very strong activity even when some of its amino acids have the non-natural D-chirality (5, 6). Work with these simple peptides led us to also think about the synthetic process used, solid-phase peptide synthesis, and develop an improved, non-toxic reagent for deprotection, a key step in the synthesis (7).
References:
(1) Capecchi, A.; Reymond, J.-L. Classifying Natural Products from Plants, Fungi or Bacteria Using the COCONUT Database and Machine Learning. Journal of Cheminformatics 2021, 13 (1), 82. https://doi.org/10.1186/s13321-021-00559-3(si apre in una nuova finestra).
(2) Capecchi, A.; Cai, X.; Personne, H.; Köhler, T.; Delden, C. van; Reymond, J.-L. Machine Learning Designs Non-Hemolytic Antimicrobial Peptides. Chem. Sci. 2021, 12 (26), 9221–9232. https://doi.org/10.1039/D1SC01713F(si apre in una nuova finestra).
(3) Zakharova, E.; Orsi, M.; Capecchi, A.; Reymond, J.-L. Machine Learning Guided Discovery of Non-Hemolytic Membrane Disruptive Anticancer Peptides. ChemMedChem 2022, e202200291. https://doi.org/10.1002/cmdc.202200291(si apre in una nuova finestra).
(4) Cai, X.; Javor, S.; Gan, B. H.; Köhler, T.; Reymond, J.-L. The Antibacterial Activity of Peptide Dendrimers and Polymyxin B Increases Sharply above PH 7.4. Chem. Commun. 2021, 57 (46), 5654–5657. https://doi.org/10.1039/D1CC01838H(si apre in una nuova finestra).
(5) Baeriswyl, S.; Personne, H.; Bonaventura, I. D.; Köhler, T.; Delden, C. van; Stocker, A.; Javor, S.; Reymond, J.-L. A Mixed Chirality α-Helix in a Stapled Bicyclic and a Linear Antimicrobial Peptide Revealed by X-Ray Crystallography. RSC Chem. Biol. 2021, 2, 1608–1617. https://doi.org/10.1039/D1CB00124H(si apre in una nuova finestra).
(6) Personne, H.; Paschoud, T.; Fulgencio, S.; Baeriswyl, S.; Köhler, T.; van Delden, C.; Stocker, A.; Javor, S.; Reymond, J.-L. To Fold or Not to Fold: Diastereomeric Optimization of an α-Helical Antimicrobial Peptide. J. Med. Chem. 2023, 66, 7570–7583. https://doi.org/10.1021/acs.jmedchem.3c00460(si apre in una nuova finestra).
(7) Personne, H.; Siriwardena, T. N.; Javor, S.; Reymond, J.-L. Dipropylamine for 9-Fluorenylmethyloxycarbonyl (Fmoc) Deprotection with Reduced Aspartimide Formation in Solid-Phase Peptide Synthesis. ACS Omega 2023, 8 (5), 5050–5056.
We have recently discovered that several residues of our short AMP can be substituted by peptoid residues, which leads to antimicrobial compounds which are entirely non-toxic but efficiently kill bacteria, however with a non-membrane disruptive mechanism similar to that of the so-called proline-rich AMPs (PrAMPs). Ongoing experiments with the PrAMP oncocin suggests that the non-membrane disruptive mechanism is compatible with conformational disorder, which is a common feature of peptoids and for PrAMPs. This hypothesis is also supported by our recent experiment with peptide dendrimers and PAMAM dendrimers pointing to intrinsically disordered bioactive conformation (8, 9). Currently we are using our genetic algorithm, as well as more recent tools such as large language models (e.g. GPT3) which seem to perform quite well (10), to guide the discovery and optimization of mixed peptide/peptoid molecules, one series being a linear sequence, and the other a cyclic peptide designed as NP analogs.
Experiments are also ongoing regarding establishing advanced biological profiling of our new antimicrobials, including profiling against multidrug-resistant bacteria, in the Galleria Mellonella test system, for toxicity with a range of human cell lines in vitro, and with biochemical assays targeting the ribosome activity, one of the most promising intracellular targets for antibiotics. This broad profiling is work intensive helps us to identify the most promising compounds. We are using these methods on several antimicrobials recently discovered in the laboratory.
(8) Cai, X.; Orsi, M.; Capecchi, A.; Köhler, T.; Delden, C. van; Javor, S.; Reymond, J.-L. An Intrinsically Disordered Antimicrobial Peptide Dendrimer from Stereorandomized Virtual Screening. Cell Rep. Phys. Sci. 2022, 3 (12). https://doi.org/10.1016/j.xcrp.2022.101161(si apre in una nuova finestra).
(9) Bonvin, E.; Reymond, J.-L. Inverse Polyamidoamine (i-PAMAM) Dendrimer Antimicrobials. Helv. Chim. Acta 2023, e202300035. https://doi.org/10.1002/hlca.202300035(si apre in una nuova finestra).
(10) Orsi, M.; Reymond, J.-L.. GPT-3 accurately predicts antimicrobial peptide activity and hemolysis. ChemRxiv 2023 https://doi.org/10.26434/chemrxiv-2023-74041(si apre in una nuova finestra)
Experiments are also ongoing regarding establishing advanced biological profiling of our new antimicrobials, including profiling against multidrug-resistant bacteria, in the Galleria Mellonella test system, for toxicity with a range of human cell lines in vitro, and with biochemical assays targeting the ribosome activity, one of the most promising intracellular targets for antibiotics. This broad profiling is work intensive helps us to identify the most promising compounds. We are using these methods on several antimicrobials recently discovered in the laboratory.
(8) Cai, X.; Orsi, M.; Capecchi, A.; Köhler, T.; Delden, C. van; Javor, S.; Reymond, J.-L. An Intrinsically Disordered Antimicrobial Peptide Dendrimer from Stereorandomized Virtual Screening. Cell Rep. Phys. Sci. 2022, 3 (12). https://doi.org/10.1016/j.xcrp.2022.101161(si apre in una nuova finestra).
(9) Bonvin, E.; Reymond, J.-L. Inverse Polyamidoamine (i-PAMAM) Dendrimer Antimicrobials. Helv. Chim. Acta 2023, e202300035. https://doi.org/10.1002/hlca.202300035(si apre in una nuova finestra).
(10) Orsi, M.; Reymond, J.-L.. GPT-3 accurately predicts antimicrobial peptide activity and hemolysis. ChemRxiv 2023 https://doi.org/10.26434/chemrxiv-2023-74041(si apre in una nuova finestra)