The transforming growth factor-β (TGF-β) signaling pathway regulates a multitude of key processes including cellular growth, differentiation, apoptosis, motility, invasion, extracellular matrix production, angiogenesis, immune responses. TGF-β is a multifunctional cytokine that acts in a cell- and context-dependent manner as a tumor promoter or tumor suppressor. As a tumor promoter, the TGF-β pathway enhances cell proliferation, migratory invasion, metastatic spread within the tumor microenvironment and suppresses immunosurveillance. Collectively, the pleiotropic nature of TGF-β signaling contributes to drug resistance, tumor escape and undermines clinical response to therapy. Based upon a wealth of preclinical studies, the TGF-β pathway has been pharmacologically targeted using small molecule inhibitors, TGF-β-directed chimeric monoclonal antibodies, ligand traps, antisense oligonucleotides and vaccines that have been now evaluated in clinical trials [1]. Additionally, Repsox-a TGF-B1 inhibitor has been found, in unbiased screen, to be mediating editing enhancement in primary human CD4+ T cells [2]. Thus, novel inhibitors of ALK5 holds promise in therapies.
Recent advances in Artificial Intelligence in Drug Discovery holds promise in terms of rapidly predicting novel molecules against a given target [3,4,5].
We used AI powered Drug Discovery platform DrugDiscoverer to predict novel chemical structures against ALK5. The platform takes an AI model or data as input and trains the machine learning algorithm to identify novel structures using the trained model (Fig. 1). The platform has already built in AI models against a number of targets as well as phenotypic models. Such models can be accessed by inputting the SMILES of compounds and the platform gives predicted activity against the target. As an example, repsox is known to be a specific binder of TgfB or ALK5 having inhibitory affinity as 4nM (IC50 value)[6] . We inserted repsox SMILES (CC1=NC(=CC=C1)C2=C(C=NN2)C3=NC4=C(C=C3)N=CC=C4) [7] as input to ALK5 inhibiton model. The resultant IC50 value which was obtained was 13.67nM suggesting a close prediction to real value (Fig.2A,B,C).
Fig.1
Fig.2A
Fig.2B
Fig.2C
Such model was used in our AI molecule generator module (Fig. 3). The generator used the built in AI models against ALK5 inhibiton and gave new molecules within 15 hours .
Fig.3
Approximately 6000 molecules were predicted and total 17 molecules had a predicted affinity of less than 30nM (IC50 value). These molecules were checked in the pubchem database and no molecule had any mapping to a CID, implying such molecules were not in pubchem database and hence novelty.
The next steps are to synthesize these molecules and check for these in assays. Such experiments are going on in our partner labs.
To check the platform visit: http://drugdiscoverer.datadiscover.co.in/
[1] Kim, BG., Malek, E., Choi, S.H. et al. Novel therapies emerging in oncology to target the TGF-β pathway. J Hematol Oncol 14, 55 (2021). https://doi.org/10.1186/s13045-021-01053-x
[2] Mishra T, Bhardwaj V, Ahuja N, Gadgil P, Ramdas P, Shukla S, Chande A. Improved loss-of-function CRISPR-Cas9 genome editing in human cells concomitant with inhibition of TGF-β signaling. Mol Ther Nucleic Acids. 2022 Mar 8;28:202-218. doi: 10.1016/j.omtn.2022.03.003. PMID: 35402072; PMCID: PMC8961078.
[3] Zhu H. Big Data and Artificial Intelligence Modeling for Drug Discovery. Annu Rev Pharmacol Toxicol. 2020 Jan 6;60:573-589. doi: 10.1146/annurev-pharmtox-010919-023324. Epub 2019 Sep 13. PMID: 31518513; PMCID: PMC7010403.
[4] Segler, Marwin HS, et al. "Generating focused molecule libraries for drug discovery with recurrent neural networks." ACS central science 4.1 (2018): 120-131.
[5] Popova M, Isayev O, Tropsha A. Deep reinforcement learning for de novo drug design. Sci Adv. 2018 Jul 25;4(7):eaap7885. doi: 10.1126/sciadv.aap7885. PMID: 30050984; PMCID: PMC6059760.
[7]https://pubchem.ncbi.nlm.nih.gov/compound/RepSox#section=InChI-Key
Copyright © 2020 Data Discover - All Rights Reserved.