TCM network pharmacology: A new trend towards combining computational, experimental and clinical approaches.

Intrinsic and acquired drug resistance is a major challenge in cancer therapy. Synergistic drug combinations could help to overcome drug resistance. However, the number of possible drug combinations is enormous, and it is infeasible to experimentally screen all drug combinations with limited resources. Therefore, computational models to predict and prioritize effective drug combinations are important for combination therapy discovery. Compared with existing models, we propose a novel deep learning model, AuDNNsynergy, to predict the synergy of pairwise drug combinations by integrating multiomics data. Specifically, three autoencoders are trained using the gene expression, copy number, and genetic mutation data of tumor samples from The Cancer Genome Atlas (TCGA). Then the gene expression, copy number, and mutation of individual cancer cell lines are coded using the three trained autoencoders. The physicochemical features of individual drugs and the encoded omics data of individual cancer cell lines are used as the input features of a deep neural network that predicts the synergy score of given pairwise drug combinations against the specific cancer cell lines. The comparison results showed the proposed AuDNNsynergy model outperforms, specifically in terms of rank correlation metric, four state-of-the-art approaches, namely, DeepSynergy, Gradient Boosting Machines, Random Forests, and Elastic Nets.

Synergistic Drug Combination Prediction by Integrating Multiomics Data in Deep Learning Models.

Drug resistance is still a major challenge in cancer therapy. Drug combination is expected to overcome drug resistance. However, the number of possible drug combinations is enormous, and thus it is infeasible to experimentally screen all effective drug combinations considering the limited resources. Therefore, computational models to predict and prioritize effective drug combinations is important for combinatory therapy discovery in cancer. In this study, we proposed a novel deep learning model, AuDNNsynergy, to prediction drug combinations by integrating multi-omics data and chemical structure data. In specific, three autoencoders were trained using the gene expression, copy number and genetic mutation data of all tumor samples from The Cancer Genome Atlas. Then the physicochemical properties of drugs combined with the output of the three autoencoders, characterizing the individual cancer cell-lines, were used as the input of a deep neural network that predicts the synergy value of given pair-wise drug combinations against the specific cancer cell-lines. The comparison results showed the proposed AuDNNsynergy model outperforms four state-of-art approaches, namely DeepSynergy, Gradient Boosting Machines, Random Forests, and Elastic Nets. Moreover, we conducted the interpretation analysis of the deep learning model to investigate potential vital genetic predictors and the underlying mechanism of synergistic drug combinations on specific cancer cell-lines.

Synergistic Drug Combination Prediction by Integrating Multi-omics Data in Deep Learning Models

Objective: Hepatocellular carcinoma (HCC) is frequently diagnosed in patients with late-stage disease who are ineligible for curative surgical therapies. The majority of patients become resistant to sorafenib, the only approved first-line therapy for advanced cancer, underscoring the need for newer, more effective drugs. The purpose of this study is to expedite identification of novel drugs against sorafenib resistant (SR)-HCC. Methods: We employed a transcriptomics-based drug repurposing method termed connectivity mapping using gene signatures from in vitro-derived SR Huh7 HCC cells. For proof of concept validation, we focused on drugs that were FDA-approved or under clinical investigation and prioritized two anti-neoplastic agents (dasatinib and fostamatinib) with targets associated with HCC. We also prospectively validated predicted gene expression changes in drug-treated SR Huh7 cells as well as identified and validated the targets of Fostamatinib in HCC. Results: Dasatinib specifically reduced the viability of SR-HCC cells that correlated with up-regulated activity of SRC family kinases, its targets, in our SR-HCC model. However, fostamatinib was able to inhibit both parental and SR HCC cells in vitro and in xenograft models. Ingenuity pathway analysis of fostamatinib gene expression signature from LINCS predicted JAK/STAT, PI3K/AKT, ERK/MAPK pathways as potential targets of fostamatinib that were validated by Western blot analysis. Fostamatinib treatment reversed the expression of genes that were deregulated in SR HCC. Conclusion: We provide proof of concept evidence for the validity of this drug repurposing approach for SR-HCC with implications for personalized medicine.

/pdf/transcriptomics-based-drug-repurposing-approach-identifies-43tkkrohfi.pdf

Transcriptomics-Based Drug Repurposing Approach Identifies Novel Drugs against Sorafenib-Resistant Hepatocellular Carcinoma

Tumor recurrence occurs in more than 70% of ovarian cancer patients, and the majority eventually becomes refractory to treatments. Ovarian Cancer Stem Cells (OCSCs) are believed to be responsible for the tumor relapse and drug resistance. Therefore, eliminating ovarian CSCs is important to improve the prognosis of ovarian cancer patients. However, there is a lack of effective drugs to eliminate OCSCs because the core signaling pathways regulating OCSCs remain unclear. Also it is often hard for biologists to identify a few testable targets and infer driver signaling pathways regulating CSCs from a large number of differentially expression genes in an unbiased manner. In this study, we propose a straightforward and integrative analysis to identify potential core signaling pathways of OCSCs by integrating transcriptome data of OCSCs isolated based on two distinctive markers, ALDH and side population, with regulatory network (Transcription Factor (TF) and Target Interactome) and signaling pathways. We first identify the common activated TFs in two OCSC populations integrating the gene expression and TF-target Interactome; and then uncover up-stream signaling cascades regulating the activated TFs. In specific, 22 activated TFs are identified. Through literature search validation, 15 of them have been reported in association with cancer stem cells. Additionally, 10 TFs are found in the KEGG signaling pathways, and their up-stream signaling cascades are extracted, which also provide potential treatment targets. Moreover, 40 FDA approved drugs are identified to target on the up-stream signaling cascades, and 15 of them have been reported in literatures in cancer stem cell treatment. In conclusion, the proposed approach can uncover the activated up-stream signaling, activated TFs and up-regulated target genes that constitute the potential core signaling pathways of ovarian CSC. Also drugs and drug combinations targeting on the core signaling pathways might be able to eliminate OCSCs. The proposed approach can also be applied for identifying potential activated signaling pathways of other types of cancers.

/pdf/core-signaling-pathways-in-ovarian-cancer-stem-cell-revealed-2pr6ggmxx2.pdf

Core signaling pathways in ovarian cancer stem cell revealed by integrative analysis of multi-marker genomics data.

Due to advances in next generation sequencing technologies and corresponding reductions in cost, it is now attainable to investigate genome-wide gene expression and variants at a patient-level, so as to better understand and anticipate heterogeneous responses to therapy. Consequently, it is feasible to inform personalized drug treatment decisions using personal genomics data. However, these efforts are limited due to a lack of reliable computational approaches for predicting effective drugs for individual patients. The reverse gene set enrichment analysis (i.e., connectivity mapping) approach and its variants have been widely and successfully used for drug prediction. However, the performance of these methods is limited by undefined mechanism of action (MoA) of drugs and reliance on cohorts of patients rather than personalized predictions for individual patients. In this study, we have developed and evaluated a computational approach, known as Mechanism and Drug Miner (MD-Miner), using a network-based computational approach to predict effective drugs and reveal potential drug mechanisms of action at the level of signaling pathways. Specifically, the patient-specific signaling network is constructed by integrating known disease associated genes with patient-derived gene expression profiles. In parallel, a drug mechanism of action network is constructed by integrating drug targets and z-score profiles of drug-induced gene expression (pre vs. post-drug treatment). Potentially effective candidate drugs are prioritized according to the number of common genes between the patient-specific dysfunctional signaling network and drug MoA network. We evaluated the MD-Miner method on the PC-3 prostate cancer cell line, and showed that it significantly improved the success rate of discovering effective drugs compared with the random selection, and could provide insight into potential mechanisms of action. This work provides a signaling network-based drug repositioning approach. Compared with the reverse gene signature based drug repositioning approaches, the proposed method can provide clues of mechanism of action in terms of signaling transduction networks.

https://link.springer.com/content/pdf/10.1186/s12918-017-0462-9.pdf

MD-Miner: a network-based approach for personalized drug repositioning.

Mutations in TDP-43, a RNA-binding protein with multiple functions in RNA metabolism, cause amyotrophic lateral sclerosis (ALS), but it is uncertain how defects in RNA biology trigger motor neuron degeneration. TDP-43 is a major constituent of ribonucleoprotein (RNP) granules, phase separated biomolecular condensates that regulate RNA splicing, mRNA transport, and translation. ALS-associated TDP-43 mutations, most of which are found in the low complexity domain, promote aberrant liquid to solid phase transitions and impair the dynamic liquid-like properties and motility of RNP transport granules in neurons. Here, we perform a comparative analysis of ALS-linked mutations and TDP-43 variants in order to identify critical structural elements, aromatic and charged residues that are key determinants of TDP-43 RNP transport and condensate formation in neurons. We find that A315T and Q343R disease-linked mutations and substitutions of aromatic residues within the α-helical domain and LARKS, show the most severe defects in TDP-43 RNP granule transport and impair both anterograde and retrograde motility. F313L and F313-6L/Y substitutions of one or both phenylalanine residues in LARKS suggest the aromatic rings are important for TDP-43 RNP transport. Similarly, W334F/L substitutions of the tryptophan residue in the α-helical domain, impair TDP-43 RNP motility (W334L) or anterograde transport (W334F). We also show that R293A and R293K mutations, which disrupt the only RGG in the LCD, profoundly reduce long-range, directed transport and net velocity of TDP-43 RNP granules. In the disordered regions flanking the α-helical domain, we find that F283Y, F397Y or Y374F substitutions of conserved GF/G and SYS motifs, also impair anterograde and/or retrograde motility, possibly by altering hydrophobicity. Similarly, ALS-linked mutations in disordered regions distant from the α-helical domain also show anterograde transport deficits, consistent with previous findings, but these mutations are less severe than A315T and Q343R. Overall our findings demonstrate that the conserved α-helical domain, phenylalanine residues within LARKS and RGG motif are key determinants of TDP-43 RNP transport, suggesting they may mediate efficient recruitment of motors and adaptor proteins. These results offer a possible mechanism underlying ALS-linked TDP-43 defects in axonal transport and homeostasis.

/pdf/sequence-determinants-of-tdp-43-ribonucleoprotein-condensate-1qynsqg3.pdf

Sequence Determinants of TDP-43 Ribonucleoprotein Condensate Formation and Axonal Transport in Neurons

Altered RNA metabolism is a common pathogenic mechanism linked to familial and sporadic Amyotrophic lateral sclerosis (ALS). ALS is characterized by mislocalization and aggregation of TDP-43, an RNA-binding protein (RBP) with multiple roles in post-transcriptional RNA processing. Recent studies have identified genetic interactions between TDP-43 and Ataxin-2, a polyglutamine (polyQ) RBP in which intermediate length polyQ expansions confer increased ALS risk. Here, we used live-cell confocal imaging, photobleaching and translation reporter assays to study the localization, transport dynamics and mRNA regulatory functions of TDP-43/Ataxin-2 in rodent primary cortical neurons. We show that Ataxin-2 polyQ expansions aberrantly sequester TDP-43 within ribonucleoprotein (RNP) condensates, and disrupt both its motility along the axon and liquid-like properties. Our data suggest that Ataxin-2 governs motility and translation of neuronal RNP condensates and that Ataxin-2 polyQ expansions fundamentally perturb spatial localization of mRNA and suppress local translation. Overall, these results indicate Ataxin-2 polyQ expansions have detrimental effects on stability, localization, and translation of transcripts critical for axonal and cytoskeletal integrity, particularly important for motor neurons.

/pdf/ataxin-2-polyglutamine-expansions-aberrantly-sequester-tdp-g6xnnpxa.pdf

Denethi Wijegunawardana

Papers

MD-Miner: a network-based approach for personalized drug repositioning.

Sequence Determinants of TDP-43 Ribonucleoprotein Condensate Formation and Axonal Transport in Neurons

Ataxin-2 polyglutamine expansions aberrantly sequester TDP-43, drive ribonucleoprotein condensate transport dysfunction and suppress local translation