GeNNius: an ultrafast drug-target interaction inference method based on graph neural networks.

MOTIVATION: Drug-target interaction (DTI) prediction is a relevant but challenging task in the drug repurposing field. In-silico approaches have drawn particular attention as they can reduce associated costs and time commitment of traditional methodologies. Yet, current state-of-the-art methods present several limitations: existing DTI prediction approaches are computationally expensive, thereby hindering the ability to use large networks and exploit available datasets and, the generalization to unseen datasets of DTI prediction methods remains unexplored, which could potentially improve the development processes of DTI inferring approaches in terms of accuracy and robustness. RESULTS: In this work, we introduce GeNNius (Graph Embedding Neural Network Interaction Uncovering System), a Graph Neural Network (GNN)-based method that outperforms state-of-the-art models in terms of both accuracy and time efficiency across a variety of datasets. We also demonstrated its prediction power to uncover new interactions by evaluating not previously known DTIs for each dataset. We further assessed the generalization capability of GeNNius by training and testing it on different datasets, showing that this framework can potentially improve the DTI prediction task by training on large datasets and testing on smaller ones. Finally, we investigated qualitatively the embeddings generated by GeNNius, revealing that the GNN encoder maintains biological information after the graph convolutions while diffusing this information through nodes, eventually distinguishing protein families in the node embedding space. AVAILABILITY AND IMPLEMENTATION: GeNNius code is available at https://github.com/ubioinformat/GeNNius.

Signature-driven repurposing of Midostaurin for combination with MEK1/2 and KRASG12C inhibitors in lung cancer.

Drug combinations are key to circumvent resistance mechanisms compromising response to single anti-cancer targeted therapies. The implementation of combinatorial approaches involving MEK1/2 or KRASG12C inhibitors in the context of KRAS-mutated lung cancers focuses fundamentally on targeting KRAS proximal activators or effectors. However, the antitumor effect is highly determined by compensatory mechanisms arising in defined cell types or tumor subgroups. A potential strategy to find drug combinations targeting a larger fraction of KRAS-mutated lung cancers may capitalize on the common, distal gene expression output elicited by oncogenic KRAS. By integrating a signature-driven drug repurposing approach with a pairwise pharmacological screen, here we show synergistic drug combinations consisting of multi-tyrosine kinase PKC inhibitors together with MEK1/2 or KRASG12C inhibitors. Such combinations elicit a cytotoxic response in both in vitro and in vivo models, which in part involves inhibition of the PKC inhibitor target AURKB. Proteome profiling links dysregulation of MYC expression to the effect of both PKC inhibitor-based drug combinations. Furthermore, MYC overexpression appears as a resistance mechanism to MEK1/2 and KRASG12C inhibitors. Our study provides a rational framework for selecting drugs entering combinatorial strategies and unveils MEK1/2- and KRASG12C-based therapies for lung cancer.

The phospholipid transporter PITPNC1 links KRAS to MYC to prevent autophagy in lung and pancreatic cancer.

BACKGROUND: The discovery of functionally relevant KRAS effectors in lung and pancreatic ductal adenocarcinoma (LUAD and PDAC) may yield novel molecular targets or mechanisms amenable to inhibition strategies. Phospholipids availability has been appreciated as a mechanism to modulate KRAS oncogenic potential. Thus, phospholipid transporters may play a functional role in KRAS-driven oncogenesis. Here, we identified and systematically studied the phospholipid transporter PITPNC1 and its controlled network in LUAD and PDAC. METHODS: Genetic modulation of KRAS expression as well as pharmacological inhibition of canonical effectors was completed. PITPNC1 genetic depletion was performed in in vitro and in vivo LUAD and PDAC models. PITPNC1-deficient cells were RNA sequenced, and Gene Ontology and enrichment analyses were applied to the output data. Protein-based biochemical and subcellular localization assays were run to investigate PITPNC1-regulated pathways. A drug repurposing approach was used to predict surrogate PITPNC1 inhibitors that were tested in combination with KRASG12C inhibitors in 2D, 3D, and in vivo models. RESULTS: PITPNC1 was increased in human LUAD and PDAC, and associated with poor patients' survival. PITPNC1 was regulated by KRAS through MEK1/2 and JNK1/2. Functional experiments showed PITPNC1 requirement for cell proliferation, cell cycle progression and tumour growth. Furthermore, PITPNC1 overexpression enhanced lung colonization and liver metastasis. PITPNC1 regulated a transcriptional signature which highly overlapped with that of KRAS, and controlled mTOR localization via enhanced MYC protein stability to prevent autophagy. JAK2 inhibitors were predicted as putative PITPNC1 inhibitors with antiproliferative effect and their combination with KRASG12C inhibitors elicited a substantial anti-tumour effect in LUAD and PDAC. CONCLUSIONS: Our data highlight the functional and clinical relevance of PITPNC1 in LUAD and PDAC. Moreover, PITPNC1 constitutes a new mechanism linking KRAS to MYC, and controls a druggable transcriptional network for combinatorial treatments.

Criteria for preclinical models of cholangiocarcinoma: scientific and medical relevance.

Cholangiocarcinoma (CCA) is a rare malignancy that develops at any point along the biliary tree. CCA has a poor prognosis, its clinical management remains challenging, and effective treatments are lacking. Therefore, preclinical research is of pivotal importance and necessary to acquire a deeper understanding of CCA and improve therapeutic outcomes. Preclinical research involves developing and managing complementary experimental models, from in vitro assays using primary cells or cell lines cultured in 2D or 3D to in vivo models with engrafted material, chemically induced CCA or genetically engineered models. All are valuable tools with well-defined advantages and limitations. The choice of a preclinical model is guided by the question(s) to be addressed; ideally, results should be recapitulated in independent approaches. In this Consensus Statement, a task force of 45 experts in CCA molecular and cellular biology and clinicians, including pathologists, from ten countries provides recommendations on the minimal criteria for preclinical models to provide a uniform approach. These recommendations are based on two rounds of questionnaires completed by 35 (first round) and 45 (second round) experts to reach a consensus with 13 statements. An agreement was defined when at least 90% of the participants voting anonymously agreed with a statement. The ultimate goal was to transfer basic laboratory research to the clinics through increased disease understanding and to develop clinical biomarkers and innovative therapies for patients with CCA.

LAMC2 Regulates Key Transcriptional and Targetable Effectors to Support Pancreatic Cancer Growth.

PURPOSE: The identification of pancreatic ductal adenocarcinoma (PDAC) dysregulated genes may unveil novel molecular targets entering inhibitory strategies. Laminins are emerging as potential targets in PDAC given their role as diagnostic and prognostic markers. Here, we investigated the cellular, functional, and clinical relevance of LAMC2 and its regulated network, with the ultimate goal of identifying potential therapies. EXPERIMENTAL DESIGN: LAMC2 expression was analyzed in PDAC tissues, a panel of human and mouse cell lines, and a genetically engineered mouse model. Genetic perturbation in 2D, 3D, and in vivo allograft and xenograft models was done. Expression profiling of a LAMC2 network was performed by RNA-sequencing, and publicly available gene expression datasets from experimental and clinical studies examined to query its human relevance. Dual inhibition of pharmacologically targetable LAMC2-regulated effectors was investigated. RESULTS: LAMC2 was consistently upregulated in human and mouse experimental models as well as in human PDAC specimens, and associated with tumor grade and survival. LAMC2 inhibition impaired cell cycle, induced apoptosis, and sensitized PDAC to MEK1/2 inhibitors (MEK1/2i). A LAMC2-regulated network was featured in PDAC, including both classical and quasi-mesenchymal subtypes, and contained downstream effectors transcriptionally shared by the KRAS signaling pathway. LAMC2 regulated a functional FOSL1-AXL axis via AKT phosphorylation. Furthermore, genetic LAMC2 or pharmacological AXL inhibition elicited a synergistic antiproliferative effect in combination with MEK1/2is that was consistent across 2D and 3D human and mouse PDAC models, including primary patient-derived organoids. CONCLUSIONS: LAMC2 is a molecular target in PDAC that regulates a transcriptional network that unveils a dual drug combination for cancer treatment.

New molecular mechanisms in cholangiocarcinoma: signals triggering interleukin-6 production in tumor cells and KRAS co-opted epigenetic mediators driving metabolic reprogramming.

BACKGROUND: Cholangiocarcinoma (CCA) is still a deadly tumour. Histological and molecular aspects of thioacetamide (TAA)-induced intrahepatic CCA (iCCA) in rats mimic those of human iCCA. Carcinogenic changes and therapeutic vulnerabilities in CCA may be captured by molecular investigations in bile, where we performed bile proteomic and metabolomic analyses that help discovery yet unknown pathways relevant to human iCCA. METHODS: Cholangiocarcinogenesis was induced in rats (TAA) and mice (JnkΔhepa + CCl4 + DEN model). We performed proteomic and metabolomic analyses in bile from control and CCA-bearing rats. Differential expression was validated in rat and human CCAs. Mechanisms were addressed in human CCA cells, including Huh28-KRASG12D cells. Cell signaling, growth, gene regulation and [U-13C]-D-glucose-serine fluxomics analyses were performed. In vivo studies were performed in the clinically-relevant iCCA mouse model. RESULTS: Pathways related to inflammation, oxidative stress and glucose metabolism were identified by proteomic analysis. Oxidative stress and high amounts of the oncogenesis-supporting amino acids serine and glycine were discovered by metabolomic studies. Most relevant hits were confirmed in rat and human CCAs (TCGA). Activation of interleukin-6 (IL6) and epidermal growth factor receptor (EGFR) pathways, and key genes in cancer-related glucose metabolic reprogramming, were validated in TAA-CCAs. In TAA-CCAs, G9a, an epigenetic pro-tumorigenic writer, was also increased. We show that EGFR signaling and mutant KRASG12D can both activate IL6 production in CCA cells. Furthermore, phosphoglycerate dehydrogenase (PHGDH), the rate-limiting enzyme in serine-glycine pathway, was upregulated in human iCCA correlating with G9a expression. In a G9a activity-dependent manner, KRASG12D promoted PHGDH expression, glucose flow towards serine synthesis, and increased CCA cell viability. KRASG12D CAA cells were more sensitive to PHGDH and G9a inhibition than controls. In mouse iCCA, G9a pharmacological targeting reduced PHGDH expression. CONCLUSIONS: In CCA, we identified new pro-tumorigenic mechanisms: Activation of EGFR signaling or KRAS mutation drives IL6 expression in tumour cells; Glucose metabolism reprogramming in iCCA includes activation of the serine-glycine pathway; Mutant KRAS drives PHGDH expression in a G9a-dependent manner; PHGDH and G9a emerge as therapeutic targets in iCCA.

PDAC as an Immune Evasive Disease: Can 3D Model Systems Aid to Tackle This Clinical Problem?

Pancreatic ductal adenocarcinoma (PDAC) is an aggressive cancer with a high mortality rate. The presence of a dense desmoplastic stroma rich in fibroblasts, extracellular matrix, and immune cells plays a critical role in disease progression, therapy response and is a distinguishing feature of PDAC. PDAC is currently treated with a combination of surgery, chemotherapy and radiation therapy in selected cases which results in long-term survival only in a small percentage of patients. Cancer therapies that incorporate immunotherapy-based techniques have become increasingly common in recent years. While such a strategy has been shown to be effective for immunogenic, "hot" tumors like melanoma and lung cancer, thus far PDAC patients display poor responses to this therapeutic approach. Various factors, such as low tumor mutational burden, increased infiltration of immunosuppressive cells, like MDSCs and Treg cells promote tolerance and immune deviation, further aggravating adaptive immunity in PDAC. In this review we will elaborate on the ability of PDAC tumors to evade immune detection. We will also discuss various 3D model system that can be used as a platform in preclinical research to investigate rational combinations of immunotherapy with chemotherapy or targeted therapy, to prime the immune microenvironment to enhance antitumor activity.

FOSL1 promotes cholangiocarcinoma via transcriptional effectors that could be therapeutically targeted.

BACKGROUND & AIMS: Cholangiocarcinoma (CCA) is a neoplasia of the biliary tract driven by genetic, epigenetic and transcriptional mechanisms. Herein, we investigated the role of the transcription factor FOSL1, as well as its downstream transcriptional effectors, in the development and progression of CCA. METHODS: FOSL1 was investigated in human CCA clinical samples. Genetic inhibition of FOSL1 in human and mouse CCA cell lines was performed in in vitro and in vivo models using constitutive and inducible short-hairpin RNAs. Conditional FOSL1 ablation was done using a genetically engineered mouse (GEM) model of CCA (mutant KRAS and Trp53 knockout). Follow-up RNA and chromatin immunoprecipitation (ChIP) sequencing analyses were carried out and downstream targets were validated using genetic and pharmacological inhibition. RESULTS: An inter-species analysis of FOSL1 in CCA was conducted. First, FOSL1 was found to be highly upregulated in human and mouse CCA, and associated with poor patient survival. Pharmacological inhibition of different signalling pathways in CCA cells converged on the regulation of FOSL1 expression. Functional experiments showed that FOSL1 is required for cell proliferation and cell cycle progression in vitro, and for tumour growth and tumour maintenance in both orthotopic and subcutaneous xenograft models. Likewise, FOSL1 genetic abrogation in a GEM model of CCA extended mouse survival by decreasing the oncogenic potential of transformed cholangiocytes. RNA and ChIP sequencing studies identified direct and indirect transcriptional effectors such as HMGCS1 and AURKA, whose genetic and pharmacological inhibition phenocopied FOSL1 loss. CONCLUSIONS: Our data illustrate the functional and clinical relevance of FOSL1 in CCA and unveil potential targets amenable to pharmacological inhibition that could enable the implementation of novel therapeutic strategies. LAY SUMMARY: Understanding the molecular mechanisms involved in cholangiocarcinoma (bile duct cancer) development and progression stands as a critical step for the development of novel therapies. Through an inter-species approach, this study provides evidence of the clinical and functional role of the transcription factor FOSL1 in cholangiocarcinoma. Moreover, we report that downstream effectors of FOSL1 are susceptible to pharmacological inhibition, thus providing new opportunities for therapeutic intervention.

p65BTK is a novel potential actionable target in KRAS-mutated/EGFR-wild type lung adenocarcinoma.

BACKGROUND: Lung cancer is still the main cause of cancer death worldwide despite the availability of targeted therapies and immune-checkpoint inhibitors combined with chemotherapy. Cancer cell heterogeneity and primary or acquired resistance mechanisms cause the elusive behaviour of this cancer and new biomarkers and active drugs are urgently needed to overcome these limitations. p65BTK, a novel isoform of the Bruton Tyrosine Kinase may represent a new actionable target in non-small cell lung cancer (NSCLC). METHODS: p65BTK expression was evaluated by immunohistochemistry in 382 NSCLC patients with complete clinico-pathological records including smoking habit, ALK and EGFR status, and in metastatic lymph nodes of 30 NSCLC patients. NSCLC cell lines mutated for p53 and/or a component of the RAS/MAPK pathway and primary lung cancer-derived cells from Kras/Trp53 null mice were used as a preclinical model. The effects of p65BTK inhibition by BTK Tyrosine Kinase Inhibitors (TKIs) (Ibrutinib, AVL-292, RN486) and first-generation EGFR-TKIs (Gefitinib, Erlotinib) on cell viability were evaluated by MTT. The effects of BTK-TKIs on cell growth and clonogenicity were assessed by crystal violet and colony assays, respectively. Cell toxicity assays were performed to study the effect of the combination of non-toxic concentrations of BTK-TKIs with EGFR-TKIs and standard-of-care (SOC) chemotherapy (Cisplatin, Gemcitabine, Pemetrexed). RESULTS: p65BTK was significantly over-expressed in EGFR-wild type (wt) adenocarcinomas (AdC) from non-smoker patients and its expression was also preserved at the metastatic site. p65BTK was also over-expressed in cell lines mutated for KRAS or for a component of the RAS/MAPK pathway and in tumors from Kras/Trp53 null mice. BTK-TKIs were more effective than EGFR-TKIs in decreasing cancer cell viability and significantly impaired cell proliferation and clonogenicity. Moreover, non-toxic doses of BTK-TKIs re-sensitized drug-resistant NSCLC cell lines to both target- and SOC therapy, independently from EGFR/KRAS status. CONCLUSIONS: p65BTK results as an emerging actionable target in non-smoking EGFR-wt AdC, also at advanced stages of disease. Notably, these patients are not eligible for EGFR-TKIs-based therapy due to a lack of EGFR mutation. The combination of BTK-TKIs with EGFR-TKIs is cytotoxic for EGFR-wt/KRAS-mutant/p53-null tumors and BTK-TKIs re-sensitizes drug-resistant NSCLC to SOC chemotherapy. Therefore, our data suggest that adding BTK-TKIs to SOC chemotherapy and EGFR-targeted therapy may open new avenues for clinical trials in currently untreatable NSCLC.

An integrative approach unveils FOSL1 as an oncogene vulnerability in KRAS-driven lung and pancreatic cancer.

KRAS mutated tumours represent a large fraction of human cancers, but the vast majority remains refractory to current clinical therapies. Thus, a deeper understanding of the molecular mechanisms triggered by KRAS oncogene may yield alternative therapeutic strategies. Here we report the identification of a common transcriptional signature across mutant KRAS cancers of distinct tissue origin that includes the transcription factor FOSL1. High FOSL1 expression identifies mutant KRAS lung and pancreatic cancer patients with the worst survival outcome. Furthermore, FOSL1 genetic inhibition is detrimental to both KRAS-driven tumour types. Mechanistically, FOSL1 links the KRAS oncogene to components of the mitotic machinery, a pathway previously postulated to function orthogonally to oncogenic KRAS. FOSL1 targets include AURKA, whose inhibition impairs viability of mutant KRAS cells. Lastly, combination of AURKA and MEK inhibitors induces a deleterious effect on mutant KRAS cells. Our findings unveil KRAS downstream effectors that provide opportunities to treat KRAS-driven cancers.