Predicting selective drug targets in cancer through metabolic networks

The interest in studying metabolic alterations in cancer and their potential role as novel targets for therapy has been rejuvenated in recent years. Here, we report the development of the first genome-scale network model of cancer metabolism, validated by correctly identifying genes essential for cellular proliferation in cancer cell lines. The model predicts 52 cytostatic drug targets, of which 40% are targeted by known, approved or experimental anticancer drugs, and the rest are new. It further predicts combinations of synthetic lethal drug targets, whose synergy is validated using available drug efficacy and gene expression measurements across the NCI-60 cancer cell line collection. Finally, potential selective treatments for specific cancers that depend on cancer type-specific downregulation of gene expression and somatic mutations are compiled.

Ciz1 is a nuclear protein that associates with cyclin A – cyclin dependent kinase 2 (CDK2) and facilitates the initiation of DNA replication. Ciz1 overexpression has been linked to common cancer types, including breast, colon, prostate, lung, and liver cancers. This suggests that identification of mechanisms that regulate Ciz1 levels may represent potential drug targets in cancer. This work identifies that CDK2 and DDK activity are required to maintain Ciz1 levels. Chemical or genetic inhibition of CDK2 or DDK (Cdc7-Dbf4) activity in murine fibroblasts reduced Ciz1 levels. Further analysis demonstrated that CDK and DDK activity promotes Ciz1 accumulation in G1 phase by reducing ubiquitin proteasome system (UPS) mediated degradation. Furthermore, Ciz1 levels are actively controlled by the proteasome, as inhibition of protein translation rapidly reduced Ciz1 levels, and this is reversed by proteasomal inhibition. The data suggest a model where Ciz1 is regulated by opposing kinase and UPS activities, leading to Ciz1 accumulation in response to rising kinase activity in G1 phase, and its degradation later in the cell cycle. Significantly, human prostate adenocarcinoma (PC3) and oestrogen receptor positive breast cancer (MCF7) cell lines require Ciz1 for efficient proliferation. The data demonstrate that Ciz1 levels can be reduced with CDK2/ DDK inhibitors via proteasomally mediated degradation in human cancer cell lines similarly to normal fibroblasts. In PC3 and MCF7 cell lines, repurposing small molecule CDK2 inhibitors efficiently reduce Ciz1 levels, decrease E2F mediated transcription and proliferation. The targeted depletion of Ciz1 via CDK2/ DDK inhibition and UPS mediated degradation requires a functional E3 ligase to be effective. As a first step towards identifying the regulatory E3 ligase(s), a biochemical fractionation and mass spectrometry approach revealed three putative E3 ligases: UBR5, FBXO8 and UBE2O, which require further characterisation. Taken together, this work suggests that deregulation of CDK activity or inactivation of UPS signalling may promote Ciz1 overexpression in specific cancers. Importantly, Ciz1 is required for proliferation of some cancer cell lines, suggesting that approaches, which reduce Ciz1 levels may be of clinical benefit. Therefore, the identification of the regulatory mechanisms that control Ciz1 levels, represent potential targets in Ciz1 dependent cancers.

Biotechnology is currently evolving through the era of big data, thanks to advances in the high-throughput technologies for rapid and inexpensive genome sequencing and other genome-wide studies [1]. With the daunting amount of data, it has been possible to put them together into a coherently organized biological network that provides counterintuitive insights on biological systems [2]. Among such biological networks, a genome-scale metabolic network model is expected to play an increasingly important role in the biopharmaceutical industry [3]. Before enumerating their specific strengths, it is important to note that principles underlying genome-scale metabolic network models are consistent with the holistic perspective of systems biology, the aim of which is to unveil hidden factors causing diseases and to find relevant treatment strategies [4]. Despite the importance of metabolism in a biological system, studies on diseases in relation to metabolism were far fewer in number than those performed on signaling and transcriptional regulatory networks [5]. However, metabolism, highly linked with observable phenotypes, is a biological network that is more comprehensively characterized when compared with the other two types of networks [6]. Metabolism is, therefore, amenable to large-scale mathematical modeling and simulation. It is with this motivation that the genome-scale metabolic simulation deserves more attention in drug discovery campaigns and optimization of a host strain for the production of biopharmaceuticals. Reconstruction and application of genome-scale metabolic network models have been forged as a major research strategy of systems biology. Over the last decade, genome-scale metabolic models have been built for almost all biologically important organisms across the domains of archea, bacteria and eukaryotes [3]. They range from simple micro organisms such as Escherichia coli [7] and Saccharomyces cerevisiae [8] to higher organisms including Chinese hamster ovary (CHO) cells [9,10] and a generic human cell [11,12]. It should be noted that all these organisms that have been subjected to metabolic modeling are important cellular hosts for biopharmaceutical production or medically meaningful organisms that need to be cured (e.g., specific cancer cells) or destroyed (e.g., pathogens). A recent notable development of importance in the genomescale metabolic modeling would be the newly updated human metabolic network Recon 2 [12]. Recon 2 is a result of efforts from a group of researchers, going over a vast amount of literature and biochemical data and reconciling conflicting information. Scope of the hitherto reconstructed genome-scale metabolic models manifest high expectations for their potential contributions to biopharmaceutical industry. Genome-scale metabolic network models are not just a simple pileup of biochemical reactions, but allow mathematical simulation under precisely defined conditions of constraints [13]. Once the experimentally Applications of genome-scale metabolic network models in the biopharmaceutical industry

Since the sequencing of the human genome, recent efforts in cancer drug target discovery have focused more on the identification of novel functions of known genes and the development of more appropriate tumor models. In the present study, we investigated in vitro transformed human adult mesenchymal stem cells (MSC) to identify novel candidate cancer drug targets by analyzing the transcriptional profile of known enzymes compared with non-transformed MSC. The identified enzymes were compared with published cancer gene expression data sets. Surprisingly, the majority of up-regulated enzymes are already known cancer drug targets or act within known druggable pathways. Only three enzymes (RNASEH2A, ADARB1, and PPAP2C) are potentially novel targets that are up-regulated in transformed MSC and expressed in numerous carcinomas and sarcomas. We confirmed the overexpression of RNASEH2A, PPAP2C, and ADARB1 in transformed MSC, transformed fibroblasts, and cancer cell lines MCF7, SK-LMS1, MG63, and U2OS. In functional assays, we show that small interfering RNA knockdown of RNASEH2A inhibits anchorage-independent growth but does not alter in vitro proliferation of cancer cell lines, normal MSC, or normal fibroblasts. Knockdown of PPAP2C impaired anchorage-dependent in vitro growth of cancer cell lines and impaired the in vitro growth of primary MSC but not differentiated human fibroblasts. We show that the knockdown of PPAP2C decreases cell proliferation by delaying entry into S phase of the cell cycle and is transcriptionally regulated by p53. These in vitro data validate PPAP2C and RNASEH2A as putative cancer targets and endorse this in silico approach for identifying novel candidates.

<div>Abstract<p>Since the sequencing of the human genome, recent efforts in cancer drug target discovery have focused more on the identification of novel functions of known genes and the development of more appropriate tumor models. In the present study, we investigated <i>in vitro</i> transformed human adult mesenchymal stem cells (MSC) to identify novel candidate cancer drug targets by analyzing the transcriptional profile of known enzymes compared with non-transformed MSC. The identified enzymes were compared with published cancer gene expression data sets. Surprisingly, the majority of up-regulated enzymes are already known cancer drug targets or act within known druggable pathways. Only three enzymes (RNASEH2A, ADARB1, and PPAP2C) are potentially novel targets that are up-regulated in transformed MSC and expressed in numerous carcinomas and sarcomas. We confirmed the overexpression of RNASEH2A, PPAP2C, and ADARB1 in transformed MSC, transformed fibroblasts, and cancer cell lines MCF7, SK-LMS1, MG63, and U2OS. In functional assays, we show that small interfering RNA knockdown of RNASEH2A inhibits anchorage-independent growth but does not alter <i>in vitro</i> proliferation of cancer cell lines, normal MSC, or normal fibroblasts. Knockdown of PPAP2C impaired anchorage-dependent <i>in vitro</i> growth of cancer cell lines and impaired the <i>in vitro</i> growth of primary MSC but not differentiated human fibroblasts. We show that the knockdown of PPAP2C decreases cell proliferation by delaying entry into S phase of the cell cycle and is transcriptionally regulated by p53. These <i>in vitro</i> data validate PPAP2C and RNASEH2A as putative cancer targets and endorse this <i>in silico</i> approach for identifying novel candidates. [Mol Cancer Ther 2009;8(1):249–60]</p></div>

<div>Abstract<p>Since the sequencing of the human genome, recent efforts in cancer drug target discovery have focused more on the identification of novel functions of known genes and the development of more appropriate tumor models. In the present study, we investigated <i>in vitro</i> transformed human adult mesenchymal stem cells (MSC) to identify novel candidate cancer drug targets by analyzing the transcriptional profile of known enzymes compared with non-transformed MSC. The identified enzymes were compared with published cancer gene expression data sets. Surprisingly, the majority of up-regulated enzymes are already known cancer drug targets or act within known druggable pathways. Only three enzymes (RNASEH2A, ADARB1, and PPAP2C) are potentially novel targets that are up-regulated in transformed MSC and expressed in numerous carcinomas and sarcomas. We confirmed the overexpression of RNASEH2A, PPAP2C, and ADARB1 in transformed MSC, transformed fibroblasts, and cancer cell lines MCF7, SK-LMS1, MG63, and U2OS. In functional assays, we show that small interfering RNA knockdown of RNASEH2A inhibits anchorage-independent growth but does not alter <i>in vitro</i> proliferation of cancer cell lines, normal MSC, or normal fibroblasts. Knockdown of PPAP2C impaired anchorage-dependent <i>in vitro</i> growth of cancer cell lines and impaired the <i>in vitro</i> growth of primary MSC but not differentiated human fibroblasts. We show that the knockdown of PPAP2C decreases cell proliferation by delaying entry into S phase of the cell cycle and is transcriptionally regulated by p53. These <i>in vitro</i> data validate PPAP2C and RNASEH2A as putative cancer targets and endorse this <i>in silico</i> approach for identifying novel candidates. [Mol Cancer Ther 2009;8(1):249–60]</p></div>

Calcium is an important intracellular messenger that regulates many physiological functions such as cellular differentiation, proliferation and apoptosis. Abnormal functioning of calcium signaling pathways may occur during tumorigenesis. Calcium signals are tightly controlled by calcium influx channels and many calcium channels, including those of the transient receptor potential (TRP) ion channel family have altered expression levels in specific cancer types, such as the upregulation of TRPM8 in some prostate cancers. Plasma membrane calcium channels belong to a protein class that can be effectively pharmacologically modulated. Pharmacological agents are available that modify specific TRP calcium channels and some voltage gated calcium channels, such as the inhibition of L-type calcium channels by the anti-hypertensive nifedipine. Both activators and inhibitors of calcium channels overexpressed in cancer cells are potential anti-cancer drugs. This is due to the duality of the calcium signal, whereby calcium can be both a promoter of proliferation but also an inducer of apoptosis. Therefore, calcium channels with increased expression and/or activity in cancer could potentially serve as anti-cancer therapeutic targets. To test this and to explore some of the likely issues related to calcium channels as drug targets in cancer, we have generated a Tet-Off inducible model system where the transient receptor potential vanilloid 1 (TRPV1) calcium channel is over-expressed in the MCF-7 Tet-Off breast cancer cell line. TRPV1 has been chosen in this model, due to the wide availability of selective inhibitors and activators to this channel. TRPV1 functional activity in the model system was measured by calcium influx using the Fluorescent Imaging Plate Reader (FLIPR) and the calcium sensitive probe Fluo-4. Capsaicin, a TRPV1 agonist had an EC50 of 2 × 10-7 M for increased cytosolic free calcium in the absence of doxycycline and an EC50 of 1 × 10-6 M in the presence of doxycycline (1µg/ml). In this same stable cell line, peak cytosolic responses were significantly higher in the absence of doxycycline compared to the presence of doxycycline. This suggests a tight regulation of TRPV1 expression in the Tet-Off system. Future studies will characterize the consequences of TRPV1 over-expression and TRPV1 pharmacological modulators on cellular proliferation and apoptosis pathways. This model system could be an important tool to address the feasibility of calcium channels as drug targets in cancer. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 3594.

There are 114.101 small molecule metabolites currently annotated in the Human Metabolome Database, which are highly connected amongst each other, with a few metabolites exhibiting an estimated number of more than 103 connections. Redundancy and plasticity are essential features of metabolic networks enabling cells to respond to fluctuating environments, presence of toxic molecules, or genetic perturbations like mutations. These system-level properties are inevitably linked to all aspects of biological systems ensuring cell viability by enabling processes like adaption and differentiation. To this end, the ability to interrogate molecular changes at omics level has opened new opportunities to study the cell at its different layers from the epigenome and transcriptome to its proteome and metabolome. In this thesis, I tackled the question how redundancy and plasticity shape adaptation in metabolic networks in evolutionary and disease contexts. I utilize a multi-omics approach to study comprehensively the metabolic state of a cell and its regulation at the transcriptional and proteomic level. One of the challenges with multi-omics approaches is the integration and interpretation of multi-layered data sets. To approach this challenge, I use genome scale metabolic models as a knowledge-based scaffold to overlay omics data and thereby to enable biological interpretation beyond statistical correlation. This integrative methodology has been applied to two different projects, namely the evolutionary adaptation towards a nutrient source in yeast and the metabolic adaptations following disease progression. For the latter, I also curated a current human genome-scale metabolic model and made it more suitable for flux predictions. In the yeast case study, I investigate the metabolic network adaptations enabling yeast to grow on an alternative carbon source – glycerol. I could show that network redundancy is one of the key features of fast adaptation of the yeast metabolic network to the new nutrient environment. Genomics, transcriptomics, proteomics, metabolomics and metabolic modeling together revealed a shift of the organism’s redox-balance under glycerol consumption as a driving force of adaption, which can be linked to the causal mutation in the enzyme Kgd1. On the other hand, the limitations of metabolic network adaptation also became apparent since all evolved and adapted strains exhibited metabolic trade-offs in other environmental conditions than the adaptation niche. Either an impaired diauxic shift (as in the case of the glycerol mutant) or an increased sensitivity towards osmotic stress (caused by mutations in the HOG pathway) was coupled with efficient use of glycerol. In the second project, the molecular phenotype of regressed breast cancer cells was studied to identify what differentiates these cells from healthy breast tissue and to characterize the potential source of tumor recurrence. Using a breast cancer mouse model with inducible oncogenes, transcriptomics together with an extensive set of different types of metabolomics (targeted and untargeted metabolomics, lipidomics and fluxomics) could show that regressed cancer cells, albeit their apparently normal morphology, possess a highly altered molecular phenotype with an oncogenic memory. While in cancer redundancy and plasticity enable the adaptation towards a proliferative state, in regressed cells, on the contrary, prolonged oncogenic signaling leads to a loss of metabolic network regulation and the entering of an irreversible metabolic state. This state appears to be insensitive to adaptation mechanisms as transcripts and metabolites reciprocally enhance each other to maintain the tumor-like metabolic phenotype. In conclusion, this work demonstrates how genome scale metabolic models can help identifying functional mechanisms from complex and multi-layered omics data. Appropriate genome scale metabolic models combined with metabolite measurements have proven particularly useful in this context. The comprehensive understanding of all integrated aspects of a cell’s physiology is a challenging endeavor and the results of this thesis might stimulate further research towards this goal.

Spatial heterogeneity is a fundamental feature of the tumor microenvironment (TME), and tackling spatial heterogeneity in neoplastic metabolic aberrations is critical for tumor treatment. Genome-scale metabolic network models have been used successfully to simulate cancer metabolic networks. However, most models use bulk gene expression data of entire tumor biopsies, ignoring spatial heterogeneity in the TME. To account for spatial heterogeneity, we performed spatially-resolved metabolic network modeling of the prostate cancer microenvironment. We discovered novel malignant-cell-specific metabolic vulnerabilities targetable by small molecule compounds. We predicted that inhibiting the fatty acid desaturase SCD1 may selectively kill cancer cells based on our discovery of spatial separation of fatty acid synthesis and desaturation. We also uncovered higher prostaglandin metabolic gene expression in the tumor, relative to the surrounding tissue. Therefore, we predicted that inhibiting the prostaglandin transporter SLCO2A1 may selectively kill cancer cells. Importantly, SCD1 and SLCO2A1 have been previously shown to be potently and selectively inhibited by compounds such as CAY10566 and suramin, respectively. We also uncovered cancer-selective metabolic liabilities in central carbon, amino acid, and lipid metabolism. Our novel cancer-specific predictions provide new opportunities to develop selective drug targets for prostate cancer and other cancers where spatial transcriptomics datasets are available.

Constraint-based modeling for genome-scale metabolic networks has emerged in the last years as a promising approach to elucidate drug targets in cancer. Beyond the canonical biosynthetic routes to produce biomass, it is of key importance to focus on metabolic routes that sustain the proliferative capacity through the regulation of other biological means in order to improve in-silico gene essentiality analyses. Polyamines are polycations with central roles in cancer cell proliferation, through the regulation of transcription and translation among other things, but are typically neglected in in silico cancer metabolic models. In this study, we analysed essential genes for the biosynthesis of polyamines. Our analysis corroborates the importance of previously known regulators of the pathway, such as Adenosylmethionine Decarboxylase 1 (AMD1) and uncovers novel enzymes predicted to be relevant for polyamine homeostasis. We focused on Adenine Phosphoribosyltransferase (APRT) and demonstrated the detrimental consequence of APRT gene silencing on different leukaemia cell lines. Our results highlight the importance of revisiting the metabolic models used for in-silico gene essentiality analyses in order to maximize the potential for drug target identification in cancer.

Microsatellite instability (MSI), a class of genetic hypermutability that arises from impaired DNA mismatch repair (MMR), contributes to the development of many malignancies including colon, endometrial, gastric, and ovarian cancers. While immune checkpoint blockade (ICB) has been an effective therapy for many patients with MSI cancers, numerous patients with MSI malignancies do not respond to ICB or the use of these agents are limited by their toxicity. Hence, there is still a pressing need to develop further therapies against MSI cancers. One approach to develop novel therapeutics is to leverage synthetic lethality, a phenomenon whereby the simultaneous occurrence of two or more genetic events lead to cell death but one event alone does not. DNA repair processes represent attractive synthetic lethal targets since many cancers exhibit an impaired DNA repair pathway, which can lead these cancers to become dependent on specific repair proteins. The success of poly (ADP ribose) polymerase (PARP) inhibitors in homologous recombination-deficient cancers highlights the potential of this approach. Hypothesizing that other DNA repair defects would give rise to alternative synthetic lethal relationships, we asked if there are specific dependencies in MSI cancers. Here, we analyzed data from large-scale CRISPR/Cas9 and RNA interference (RNAi) functional genomic screens and found that the RecQ DNA helicase WRN was selectively essential in MSI models in vitro and in vivo, but dispensable in microsatellite stable models. WRN silencing induced double-strand DNA breaks, activated a DNA-damage response, and promoted apoptosis and cell cycle arrest preferentially in MSI models. MSI cancer models specifically required WRN’s helicase activity, but not its exonuclease activity. These findings expose WRN as a synthetic lethal vulnerability and promising drug target for MSI cancers. Citation Format: Edmond M. Chan, Tsukasa Shibue, James McFarland, Benjamin Gaeta, Francisca Vazquez, Adam J. Bass. WRN helicase is a synthetic lethal target in microsatellite unstable cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr LB-048.

Non-mitotic functions of polo-like kinases in cancer cells

Tyrosine phosphorylation has a key role in intracellular signaling. Inappropriate proliferation and survival cues in tumor cells often occur as a consequence of unregulated tyrosine kinase activity. Much of the current development of anti-cancer therapies tries to target causative proteins in a specific manner to minimize side-effects. One attractive group of target proteins is the kinases. c-Kit is a receptor tyrosine kinase that normally controls the function of primitive hematopoietic cells, melanocytes and germ cells. It has become clear that uncontrolled activity of c-Kit contributes to formation of an array of human tumors. The unregulated activity of c-Kit may be due to overexpression, autocrine loops or mutational activation. This makes c-Kit an excellent target for cancer therapies in these tumors. In this review we will highlight the current knowledge on the signal transduction molecules and pathways activated by c-Kit under normal conditions and in cancer cells, and the role of aberrant c-Kit signaling in cancer progression. Recent advances in the development of specific inhibitors interfering with these signal transduction pathways will be discussed.

The phosphoinositide 3-kinase/3-phosphoinositide-dependent kinase 1 (PDK1)/Akt signaling pathway plays a key role in cancer cell growth, survival, and tumor angiogenesis and represents a promising target for anticancer drugs. Here, we describe three potent PDK1 inhibitors, BX-795, BX-912, and BX-320 (IC(50) = 11-30 nm) and their initial biological characterization. The inhibitors blocked PDK1/Akt signaling in tumor cells and inhibited the anchorage-dependent growth of a variety of tumor cell lines in culture or induced apoptosis. A number of cancer cell lines with elevated Akt activity were >30-fold more sensitive to growth inhibition by PDK1 inhibitors in soft agar than on tissue culture plastic, consistent with the cell survival function of the PDK1/Akt signaling pathway, which is particularly important for unattached cells. BX-320 inhibited the growth of LOX melanoma tumors in the lungs of nude mice after injection of tumor cells into the tail vein. The effect of BX-320 on cancer cell growth in vitro and in vivo indicates that PDK1 inhibitors may have clinical utility as anticancer agents.

Metabolic reprogramming is considered a hallmark of cancers, which plays an important role in cancer progression and development, partly as a consequence of somatic mutations. A representative product of metabolic reprogramming in cancers is oncometabolites that show abnormal accumulation in a cancer cell, induce malignancy, and are generated upon mutations in a metabolic gene, often IDH1, SDH or FH. Identification of novel mutation-associated metabolites will facilitate developing biomarkers and treatment strategies for cancers, as in the case of ivosidenib, an FDA approved drug for treating acute myeloid leukemia (AML) having the IDH1 mutant. To this end, we develop a computational workflow that predicts so-called metabolite-gene-pathway sets (MGPs) that present metabolites and metabolic pathways significantly associated with gene mutations in cancers. The computational workflow uses cancer patients’ mutation data and a computational model of cellular metabolism called a genome-scale metabolic models (GEM). In this study, the computational workflow was demonstrated using 943 cancer patient-specific GEMs representing 24 different cancer types based on the Pan-Cancer Analysis of Whole Genomes (PCAWG) data [1], which, as a result, predicted 4,135 MGPs for these multiple cancer types. The computational workflow was shown to generate biologically meaningful MGPs on the basis of multi-omics data from 17 AML samples collected in this study as well as previous published studies. For the AML samples, metabolites of the MGPs predicted using the computational workflow showed significantly different intracellular concentrations, depending on mutation of genes involved in the MGP. Moreover, for the 115 MGPs predicted for CNS-GBM/Oligo that are associated with mutation of CIC, EGFR, IDH1 or TP53, 69% of the MGPs were supported by previous studies. This two-level validation indeed showed that it is possible to characterize metabolic pathways and their metabolites that are affected in response to somatic mutations in a cancer cell. Our computational workflow and its prediction outcomes will help better understand the mutation-associated metabolic reprogramming in cancers, and serve as a valuable resource for further extended studies on cancer metabolism. [1] The ICGC/TCGA Pan-Cancer Analysis of Whole Genomes Consortium. Pan-cancer analysis of whole genomes. Nature 578, 82-93 Citation Format: Sang Mi Lee, GaRyoung Lee, Sungyoung Lee, Hyojin Song, Sung Soo Yoon, Hongseok Yun, Youngil Koh, Hyun Uk Kim. A computational model of cellular metabolism with mutation data predicts metabolites associated with somatic mutations in cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 2733.