Personalization of cancer treatment using predictive simulation.

BACKGROUND: The personalization of cancer treatments implies the reconsideration of a one-size-fits-all paradigm. This move has spawned increased use of next generation sequencing to understand mutations and copy number aberrations in cancer cells. Initial personalization successes have been primarily driven by drugs targeting one patient-specific oncogene (e.g., Gleevec, Xalkori, Herceptin). Unfortunately, most cancers include a multitude of aberrations, and the overall impact on cancer signaling and metabolic networks cannot be easily nullified by a single drug. METHODS: We used a novel predictive simulation approach to create an avatar of patient cancer cells using point mutations and copy number aberration data. Simulation avatars of myeloma patients were functionally screened using various molecularly targeted drugs both individually and in combination to identify drugs that are efficacious and synergistic. Repurposing of drugs that are FDA-approved or under clinical study with validated clinical safety and pharmacokinetic data can provide a rapid translational path to the clinic. High-risk multiple myeloma patients were modeled, and the simulation predictions were assessed ex vivo using patient cells. RESULTS: Here, we present an approach to address the key challenge of interpreting patient profiling genomic signatures into actionable clinical insights to make the personalization of cancer therapy a practical reality. Through the rational design of personalized treatments, our approach also targets multiple patient-relevant pathways to address the emergence of single therapy resistance. Our predictive platform identified drug regimens for four high-risk multiple myeloma patients. The predicted regimes were found to be effective in ex vivo analyses using patient cells. CONCLUSIONS: These multiple validations confirm this approach and methodology for the use of big data to create personalized therapeutics using predictive simulation approaches.

ß-Caryophyllene oxide inhibits constitutive and inducible STAT3 signaling pathway through induction of the SHP-1 protein tyrosine phosphatase.

Constitutive activation of STAT3 is frequently observed and closely linked with proliferation, survival, invasion, metastasis and angiogenesis in tumor cells. In the present study, we investigated whether ß-caryophyllene oxide (CPO), a sesquiterpene isolated primarily from the essential oils of medicinal plants such as guava (Psidium guajava), and oregano (Origanum vulgare L.), can mediate its effect through interference with the STAT3 activation pathway in cancer cells. The effect of CPO on STAT3 activation, associated protein kinases and phosphatase, STAT3-regulated gene products and apoptosis was investigated using both functional proteomics tumor pathway technology platform and different tumor cell lines. We found that CPO suppressed constitutive STAT3 activation in multiple myeloma (MM), breast and prostate cancer cell lines, with a significant dose- and time-dependent effects observed in MM cells. The suppression was mediated through the inhibition of activation of upstream kinases c-Src and JAK1/2. Also, vanadate treatment reversed CPO-induced down-regulation of STAT3, suggesting the involvement of a tyrosine phosphatase. Indeed, we found that CPO induced the expression of tyrosine phosphatase SHP-1 that correlated with the down-regulation of constitutive STAT3 activation. Interestingly, deletion of SHP-1 gene by siRNA abolished the ability of CPO to inhibit STAT3 activation. The inhibition of STAT3 activation by CPO inhibited proliferation, induced apoptosis and abrogated the invasive potential of tumor cells. Our results suggest for the first time that CPO is a novel blocker of STAT3 signaling cascade and thus has an enormous potential for the treatment of various cancers harboring constitutively activated STAT3.

In silico modeling predicts drug sensitivity of patient-derived cancer cells.

BACKGROUND: Glioblastoma (GBM) is an aggressive disease associated with poor survival. It is essential to account for the complexity of GBM biology to improve diagnostic and therapeutic strategies. This complexity is best represented by the increasing amounts of profiling ("omics") data available due to advances in biotechnology. The challenge of integrating these vast genomic and proteomic data can be addressed by a comprehensive systems modeling approach. METHODS: Here, we present an in silico model, where we simulate GBM tumor cells using genomic profiling data. We use this in silico tumor model to predict responses of cancer cells to targeted drugs. Initially, we probed the results from a recent hypothesis-independent, empirical study by Garnett and co-workers that analyzed the sensitivity of hundreds of profiled cancer cell lines to 130 different anticancer agents. We then used the tumor model to predict sensitivity of patient-derived GBM cell lines to different targeted therapeutic agents. RESULTS: Among the drug-mutation associations reported in the Garnett study, our in silico model accurately predicted ~85% of the associations. While testing the model in a prospective manner using simulations of patient-derived GBM cell lines, we compared our simulation predictions with experimental data using the same cells in vitro. This analysis yielded a ~75% agreement of in silico drug sensitivity with in vitro experimental findings. CONCLUSIONS: These results demonstrate a strong predictability of our simulation approach using the in silico tumor model presented here. Our ultimate goal is to use this model to stratify patients for clinical trials. By accurately predicting responses of cancer cells to targeted agents a priori, this in silico tumor model provides an innovative approach to personalizing therapy and promises to improve clinical management of cancer.

Amplification and demultiplexing in insulin-regulated Akt protein kinase pathway in adipocytes.

Akt plays a major role in insulin regulation of metabolism in muscle, fat, and liver. Here, we show that in 3T3-L1 adipocytes, Akt operates optimally over a limited dynamic range. This indicates that Akt is a highly sensitive amplification step in the pathway. With robust insulin stimulation, substantial changes in Akt phosphorylation using either pharmacologic or genetic manipulations had relatively little effect on Akt activity. By integrating these data we observed that half-maximal Akt activity was achieved at a threshold level of Akt phosphorylation corresponding to 5-22% of its full dynamic range. This behavior was also associated with lack of concordance or demultiplexing in the behavior of downstream components. Most notably, FoxO1 phosphorylation was more sensitive to insulin and did not exhibit a change in its rate of phosphorylation between 1 and 100 nm insulin compared with other substrates (AS160, TSC2, GSK3). Similar differences were observed between various insulin-regulated pathways such as GLUT4 translocation and protein synthesis. These data indicate that Akt itself is a major amplification switch in the insulin signaling pathway and that features of the pathway enable the insulin signal to be split or demultiplexed into discrete outputs. This has important implications for the role of this pathway in disease.

Targeting chromosome 12q amplification in relapsed glioblastoma: the use of computational biological modeling to identify effective therapy-a case report.

Background: Relapsed glioblastoma (GBM) is often an imminently fatal condition with limited therapeutic options. Computation biological modeling, i.e., biosimulation, of comprehensive genomic information affords the opportunity to create a disease avatar that can be interrogated in silico with various drug combinations to identify the most effective therapies. Case Description: We report the outcome of a GBM patient with chromosome 12q amplification who achieved substantial disease remission from a novel therapy using this approach. Following next generation sequencing (NGS) was performed on the tumor specimen. Mutation and copy number changes were input into a computational biologic model to create an avatar of disease behavior and the malignant phenotype. In silico responses to various drug combinations were biosimulated in the disease network. Efficacy scores representing the computational effect of treatment for each strategy were generated and compared to each other to ascertain the differential benefit in drug response from various regimens. Biosimulation identified CDK4/6 inhibitors, nelfinavir and leflunomide to be effective agents singly and in combination. Upon receiving this treatment, the patient achieved a prompt and clinically meaningful remission lasting 6 months. Conclusions: Biosimulation has utility to identify active treatment combinations, stratify treatment options and identify investigational agents relevant to patients' comprehensive genomic abnormalities. Additionally, the combination of abemaciclib and nelfinavir appear promising for GBM and potentially other cancers harboring chromosome 12q amplification.

Genomic clustering analysis identifies molecular subtypes of thymic epithelial tumors independent of World Health Organization histologic type.

Further characterization of thymic epithelial tumors (TETs) is needed. Genomic information from 102 evaluable TETs from The Cancer Genome Atlas (TCGA) dataset and from the IU-TAB-1 cell line (type AB thymoma) underwent clustering analysis to identify molecular subtypes of TETs. Six novel molecular subtypes (TH1-TH6) of TETs from the TCGA were identified, and there was no association with WHO histologic subtype. The IU-TAB-1 cell line clustered into the TH4 molecular subtype and in vitro testing of candidate therapeutics was performed. The IU-TAB-1 cell line was noted to be resistant to everolimus (mTORC1 inhibitor) and sensitive to nelfinavir (AKT1 inhibitor) across the endpoints measured. Sensitivity to nelfinavir was due to the IU-TAB-1 cell line's gain-of function (GOF) mutation in PIK3CA and amplification of genes observed from array comparative genomic hybridization (aCGH), including AURKA, ERBB2, KIT, PDGFRA and PDGFB, that are known upregulate AKT, while resistance to everolimus was primarily driven by upregulation of downstream signaling of KIT, PDGFRA and PDGFB in the presence of mTORC1 inhibition. We present a novel molecular classification of TETs independent of WHO histologic subtype, which may be used for preclinical validation studies of potential candidate therapeutics of interest for this rare disease.

Computational Biological Modeling Identifies PD-(L)1 Immunotherapy Sensitivity Among Molecular Subgroups of KRAS-Mutated Non-Small-Cell Lung Cancer.

PURPOSE: KRAS-mutated (KRASMUT) non-small-cell lung cancer (NSCLC) is emerging as a heterogeneous disease defined by comutations, which may confer differential benefit to PD-(L)1 immunotherapy. In this study, we leveraged computational biological modeling (CBM) of tumor genomic data to identify PD-(L)1 immunotherapy sensitivity among KRASMUT NSCLC molecular subgroups. MATERIALS AND METHODS: In this multicohort retrospective analysis, the genotype clustering frequency ranked method was used for molecular clustering of tumor genomic data from 776 patients with KRASMUT NSCLC. These genomic data were input into the CBM, in which customized protein networks were characterized for each tumor. The CBM evaluated sensitivity to PD-(L)1 immunotherapy using three metrics: programmed death-ligand 1 expression, dendritic cell infiltration index (nine chemokine markers), and immunosuppressive biomarker expression index (14 markers). RESULTS: Genotype clustering identified eight molecular subgroups and the CBM characterized their shared cancer pathway characteristics: KRASMUT/TP53MUT, KRASMUT/CDKN2A/B/CMUT, KRASMUT/STK11MUT, KRASMUT/KEAP1MUT, KRASMUT/STK11MUT/KEAP1MUT, KRASMUT/PIK3CAMUT, KRAS MUT/ATMMUT, and KRASMUT without comutation. CBM identified PD-(L)1 immunotherapy sensitivity in the KRASMUT/TP53MUT, KRASMUT/PIK3CAMUT, and KRASMUT alone subgroups and resistance in the KEAP1MUT containing subgroups. There was insufficient genomic information to elucidate PD-(L)1 immunotherapy sensitivity by the CBM in the KRASMUT/CDKN2A/B/CMUT, KRASMUT/STK11MUT, and KRASMUT/ATMMUT subgroups. In an exploratory clinical cohort of 34 patients with advanced KRASMUT NSCLC treated with PD-(L)1 immunotherapy, the CBM-assessed overall survival correlated well with actual overall survival (r = 0.80, P < .001). CONCLUSION: CBM identified distinct PD-(L)1 immunotherapy sensitivity among molecular subgroups of KRASMUT NSCLC, in line with previous literature. These data provide proof-of-concept that computational modeling of tumor genomics could be used to expand on hypotheses from clinical observations of patients receiving PD-(L)1 immunotherapy and suggest mechanisms that underlie PD-(L)1 immunotherapy sensitivity.

A feedforward relationship between active Rac1 and phosphorylated Bcl-2 is critical for sustaining Bcl-2 phosphorylation and promoting cancer progression.

Active GTPase-Rac1 is associated with cellular processes involved in carcinogenesis and expression of Bcl-2 endows cells with the ability to evade apoptosis. Here we provide evidence that active Rac1 and Bcl-2 work in a positive feedforward loop to promote sustained phosphorylation of Bcl-2 at serine-70 (S70pBcl-2), which stabilizes its anti-apoptotic activity. Pharmacological and genetic inactivation of Rac1 prevent interaction with Bcl-2 and reduce S70pBcl-2. Similarly, BH3-mimetic inhibitors of Bcl-2 could disrupt Rac1-Bcl-2 interaction and reduce S70pBcl-2. This effect of active Rac1 could also be rescued by scavengers of intracellular superoxide (O2.-), thus implicating NOX-activating activity of Rac1 in promoting S70pBcl-2. Moreover, active Rac1-mediated redox-dependent S70pBcl-2 involves the inhibition of phosphatase PP2A holoenzyme assembly. Sustained S70pBcl-2 in turn secures Rac1/Bcl-2 interaction. Importantly, inhibiting Rac1 activity, scavenging O2.- or employing BH3-mimetic inhibitor significantly reduced S70pBcl-2-mediated survival in cancer cells. Notably, Rac1 expression, and its interaction with Bcl-2, positively correlate with S70pBcl-2 levels in patient-derived lymphoma tissues and with advanced stage lymphoma and melanoma. Together, we provide evidence of a positive feedforward loop involving active Rac1, S70pBcl-2 and PP2A, which could have potential diagnostic, prognostic and therapeutic implications.

Computational Models Accurately Predict Multi-Cell Biomarker Profiles in Inflammation and Cancer.

Individual computational models of single myeloid, lymphoid, epithelial, and cancer cells were created and combined into multi-cell computational models and used to predict the collective chemokine, cytokine, and cellular biomarker profiles often seen in inflamed or cancerous tissues. Predicted chemokine and cytokine output profiles from multi-cell computational models of gingival epithelial keratinocytes (GE KER), dendritic cells (DC), and helper T lymphocytes (HTL) exposed to lipopolysaccharide (LPS) or synthetic triacylated lipopeptide (Pam3CSK4) as well as multi-cell computational models of multiple myeloma (MM) and DC were validated using the observed chemokine and cytokine responses from the same cell type combinations grown in laboratory multi-cell cultures with accuracy. Predicted and observed chemokine and cytokine responses of GE KER + DC + HTL exposed to LPS and Pam3CSK4 matched 75% (15/20, p = 0.02069) and 80% (16/20, P = 0.005909), respectively. Multi-cell computational models became 'personalized' when cell line-specific genomic data were included into simulations, again validated with the same cell lines grown in laboratory multi-cell cultures. Here, predicted and observed chemokine and cytokine responses of MM cells lines MM.1S and U266B1 matched 75% (3/4) and MM.1S and U266B1 inhibition of DC marker expression in co-culture matched 100% (6/6). Multi-cell computational models have the potential to identify approaches altering the predicted disease-associated output profiles, particularly as high throughput screening tools for anti-inflammatory or immuno-oncology treatments of inflamed multi-cellular tissues and the tumor microenvironment.

Computational modeling of early T-cell precursor acute lymphoblastic leukemia (ETP-ALL) to identify personalized therapy using genomics.

Early T-cell precursor acute lymphoblastic leukemia (ETP-ALL) is an aggressive hematological malignancy for which optimal therapeutic approaches are poorly characterized. Using computational biology modeling (CBM) in conjunction with genomic data from cell lines and individual patients, we generated disease-specific protein network maps that were used to identify unique characteristics associated with the mutational profiles of ETP-ALL compared to non-ETP-ALL (T-ALL) cases and simulated cellular responses to a digital library of FDA-approved and investigational agents. Genomics-based classification of ETP-ALL patients using CBM had a prediction sensitivity and specificity of 93% and 87%, respectively. This analysis identified key genomic and pathway characteristics that are distinct in ETP-ALL including deletion of nucleophosmin-1 (NPM1), mutations of which are used to direct therapeutic decisions in acute myeloid leukemia. Computational simulations based on mutational profiles of 62 ETP-ALL patient models identified 87 unique targeted combination therapies in 56 of the 62 patients despite actionable mutations being present in only 37% of ETP-ALL patients. Shortlisted two-drug combinations were predicted to be synergistic in 11 profiles and were validated by in vitro chemosensitivity assays. In conclusion, computational modeling was able to identify unique biomarkers and pathways for ETP-ALL, and identify new drug combinations for potential clinical testing.

Computational drug treatment simulations on projections of dysregulated protein networks derived from the myelodysplastic mutanome match clinical response in patients.

Although the majority of MDS patients fail to achieve clinical improvement to approved therapies, some patients benefit from treatment. Predicting patient response prior to therapy would improve treatment effectiveness, avoid treatment-related adverse events and reduce healthcare costs. Three separate cohorts of MDS patients were used to simulate drug response to lenalidomide alone, hypomethylating agent (HMA) alone, or HMA plus lenalidomide. Utilizing a computational biology program, genomic abnormalities in each patient were used to create an intracellular pathway map that was then used to screen for drug response. In the lenalidomide treated cohort, computer modeling correctly matched clinical responses in 37/46 patients (80%). In the second cohort, 15 HMA patients were modeled and correctly matched to responses in 12 (80%). In the third cohort, computer modeling correctly matched responses in 10/10 patients (100%). This computational biology network approach identified GGH overexpression as a potential resistance factor to HMA treatment and paradoxical activation of beta-catenin (through Csnk1a1 inhibition) as a resistance factor to lenalidomide treatment. We demonstrate that a computational technology is able to map the complexity of the MDS mutanome to simulate and predict drug response. This tool can improve understanding of MDS biology and mechanisms of drug sensitivity and resistance.

Identification of myeloproliferative neoplasm drug agents via predictive simulation modeling: assessing responsiveness with micro-environment derived cytokines.

Previous studies have shown that the bone marrow micro-environment supports the myeloproliferative neoplasms (MPN) phenotype including via the production of cytokines that can induce resistance to frontline MPN therapies. However, the mechanisms by which this occurs are poorly understood. Moreover, the ability to rapidly identify drug agents that can act as adjuvants to existing MPN frontline therapies is virtually non-existent. Here, using a novel predictive simulation approach, we sought to determine the effect of various drug agents on MPN cell lines, both with and without the micro-environment derived inflammatory cytokines. We first created individual simulation models for two representative MPN cell lines; HEL and SET-2, based on their genomic mutation and copy number variation (CNV) data. Running computational simulations on these virtual cell line models, we identified a synergistic effect of two drug agents on cell proliferation and viability; namely, the Jak2 kinase inhibitor, G6, and the Bcl-2 inhibitor, ABT737. IL-6 did not show any impact on the cells due to the predicted lack of IL-6 signaling within these cells. Interestingly, TNFα increased the sensitivity of the single drug agents and their use in combination while IFNγ decreased the sensitivity. In summary, this study predictively identified two drug agents that reduce MPN cell viability via independent mechanisms that was prospectively validated. Moreover, their efficacy is either potentiated or inhibited, by some of the micro-environment derived cytokines. Lastly, this study has validated the use of this simulation based technology to prospectively determine such responses.

Overexpression of Bcl-2 induces STAT-3 activation via an increase in mitochondrial superoxide.

We recently reported a novel interaction between Bcl-2 and Rac1 and linked that to the ability of Bcl-2 to induce a pro-oxidant state in cancer cells. To gain further insight into the functional relevance of this interaction, we utilized computer simulation based on the protein pathway dynamic network created by Cellworks Group Inc. STAT3 was identified among targets that positively correlated with Rac1 and/or Bcl-2 expression levels. Validating this, the activation level of STAT3, as marked by p-Tyr705, particularly in the mitochondria, was significantly higher in Bcl-2-overexpressing cancer cells. Bcl-2-induced STAT3 activation was a function of GTP-loaded Rac1 and NADPH oxidase (Nox)-dependent increase in intracellular superoxide (O2•-). Furthermore, ABT199, a BH-3 specific inhibitor of Bcl-2, as well as silencing of Bcl-2 blocked STAT3 phosphorylation. Interestingly, while inhibiting intracellular O2•- blocked STAT3 phosphorylation, transient overexpression of wild type STAT3 resulted in a significant increase in mitochondrial O2•- production, which was rescued by the functional mutants of STAT3 (Y705F). Notably, a strong correlation between the expression and/or phosphorylation of STAT3 and Bcl-2 was observed in primary tissues derived from patients with different sub-sets of B cell lymphoma. These data demonstrate the presence of a functional crosstalk between Bcl-2, Rac1 and activated STAT3 in promoting a permissive redox milieu for cell survival. Results also highlight the potential utility of a signature involving Bcl-2 overexpression, Rac1 activation and STAT3 phosphorylation for stratifying clinical lymphomas based on disease severity and chemoresistance.

Predictive simulation approach for designing cancer therapeutic regimens with novel biological mechanisms.

Introduction Ursolic acid (UA) is a pentacyclic triterpene acid present in many plants, including apples, basil, cranberries, and rosemary. UA suppresses proliferation and induces apoptosis in a variety of tumor cells via inhibition of nuclear factor kappa-light-chain-enhancer of activated B cells (NFκB). Given that single agent therapy is a major clinical obstacle to overcome in the treatment of cancer, we sought to enhance the anti-cancer efficacy of UA through rational design of combinatorial therapeutic regimens that target multiple signaling pathways critical to carcinogenesis. Methodology Using a predictive simulation-based approach that models cancer disease physiology by integrating signaling and metabolic networks, we tested the effect of UA alone and in combination with 100 other agents across cell lines from colorectal cancer, non-small cell lung cancer and multiple myeloma. Our predictive results were validated in vitro using standard molecular assays. The MTT assay and flow cytometry were used to assess cellular proliferation. Western blotting was used to monitor the combinatorial effects on apoptotic and cellular signaling pathways. Synergy was analyzed using isobologram plots. Results We predictively identified c-Jun N-terminal kinase (JNK) as a pathway that may synergistically inhibit cancer growth when targeted in combination with NFκB. UA in combination with the pan-JNK inhibitor SP600125 showed maximal reduction in viability across a panel of cancer cell lines, thereby corroborating our predictive simulation assays. In HCT116 colon carcinoma cells, the combination caused a 52% reduction in viability compared with 18% and 27% for UA and SP600125 alone, respectively. In addition, isobologram plot analysis reveals synergy with lowered doses of the drugs in combination. The combination synergistically inhibited proliferation and induced apoptosis as evidenced by an increase in the percentage sub-G1 phase cells and cleavage of caspase 3 and poly ADP ribose polymerase (PARP). Combination treatment resulted in a significant reduction in the expression of cyclin D1 and c-Myc as compared with single agent treatment. Conclusions Our findings underscore the importance of targeting NFκB and JNK signaling in combination in cancer cells. These results also highlight and validate the use of predictive simulation technology to design therapeutics for targeting novel biological mechanisms using existing or novel chemistry.

A kinetic platform for in silico modeling of the metabolic dynamics in Escherichia coli.

BACKGROUND: A prerequisite for a successful design and discovery of an antibacterial drug is the identification of essential targets as well as potent inhibitors that adversely affect the survival of bacteria. In order to understand how intracellular perturbations occur due to inhibition of essential metabolic pathways, we have built, through the use of ordinary differential equations, a mathematical model of 8 major Escherichia coli pathways. RESULTS: Individual in vitro enzyme kinetic parameters published in the literature were used to build the network of pathways in such a way that the flux distribution matched that reported from whole cells. Gene regulation at the transcription level as well as feedback regulation of enzyme activity was incorporated as reported in the literature. The unknown kinetic parameters were estimated by trial and error through simulations by observing network stability. Metabolites, whose biosynthetic pathways were not represented in this platform, were provided at a fixed concentration. Unutilized products were maintained at a fixed concentration by removing excess quantities from the platform. This approach enabled us to achieve steady state levels of all the metabolites in the cell. The output of various simulations correlated well with those previously published. CONCLUSION: Such a virtual platform can be exploited for target identification through assessment of their vulnerability, desirable mode of target enzyme inhibition, and metabolite profiling to ascribe mechanism of action following a specific target inhibition. Vulnerability of targets in the biosynthetic pathway of coenzyme A was evaluated using this platform. In addition, we also report the utility of this platform in understanding the impact of a physiologically relevant carbon source, glucose versus acetate, on metabolite profiles of bacterial pathogens.