Greg Villareal

Precision medicine using individualized biosimulations of drug dosing: Alzheimer's disease

Precision medicine requires the right drug at the right dose for the right patient at the right time. This study used a computational biology model of 30 metabolic and transport pathways and multiple compartments to simulate oral dosing of pioglitazone that is currently in clinical trials to delay onset of Alzheimers disease. The Transcriptome-To-MetabolomeTM Method was used to simulate individual human subjects by using their gene expression profiles to determine parameters for the kinetic biosimulation. The in silico plasma profiles for multiple doses matched those for in vivo results from literature. Individual ED50 values were determined on each subject for the mitochondrial pyruvate carrier bound by pioglitazone as the target. This approach will allow determination of effective dosing for individual subjects in clinical trials and patients for treatments.

MSDC-0160 and MSDC-0602 Binding with Human Mitochondrial Pyruvate Carrier (MPC) 1 and 2 Heterodimer: PPARγ Activating and Sparing TZDs as Therapeutics

Themitochondrialpyruvatecarrier(MPC)isanoveltargetfortherapeuticdrugstotreatAlzheimer’s andParkinson’sdisease, diabetesmellitus, andnon-alcoholic steatohepatitis (NASH).Metabolic SolutionsDevelopmentCompany(MSDC)hastwothiazolidinediones,MSDC-0160andMSDC-0602, inthepipeline.ThisreportdescribesresultsforaMPC1/2heterodimerhomologymodel.TheFASTA sequencesforMPC1andMPC2wereaccessedfromUniProtandsubmittedtoRaptorX,resultingin bestcandidatemonomeric“proteindatabase”filesforeach.OnemutantformofMPC1,L36I,was alsoprocessed.TheseweresubmittedtoPyDocktogeneratebestcandidateMPC1/2heterodimer modelsthatwereusedforliganddockinganalyseswithAutoDockVinaand“RosettaOnlineServer thatIncludesEveryone”(ROSIE).Multiplebindingsitesforpyruvateandbothdrugswerefoundon bothMPC1andMPC2subunitswithdrugshavingnearlydoubletheaffinityineachcaseexceptthe intermediateandopen-instatesfortheL36Imutanttransporter. KeywoRDS Protein Homology Modeling, Inhibitor Docking, L36I Mutant, Michael Addition, MPC1, MPC2, Protein Protein Docking, Pyruvate, Single Nucleotide Polymorphism, Thiohemiacetal, UK-5099, Wild Type

Modeling non-clinical and clinical drug tests in Gaucher disease

There is need for modeling biological systems to accelerate drug pipelines for treating metabolic diseases. The eliglustat treatment for Gaucher disease is approved by the FDA with a companion genomic test. The Transcriptome-To-Metabolome™ biosimulation technology was used to model, in silico, a standard non-clinical eliglustat test with an in vitro canine kidney cell system over-expressing a human gene; and a clinical test using human fibroblasts from control and Gaucher disease subjects. Protein homology modeling and docking studies were included to gather affinity parameters for the kinetic metabolic model. Pharmacodynamics and metabolomics analyses of the results replicated published findings and demonstrated that processing and transport of lysosomal proteins alone cannot explain the metabolic disorder. This technology shows promise for application to other diseases.

Low Dose Pioglitazone Attenuates Oxidative Damage in Early Alzheimer's Disease by Binding mitoNEET: Transcriptome-To-ReactomeTM Biosimulation of Neurons

Oxidative damage OD is considered to be a central component in the progression of Alzheimers disease AD. 8-hydroxyguanosine 8-OHG, a readily oxidized ribonucleic acid found in AD, was used as a biomarker to investigate the role of OD in the progression of the disease. A disruption in two critical Thioredoxin-Dependent Peroxiredoxin System components, peroxiredoxin-3 Prx-3 and thioredoxin Trx, may serve as a source of the increased accumulation of OD observed in AD. We demonstrate that OD, in the form of 8-OHG, was quantitatively most significant during the earliest stage of AD [F 3, 25 = 5.08, p <.01]. A drastic decline in mitochondrial protein levels of Prx-3 [F 3, 25 = 8.74, p. < 01] and Trx [F 3, 25 = 4.33, p. < 05] were also observed across the progression of the disease. We then tested the efficacy of pioglitazone, a thiazolidinedione class drug aimed to delay onset of AD by acting on mitoNEET. Our results showed a significant reduction in the oxidized variant of mitoNEET within the incipient population when a 0.8mg dose was simulated in silico p = 0.0242; a. < 05.

Biomarkers from biosimulations: Transcriptome-To-Reactome™ Technology for individualized medicine.

We validated a model of the TGF-β signaling pathway using reactions from Reactome. Using a patentpending technique, gene expression profiles from individual patients are used to determine model parameters. Gene expression profiles from 45 women, normal, or benign tumor and malignant breast cancer were used as training and validating sets for assessing clinical sensitivity and specificity. Biomarkers were identified from the biosimulation results using sensitivity analyses and derivative properties from the model. A membrane signaling marker had sensitivity of 80% and specificity of 60%; while a nuclear transcription factor marker had sensitivity of 80% and specificity of 90% to predict malignancy. Use of Fagans nomogram increased probability from 7.5% for positive mammogram to 39% with positive results of the biosimulation for the nuclear marker. Our technology will allow researchers to identify and develop biomarkers and assist clinicians in diagnostic and treatment decision making.

In Vivo and In Silico Evidence: Hippocampal Cholesterol Metabolism Decreases with Aging and Increases with Alzheimers Disease -- Modeling Brain Aging and Disease

Genome wide association studies revealed genetic evidence for involvement of cholesterol metabolism in the etiology of Alzheimers disease (AD). The present study used gene expression profiles on human Cornu Ammonis 1(CA1) for subjects with severe AD and an age-matched group to determine the enzyme reaction rate constants for 16 core metabolic pathways including cholesterol biosynthesis, isoprenoid production, and cholesterol catabolism for removal from brain. The core metabolic model was used to simulate a young hippocampus (20-39yo) to compare with age-matched control group for our AD study (mean= 85.3y). In the aged human brain, the flux through the rate limiting step in the simulation for aged human hippocampus was lower by 9.5%, the cholesterol level was 52.3% lower in the simulation and 33.6% lower in the aged human brain, validating the in silico method. Data was also used to evaluate sterol regulatory element binding protein 1 and 2 (SREBP1 & SREBP2) showing the levels were increased significantly in the severe AD samples versus age-matched control. We predicted that the core metabolism simulation of severe AD versus age-matched control would show corresponding results and they do. The sensitivities analyses for incipient and severe AD demonstrated how they differ: Most reactions are insensitive for severe AD and two sensitive peaks are obvious, cholesterol and ubiquinone levels are most sensitive to cholesterol 24-hydroxylase, CYP46a1. These findings are consistent with statins being ineffective in clinical trials for treatment of AD, post-diagnosis.

Integrating -omics information with biosimulations to assist diagnosis and treatment of diabetes mellitus in silico

Genomic test results were used to predict protein structure for the mitochondrial pyruvate carrier (MPC) that is a novel target for an FDA approved anti-diabetic drug, pioglitazone. Transcriptomic test results were used along with in silico drug-protein binding results to determine parameters for oral glucose tolerance and metabolic biosimulation models used to diagnose diabetes mellitus and determine the best dose for treatment, respectively. The metabolomic and fluxomic biosimulation results were used to determine the proper dose of pioglitazone. This -omic and in silico information can be integrated into comprehensive health information to assist clinicians in medical decision making.

Alzheimer's and Parkinson's Disease Novel Therapeutic Target: The Mitochondrial Pyruvate Carrier - Ligand Docking to Screen Natural Compounds Related to Classic Inhibitors

Thiazolidinedione (TZD) drugs (Takeda Pharmaceuticals and Metabolic Solutions Development Company) targeting inhibition of the mitochondrial pyruvate carrier (MPC) are currently being tested in clinical trials to prevent progression into mild cognitive impairment of Alzheimer’s disease (AD) or in the pipeline to prevent neurodegeneration in Parkinson’s disease (PD). These have Ki values in the μM range. This study was focused on identifying candidate drug precursors of the natural cinnamic acid products that might have good bioavailability in the nM ranges forming covalent thiol bonds with targets. In silico protein homology modeling and ligand docking has demonstrated that binding cysteine residues within the transport channel is a key part of the inhibitory mechanism. These are covalent thiohemiacetal bonds with the alpha-carbon, carboxylate group, off a phenol ring. Like the classic MPC inhibitors, these natural derivatives of hydroxycinnamic acid have a conjugated pi-system used to form thiol bonds with the cysteine residue via Michael addition. Alzheimer’s and Parkinson’s Disease Novel Therapeutic Target: The Mitochondrial Pyruvate Carrier Ligand Docking to Screen Natural Compounds Related to Classic Inhibitors

Molecular neural model recreates electrophysiology: Transcriptome-To-Physiome™ NeurobioSimulations using COPASI ® software

A multi-compartmental molecular model has been developed for rodent basal forebrain cholinergic neurons with established gene expression levels. Reconstruction of neurons and network function were acquired using the Transcriptome-To-Physiome™ (TTP™) NeurobioSimulation. Gene expression values [NCBI GEO GSE 13379] were used to derive protein level and kinetic parameters for ligand and voltage gated ion channels in the TTP™ NeurobioSimulator Model using COPASI® software. Global parameters for membrane potential used permeability and ion concentrations inside and outside of the membrane in the Goldman-Hodgkin-Katz equation. Four compartments of the model neuron are included: glutamate synapse, distal dendrite, proximal dendrite, and axon hillock. The simulation of a voltage-gated sodium channel activation, and inactivated states of distal dendrites of cholinergic modeled neurons depends on the excitatory postsynaptic potential (EPSP) event. This distally activated event yielded temporally relevant proximal dendritic activation and inactivation of voltage-gated sodium and potassium channels in the reconstructed neuron. Graded potentials showed temporal summation and a classic action potential occurs at the axon hillock with sodium and potassium fluxes as expected. In future studies, we will reconstruct the electrophysiology of vulnerable neuronal populations in the diseased brain and compare them to controls thus lending substantial insight into molecular and network function corollary to neuropathogenesis.

Abstract 5224: Breast cancer biomarkers from biosimulations: transcriptome-to-reactome™ technology.

Proceedings: AACR 104th Annual Meeting 2013; Apr 6-10, 2013; Washington, DC Genome-wide gene-expression profiles, transcriptomes, from microarray tests are used for clinical decision-making; novel uses of transcriptomes will continue to emerge. We are commercializing a patent-pending technology using transcriptomes to determine parameters for deterministic-kinetic models of biological pathways, Transcriptome-To-Reactome:Transforming Growth Factor Beta Signaling. We demonstrate how biosimulation results for proteomics, flux, and a derivative property of the network system can be utilized as biomarkers for breast cancer. The TGFbeta Signaling Model was accessed from Reactome.org; manual curation was performed to add beta-induced-gene-human-clone-3 as a target gene for Smad transcription factors. Biosimulations were run with COPASI. Transcriptomes were accessed from NCBI GEO archives, where peripheral blood mononuclear cells (PBMC) were collected from women with suspect initial mammograms, prior to undergoing diagnostic biopsy to differentiate benign from malignant cases and from women with normal initial mammograms as negative controls. Ten individual biosimulations were run in each category. Candidate biomarkers were identified by differences in simulated proteomic levels, reaction fluxes, and sensitivities analyses of simulated reactions. A 10 fold cross over design for training and testing data sets was used to assess biomarker candidates, a TGFbeta1-receptor complex, Smads, and slope of the target gene expression flux. Our results show sensitivities analyses showed differences between control and benign versus malignant biosimulation reactions. Ten reactant-reaction sensitivities were greater in malignant versus both control and benign categories. The TGFβ1:type-II-receptor:Phospho-type-I-receptor:SARA-complex was selected for analysis as a biomarker; also the reaction for BIGH3 expression was identified and the slope of the rising phase of BIGH3 expression was assessed; as were Smad3 and Smad4 levels. For predicting malignancy versus control, accuracy was 0.60, 0.85, 0.80, and 0.90; sensitivity was 0.30, 0.90, 0.70, and 0.80; specificity was 0.90, 0.80, 0.90, and 1.00; PPV was 75%, 82%, 88%, and 100%; NPV was 56%, 89%, 75%, and 83%; AUC was 0.60, 0.77, 0.72, and 0.82, respectively. These biomarkers were effective for predicting benign vs. control, but worse for distinguishing malignant from benign classifications. We conclude this method is validated for identifying candidate biomarkers for cancer. This retrospective study satisfies the requirement for this phase of discovery and evaluation of cancer biomarkers. Prospective studies are needed for successful commercialization. Citation Format: Richard LeBaron, Greg Villareal, Dawnlee J. Roberson, Clyde Phelix. Breast cancer biomarkers from biosimulations: transcriptome-to-reactome™ technology. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 5224. doi:10.1158/1538-7445.AM2013-5224