|
|
|
|
||
|
|
|
|
|
| January - March | April - June | July - August | September - December | |
|
2006
|
Summer
|
|||
|
2005
|
||||
|
2004
|
||||
|
2003
|
Winter
|
Summer
|
Fall
|
|
|
2002
|
Winter
|
Spring
|
Summer
|
Meets Fridays from 4-6pm, Boelter Hall 4440 (Biocybernetics
Lab)
This workshop/graduate seminar meets year-round to
discuss basic and cutting edge developments in integrative systems biology of
regulatory mechanisms, primarily at intra- and inter- cell signaling levels.
Our current focus is on pathway, network and other structured and data-driven
mathematical and simulation models of candidates for emerging hypotheses in
biology and medicine. Topics this summer include gene/protein network regulation
(p53), developmental cell biology (neural stem cell differentiation), receptor
signaling pathways (insulin), hormone secretion mechanisms (thyroid, pituitary),
viral infection (HCV), dynamic system biomodeling using gene expression kinetic
data, and protein engineering dynamics and optimal therapy in cancer (leukemia
and lymphoma).
Friday, Feb. 17
Pep Charusanti, Dept. of Chemistry
pep[AT]chem.ucla.edu
Chronic myeloid leukemia (CML) is a blood stem cell disorder characterized by
overproduction and accumulation of white blood cells, particularly neutrophils.
At the molecular level, the hallmark of CML is the presence of BCR-ABL, an oncoprotein
that constitutively activates numerous cell signaling pathways. BCR-ABL must
be phosphorylated to become active. In this talk I will present a math model
controlling BCR-ABL phosphorylation, one that postulates that there are uncharacterized
dynamics in the system.
Ribavirin is a nucleoside analog with antiviral effects on many RNA and DNA
viruses, including hepatitis C virus (HCV). We develop a PK/PD model for analyzing
effects of ribavirin monotherapy. We fit this model to ribavirin PK and viral
PD data, and also show that the model accounts for observed clinical effects
of ribavirin. Finally, we demonstrate how our model can be used to design more
effective, individualized ribavirin treatment plans.
Jason Lee, Electrical Engineering undergraduate
jasontlee[at]ucla.edu
Visual Interactive On-line Navigation Biomodeling Data Base
(VISION-BIOMODBASE)
Over the summer, we began
setting up another internet-accessible biomodeling module to be included under
the PIEallaMODE umbrella, coupled with its analytical functionality, and hosted
on our Lab server. VISION's purpose is to provide end-users with a visually-pleasing
and interactive means for facilely displaying, documenting and updating biomodels,
in any and all forms - cartoon, schematic, mathematical etc, with ready-access
to all embedded supportive data and reference material …all in one place. User
navigation is via graphical mappings from the biosystem diagram (e.g. a cartoon
or schematic of cell-signal interactions, or compartmentalizations, etc), and
a menu that explains the linkages of the interactive interface. It should be
quite useful for teaching as well as research. Users will be able to query the
online biomodel databases, and also store new ones. Tools for building new biomodel
databases are being provided, currently in HTML, and later in XML and other
community-accepted web-oriented authoring languages, including Flash animation.
This presentation will demonstrate a nearly complete ver.1 of the first biomodel
database, currently being installed for our new PK/PD model of mAb/FcRn/apoptosis
dynamics. Audience input will greatly facilitate in making VISION user-friendly
and forward-looking.
Robyn Javier, Cybernetics undergraduate
neurobyn[AT]ucla.edu
Friday, July 15
Simon Galbraith, PhD student, Dept of Computer Science
sgalbrai[AT]ucla.edu
Gene expression is regulated by transcription factors (TFs) that bind cis-regulatory elements in the promoter region of genes in order to induce or repress transcription. Some genes are regulated by multiple factors and together these form a network over the space of TF-gene interactions. Accurate models of the regulation and structure of transcriptional regulatory networks provide insight into these underlying processes and helps elucidate the current “state” of the cell. The majority of models of gene expression dynamics determine how inputs (a regulatory signal, or perturbation) produce transcriptional output (mRNA expression). In this talk, I will highlight state of the art approaches for transcriptional regulatory network modeling, and describe one such method, Network Component Analysis, that models transcriptional regulatory networks to compute transcription factor activities from DNA microarray data.
Paper to be presented:
"Simulation and validation of modelled sphingolipid metabolism in Saccharomyces
cerevisiae"
Fernando Alvarez-Vasquez, Jellie J. Sims, L. Ashley Cowart, Yasuo Okamoto, Ederhad
O. Voit, and Usuf A. Hannun.
Nature 443, 425-430 (2005).
Here, the authors validate experimentally a biochemical system theoretical model
of sphingolipid metabolism in yeast. Simulations of metabolic fluxes, enzyme
deletion, and the effects of inositol led to predictions that show significance
concordance with experimental results. The model also can also simulate the
effects of acute perturbations in fatty-acid precursors of sphingolipids. These
results demonstrate that modelling can be used to make testable predictions,
as well as assist in designing and developing in new experiments for this system.
Continuation of Simon's July 15 talk ...
Continuation of Simon's July 15 talk ...
Building a website for a B-cell database: We will show navigation methods and several VRML worlds which describe 3D interactions between the particles of endoplasmic reticulum and mitochondrion cells. To improve searching, filtering and maintaining the information on the website, an XML-based definition of metadata formats will be introduced. We will also show a new method to describe the active scene for better communication between the domain expert and software developer.
A predictive model for dually-labeled monoclonal antibody (mAb) kinetics. We discuss construction and implementation of a kinetic model of the biodistribution of wild type and Fc mutated mAbs labeled with different isotopes. mAb distribution and degradation rates in several organs are estimated and several new insights are gleaned from the model.
Population based studies have yielded rich, very high dimensional datasets that promise to yield clues to the pathophysiology of Chronic Fatigue Syndrome: The CDC is currently working with a very large and rich dataset to try and understand Chronic Fatigue Syndrome (CFS) on a molecular basis, and its impact on the endocrine, immune, HPA (hypothalamus-pituitary-adrenal axis), and neuro-cognitive systems. CFS is an extremely debilitating, long duration disease that is apparently non-fatal. Suffers manifest physical, physiological, and cognitive symptoms, and are generally severely impaired. There are no known demographic risk factors for CFS, except that it preferentially affects women. The CDC conservatively estimates that currently between 200 and 400,000 people in the U.S. are or have been affected by the disease (other academic groups estimate at least twice that). In terms of the complexity of its pleotropic affects on the individual and its subtlety, CFS is one of the biggest medical mysteries/challenges of our time. The CDC has been collecting data on CFS for roughly 15 years, much of it waiting to be scrutinized in detail. We are taking a two pronged approach of data-mining the very large dataset we already have, and biomodeling certain key systems (like the HPA axis) that we think are at the center of the disease to try and understand CFS. For the data-mining we are developing new methods for ultra high-dimensional data reduction and biostatistics. For the biomodeling we are pursuing the method of object models and state charts, coupled with traditional compartmental and differential equation modeling techniques.
Chronic myeloid leukemia (SML) is a blood stem cell disorder characterized by the overproduction and accumulation of white blood cells, particularly neutrophils. At the molecular level, the hallmark of CML is the presence of BCR-ABL, an oncoprotein that activates numerous cellular signaling pathways in a deregulated manner. BCR-ABL exists in two states, an inactive state and an active state; autophosphorylation of several tyrosine residues switches the oncoprotein from the inactive to active state. In this talk I will present a math model of BCR-ABL autophosphorylation, one that includes effects from the anti-CML drug Gleevec. Results from the model predict the existence of a novel Gleevec-resistance mechanism, and show that other mechanisms, besides autophosphorylation, contribute significantly to BCR-ABL activation. These model results suggest testable experimental hypotheses which, if confirmed, would have clinical implications by providing new, specific strategies to combat this disease.
A predictive model for dually-labeled monoclonal antibody (mAb) kinetics. We
discuss construction and implementation of a kinetic model of the biodistribution
of wild type and Fc mutated mAbs labeled with different isotopes. mAb distribution
and degradation rates in several organs are estimated and several new insights
are gleaned from the model. A discussion of issues raised during the Friday,
April 8th talk. Specifically, model changes were introduced that result in improved
fits to kidney and tumor concentration-time profiles.
Systems Biology of p53: The p53 protein is a tumor suppressor often mutated in cancer. Its primary regulator is MDM2. We have been formulating a mechanistic model of p53-MDM2 dynamics. Our current work focuses on integrating our model with previously published models as well as new knowledge of the mechanisms involved.
Ribavirin is a synthetic guanosine analogue that is known to have a transient
antiviral effect on HCV patients, and increases sustained viral response when
used in combination with peglyated IFN-a2a. Though the precise mechanism for
the antiviral actions of ribavirin is not known, ribavirin is known to induce
mutagenesis in HCV RNA which may decrease HCV infectivity. We have devised more
mechanistic models for predicting PK and PD effects of ribavirin treatment.
Our models fit ribavirin data from blood and urine, and accounts for the transient
decline in viral load following ribavirin treatment.
Applying VRML to dynamic interactions of Rituxan Pharmacokinetic models:
VRML is a practical tool for showing dynamic 3D scenes and objects on the Internet
for educational purposes. We demonstrate some VRML worlds to interactively present
the mechanisms underlying pharmacokinetic models. After introducing the methods
used to enhance visional effects, we discuss the data exchange strategy between
the modeling building, storage and displaying tool, and suggest a definition
for representing scenarios which may be supported as a VRML extension.
Friday, January
14
Pep Charusanti, PhD Student, Biocybernetics Lab and
Dept of Chemistry
pep[AT]chem.ucla.edu
Chronic myeloid leukemia (CML) is a blood disorder characterized by the presence of the Bcr-Abl oncoprotein. Gleevec, a small molecule inhibitor of Bcr-Abl, induces durable remission in CML patients in the chronic phase of the disease, but patients in more advanced phases of CML often develop resistance to Gleevec and relapse within several months. Several possible mechanisms of resistance exist, one of which is the presence of P-glycoprotein (PGP). To examine the effects of the presence of PGP, a mathematical model has been developed that accounts for Bcr-Abl, Gleevec, and PGP. In agreement with experimental results, the model shows that PGP reduces considerably the intracellular concentration of Gleevec. In addition, simulation results will be presented showing that the PGP alone is likely not sufficient to account for resistance, leading to hypotheses regarding other possible molecular resistance mechanisms to Gleevec.
I'll be giving a brief update on the status of our 2-tier physiologically-based
pharmacokinetic model for monoclonal antibodies. Currently our model is able
to accurately fit one of three data sets, but problems remain with the other
two. This talk will be given in a workshop format, opening up modeling problems
for discussion by the group.
Ribavirin is a synthetic guanosine analogue that is known to have a transient antiviral effect on HCV patients, and increases sustained viral response when used in combination with peglyated IFN-a2a. Though the precise mechanism for the antiviral actions of ribavirin is not known, ribavirin is known to induce mutagenesis in HCV RNA which may decrease HCV infectivity. My next goal is to extend the mathematical model describing HCV kinetics to account for the effects of ribavirin. Current results indicate that the pharmacokinetics of ribavirin can be described by a 3-compartment model, and its effects can be modeled by a simple precursor-dependent indirect pharmacodynamic response model.
Robyn
Javier, Undergraduate Cybernetics Major
neurobyn[AT]ucla.edu
Modeling Regulation of the Tumor Suppressor p53. In response to DNA damage, the protein p53 induces transcription of genes involved in cell cycle arrest and apoptosis. We have developed a mathematical model of this system, including p53 and its primary regulator, MDM2. Our most recent work has focused on expanding this model to include more mechanistic details.
The blood-brain barrier (BBB), formed from tightly sealed endothelial cells, allows oxygen, carbon dioxide and glucose into the brain, while preventing entry of toxins and other harmful substances. Consequently, because the BBB is impenetrable to most drugs, many brain diseases cannot be treated via oral or intracranial administration. We will examine the involvement of active efflux transporters (AET), such as the multi-drug resistance protein P-glycoprotein, which use the energy from ATP to eject substances from brain tissue back to blood. We will also discuss the Brain Efflux Index method, which is used to examine AET systems, and describe the development of mathematical models in predicting the transport of radiolabeled compounds across the BBB.
We'll give a brief introduction to our interactive database for rituximab and CD20 signaling in B cells. First, we'll discuss the process of transforming the database from its current format to an online format. Next, we'll give a demo of a few pages that were created using a VRML browser plugin. Finally, we'll briefly discuss the latest developments of web3D.
Jason Lee, Undergraduate Electrical Engineering w/Biomed Major
virgil[AT]ucla.edu
FALL
2004 Schedule
CS 298 - Workshop on Pathway/Network Receptor/Cell
Signaling Mechanism Modeling
What can we do
for Biomodeling with VRML?
VRML, or Virtual
Reality Modeling Language, was developed to define 3D worlds and connect them
over the Internet. In this talk, we first discuss the major distinguishing features
of VRML,accompanied by several application examples in chemistry, biology, aerospace
engineering and education. Then we discuss the basic concepts of VRML in a little
greater depth, and the process of building VRML worlds. Finally, I will present
my initial thoughts on how to use VRML in biological modeling software, to enhance
its visualization, and then we open the floor to discussion for further ideas
on the subject.
Friday, October
29
Greg
Ferl, PhD student, Biomedical Engineering IDP
gzferl[AT]ucla.edu
Multiscale Approach
to Modeling Monoclonal Antibody Kinetics
Summary
of our recently submitted manuscript, "A predictive FcRn-based mathematical
model for anti-CEA monoclonal antibodies in nude mice", plus a presentation
of our current research on semi-compartmental models of dually-labeled monoclonal
antibodies, used to predict IgG degradation rates in liver, tumor, spleen and
"slow uptake" organ pools such as muscle, skin and bone.
RETURN
TO TOP
MODELING
CORTICOSTEROID (CS) PHARMACOGENOMICS IN RAT LIVER USING GENE MICROARRAYS –
JY Jin, RR Almon, DC DuBois & WJ Jusko, JPET 307:93-109, 2003
This paper is apparently a pioneering one, being the first to successfully use
microarrays for analyzing the temporal patterns of extensive mRNA data (8000
genes) describing the time dependence of in vivo responses of a tissue to any
drug. It’s a lengthy paper, with lots of detail, so we are presenting
it in two parts, on two Fridays.
Part 1: Motivation, Experimental Design, PK, PD and Alternative Pharmacogenomic
Model Development
The experimental methods and design, alternative model development and model
discrimination methods are of particular interest to our group and these are
presented first. The results and broader implications of this work, to be presented
on September 10, include cluster analysis, which revealed six different temporal
patterns consisting of 197 CS-responsive probes representing 143 genes. Six
different mechanistic models are shown to fit the different clusters, revealing
a marked diversity of genes regulated by CS via a limited array of mechanisms,
and new hypotheses about these mechanisms of CS receptor-gene mediated action.
MODELING
CORTICOSTEROID (CS) PHARMACOGENOMICS IN RAT LIVER USING GENE MICROARRAYS –
JY Jin, RR Almon, DC DuBois & WJ Jusko, JPET 307:93-109, 2003
This paper is apparently a pioneering one, being the first to successfully use
microarrays for analyzing the temporal patterns of extensive mRNA data (8000
genes) describing the time dependence of in vivo responses of a tissue to any
drug. It’s a lengthy paper, with lots of detail, so we are presenting
it in two parts, on two Fridays.
Part 1: Motivation, Experimental Design, PK, PD and Alternative Pharmacogenomic
Model Development
The experimental methods and design, alternative model development and model
discrimination methods are of particular interest to our group and these are
presented first. The results and broader implications of this work, to be presented
on September 10, include cluster analysis, which revealed six different temporal
patterns consisting of 197 CS-responsive probes representing 143 genes. Six
different mechanistic models are shown to fit the different clusters, revealing
a marked diversity of genes regulated by CS via a limited array of mechanisms,
and new hypotheses about these mechanisms of CS receptor-gene mediated action.
EXPERIMENT DESIGN FOR COMPLETE, ORGAN-SPECIFIC 4-D QUANTIFICATION OF THYROID HORMONE DISTRIBUTION DYNAMICS BY PET FUNCTIONAL IMAGING & SIMULTANEOUS BLOOD SAMPLING.
We investigated
the properties and experimental usefulness of a PET functional imaging experiment
model, to establish a multi-organ-specific, space and time characterization
of the dynamics of T4 and its metabolites in mammals. The 21-plus-compartment
biomodel maps the cascade of metabolites formed following injection of radioactive
tracer T4, with distribution and metabolism occurring in all organs observable
under PET. We did a comprehensive a priori identifiability analysis of this
model structure, to uncover which kinetic parameters can be quantified, for
different combinations of exogenous test-input/measurement-output candidates,
involving multiple imaging experiments with outer-ring labeled T4 and its 6
labeled metabolites. The potentially useful candidate experiment configurations
were also Monte-Carlo-simulated using a prospective approach, thus completing
the analysis numerically. The outer-ring deiodination cascade - from radiolabeled
T4 to free iodide, measurable (in toto) in all organs during PET imaging, is
represented by a 7-tier multidirectionally cascading model, with test-input
in each tier plasma compartment and summed PET measurements from like numbered
plasma and organ compartments in each tier. This experiment design model was
shown to have intrinsically unidentifiable inter-level iodothyronine conversion
(e.g. T4 to T3, etc) parameters in individual organs. However, if plasma samples
are collected in addition to PET measurements, and each iodothyronine metabolite
and iodide are measured in these samples, all parameters of this extended experiment
model were shown numerically to be uniquely quantifiable. Augmentation of the
PET output with blood sampling data is thus the key to practical estimation
of all kinetic parameters. The resulting design is readily adaptable to small
animal or human PET studies; and the overall scheme is also applicable to other
molecular species in multilevel, multidirectional, precursor-multiproduct chains.
RETURN
TO TOP
Nearly 100% of large molecule drugs are unable to pass through the blood-brain barrier (BBB), a layer of tightly packed endothelial cells that prevents toxins and other substances in blood from entering the brain. Roughly 2% of low molecular weight lipid-soluble drugs can pass through the BBB, but only a few brain diseases respond to such drugs. The difficulty in transporting drugs across the BBB has been a major obstacle within the brain drug industry, requiring a new approach to brain drug targeting. Three types of transport systems (carrier-mediated transport, active efflux transport and receptor-mediated transport) have been studied in an attempt to facilitate the entry of drugs into the brain. We plan to develop a series of mathematical models that will compare various mechanisms of BBB drug transport in hopes of determining the most effective route of drug entry into the brain.
*Based on work by William M. Pardridge (reviewed in Mol Interv 3(2) 90-105, 2003)
Modeling Regulation of T3 and T4 in the Thyroid
Thyroid hormone (T3, T4) plasma levels are highly dependent on regulatory control in the hypothalamo-pituitary-thyroid axis; drastically high or low plasma thyroid hormone levels is indicatory of pathology of control. During this presentation, I will address the task of modeling regulation of T3 and T4 in the thyroid including, but not limited to, thyrotropin (TSH)-mediated stimulation of hormone production in and secretion from the thyroid. Other minor controlling factors, such as monodeiodination via selenedeiodinases D1 & D2 and carrier-mediated uptake and efflux of thyroid hormones, will also be discussed. Eventually, this thyroid "submodel" will serve as a vital part in a larger, full body human model used to simulate and predict pathophysiological states, and to determine levothyroxine bioequivalence standards.
The Hepatitis C Virus (HCV) is currently the leading cause of chronic hepatitis, liver cirrhosis, and hepatocellular carcinoma in the world, and it has been estimated that about 170 million people are infected with the virus worldwide. Standard treatments include interferon (IFN) monotherapy, or combination therapy with Ribavirin. Neumann and Perelson have devised a standard predator-prey type model of viral infection. My goal is to evaluate and expand on this model, and develop new, more mechanistic models of viral kinetics based on jointly-designed collaborative studies in chimpanzees and humans.
Thyroid hormone (T3, T4) plasma levels are highly dependent on regulatory control in the hypothalamo-pituitary-thyroid axis; drastically high or low plasma thyroid hormone levels is indicatory of pathology of control. During this presentation, I will address the task of modeling regulation of T3 and T4 in the thyroid including, but not limited to, thyrotropin (TSH)-mediated stimulation of hormone production in and secretion from the thyroid. Other minor controlling factors, such as monodeiodination via selenedeiodinases D1 & D2 and carrier-mediated uptake and efflux of thyroid hormones, will also be discussed. Eventually, this thyroid "submodel" will serve as a vital part in a larger, full body human model used to simulate and predict pathophysiological states, and to determine levothyroxine bioequivalence standards.
This talk will focus
on the union of cancer, monoclonal antibodies, and mathematical modeling. Since
the approval of Rituxan in 1997, monoclonal antibodies (Mabs) have become standard
in the treatment of several types of cancer, including Non-Hodgkin's Lymphoma
(NHL), Acute Myelogenous Leukemia, Chronic Lymphocytic Leukemia, Breast Cancer,
and Colorectal Cancer. Currently three Mabs have been approved for the treatment
of NHL: Rituxan, Zevalin, and Bexxar. Rituxan is used for immunotherapy of NHL
with or without chemotherapy (e.g. CHOP), while Zevalin and Bexxar are radiolabeled
Mabs (Y90, I131 resp.) used for radioimmunotherapy of NHL. Given the paucity
of published mathematical models in this area, the development of a model describing
the biodistribution of these Mabs represents a significant contribution to the
field and should aid in increasing drug efficacy.
The following areas will be covered:
1) Development of a mathematical model describing the biodistribution of anti-CEA
Mabs in mice. We augmented a previously published model in order to account
for FcRn (neonatal Fc receptor) mediated IgG half-life regulation and variable
tumor mass over the course of the experiment. Mouse biodistribution data was
used to parameterize and validate this in silico model. We were able to successfully
fit the model to intact anti-CEA Mab biodistribution data and then use the fitted
model to successfully predict the biodistribution of an anti-CEA F(ab')2 Mab
fragment.
2) Discussion of a newly formed collaboration with City of Hope. Our goal is
to modify the aforementioned mathematical model in order to optimize treatment
of NHL using Rituxan and Zevalin in the same patient. This will involve simplification
of the mouse model and scaling of species specific parameters from mouse to
human. 3) Model development and validation process. Discussion of the significant
challenges involved in the development of this type of mathematical model
RETURN
TO TOP
Algorithmic Noise Generated by Quantitation of Affymetrix Gene
Data obtained with Affymetrix gene expression arrays are widely used to characterize genes, classify disease tissues, classify normal tissues and characterize the response of cells or tissues to drugs and toxins. The end result of analysis of this data is typically a list genes that characterize a biological state or states. In my talk, I will compare and evaluate standard quantitation techniques and will show that the uncertainty introduced by the quantitation method alone is comparable to the total uncertainty introduced by processing of samples through all the wet lab techniques combined (tissue procurement, RNA extraction, RNA labeling, hybridization, chip variability). These comparisons will be made both qualitatively, by use of tissue ratio plots and, quantitatively, via t-test overlap comparisons and measures of the minimum detectable fold change difference increase.
Mechanistic Modeling, Pharmacokinetics, and Pharmacodynamics of Therapeutic Monoclonal Antibodies
Rituxan (Rituximab / IDEC-C2B8 / MabThera) is a therapeutic anti-CD20 monoclonal antibody (mAb) used to treat certain types of B-cell lymphomas and leukemias. Rituxan has a high binding specificity for the cell surface receptor CD20, which is expressed exclusively on B-cells from the pro-B-cell stage through the mature B-cell stage and is thought to play a role in B-cell development. Three putative mechanisms of action by which Rituxan eliminates B-lymphocytes are 1) antibody-dependent cellular cytotoxicity (ADCC), 2) complement-mediated cytotoxicity (CDC), and 3) apoptosis. The primary goal of this work is to elucidate and establish the relative importance of these mechanisms in vivo. To accomplish this, we first propose a new mathematical model for the pharmacokinetics (PK) and pharmacodynamics (PD) of Rituxan and other monoclonal antibodies in mice and humans. An augmentation of a previously published physiologically-based pharmacokinetic (PBPK) model of monoclonal antibody biodistribution is proposed, accounting for varying tumor mass and FcRn-mediated recycling of immunoglobulin G (IgG) through capillary endothelial cells. The complete PK/PD model can be used in a clinical setting to design and optimize mAb infusion schedules for patients and provide insight into the type of antibody best suited for treatment (complete antibody, Fab, F(ab’)2, Fv, diabody, or minibody), as well as providing insight into in vivo mechanisms of action.
Model Discrimination Analysis of GLUT4 Translocation Mechanisms and Sorting
Facilitative glucose transporter translocation from intracellular compartments to the plasma membrane is a key process cells undergo in response to insulin hormone stimulation. The details of the mechanisms involved in this movement and methods of intracellular sorting, however have not yet been fully elucidated. In this study, I construct three compartmental models depicting three different modes for sorting and translocating glucose transporters. Using experimental data from gold particle endosomal markers (S. Martin et al 2000), I will attempt to find translocation rates between compartments using SAAMII (a program for kinetic analysis) and eventually propose justifications for choosing one model over the other two, or provide improvements for an alternate mechanistic model. I will present my progress in this presentation.
Modeling Hematopoiesis in Chronic Myeloid Leukemia
Chronic myeloid leukemia (CML) is a myeloproliferative disorder characterized by the increased production and accumulation of various types of blood cells. Neutrophils show the greatest increase in numbers, although cells of other lineages are affected to a lesser extent. The disease consists of three phases: a chronic phase, an accelerated phase, and blast crisis. Mortality during blast crisis is very high, with a median survival time of 3-6 months after onset. Recently a new drug, Gleevec, has been introduced and proven effective for the treatment of CML during the chronic phase. The drug has proven less effective, however, during blast crisis due to the presence of cells resistant to the drug. With the goal of predicting the dynamics underlying how these mutants arise, a math model of normal and chronic phase CML hematopoiesis will be presented, as well as initial efforts to model blast crisis.
A Mathematical Model of Insulin Receptor Activation, Deactivation, and Recycling in Hepatocytes
Insulin is the prominent hormone for glucose uptake and hence, cellular maintenance. Upon binding to its cell surface receptor, insulin instigates a signaling transduction pathway through which glucose transporters are translocated to the plasma membrane from an intracellular pool. If either insulin or its intact receptor is unavailable at the membrane, glucose uptake will not occur and the cell will suffer, resulting in a deficiency called insulin resistance. Mathematical models of insulin receptor binding kinetics and insulin receptor recycling have been elucidated, but the mechanism of receptor activation and deactivation has yet to be simulated. While insulin binding and subsequent receptor activation occur at the plasma membrane, deactivation occurs at the endoplasmic reticulum. The deactivated receptors are either transcytosed to the membrane for further activation, or are sent to lysosomes for degradation. Our complete model links the mechanistic processes of insulin receptor activation, deactivation, and recycling. The model gives a more thorough portrait of the life cycle of the insulin receptor as well as the interrelationships between insulin signaling and insulin resistance, becoming a more useful tool in biology and medicine.
Induced Secretion of TSH from the Anterior Pituitary Gland in Humans
Ligand binding with
G-protein coupled receptors and subsequent downstream messaging is a ubiquitous
method of cell communication. This mechanism is manifest in the anterior pituitary
(AP) gland, one important example being the signaling pair: thyrotropin releasing
hormone (TRH), the ligand, and its G-protein coupled receptor, TRH-R. The primary
effect of this TRH binding on the feedback regulatory hypothalamo-pituitary-thyroid
axis is to induce secretion of thyrotropin (TSH) from the AP into the blood
circulation. Although G-protein receptors have been modeled somewhat extensively,
to this date no model has been recovered mathematically representing protein
secretion, let alone the upstream events stimulating release. My presentation
will cover work done this quarter elucidating these mechanisms and developing
mathematical relationships between TRH input stimulation followed downstream
by output secretion of TSH from the human pituitary. We intend to represent
this input-output mechanism mathematically at a mechanistic level, with enough
complexity to capture its essential dynamics in health and common disease states,
rendering it “as simple as possible, but not too simple” (alla A. Einstein).
RETURN
TO TOP
I examine different hypotheses for GLUT4 activation and translocation in response to the insulin cascade, with the goal of integrting all aspects of these hypotheses into a single, testable model.
Molecular Mechanism of CD20 Mediated Apoptosis in B Lymphocytes
Crosslinking of CD20, a cell surface molecule, by monoclonal antibodies has been shown to activate intracellular machinery that results in cell death. The death signal pathway, as currently understood in the literature, will be presented along with a mathematical model describing the initial steps in the pathway. CD20 is a target for therapeutic monoclonal antibodies (Rituxan / Rituximab / IDEC-C2B8) used to treat lymphoma and leukemia.
Cell Dynamics Modeling of Drug-Resistance in CML Leukemia
Current therapy for patients with chronic myeloid leukemia (CML) involves administration of Gleevec, a novel inhibitor of the Bcr-Abl tyrosine kinase responsible for CML. Though extremely effective, patients on Gleevec eventually develop resistance to the drug; mutations in Bcr-Abl arise that do respond to treatment. Thus, it is important to understand how Gleevec affects the growth of drug-resistant cells. To this end, a mathematical model describing the growth of normal blood cells, the growth of blood cells with drug sensitive Bcr-Abl, and the growth of blood cells with drug resistant Bcr-Abl will be presented. Preliminary work to describe the effect of Gleevec on the growth dynamics of these cells will also be presented.
A Mathematical Model of HCV Infection and IFN anti-viral Action
Hepatitis C is the most common of the recognized chronic hepatitis viruses in the United States. Interferon monotherapy is currently the accepted treatment, however it achieves sustained viral clearance in only 15-20% of patients and its exact mechanism of action is unknown. I will present a mathematical model derived from a hypothesis on the HCV entry site and the mechanism of IFN's anti-HCV action.
Overview of parameter estimation algorithms for dynamic systems, including a new hybrid global search algorithm
The problem of parameter estimation is common to various scientific and engineering fields from which different strategies for solving this problem have been developed. Four general methods are reviewed: 1) direct single-shooting approach; 2) multiple-shooting approach; 3) joint state-parameter estimator; 4) a category of continuous-time identification methods. The optimization problem resulting from the above parameter estimation methods is a difficult one with a complicated "fitness landscape" and justifies use of global optimization techniques. How to complement a simulated annealing algorithm and a hyper-cone random search algorithm is discussed with numerical results on some standard test problems.
Mechanisms of Insulin Resistance: A Model of Functional and Defective Insulin Receptor Binding Kinetics
The binding of plasma insulin to its receptor on the plasma membrane is the primary step in a signaling cascade that culminates with uptake of glucose into the cell. Depending on effectiveness of the receptor, we may be able to predict the effectiveness of the insulin signal in activating GLUT4 transporters and translocating them from the intracellular pool to the plasma membrane. I will discuss alternative models of receptor binding states and how defects can alter the effectiveness of the signaling cascade .
An Overview of Methods for Continuous-Time Dynamic Model Identification from Experimental Signals
Abstract: The problem of identifying (estimating) unknown parameters in ODE and differential algebraic equations (DAEs) given only discrete measurements of the output is a difficult one compared to discrete-time system model identification due to the presence of the differential terms. Various methods to tackle the problem have been proposed and established in different research fields. In this talk, four types of methods of different origins will be discussed including their basic theories, scopes of application and limitations. The 4 methods are: a direction single-shooting approach, a multiple-shooting approach, a joint-state-parameter estimator and a category of continuous-time identification methods. A numerical example is presented at the end for illustrating the major methods in action.