Workshop 6: Targeting Cancer Cell Proliferation and Metabolism Networks

(March 23,2015 - March 27,2015 )

Organizers


Baltazar Aguda
Founder & CEO, Disease Pathways, LLC
Robert Gatenby
H. Lee Moffitt Cancer Center & Research Institute
Vito Quaranta
Department of Cancer Biology, Vanderbilt University
Santiago Schnell
Brehm Center for Diabetes Research, University of Michigan Medical School

A fundamental property of cancer is uncontrolled cell proliferation. Much knowledge has accumulated on altered genetic and signaling networks that drive uncontrolled proliferation. Recently, there has been a resurgence of interest in the intimate link between proliferation and metabolism, absolutely required to fulfill energy and biomass demands for cell division. In cancer, metabolic networks are highly adaptable, and often metabolism of cancer cells relies largely on aerobic glycolysis, a property referred to as the Warburg effect and akin to fermentation: even in the presence of oxygen, energy metabolism bypasses mitochondrial respiration. The dysregulated interface between metabolic networks and oncogene-modified proliferation networks is emerging as a fertile area to identify critical target nodes, or strategies to defy the drive to ever-adaptable uncontrolled proliferation. This workshop will encompass a mix of experimentalists and mathematicians. Ideally, the former will be engaged on the production of large datasets on cancer cell proliferation, both at the cell population and single-cell level, and in response to microenvironment perturbations including anti-proliferative drugs. The latter will focus on mathematical models of proliferation and metabolism at several scales, including genetic, signaling and cellular, including a focus on the ability of cancer cell populations to regenerate and reprogram in response to hostile microenvironment and to targeted treatment, ultimately persisting in their proliferative state. Multi-scale models connecting the growth of cultured cancer cells and/or individual tumors to epidemiological data will also be considered. Although tumor growth and cancer cell proliferation have been modeled mathematically for decades, adequate datasets have been scarce and fragmentary due to experimental limitations. Recently, several game-changing high-throughput technologies, including genomics, proteomics, and automated microscopy, have created remarkable opportunities for renewed modeling efforts. Furthermore, small-molecule drugs with exquisite specificity for signaling network nodes are in an intensive phase of development and deployment into clinical trials. As these targeted agents increasingly enter standard clinical practice, a major challenge is to improve outcomes by rational drugging strategies. Sheer combinatorics makes drug strategy testing in the field prohibitively expensive, both financially and temporally, opening avenues for mathematical and statistical approaches that, combined with experimentation, have the power to streamline testing.

Accepted Speakers

Lilia Alberghina
Biotecnologie Bioscienze, University Milano Bicocca
Brian Altman
Abramson Family Cancer Research Institute, University of Pennsylvania Perelman School of Medicine
Steven Altshuler
Helen Byrne
Centre for Mathematical Medicine and Biology, University of Nottingham
Riccardo Colombo
Department of Informatics, Systems and Communication, University of Milan - Bicocca
Olivier Elemento
Ido Goldstein
Lab of Receptor Biology and Gene Expression, National Cancer Institute
Sui Huang
Cell and Molecular Biology, Northwestern University Medical School
Peng Huang
Paul Hwang
Center for Molecular Medicine, NHLBI-NIH
Rainer König
Radhakrishnan Mahadevan
Sofia Merajver
Internal Medicine, Cancer and Molecular Biology Program, University of Michigan
Rafael Moreno-Sanchez
Biochemistry, Instituto Nacional de Cardiologia and Universidad Nacional Autonoma de Mexico
Jacques Pouysségur
Institute for Research on Cancer & Aging , nice (IRCAN),
Jeffrey Rathmell
Pharmacology and Cancer Biology, Duke University
R. Brooks Robey
Research & Development Service, Veterans Afaairs Medical Center
Rosalie Sears
Molecular and Medical Genetics, Oregon Health and Science University
Siv Sivaloganathan
Applied Mathematics, University of Waterloo
Lani Wu
Thomas Yankeelov
Radiology, Vanderbilt University
Monday, March 23, 2015
Time Session
08:00 AM

Shuttle to MBI

08:15 AM
08:45 AM

Breakfast

08:45 AM
09:00 AM

Introductions and Information - Marty Golubitsky

09:00 AM
09:45 AM
Baltz Aguda
09:45 AM
10:30 AM
Paul Hwang - p53 Regulation of Mitochondria

p53, one of the most commonly mutated tumor suppressor genes in human cancers, promotes oxidative metabolism by regulating mitochondrial biogenesis through various mechanisms. These include the transactivation of mitochondrial biogenesis genes in the nucleus and the maintenance of mitochondrial genomic DNA by the translocation of p53 protein into the mitochondria. While oxygen is essential for oxidative metabolism, it also serves as the essential substrate for reactive oxygen species (ROS) that can damage the genome. Given the well-established role of p53 in maintaining genomic stability, its promotion of respiration that efficiently converts reactive oxygen to water may serve to contribute to the antioxidant activities of p53. Interestingly, one physiological impact of p53 promoting mitochondrial respiration is increased aerobic exercise capacity that parallels the strong inverse relationship between cardio-respiratory fitness and cancer-free survival observed in large epidemiologic studies. On the other hand, our recent observations from mouse and human studies of Li-Fraumeni syndrome, a premature cancer condition caused by germline mutations of p53, indicate that mutant p53 can also increase oxidative metabolism. We suggest that these disparate findings can be reconciled by the dissociation of the cell cycle and mitochondrial activities of p53. Our observations are also consistent with the growing evidence that cancer cells depend not only on aerobic glycolysis (Warburg effect) but also on mitochondrial metabolism which may contribute to tumorigenesis in the setting of defective cell cycle regulation by mutated p53.

10:30 AM
11:00 AM

Break

11:00 AM
11:45 AM
Riccardo Colombo - Modeling metabolic networks with an ensemble evolutionary flux balance analysis approach

Metabolism can be seen as the biochemical factory producing building blocks and energy for cellular functioning. The study of its control mechanisms and dysregulation is today an area of intense application of modeling efforts. Due to the complexity and dimension of metabolic networks, mechanistic modeling of their dynamics becomes unfeasible. For this reason, the computational investigation of metabolic models makes typically use of constraint-based approaches, which exploits the knowledge about the structure of cell metabolism, while disregarding dynamic intracellular behavior, on the basis of a pseudo-steady state assumption.

Assuming also that cell behavior is optimal with respect to a “metabolic objective”, flux balance analysis (FBA) is widely used to calculate a single optimal flux distribution. This approach has proven to be effective in implementing metabolic engineering design goals, such as the maximization of the cell production of metabolites of industrial interest. However, FBA has recently received increasing attention in Systems Biology, to gain novel knowledge about the physiological state of a cell. In this regard, the assumption of maximization of biomass yield as objective function has revealed successful in predicting some phenotypical characteristics of microorganisms. Nevertheless, when dealing with multicellular organisms, the definition of a plausible objective function is not straightforward and, besides, even if we knew the true objective function, we could not still exclude that the systems is in a sub-optimal space. For this reason, new approaches aimed at describing global network properties are emerging for an unbiased analysis of the solutions space. In this context, we proposed an extension of the classic constraint-based modeling approach in order to overcome the problem of defining an objective function and to analyze the space of fluxes distribution using clustering techniques.

In particular, the developed method does not focus on a given flux distribution assuming that it corresponds to the real one, but is rather designed to explore the solution space sampling it by means of several distinct random objective functions and looking for an ensemble solutions that match a given metabolic phenotype (the system expected behavior). The identified ensemble may indeed be capable of greater prediction accuracy than any of their individual members. By analyzing the generic properties of the ensemble of matching solutions we may identify some patterns responsible for the emergent phenotype. If it is the case that a metabolic phenotype may result form alternative sub-phenotypes. By finding clusters of similar solutions and analyzing their common/distinguishing properties we may therefore gain information about such possible sub-phenotypes. The approach reveals even more effective when two metabolic responses, let say for example a physiological against a pathological one, are compared. In this case, two different ensembles will therefore be obtained: one that matches the former conditions and the other that matches the latter. By comparing the generic properties of the two ensembles the pathways mainly involved in the differential response will be disentangled. It goes without saying that the expected behavior must be abstracted and formalized, and we propose to express the definition in terms of a metabolic response to a condition variation (e.g. redistribution of fluxes as a consequence of a variation in nutrient availability). Depending on the specific problem, the match between a solution and the metabolic response definition can be Boolean (the condition is either met or not) or expressed in term of a fitness (how close the condition is) exploiting an evolutionary algorithm.

We tested our method on a yeast core metabolic model, from which we identified two ensembles of solutions, the first in agreement with a definition for “Crabtree-positive yeasts” and the second in agreement with a definition for “Crabtree-negative yeasts”. In a further step we showed how the ensembles can be further characterized and refined by means of a cluster analysis in order to identify sub-phenotypes. Finally the fluxes that significantly differ between the two ensembles have been identified according to a Kolmogorov-Smirnov test. The aim now, in SysBio – Italy, is to apply this method to the experimentally validated network structures of cancer metabolic rewiring.

11:45 AM
12:30 PM
Rosalie Sears
12:30 PM
02:00 PM

Lunch Break

02:00 PM
02:45 PM
Ido Goldstein - Massive re-organization of the liver chromatin landscape following metabolic and inflammatory signals

Diabetes and chronic inflammation are major risk factors for developing various cancer types, including liver cancer. We set out to study the chromatin and transcriptional changes in liver following metabolic and inflammatory signals that are associated with a cancer-promoting phenotype.

One of the hallmarks of diabetes is a hyper-activated, de-regulated response to fasting orchestrated mainly by the liver. In non-pathological states, this response is achieved by eliciting a comprehensive and elaborate transcriptional program leading to fuel production in the form of glucose and ketone bodies. These changes in transcription are mediated by alterations in chromatin structure and transcription factor occupancy. To evaluate the alterations in chromatin landscape and gene expression following fasting, we analyzed livers from fasted mice in three high-throughput experiments. We employed the DNase I Hyper-Sensitivity assay followed by sequencing (DHS-seq) to globally map the accessible regions of chromatin; thus detecting transcriptional regulatory nodes in chromatin (mainly promoters and enhancers). We have found ~4,000 sites in liver chromatin in which accessibility was altered following fasting. These altered regulatory regions were enriched in binding sites for transcription factors known to regulate the fasting response. We obtained similar results by globally mapping active enhancers (by chromatin immunoprecipitation of the active enhancer mark H3K27Ac followed by sequencing – ChIP-seq). Moreover, we determined the alterations in gene expression by profiling the transcriptome of those livers using RNA-seq, with the aim of linking between alterations in chromatin landscape and gene expression. Regions with increased accessibility and increased H3K27 acetylation following fasting (suggesting active transcriptional regulation) were evidenced proximally to fasting induced genes. We present surprising findings showing that a very common metabolic perturbation (i.e. fasting) leads to a massive re-organization of liver chromatin which supports the onset of a complex, temporally organized, mutli-stage transcriptional response.

In a parallel effort, we characterized the chromatin and gene expression patterns of primary hepatocytes in response to three major pro-inflammatory cytokines associated with hepato-carcinogenesis (IL-6, IL-1 beta and TNF alpha). This analysis revealed a complex and diverse pattern of chromatin and gene regulation in which inflammatory cytokines play complementing roles whereby a certain cytokine ‘primes’ an enhancer while the second cytokine induces gene expression.

We believe that understanding the transcriptional response and its underlining chromatin regulation following cancer-promoting stimuli is critical to elucidating the mechanisms initiating carcinogenesis and is expected to aid in combating cancer as well as interconnected metabolic disorders

02:45 PM
03:30 PM
Sui Huang
03:30 PM
04:15 PM

Break

04:15 PM
05:45 PM

Poster Session

05:45 PM
07:00 PM

Reception

07:15 PM

Shuttle pick-up from MBI

Tuesday, March 24, 2015
Time Session
08:00 AM

Shuttle to MBI

08:15 AM
09:00 AM

Breakfast

09:00 AM
09:45 AM
Vito Quaranta
09:45 AM
10:30 AM
Eytan Ruppin
10:30 AM
11:00 AM

Break

11:00 AM
11:45 AM
Rainer König
11:45 AM
12:30 PM
Lilia Alberghina - Cancer metabolic rewiring in cancer cells

The investigation on metabolic profiling of normal and cancer cells is recently gaining more interest in molecular oncology due to the understanding that a metabolic rewiring underlies the ability of uncontrolled proliferation of cancer cells. It is not only the well-known Warburg effect, but also an increased utilization of glutamine by reductive carboxylation that takes place in cancer cells.

We have developed transcriptional and metabolic analysis in various cancer cell lines, to reach a more detailed definition of the metabolic steps involved in cancer metabolic rewiring and on its redox regulation. These data will be discussed in the logic of a drug discovery attempt. The construction of mathematical models of the indicated cancer metabolic rewiring is in progress.

12:30 PM
02:00 PM

Lunch Break

02:00 PM
02:45 PM
Olivier Elemento
02:45 PM
03:30 PM
Lani Wu
03:30 PM
04:15 PM

Break

04:15 PM
05:45 PM

Poster Chalk Talks

05:45 PM

Shuttle pick-up from MBI

Wednesday, March 25, 2015
Time Session
08:00 AM

Shuttle to MBI

08:15 AM
09:00 AM

Breakfast

09:00 AM
09:45 AM
Peng Huang
09:45 AM
10:30 AM
Siv Sivaloganathan
10:30 AM
11:00 AM

Break

11:00 AM
11:45 AM
Brian Altman
11:45 AM
12:30 PM
Rafael Moreno-Sanchez - Kinetic modeling of cancer glycolysis

Glycolysis provides cytosolic ATP and NADH as well as precursors for several anabolic pathways. These are probably the reasons why most cancer cells have an enhanced glycolytic capacity. To have a better understanding of the controlling mechanisms of this essential pathway, and to unveil suitable and alternative therapeutic targets, kinetic models were built up for glycolysis in cancer cells under a variety of experimental conditions. The kinetic models were constructed using : (i) all the kinetic parameters of all enzymes and transporters involved in the pathway, all determined under near-physiological conditions of pH, temperature and medium composition; (ii) the enzyme activities, metabolite concentrations and fluxes determined in living cells; (iii) the description of appropriate rate equations for all pathway steps, including reversibility or equilibrium constants; and (iv) an iterative process of re-experimentation and refinement of the kinetic models. Once the models were validated, by comparing the model predicted pathway behavior regarding intermediary concentrations and fluxes with that determined experimentally in living cells, they were further used to establish the pathway control.

In cancer cells, kinetic modeling indicated that glucose transporter (GLUT), hexokinase (HK), glycogen degradation (GlycDeg) and/or hexosephosphate isomerase (HPI) were the main flux- and ATP concentration-controlling steps. Although the specific contribution of each of these steps varied, depending on the O2 level (normoxia, severe hypoxia), initial external glucose concentration (25 mM, 2.5 mM), or cancer cell type (AS-30D, HeLa), the same steps kept the pathway control, i.e. the mechanisms that govern the control of cancer glycolysis are preserved and are highly robust. The glycolytic flux increased under low glucose (+ normoxia), which was accompanied by no significant variation in total GLUT and HK activities; instead an increased affinity for glucose emerged as a consequence of a shift in activity from low to high affinity isoforms (GLUT-3 over GLUT-1; and HK-I over HK-II). Modeling appointed GLUT as the principal controlling step in HeLa cells exposed to low glucose; indeed, glycolysis in these cells were more sensitive to GLUT inhibition than cells exposed to high glucose. The glycolytic flux also increased under hypoxia (+ high glucose), but in this case most glycolytic proteins were over-expressed, including the low affinity GLUT and HK isoforms, and with no variation in the high affinity isoforms. Thus, pathway flux can be modulated by changing the isoform pattern (low glucose) and over-expressing (hypoxia) the most controlling steps.

Kinetic modeling identified GLUT, HK, GlycDeg and/or HPI as the foremost therapeutic targets; their simultaneous and partial, not complete, inhibition will have greater deleterious effects on cancer glycolysis, and possibly on cell growth and viability.

This work was partly supported by CONACyT-Mexico grants Nos. 180322, 107183, and 178638.

12:30 PM
02:00 PM

Lunch Break

02:00 PM
02:45 PM
Radhakrishnan Mahadevan
02:45 PM
03:30 PM
Helen Byrne
03:30 PM
04:15 PM

Break

04:15 PM
05:45 PM

Informal Discussion

05:45 PM

Shuttle pick-up from MBI

Thursday, March 26, 2015
Time Session
08:00 AM

Shuttle to MBI

08:15 AM
09:00 AM

Breakfast

09:00 AM
09:45 AM
Steven Altshuler - Relating cancer cell heterogeneity to drug resistance

Cancer cell populations can be highly heterogeneous. We will discuss progress in understanding the origins of this heterogeneity and its implications in predicting drug response.

09:45 AM
10:30 AM
R. Brooks Robey
10:30 AM
11:00 AM

Break

11:00 AM
11:45 AM
Jacques Pouysségur - Targeting pHi control and Bioenergetics in rapidly growing Hypoxic Tumours

Intense interest in the “Warburg effect” has been revived by the discovery that hypoxia-inducible factor 1 (HIF1) re-programs pyruvate oxidation to lactic acid conversion. The most aggressive and invasive cancers, which are often hypoxic, rely on exacerbated glycolysis to meet their increased ATP and biosynthetic precursors demands and on robust pHi regulating systems to combat excessive generation of lactic and carbonic acids.

We and others hypothesized (Pouyssgur et al., Nature, 2006, 441:437) that disruption of the key pHi regulating systems Na+-H+ Exchangers (NHE), Na-dependent Bicarbonate Transporters (NBTs), Carbonic Anhydrases CAIX, XII (CAs), MonoCarboxylate Transporters (MCT1/4) might offer a H+-mediated killing strategy to eradicate fast growing tumours. This approach has demonstrated efficient tumor growth arrest but a rather limited success as far as tumor cell killing is considered. Here, we will discuss the limitations of this strategy and present the concept of ‘metabolic catastrophe’ as an alternative way to kill tumor cells as recently discussed by Parks et al. in Nature Reviews Cancer, 2013, 13, 611.

We will illustrate, in colon adenocarcinoma and glioblastoma cell lines (LS174 and U87), that blocking lactic acid export by genetic disruption of MCT1 and MCT4 or of basigin/MCTs complexes does compromise tumor growth but not viability since tumor cells rapidly re-activate OXPHOS. However we will demonstrate that blocking lactic acid export provides an efficient anticancer approach when combined with phenformin, a mitochondrial complex I inhibitor (Marchiq I. et al., Cancer Research, 2014, Nov.)

11:45 AM
12:30 PM
Jeffrey Rathmell - The Metabolism of Proliferating and Leukemic Lymphocytes

Abstract not submitted.

12:30 PM
02:00 PM

Lunch Break

02:00 PM
02:45 PM
Thomas Yankeelov - Quantitative Imaging to Drive Biophysical Models of Tumor Growth

The ability to identify—early in the course of therapy—patients that are not responding to a given therapeutic regimen is highly significant. In addition to limiting patients’ exposure to the toxicities associated with unsuccessful therapies, it would allow patients the opportunity to switch to a potentially more efficacious treatment. In this presentation, we will discuss ongoing efforts at using data available from advanced imaging technologies to initialize and constrain predictive biophysical and biomathematical models of tumor growth and treatment response.

02:45 PM
03:30 PM
Sofia Merajver
03:30 PM
04:15 PM

Break and Informal Discussions

04:15 PM

Shuttle pick-up from MBI

05:30 PM
06:00 PM

Cash Bar - Crowne Plaza

06:00 PM
06:00 PM

Banquet - Crowne Plaza

Friday, March 27, 2015
Time Session
08:00 AM

Shuttle to MBI

08:15 AM
09:00 AM

Breakfast

09:00 AM
09:45 AM
Santiago Schnell
09:45 AM
10:30 AM

TBD

10:30 AM
11:00 AM

Break

11:00 AM
12:15 PM

Panel Discussion

12:15 PM

Shuttle pick-up from MBI (One to airport and one back to hotel)

Name Email Affiliation
Aguda, Baltazar bdaguda@gmail.com Founder & CEO, Disease Pathways, LLC
Alberghina, Lilia lilia.alberghina@unimib.it Biotecnologie Bioscienze, University Milano Bicocca
Altman, Brian altman@mail.med.upenn.edu Abramson Family Cancer Research Institute, University of Pennsylvania Perelman School of Medicine
Altshuler, Steven steven.altschuler@utsouthwestern.edu
Byrne, Helen byrneh@maths.ox.ac.uk Centre for Mathematical Medicine and Biology, University of Nottingham
Colombo, Riccardo riccardo.colombo@disco.unimib.it Department of Informatics, Systems and Communication, University of Milan - Bicocca
Elemento, Olivier ole2001@med.cornell.edu
Fallahi-Sichani, Mohammad Mohammad_FallahiSichani@hms.harvard.edu
Gaglio, Daniela daniela.gaglio@ibfm.cnr.it
Goldstein, Ido goldstein.ido@gmail.com Lab of Receptor Biology and Gene Expression, National Cancer Institute
Huang, Peng phuang@mdanderson.org
Huang, Sui sui.huang@systemsbiology.org Cell and Molecular Biology, Northwestern University Medical School
Hwang, Paul hwangp@mail.nih.gov Center for Molecular Medicine, NHLBI-NIH
Knig, Rainer r.koenig@dkfz.de
Leander, Rachel Rachel.Leander@mtsu.edu Mathematics, Middle Tennessee State University
Mahadevan, Radhakrishnan mahadevan@chem-eng.utoronto.ca
Merajver, Sofia smerajve@umich.edu Internal Medicine, Cancer and Molecular Biology Program, University of Michigan
Moreno-Sanchez, Rafael rafael.moreno@cardiologia.org.mx Biochemistry, Instituto Nacional de Cardiologia and Universidad Nacional Autonoma de Mexico
Pouyssegur, Jacques pouysseg@unice.fr Institute for Research on Cancer & Aging , nice (IRCAN),
Quaranta, Vito vito.quaranta@vanderbilt.edu Department of Cancer Biology, Vanderbilt University
Rathmell, Jeffrey jeff.rathmell@duke.edu Pharmacology and Cancer Biology, Duke University
Reznik, Ed reznik.ed@gmail.com Computational Biology, Memorial Sloan Kettering Cancer Center
Robey, R. Brooks R.Brooks.Robey@Dartmouth.edu Research & Development Service, Veterans Afaairs Medical Center
Schnell, Santiago schnells@umich.edu Brehm Center for Diabetes Research, University of Michigan Medical School
Sears, Rosalie searsr@ohsu.edu Molecular and Medical Genetics, Oregon Health and Science University
Sivaloganathan, Siv ssivalog@math.uwaterloo.ca Applied Mathematics, University of Waterloo
Wu, Lani lani.wu@utsouthwestern.edu
Wynn, Michelle mlwynn@umich.edu Molecular & Integrative Physiology, University of Michigan
Yankeelov, Thomas thomas.yankeelov@vanderbilt.edu Radiology, Vanderbilt University
Cancer metabolic rewiring in cancer cells

The investigation on metabolic profiling of normal and cancer cells is recently gaining more interest in molecular oncology due to the understanding that a metabolic rewiring underlies the ability of uncontrolled proliferation of cancer cells. It is not only the well-known Warburg effect, but also an increased utilization of glutamine by reductive carboxylation that takes place in cancer cells.

We have developed transcriptional and metabolic analysis in various cancer cell lines, to reach a more detailed definition of the metabolic steps involved in cancer metabolic rewiring and on its redox regulation. These data will be discussed in the logic of a drug discovery attempt. The construction of mathematical models of the indicated cancer metabolic rewiring is in progress.

Relating cancer cell heterogeneity to drug resistance

Cancer cell populations can be highly heterogeneous. We will discuss progress in understanding the origins of this heterogeneity and its implications in predicting drug response.

Modeling metabolic networks with an ensemble evolutionary flux balance analysis approach

Metabolism can be seen as the biochemical factory producing building blocks and energy for cellular functioning. The study of its control mechanisms and dysregulation is today an area of intense application of modeling efforts. Due to the complexity and dimension of metabolic networks, mechanistic modeling of their dynamics becomes unfeasible. For this reason, the computational investigation of metabolic models makes typically use of constraint-based approaches, which exploits the knowledge about the structure of cell metabolism, while disregarding dynamic intracellular behavior, on the basis of a pseudo-steady state assumption.

Assuming also that cell behavior is optimal with respect to a “metabolic objective”, flux balance analysis (FBA) is widely used to calculate a single optimal flux distribution. This approach has proven to be effective in implementing metabolic engineering design goals, such as the maximization of the cell production of metabolites of industrial interest. However, FBA has recently received increasing attention in Systems Biology, to gain novel knowledge about the physiological state of a cell. In this regard, the assumption of maximization of biomass yield as objective function has revealed successful in predicting some phenotypical characteristics of microorganisms. Nevertheless, when dealing with multicellular organisms, the definition of a plausible objective function is not straightforward and, besides, even if we knew the true objective function, we could not still exclude that the systems is in a sub-optimal space. For this reason, new approaches aimed at describing global network properties are emerging for an unbiased analysis of the solutions space. In this context, we proposed an extension of the classic constraint-based modeling approach in order to overcome the problem of defining an objective function and to analyze the space of fluxes distribution using clustering techniques.

In particular, the developed method does not focus on a given flux distribution assuming that it corresponds to the real one, but is rather designed to explore the solution space sampling it by means of several distinct random objective functions and looking for an ensemble solutions that match a given metabolic phenotype (the system expected behavior). The identified ensemble may indeed be capable of greater prediction accuracy than any of their individual members. By analyzing the generic properties of the ensemble of matching solutions we may identify some patterns responsible for the emergent phenotype. If it is the case that a metabolic phenotype may result form alternative sub-phenotypes. By finding clusters of similar solutions and analyzing their common/distinguishing properties we may therefore gain information about such possible sub-phenotypes. The approach reveals even more effective when two metabolic responses, let say for example a physiological against a pathological one, are compared. In this case, two different ensembles will therefore be obtained: one that matches the former conditions and the other that matches the latter. By comparing the generic properties of the two ensembles the pathways mainly involved in the differential response will be disentangled. It goes without saying that the expected behavior must be abstracted and formalized, and we propose to express the definition in terms of a metabolic response to a condition variation (e.g. redistribution of fluxes as a consequence of a variation in nutrient availability). Depending on the specific problem, the match between a solution and the metabolic response definition can be Boolean (the condition is either met or not) or expressed in term of a fitness (how close the condition is) exploiting an evolutionary algorithm.

We tested our method on a yeast core metabolic model, from which we identified two ensembles of solutions, the first in agreement with a definition for “Crabtree-positive yeasts” and the second in agreement with a definition for “Crabtree-negative yeasts”. In a further step we showed how the ensembles can be further characterized and refined by means of a cluster analysis in order to identify sub-phenotypes. Finally the fluxes that significantly differ between the two ensembles have been identified according to a Kolmogorov-Smirnov test. The aim now, in SysBio – Italy, is to apply this method to the experimentally validated network structures of cancer metabolic rewiring.

Massive re-organization of the liver chromatin landscape following metabolic and inflammatory signals

Diabetes and chronic inflammation are major risk factors for developing various cancer types, including liver cancer. We set out to study the chromatin and transcriptional changes in liver following metabolic and inflammatory signals that are associated with a cancer-promoting phenotype.

One of the hallmarks of diabetes is a hyper-activated, de-regulated response to fasting orchestrated mainly by the liver. In non-pathological states, this response is achieved by eliciting a comprehensive and elaborate transcriptional program leading to fuel production in the form of glucose and ketone bodies. These changes in transcription are mediated by alterations in chromatin structure and transcription factor occupancy. To evaluate the alterations in chromatin landscape and gene expression following fasting, we analyzed livers from fasted mice in three high-throughput experiments. We employed the DNase I Hyper-Sensitivity assay followed by sequencing (DHS-seq) to globally map the accessible regions of chromatin; thus detecting transcriptional regulatory nodes in chromatin (mainly promoters and enhancers). We have found ~4,000 sites in liver chromatin in which accessibility was altered following fasting. These altered regulatory regions were enriched in binding sites for transcription factors known to regulate the fasting response. We obtained similar results by globally mapping active enhancers (by chromatin immunoprecipitation of the active enhancer mark H3K27Ac followed by sequencing – ChIP-seq). Moreover, we determined the alterations in gene expression by profiling the transcriptome of those livers using RNA-seq, with the aim of linking between alterations in chromatin landscape and gene expression. Regions with increased accessibility and increased H3K27 acetylation following fasting (suggesting active transcriptional regulation) were evidenced proximally to fasting induced genes. We present surprising findings showing that a very common metabolic perturbation (i.e. fasting) leads to a massive re-organization of liver chromatin which supports the onset of a complex, temporally organized, mutli-stage transcriptional response.

In a parallel effort, we characterized the chromatin and gene expression patterns of primary hepatocytes in response to three major pro-inflammatory cytokines associated with hepato-carcinogenesis (IL-6, IL-1 beta and TNF alpha). This analysis revealed a complex and diverse pattern of chromatin and gene regulation in which inflammatory cytokines play complementing roles whereby a certain cytokine ‘primes’ an enhancer while the second cytokine induces gene expression.

We believe that understanding the transcriptional response and its underlining chromatin regulation following cancer-promoting stimuli is critical to elucidating the mechanisms initiating carcinogenesis and is expected to aid in combating cancer as well as interconnected metabolic disorders

p53 Regulation of Mitochondria

p53, one of the most commonly mutated tumor suppressor genes in human cancers, promotes oxidative metabolism by regulating mitochondrial biogenesis through various mechanisms. These include the transactivation of mitochondrial biogenesis genes in the nucleus and the maintenance of mitochondrial genomic DNA by the translocation of p53 protein into the mitochondria. While oxygen is essential for oxidative metabolism, it also serves as the essential substrate for reactive oxygen species (ROS) that can damage the genome. Given the well-established role of p53 in maintaining genomic stability, its promotion of respiration that efficiently converts reactive oxygen to water may serve to contribute to the antioxidant activities of p53. Interestingly, one physiological impact of p53 promoting mitochondrial respiration is increased aerobic exercise capacity that parallels the strong inverse relationship between cardio-respiratory fitness and cancer-free survival observed in large epidemiologic studies. On the other hand, our recent observations from mouse and human studies of Li-Fraumeni syndrome, a premature cancer condition caused by germline mutations of p53, indicate that mutant p53 can also increase oxidative metabolism. We suggest that these disparate findings can be reconciled by the dissociation of the cell cycle and mitochondrial activities of p53. Our observations are also consistent with the growing evidence that cancer cells depend not only on aerobic glycolysis (Warburg effect) but also on mitochondrial metabolism which may contribute to tumorigenesis in the setting of defective cell cycle regulation by mutated p53.

Kinetic modeling of cancer glycolysis

Glycolysis provides cytosolic ATP and NADH as well as precursors for several anabolic pathways. These are probably the reasons why most cancer cells have an enhanced glycolytic capacity. To have a better understanding of the controlling mechanisms of this essential pathway, and to unveil suitable and alternative therapeutic targets, kinetic models were built up for glycolysis in cancer cells under a variety of experimental conditions. The kinetic models were constructed using : (i) all the kinetic parameters of all enzymes and transporters involved in the pathway, all determined under near-physiological conditions of pH, temperature and medium composition; (ii) the enzyme activities, metabolite concentrations and fluxes determined in living cells; (iii) the description of appropriate rate equations for all pathway steps, including reversibility or equilibrium constants; and (iv) an iterative process of re-experimentation and refinement of the kinetic models. Once the models were validated, by comparing the model predicted pathway behavior regarding intermediary concentrations and fluxes with that determined experimentally in living cells, they were further used to establish the pathway control.

In cancer cells, kinetic modeling indicated that glucose transporter (GLUT), hexokinase (HK), glycogen degradation (GlycDeg) and/or hexosephosphate isomerase (HPI) were the main flux- and ATP concentration-controlling steps. Although the specific contribution of each of these steps varied, depending on the O2 level (normoxia, severe hypoxia), initial external glucose concentration (25 mM, 2.5 mM), or cancer cell type (AS-30D, HeLa), the same steps kept the pathway control, i.e. the mechanisms that govern the control of cancer glycolysis are preserved and are highly robust. The glycolytic flux increased under low glucose (+ normoxia), which was accompanied by no significant variation in total GLUT and HK activities; instead an increased affinity for glucose emerged as a consequence of a shift in activity from low to high affinity isoforms (GLUT-3 over GLUT-1; and HK-I over HK-II). Modeling appointed GLUT as the principal controlling step in HeLa cells exposed to low glucose; indeed, glycolysis in these cells were more sensitive to GLUT inhibition than cells exposed to high glucose. The glycolytic flux also increased under hypoxia (+ high glucose), but in this case most glycolytic proteins were over-expressed, including the low affinity GLUT and HK isoforms, and with no variation in the high affinity isoforms. Thus, pathway flux can be modulated by changing the isoform pattern (low glucose) and over-expressing (hypoxia) the most controlling steps.

Kinetic modeling identified GLUT, HK, GlycDeg and/or HPI as the foremost therapeutic targets; their simultaneous and partial, not complete, inhibition will have greater deleterious effects on cancer glycolysis, and possibly on cell growth and viability.

This work was partly supported by CONACyT-Mexico grants Nos. 180322, 107183, and 178638.

Targeting pHi control and Bioenergetics in rapidly growing Hypoxic Tumours

Intense interest in the “Warburg effect” has been revived by the discovery that hypoxia-inducible factor 1 (HIF1) re-programs pyruvate oxidation to lactic acid conversion. The most aggressive and invasive cancers, which are often hypoxic, rely on exacerbated glycolysis to meet their increased ATP and biosynthetic precursors demands and on robust pHi regulating systems to combat excessive generation of lactic and carbonic acids.

We and others hypothesized (Pouysségur et al., Nature, 2006, 441:437) that disruption of the key pHi regulating systems Na+-H+ Exchangers (NHE), Na-dependent Bicarbonate Transporters (NBTs), Carbonic Anhydrases CAIX, XII (CAs), MonoCarboxylate Transporters (MCT1/4) might offer a H+-mediated killing strategy to eradicate fast growing tumours. This approach has demonstrated efficient tumor growth arrest but a rather limited success as far as tumor cell killing is considered. Here, we will discuss the limitations of this strategy and present the concept of ‘metabolic catastrophe’ as an alternative way to kill tumor cells as recently discussed by Parks et al. in Nature Reviews Cancer, 2013, 13, 611.

We will illustrate, in colon adenocarcinoma and glioblastoma cell lines (LS174 and U87), that blocking lactic acid export by genetic disruption of MCT1 and MCT4 or of basigin/MCTs complexes does compromise tumor growth but not viability since tumor cells rapidly re-activate OXPHOS. However we will demonstrate that blocking lactic acid export provides an efficient anticancer approach when combined with phenformin, a mitochondrial complex I inhibitor (Marchiq I. et al., Cancer Research, 2014, Nov.)

The Metabolism of Proliferating and Leukemic Lymphocytes

Abstract not submitted.

Quantitative Imaging to Drive Biophysical Models of Tumor Growth

The ability to identify—early in the course of therapy—patients that are not responding to a given therapeutic regimen is highly significant. In addition to limiting patients’ exposure to the toxicities associated with unsuccessful therapies, it would allow patients the opportunity to switch to a potentially more efficacious treatment. In this presentation, we will discuss ongoing efforts at using data available from advanced imaging technologies to initialize and constrain predictive biophysical and biomathematical models of tumor growth and treatment response.