Go to the About Us section
Go to the Courses section
Go to the Research section
Go to the Specialist Services & Consultancy section
Go to the Outreach section
You can find details about all the staff in the department below.
return to full list
Department of Biological and Medical Sciences
Faculty of Health and Life Sciences
After studying biochemistry at Oxford University, followed by a DPhil on the physical biochemistry of yeast pyruvate kinase, David started lecturing at Oxford Polytechnic. His research gradually moved from experimental biochemistry into computer simulation and theoretical analysis of metabolic control, and he has written the only textbook on metabolic control analysis, Understanding the Control of Metabolism. In 2001, he helped to found and became part-time Chief Scientific Officer of the Oxford company, Physiomics plc, which is using computer simulation of cellular systems for the development and analysis of therapeutic strategies for the pharmaceutical industry.
Physiomics has firmly established itself as a leading light in systems biology approaches to drug discovery and latterly in therapy design, demonstrable through contracts with three major international pharmaceutical companies. Through its strong advocacy of this approach the sector has invested in and adopted new computational biology processes. As Physiomics has continued to grow, it has expanded its own specialist research team, in many cases recruiting scientists trained within Fell’s Brookes-based research group at Brookes.
David is chairman of the Policy Committee of the Biochemical Society, and has been a member of several panels and committees of the Biotechnology and Biological Sciences Research Council.
David’s group formed nearly thirty years ago with initial interests in computer simulation of metabolism and the theory of metabolic control. To these it has since added interests in modelling signal transduction, in various different approaches to network analysis of metabolism, and in reconstructing metabolic networks from genomic data. In the course of this research, he has addressed problems in microbial, plant and mammalian metabolism, often in conjunction with collaborators who have contributed experimental results.
His work forms part of the emerging field of Systems Biology, in that we are concerned with understanding how biological function arises from the interactions between many components, and with building predictive models. Potential applications of our work include the design of changes in cellular metabolism to improve the output of product such as antibiotics, detecting vulnerable sites in cellular networks that could be targets for drugs to control disease-causing organisms, and improved understanding of how organisms manage to adjust their metabolism in response to environmental changes and other signals.
Clostridium autoethanogenum is an industrial microbe used for the commercial-scale production of ethanol from carbon monoxide. While significant progress has been made in the attempted diversification of this bioprocess, further improvements are desirable, particularly in the formation of the high-value platform chemicals, such as 2,3-butanediol. A new, experimentally parameterised genome scale model of C. autoethanogenum predicts dramatically increased 2,3-butanediol production under non-carbon-limited conditions when thermodynamic constraints on hydrogen production are considered.
Analysis of the impact of photorespiration on plant metabolism is usually based on manual inspection of small network diagrams. Here we create a structural metabolic model that contains the reactions that participate in photorespiration in the plastid, peroxisome, mitochondrion and cytosol and the metabolite exchanges between them. This model was subjected to elementary flux modes analysis, a technique that enumerates all the component, minimal pathways of a network. Any feasible photorespiratory metabolism in the plant will be some combination of the elementary flux modes (EFMs) that contain the Rubisco oxygenase reaction. Amongst the EFMs we obtained was the classic photorespiratory cycle, but there were also modes that involve photorespiration coupled with mitochondrial metabolism and ATP production, the glutathione‐ascorbate (GSH‐ASC) cycle and nitrate reduction to ammonia. The modes analysis demonstrated the underlying basis of the metabolic linkages with photorespiration that have been inferred experimentally. The set of reactions common to all the elementary modes showed good agreement with the gene products of mutants that have been reported to have a defective phenotype in photorespiratory conditions. Finally, the set of modes provided a formal demonstration that photorespiration itself does not impact on the CO2:O2 ratio (assimilation quotient, AQ), except in those modes associated with concomitant nitrate reduction.
Previously we have used a genome scale model of rice metabolism to describe how metabolism reconfigures at different light intensities in an expanding leaf of rice. Although this established that the metabolism of the leaf was adequately represented, in the model, the scenario was not that of the typical function of the leaf—to provide material for the rest of the plant. Here we extend our analysis to explore the transition to a source leaf as export of photosynthate increases at the expense of making leaf biomass precursors, again as a function of light intensity. In particular we investigate whether, when the leaf is making a smaller range of compounds for export to the phloem, the same changes occur in the interactions between mitochondrial and chloroplast metabolism as seen in biomass synthesis for growth when light intensity increases. Our results show that the same changes occur qualitatively, though there are slight quantitative differences reflecting differences in the energy and redox requirements for the different metabolic outputs.
Produce rich in phytochemicals may alter postprandial glucose and insulin responses by interacting with the pathways that regulate glucose uptake and insulin secretion in humans. The aims of the present study were to assess the phytochemical constituents of red beetroot juice and to measure the postprandial glucose and insulin responses elicited by either 225 ml beetroot juice (BEET), a control beverage matched for macronutrient content (MCON) or a glucose beverage in healthy adults. Beetroot juice was a particularly rich source of betalain degradation compounds. The orange/yellow pigment neobetanin was measured in particularly high quantities (providing 1·3 g in the 225 ml). A total of sixteen healthy individuals were recruited, and consumed the test meals in a controlled single-blind cross-over design. Results revealed a significant lowering of the postprandial insulin response in the early phase (0–60 min) (P < 0·05) and a significantly lower glucose response in the 0–30 min phase (P < 0·05) in the BEET treatment compared with MCON. Betalains, polyphenols and dietary nitrate found in the beetroot juice may each contribute to the observed differences in the postprandial insulin concentration.
We describe the construction and analysis of a genome-scale metabolic model representing a developing leaf cell of rice (Oryza sativa) primarily derived from the annotations in the RiceCyc database. We used flux balance analysis to determine that the model represents a network capable of producing biomass precursors (amino acids, nucleotides, lipid, starch, cellulose, and lignin) in experimentally reported proportions, using carbon dioxide as the sole carbon source. We then repeated the analysis over a range of photon flux values to examine responses in the solutions. The resulting flux distributions show that (1) redox shuttles between the chloroplast, cytosol, and mitochondrion may play a significant role at low light levels, (2) photorespiration can act to dissipate excess energy at high light levels, and (3) the role of mitochondrial metabolism is likely to vary considerably according to the balance between energy demand and availability. It is notable that these organelle interactions, consistent with many experimental observations, arise solely as a result of the need for mass and energy balancing without any explicit assumptions concerning kinetic or other regulatory mechanisms.
Flux balance models of metabolism generally utilize synthesis of biomass as the main determinant of intracellular fluxes. However, the biomass constraint alone is not sufficient to predict realistic fluxes in central heterotrophic metabolism of plant cells because of the major demand on the energy budget due to transport costs and cell maintenance. This major limitation can be addressed by incorporating transport steps into the metabolic model and by implementing a procedure that uses Pareto optimality analysis to explore the trade-off between ATP and NADPH production for maintenance. This leads to a method for predicting cell maintenance costs on the basis of the measured flux ratio between the oxidative steps of the oxidative pentose phosphate pathway and glycolysis. We show that accounting for transport and maintenance costs substantially improves the accuracy of fluxes predicted from a flux balance model of heterotrophic Arabidopsis cells in culture, irrespective of the objective function used in the analysis. Moreover, when the new method was applied to cells under control, elevated temperature and hyper-osmotic conditions, only elevated temperature led to a substantial increase in cell maintenance costs. It is concluded that the hyper-osmotic conditions tested did not impose a metabolic stress, in as much as the metabolic network is not forced to devote more resources to cell maintenance.
The active, yet energetically inefficient electron transport chain of the ethanologenic bacterium Zymomonas mobilis could be used in metabolic engineering for redox-balancing purposes during synthesis of certain products. Although several reconstructions of Z. mobilis metabolism have been published, important aspects of redox balance and aerobic catabolism have not previously been considered. Here, annotated genome sequences and metabolic reconstructions have been combined with existing biochemical evidence to yield a medium-scale model of Z. mobilis central metabolism in the form of COBRA Toolbox model files for flux balance analysis (FBA). The stoichiometric analysis presented here suggests the feasibility of several metabolic engineering strategies for obtaining high-value products, such as glycerate, succinate, and glutamate that would use the electron transport chain to oxidize the excess NAD(P)H, generated during synthesis of these metabolites. Oxidation of the excess NAD(P)H would also be needed for synthesis of ethanol from glycerol. Maximum product yields and the byproduct spectra have been estimated for each product, with glucose, xylose, or glycerol as the carbon substrates. These novel pathways represent targets for future metabolic engineering, as they would exploit both the rapid Entner–Doudoroff glycolysis, and the energetically uncoupled electron transport of Z. mobilis.
Organisms share a common core to their metabolic networks. But what determined this: chance, chemical necessity, or evolutionary optimization? In this issue of Molecular Cell, Noor et al. (2010) provide new evidence for selection of a network with optimal features from a broader set of possibilities.
Reconstructing a model of the metabolic network of an organism from its annotated genome sequence would seem, at first sight, to be one of the most straightforward tasks in functional genomics, even if the various data sources required were never designed with this application in mind. The number of genome-scale metabolic models is, however, lagging far behind the number of sequenced genomes and is likely to continue to do so unless the model-building process can be accelerated. Two aspects that could usefully be improved are the ability to find the sources of error in a nascent model rapidly, and the generation of tenable hypotheses concerning solutions that would improve a model. We will illustrate these issues with approaches we have developed in the course of building metabolic models of Streptococcus agalactiae and Arabidopsis thaliana.
Flux is a key measure of the metabolic phenotype. Recently, complete (genome-scale) metabolic network models have been established for Arabidopsis (Arabidopsis thaliana), and flux distributions have been predicted using constraints-based modeling and optimization algorithms such as linear programming. While these models are useful for investigating possible flux states under different metabolic scenarios, it is not clear how close the predicted flux distributions are to those occurring in vivo. To address this, fluxes were predicted for heterotrophic Arabidopsis cells and compared with fluxes estimated in parallel by 13C-metabolic flux analysis (MFA). Reactions of the central carbon metabolic network (glycolysis, the oxidative pentose phosphate pathway, and the tricarboxylic acid [TCA] cycle) were independently analyzed by the two approaches. Net fluxes in glycolysis and the TCA cycle were predicted accurately from the genome-scale model, whereas the oxidative pentose phosphate pathway was poorly predicted. MFA showed that increased temperature and hyperosmotic stress, which altered cell growth, also affected the intracellular flux distribution. Under both conditions, the genome-scale model was able to predict both the direction and magnitude of the changes in flux: namely, increased TCA cycle and decreased phosphoenolpyruvate carboxylase flux at high temperature and a general decrease in fluxes under hyperosmotic stress. MFA also revealed a 3-fold reduction in carbon-use efficiency at the higher temperature. It is concluded that constraints-based genome-scale modeling can be used to predict flux changes in central carbon metabolism under stress conditions.
Motivation: In recent years, several methods have been proposed for determining metabolic pathways in an automated way based on network topology. The aim of this work is to analyse these methods by tackling a concrete example relevant in biochemistry. It concerns the question whether even-chain fatty acids, being the most important constituents of lipids, can be converted into sugars at steady state. It was proved five decades ago that this conversion using the Krebs cycle is impossible unless the enzymes of the glyoxylate shunt (or alternative bypasses) are present in the system. Using this example, we can compare the various methods in pathway analysis. Results: Elementary modes analysis (EMA) of a set of enzymes corresponding to the Krebs cycle, glycolysis and gluconeogenesis supports the scientific evidence showing that there is no pathway capable of converting acetyl-CoA to glucose at steady state. This conversion is possible after the addition of isocitrate lyase and malate synthase (forming the glyoxylate shunt) to the system. Dealing with the same example, we compare EMA with two tools based on graph theory available online, PathFinding and Pathway Hunter Tool. These automated network generating tools do not succeed in predicting the conversions known from experiment. They sometimes generate unbalanced paths and reveal problems identifying side metabolites that are not responsible for the carbon net flux. This shows that, for metabolic pathway analysis, it is important to consider the topology (including bimolecular reactions) and stoichiometry of metabolic systems, as is done in EMA.
A decrease in retinoic acid levels due to alcohol consumption has been proposed as a contributor to such conditions as fetal alcohol spectrum diseases and ethanol-induced cancers. One molecular mechanism, competitive inhibition by ethanol of the catalytic activity of human alcohol dehydrogenase (EC 22.214.171.124) (ADH) on all-trans-retinol oxidation has been shown for the ADH7 isoform. Ethanol metabolism also causes an increase in the free reduced nicotinamide adenine dinucleotide (NADH) in cells, which might reasonably be expected to decrease the retinol oxidation rate by product inhibition of ADH isoforms. To understand the relative importance of these two mechanisms by which ethanol decreases the retinol oxidation in vivo we need to assess them quantitatively. We have built a model system of 4 reactions: (1) ADH oxidation of ethanol and NAD(+), (2) ADH oxidation of retinol and NAD(+), (3) oxidation of ethanol by a generalized Ethanol(oxidase) that uses NAD(+), (4) NADH(oxidase) which carries out NADH turnover. Using the metabolic modeling package ScrumPy, we have shown that the ethanol-induced increase in NADH contributes from 0% to 90% of the inhibition by ethanol, depending on (ethanol) and ADH isoform. Furthermore, while the majority of flux control of retinaldehyde production is exerted by ADH, Ethanol(oxidase) and the NADH(oxidase) contribute as well. Our results show that the ethanol-induced increase in NADH makes a contribution of comparable importance to the ethanol competitive inhibition throughout the range of conditions likely to occur in vivo, and must be considered in the assessment of the in vivo mechanism of ethanol interference with fetal development and other diseases.
The following report selects and summarises some of the conclusions and recommendations generated throughout a series of workshops and discussions that have lead to the publication of the Science Policy Briefing (SPB) Nr. 35, published by the European Science Foundation. (Large parts of the present text are directly based on the ESF SPB. Detailed recommendations with regard to specific application areas are not given here but can be found in the SPB. Issues related to mathematical modelling, including training and the need for an infrastructure supporting modelling are discussed in greater detail in the present text.)The numerous reports and publications about the advances within the rapidly growing field of systems biology have led to a plethora of alternative definitions for key concepts. Here, with ‘mathematical modelling’ the authors refer to the modelling and simulation of subcellular, cellular and macro-scale phenomena, using primarily methods from dynamical systems theory. The aim of such models is encoding and testing hypotheses about mechanisms underlying the functioning of cells. Typical examples are models for molecular networks, where the behaviour of cells is expressed in terms of quantitative changes in the levels of transcripts and gene products. Bioinformatics provides essential complementary tools, including procedures for pattern recognition, machine learning, statistical modelling (testing for differences, searching for associations and correlations) and secondary data extracted from databases.Dynamical systems theory is the natural language to investigate complex biological systems demonstrating nonlinear spatio-temporal behaviour. However, the generation of experimental data suitable to parameterise, calibrate and validate such models is often time consuming and expensive or not even possible with the technology available today. In our report, we use the term ‘computational model’ when mathematical models are complemented with information generated from bioinformatics resources. Hence, ‘the model’ is, in reality, an integrated collection of data and models from various (possibly heterogeneous) sources. The present report focuses on a selection of topics, which were identified as appropriate case studies for medical systems biology, and adopts a particular perspective which the authors consider important. We strongly believe that mathematical modelling represents a natural language with which to integrate data at various levels and, in doing so, to provide insight into complex diseases: 1. Modelling necessitates the statement of explicit hypotheses, a process which often enhances comprehension of the biological system and can uncover critical points where understanding is lacking. 2. Simulations can reveal hidden patterns and/or counter-intuitive mechanisms in complex systems. 3. Theoretical thinking and mathematical modelling constitute powerful tools to integrate and make sense of the biological and clinical information being generated and, more importantly, to generate new hypotheses that can then be tested in the laboratory.Medical Systems Biology projects carried out recently across Europe have revealed a need for action: 4. While the need for mathematical modelling and interdisciplinary collaborations is becoming widely recognised in the biological sciences, with substantial implications for the training and research funding mechanisms within this area, the medical sciences have yet to follow this lead. 5. To achieve major breakthroughs in Medical Systems Biology, existing academic funding schemes for large-scale projects need to be reconsidered. 6. The hesitant stance of the pharmaceutical industry towards major investment in systems biology research has to be addressed. 7. Leading medical journals should be encouraged to promote mathematical modelling.
We describe the construction and analysis of a genome-scale metabolic model of Arabidopsis (Arabidopsis thaliana) primarily derived from the annotations in the Aracyc database. We used techniques based on linear programming to demonstrate the following: (1) that the model is capable of producing biomass components (amino acids, nucleotides, lipid, starch, and cellulose) in the proportions observed experimentally in a heterotrophic suspension culture; (2) that approximately only 15% of the available reactions are needed for this purpose and that the size of this network is comparable to estimates of minimal network size for other organisms; (3) that reactions may be grouped according to the changes in flux resulting from a hypothetical stimulus (in this case demand for ATP) and that this allows the identification of potential metabolic modules; and (4) that total ATP demand for growth and maintenance can be inferred and that this is consistent with previous estimates in prokaryotes and yeast.
Motivation: Metabolic modelling provides a mathematically rigorous basis for system-level analysis of biochemical networks. However, the growing sizes of metabolic models can lead to serious problems in their construction and validation. In this work, we describe a relatively poorly investigated type of modelling error, called stoichiometric inconsistencies. These errors are caused by incorrect definitions of reaction stoichiometries and result in conflicts between two fundamental physical constraints to be satisfied by any valid metabolic model: positivity of molecular masses of all metabolites and mass conservation in all interconversions. Results: We introduce formal definitions of stoichiometric inconsistencies, inconsistent net stoichiometries, elementary leakage modes and other important fundamental properties of incorrectly defined biomolecular networks. Algorithms are described for the verification of stoichiometric consistency of a model, detection of unconserved metabolites and inconsistent minimal net stoichiometries. The usefulness of these algorithms for effective resolving of inconsistencies and for detection of input errors is demonstrated on a published genome-scale metabolic model of Saccharomyces cerevisiae and one of Streptococcus agalactiae constructed using the KEGG database.
Stoichiometric analysis of metabolic networks allows the calculation of possible metabolic flux distributions in the absence of kinetic data. In order to predict which of the possible fluxes are present under certain conditions, additional constraints and optimization principles can be applied. One approach of calculating unknown fluxes (frequently called flux balance analysis) is based on the optimality principle of maximizing the molar yield of biotransformations. Here, the relevance and applicability of that approach are examined, and it is compared with the principle of maximizing pathway flux. We discuss diverse experimental evidence showing that, often, those biochemical pathways are operative that allow fast but low-yield synthesis of important products, such as fermentation in Saccharomyces cerevisiae and several other yeast species. Together with arguments based on evolutionary game theory, this leads us to the conclusion that maximization of molar yield is by no means a universal principle.
Motivation: In recent years, several methods have been proposed for determining metabolic pathways in an automated way based on network topology. The aim of this work is to analyse these methods by tackling a concrete example relevant in biochemistry. It concerns the question whether even-chain fatty acids, being the most important constituents of lipids, can be converted into sugars at steady state. It was proved five decades ago that this conversion using the Krebs cycle is impossible unless the enzymes of the glyoxylate shunt (or alternative bypasses) are present in the system. Using this example, we can compare the various methods in pathway analysis.
Of all the molecular-interaction networks in plants, metabolism is by far and away the most completely described both in terms of network topology and in terms of the properties of its molecular components. Moreover, experimental tools exist that allow systematic measurement of its behaviour and there is now a wealth of data describing metabolic systems at the molecular level. Nevertheless, we remain some way off being able to exploit this knowledge to rationally engineer metabolism and we are unable to predict the nature of metabolic change under different conditions or in response to genetic intervention. One of the great problems is the persistence of the metabolic pathway paradigm when in reality, metabolism is a highly interconnected network and fixed pathways do not exist. To generate a more realistic view of metabolism, one must encapsulate information about metabolic behaviour into a model of the network as a whole. The construction of a mathematical model that describes the metabolic network of plants at ‘genome scale’ is now a plausible goal (especially in the case of model plants such as Arabidopsis). We have begun the construction of such a model and in my talk I will outline our current progress.
We describe a method by which the reactions in a metabolic system may be grouped hierarchically into sets of modules to form a metabolic reaction tree. In contrast to previous approaches, the method described here takes into account the fact that, in a viable network, reactions must be capable of sustaining a steady-state flux. In order to achieve this decomposition we introduce a new concept-”the reaction correlation coefficient, Ï†, and show that this is a logical extension of the concept of enzyme (or reaction) subsets. In addition to their application to modular decomposition, reaction correlation coefficients have