Metabolism in bacteria occurs via a set of complex, dynamic, and interconnected metabolic steps (metabolic modules). The outputs of such metabolic systems depend on the interplay between the genetic circuits in the cell, which generate proteins, and the metabolic circuits, which generate flux and metabolite pools (Fig. 1). Enzymes drive the metabolic circuits, while metabolites control enzyme activity and transcription, creating a complex interlocked set of circuits. Understanding how this complex system functions in the cell is the goal of integrated metabolism studies. Along with classical genetic studies, global "omics"-based approaches coupled to metabolic models and physiological insight are setting the stage for understanding how the metabolic network functions as an integrated system. Methylobacterium extorquens is a facultative methylotrophic bacterium, which has the property of two dramatically different modes of growth: growth on one-carbon (C1) compounds is reducing-power limited and involves a toxic intermediate (formaldehyde), while growth on multi-carbon compounds is energy-limited and involves standard heterotrophic metabolism. In addition, a suite of computational and genetic tools and multivariate "omics" datasets have been generated by our group. Therefore, this bacterium is becoming an attractive model system in which to ask fundamental questions regarding metabolic integration, using comparative studies of different metabolic conditions.
|Fig. 1. Circuitry in biology. Genetic circuits generate proteins, which drive the metabolic circuits and regulate transcription. Metabolic circuits generate flux and metabolite pools, and metabolites regulate enzyme activity and transcription.|
We propose to analyze how methylotrophic metabolism functions as an integrated system in M. extorquens. Our data to date show that at steady-state, the cells are in balance, while during perturbations, the metabolic network is shifted out of balance and the cells respond dramatically at the transcriptional, flux and metabolite levels. We have begun to define steady-state metabolic states, when the cell is in balance, for instance, a reducing power-limited state, an energy-limited state, and a stressed state. The response “resets” the metabolic state at a different level, and the cells return to a balanced state. Currently, we are now addressing how this response occurs at the metabolic module level, carrying out a set of manipulations to probe specific metabolic states and response scenarios.
When succinate-grown cells are switched to methanol, hundreds of genes change expression dramatically. These changes reflect the major shift in metabolism as the cells transition from energy-limited to reducing power-limited growth, and begin to generate the toxic intermediate formaldehyde at high rates. During that shift, succinate uptake systems, the TCA cycle, and NADH dehydrogenase shut down, while the C1 dissimilatory and assimilatory modules ramp up in stages, with the dissimilatory modules coming up first. Transcripts and proteins change dramatically, with some transcripts spiking up 50-fold, then dropping back down to a steady-state level higher than initially. Likewise, nucleotide pools change within minutes. ATP drops a little initially, then builds back up, while the reduced nicotinamide adenine dinucleotide pool spikes up, then drops back down to less than pre-stress levels. In the medium, formaldehyde spikes up early, then drops, to be followed by a spike of formate, which also drops. The overall picture is of cells in balance (at a metabolic set point), tipped out of balance with a dramatic response, then transitioning to a new metabolic setpoint.
Ongoing research. From this work, we now have multivariate datasets (transcripts, enzyme activities, fluxes, metabolites, and preliminary protein data) in which all of the growth conditions have been rigorously controlled for maximum reproducibility and in which the data have been generated from cells grown in the same laboratory under the same conditions. These datasets, coupled to our extensive physiological knowledge of this bacterium are allowing us to gain insights into how the output of the genetic circuits (or, gene expression) results in the output of the metabolic circuits (or, phenotype). For instance, what is happening to central metabolism during the typical 2-hr lag when cells are transitioned from succinate to methanol? Now we know: there is huge transcriptional activity of select genes that results in a rapid increase in activity of key metabolic modules. Nucleotide pools change, formaldehyde is overproduced and excreted, and the assimilatory modules sluggishly kick into gear. Once all of the modules are up and running, the cell starts growing, and a new metabolic state is established.
This background information and set of datasets and tools are now allowing us to address fundamental questions regarding the ability to grow on C1 compounds, in the context of metabolic integration. For the first time, we can assess cellular state and cellular response on a network-wide scale. Chemostat cultivation allows us to define specific steady-state metabolic set-points. Computational and visualization tools allow us to work with balanced networks and follow how each module responds to perturbations.
A series of computational and experimental studies we have carried out are creating an emerging picture of integrated metabolism: cells are tuned to a finite group of rough set points, or metabolic states, and the metabolic network functions within fairly broad constraints around those set points such that fluxes through key dissimilatory and assimilatory modules are constant, for a given growth rate. When cells encounter metabolic perturbations, the metabolic network is thrown out of balance, and they adapt to the perturbation by resetting the metabolic state to correct the imbalance. Fluxes come back to one of the set points. Robustness would then be achieved by a combination of redundant pathways with internal feedback loops (such as the formaldehyde buffering system we discovered in M. extorquens) and the intrinsic complex balancing of all cofactors and metabolites (the network topology). In this project, we are using this emerging model of metabolic integration as a framework to study metabolic integration of methylotrophy in M. extorquens.
The focus of this project is to address these questions of metabolic state and response to perturbations in the metabolic mode of methylotrophy, in M. extorquens. We are taking an experimental approach, supplemented by computational approaches that are constrained by the experimental data. Our experimental approach will directly address these fundamental issues, and we will use the computational approaches to test results from the experiments. The combination of experiment and computation will lead to an understanding that neither alone can provide. For instance, many combinations of metabolic solutions exist that will bring the metabolic network into balance. It is difficult to use solely computational tools to identify which of these many possible states actually exists in the cells. In addition, computational approaches do not predict the importance of modules involved in acquisition of low level components, such as vitamin synthesis or iron acquisition. Likewise, a strictly experimental approach limits the ability to understand where shifts in balances occur, to compensate for perturbations.
First, we are defining metabolic states by analyzing existing datasets at the modular level, defining specific metabolic states that might overlap in our studies, including stresses and slow growth rate, and by further examining linkages that have been suggested by our current studies. This work will set the stage for testing a set of predictions, in studies that will involve assessing module-level response to metabolic perturbations. In that case, we will perturb metabolism by both physiological and genetic manipulation. The long-term result will be a system-level understanding of this metabolic mode as a complex, dynamic, and interconnected metabolic network.
In all cases, we are measuring outputs of the genetic circuits (transcripts by microarrays and proteins by proteomics), and outputs of the metabolic circuits (fluxes, a suite of enzyme activities and metabolites including extracellular formaldehyde and formate). We will also run a flux balance simulation for all conditions, constrained by experimental data. These comparative datasets will provide both specific and general profiles of the metabolic state of the cells, and will provide the data to predictions about how metabolism is integrated.