Informing the improvement and biodesign of crassulacean acid metabolism via system dynamics modelling

Authors


Achieving sustainable increases in plant productivity that will meet global demands for food, fuel and fibre in the drier climates predicted for our warming planet is a major challenge for the twenty-first century. Over 35% of the world's land area is considered arid or semi-arid with precipitation that is inadequate for most agricultural uses, and nearly 80% of the world's population is exposed to high levels of threat related to water security (Vörösmarty et al., 2010). The inherently high water-use efficiency (WUE) of plants with the photosynthetic specialization of crassulacean acid metabolism (CAM) has highlighted their potential for sustainable production of biomass in warmer and drier environments (Borland et al., 2009, 2011). CAM plants are adapted to uptake CO2 at night which means that they can use 20–80% less water to produce similar amounts of biomass compared with C4 and C3 plants, respectively. In a New Phytologist Tansley review published over 20 yr ago, Nobel (1991) challenged the common misconception that all CAM plants grow slowly and instead showed that the productivity of some CAM species of Agave and Opuntia can exceed that of most C3 and C4 plants in certain habitats and under particular agronomic conditions. Nobel developed an environmental productivity index (EPI), which represents a first-order approximation of the combined influence of environmental factors (i.e. water, temperature, light, nutrients) on net CO2 uptake over 24 h. The EPI represented the first modelling framework to inform and improve agronomic practice for the cultivation of CAM crops and predicted the expanding geographical regions that CAM plants might successfully exploit in a warmer and drier world (Nobel, 2000). In this issue of New Phytologist, Owen & Griffiths (pp. 1116–1131) describe a system dynamics (SD) approach to modelling CAM which encapsulates the key biochemical and physiological parameters of this photosynthetic specialization to provide higher resolution (compared with EPI) for predicting and quantifying patterns of net CO2 uptake across the day/night cycle. The SD model takes an important step towards setting up a mechanistic framework to identify the parameters that rate-limit net CO2 uptake over the CAM cycle. This is a topic which looks set to grow in importance, given the recently raised profile of CAM as a target for synthetic biology to engineer improved WUE in C3 biomass crops (http://cambiodesign.org/), and for the adoption of certain succulent CAM species as dedicated bioenergy feedstocks (Davis et al., 2011).

The system dynamics model by Owen & Griffiths provides a foundation onto which molecular and signalling components can be added in a heuristic integrative manner.’

Capturing the dynamics of CAM

Significant progress has been made in computational modelling of C3 and C4 photosynthesis over the past 30+ yr with integrative models encompassing a suite of physiological and biochemical reactions and transport processes involved in CO2 uptake, photosynthesis, Calvin cycle activity, and carbohydrate partitioning (von Caemmerer, 2013; Zhu et al., 2013). Previous efforts to model CAM have been limited to a few variables related to CO2 concentration, malate concentration, temperature and light intensity (Blasius et al., 1999; Matsuo et al., 2013). The SD model presented by Owen & Griffiths captures the inherent plasticity that exists both within and between CAM species in the magnitude and duration of nocturnal and day-time gas exchange (Borland et al, 2011). CAM is commonly described in four phases (Osmond, 1978) that encapsulate temporal separation of carboxylases (phosphoenolpyruvate carboxylase (PEPC) and Rubisco), malate accumulation and mobilization from the vacuole, and diel carbohydrate processing (Fig. 1). Owen & Griffiths have successfully used this four-phase framework to inform the parameterization of an SD model with measured biochemical and physiological constants and state-dependent feedback which interact to generate predicted outputs of net CO2 uptake and malate turnover on a temporal basis. Thus, the SD model captures the ‘CAM system’ and establishes the causal relationship between defined biochemical/physiological constants and potential carbon gain. The SD model demonstrates that manipulations of several key parameters (e.g. vacuole capacity, stomatal and mesophyll conductance, maximum rate of PEPC activity, Michaelis-Menten constant for PEPC) could alter the diel patterns of carbon uptake signatures encountered in CAM. Published data on the dicot CAM species Kalanchoë daigremontiana, which typically displays all four phases of CAM, provided the basis for parameterization and development of the SD model. With the manipulation of just three factors (i.e. delayed activation of Rubisco, rates of malate decarboxylation and mitochondrial respiration), the model was also able to simulate net CO2 uptake in a monocot CAM species, Agave tequilana, in which daily carbon gain tends to be dominated by phase I net CO2 uptake. Such findings highlight the potential of the SD model for predicting diel patterns of net CO2 output across physiologically and taxonomically divergent CAM species.

Figure 1.

The four phases of crassulacean acid metabolism (CAM) describe the interplay and coordination of carboxylases (phosphoenolpyruvate carboxylase (PEPC) and Rubisco) for taking up CO2 over the day/night cycle and generating nocturnal accumulation and day-time mobilization of malate from the vacuole. The diel cycles of net CO2 uptake and malate turnover are accompanied by a reciprocal cycle of carbohydrate turnover (adapted from Osmond, 1978).

Integrating the model within a systems approach for bioengineering CAM

The SD model created by Owen & Griffiths provides a platform for moving towards computational modelling of CAM which would include gene regulatory networks, signaling pathways, and metabolic network fluxes (Fig. 2). A future challenge for modelling CAM at the molecular level will be to determine the spatial and temporal concentration of proteins, the kinetics of protein–protein interaction, and the post-translational status of key proteins. One approach to resolve the molecular dynamics of CAM in a time, cost and resource-efficient manner will be to model the small-scale network modules, which include a small number of functionally associated genes, separately, and then integrate these modules with the biochemical and physiological components. The ultimate goal of CAM-modelling is to accurately predict the consequences of modifying different molecular, signalling, cellular and leaf level properties on canopy CO2 uptake, water- and nitrogen-use efficiencies, and ultimately, productivity. The model by Owen & Griffiths provides a foundation onto which molecular and signalling components can be added in a heuristic integrative manner.

Figure 2.

The major elements of a systems model of crassulacean acid metabolism (CAM). The model includes environmental, physiological and biochemical inputs along with temporal control switches that activate key regulatory processes for carboxylation and decarboxylation. The ‘fluxes’ of metabolites that are calculated from these inputs interact with state-dependent feedback to generate predicted outputs of net CO2 uptake and malate turnover on a 24 h basis (adapted from Owen & Griffiths, this issue of New Phytologist, pp. 1116–1131). The biochemical and physiological components of the model are underpinned by dynamic models of metabolic networks, gene regulatory networks and signalling pathways. PAR, photosynthetically active radiation; PEP, phosphoenolpyruvate; PEPC, phosphoenolpyruvate carboxylase.

Computational modelling of CAM could accelerate the improvement of CAM crops in terms of biomass productivity and quality-related attributes. Whilst focusing on the establishment of a SD framework in a CAM model species (i.e. Kalanchoë daigremontiana), Owen & Griffiths also applied their model to an economically important CAM crop, Agave tequilana, that has been grown commercially for centuries as a source of alcohol (Tequila) and is also currently being considered as a potential feedstock for biofuel production. A recent life cycle energy and greenhouse gas (GHG) analysis suggests that ethanol derived from Agave is likely to be superior, or at least comparable, to that from corn, switchgrass and sugarcane in terms of energy and GHG balances, as well as in ethanol output and net GHG offset per unit land area (Yan et al., 2011). However, several challenges need to be addressed to materialize the potential of Agave for biofuel production. Agave biomass yield in the United States (c. 2 Mg ha−1) is relatively low compared with the highest levels (> 20 Mg ha−1) worldwide (Davis et al., 2011), and some productive Agave species are cold-sensitive. Understanding the causal relationships and rate-limiting processes in CAM expression through dynamic modelling could maximize biomass yield and enhance tolerance to abiotic and biotic stresses.

Food and biofuels are largely produced from non-CAM crops, yet it is imperative to improve crop WUE in the face of increasing frequency of drought and heat stress. A multi-institutional project, recently funded by the US Department of Energy, is embarking on the ambitious goal of introducing the CAM machinery into the C3 bioenergy crop Populus to improve WUE (http://cambiodesign.org/). A major aim of the project is to model and introduce core CAM gene modules in a C3 background. This will require a major expansion of the model developed by Owen & Griffiths with a focus on predicting the compatibility and efficiency of the synthetic CAM/C3 hybrid system. Comparative genomics and transcriptomics analyses of taxonomically diverse CAM and non-CAM species will be critical for deciphering how many genes are shared between CAM and C3 photosynthesis and establishing which genes are required for the C3-to-CAM transition. Genomic resources for CAM species are expanding and will facilitate the identification of key genes (e.g. enzymes, transcription factors) involved in core CAM modules (e.g. carboxylation, decarboxylation, stomatal movement). These molecular components could be incorporated into the SD model developed by Owen & Griffiths to establish a comprehensive mechanistic model of CAM photosynthesis.

Ultimately, the practicalities for maximizing CAM-based biomass and carbon sequestration, as well as the bioengineering of CAM into C3 crops, must be informed by underlying molecular, physiological and ecological processes, which will require the collaboration of scientists from diverse areas of research in the CAM field and beyond. The upcoming 34th New Phytologist Symposium on ‘Systems biology and ecology of CAM plants’ (http://www.newphytologist.org/symposiums/view/5) will provide an opportunity for integrating functional genomics with biochemistry, physiology, development, ecology, and evolutionary studies to gain new insights into the regulatory mechanisms and evolutionary origins of the pathway. Progress in CAM genomics research and this special New Phytologist CAM Symposium look set to increase our understanding of this photosynthetic specialization and facilitate the future development of integrative frameworks for modelling CAM.

Acknowledgements

Oak Ridge National Laboratory is managed by UT-Battelle, LLC for the US Department of Energy under Contract Number DE–AC05–00OR22725.

Ancillary