«Integrative Computational Approaches to Complex Ecophysiological Systems Andreas Bohn (Oeiras, Portugal) With 6 Figures Abstract The present work ...»
Nova Acta Leopoldina NF 96, Nr. 357, XXX–XXX (2009)
Integrative Computational Approaches
to Complex Ecophysiological Systems
Andreas Bohn (Oeiras, Portugal)
With 6 Figures
The present work highlights the application of integrative computational tools in ecophysiological studies. With the
example of circadian rhythms in Crassulacean acid metabolism, modeling approaches for the integration of diverse
levels of biological organization, as well as different time- and space scales are assessed. The integration of heterogeneous data sources is discussed with the case of a web-based computational infrastructure in a multinational, multidisciplinary project on phototrophic biofilms. Both examples underline the importance of aligning the respective scientific cultures and communication forms of the project partners for the successful application of computational techniques in the research on complex biological systems.
Zusammenfassung Die vorliegende Arbeit behandelt die Anwendung integrativer Computerwerkzeuge in ökophysiologischen Studien.
Am Beispiel circadianer Rhythmen im Crassulaceen-Säurestoffwechsel werden Modellierungsansätze zur Integration verschiedener Ebenen biologischer Organisation sowie verschiedener Zeit- und Längenskalen erörtert. Die Integration heterogener Datenquellen wird anhand der Fallstudie einer internetbasierten Dateninfrastruktur im Rahmen eines multinationalen, multidisziplinären Projekts über phototrophe Biofilme diskutiert. Beide Beispiele unterstreichen die Bedeutung der wechselseitigen Abstimmung der wissenschaftlichen Kulturen und Kommunikationsformen der Projektpartner für die erfolgreiche Anwendung von computerbasierten Techniken in der Erforschung komplexer biologischer Systeme.
1. Introduction In the second half of the 20th century, technological breakthroughs in biochemistry, imaging and information processing have triggered a tremendous increase in the generation of information about the constituent parts of living organisms. Until the full sequencing of the human genome, the predominant approach to understanding the complex nature of biological systems followed a reductionist paradigm, relating organismal functionality and dynamics to the activity of individual molecules (Keller 2005, Hütt and Lüttge 2005). The evidence that the functionality or disease of entire biological systems could not be explained by deciphering solely the letters of the ‘Book of Life’ (Noble 2003), lead to a massive paradigmatic change in life sciences and the emergence of a large number of novel research approaches, attempting to unravel how whole-organismic function surges from the interactions between the parts of the system (Busch and Eils 2005). One of the most prominent approaches taking a systemwide perspective on life has been entitled systems biology (Ideker et al. 2001, Kitano Andreas Bohn 2001). Since its beginnings, this field has attracted scientists of many different disciplines, from experimental biology over engineering to computer sciences and physics (Keller 2005). Despite the existing lack of a clear definition of what exactly constitutes systems biology, common elements to most suggestions for a definition include the quantitative modeling of biological systems across different levels of organization, the integration of heterogeneous data sets, and the interdisciplinary networking of experimental biologists and quantitatively trained scientists (Morris et al. 2005).
The inherently quantitative character of systems biology, together with the traditionally strong connection between molecular high-throughput studies and bioinformatics, has founded an implicit tendency to understand systems biology as the genome-wide, or generally ‘ome-wide’, study of cellular and organismal function emerging from the interactions of its molecular parts (Westerhoff and Palsson 2004). Yet, it has been suggested that in studies of multicellular organisms and their environmental interactions, e.g., in crop development, a dialectic between bottom-up and top-down approaches provides a more efficient approach to biocomplexity and biotechnological developments, than hierarchical, unidirectional advances, working from the molecular level up to higher levels of organization (Hammer et al.
2004). These arguments gain even more weight if one understands systems biology as a truly holistic attempt to integrate all space-scales of the biosphere from molecules to ecosystems.
Bridging the entire spectrum of scales will also increase the spectrum of computational tools to be applied. While molecular studies are mainly challenged by the large volumes of information to be processed, quantitative ecological studies also face a bewildering variety of data types, sources and logical structures which need to be integrated (Jones et al. 2006).
Situated between the molecular and the ecological scale, computational ecophysiology is about to become an interesting meeting point of bottom-up and top-down approaches in fullscale systems biology.
By means of two case studies, the present work discusses the implementation of the basic elements of systems biology in ecophysiological studies. Section 2 discusses multilevel modeling with the example of circadian rhythms of whole-leaf gas exchange in a Crassulacean acid metabolism (CAM) plant. It is demonstrated how the development of logically connected models with differing degrees of abstraction can elucidate the connection of dynamical processes at the cellular and organismal level, and enhance related experimental studies.
Section 2.3 draws some general conclusions on multilevel modeling approaches in environmental physiology.
The second example, presented in Section 3, highlights the integration and analysis of heterogeneous data sources in ecophysiology, generated in the context of a multinational, multidisciplinary project on phototrophic biofilms. It stresses the importance of metadata management to balance flexibility with consistence in data integration, and the importance of matching the applied analysis tools with the size and structure of the given data pool. Section 3.4 discusses general aspects in the creation of effective scientific data workflows and underlines the importance of the interdisciplinary communication between the involved partners. The conclusions on both examples are integrated in Section 4, discussing the necessity to complement new technologies for knowledge discovery and hypothesis testing with cultural and sociological advances in networked team science, to promote successful interdisciplinary research in biocomplexity.
2. Circadian Rhythms in Crassulacean Acid Metabolism: Integrating Data and Hypotheses on Different Levels of Organization
2.1 Chronobiology of CAM Cyclic, oscillatory dynamics are deeply entrenched into the temporal organization of living organisms. An elusive example is the adaptation to geophysical cycles, in particular the 24 h-cycle of day and night (Pittendrigh 1993). These co-called circadian rhythms are ubiquitously observed in plants, animals and microorganisms, and there is increasing evidence that the coordinated timing driven by endogenous, circadian clocks enhances organismal fitness (Paranjpe and Sharma 2005, Dodd et al. 2005). A prominent example for how the temporal organization of metabolic processes by a circadian system can provide ecological advantages to plants is Crassulacean acid metabolism (CAM), an adaptation of plants to drought stress (Black and Osmond 2003, Lüttge 2004, and references therein): Governed by an endogenous circadian system, the uptake of CO2 from the environment is shifted to occur predominantly at nighttime. The temporal separation of CO2 uptake from its fixation and storage as starch via the light-dependent C3 pathway, allows the use of the internal CO2 store accumulated during the night. Diurnal photosynthesis can then take place behind closed stomata during the hottest and driest phase of the day, yielding an overall improvement of water-use efficiency.
Over the last two decades, two principal hypotheses about the origin of the endogenous oscillations in the carbon metabolism of CAM plants have been proposed. One is based on a molecular feedback system which hierarchically drives metabolic rhythmicity by modulating the activity of key enzymes in CAM carbon metabolism (Hartwell 2005, and references therein). The second approach features a biophysical pacemaker localized at the the vacuolar membrane, the tonoplast. The principal feedback mechanism is based on the nonlinear interdependence of the efflux rate of vacuolar malic acid, the principal store for nocturnally acquired CO2 and the order of the vacuolar membrane (Lüttge 2000). While the former hypothesis to date has not been modeled in a quantitative fashion, the latter mechanism was subject to an ongoing iteration of experimental studies and quantitative multi-level modeling.
2.2 Modeling CAM Rhythmicity: From Single- to Multi-Oscillator Systems and Back The first quantitative model of CAM, based on the experimental knowledge available at that time was presented by Nungesser et al. (1984). By interdisciplinary collaboration between engineers and botanists, a computational model was developed, featuring 6 ordinary differential equations (ODEs), representing 6 metabolic pools, interacting by first order reaction and regulation terms. Already in that model, the principal point of impact for environmental parameters like light intensity was the transport of malic acid at the tonoplast. As described in detail by Lüttge (2000), this model evolved in several steps in alignment with surging experimental evidences. The hitherto final point in the evolution of cellular CAM models was reached with the model by Blasius et al. (1999), quantifying the mentioned nonlinear interdependency of the efflux of vacuolar malic acid and its level of accumulation.
A principal merit of this model is the representation of the conditionality of the CAM cycle in continuous light: here, the circadian cycle is arrested in steady states with a filled vacuole at low temperatures, and an empty vacuole when the plant is exposed to high temperNova Acta Leopoldina NF 96, Nr. 357, XX–XX (2009) Andreas Bohn atures (Grams et al. 1997). Starting to lower the temperature from the latter arrested state, the model predicts the onset of circadian oscillations once the temperature crosses the bifurcation threshold, independent on the rate of temperature change. This prediction was contradicted by experiments by Rascher et al. (1998): rhythm re-initiation could only be observed experimentally in response to fast temperature changes, while a slow transition between the two temperature regimes maintained the gas-exchange cycle arrested in the arrhythmic state.
This finding induced a fundamental change in the modeling approach to CAM rhythms.
The experimental results by Rascher et al. (1998) became interpretable by considering populations of several copies of the CAM model (Blasius et al. 1999) with an additional noise term (Beck et al. 2001). Taking into account the multi-cellular nature of the measured wholeleaf gas exchange, and the stochastic dynamics of the oscillations emerging from omnipresent noise in real systems, rhythm re-initiation after a fast temperature transition could be understood as the synchronization of a population of noisy oscillators by a quick common transition of all oscillators from a fixed-point to a limit-cycle regime. Slow temperature transitions would yield an onset of oscillations in each individual oscillator, however, phase desynchronization of the population would maintain the arrhythmic global signal. In addition to that, this multi-level approach gave rise to interpret macroscopic arrhythmicity and rhythm damping as a noise-induced loss of phase coherence among the microscopic oscillating elements of the system. The success of this multi-oscillator model, introducing the “clockshop” hypothesis (Winfree 1975) to CAM rhythms, lead to a new experimental approach to whole-leaf rhythmicity. By implementing a chlorophyll fluorescence imaging facility, image sequences of photosynthetic efficiency could be recorded to assess the spatio-temporal metabolic dynamics in CAM leaves in day-night cycles and in continuous light (Rascher et al. 2001). This new technique unraveled a significant amount of spatio-temporal heterogeneity in the CAM leaves in day-night cycles and continuous light conditions, which was a new and unexpected result for a physiologically and anatomically homogeneous leaf.