Microbial ecology as a discipline
The term “ecosystem” was coined in 1935 by the botanist Tansley; its most consensual acceptation designates the set of all the individuals (also called community, or biocenosis) living in a delimited environment called “biotope”. Microbial ecology is then the study of a microbial community, and the interactions between its members and with its biotope. The origin of this disciplin can be traced back to the work of the microbiologist Winogradsky, starting in the late XIXth century. Indeed he formulated what is now envisioned as microbial ecology problems, by emphasizing the in-situ study of microbes and their relationship with their geochemical environment (Zavarzin, 2006). However, the development of this discipline was hindered by a technical deadlock until the end of the XXth century; culture was the only way to study microorganisms. Although some coculture experiments run in this context led to insights regarding species interactions (such as the formulation of the competitive exclusion principle (Gause, 1934)), microbiology mainly studied microbes under the perspective of pure culture.
The development of metagenomics approaches at the very end of the XXth century totally changed this persective. It was observed that populations in microbial communities interacts in ways comparable as those observed in macrobiology; competition for resources, predation, parasitism and mututalism occur (Little et al., 2008). Moreover, while unicellular organisms have initially been thought of as individualistic and disorganized, high levels of self-organization have also been evidenced through inter and intra specie interactions. Indeed, multiple observations shows that a microbe collective can self-organize, thus making large scale patterns to emerge to the point their behavior have been compared to this of a multicellular organism (Velicer et al., 1998; Velicer, 2003;
Jacob et al., 2004; Gloag et al., 2015). To summarise, those observations suggest that microbes can assemble into complex communities harboring interactions already described in macrobial community ecology. Those similarities foster the use of tools and concepts developed for macrobial ecology in microbial ecology.
However microbial communities also carry specificies which make it a field distinct from its macrobial coun-terpart. A list of diﬀerences between microbes and macrobes which are expected to impact community assembly is given by Nemergut and collaborators. This list includes the higher dispersal potential of microbes, the (poten-tially) vast genetic reservoir induced by their ability to stay dormant for long time periods, their important genetic plasticity. This latter feature implies that microbes are able to share genetic material even with distantly related kins, and also to adapt their metabolism to adapt to their environment far more easily than most macrobes could (Nemergut et al., 2013).
The study of microbial ecology is particularly relevant in multiple regards; the understanding of the behavior of biogeochemical fluxes in the environment, the understanding of the dynamics of human-associated microbiota such as this of the intestinal tract, and the improvement of industrial processes based on microbial communities. The work undertaken during this thesis particularly focuses on the latter.
The role of microbes in the geochemical fluxes
Microbial communities plays a crucial role in the regulation of biogeochemical processes, literally controlling the habitability conditions of the earth (Falkowski et al., 2008; Rousk and Bengtson, 2014; Treseder et al., 2012).
It has been estimated that the quantity of carbon stored in microorganisms correspond to 60-100% of this stored in plant biomass, and prokaryotes are estimated to constitute by far the largest pool of nitrogen and phosphorus of the biosphere (Whitman et al., 1998).
The fact that microorganisms modify their environment and that such phenomena is important at a global scale has also been intuited long before it could be measured, by microbiologists such as Winogradsky (Zavarzin, 2006) or Baas Becking (De Wit and Bouvier, 2006). Most of microbes are growing by carrying out redox reactions. While every living organisms community performs biochemical interactions with its environment, microbes are performing chemical interactions in the most direct way (even sometimes interacting directly with electrons (Strycharz-Glaven et al., 2011)). Therefore the functional and taxonomic structure of a microbial community is directly influenced, and directly influencing, the physicochemistry of its environment. Moreover, the microbes consitutes an important part of the total biosphere.
The human population recently realized that its own influence on earth’s geochemical cycles may impede the development of its future generations (Griggs et al., 2013). The study of the interactions between microbes and their environment is then of paramount importance for the sustainable evolution of the biosphere.
The use of microbial ecosystems in bioprocesses
Microbial ecosystems being ubiquitous, they have been serendipituously used by humankind to perform bio-chemical transformations since at least the neolithic. Indeed, the chemical analysis of potteries estimated to date from 7000 years B.C. revealed they contained a mixed fermented beverage of rice, honey and fruit. (McGovern et al., 2004). The first unicellular organisms have been described during the XVIIth century by van Leeuwenhoek and naturalists such as Swammerdam, however, the meaning to give to these observations, the notion of cell, and the actual relevance of the study of biology at microscopic level was debatted until the XIXth century. At that time, vitalism progressively became obsolete, industrial processes actually involving microbes gradually became perceived as such, and the transformations carried out by microbes became perceived as chemical reactions. Some important events marking this paradigm shift in biology are the synthesis of urea by W¨ohler in 1828, the discovery of the first enzyme (diastase) by Payen and Perzoz in 1833, and the dispelling of spontaneous generation by Pasteur’s swan-necked bottles experiment in 1859.
The main industrial bioprocesses relying on a community of microbes are wastewater treatment and anaerobic digestion. Other examples of bioprocesses are related to food industry, such as the production of wine, beer, cheese and vinegar, and the production of fodder through silaging. Another non-food related industrial use of fermentation is the exploitation of the acetone-butanol-ethanol metabolic pathway (Moon et al., 2016). This specific fermentation has been used to produce acetone until world war II when petrochemistry-based acetone production became cheaper. The possibilities oﬀered by microbial ecosystems engineering also goes beyond the improvement of the current bioprocesses; the use of this discipline is also envisioned as a mean to cure microbiota- related diseases (Shen et al., 2015) or to depollute an environment (Crawford and Crawford, 1996).
Many bioprocesses tend to use pure culture or low diversity cultures of microbes, because it makes the process more controlable. However in some cases the sterility of the process cannot be assured at viable costs; mixed culture then appears as an attractive solution despite its complexity (Kleerebezem and van Loosdrecht, 2007). This is notably the case with wastewater treatment and anaerobic digestion, where the input is non-sterile. In this case, diverse microbial ecosystems have to be used.
Addressing the technical deadlock on microbial community engineering
The development of novel technologies based on mixed microbial cultures, as well as the improvement of existing bioprocesses, is currently uneasy because of the lack of a solid theoretical basis to understand and predict the trajectories of microbial communities. All the models used to predict the behavior of a microbial community (such as ADM1 (Batstone, 2001)) are based on empirical equations calibrated after years of experiments. Moreover, those models do not provide knowledges useful to the modelling of other bioprocesses. Consequently the development of new bioprocesses, or the modification of existing ones, requires experimental work costly in time, money and engineering eﬀort.
On the other hand, products such as cheese or beer have been consistently reproduced from rudimentary and varied material, with few to none microbiological knowledge. This observation evidences that environmental factors (such as temperature and concentrations) exert an important constraint on microbial communities’ structure. Evidences of the deterministic eﬀect of environmental factors, and more specifically physicochemical factors, on the communities’ expressed functions, can also be found in natural environments. Indeed, remarkable spatial structuration patterns of microbial communities have been explained by thermodynamics in the case of Winograd-sky columns (Zavarzin, 2006), aquifers (Chapelle and Lovley, 1992) and lake hypolimnia (Boehrer and Schultze, 2008; M¨uller et al., 2012). The aforementioned cases feature ecological successions reflecting a gradient of available energy in the environment, this is also called a “redox tower”. Even in the absence of an obvious gradient, recent studies evidenced that physicochemical factors have a deterministic influence on the functional structure (Louca et al., 2016b,a; Louca and Doebeli, 2017). Those results are in line with other results showing microbial communities’ functional stability over taxonomic unstability (Fernandez et al., 2000; Burke et al., 2011). These results suggest that the functional structure of microbial communities is more sensitive to selection and less sensi-tive to invasion processes when compared to the taxonomic structure. Those evidences lead some authors to argue that functions may be a descriptive unit more relevant than taxon in the search for generic principles explaining microbial community structuration (Burke et al., 2011; Louca et al., 2016b; Lemanceau et al., 2017). Thermochemistry, which is the study of energy transfers in chemical systems, then appears as a good starting point in the search for generic organisational principles in microbial functional communities’ structuration.
The emergence of microbial thermodynamics
In 1960, Bauchop and Elsden suggested that since microbial growth was constrained by thermodynamics, the study of its energetics could lead to an eﬀective method to predict growth yields (Bauchop and Elsden, 1960). This fostered two decades of attempts to empirically correlate growth yield with physicochemical parameters of metabolisms (Mayberry et al., 1967, 1968; Prochazka et al., 1970; Minkevich and Eroshin, 1973; Stouthamer, 1973; Linton and Stephenson, 1978). Those attempts later cristallized in models accounting for growth energetics in order to predict growth yields and rates. This emerging field will later be referred to as “microbial thermodynamics”.
These are presented in the bibliography review of this thesis (chapter 2), in order to provide the reader with a firm grasp on the state of the art in microbial thermodynamics. This review is then concluded by a presentation of the objectives of the thesis (last section of chapter 2).
Objectives and organization of this memoir
The work undertaken during this thesis is purposed at improving the current theoretical knowledge on microbial ecosystem structuration. It is done by studying the capability of a model, the Microbial Transition State (MTS) model (Desmond-Le Qu´em´ener and Bouchez, 2014; Wade et al., 2016). This model has been used during this thesis to simulate the growth of multiple microbial populations catalyzing diﬀerent metabolisms to test to which extent the theory behind the MTS model is able to explain the functional structuration of microbial communities.
The next chapter is the bibliography review of this memoir. It is followed by the materials and methods, which describe the implemention details of the MTS model. Then two articles based on predictions made with the model are presented. The first article is a demonstration of trophic and ecological patterns predictable by the MTS model through increasingly complex implementations. The second papers features an implementation of a microbial community related to a bioprocess (activated sludge), along with the calibration of the model’s parameters from experimental data. The third article leverages on a body of experimental data collected from the literature to question the link between the growth yield and physicochemical properties of the metabolisms. The final chapter is a general discussion which summarises what insight has been earned on the structuration of microbial communities and which further developments are required to step forward in the development of microbial community engineering.
Modeling the energy dependence of microbial growth
This chapter is the bibliography synthesis of the memoir. It corresponds to a bibliography review made at the invitation of the FEMS Microbiology Ecology journal. The purpose of this review is to introduce some of the models proposed to describe the role of energy in microbial growth. This review focuses on models predicting growth yields and dynamics at population scale.
The review is split into two parts; the first part concerns the growth yield prediction models, which make predictions about the energy balance of microbial growth and do not provide a formulation for dynamic variables such as growth rate. The second part concerns population dynamics models, which express dynamic variables (such as the growth rate) from energy-based considerations on microbial growth.
As a chapter of this memoir, this review presents the MTS model in the context of other microbial thermodynamics models. Firstly, this helps to understand the relevance of the contribution of the MTS theory to the field of microbial thermodynamics, as the properties of the MTS theory in terms of population dynamics prediction are studied in the chapter 4 and 5 of this memoir. Secondly, this emphasizes the contribution of previous models to the implementations of the MTS model used in this memoir. Indeed, since the MTS theory, in itself, does not define some variable (dissipated energy) required to close the energy balance of growth, it borrows an empirical expression allowing to compute this variable from a previous model (Heijnen’s energy dissipation model, introduced in this chapter). The calibration of a new empirical relationship to close the energy balance of microbial growth is attempted in chapter 6.
Microbial communities are ubiquitous and play a crucial role in the regulation of biogeochemical processes, literally controlling habitability conditions on earth (Falkowski et al 2008). Moreover, they are used in many bioprocesses such as wastewater treatment and many food production processes. Despite their importance, studies on microbial growth communities only recently gained momentum, with the development of molecular microbial ecology methods in the late 1990s, which made possible to study microbial communities in situ. It was observed that the populations in microbial communities interact in ways that are comparable to those observed in macrobiology, including competition for resources, predation, mutualism (Little et al 2008). However microbial communities also have specificities that make the field distinct from its macrobial counterpart, including higher dispersal of individuals and important genetic plasticity (Nemergut et al 2013). Before the study of microbial communities took off, models were designed predicting their mass balance and kinetics, notably for the use of mixed culture waste management bioprocesses (Activated Sludge Model (Henze et al 1987, Henze et al 2000), Anaerobic Digester Model 1 (Batstone et al 2001)). They use an approach similar to the study of pure cultures, where the dependency between the system’s variables (biomass, growth rate as a function of substrate concentration etc.) is captured in the parameters of empirical equations calibrated from observations. However, a model designed using such an approach is only to a limited extend applicable if the culture conditions differ from those in which the model was calibrated. Moving outside the calibrated area of the model may result in a complete change in microbial community structure and corresponding stoichiometric and kinetic properties. Notably, the calibration of the biomass composition may be a key limitation. Indeed, the ecological structure reproduced by such models is the consequence of expert knowledge implemented into it, but does not emerge from the model itself. In fact, the mechanistic approach to modelling microbial communities is not very advanced as ecology in general did not yield many generic rules, to the point that its ability as a discipline to produce such rules has been called into question (Lawton 1999).
These engineering models do not aim at providing a theory in which invariants of microbial community structure can be formulated. Consequently, such theory is still lacking. Meanwhile, the amount of data that can be collected from the study of microbial communities is increasing exponentially (Nature Methods Editorial 2009). To quote Henry Poincarré; “Science is built up with facts, as a house is with stones. But a collection of facts is no more a science than a heap of stones is a house”. Microbial ecology as a science has indeed reached a stage at which a firmer theoretical footing is needed to integrate its numerous observations into a coherent picture.
One reason for the apparent lack of generic rules in microbial community structure is the use of taxon as the descriptive unit for microbial communities. Indeed, multiple experimental reports provide evidence that microbial communities have stable functions despite an unstable taxonomic profile (Burke et al 2011, Fernández et al 1999, Huttenhower et al 2012). Moreover, experimental observations suggest that physicochemical factors (pH, temperature, salinity etc.) exert a determining influence on the functional structure of microbial communities (Louca et al 2016a, Louca et al 2016b, Raes et al 2011). According to those results, the functional structure is more sensitive to selection and less sensitive to invasion than the taxonomic structure.
Thermochemistry appears to be a good starting point to build a theoretical framework to integrate generic assembly rules for microbial communities. Indeed, thermochemistry makes it possible to model the influence of physicochemical factors on growth. While the above-mentioned results were obtained using metagenomics, for decades, remarkable spatial patterns of microbial communities have been explained by thermodynamics (like Winogradsky columns (Zavarzin 2006) or aquifers (Chapelle and Lovley 1992)). From the 1960s on, the observation of invariants in the structure of microbial community encouraged the development of models linking variables such as growth rate and yield to physicochemical factors (Bauchop and Elsden 1960, McCarty 1965). Since then, thermodynamic theories of microbial growth have been published in a somewhat disseminated fashion. The ensuing problems of redundancy, notation mismatch and internal inconsistency (Heijnen and Dijken 1991) make it hard to draw a clear map of the existing theories on the subject. The aim of this review is consequently to provide a clear map of the discipline, hereafter referred to as “microbial thermodynamics”. We focus on models intended to enable conclusions to be drawn at population scale, although models also exist at the intracellular scale (González-Cabaleiro et al. 2013), and also at the scale of whole ecosystems (Ludovisi 2009, Odum 1969, Svirezhev 2004). In next section, we present the classical differential equation framework to model microbial population growth, along with ideas concerning the formalization of microbial metabolism. In the following section, we review models intended to predict microbial growth yield by accounting for the energy and matter balance in growth. We then present models that apply thermodynamic considerations for the prediction of microbial population growth dynamics.
Table of contents :
Chapter 1: Introduction
1.1 Microbial ecology as a discipline
1.2 The role of microbes in the geochemical fluxes
1.3 The use of microbial ecosystems in bioprocesses
1.4 Addressing the technical deadlock on microbial community engineering
1.5 The emergence of microbial thermodynamics
1.6 Objectives and organization of this memoir
Chapter 2: Bibliography review – Modelling the energy dependence of microbial growth
Chapter 3: Materials And Methods
3.2 Specifications of the MTS model
3.2.1 Units and dimensions
3.2.2 Thermodynamic standard conditions
15 3.2.3 Metabolic reactions computations
3.2.4 Microbial dynamics
3.2.5 Additional dynamics
3.2.6 Chemostat dynamics
3.2.7 Point-settler dynamics
20 3.3 Description of the MTS simulation framework
3.3.2 Files organisation
3.3.3 Basic principle
3.3.4 Parameters computation
3.3.5 User-definable parameters
3.3.8 Definition of simulations’s conditions
3.3.9 Calibration of the parameters of the MTS model
Chapter 4: Consistent microbial dynamics and functional community patterns derived from first physical principles
Chapter 5: Consistent dynamics of simplified activated sludge ecosystem obtained by the calibration of a parameter-parsimonious thermodynamic model
Chapter 6: Predicting microbial growth yield from the nature of metabolic reactions
Chapter 7: Discussion And Perspectives
7.1.1 Prediction of the “redox tower” by the MTS model
7.1.2 Consequences of the growth rate function of the model
7.1.3 Sensitivity of the MTS model’s predictions to the value of its parameters
7.1.4 Use of the MTS model for simulating mixed culture bioprocesses
7.1.5 The estimation of dissipated energy to predict microbial growth yield
7.2 Perspectives and conclusion
7.3 Spatialization of the MTS model
7.4 Accounting for the metabolic versatility of microbial populations
Chapter 8: Bibliography
Chapter 9: Appendix
9.1 Growth limitations by nutrients predicted by the MTS model
9.2 Example of getExperimentalDesign files
9.3 Specificities of the MTS model’s growth function
Superimposition of Monod’s and MTS’s growth curves
Effect of the value of KS and VH on the shape of Monod’s and MTS’s growth curve respectively 227
Difference between Monod’s and MTS’s growth curves
Comparison between Monod’s and MTS’s growth curves considering their usual parameters value range
9.4 Compliance of a population simulated by MTS to Liebig’s law
9.5 Relationship between growth yield and rate in some microbial thermodynamics models
Heijnen’s dynamic model
60 Jin and Bethke’s model
9.6 Supplementary material of chapter 4
9.7 Supplementary material of chapter 5
9.8 Supplementary material of chapter 6
9.9 R´esum´e substanciel de la th`ese en fran¸cais