Funded Grants

Community modeling of bacterial metabolic interactions

In natural environments, individual organisms do not live in isolation. Bacterial species occupy ecological niches, in most cases forming together dynamic communities. The composition and interactions within these communities change over time and in response to environmental stimuli [1‐3]. Two species co‐inhabiting an ecological niche can compete for resources, cooperate to maximize resource utilization, divide the available resources between them, or combine competition, cooperation and division to various degrees. The relative fitness of a species in a certain ecological niche and, hence, the structure and composition of that particular community, will be to a large extent determined by the ecological strategies taken by all organisms in the community and by their interactions.

The composition of bacterial communities is a major factor in human health. The microorganisms that live inside and on humans are estimated to outnumber human somatic and germ cells by a factor of ten. The genomes of these microbial symbionts provide functions that humans did not evolve on their own [4]. The microbiome of the human gut, for example, is composed of different lineages with a capacity to communicate with one another and with the host; it consumes, stores, and redistributes energy; and it mediates physiologically important chemical transformations [5]. Variations in the identity and abundance of species within this community affect its metabolic potential and hence have important medical implications. Obesity, for example, is associated with changes in the relative abundance of bacterial phyla [6]. Moreover, variations in the relative abundance of species in a community may shift the ecological balance of a given niche allowing, for example, the outburst of pathogens [7]. Such variations may also affect managed ecosystems such as bioreactors and agricultural fields [8,9]. Thus, considering its medical and ecological implications, successful modeling of bacterial communities is thus likely to have broad consequences. Experimental and computational tools are now becoming available for the modeling of bacterial interactions, taking into account the identity of the interacting species and the available nutritional supplies.

(I)Background: from genomic information to ecological information

Progress in the reconstruction of genome‐wide metabolic networks has led to the development of network based computational approaches for linking an organism with its biochemical habitat. Beyond predicting the biochemical habitat, such approaches are further exploited to study the interactions of microbes with other species thriving in similar habitats. Here we provide a short review of these innovative methods, some developed by us, for processing genomic information into environmental and ecological information.

From genomic information to environmental information: the consecutive nature of the reactions in metabolic pathways means that they can be modeled in the form of a network of enzymes and chemical transformations, and graph theory can be used to represent and understand metabolism [10]. The availability of many completely sequenced genomes together with several generic metabolic schemes [11] has led to the reconstruction of the metabolic networks of hundreds of species across the tree of life [12]. A recently published “seed algorithm” (developed as part of our previous work [13]) allows to use metabolic networks for predicting the set of metabolites an organism consumes from its surroundings. This algorithm takes as an input the metabolic network of a given species and looks for source components (i.e., components with no incoming edges); those components cannot be synthesized by the species and are hence exogenously acquired. This set of compounds reflects the metabolic environment of a species [14].

From environmental information to ecological information: the seed algorithm allows the computation of species‐specific habitats. Aggregating the habitats computed for a set of species (for which a metabolic network is available) we previously obtained an ensemble of predicted natural habitats corresponding to hundreds of species across the tree of life. This ensemble represents the broadest ecological view provided by current data, allowing us to examine the viability of species across a wide‐range of ecological niches. Traditionally, flux‐balance approaches were used for estimating the ability of a species to grow on a given medium [15], but the underlying stoichiometric metabolic models are available for only a few selected species. Topological‐driven approaches (requiring only the network topological backbone and not a full‐blown stoichiometric model) have been recently shown to be sufficient for estimating growth by studying the ability of an organism to successfully expand its metabolic network so it produces a set of target metabolites that are essential for growth [16, 17]. Repeating this procedure for all species over all environments, we have provided the first large‐scale ecological model describing environmental characteristics from both a species and a habitat point of view [18]. From a species perspective, the model describes the range of environments it can inhabit. From an environmental perspective, it estimates for each environment the range of species it can contain. There are several indications testifying to the ecological plausibility of our model. First, the level of population of different environments is compatible with ecological knowledge: soil bacteria inhabit the most densely populated environments, and obligatory symbionts inhabit sparsely populated environments [19‐21] (Figure 1, Preliminary results). Second, the species‐specific level of environmental diversity is in strong agreement with acceptable, general measures of environmental diversity [20, 22]. This new ability to produce a genomic driven environmental model lays down a computational foundation for the study of a variety of aspects of the communal metabolic life. Notably, co‐inhabitation involves additional facets of interspecies interactions beyond competition, such as cooperation and symbiosis. In the following section we will discuss our vision for further improving this environmental model to describe more complex community structures, as well as promising applications of this extended model.

(II) Modeling bacterial community structure

In previous studies we have developed an ecological model predicting the range of species an ecological niche can populate. This list describes potential inhabitants but it ignores the complex relationships within a bacterial community. Bacterial species have been long thought of as freeswimming planktonic organisms which live relatively independent unicellular lives [2]. Unlike this perception of bacterial life, a rapidly expanding body of research shows that microbes exhibit a variety of social behaviors involving complex systems of cooperation and labor division [2, 23]. Hence, relationships within a community of species sharing limited resources can be described in terms of competition and cooperation where the identity of the actual community members changes according to the pattern of the interactions. We aim to develop an eco‐metabolic model, describing communities according to both the characteristics of the environment and the characteristics of the inhabitants. In a given community, we aim to obtain a context‐dependent description of the community structure and composition, considering environmental variations (i.e., changes in the available metabolic resources). By estimating the relative fitness of the members of the community under given conditions we will produce a description of the community structure and composition. We will analyse the ecological strategies that contribute to the organisms’ relative fitness and analyse their evolutionary implications. The ultimate goal is to be able to manipulate bacterial communities to our advantage. The approaches taken toward realizing this vision are broadly described in the Research Plan.

(III) Implementation of the ecological metabolic model for studying the human gut community structure.

As a case‐study we will apply our modeling approach to study shifts in balance of the bacterial gut community; at the same time we will analyze the medical implications of variations in community composition. We have chosen to focus on the gut because, first, metagenomic sampling provides a relatively comprehensive description of the nature of the species occupying this niche [5]. Second, tissue‐specific metabolic models for humans are now becoming available [24]; such models can provide information on the nature of compounds produced and consumed by the gut tissue, hence providing a description of the metabolic environment. Third, variations within the gut community are of considerable medical importance, making its modeling a high priority [6, 25‐27].