In nature, microorganisms live by interacting with each other. Microbiological studies that only consider pure cultures are not sufficient to adequately describe the natural behaviour of microbes. Several microbial interactions have been recognized to affect the growth or metabolism of others; e.g. syntrophic cometabolism, competition, production of inhibitors or activators, and predation. It is believed that third-party organisms easily affect the two-species relationships and these relationships form the basis of interspecies networks within microbial communities. A microbial network contributes to ‘functional redundancy’ or ‘structural diversity’ and the microbial communities effectively act as a multicellular organism. It is necessary to understand not only the physiological activity of members within microbial communities but also their roles to regulate the activity or population of others. To access the microbial network, we require (i) comprehensive determination of all possible interspecies relationships among microbes, (ii) knock-out experiments by which certain members can be removed or suppressed, and (iii) supplemental addition of microbes or activation of certain members. Microbial network studies have started using defined microbial communities, i.e. a mixed culture that is composed of three or four species. In order to expand these studies to microflora in nature, microbial ecology requires the help of mathematical biology.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Various molecular techniques have been applied to microbial ecology in the past two decades. These studies have provided new insight into environmental microbiology, i.e. by revealing the hidden diversity of the prokaryotic world and by identifying a variety of interspecies relationships. It has been suggested that over 10 000 prokaryotic species exist in 1 g of soil (Curtis and Sloan, 2004) although for prokaryotes approximately only 7000 species have been formally described for the whole planet. Metagenomic studies are expected to reveal even greater microbial diversity (Rodríquez-Valera, 2002; Hugenholtz and Tyson, 2008).
Since the 1950s, some researchers have noticed that physiological characteristics, such as growth rate, will differ between the test tube at the laboratory and in a natural environment (see for example Hattori and Furusaka, 1959). Consequently, many microbiologists have recently focused on ecophysiology in addition to the discovery of novel microbes. The discrepancy between the culture-tube and the natural environment may be due to complex and unexpected environmental physicochemical factors, for example, heterogeneous soil particles create microniches. In addition, the effects of other organisms in the environment need to be considered. Antagonistic relationships through competition for nutrients or production of antibiotics have been well recognized to understand survival advantage of a prokaryote in nature. Today, cooperative interactions such as syntrophy and growth promotion by chemicals are also of interest. Recent research using two-species mixed cultures has identified many types of signalling molecules responsible for cell–cell communication (Camilli and Bassler, 2006). Microbial ecology is moving into the next era by taking on the challenge of comprehending microbial relationships as they exist among numerous highly diverse microbial species in a natural environment.
Systems analysis has been applied as a method to tackle diverse complex systems, including metabolic flux, economics and computer science. This methodology has also been applied to biology; life science in the 21st century is characterized by systems biology. However, in order to unveil microbial networks, microbiologists have largely pursued a reductionistic approach by determining the physiology of individual species and describing single interactions. Once sufficient knowledge is accumulated, it is then combined piecemeal to build up a description of the microbial network. Such an approach is practical but it is not enough to properly describe natural microbial communities. Here we review advanced research attempting to expand this conventional reductionistic methodology in order to more accurately elucidate the nature of microbial networks.
Microbial interspecies interaction
The relationship between two microbial species has been described as competition for nutrients or as a prey–predator interaction. Most research attempts to assess the population size of a member in the microbial community by comparing the growth yield of each member. These approaches are supported by mathematical biology which utilizes simulation models such as the Lotka–Volterra equation.
Today, however, we notice that microbes affect each other not only through growth yield or rate but also through their metabolism (e.g. symbiosis) or via signalling molecules (e.g. antibiotics, growth promoting factors, pheromones). Moreover, direct contact often changes bacterial behaviour, such as in biofilm formation. Bdellovibrio sp. predates the Gram-negative bacteria Escherichia coli by directly attaching to the prey cell and subsequently replicating within the prey (Koval and Bayer, 1997), and certain E. coli strains inhibit the growth of other E. coli strains in a contact-dependent manner (Aoki et al., 2005). Interkingdom interactions of these types are also observed (see for example, Hogan and Kolter, 2000). Recently, molecular (micro- or nano-) scale studies provided new insights into interspecies interactions. For example, direct communication in symbiotic relationships is found in several anaerobic environments. Ishii et al. observed a cell-to-cell connection using nanowires in syntrophic methanogenic cocultures (Ishii et al., 2005; Gorby et al., 2006). There have been numerous studies of small-molecule signalling as reviewed by Camilli and Bassler (2006). Antibiotics have also been demonstrated to work as signalling molecules at low concentrations that affect the biofilm formation of other microbes, even though they are traditionally believed to work as weapons to remove others (Linares et al., 2006).
Coexistence with other organisms induces changes in the environment and affects microbial behaviours. In a wide sense, these are also addressed as an incidental interaction among species. Microorganisms adapt to environmental changes not only through chemotaxis or metabolic adaptation but also through genetic evolution (e.g. Ibara et al., 2002; Baker et al., 2006). Organisms connected by interspecies interactions have probably co-evolved as found in the mutualistic relationship between hummingbirds and ornithophilous flowers. Co-evolution may also take place in prokaryotes, which are in close association with other organisms or species such as symbioses (e.g. nitrogen-fixing bacteria in leguminous plants and prokaryotes with certain protists) (Okazaki et al., 2004; Ohkuma, 2008) and morphologically structured consortia (e.g. biofilm and cell aggregate) (Hansen et al., 2007). The current emphasis of interspecies communication is to understand theoretical and evolutionary aspects in addition to the signalling mechanisms and novel signal discoveries (for recent review see von Bodman et al., 2008).
Despite these advances, little effort has been made to answer some important questions: How do interspecies interactions regulate population size? How do interspecies interactions work in the microbial community? How do interspecies interactions affect the whole community? Mathematical simulations propose some theories; for example, positive interactions among competitors may contribute to biodiversity (Gross, 2008). One possible approach to obtain answers to these questions is the specific interference of certain interactions in the microbial community. If a defined microbial community is available, we can readily inhibit or block the interaction of interest, e.g. replace a member with a mutant strain deficient in some way, or by physically separating members with a membrane filter. Clarification of the mechanisms of interspecies relationships makes it possible to break a particular connection between microbes. Enzymes capable of inactivating signalling molecules will be a useful tool to realize this approach, as quorum sensing is quenched by lactonases (Riaz et al., 2008). The effects of suppression of certain interactions in the community can then be evaluated by molecular ecological techniques such as denaturing gradient gel electrophoresis (DGGE), single strand conformation polymorphism (SSCP), T-RFLP and so on (Kowalchuk et al., 2004). In addition, quantitative analyses by quantitative PCR or fluorescence in situ hybridization (FISH) make it possible to determine the growth curve of each member in the microbial community.
A third party changes the two-species relationship
Ecosystems composed of three species are completely different from two-species relationships. Three-species communities are often difficult to simulate mathematically (Huisman and Weissing, 2001; Li, 2001). Involvement of the third party produces unpredictable indirect effects among the members. Such effects have been described in macroecology, e.g. the predatory relationship found between sea stars and several shellfish species in a bay is the most representative example demonstrated by field experiments, where the sea star Pisaster was shown to indirectly maintain the coexistence of diverse species of shellfish by its selective predation (Paine, 1966). Cabbage plant (Brassica oleracea) – larvae (Pieris rapae and Plutella xylostella) – parasitic wasps (Cotesia plutellae and Cotesia glomerata) is another striking example. Shiojiri and colleagues (2002) untangled this interaction web in which organisms are connected by competition, interkingdom signalling molecules, predatory interaction and parasitism, showing the indirect effect among organisms on the cabbage plant. Microcosm experiments using zooplankton have been widely applied to understand the population dynamics of ecological systems. These studies have recently expanded from a two-species mixed community to a three- or four-species community. Yoshida and colleagues (2003) built a microcosm with two zooplankton (rotifier) genotypes and two green algal species to show the mechanism for coexistence of these organisms not only from their selective prey and the resulting growth yields but also from evolutional aspects.
It is easy to imagine that interspecies interactions are affected by third-party organisms in prokaryotic ecosystems. However, only a few studies have focused on the importance of third-party organisms in the prokaryotic community. Non-transitive relationships like rock-paper-scissors have been demonstrated with three strains of E. coli; a colisin producing strain (P), a colisin resistant strain (R) and a colisin sensitive strain (S) (Kerr et al., 2002). The order of their growth rate was strain S > strain R > strain P, but strain P could overcome strain S by colisin production. Their coexistence was achieved by local dispersal of each strain on a laboratory dish. Similarly, Narisawa and colleagues (2008) demonstrated that the local dispersal of bacterial species made it possible to coexist in a biofilm. They showed that an antibiotic producing bacterium and a sensitive bacterium coexisted in the biofilm in the presence of a third antibiotic-resistant species. In the biofilm, the sensitive species was separated from the antibiotic producer by a covering layer of the resistant species. Biogeography for macro-fauna and macro-flora has traditionally studied on these spatial effects to explain the biodiversity through both experimental and theoretical approaches (MacArthur and Wilson, 1967). We may apply their findings to microbiogeography.
A simple example under planktonic conditions was reported for predation by Bdellovibrio sp. (Hobley et al., 2006). Predation efficiency was reduced by the existence of a non-susceptible decoy bacterium such as Bacillus sp. Moreover, protease production by the decoy increased the concentration of amino acids which promoted the growth of the prey. Kato and colleagues (2008) also found that lethal effects detected between the members of a mixed culture were suppressed by a third-party bacterium. This bacterium in the culture improved fitness of the weak by supplementation of nutrients to the weak (K. Yamamoto, S. Haruta, S. Kato, M. Ishii and Y. Igarashi, unpublished).
As shown by these studies, the existence of third-party organisms can change two-species interactions in microbial systems. These three-species interactions thus form the smallest unit considered to be an intertwined network. When we focus on third-party organisms in various microflora, novel observations will be obtained, although experiments designed to evaluate such third-party interactions in the prokaryotic community are currently limited to a defined mixed culture. This bottom-up approach to increase the number of species in a defined community will challenge us to untangle the complex microbial network.
Approach to interspecies interaction within microbial communities
The fundamental question for microbial ecologists is how microbes and their interspecies interactions affect the whole community. Microbial ecologists can generally follow the approaches utilized for biochemical study to answer this question. In order to clarify the function of a protein of an organism, biochemists apply the following approach: (i) isolate the protein or gene encoding the protein, and (ii) characterize the native or recombinant protein in vitro. An analogue of these two approaches is applicable to the microbial community, i.e. isolation and characterization of the microbe desired (although we sometimes fail to obtain isolated cultures). However, as mentioned previously, physiological properties determined under pure culture conditions do not always reflect the properties within a community (see for example, Nakamura et al., 2004).
In vivo function of a protein or gene of interest is evaluated by analyses of mutants in which the target gene is knocked out or in which the target gene is overexpressed. The latter approach seems to be practicable in microbial ecology by the addition or activation of a microbe in the microbial community. Bioremediation utilizes this type of approach, i.e. bioaugmentation and biostimulation. However, the effects resulting from bioremediation on the indigenous microflora are too diverse to be understood totally. Furthermore, additional microbes, even those originally derived from the same microflora, often fail to establish or do not persist for very long (Watanabe, 2001). As a result, the addition of microbes or activation of a member species does not appreciably affect the indigenous microflora, indicating that complex microbial ecosystems have structural resilience. What is needed is a way for the addition or activation of microbes that has considerable and persistent effects on the indigenous microflora. One potential compromise to allow the study of such effects is to utilize non-natural uncomplicated microflora such as enrichment cultures in laboratories (e.g. Narisawa et al., 2007).
The other approach, equivalent to a knock-out experiment, will directly tell us the roles of the component microbes in the microflora. How do we remove or suppress a microbe within the microflora? One of the ways is the specific suppression or elimination of a member using a chemical approach. Antibiotics are readily available, but their efficacy and specificity are not sufficient. Specific types of metabolism can be inhibited by chemicals, such as 2-bromoethanesulfonate (an inhibitor for the methanogenic reaction) (Glissmann and Conrad, 2000) and molybdate (an inhibitor for biological sulfate-reduction) (Okabe et al., 2003). The involvement of viruses in microbial population dynamics has been recognized, and species-specific viruses may offer a way to manipulate microbial assemblages (Weinbauer and Rassoulzadegan, 2004). Unfortunately, it is not easy to find viruses that infect specific target microorganisms.
If we identified all the members or all the environmental genes completely, we would see the light at the end of the tunnel. However, such identification is not practical. As the pioneers of environmental microbiology mentioned, understanding microbial diversity is a formidable issue (Brock, 1987; Wilson, 1994). The first step towards this understanding will be the analyses of defined mixed cultures. Such a model community will be useful to look for the way to analyse (e.g. to find a hub in a complex web of microflora) and to develop and test primitive theories about microbial networks.
A valuable model system requires the following properties: (i) all the members are isolated; (ii) the members stably coexist under certain conditions; (iii) the community has a simple metabolic function; and (iv) cultivation conditions are homogenous (no specific niche). Practically, a chemostat culture is a favourable feature. Unfortunately, we are not aware of any prokaryotic community which satisfies all these properties, although the first three are defined for a stable mixed culture reported by Kato and colleagues (2005; 2008). They reported a stable mixed culture system composed of four or five bacterial species. Their study is an attempt to assess the roles of each member in the mixed culture using a knock-out approach. In their approach, knock-out communities such as a four-species mixed culture in which one of the members was eliminated were compared with the original five-species mixed culture. In situ roles of each member and relationships among the members of the community were evaluated. The network relationships proposed for the community indicated that suppressive interactions regulated the overgrowth of a particular member leading to the structural and functional stability.
As the cell density of a pure culture is determined by intraspecies interactions, i.e. quorum sensing (Juhas et al., 2005), space occupied by one species seems to reduce the population of the other. Competition for space can easily be imagined to occur, for example under biofilm conditions. It may also occur in homogenous liquid cultures. Kato and colleagues (2008) showed that the maximum total cell density in liquid medium was not affected by the number of species in the defined mixed cultures (Fig. 1). Their findings, through comprehensive analyses of microbial interactions, indicated that the microbial network in their in situ experiments possessed feedback and/or feedforward loops that provided elements of regulation to the system. This is a top-down approach using a rather simple defined community. These top-down approaches will be more useful when more complicated defined community is available.
Characterization of a complex microbial community
Although useful, experiments using defined mixed cultures are not enough to properly study microbial ecology. We should keep in mind the fact that global reservoirs of diversity are an important feature of natural ecosystems (Curtis and Sloan, 2004). In the case of a highly diverse community, the effects of a knock-out of a single species may be compensated by alternative species. This was partly suggested by the knock-out experiments of a four-species mixed culture (Kato et al., 2005). The local dispersal of microbial cells is also one of the factors that raise profound questions in microbial ecology, as described above.
Experimental studies performed on a natural microbial community rather than a defined mixed culture should be developed as a fundamental approach to microbial ecology. This type of study can pay some attention to the properties of microflora as a system, such as stability, redundancy and resilience. For example, the effect of diversity and flexibility of microflora structure on the stability of community function has been discussed with a natural microbial community by Fernandez and colleagues (2000). Socio-microbiology has been proposed as a keyword to describe the multicellular nature of microbial communities (Jacob et al., 2004; Parsek and Greenberg, 2005). Accordingly, physiological profiles of microflora have been determined by community level physiological profile (Insam et al., 2001).
In the 1990s, artificial neural networks (ANN) were applied to ecological modelling (Lek and Guégan, 1999). Artificial neural networks predict the output of input data through the learning process of non-linear and complex data. Process engineers initially applied ANNs to microbial communities in bioreactors (e.g. Holubar et al., 2002). The ANN treats microbial communities as a black box to predictively characterize their activity; the ANN tells us what the best condition (pH, temperature, agitation, supplemental nutrients, the loading rate and so on) for the reactor is. The quality of the results absolutely depends on the quantity and variety of data for learning. Biology has entered the omics age to comprehensively access the black box, i.e. metagenome, meta-transcriptome, meta-proteome, and meta-metabolome. Studies on biodegradation of chemical pollutants have started to utilize these omics approaches (Gómez et al., 2007; de Lorenzo, 2008). Their systemic analyses provide a global metabolic network encompassing all possible reactions of the microbial community. This will predict the biodegradability and potential fate of pollutants in the environment, thereby helping to devise appropriate strategies for bioremediation. Mathematical simulation of population dynamics, applicable under limited conditions, is also a tool useful for improving our understanding of the behaviour of microbial communities (see for example, Matsumoto et al., 2007). Even so, it is still hard to explore the actual population dynamics of microbial communities because of the complexity of interspecies interaction and the fluctuation of microbial behaviour (discussed below).
From a conceptually different point of view, a three-species mixed culture was studied by Becks and colleagues (2005). They successfully observed chaotic behaviour of the populations in competitive relationships indicating the unpredictable behaviour of the population dynamics with great sensitivity to subtle changes in conditions. Studies on gene regulatory networks are focusing on noise in the system (Thattai and van Oudenaarden, 2001; Sato et al., 2003). A microbial community that has an intrinsically noisy network may show greater adaptability to natural environments. Complex systems research will help to develop new insights into microbial ecology. In addition, theoretical studies on complex ecosystems may completely supersede the reductionistic analyses of interaction between organisms. For example, Hubbell (2001) proposed the neutral theory that states biodiversity is not derived from interspecies interactions. Biologists continue to discuss the possibility of theoretical studies in microbial ecology (Prosser et al., 2007).
We are struggling to comprehensively determine all possible interspecies relationships within microbial communities. These efforts certainly bring us novel findings about microbial relationships. The knock-out or addition of a member within defined mixed cultures promotes a better understanding of the structure of connections between microbes. These studies will steadily lead us towards accurate descriptions of both microbial networks and their component interactions. Network analysis will tell us the basic organizational principles of microbial ecological systems with the help of network theory as proposed by Trigo and colleagues (2009). The visualization of microbial networks will illuminate hubs, which are key microbes or microbial behaviours that let us comprehend the complex microbial systems in which they are found.
Microbial ecologists should not dismiss the complexity of microbial behaviours as represented by the following two aspects. (i) Behaviours of microorganisms are not invariable during their growth even under constant conditions. Quorum sensing and sporulation are well-known examples. Chemotaxis, metabolic adaptation and genetic evolution are fascinating topics for microbiologists (e.g. Ibara et al., 2002; Baker et al., 2006; Hansen et al., 2007). (ii) Microbial cells are not homogenous even in a clonal population. Until now, most data we have obtained for microbial cultures are averages of huge numbers of cells, but microbial cells in a clonal population have phenotypic diversity (Smits et al., 2006). For a marked example, some Bacillus subtilis cells in culture transiently become competent for DNA uptake (Nester and Stocker, 1963). Suel and colleagues (2006) utilized fluorescent proteins as a reporter to visualize the gene regulation at a living single-cell level as a time-lapse photography, and clearly demonstrated that the transient cellular differentiation in B. subtilis was induced by excitable dynamics driven by noise in gene regulation. Their study using single-cell analyses successfully proceeded in a close collaboration between biology and mathematics. Advanced microbiological research like these examples makes us aware that microbial ecology should move towards characterizing the physiological diversity of cells in situ in addition to obtaining the phylogenetic diversity of species or strains. These may give us more profound insights into microbial relationships in microflora.
It had been somewhat challenging to apply single-cell or in situ analyses to microbial communities until recent technical developments, i.e. specific labelling (e.g. in situ hybridization, in situ RT-PCR and microautoradiography) and in situ detection (e.g. microscopy, flow cytometry, density centrifugation and microelectrophoresis). Even reporter protein analyses will be applied to microbial communities in the future by the deliberate addition of a genetically modified microbe and by observation of the cells during cultivation on a microscope stage. Conceivably, phages would be applicable to introduce the gene encoding a reporter protein into a targeted microbe within the microbial community. In the immediate future, the approach outlined here will allow researchers to obtain enormous amount of high-quality data about microbial behaviour. From there, bioinformatics and novel theories will be required for analysis. Systems microbiology treats the community as a whole by integrating fundamental biological knowledge about the components of the network. As we continue to progress in ecophysiological studies of component strains or cells, we need to begin to communicate with research groups in the fields of mathematical biology, systems biology and computer science. Through this interdisciplinary approach, we can develop novel analytical tools and theoretical interpretations of our experimental observations.
We are very grateful to Craig Everroad for critical reading of the manuscript.