Principles for Designing Synthetic Microbial Communities

Abstract

Advances in synthetic biology to build microbes with defined and controllable properties are enabling new approaches to design and program multispecies communities. This emerging field of synthetic ecology will be important for many areas of biotechnology, bioenergy and bioremediation. This endeavor draws upon knowledge from synthetic biology, systems biology, microbial ecology and evolution. Fully realizing the potential of this discipline requires the development of new strategies to control the intercellular interactions, spatiotemporal coordination, robustness, stability and biocontainment of synthetic microbial communities. Here, we review recent experimental, analytical and computational advances to study and build multispecies microbial communities with defined functions and behavior for various applications. We also highlight outstanding challenges and future directions to advance this field.

Introduction

Genetically modified microbial organisms are used in many applications in industrial and environmental biotechnology, from synthesis of materials, chemicals medicines, and fuels, to remediation of waste products and toxins. Recent advances in synthetic biology have substantially improved our ability to program these microbes quickly and cheaply on a large scale with greater control [1,2]. While many successes are documented for single-step microbial bioconversion reactions [3], potential applications that involve complex substrates may require the use of multiple pathways and processes, which may be difficult or impossible to execute efficiently using single strains. These and other complex applications may be best tackled by cohorts of different microbes, each programmed with specialized sub-functions that synergize towards an overall population-level function. This fact is evident in natural systems where single species do not occupy all niches in an environment, but rather multiple species coexist and perform complementary roles, creating intricate ecological networks [4].

With a greater understanding of natural microbial interactions, dynamics, and ecology, we are poised to expand microbial engineering to mixed consortia in order to perform more complex and challenging functions in both closed and defined bioreactors as well as open and natural environments. This emerging field of synthetic ecology builds upon gene circuit design strategies [5] and further integrates ecological and evolutionary principles [6]. These population-scale considerations involve microbial interactions with complex dynamics and stability properties manifesting over different time and length scales. In this perspective, we explore how these properties can be applied to design and construct microbial communities relevant to emerging biotechnology applications (Figure 1). We specifically discuss four key considerations for building synthetic microbial communities: engineering various interspecies and intraspecies interactions, constructing spatiotemporal dynamics, modeling and maintaining community-wide functional robustness, and developing population control and biocontainment measures. We discuss recent examples of experimental and quantitative modeling advances that have enhanced our foundational capabilities to understand, develop and exploit synthetic microbial consortia in different settings.

Figure 1.

A summary of the design and utility of synthetic microbial communities. In addition to principles used in single-strain engineering, community engineering allows for diversification of biochemical roles in breaking down complex substrates, and optimized compartmentalization of pathways between individuals for simultaneous execution of multiple functions with reduced individual burden. Synthetic communities can be further engineered with increased robustness through interdependencies and spatiotemporal control.

Engineering Intercellular Interactions

Organisms in nature interact with one another through a variety of modes ranging from competitive or predatory behaviors to commensal and mutualistic exchanges that have been extensively explored in ecological studies [7]. Microbes living within communities are involved in many interactions simultaneously – competing for some resources while exchanging others. Over time, these tradeoffs create interspecies dependencies manifested by differing specialized phenotypes across various microbes. A key challenge for engineering consortia with stable interactions has been to understand how functions can be partitioned across a microbial population in productive compartments to achieve desirable population-level behaviors (Figure 2a).

Figure 2.

Key principles for engineering microbial communities. (A) Various metabolic interactions can be designed and leveraged for multispecies production of a desired product. (B) Communities can be spatially and temporally coordinated through engineered environments and programmed aggregation behavior. Quantitative in silico modeling of structured environments will improve the design of these consortia. (C) Population robustness can be maintained through strategies that enhance cooperation, avoid cheating, and promote non-metabolic stabilizing interactions, such as antibiotic antagonism. Various modeling approaches are need to study the dynamics and stability of the systems. (D) Biocontainment methods use synthetic auxotrophies or kill switches to control growth and function of engineered microbial communities.

Several studies have recently explored strategies to divide metabolic roles across different individuals in a consortium toward generation of a desired biochemical product. Minty et al. showed that a two-member microbial consortium containing a cellulase-secreting fungi, Trichoderma reesei, and an engineered E. coli strain that produced isobutanol could be used for the direct conversion of plant biomass into biofuels [8]. Zhou et al. engineered an E. coli and S. cerevisiae consortium that can more effectively produce natural products than the individual strains alone [9]. In this study, the biosynthetic pathway for oxygenated taxanes, a medically valuable diterpene chemotherapeutic, was partitioned into two separate pathways in E. coli and S. cerevisiae. A taxadiene intermediate was produced and secreted by E. coli and then taken up by the yeast to complete the necessary oxidation steps using more optimal eukaryotic cytochromes. Studies like these demonstrate that consortia containing members with specialized tasks can succeed in applications where single-strains would struggle.

A number of studies have explored syntrophic interactions using model bacterial and yeast systems that exchange essential metabolites [6]. Each strain is engineered to produce some but not all essential metabolites (e.g. amino acids). When these different auxotrophic strains are grown together, those with complementary metabolic functions are able to support the growth of one another as a syntrophic co-culture. These systems have been characterized in synthetic communities of two [10,11], three, and up to 14 members [12]. A key observation from these studies is that metabolically costly resources are more likely to be involved in syntrophic exchanges. While metabolically-dependent specialist strains may be sensitive to some environmental perturbations (e.g. nutrient depletion), they can outcompete generalist cells in some environments due to more optimized metabolic configurations [13]. In fact, these metabolic dependencies may be prevalent in natural communities, with many sequenced genomes missing multiple essential biosynthetic pathways [12,14]. Furthermore, the magnitude and direction of metabolic exchange is potentially tunable by modulating membrane transporters and intercellular nanotubes [15].

Beyond metabolism, many intercellular interactions in bacterial ecosystems are mediated by the secretion, diffusion and exchange of diverse molecules including peptides, small-molecules and natural products. These compounds are used by bacteria to sense their environment and communicate with surrounding cells. Quorum sensing, a general mechanism by which bacteria produce and respond to specific signaling molecules in a density-dependent fashion, offers a means to program cell-cell communication and coordinate population-level behavior. These simple bacterial communication systems possess modular and engineerable features that have been exploited extensively for rational design [16]. In a recent example, Saeidi et al. engineered an E. coli strain that was able to sense naturally produced quorum sensing molecules from P. aeruginosa and respond by turning on a self-lysis kill-switch to release a pyocin compound that inhibited the growth of the target pathogen [17]. Many opportunities exist to extend these approaches for cooperative intercellular interactions.

To model these microbial interactions, computational and genome-scale approaches may be used to inform the rational engineering of simple and complex consortia. Constraint-based methods such as flux balance analysis (FBA) use in silico metabolic reconstructions of cellular metabolism based on a set of known stoichiometrically balanced reactions to assess steady-state metabolite fluxes within the cell during growth. When extending these approaches to model synthetic or natural communities, each cell can be compartmentalized and fluxes between the compartments can be evaluated across the population to predict community-wide behaviors [18]. Currently, for most microbial species, incomplete or missing information about the metabolic network components and their biochemical functions make FBA-based methods challenging. Furthermore, drawing realistic inferences from these models requires constraints for maximizing community-wide objective functions (e.g. growth), which may be difficult to define or can change on an individual level across a dynamic community. Nevertheless, these approaches can be useful for predicting metabolite flux and exchange in microbial communities. Nagarajan et al. recently constructed a metabolic model of two Geobacter species that parameterized their metabolite exchange and direct electron transfer to characterize their syntrophic growth dynamics [19]. Such a model system may have useful applications in bioremediation and microbial fuel cells. Beyond studying physiology of naturally interacting microbes, computational tools are needed to better predict the behavior and impact of engineered genetic pathways on community dynamics. McClymont and Soyer developed Metabolic Tinker, a graph-based tool that can identify thermodynamically-feasible biochemical routes to a desirable compound [20]. Such approaches can help predict and design new metabolic interactions between synthetically engineered microbial consortia. Integration of biochemical, transcriptomic, proteomic, and metabolomic data [21] will help to better parameterize these in silico models to improve predictions for rationally designed communities [22,23].

An emerging area of quantitative models is the use of economic principles to study microbial trade [24]. The vast diversity of microbial metabolic capabilities offers opportunities for production and exchange of specific metabolites between two or more microbes that can be mutually beneficial. Similar processes underlie modern economic markets where businesses and nations produce and consume goods and can improve efficiency through trade. Furthermore, economic concepts such as specialization, vertical integration, market competition and even farming have analogous processes in the microbial world [25]. The extensive literature in economic theories and frameworks can be adapted to model trade in microbial communities using analogous parameterizations for productivity, utility, import and export rates, and growth maximization. We and others recently applied the economic principle of comparative advantage to microbial trade [26,27]. Based on general equilibrium theory, we showed that trade in microbial community can be stabilized when trading agents benefit from the exchange of resources that they are relatively more efficient at producing [26]. These analytical predictions can be experimentally explored using simple bacterial models, yielding important insights into the stability, efficiency, and design principles of microbial metabolic exchange. A key challenge moving forward will be parameterizing these models using data from realistic environmental settings [28]. On the other hand, these model microbial communities can offer economists a new tool to test economic theories and hypotheses that are difficult to implement in real economic settings [29].

Understanding Interactions in Space and Time

While most synthetic microbial communities developed to date have been studied in well-mixed co-cultures, many microbes in nature exist in spatially defined structures such as surface-attached biofilms. Spatial assortment of cells creates locally heterogeneous subpopulations with varying resource availabilities that strengthens local interactions, avoids global catastrophes such as the tragedy of the commons [30], and improves resilience to environmental stresses [31].

Several general approaches have been explored to build spatially defined microbial communities by organizing the physical environment, patterning specific community structures, or engineering cells with programmed aggregation behaviors (Figure 2b). Microfluidic and microwell devices have been used to build microbial communities where individual species are grown in separated chambers that allow metabolites to exchange freely, but restrict physical contact between cells [32]. Other strategies using micro-contact printing techniques allow specific members to be arranged in defined geometric patterns on two-dimensional surfaces [33]. Additionally, 3D-printing technologies have recently enabled construction of microbial communities with more complex structures [34]. To engineer surface attachment interactions, Nguyen et al. developed a nanofiber display platform to functionalize the extracellular matrix of microbial biofilms [35]. By fusing metal and nanoparticle binding peptides to an amyloid protein, they were able to program E. coli biofilms to adhere to specific abiotic surfaces and particles. Furthermore, programmed aggregation behaviors have been demonstrated in a two-member consortium that sequentially colonized a surface and could be inducibly dispersed to clear the engineered biofilm.

Beyond experimental systems for spatial control of engineered communities, a number of computational and in silico methods have emerged to better model communities in structured environments. Harcombe et al. recently developed COMETS, a dynamic flux balance framework that simulates microbial growth on a two-dimensional surface [36]. This approach accurately predicted the steady-state abundances of a three-species consortium grown in defined medium. Furthermore, experiments validated predictions that certain spatial distributions of competing colonies and crossfeeding partners can lead to counter-intuitive growth benefits. Other approaches using agent-based modeling frameworks have been explored to assess metabolic and population feedbacks in structured environments [37] and to model interspecies interactions during biofilm formation [38]. These emerging computational and experimental advances will enable more sophisticated design and control of consortia across space and time.

Maintaining Community Robustness

An important engineering consideration is the long-term stability of microbial consortia in challenging and open environments where engineered populations may experience changing conditions and exposure to competitive species. Furthermore, these consortia will change over time due to genome evolution and horizontal gene transfer [39]. In individual strains, engineered genetic circuits can lose function even on short timescales [40]. Loss of engineered function can lead to cheating (i.e. utilization of common goods without reciprocal contribution) and nonproductive phenotypes, which decrease population-level performance. To mitigate evolutionary decay of genetic circuits, strategies to reduce host mutation rates and avoid mutation-prone designs have been suggested [41]. The maintenance of community robustness and function over operationally-useful timescales is a key challenge for deployment of multispecies consortia in complex settings (Figure 2c).

To engineer stable consortia with defined function, strategies need to be developed for surveilling and enforcing cooperative or synergistic community properties at the level of individual members. While cheating phenotypes can contribute to population instability and reduced consortia performance, various strategies have been explored to undermine their emergence. Spatial self-organization can promote cooperative behaviors by excluding cheaters and invaders [42,43]. Cooperators can also avoid cheaters by responding to environmental cues for advantageous times to produce public goods [44]. Furthermore, cooperative interactions can be reinforced when genes encoding these traits are actively transferred between cells [45]. Engineering biosensors that monitor the presence of trading partners and privatizing metabolic exchange through intercellular nanotubes between cells [46] could also be used to enhance cooperative interactions.

Competitive and antagonistic interactions from native species pose another challenge for engineered consortia. For relatively simple consortia composed of a small number of species, unoccupied metabolic niches may lead to colonization of invasive species. A recent theoretical analysis suggests that competitive interactions are crucial for population stability in highly diverse communities [47], posing a challenge for current synthetic consortia of limited population diversity. In addition to competition, antagonism (e.g. antibiotic production and degradation) also promotes coexistence of competing species [48], highlighting the importance of non-metabolic interactions. Furthermore, stochastic events can also create population fluctuations in mixed communities that destabilize community structure and composition [49]. Further basic research is needed to understand these dynamics and to develop engineering solutions to mitigate their adverse effects on desired community function.

Biocontainment and Population Control

Deployment of engineered microbial communities in open environments will require precise control of population growth. Furthermore, biocontainment of these systems will ensure that engineered functions are not released into and do not disrupt natural ecosystems, yielding unintended negative consequences [50]. To address these potential concerns, several groups have attempted to develop biocontrol and containment strategies to modulate growth rate, yield, and function (Figure 2d). For example, E. coli and yeast have been recoded to require supplementation of non-natural amino acids [51,52] or defined small molecules for growth [53]. By changing the levels of these externally supplied molecules, population growth could be precisely controlled. To further prevent the potential for engineered populations and their genes to escape into the environment, kill switch gene circuits have been developed to contain modular multilayered programmable input logic that can respond to different environmental conditions [54]. Ongoing concerns for the dissemination of engineered traits to natural populations through horizontal DNA transfer have led to strategies using CRISPR systems to precisely target and degrade defined sequences in the genome to prevent their possible escape [55].

Beyond biocontainment, precise population control may be desirable for executing cell density-dependent functions. To coordinate population-level function across various length and time scales, several approaches including the using of quorum sensing have been exploited to develop gene circuits that respond to population densities and modulate growth and function accordingly [56–58]. Regulation of amino acid export has been used to tune the abundance and membership ratios of a crossfeeding microbial consortium to control community-wide function [15]. Recent approaches to develop multicellular gene circuits enable execution of complex tasks and logic functions across multiple independently tunable and modular strains [59,60]. Scale-up and synthesis of these approaches for higher order systems and diverse hosts will be a new frontier for engineering precisely controllable microbial ecosystems.

Conclusions and Future Outlook

Synthetic ecology presents an exciting opportunity to leverage recent theoretical and experimental advances in synthetic biology, ecology, and computational biology to rationally engineer useful microbial consortia in a variety of environmental and biotechnological applications. With the recent revolution in genome engineering capabilities to manipulate microbes and higher-order organisms [2], the scale up of microbiome engineering to systems with more complex functions in dynamic environments is poised to become an exciting and fruitful endeavor for synthetic biology. New opportunities to engineer microbial communities in open and changing environments will require next-generation in situ approaches [61]. The bottom-up study of synthetic communities will likely yield a better understanding for natural microbial ecology by systematically evaluating individual parameters in a controlled environment in an iterative design-test-learn cycle. In turn, the exploration and characterization of new microbial ecosystems will further lend insights into the fundamental principles that enable the modeling and engineering of synthetic communities in many useful applications.

Highlights.

Synthetic microbial communities have many potential uses in industrial biotechnology.

Defined intercellular and spatiotemporal coordination enable precise consortia engineering.

Community robustness and stability should be assessed and designed into synthetic systems.

Deployment of biocontainment and population control measures are needed for real-world applications.