Engineering microbial consortia by division of labor
During microbial applications, metabolic burdens can lead to a significant drop in cell performance. Novel synthetic biology tools or multi-step bioprocessing (e.g., fermentation followed by chemical conversions) are therefore needed to avoid compromised biochemical productivity from over-burdened cells. A possible solution to address metabolic burden is Division of Labor (DoL) via natural and synthetic microbial consortia. In particular, consolidated bioprocesses and metabolic cooperation for detoxification or cross feeding (e.g., vitamin C fermentation) have shown numerous successes in industrial level applications. However, distributing a metabolic pathway among proper hosts remains an engineering conundrum due to several challenges: complex subpopulation dynamics/interactions with a short time-window for stable production, suboptimal cultivation of microbial communities, proliferation of cheaters or low-producers, intermediate metabolite dilution, transport barriers between species, and breaks in metabolite channeling through biosynthesis pathways. To develop stable consortia, optimization of strain inoculations, nutritional divergence and crossing feeding, evolution of mutualistic growth, cell immobilization, and biosensors may potentially be used to control cell populations. Another opportunity is direct integration of non-bioprocesses (e.g., microbial electrosynthesis) to power cell metabolism and improve carbon efficiency. Additionally, metabolic modeling and 13C-metabolic flux analysis of mixed culture metabolism and cross-feeding offers a computational approach to complement experimental research for improved consortia performance.
Keywords13C-metabolic flux analysis Cross-feeding Metabolite channeling Reporter protein Subpopulations
Microbial consortia can break up metabolic load among partners [11, 12, 13]. In the past, microbial consortia have been used for waste treatment, anaerobic digestion, the food/nutrient industry (e.g., dairy products, soy sauce, wines, and vitamins), as well as medical applications (e.g., gut microbiome). Recently, synthetic biology has successfully engineered co-cultures for the production of commodity chemicals (cis,cis-muconic acid)  as well as drugs (oxygenated taxanes) . Synthetic consortia can be formed from several strains of the same species, such as engineered E. coli strains which together produce high-value chemicals such as flavan-3-ols , curcuminoids , and anthocyanins . Current industrial applications of DoL using microbial consortia still face challenges in controlling population dynamics and optimizing productions. This review presents an overview of microbial consortia applications, limitations, and opportunities.
Microbial consortia interactions, maintenance and stability
Due to the complex interactions and dynamics of species within a community, consortia maintenance and stability is crucial for any successful applications (Fig. 1). Different microbes in consortia must grow in the same environment (temperature, media, pH, and oxygen) and the growth of one species must not destroy the other members in a short period. Since the growth rates of different species or different partner strains from the same species will not be identical, one species can eventually take over the culture. To balance the subpopulations, several approaches may be employed. First, the inoculation ratios for different partners must be optimized. Second, intermittent supplementation of underdog subpopulations may elongate the period of co-cultivation. With the aid of real-time detection systems , optimized population composition and bioprocess parameters can be closely monitored and maintained across the fermentation by feeding the desired cultures. Third, biosensors (e.g., quorum sensing through cell-to-cell communication) may potentially be used to control cell sub-population . Fourth, cell immobilization can be attempted (e.g., growing free cells of Pichia stipites and immobilized Zymomonas mobilis together for ethanol production) . Fifth, coexistence partners often compete for substrates, but nutritional divergence or syntrophy (one species lives off the products of another species) can be employed to avoid substrate competition. Such concepts have been widely applied for utilization of mixed substrates or cascade biodegradation of recalcitrant feedstock. Importantly, mutualistic growth is desirable for stable consortia applications. During this cooperation, a species benefits from the waste of another, while the waste producer might also receive costly resources in return. Such mutualistic consortia can show stable performance over prolonged cultivations . Moreover, in situ evolution can improve the robustness of mutualistic systems. For example, in a stable co-culture, Salmonella enterica evolved to secrete methionine, a costly amino acid, for an E. coli strain, while the E. coli evolved novel secretion of sugar to feed S. enterica [24, 25].
Applications of DoL for substrate utilizations
Microbial communities have been extensively studied for microbial ecology and waste treatment. Industrial biotechnology has applied the same principles to increase bioproduction from cheap feedstocks such as CO2 and cellulosic biomass. Synthetic consortia composed of bacteria and fungi are a particularly promising method for utilizing agricultural wastes to achieve waste-to-fuel/waste-to-food processes .
Mixed sugar fermentations
For cellulosic feedstock, xylose derived from the hemi-cellulose component cannot be used by model yeast platforms. To address this, a xylose-fed co-culture of E. coli and Saccharomyces cerevisiae has been developed, wherein E. coli (which can naturally use xylose) generate acetate as waste product; S. cerevisiae consumes the acetate and reduces its inhibition of E. coli growth . Similarly, consortia with different engineered S. cerevisiae strains can ferment glucose–xylose–arabinose mixtures .
C1 carbon utilizations
Consortia can serve as platforms for utilization of C1 substrates (such as methane, CO2, and CO). For example, Citrobacter amalonaticus can assist Sporomusa ovata to produce acetic acid from carbon monoxide . Moreover, methane is an abundant feedstock from anaerobic digestion or oil reservoirs. Archaeal–bacterial symbiosis could couple methane oxidation with sulphate reduction in gas-hydrate-rich sediments , which makes the anaerobic conversion of methane into bio-products possible. Among C1 substrates, biosequestration of CO2 has great potential. Algae, including both prokaryotic cyanobacteria and eukaryotic microphytes, can fix inorganic carbon for bio-production. However, algal growth and genetic tool development have lagged far behind that of classic heterotrophic hosts like E. coli and S. cerevisiae . It is possible, however, for the autotrophy of microalgae to be exploited more indirectly, in a modular system that supplies organic carbon for a chemical-producing heterotrophic host (Fig. 3). Such modularity (algae-bacteria system) has been laid in the development of a sucrose-exporting strain of the cyanobacteria Synechococcus elongatus PCC 7942, which heterologously expresses the proton-sucrose symporter cscB. Under conditions of osmotic stress, this cscB+ strain irreversibly secretes up to 85% of fixed carbon as sucrose . This cyanobacterial host has been utilized in synthetic microbial consortia with a variety of microbes, including model strains (E. coli, Bacillus subtilis, and S. cerevisiae), soil bacteria (Pseudomonas putida, Halomonas boliviensis), and diazotrophs (Azotobacter vinelandii) to produce a variety of products without additional sugar input [39, 40, 41, 42]. Autotroph-heterotroph coupling is not limited to cyanobacteria—binary culturing of the eukaryotic algae Haematococcus pluvialis with B. subtilis demonstrated a coupling of oxygenic photosynthesis with respiration, such that neither external CO2 nor O2 supplies were necessary, suggesting that some of the overhead costs associated with autotrophs could be reduced in a properly-configured production system . Similarly, co-cultures of eukaryotic algae Chlorella minutissima with E. coli can increase neutral lipid production and improve biodiesel quality . These studies facilitate the application of CO2-to-fuel technologies.
Nutrient crossing feeding using companion strains
Interacting partners can be designed to maintain stable interactions through detoxifying inhibitory substances. For instance, Dehalococcoides mccartyi is an important bacterium involved in the bioremediation of chlorinated solvents, but its incomplete Wood–Ljungdahl pathway generates toxic CO . Co-cultures with CO-consuming bacterium Desulfovibrio vulgaris Hildenborough as the companion can significantly support D. mccartyi growth and TCE degradation ability. In another example, l‐ascorbic acid (vitamin C) is produced via a two‐step fermentation process to reduce costs and increase product quality. In the second step, a mixed fermentation consisting of a production strain (Ketogulonicigenium vulgare) and a companion strain (e.g. Bacillus spp.) is used to convert l‐sorbose to 2‐keto‐l‐gulonic acid and ascorbic acid . The interaction mechanisms between K. vulgare and the companion strain have been studied, and it was found that the companion strain secreted metabolites to support K. vulgare growth (including proteins, some amino acids, and other unknown substances) while K. vulgare can also inhibit growth of the companion strain by lysis of companion cells (thus both mutualism and amensalism exist in this artificial ecosystem).
Division of long-step biosynthesis pathways among different species
Limitations of DoL and future directions
Culturing microbial workhorses as a community faces several key limitations. First, many microbial community systems are dynamic and cannot achieve long-term production stability. Applications of consortia systems require identifying both an optimal inoculation ratio and an ideal product induction time. Nevertheless, the stability of the co-culture is expected to decrease in large volume and long-term fermentation processes due to spatiotemporal dynamics inside of reactors , incompatible growth requirements (e.g., the cross-feeding is insufficient to sustain net growth of both partners), or metabolite dilutions . Second, two different species in co-cultures have to grow under suboptimal or compromised conditions. For example, to stabilize a co-culture of E. coli and S. cerevisiae, S. cerevisiae must be grown with acetate to avoid producing toxic ethanol . Since acetate is not a favorable carbon source for yeast, the final bio-manufacturing capability is impaired. Third, metabolite transport can be an issue if a pathway is divided, such that the first product must be transported into the second host for further conversion. If the first product (e.g., a phosphorylated molecule) is unstable or unable to cross the cell membrane, or the second host lacks a transporter to uptake the precursor from the first host, the DoL will fail. Engineering effective transporters and efflux systems is not easy and may also add metabolic burdens . Furthermore, long-step biosynthesis pathways often require proximity channeling to overcome in vivo diffusion limitation, metabolite loss, and thermodynamic barriers. DoL of a biosynthesis pathway between two different species may break innate channels and dilute intermediate concentrations, which unavoidably hinders biosynthesis efficiency  (Fig. 4). Fourth, microbial species may naturally evolve into distinct phenotypic subpopulations (e.g., E. coli may form two distinct cell populations with acetate cross-feeding) . In mutualistic consortia, cheaters (individual cells that do not help the other species) can be observed—cheaters face less metabolic burden and can achieve a faster growth rate, but when they become dominant, the co-culture will be unable to survive . For example, in a CBP-based isobutanol fermentation, fungus T. reesei and engineered E. coli consortia form stable cooperator–cheater dynamics . Interestingly, even in a pure culture of synthetic hosts, nongenetic cell-to-cell variations can be high, allowing low-production cheaters to eventually become dominant .
Current community-based biomanufacturing mainly leverages natural systems and focuses on utilization of recalcitrant or cheap feedstock. In these desired platforms, metabolic interactions (e.g., cross feeding) can be robustly evolved. Still, there are few industrial applications based on the division of biosynthesis pathways among different species. For future DoL developments, a few approaches are promising. First, biosensors can be used to monitor and control cell sub-populations. In addition to quorum sensing for population monitoring, an RNA riboswitch-based biosensor module has been developed to provide real-time screening for overproducing cells in co-cultures . Biosensor-regulator systems may be used for in vivo population quality control to continuously select for high-performing variants by suppressing the growth of cheaters . Second, state-of-the-art bioreactor operations with the aid of readily-available high-throughput tools and real-time detection systems allow for feedback control/optimization of population composition (supplements of desired culture) and bioprocess parameters (e.g., substrate feeding) to maintain co-culture fermentation for stable and efficient biomanufacturing . Third, the integration of non-bioprocess methods (Chem + electrobio) is another future direction . Microbial electrosynthesis is a novel technology which combines metabolic engineering and electrochemistry within a biological reactor and has the ability to achieve 100% carbon efficiency by generating reducing equivalents (in the form of NADH) from electrons . This type of electrochemical bioreactor may also capture and reduce CO2, thus moving carbon into the desired metabolic pathway. A similar concept has been used to combine hybrid inorganic and biological organisms for solar-to-chemical production [59, 60].
Analysis and modeling of consortia and metabolic interactions between species
There are many tools being developed to study consortia and DoL applications. Kinetic modeling is the most common method to analyze nutrient cross-feeding and microbial interactions . For example, E. coli and photoheterotrophic Rhodopseudomonas palustris cross-feed carbon (organic acids) and nitrogen (ammonium) to form a stable co-culture. Monod-based differential equations have been developed to reveal fermentation conditions for stabilizing mutualism [61, 62]. The model results indicate that organic acids from E. coli can be inhibitory and thus decrease carbon utilization efficiency and population equilibrium. For molecular-level understanding, functional genomics and high-quality metabolic modeling are needed to decipher cellular regulatory networks and integrated functions. For consortia, functional analysis can rely on typical microbiome tools, including targeted cell population analysis (such as 16S rRNA), metagenomic sequencing (DNA is recovered in an untargeted manner), metatranscriptomics (study the gene expression of microbial communities), and metaproteomics (study proteins expressed by microbiota to gain insight into functional potential). Microfluidic cell sorting can assist in the separation of individual species to study their DNA–RNA–Protein profiles under a community system. Moreover, measurement of metabolites within community systems can monitor the global community outcome and provide a snapshot of the general physiological state of microbes [46, 63].
To date, methods applicable for analyzing metabolic fluxes between two species (especially for two engineered strains derived from the same parent strain) are still rare. Currently, genome-scale models (GSMs) predict the feasible fluxes based on stoichiometry of the metabolic reactions, cellular objective functions, and constraints [64, 65, 66, 67, 68]. Co-culture/tri-culture stoichiometric models have been developed to analyze compositions of microbial communities for the biogas process . Such models can potentially combine gene expression data or translational profiles as reaction constraints to capture information passage from DNA → RNA → proteins . For example, GSM and metatranscriptomic data are used to study the uncultured bacterial symbiont “Candidatus Endobugula sertula” and identify their metabolic deficiencies . Flux balance analysis (FBA) using reconstructed metabolic networks was attempted for the syntrophic Desulfovibrio vulgaris/Methanococcus maripaludis chemostat system . The modeling results were compared to their continuous bioreactor data on lactate cultures in the absence of sulfate, and were found to accurately predict ecologically-relevant characteristics and growth parameters of bacterial communities. FBA, however, requires the assumption of a pseudo steady-state system. To capture complex dynamics in time and space of microbial communities, dynamic flux balance analysis (dFBA) is employed. In a co-culture model for cellulolytic Clostridium cellulolyticum and the solventogenic Clostridium acetobutylicum, an adapted dFBA was used where the key parameters required are km and Vmax for cellulose solubilization and cell metabolism, as well as starting biomass concentrations for the two co-culture strains to set the input conditions for the growth dynamics . dFBA can simulate the time-step interaction of the individual steady state models, and measurements of key central metabolites can be used to test the accuracy of the co-culture simulation. In another example, dFBA for multiple organisms was developed to address whether ecosystem-level behavior of structured communities can be predicted. The modeling framework with experimentally-confirmed species ratio data simulated the community behavior and metabolic interplay among two or three species inside of a colony . This result can be integrated with economic analysis for a priori estimation of synthetic co-culture performance using cellulosic sugars and flue gas . In parallel, new algorithms are proposed for designing synthetic microbial communities with desired features [74, 75]. Recently, the constraint-based modeling approach was applied to the human gut microbiome, resulting in a collection of 773 GSMs for gut-associated microorganisms . In addition to the computational advances for modeling multi-species systems, other research has sought to expand the depth of modeling to include additional molecular details. An example of these computational advances is a modularized approach using multiple mathematical approaches to model the whole cell function of Mycoplasma genitalium . Another significant study was the implementation of transcriptional and translational machinery into constraint-based models to predict gene expression levels . Together, advances in single organism models and approaches for modeling microbial consortia will enable improved understanding of consortia dynamics and the ability to prospectively design and control consortia function.
13C-metabolic flux analysis (MFA) can measure intracellular carbon and energy fluxes. As a simple approach, MFA can treat co-culture as a single system and determine the bulk fluxomes of this system using traditional 13C-MFA approaches via labeling in protein-derived amino acids. This bulk MFA only describes the system as a whole, without metabolic exchanges. The full separation of metabolites from individual species is required to reveal subpopulation fluxes and interactions between two species. Separation of subpopulation is technologically difficult, even with fast cell-sorting. However, a “reporter protein” synthesized by a specific cell type can store 13C-fingerprints to investigate the microbial community. For mixed culture MFA, instead of detecting the 13C labeling patterns in amino acids from the total cellular protein, 13C patterns in amino acids from the reporter proteins can provide labeling information for each individual species. For example, photosystem I is a reliable reporter protein to probe symbiotic interactions in algae-heterotrophic bacterial communities . Specifically, photosystem I proteins are abundant in algae and these proteins form large complexes that can be separated from the culture via ultracentrifuge and used as the reporter to study algae-bacteria cultures. For engineered species, the reporter protein can be overexpressed with a His-tag, allowing affinity purification from bulk culture [79, 80]. For example, the recombinant fusion glutathione S-transferase and green fluorescent protein can be used as the reporters to study mixed-culture of E. coli mutants. The reporter method quantitatively resolved the expected mutant-specific phenotypes down to subpopulation fractions of ~ 1% . Similarly, peptide-based 13C-MFA has been developed to measure intracellular metabolic fluxes and inter-species metabolite exchange for complex microbial communities. Peptide identity and labeling patterns can be obtained in a high-throughput manner from modern proteomics techniques, which can recover metabolic fluxes in the same way as through the standard amino acid-based 13C-MFA . Another approach for co-culture 13C-MFA is proposed by the Antoniewicz group, which determines metabolic flux distributions without the need for physical separation of cells or species-specific products via measurement of isotopic labeling of total biomass and elegant mathematical calculations. This approach can simulate both fluxes and the relative population size of each species in a mixed culture as well as inter-species metabolite exchange .
Division of labor via mixed cultures can theoretically reduce metabolic burden to enable a system to utilize recalcitrant feedstock or generate products which require long-step heterologous pathways. However, microbial consortia represent complex and dynamic systems that are difficult to operate. Community based strain engineering, metabolic modeling, and flux analysis tools are rapidly developed recently, which may facilitate the wide application of DoL concept for biomanufacturing.
GWR, RRC and YJT wrote introduction, consortia interactions, and DoL applications for feedstock utilizations. JZ and MAK wrote DoL for developing synthetic pathways. SSF and YJT wrote modeling section. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Availability of data and materials
Consent for publication
This work is supported by NSF (MCB 1616619) and DOE (DESC0018324).
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- 20.Shuler ML, Kargi F, DeLisa M. Bioprocess engineering: basic concepts. 3rd ed. NJ: Prentice Hall Englewood Cliffs; 2017.Google Scholar
- 34.Verhoeven MD, de Valk SC, Daran J-MG, van Maris AJA, Pronk JT. Fermentation of glucose–xylose–arabinose mixtures by a synthetic consortium of single-sugar-fermenting Saccharomyces cerevisiae strains. FEMS Yeast Res. 2018;18:foy075.Google Scholar
- 42.Li T, Li CT, Butler K, Hays SG, Guarnieri MT, Oyler GA, Betenbaugh MJ. Mimicking lichens: incorporation of yeast strains together with sucrose-secreting cyanobacteria improves survival, growth, ROS removal, and lipid production in a stable mutualistic co-culture production platform. Biotechnol Biofuels. 2017;10:55.PubMedPubMedCentralCrossRefGoogle Scholar
- 46.Yang WXH. Industrial fermentation of vitamin C. In: Vandamme ERJ, editor. Industrial biotechnology of vitamins, biopigments, and antioxidants. Weinheim: Wiley; 2016.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.