Team:Toulouse INSA-UPS/Model

Introduction

Our project involves a phototrophic community between a sucrose-secreting cyanobacterium S. elongatus UTEX 2973 (CscB +) and the yeast S. cerevisiae, these two organisms sharing the production of the different molecules of the violet fragrance. Synthetic microbial communities have many advantages compared to mono-culture based processes: reduced metabolic burden due to division of labour, enhanced complexity of the tasks that can be achieved, exchange of resources and information, as well as increased robustness to environmental changes (Johns et al. 2016; McCarty and Ledesma-Amaro 2019; Ibrahim et al. 2021). Synthetic microbial consortia with programmed behaviors therefore hold a great potential for various biotechnological applications (Johns et al. 2016; McCarty and Ledesma-Amaro 2019; Ibrahim et al. 2021).

Such systems inevitably involve a greater level of complexity, originating from the interactions between the different species. In order to achieve the intended engineering objectives, careful analysis and design are therefore necessary. Models proved valuable tools to accelerate the DBTL (Design, Build, Test, Learn) cycle of multi-organisms synthetic biology projects (Johns et al. 2016; McCarty and Ledesma-Amaro 2019; Ibrahim et al. 2021). In our case, modeling was vital to address the following question:

How to establish the co-culture system in a stable manner while ensuring the maximal production of a custom mix of the molecules of interest?

This general question was expanded in the form of different sections during our modeling work, allowing a strong link with our wet lab work as well as with our Supporting Entrepreneurship section by addressing the following aspects of our project:

Demonstration of the Feasibility of the project based on a dynamic model of our microbial consortium
Optimization of the process by model-driven identification of key controlling parameters
Driving our entrepreneurship vision by evaluating the Dimensioning and profitability at an industrial scale

When discussing with other iGEM teams, we soon realized that we could establish a partnership with iGEM IISER Pune team, whose project also relied on a microbial consortium exploiting a sucrose-secreting cyanobacteria for production of butanol. We thus established a partnership which proved highly fruitful for both teams!

These different elements have allowed us to give modeling a central place in our project by integrating it at different levels as shown in Figure 1.

Figure 1: Overview of the integration of modeling in our project.

A dynamic coculture model based on mechanistic coupling of genome scale metabolic models

To understand, predict and ultimately control the behavior of the synthetic microbial consortium, we developed a mechanistic, dynamic model of the microbial consortium. The balance of concentration of extracellular substrates and/or products of our microbial consortium are described by a system of ordinary differential equations (ODEs), and intracellular processes are modeled using genome scale metabolic models (GSMs) that contain all metabolic capabilities of each organism. The final model integrates 2593 species (metabolites and cells) and 2982 reactions representing a broad range of processes (growth of each microorganism, metabolite production and utilization, transport between compartments, diffusion, etc). This model was implemented in Python and is provided as a Jupyter notebook to ensure reproducibility and reusability.

Representation of the model

Before constructing the mathematical model, we used the standardized Systems Biology Graphical Notation (SBGN) to represent all the elements of the system and their interactions (Figure 2). More details on the meaning of the different symbols can be found here.

Figure 2: Representation of the model in Systems Biology Graphical Notation (Le Novère et al. 2009). The autotrophic module (Synechococcus elongatus) is in green and the heterotrophic module (Saccharomyces cerevisiae) is in purple.

Construction of the system of ODEs

In this section, we describe the structure of the model, the rate laws used to represent each reaction, and the balance equations for each component of the system. Model units are litre (L) for volumes, hour (hr) for time, and millimole (mmol) and gram dry weight (g_DW) for amounts of metabolites and biomass, respectively.

Growth

First the growth of each of the microorganisms is modeled using their growth rate μ (in h^-1):

\( \frac{ d_{X_{cyano}} }{ d_{t} } = X_{cyano} * μ_{cyano} \) (1)

\( \frac{ d_{X_{yeast}} }{ d_{t} } = X_{yeast} * μ_{yeast} \) (2)

Where \( X_{cyano} \) and \( X_{yeast} \) denote the biomass concentration for cyanobacteria and yeasts, respectively.

CO₂ gas transfer

The CO₂ balance can be represented by considering three processes: the input of CO₂ in the reactor provided by the bubbling of CO₂-enriched air (I_CO₂), the transfer of CO₂ from the gas to the liquid phase (T_CO₂) and finally the output of gas CO₂ outside of the reactors (O_CO₂) (Figure 2).

This can be summarized as:

\( \frac{d_{CO_{2},g}}{d_{t}} = I_{CO_{2}} - T_{CO_{2}} - O_{CO_{2}} \) (3)

In which:

\( I_{CO_{2}} = Q_{gas} * CO_{2,input} \)

\( T_{CO_{2}} = κ_{l}α( β * CO_{2,g} - CO_{2,l} ) \)

\( O_{CO_{2}}=Q_{gas} * CO_{2,g} \)

Where \( CO_{2,g} \) and \( CO_{2,l} \) are the CO₂ concentrations in the gas and liquid phase respectively (in mM), \( Q_{gas} \) is the flow of air bubbled in the reactor (in L.h^-1), \( κ_{l}α \) is the global mass transfer coefficient of CO₂ (in h^-1) and β is Henry’s law constant which models the gas-liquid equilibrium (Henry and Banks 1832).

Carbon source uptake

To ensure that each microorganism responds to changes of extracellular nutrient concentrations (sucrose and CO₂), the uptake of carbon sources is defined for both organisms by Monod equations (Monod 1949) which allow the calculation of specific consumption rates \(q\) (mmol.gDCW-1.h-1) based on the maximum uptake rates of the micro-organism \(q^{max}\) (mmol.g_DCW^-1.h^-1) and a half-velocity constant K_M (mM):

\( q_{CO_{2},cyano} = q_{CO_{2},cyano}^{max} * (\frac{CO_{2,l}}{K_{CO_{2}} + CO_{2,l}}) \)

\( q_{sucrose,yeast} = q_{sucrose,yeast}^{max} * (\frac{Sucrose}{K_{Sucrose} + Sucrose}) \)

As explained below, these fluxes will be used at each timepoint to constrain the genome-scale models in the Flux Balance Analysis (FBA) performed by our algorithm (Orth et al. 2010).

Based on this representation, the ODEs representing the evolution of the liquid CO₂ and sucrose respectively can be written by considering that the cyanobacteria consumes CO₂ and produces sucrose while the yeast does the opposite:

\( \frac{d_{CO_{2}}}{d_{t}} = T_{CO_{2}} - q_{CO_{2},cyano} * X_{cyano} + q_{CO_{2},yeast} * X_{yeast} \) (4)

\( \frac{d_{Sucrose}}{d_{t}} = q_{Sucrose,cyano} * X_{cyano} - q_{sucrose,yeast} * X_{yeast} \) (5)

Light

We took explicitly into consideration the light since it is the source of energy which allows CO₂ fixation, thus ensuring the sustainability of the process. This aspect was also important for dimensioning the system. First, we converted the experimentally measured light intensity \(I_{light} \) which is in μmoles_photon.m^-2.s^-1 into a specific flux q_photon in mmol_photon.g_DCW^-1.h^-1 using the reaction rate proposed by Clark et al. (2018):

\( q_{photon} = I_{light} * \frac{ 3600 * Surface }{ 1000 * X_{cyano} * Volume_{reactor} } \)

Production of the molecules of interest

Finally for the four terpenes (produced by the yeast) and the two violet leaf aldehydes (produced by the cyanobacteria), the following differential equations were established:

\( \frac{ d_{α-ionone} }{d_{t}} = q_{α-ionone} * X_{yeast} \) (6)

\( \frac{ d_{β-ionone} }{d_{t}} = q_{β-ionone} * X_{yeast} \) (7)

\( \frac{ d_{Dihydro-β-ionone} }{d_{t}} = q_{Dihydro-β-ionone} * X_{yeast} \) (8)

\( \frac{ d_{Linalool} }{d_{t}} = q_{Linalool} * X_{yeast} \) (9)

\( \frac{ d_{Nonadienol} }{d_{t}} = q_{Nonadienol} * X_{cyano} \) (10)

\( \frac{ d_{Nonadienal} }{d_{t}} = q_{Nonadienal} * X_{cyano} \) (11)

Where \(q_{M} \) (in mmol.g_DW^-1.h^-1) represents the production rate of each molecule M of interest.

Genome Scale Models

In the above equations, several fluxes such as growth rates or specific fluxes of odorant molecule production describe phenomena that influence the entire production process. The representation of the intracellular dynamics in our model is done by Flux Balance Analysis (FBA) using genome scale metabolic models (GSMs) (Orth et al. 2010). FBA uses the GSM that contains the metabolic network reconstruction to predict the phenotypic responses imposed by environmental constraints. It is one of the main methods to simulate metabolic fluxes in silico (Orth et al. 2010).

The production of the violet terpenes by S. cerevisiae is modeled using the GSM iAZ900 (Zomorrodi and Maranas 2010), which contains 1404 metabolites and 1761 reactions. Since S. cerevisiae was engineered to produce different fragrances, 15 metabolites and 19 reactions that represent the production pathways were added in this model, resulting in a new model iMH919 that represents our engineered LycoYeast-VIOLETTE-FLEUR-FRAMBOISE strain. For S. elongatus UTEX 2973, the model developed by Mueller et al. (2017) was used and further extended with 17 metabolites and 26 reactions to represent the sucrose secretion pathway as well as the production of the violet leaf aldehydes, here again resulting in a new model iRD2999 that represents the sucrose-producing S. elongatus strain.

Coupled algorithm workflow

To couple these GSMs in our bioprocess and make our model dynamic we developed an original algorithm to constitute a unified, consistent modeling framework for simulating the functioning of microbial consortium. This model is made available in the form of a heavily commented Jupyter Notebook. Besides ensuring the reproducibility of our in silico analyses, this notebook can be reused by the future iGEM teams willing to obtain a dynamic insight on their synthetic microbial consortiums.

Notebook
Click here to download our model

The algorithm is presented in detail in Figure 3. Briefly, this algorithm takes as input i) the genome scale models of each organism, ii) the model parameters, and iii) the time and calculates the balance on each component of the system at a given time. Carbon (CO₂) and energy (Light) transfer rates are calculated based on process equations. Uptake fluxes of each carbon source are determined based on their concentration in the environment, using the Monod equations described above. The production rate of the molecules of interest is then constrained according to the experimental yields described in the literature. Finally, FBA is performed to maximize the growth of the organisms with these constraints. Having determined the value of all fluxes, the mass balance on each species can be calculated. We have also implemented events that are triggered by the (in silico) addition of each inducer, hence opening the way to the optimization of the production for tailored fragrances. We used the odeint function of the (SciPy package of Python) to solve this system of ODEs, thereby obtaining dynamical simulations of our production system.

Figure 3: Algorithm to simulate the dynamics of our microbial consortium.

Parameters

Overall, our model contains 17 biochemical and physical parameters which are listed in Table 1. The values of most of these parameters were taken from the literature. We have also measured some key parameters to enhance the predictive capabilities of the model. For example, we chose to measure the maximal growth and sucrose uptake rates of our highly engineered LycoYeast strain (Figure 4). These key parameters are highly strain-dependent (Rodrigues et al. 2021) and have indeed been shown to be critical for the stabilization of the consortium with S. elongatus (Hays et al. 2017).

Figure 4: Raw and fitted data of the growth of LycoYeast using sucrose as a carbon source obtained in the Wet lab. The biomass concentration (left) is expressed in g_DCW/L while the sucrose concentration (right) is in mM. The time unit is in hour. Parameter values (and standard deviation) obtained from the fits are shown in the graphs.

The values we previously used for our model were those determined by Rodrigues et al. (2021) for S. cerevisiae CEN.PK113-7D who had measured a q_{sucrose,yeast}^max of -8.2 ± 0.4 mmol.g_DCW^-1.h^-1. In our conditions, this value was significantly lower (-3.0 ± 0.3 mmol.g_DCW^-1.h^-1). This shows that it was indeed critical to perform this measurement to make our model more realistic. This approach furthermore allowed us to test whether our model could predict the growth rate of S. cerevisiae on sucrose. The excellent agreement between predictions and measurements (see details in the FBA & GSM box) demonstrates that updating this parameter improves model predictions. The complete list of parameters is given below. More details on the determination of some of these parameters are available in our Assumptions and Parameter Calculations Appendix.

Table 1: Model parameters and associated values.

Demonstration of the feasibility of the project

To assess the feasibility of our project, we simulated the dynamics of our complex synthetic microbial community using the parameters described above and initial conditions that correspond to realistic environmental conditions of our system. The dimensions of the chosen reactor correspond to the 180 L flat panel bioreactor detailed in the Supporting Entrepreneurship section.

Figure 5: Simulation of one week of coculture in a 180 L flat panel bioreactor. A) Dynamics of S. cerevisiae (purple) and S. elongatus (green) B) Dynamics of sucrose concentration (mM) C) Dynamics of fragrance molecules concentrations (mM) D) Dynamics of CO₂ concentrations (mM).

Model simulations (Figure 5) show that, as expected, cyanobacteria use CO₂ to grow, produce the violet leaf aldehydes and secrete sucrose. This sucrose is consumed by the yeast to support its growth and the production of the different terpenoids. Overall, these simulations are consistent and demonstrate the theoretical feasibility of our project. This preliminary work was very positive and encouraged us to continue in this direction!

On this basis, we chose to carry out the coculture in silico for three weeks, in agreement with the calculations made in our Supporting Entrepreneurship section, which show that the culture medium represents an important part of the operating cost of a batch. This is consistent with previous studies showing that co-culture of yeast with sucrose-secreting S. elongatus can be maintained stable for up to one month (Hays et al. 2017; Li et al. 2017). Under these conditions, the model predicts that our coculture would produce 434.2 g of violet fragrance ingredient in one batch. This motivated us to analyze further this model to optimize the system, as explained below.

Optimization of the system by Metabolic Control Analysis (MCA)

Since our model predicts the possibility to grow a stable yeast/cyanobacteria consortium, we aimed at identifying key parameters that determine production by taking advantage of the mathematical tools of Metabolic Control Analysis (MCA) (Moreno-Sánchez et al. 2008). A key point of this theory is the concept of control coefficients which quantify the degree of control of a given parameter on a defined metric. Usually used to investigate the control of specific metabolite concentrations or metabolic fluxes, we extend here this concept to investigate the degree of control exerted by each of the parameters of our system on i) the time taken by our consortium to produce a selected mass of violet fragrance (thereby optimizing productivity) and ii) the energetic cost per gram of our violet fragrance ingredient (thereby optimizing sustainability).

These metrics were estimated using the calculations of our Supporting Entrepreneurship section. For instance, the second metrics allow us to quantify the influence of each parameter on the operational energy cost of our process. Examples include the lighting of the reactor, the pump controlling the bubbling of the air in the reactor or the volume of the reactor which determines the energy needed for sterilization and heat transfer. To be able to compare different conditions with respect to production, we then divide this energetic cost score by the mass of violet ingredient produced at the end of the cultivation process.

The control coefficients for both the production time metric \( C_{prod_time} \) and for the energetic cost metric \( C_{energetic_cost} \) were calculated for each parameter using numerical differentiation. Simply, for a given parameter p, the value of each metric is compared between a reference simulation (the parameter value is the one presented in Table 1) and for a simulation for which we impose a relative perturbation of \(Δp\) to the parameter value. Control coefficients are then calculated as:

\( C_{prod \_ time} = \frac{ p }{ prod \_ time } \frac{Δprod \_ time}{Δp} \)

\( C_{energetic \_ cost} = \frac{ p }{ energetic \_ cost } \frac{Δenergetic \_ cost}{Δp} \)

If the metric is not controlled by the parameter p, the corresponding coefficient will be zero. A positive (or negative) value indicates that an increase in p increases (or reduces) the metric of interest. A coefficient of 1 indicates that a change in x % of the parameter results in a change of x % of the metric value. MCA results for these two metrics are shown as heatmaps in Figure 6.

Figure 6: Heatmap of the control coefficients of each parameter on (A) the production time and (B) the energetic cost per gram of our violet fragrance ingredient. Positive values of control coefficient (purple) mean that increasing the value of the parameter increases the metric value. Conversely, negative values of the control coefficient (green) mean that increasing the value of the parameter reduces metric value.

Based on these quantitative results, we could investigate the control exerted by each parameter on the productivity and sustainability of our process, as detailed below.

CO₂ gas transfer: β, CO₂^gas_in, kLA

The first important observation made from these results is that under the current conditions, the transfer of CO₂ from the gas phase to the liquid phase T_CO₂ is an element that significantly limits the productivity of our system. It is not possible to act on β which is the thermodynamic coefficient of CO₂ gas/liquid equilibrium. However, increasing the concentration of CO₂ of the input gas could easily be achieved. Increasing the volumetric CO₂ mass-transfer coefficient kLa is another possibility. To do so, it would be possible to use optimized bubbling systems such as air stones as described by Falinski et al. (2018).

CO₂ uptake by the cyanobacteria: K_CO₂, q_{CO_2,cyano}^max

Consistently, the productivity of the system is also impacted by the rate of CO₂ uptake by cyanobacteria and thus by the parameters of Monod's law K_CO₂ and q_{CO₂,cyanomax}. This rate of CO₂ fixation will indeed influence the transfer term T_CO₂> by modifying the gradients of concentration between the different phases. Although these parameters are controlling, we can hardly optimize them because they are intrinsic, systemic properties of the metabolism of S. elongatus UTEX 2973.

Sucrose secretion: R_{CO₂,sucrose}, X_cyano, t_IPTG

Counter-intuitively, it appears that increasing the yield of sucrose production R_{CO₂,Sucrose} by the cyanobacterium has a deleterious effect on our two metrics of interest! After careful analysis of the model response to this yield value, we found that using too much of the fixed carbon for sucrose production indeed limits the growth of cyanobacteria. The sucrose secretion flux per cell is higher but since the amount of cells in the reactor is lower, the overall sucrose production is lower. This is consistent with the observation that inoculating the reactor initially with more cyanobacterial biomass or delaying the addition of IPTG (which induces sucrose secretion) improves the overall productivity. Thus, a balance must be found between cyanobacterial growth and sucrose production. Using our model, we were able to show that the maximum productivity of the coculture is obtained when 60% to 65% percent of the fixed CO₂ is converted to sucrose (Figure 7). To give a point of comparison, it has been shown that the strains we use experimentally can secrete up to 88% of the fixed CO₂ in the form of sucrose (Lin et al. 2020).

Though it was initially counter-intuitive that reducing sucrose production in our process would overall enhance the production of our fragrances, a similar conclusion could be obtained by our partners, as detailed in the Partnership section.

Figure 7: Influence of the sucrose production yield on the productivity of the fragrance ingredient. (the production of the fragrance molecules produced by the cyanobacteria were not induced in this study to avoid biasing the results)

Light intensity: I_{light_measured}

So far, the parameters had the same qualitative impact for both metrics studied, which are indeed related since the estimated energy cost is reduced to one gram of violet fragrance. Thus, if we decrease the time needed to produce a given quantity of ingredient (first metric), we will produce more in the end, which, if the use of energy is the same, gives a more interesting energy cost per gram produced (second metric). Still, we found that light intensity exerted a different control on the Productivity and the Sustainability of our process. Indeed, light greatly impacts energy use during the process. Since increasing the light intensity does not increase the productivity as shown with the first metric, we investigated whether we could decrease the bioreactor lighting to save energy (Figure 8). By performing several simulations with our model, we were able to show that it would actually be possible to decrease the intensity of the light used by one third without altering the amount of violet perfume produced at the end of the culture.

Figure 8: Influence of the light intensity on the energetic cost per gram of violet fragrance ingredient.

Dimensioning and profitability of the process

As explained before, our idea was to exploit our model to provide quantitative data to design an industrial biotechnology plant to produce our fragrances. Here are briefly described several elements for which our model has proven to be a useful predictive tool:

Dimensioning our pilot plant: After discussion with our industrial partner Robertet, we set a production objective of 250 kg per year of our violet fragrance ingredient. Our dynamic model predicted that each three-weeks-long batch would produce 434.2 g of product per bioreactor. As a result, we were able to calculate that 40 reactors would be required to complete the target production in a year.
Assist in the design of the model bioreactor: The MCA performed on the production time highlighted that the transfer of CO₂ was limiting in the first simulations. To increase this transfer, it is possible to choose to use air stones for the bubbling of CO₂ enriched air in the reactor (Falinski et al. 2018).
Guide future strain engineering strategies: This same MCA strategy demonstrates that counter-intuitively, it would be necessary to decrease the yield of sucrose production to about 60% to 65% of the CO₂ uptake to maximize the productivity of the consortium. This could be used for future strain engineering optimization, for example by choosing to place the CscB sucrose transporter under the control of a tightly inducible promoter controllable by inducer concentration, such as theophylline-inducible riboswitch theoE* (Li et al. 2018).
Minimizing the energetic cost of the process: It was estimated in our Supporting Entrepreneurship section that lighting accounted for 39% of energy consumption. Using our MCA approach, we found that light was in large excess under the previous conditions and that energy could be saved by reducing its intensity by one-third. This result is therefore particularly relevant considering our primary goal of sustainability.
Carbon footprint calculation: Using our model, we were able to determine that in one year, our system would produce 260.5 kg of violet fragrance, 270 kg of yeast biomass and 4320 kg of cyanobacterial biomass. Each of these elements comes from the CO₂ fixed by the cyanobacteria. Thus, this represents a total of 8.6 tons of CO₂ equivalent that is captured from the atmosphere by our production plant each year (this term is therefore counted negatively in the global carbon balance). This covers approximately 50% of the operational carbon footprint of our production process, demonstrating the interest of our consortium for the sustainability of the process.
Determine how much GMO biomass has to be treated at the end of a batch: Genetically modified biomass must be inactivated at the end of an industrial biotechnology process to ensure biocontainment before further valorization. Such a process consumes energy, and it was therefore necessary in our Supporting Entrepreneurship calculations to be able to estimate the amount of waste to be treated.

Dry lab partnership with iGEM IISER Pune

As part of our partnership with the iGEM IISER Pune team, we collaborated on several aspects of our dry lab work. Firstly, we compared our implementation of the sucrose secretion pathways on our respective GSMs. Using FBA, we were able to verify that the results were similar and thus strengthened the reproducibility, reusability and correctness of our updated models.

Then, as a proof of the modularity of our approach to dynamically model synthetic microbial consortiums, we have shared our code with their team. They were able to use it to predict and understand the behavior of their butanol-producing microbial community (Figure 9).

Figure 9: Example of simulation obtained by the iGEM IISER Pune team using our dynamic coculture modeling method.

Both our teams initially had the preconceived notion that boosting sucrose production yield by cyanobacteria would improve the productivity of our respective microbial consortia. Using Flux Scanning based on Enforced Objective Flux (FSEOF) (Choi et al. 2010) and Optknock (Burgard et al. 2003), two different strain optimization approaches, they were able to identify interesting targets for gene overexpression or deletion to improve the sucrose secretion. In fact, our dynamic coculture model actually demonstrates that boosting sucrose production is counter-productive to enhance the production of our violet fragrance. These results, which were confirmed by complementary in silico analyses carried out by our partner, will be of interest for other applications that exploit sucrose-producing cyanobacteria.

Finally they also worked on an alternative approach to model microbial communities: Steadycom (Chan et al. 2017). Using the code they provided, we were able to validate the existence of a steady state with positive growth values for both microorganisms and the production of all our final products, which further confirms the feasibility of our project. Still, this approach has some limitations, as it can only represent a steady state of the community, which is not always the norm in this type of community configuration. This is notably attested by the existence of microbial systems with oscillating populations (Di and Yang 2019). Overall, the comparison of this static method with our dynamic approach has strengthened the value of the dynamic modeling approach we have developed and exploited throughout our project.

Conclusion

To summarize, using our dynamic and predictive model, we were able to demonstrate the feasibility of our project, optimize it and dimension a pilot production unit. The key outcomes obtained by our modeling approach are summarized here:

Feasibility: Our dynamic coculture model demonstrated the theoretical feasibility of our project with the production of 434.2 g of violet fragrance ingredient in one batch per bioreactor.

Optimization: Using the mathematical tools of MCA, we highlighted that in the previous conditions, i) CO₂ transfer was limitant, ii) decreasing the yield of sucrose production by the cyanobacteria would actually improve the productivity of the system and iii) the lighting was in excess. These findings provide perspectives for optimization of both the strains used and the design of the model bioreactor used on an industrial scale.

Supporting entrepreneurship: Modeling was used to provide key quantitative data to design an industrial biotechnology plant to produce our fragrances.

Contribution: The modeling approach we developed is fully modular and can therefore be reused by any team interested in modeling the behavior of synthetic consortia. It is made available in the form of a Jupyter Notebook that can be downloaded here.

Partnership: Finally, modeling has been a pillar of our fruitful partnership with the iGEM IISER Pune team.

Overall, our strategy demonstrates how modeling can be used as a tool to understand and optimize microbial consortia for biotechnological applications, as well as supporting the implementation of an industrial biotechnology project.

References

Burgard AP, Pharkya P, Maranas CD. 2003. Optknock: a bilevel programming framework for identifying gene knockout strategies for microbial strain optimization. Biotechnol Bioeng. 84(6):647–657. doi:10.1002/bit.10803.

Chan SHJ, Simons MN, Maranas CD. 2017. SteadyCom: Predicting microbial abundances while ensuring community stability. PLOS Computational Biology. 13(5):e1005539. doi:10.1371/journal.pcbi.1005539.

Choi H, Lee SY, Kim TY, Woo H. 2010. In Silico Identification of Gene Amplification Targets for Improvement of Lycopene Production. Applied and environmental microbiology. 76:3097–105. doi:10.1128/AEM.00115-10.

Clark RL, McGinley LL, Purdy HM, Korosh TC, Reed JL, Root TW, Pfleger BF. 2018. Light-optimized growth of cyanobacterial cultures: Growth phases and productivity of biomass and secreted molecules in light-limited batch growth. Metabolic Engineering. 47:230–242. doi:10.1016/j.ymben.2018.03.017.

Di S, Yang A. 2019. Analysis of productivity and stability of synthetic microbial communities. J R Soc Interface. 16(150):20180859. doi:10.1098/rsif.2018.0859.

Falinski K, Timmons M, Callan C, Laidley C. 2018. Response of Tisochrysis lutea [Prymnesiophycidae] to aeration conditions in a bench-scale photobioreactor. Journal of Applied Phycology. 30. doi:10.1007/s10811-018-1453-y.

Hays SG, Yan LLW, Silver PA, Ducat DC. 2017. Synthetic photosynthetic consortia define interactions leading to robustness and photoproduction. J Biol Eng. 11(1):4. doi:10.1186/s13036-017-0048-5.

Henry W, Banks J. 1832. Experiments on the quantity of gases absorbed by water, at different temperatures, and under different pressures. Abstracts of the Papers Printed in the Philosophical Transactions of the Royal Society of London. 1:103–104. doi:10.1098/rspl.1800.0063.

Ibrahim M, Raajaraam L, Raman K. 2021. Modelling microbial communities: Harnessing consortia for biotechnological applications. Computational and Structural Biotechnology Journal. 19:3892–3907. doi:10.1016/j.csbj.2021.06.048.

Johns NI, Blazejewski T, Gomes AL, Wang HH. 2016. Principles for designing synthetic microbial communities. Current Opinion in Microbiology. 31:146–153. doi:10.1016/j.mib.2016.03.010.

Li S, Sun T, Xu C, Chen L, Zhang W. 2018. Development and optimization of genetic toolboxes for a fast-growing cyanobacterium Synechococcus elongatus UTEX 2973. Metabolic Engineering. 48:163–174. doi:10.1016/j.ymben.2018.06.002.

Li T, Li C-T, Butler K, Hays S, Guarnieri M, Oyler G, Betenbaugh M. 2017. 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. Biotechnology for Biofuels. 10. doi:10.1186/s13068-017-0736-x.

Lin P-C, Zhang F, Pakrasi HB. 2020. Enhanced production of sucrose in the fast-growing cyanobacterium Synechococcus elongatus UTEX 2973. Scientific Reports. 10(1):390. doi:10.1038/s41598-019-57319-5.

McCarty NS, Ledesma-Amaro R. 2019. Synthetic Biology Tools to Engineer Microbial Communities for Biotechnology. Trends in Biotechnology. 37(2):181–197. doi:10.1016/j.tibtech.2018.11.002.

Monod J. 1949. The Growth of Bacterial Cultures. Annual Review of Microbiology. 3(1):371–394. doi:10.1146/annurev.mi.03.100149.002103.

Moreno-Sánchez R, Saavedra E, Rodríguez-Enríquez S, Olín-Sandoval V. 2008. Metabolic Control Analysis: A Tool for Designing Strategies to Manipulate Metabolic Pathways. J Biomed Biotechnol. 2008:597913. doi:10.1155/2008/597913.

Mueller TJ, Ungerer JL, Pakrasi HB, Maranas CD. 2017. Identifying the Metabolic Differences of a Fast-Growth Phenotype in Synechococcus UTEX 2973. Scientific Reports. 7(1):41569. doi:10.1038/srep41569.

Novère NL, Hucka M, Mi H, Moodie S, Schreiber F, Sorokin A, Demir E, Wegner K, Aladjem MI, Wimalaratne SM, et al. 2009. The Systems Biology Graphical Notation. Nat Biotechnol. 27(8):735–741. doi:10.1038/nbt.1558.

Orth JD, Thiele I, Palsson BØ. 2010. What is flux balance analysis? Nat Biotechnol. 28(3):245–248. doi:10.1038/nbt.1614.

Rodrigues C, Wahl A, Gombert A. 2021. Aerobic growth physiology of Saccharomyces cerevisiae on sucrose is strain-dependent.

Zomorrodi A, Maranas C. 2010. Improving the iMM904 S. cerevisiae metabolic model using essentiality and synthetic lethality data. BMC systems biology. 4:178. doi:10.1186/1752-0509-4-178.

Go to Results

Experimental growth rate (h^-1)	Predicted growth rate with q_S from literature (h^-1)	Predicted growth rate with experimental q_S (h^-1)
0.33 ± 0.02	1.3	0.44