<!DOCTYPE html> Title



Mathematical modeling is an integral part of the field of metabolic engineering as it can provide a crucial understanding of a broad range of features of the system under study - from its general dynamics to how it responds to specific genetic modifications. Computational models can be leveraged to select promising avenues for wet-lab experimentation.

SynBactory benefits tremendously from the insights gained by the modeling of various features of the organisms such as their growth rates, metabolic interactions, etc.

The main objectives of our dry lab team this year were:

  1. Predict individual growth rates and product yields of our organisms under different stresses and environmental conditions.
  2. Identify key gene targets to overexpress or knockout S. elongatus and E. coli for improved yields of sucrose and butanol respectively.
  3. Understand interactions in the co-culture system and identify optimal parameters for the maximum yield of butanol.

To predict growth rates and product yields of our individual models we used Flux Balance Analysis (FBA) and Minimisation of Metabolic Adjustments (MoMA). To identify gene targets for overexpression and deletion we used Flux Scanning based on Enforced Objective Flux (FSEOF) and Optknock. The results of these for each of our individual species can be found on the E. coli and S. elongatus tabs respectively.

To analyze interactions in the co-culture and detect optimal parameters we used SteadyCom and a Dynamic Model, based on FBA. The results of these can be found in the Co-Culture tab.

We have used the COBRA package in MATLAB[1] for almost all of our models. The core technique for all of the methods mentioned above is the constraint-based modelling technique, Flux Balance Analysis.

Flux Balance Analysis

FBA is a steady-state analysis algorithm that is widely used in metabolic predictions and optimizations. FBA assumes the organism to be in a metabolically steady state i.e., there is no accumulation of metabolites within the organism. To perform FBA, a metabolic model (GSM - Genome Scale Model) of the organism is required.

A metabolic model of an organism contains information about all (or most) reactions that exist within the organism, the metabolites involved, the genes present, the gene-reaction relations, and a calculated biomass production reaction (growth rate reaction) that gives the growth rate when analyzed.

The core elements of FBA are2:

  1. S Matrix or the stoichiometric matrix describes all the reactions present in the GSM
  2. The objective function or 'c' - This is a column vector that contains the information about the metabolic network that allows it to attain the aim of the analysis, for example biomass optimization.
  3. Lower bounds and upper bounds of the fluxes of the reactions are a set of values that help constrain the solution space
  4. Solution flux space - This is a set of fluxes that the algorithm converges to when optimizing the objective function.

In an organism, it is generally observed that the same metabolites participate in multiple reactions, and the number of reactions is greater than the number of metabolites. This can be represented by a system of linear equations, where the number of equations is greater than the number of variables. In such a case the solution space is infinite, thus we must introduce constraints to constrain the solution space, in the form of bounds.

A pictorial depiction of the algorithm

Equations of FBA and their solutions2


  • Sv = 0
  • Lower Bounds
  • Upper Bounds

Objective Function2:

  • cTv


  • Minimise or Maximise Objective function


  • Plausible flux distributions for the above mentioned objective function

However, when the metabolic network is perturbed, for example in stress conditions, the organism might not necessarily optimize their growth rate, as assumed by FBA.

Minimisation of Metabolic Adjustments (MoMA)

MoMA is an extension of Flux Balance Analysis that imposes additional constraints that make the model a better representation of reality.It minimizes the euclidean distance (L2 norm) between the wild-type flux distribution and the flux distribution of the modified organism in the flux vector space.3.

The algorithm3:

  1. Find the wild-type/base strain flux distribution using FBA with maximizing growth rate as the objective function.
  2. Simulate the deletion, overexpression or a particular stress condition.
  3. Impose an additional constraint that minimizes the L2 norm of the solution and the wild type flux; which would look like this - ||vwild - vmod||2 is minimized. Where vwild is the flux distribution of the wildtype strain and vmod is a flux distribution of the modified strain.
  4. Run FBA multiple times to find such a vmod.

When one does not know of the modifications that need to be made in order to increase the yield of a particular metabolite, techniques such as flux scanning based on enforced objective flux (FSEOF) and OptKnock come in handy.

Flux Scanning based on Enforced Objective Flux (FSEOF)

This algorithm was developed to identify gene targets that can be amplified for increased production of selected metabolites. We have used this technique to identify the reactions and further genes that we can over-express to increase sucrose production in S. elongatus and butanol production in E. coli.

A useful analogy to understand FSEOF is to imagine the organism as a city; the reactions as roads, and the flux through the reactions as the traffic on each road. Suppose the objective function for our case is the most efficient evacuation possible of the city; then the roads, when widened, will allow larger traffic flows to pass through them. The problem is to select which roads to widen to quickly optimize the city’s architecture for evacuation. In the biological scenario of our interest, the objective function corresponds to maximizing the production of a particular metabolite. The problem can then be recast as the question of which fluxes to upregulate and downregulate to attain this maximum.

In the form of an algorithm it would look something like this4:

  1. Find the maximum growth under a particular media composition using FBA.
  2. Find the maximum production of the metabolite of intrest using FBA.
  3. Constrain the production of the metabolite to some fraction of the maximum amount of metabolite that can be produced.
  4. Run FBA for maximum growth rate.
  5. For a given constraint, calculate the corresponding growth rate value, and its product with the metabolite production value.
  6. Plot how the flux values of each reaction changes as the fractional value of the constraint on the metabolite is increased.
A pictorial depiction of the algorithm

Since FBA and all of its derivative algorithms work with reactions and not genes, it is necessary to trace the reactions back to their regulating genes and verify if those genes are essential for the organism’s growth or survival. Once such gene candidates are eliminated, we can obtain a list of genes that could, in theory, lead to increased metabolite production upon over-expression.


OptKnock is an algorithm that finds and suggests which set of reactions to remove from a metabolic network to obtain a mutant that will produce a particular target of interest at a higher rate than the wild-type strain.5. The algorithm works as follows-

  1. Find the set of reactions that you wish to delete.
  2. Change their bounds to 0, which is how gene deletions are simulated in the model, by specifying the maximum number of deletions in each iteration.
  3. Run FBA to maximise growth rate.
  4. Trace the set of reactions, whose deletions result in increased metabolite production, back to their respective genes
  5. Eliminate all the synthetic lethals from the set of identified genes. Synthetic lethals are essential genes that are required for the organism’s survival.

A pictorial depiction of the algorithm

Once the synthetic lethals have been eliminated, we obtain a set of reactions and their corresponding genes that can be knocked out for the enhanced production of the metabolite of interest.


SteadyCom[6] is an algorithm used to analyze the steady-state behaviour of co-cultures, microbiomes, and host-microbe interactions. Similar to FBA, SteadyCom assumes that there is no accumulation of metabolites and only works with reaction fluxes. A second assumption is that at community steady-state if such a state exists, the growth rates of all the organisms is the same, such that no one organism dominates and takes over the population. Based on the constraints in the model, SteadyCom tries to maximize the community growth rate. It also provides results such as the biomass ratio of the constituent organisms at steady-state growth, and the fluxes of various reactions in the multi-species network.

The Logic:

  1. Create a joint model which has all the reactions of every organism’s genome-scale models and a separate community space to allow for the transfer of common extracellular metabolites between the separate extracellular compartments of the individual models
  2. The algorithm predicts steady-state fluxes which maximizes the community growth rate, such that all organisms grow at this rate
  3. The fluxes of reactions out of individual organisms are multiplied by their biomass so that the abundances of the organisms are taken into account in their interactions
A pictorial depiction of the algorithm

Dynamic Modeling


Dynamic modelling here involves using fluxes obtained from FBA to form and solve ordinary differential equations that predict the concentrations of metabolites and the amount of biomass in the bioreactor over time. Essentially, these fluxes act as the first-order time derivatives in the system of differential equations.

When we optimize a model using FBA, we obtain the fluxes of all the reactions that allow the model to optimize its objective function. So, given an initial concentration of a metabolite, the FBA flux gives the derivative of the concentration at that point. Using this derivative, we can calculate the concentration of the metabolite at the next time step, and use this as the new initial concentration. This process can be repeated iteratively within the required time frame.

The algorithm:

  1. Set the flux restrictions of the cyanobacteria uptake rates of carbon dioxide and light, based on the concentration of these present in the bioreactor available for uptake.
  2. Optimize the cyanobacteria model with sucrose production as its objective using FBA. This will naturally make the biomass flux of the cyanobacteria set to its lower bound 0. Biomass flux set to zero implies that the rate of growth of biomass is 0. Hence, the maximized sucrose flux obtained from this result is the theoretical maximum sucrose yield possible from the model within the given constraints.
  3. It is known that attaining the theoretical maximum yields of sucrose is not biologically feasible. So it is assumed that the salt stress provided to force the cyanobacteria to produce sucrose as an adaptive response, allows it to export sucrose at some fraction of the maximum sucrose yield calculated above. This is referred to as the percentage productivity of sucrose flux in the cyanobacteria.
  4. This productivity rate, x% of max sucrose flux, is now set as the lower bound for the sucrose export reaction in the cyanobacteria model, and the objective function is changed to biomass flux. An FBA is performed to maximize the biomass flux given this new constraint of sucrose production.
  5. The final results from this FBA are taken as the fluxes of the cyanobacteria model.
  6. The same process is repeated for the E. coli model, where the available sucrose in the bioreactor comes from the cyanobacteria flux calculated above. Butanol production is first optimized to find the maximum flux possible, then as above, some productivity for this reaction is set and biomass is optimized.
  7. All the fluxes from both models provide the time derivatives of the corresponding metabolite concentrations, which are integrated and plotted.


  1. Heirendt, L., Arreckx, S., Pfau, T. et al. Creation and analysis of biochemical constraint-based models using the COBRA Toolbox v.3.0. Nat Protoc 14, 639–702 (2019).
  2. Orth, J. D., Thiele, I., & Palsson, B. Ø. (2010). What is flux balance analysis?. Nature biotechnology, 28(3), 245–248.
  3. 25 Segrè, D., Vitkup, D., & Church, G. M. (2002). Analysis of optimality in natural and perturbed metabolic networks. Proceedings of the National Academy of Sciences of the United States of America, 99(23), 15112–15117.
  4. Choi, H. S., Lee, S. Y., Kim, T. Y., & Woo, H. M. (2010). In silico identification of gene amplification targets for improvement of lycopene production. Applied and environmental microbiology, 76(10), 3097–3105.
  5. Burgard, A. P., Pharkya, P., & Maranas, C. D. (2003). Optknock: a bilevel programming framework for identifying gene knockout strategies for microbial strain optimization. Biotechnology and bioengineering, 84(6), 647–657.
  6. Chan, S. H. J., Simons, M. N., & Maranas, C. D. (2017). SteadyCom: predicting microbial abundances while ensuring community stability. PLoS computational biology, 13(5), e1005539.


Cyanobacteria is the green element of our project, both figuratively and literally. Modeling its behavior both in a monoculture and in the co-culture is of significance with regard to identifying productive avenues for further exploration.

We’ve made use of FBA and MoMA to predict the growth rates and sucrose production rates of S. elongatus, both with and without salt stress. Dr. Costas Maranas was generous enough to provide the Genome-scale metabolic model for Synechococcus elongatus UTEX 2973 which his lab uses - iSyu683[1].

Modelling Cyanobacteria for Optimum Sucrose Production

Simulating salt stress was a major hurdle since it is difficult to account for the regulatory changes that would affect the sucrose production pathway. In consultation with our mentors, we came up with the following simplifying assumptions to simulate salt stress.


  1. Intracellular sucrose levels remain constant to maintain tonicity corresponding to a particular salt concentration in the medium.
  2. The biomass reaction of the model does not take into account the intracellular sucrose that accumulates in response to salt stress. Since FBA does not allow the accumulation of any metabolites within the model, to account for this accumulated sucrose we need to create a sink reaction. The flux through this sink reaction corresponds to the accumulated sucrose. Following from assumption 1, this flux is a constant ratio of the flux through the biomass reaction. This couples the reactions by a constant, corresponding to the ratio of their fluxes.
  3. Sucrose transport efficiency through the cscB transporter (referred to as cscB efficiency henceforth) remains constant.
  4. This means that the flux through the cscB transporter reaction is a constant proportion of the flux through the sucrose sink reaction.

Apart from these assumptions, we had to make certain modifications to the model. The model did not include the reactions for the sps and spp genes and we had to add them manually, referred to as “SPS” and “SPP” respectively. We also added the reaction for the sucrose permease gene, referred to as “cscB”.


  • Upon adding “SPP”, “SPS”, and “cscB” and running FBA with an objective to maximize growth rate we obtained a maximum growth rate of 0.3033 h-1 and a sucrose production and transport flux of 0. This value is very similar to the experimental growth rate of 0.33 ± 0.05 h[2] , which gives us confidence in the model we’re using.
  • We obtained a growth rate of 0.2012 h−1 and a sucrose output flux of 0.3079 mmol/gDW/h when we performed MoMA for the following parameters:
    • For a salt concentration of 150mM in the medium, Song et al[3] report an internal sucrose concentration of 0.06 gram per gram dry weight (g/gDW) equivalent to 0.17 mmols per gram dry weight. Thus, to maintain a constant flux ratio of the biomass and sucrose sink reactions, the coupling constant must be 0.17.
    • We set a CO2 uptake rate of -12.2 mmol/gDW/h based on ref. 2.
    • We set a cscB efficiency of 90% based on ref. 4.

We performed MoMA by varying CO2 uptake rates in the range: 10.2 mmol/gDW/hr to 14.2 mmol/gDW/hr, and varying cscB efficiency values in the range: 75% to 94%.

Growth Rate vs CO2

Biomass vs CO_2 graph for different constant cscB efficiency values and a constant salt stress of 150mM. A linear pattern is observed

Fig 1: Relationship between Growth rate and CO2 uptake rates when run at a salt conc. 150mM under varying cscB efficiencies


  1. Biomass production rates decrease with decreasing CO2 uptake rates for a constant cscB efficiency
  2. Biomass increases for a decreasing cscB efficiency at a given CO2 intake value.


An increase in CO2 uptake rate naturally results in a higher biomass production rate as more carbon is sequestered to be photosynthesized and converted to biomass as well as sucrose.
Carbon flux through the biomass reaction decreases as the efficiency of cscB increases. Since the objective function is to maximize sucrose production we can infer that cscB acts as a sucrose sink and therefore increases flux through sucrose production reaction at higher efficiencies.

Sucrose vs Growth Rate

Biomass vs Sucrose graph for different CO_2 uptake rates. We observe a linear pattern yet again where sucrose production increases with decrease in growth rates

Fig 2: Relationship between sucrose production and growth rate under varying CO2 uptake rates for a salt concentration of 150mM


  1. Sucrose production rates decrease with increasing growth rates
  2. Both sucrose production rates and growth rate increase with increase in CO2 uptake rate


Since sucrose production directs carbon flux away from growth, an increase in sucrose production naturally decreases the growth rate. Therefore, our aim is to direct carbon flux through both the reactions in a fixed ratio we optimize for sucrose production, but not at the cost of a viable growth rate.

Sucrose production

This is a graphical representation of our combined results. We have eliminated all the metabolic states that do not agree with our MoMA results. The color bar shows the sucrose production fluxes in mmol/gDW/h.

A graph respresting varaition of sucrose production with increase in growth rates, CO_2 Uptake rates and cscB efficieny values

Fig 3: A graph showing the production of sucrose for varying CO2 uptake rates and cscB efficiencies
Note - the colour bar on the right shows the mmol/gDW/h values of sucrose

Sucrose production is maximum when there is a high CO2 uptake rate and a high cscB efficiency. For enhanced sucrose production, it would be best to engineer S. elongatus to function in these regimes, while still maintaining viable growth rates.
However, improving cscB efficiency beyond its highest reported efficiency of around 90% might be difficult and may not have a substantial payoff. On the other hand, we have looked at methods to increase CO2 uptake rate as a part of our proposed implementation.

Identifying Over-expressions and Knockouts for enhanced sucrose production

We have used OptKnock to identify knockout targets and FSEOF to identify over-expression targets.

Targets for Over-expression

Upon running FSEOF under the following conditions we obtained a few targets for over-expression.

Problems to note:

  1. The model does not have all the reactions in the organism and thus certain targets corresponding to those reactions may not be identified.
  2. The genes have no names, making it difficult to correlate genes in the model with genes in the organism.
  3. Some reactions have no gene-protein relations. (Gene-protein-relations (GPRs) are rules that bind reactions in the organism to particular genes or a combination of genes. For example, if reaction R1 requires both genes G1 and G2 then in the absence of either one of the genes the reaction ceases to work.)


  1. We have obtained a total of 80 reactions corresponding to 87 genes that can potentially be overexpressed to increase sucrose production.
  2. These 87 genes are responsible for a total of 555 reactions in the organism. Few of them are the same reaction but in different parts of the organism, for example, if the cytoplasm and periplasm both contain a reaction X, the reaction X is counted twice.
  3. Accounting for this redundancy in counting the reactions, there are 107 unique reactions.
  4. 25 out of the 80 target reactions have no gene-protein-relations so we cannot test these in the lab.
  5. Although upon overexpression of these targets, sucrose production is maximized, growth rates decrease, which is a tradeoff we need to keep in mind as aforementioned.

Given below are 4 graphs each with 5 reactions plotted against varying sucrose production fluxes. This is just to give a comparison of the various results we obtained. If you would like to download the entire set of interactive graphs representing these results click here. We also have graphs which depict the variations of growth rate along with sucrose and these reactions (click here)

However, these over-expression targets require experimental validation.
We have also ranked the results in order of decreasing slope because we believe that the ones with higher slopes have a higher flux handling capacity i.e., they can hold more flux for the desired purpose. In terms of the analogy between FSEOF and traffic in a city, these are the roads that can take a good chunk of traffic away from other smaller roads and still not get jammed.

These targets of overexpression are in terms of reactions, but in the lab, we can not selectively overexpress reactions, only genes. So naturally, we looked into how other reactions in the model get affected if and when we overexpress one of the above genes. The table below shows how the genes responsible for the above reactions will affect other reactions in the organism. Note - the “fig x” in each row indicates which graph in the folder corresponds to the said reaction.

Over-expression Table

Target Reaction
Other reactions controlled by the gene
Proton exchange (Fig 1)
Water transport pp (Fig 2)
Water transport ltr
Water transport ext (Fig 3)
ChorismateL-glutamine aminotransferase (Fig 4)
M744_14170 or M744_09950
M744_14170 - Xanthosine-5'-phosphateL-glutamine amido-ligase (AMP-forming)
M744_09950 - NA
Pyrophosphate phosphohydrolase (Fig 5)
Triphosphate phosphohydrolase
Asp-tRNA(Asn)L-glutamine amido-ligase (ADP-forming) (Fig 6)
M744_02485 and M744_04675 and M744_05790
M744_02485 - Glu-tRNA(Gln)L-glutamine amido-ligase(ADP-forming)
M744_04675 - Glu-tRNA(Gln)L-glutamine amido-ligase(ADP-forming)
M744_05790 - Glu-tRNA(Gln)L-glutamine amido-ligase(ADP-forming)
L-AsparaginetRNA(Asn) ligase (AMP-forming) (Fig 7)
L-Aspartateammonia ligase (AMP-forming) (Fig 8)
ATPAMP phosphotransferase (Fig 9)
Cytochrome b6/f complex periplasm (Fig 10)
Cytochrome b6/f complex, Plastoquinone Oxidized-plastocyanin oxidoreductase
NAD(P)H dehydrogenase (plastoquinone-8 & 3 protons) (periplasm) (Fig 11)
M744_06540 and M744_06535 and
M744_09900 and M744_09515 and
M744_10750 and M744_09885 and
M744_09890 and M744_07765 and
M744_09895 and M744_10740
and M744_10745 and M744_06510
and M744_05170 and M744_01630 and
M744_01625 and M744_03420 or
M744_09385 and M744_06535 and
M744_09900 and M744_09515 and
M744_10750 and M744_09885 and
M744_09890 and M744_07765 and
M744_09895 and M744_10740 and
M744_10745 and M744_06510 and
M744_05170 and M744_01630 and
M744_01625 and M744_03420
M744_06540, M744_06535, M744_09385, M744_06535 - NAD(P)H dehydrogenase (plastoquinone-8 & 3 protons) (tilacoide)
M744_09900, M744_09515, M744_09885, M744_06510, M744_05170, M744_09900, M744_09515, M744_09885, M744_06510, M744_05170 - NAD(P)H dehydrogenase (plastoquinone-8 & 3 protons) (tilacoide), Active co2 transporter facilitator (tilacoide), Active co2 transporter facilitator (periplasm)
M744_10750, M744_09890, M744_07765, M744_09895, M744_10740, M744_10745, M744_10750, M744_09890, M744_07765, M744_09895, M744_10740, M744_10745 - NADPHp-benzoquinone oxidoreductase, NADHquinone oxidoreductase, NAD(P)H dehydrogenase (plastoquinone-8 & 3 protons) (tilacoide), Active co2 transporter facilitator (tilacoide), Active co2 transporter facilitator (periplasm)
M744_01630, M744_01625, M744_03420, M744_01630, M744_01625, M744_03420 - Bidirectional [NiFe] Hydrogenase, NAD(P)H dehydrogenase (plastoquinone-8 & 3 protons) (tilacoide), Active co2 transporter facilitator (tilacoide), Active co2 transporter facilitator (periplasm)
Cytochrome C Oxidase, Synechocystis (2 protons) (Fig 12)
M744_02050 or ( M744_03160 and M744_03155 and M744_03150 )
M744_02050, M744_03160, M744_03155, M744_03150 - cytochrome c oxidase, synechocystis (2 protons periplasm)
2,3,4,5-TetrahydrodipicolinateNAD+ oxidoreductase (Fig 13)
2,3,4,5-TetrahydrodipicolinateNADP+ oxidoreductase
2,3,4,5-TetrahydrodipicolinateNADP+ oxidoreductase (Fig 14)
2,3,4,5-TetrahydrodipicolinateNAD+ oxidoreductase
Sedoheptulose 1,7-bisphosphate D-glyceraldehyde-3-phosphate-lyase (Fig 15)
D-Fructose 1-phosphate D-glyceraldehyde-3-phosphate-lyase,
D-Fructose-1,6-bisphosphate D-glyceraldehyde-3-phosphate-lyase
FerredoxinNADP+ oxidoreductase (Fig 16)
Oxygen transport ext (Fig 17)
ATP synthase (four protons for one ATP) (periplasm) (Fig 18)
M744_01360 and M744_01350 and M744_01345 and M744_01335 and M744_01340 and M744_01330 and M744_04715 and M744_04710 and M744_01355 and M744_01365
ALL - ATP synthetase(u)
2-Phospho-D-glycerate 2,3-phosphomutase (Fig 19)
M744_00620 and M744_06000 and M744_00535 and M744_08585
M744_00620 - 3-Phospho-D-glyceroylphosphate 2,3-phosphomutase
Carbonic acid hydro-lyase (Fig 21)
M744_09475 or M744_09655 or M744_09335 or M744_14040
M744_09335 - Carbonic anhydrase
Nitrate exchange (Fig 22)
ATPnucleoside-diphosphate phosphatransferase (Fig 23)
D-Ribulose-5-phosphate 3-epimerase (Fig 24)
Acetyl-CoAenzyme N6-(dihydrolipoyl)lysine S-acetyltransferase (Fig 25)
PyruvateNAD+ 2-oxidoreductase (CoA-acetylating)
L-Valinepyruvate aminotransferase (Fig 26)
Sucrose exchange (Fig 27)
SPS (Fig 28)
SPP (Fig 29)
cscB (Fig 30)
UTP-glucose-1-phosphate uridylyltransferase (Fig 31)
ATPnucleoside-diphosphate phosphate transferase (Fig 32)
L-Aspartate2-oxoglutarate aminotransferase (Fig 33)
L-Phenylalanine2-oxoglutarate aminotransferase, L-Tyrosine2-oxoglutarate aminotransferase, L-Cysteine2-oxoglutarate aminotransferase, 4-Hydroxy-L-glutamate2-oxoglutarate aminotransferase
L-Phenylalanine2-oxoglutarate aminotransferase (Fig 56)
M744_03485 or M744_12480
M744_03485 - Same as above
M744_12480 - L-Tyrosine2-oxoglutarate aminotransferase, LL-2,6-diaminoheptanedioate2-oxoglutarate aminotransferase
L-Tyrosine2-oxoglutarate aminotransferase (Fig 58)
M744_03485 or M744_12480
Same as above
(S)-Malate hydro-lyase (Fig 34)
L-HomoserineNAD+ oxidoreductase (Fig 35)
(3R)-3-Hydroxyhexanoyl-[acyl-carrier-protein]NADP+ oxidoreductase (Fig 36)
M744_13375 or M744_08910
M744_13375, M744_08910 - (3R)-3-Hydroxybutanoyl-[acyl-carrier protein]NADP+ oxidoreductase, (3R)-3-Hydroxytetradecanoyl-[acyl-carrier-protein]NADP+, 3oxoacyl-acp reductase (c180), (3R)-3-Hydroxyoctanoyl-[acyl-carrier-protein]NADP+ oxidoreductase, 3-oxoacyl-[acyl-carrier-protein] reductase (n-C121), 3-oxoacyl-[acyl-carrier-protein] reductase (n-C141), 3-oxoacyl-[acyl-carrier-protein] reductase (n-C161), 3-oxoacyl-[acyl-carrier-protein] reductase (n-C181)
(3R)-3-Hydroxybutanoyl-[acyl-carrier protein]NADP+ oxidoreductase (Fig 37)
M744_13375 or M744_08910
Same as above
(3R)-3-Hydroxydodecanoyl-[acyl-carrier-protein]NADP+ oxidoreductase (Fig 38)
M744_13375 or M744_08910
ame as above
(3R)-3-Hydroxyoctanoyl-[acyl-carrier-protein]NADP+ oxidoreductase (Fig 39)
M744_13375 or M744_08910
Same as above
(3R)-3-Hydroxypalmitoyl-[acyl-carrier-protein]NADP+ oxidoreductase (Fig 42)
M744_13375 or M744_08910
Same as above
L-Leucine2-oxoglutarate aminotransferase (Fig 40)
M744_09945 or M744_11540
M744_09945 - ​​L-Valine2-oxoglutarate aminotransferase, L-Isoleucine2-oxoglutarate aminotransferase, 4-amino-4-deoxychorismate pyruvate-lyase M744_11540 - ​​L-Valine2-oxoglutarate aminotransferase, L-Isoleucine2-oxoglutarate aminotransferase
3-Isopropylmalate hydro-lyase (Fig 41)
M744_06945 and M744_03470
M744_06945, M744_03470 - (R)-2-Methylmalate hydro-lyase, 2-Isopropylmalate hydro-lyase, 3-Isopropylmalate hydro-lyase
L-Isoleucine2-oxoglutarate aminotransferase (Fig 49)
M744_09945 or M744_11540
Same as above
IMP 1,2-hydrolase (decyclizing) (Fig 43)
5,6,7,8-TetrahydrofolateNADP+ oxidoreductase (Fig 44)
N2-Acetyl-L-ornithine2-oxoglutarate aminotransferase (Fig 45)
M744_02940 or M744_11990
M744_02490 - S-Adenosyl-L-methionine8-amino-7-oxononanoate aminotransferase, N-Succinyl-L-2,6-diaminoheptanedioate2-ocoglytarate M744_11990 - L-Alanine3-oxopropanoate aminotransferase, 4-Aminobutanoate2-oxoglutarate aminotransferase
N-Succinyl-L-2,6-diaminoheptanedioate2-ocoglytarate (Fig 50)
Same as above
N-Acetyl-L-glutamate-5-semialdehydeNADP+ 5-oxidoreductase (Fig 46)
5,10-methylenetetrahydrofolate,FADH2dUMP C-methyltransferase (Fig 47)
1-(5-Phospho-D-ribosyl)-5-amino-4-imidazolecarboxylate carboxy-lyase (Fig 48)
M744_00760 and M744_09450
sn-Glycerol-3-phosphateNADP+ 2-oxidoreductase (Fig 51)
Phosphate exchange (Fig 52)
Orotidine-5'-phosphatepyrophosphate phosphoribosyltransferase (Fig 53)
(S)-Dihydroorotate amidohydrolase (Fig 54)
M744_00530 and M744_00520
L-1-Pyrroline-5-carboxylateNAD+ oxidoreductase (Fig 55)
3-OxopropanoateNAD+ oxidoreductase (decarboxylating),
AcetaldehydeNAD+ oxidoreductase,
beta-Aminopropion aldehydeNAD+ oxidoreductase,
glycolaldehydeNAD+ oxidoreductase,
D-GlyceraldehydeNAD+ oxidoreductase,
4-AminobutyraldehydeNAD+ oxidoreductase,
Propane-1,2-diolNAD+ oxidoreductase,
PhenylacetaldehydeNAD+ oxidoreductase,
4-aminobutanalNAD+ 1-oxidoreductase,
midazole acetaldehydeNAD+ oxidoreductase,
N4-AcetylaminobutanalNAD+ oxidoreductase,
Methylmalonate - semialdehyde dehydrogenase (propanol),
Indole-3-acetaldehydeNAD+ oxidoreductase,
L-1-Pyrroline-5-carboxylateNAD+ oxidoreductase
3-OxopropanoateNAD+ oxidoreductase (decarboxylating) (Fig 77)
Same as above
Sulfate exchange (Fig 57)
Potassium exchange (Fig 59)
diacylglycerol kinase (n-C160) (Fig 60)
diacylglycerol kinase (n-C120),
diacylglycerol kinase (n-C161),
diacylglycerol kinase (n-C180),
diacylglycerol kinase (n-C181),
diacylglycerol kinase (n-C160)
diacylglycerol kinase (n-C161) (Fig 62)
Same as above
diacylglycerol kinase (n-C181) (Fig 78)
Same as above
diacylglycerol kinase (n-C180) (Fig 80)
Same as above
D-Glucosamine 1-phosphate 1,6-phosphomutase (Fig 61)
D-Ribose 1,5-phosphomutase
Magnesium exchange (Fig 63)
UDP-N-acetylmuramate NADP+ oxidoreductase (Fig 64)
Fe2+ exchange (Fig 65)
Fe3+ exchange (Fig 66)
Calcium exchange (Fig 67)
Calcium (Ca+2) transport out via proton antiport (periplasm) (Fig 68)
Sodium exchange (Fig 69)
Sodium transport out via proton antiport (cytoplasm to periplasm) (Fig 70)
(M744_12695 or M744_10315) and M744_04470
Co2+ exchange (Fig 71)
Molybdate exchange (Fig 72)
Mn2+ exchange (Fig 73)
Cu2+ exchange (Fig74)
Zinc exchange (Fig 75)
Nicotinate-nucleotide pyrophosphatase
Phosphoribosyltransferase (Fig 76)
ATPpolyphosphate phosphotransferase (Fig 79)

Targets for Knockouts

We ran Optknock like we did for E. coli, but we were unable to obtain any significant results. All our runs gave a 0 growth rate. We believe the problem may be with the model, as it is incomplete.


  1. Mueller, T. J., Ungerer, J. L., Pakrasi, H. B., & Maranas, C. D. (2017). Identifying the metabolic differences of a fast-growth phenotype in Synechococcus UTEX 2973. Scientific reports, 7(1), 1-8.
  2. Abernathy, M. H., Yu, J., Ma, F., Liberton, M., Ungerer, J., Hollinshead, W. D., ... & Tang, Y. J. (2017). Deciphering cyanobacterial phenotypes for fast photoautotrophic growth via isotopically nonstationary metabolic flux analysis. Biotechnology for biofuels, 10(1), 1-13.
  3. Song, K., Tan, X., Liang, Y., & Lu, X. (2016). The potential of Synechococcus elongatus UTEX 2973 for sugar feedstock production. Applied microbiology and biotechnology, 100(18), 7865-7875.
  4. Lin, P. C., Zhang, F., & Pakrasi, H. B. (2020). Enhanced production of sucrose in the fast-growing cyanobacterium Synechococcus elongatus UTEX 2973. Scientific reports, 10(1), 1-8.


The economic value of our co-culture heavily depends on butanol production by E. coli, making it important to model its growth and butanol production. We used Flux Balance Analysis (FBA) to study the growth dynamics and butanol production in E. coli. We also aimed to optimize the strain by identifying gene targets for overexpression and knockouts in order to improve butanol yields using OptKnock and FSEOF.

Model Construction

We aimed to replicate the KJK01 butanol-producing strain in silico that we received from Dr. ‪Syed Shams Yazdani‬ at ICGEB, Delhi.

We used the iML1515 model from the BIGG database1 which is a metabolic flux model of E. coli MG1655 that contains 1516 genes. This model was the base strain that we further modified to replicate the KJK01 strain. All modifications were made using the COBRA Toolbox in MATLAB and all optimization problems were solved using the gurobi solver2.

We first deleted the reactions corresponding to the following genes that were deleted in the KJK01 strain3, by setting the fluxes through those reactions to zero. The reactions corresponding to the genes in the model were taken from the EcoCyc database4.

  • pta - codes for an enzyme that catalyses the conversion of acetyl-CoA to acetate.
  • adhE - codes for an enzyme that catalyses the conversion of acetyl-CoA to ethanol.
  • frdA - codes for an enzyme that catalyzes the conversion of fumarate to succinate.
  • ldhA - codes for an enzyme that catalyses the conversion of pyruvate to lactate.

These genes catalyze reactions that direct flux away from the butanol producing pathway and produce acetate, ethanol, succinate and lactate respectively, which impact butanol yields. Thus their deletion will reroute flux back into the butanol producing pathway and increase yields.

We then added reactions catalyzed by the heterologous genes expressed in E. coli in order to construct a butanol-producing pathway in the KJK01 strain. We added the reactions using the MetaCyc5 database and the KEGG6,7,8 database.

These were reactions catalyzed by the following genes:

  • From Clostridium acetobutylicum:
    • hbd - codes for beta-hydroxybutyryl-CoA dehydrogenase which converts acetoacetyl-coA to (S)-3-hydroxybutanoyl-CoA
    • crt - codes for 3-hydroxy-butyryl-CoA dehydratase that converts (S)-3-hydroxybutanoyl-CoA to crotonyl-CoA
    • adhE2 - codes for butyraldehyde/butanol dehydrogenase, which is a multifunctional enzyme with both alcohol dehydrogenase and acetaldehyde dehydrogenase activities.
  • From Treponema denticola
    • ter - codes for trans-enoyl-coenzyme A (CoA) reductase that converts acyl-CoA to trans-2,3-dehydroacyl-CoA
  • From E. coli
    • atoB - coding for acetyl-CoA acetyltransferase that converts acetyl-coA to acetoacetyl-coA.

We also added the reactions catalyzed by the cscA (invertase), cscB (sucrose permease), cscK (fructokinase) genes from E. coli W to allow the strain to be able to consume and metabolize sucrose9.

However, we noticed that some of the heterologous gene reactions corresponding to the hbd, crt, cscK, and atoB genes already existed in the base iML1515 strain, we first decided to consider the flux through those reactions as overexpressions of the corresponding genes. However, after simulating these overexpressions and after further consultation with our mentors, we realized that this did not have any particular relevance in simulating IPTG induction of the butanol production pathway. We instead set butanol production as the objective function and constrained growth to a particular percentage of the optimal growth rate for a particular set of conditions, as a way to simulate IPTG induction.

Flux Variability Analysis for Overexpression Targets

We used flux variability analysis (FVA)11 to find the range of possible fluxes through the reactions of the crt, atoB, cscK genes and the butanol exchange reaction at optimal production of butanol.

FVA can be used to find the minimum and maximum fluxes for any reaction for a particular state of the metabolic network. The FVA for these reactions were performed in anaerobic conditions, when growth was set to 60% of its optimal value, with butanol production as the objective function.

Reaction Name
Minimum Flux
Maximum Flux
ECHOAH1 (CRT2) (by crt gene)
ECOAH2 (CRT1) (by crt gene)
ACACT1r (by adhE gene)
Butanol exchange reaction
HACD1 (by hbd gene)
HEX7 (by cscK gene)

Thus we can see what optimal overexpression fluxes are required to maximize butanol yields at 60% of optimal growth rate.


Growth Rate v/s Sucrose

A graph representing the relation between varying sucrose uptaek values and growth rates. We obsereve a linear relation between the two, where growth rate increases with increase in sucrose uptake value for both aerobic and anaerobic scenario. The aerobic system has a steeper rise

Fig 1: Relationship between sucrose uptake and growth rate.


  1. Growth rate increases with increase in sucrose uptake rate
  2. Growth rate is higher under aerobic conditions for all sucrose uptake rates as compared to anaerobic conditions


Growth rate increases linearly with sucrose uptake rate as expected, since the carbon flux from sucrose metabolism will be directed to the biomass reaction, leading to an increase in growth rate. This confirms that the cscA, cscB, cscK reactions added to the model work in the metabolic network and the model can grow on sucrose as a carbon source.

Butanol Production vs Growth Rate

The objective function was first set to maximize the growth rate of the model. Butanol production was then set as the objective function for different percentages of the optimal growth rate at a sucrose uptake rate of 10 mmol/gDW/h to obtain the theoretical maximum yield of butanol if E. coli were growing at X% of its optimal growth rate.

A graph showing the relation between butanol production and growth rate. Growth rate decreases linearly with increase in butanol production, for both aerobic and anaerobic systems. The line for aerobic system has a steeper decline than the anaerobic system

Figure 2: Relationship between maximum butanol production and growth rate.


  1. The maximal butanol yield decreases as the growth rate increases.
  2. Maximal butanol production is much higher for anaerobic conditions as compared to aerobic conditions.
  3. The optimal growth rate (100% on the graph) leads to zero butanol production in both aerobic and anaerobic conditions.


If E.coli grows at its optimal growth rate, butanol production is zero as would be expected, since it is a secondary metabolite that takes carbon flux away from the growth.

As the growth rate is decreased from its optimal value, the maximal theoretical yield of butanol increases implying that carbon flux is rerouted to the butanol producing pathway in the network.

This confirms that the butanol production pathway in the model is functional. Butanol production happens under anaerobic or microaerobic conditions 10. We wanted to thus test the effect of oxygenation on butanol production in the model, and as expected found that butanol production was significantly higher in anaerobic conditions as compared to aerobic conditions at all growth rates.

Butanol Production vs Sucrose Uptake

A grpah depicting the relation between the maximum butanol production given a sucrose uptaek rate. We observe a linear pattern with the max butanol production increasing with increase in sucrose uptake values for both aerobic and anaerobic systems. The anaerobic line has a steeper rise.

Figure 3: Relationship between maximum butanol production and sucrose uptake when growth is 60% of its optimal value


  1. Maximum butanol production increases linearly with increasing sucrose uptake
  2. Maximum butanol production is higher for anaerobic conditions as compared to aerobic conditions at all sucrose uptake rates


Increasing sucrose uptake rates lead to an increase in butanol production when butanol is the objective function. This implies that with changing sucrose concentrations, both growth rates and the sucrose production rate (per gram dry weight) will change accordingly.

Butanol Production vs Oxygen Uptake

Our co-culture modeling results indicated that butanol production in the co-culture is optimum at microaerobic conditions as compared to anaerobic conditions and we wanted to test whether that would hold true for a monoculture as well, by testing the maximum butanol production at varying oxygen uptake rates, when growth was set to 60% of the optimal value.

A graph depicting the relation between butanol production and oxygen uptake rates. We observe a linear relation between the two, where as the oxygen uptake value increases, the butanol production value decreases

Figure 4: Relationship between maximum butanol production and oxygen uptake when growth is 60% of its optimal value


  1. As oxygen uptake rates increase, maximum butanol production decreases


In a monoculture, there is a slight decrease in maximal butanol production in microaerobic as compared to anaerobic conditions.

Side Products

Abdelaal el. al3 observed pyruvate, ethanol, and butyrate to be the major side products in the KJK01 butanol producing strain, which divert flux away from the butanol production pathway and reduce yields. Pyruvate was the major side product that indicated a redox imbalance, while ethanol was produced in equal quantities even after the deletion of the endogenous alcohol dehydrogenase gene (adhE).

We wanted to estimate the maximal side product formation at different growth levels in anaerobic and aerobic conditions and see whether our results matched with those reported by Abdelaal el. al3.

We used flux variability analysis (FVA)11 to find the maximum possible fluxes through the side product reactions under a growth rate constraint. FVA can be used to find the minimum and maximum fluxes for any reaction for a particular state of the metabolic network. For example, it can be used to find the minimum and maximum fluxes for a reaction such that X% of the optimal growth rate is maintained.

We obtained the optimal growth rates for aerobic and anaerobic conditions and for a particular percentage of that growth rate, FVA was performed for the exchange reactions of butanol, ethanol, pyruvate, butyrate and sucrose uptake to find the maximal fluxes through those reactions. The minimum fluxes for these reactions will be zero since all the carbon flux will be directed toward growth.

Fig 13: Maximal side product formation as a function of growth in anaerobic conditions

Fig 14: Maximal side product formation as a function of growth in aerobic conditions


  1. In anaerobic and aerobic conditions, maximal pyruvate production is consistently higher than the other side products.
  2. Maximal butyrate production seems to be almost equal to maximal butanol (particularly in aerobic conditions) whereas in both anaerobic and aerobic conditions, ethanol production is zero.
  3. Sucrose consumption in both anaerobic and aerobic conditions is quite similar.


Consistent with the observations of Abdelaal el. al3, pyruvate is a major possible byproduct. Ethanol was zero, which is to be expected since there is only one gene (adhE) that produces ethanol from acetyl-coA, whose flux we set to zero to duplicate the KJK01 strain knockout of adhE. In reality, there must be other pathways that produce ethanol that are not present in this genome-scale model. These side-products pose as possible targets of deletions to increase butanol production.


We ran OptKnock on our base model without the knockouts in the KJK01 strain, to validate that the algorithm works, and actually results in knockout targets that are known to increase butanol production.

For OptKnock, we limited the number of knockout combinations to a set of 5. Since the metabolic pathway for butanol production is linked to the TCA cycle, as well as the glycolytic pathway in E. coli, we decided to look for knockouts corresponding to these pathways. We were hoping to find the same deletions as in our constructed model, plus some additional ones. We selected the top 5 sets of knockouts. The results are summarized in the table below.

Optknock Table

S. No.
Knockouts suggested by OptKnock
Butanol production
Acetate kinase, Alcohol dehydrogenase (ethanol), D-lactate dehydrogenase, Triose-phosphate isomerase
Acetaldehyde dehydrogenase (acetylating), Fumarate reductase, Acetate kinase, D-lactate dehydrogenase, Triose-phosphate isomerase
Acetaldehyde dehydrogenase (acetylating), Phosphotransacetylase, Alcohol dehydrogenase (ethanol), D-lactate dehydrogenase, Triose-phosphate isomerase
Fumarate reductase, Phosphotransacetylase, Alcohol dehydrogenase (ethanol), D-lactate dehydrogenase, Triose-phosphate isomerase
Acetaldehyde dehydrogenase (acetylating), Fumarate reductase, Phosphotransacetylase, D-lactate dehydrogenase, Triose-phosphate isomerase

We observed that all the deletions (except ‘ACLD19’) we made in our constructed model showed up in some or the other combination of the OptKnock results. We observed a growth rate of 0.17 and butanol production of 18.6 in our suggested deletions. Furthermore, D-lactate dehydrogenase (LDH_D) and Triose-phosphate isomerase (TPI) were suggested in all the top 5 OptKnock deletions. While the suggested reactions have been knocked out in our constructed model, TPI is a new deletion suggested by OptKnock. TPI catalyzes the conversion of Dihydroxyacetone phosphate (DHAP) to 3-Phosphoglyceraldehyde (PGAL). It isn’t clear how this helps in butanol production yet.


We ran FSEOF on our constructed model to identify reactions whose fluxes increased with an increase in the enforced flux through the butanol exchange reactions. The fluxes through a total of 65 reactions showed a strict increase with an increase in enforced butanol flux. Some of the reactions showing a large increase in flux include, proton transfer, CO2 exchange but also reactions of genes like pyruvate dehydrogenase, pyruvate kinase, Acyl-CoA dehydrogenase, 3-hydroxyacylCoA dehydratase, 3-hydroxyacylCoA dehydrogenase. Apart from the transport and exchange reactions of butanol, we did not observe any increase in flux in the reactions added by us. The full list of reactions is available here.


  1. King, Z. A., Lu, J., Dräger, A., Miller, P., Federowicz, S., Lerman, J. A., ... & Lewis, N. E. (2016). BiGG Models: A platform for integrating, standardizing and sharing genome-scale models. Nucleic acids research, 44(D1), D515-D522.
  2. Gurobi Optimization, L.. (2021). Gurobi Optimizer Reference Manual.
  3. Abdelaal, A. S., Jawed, K., & Yazdani, S. S. (2019). CRISPR/Cas9-mediated engineering of Escherichia coli for n-butanol production from xylose in defined medium. Journal of Industrial Microbiology and Biotechnology, 46(7), 965-975.
  4. Keseler, I. M., Collado-Vides, J., Santos-Zavaleta, A., Peralta-Gil, M., Gama-Castro, S., Muñiz-Rascado, L., ... & Karp, P. D. (2010). EcoCyc: a comprehensive database of Escherichia coli biology. Nucleic acids research, 39(suppl_1), D583-D590.
  5. Caspi, R., Altman, T., Billington, R., Dreher, K., Foerster, H., Fulcher, C. A., ... & Karp, P. D. (2014). The MetaCyc database of metabolic pathways and enzymes and the BioCyc collection of Pathway/Genome Databases. Nucleic acids research, 42(D1), D459-D471.
  6. Kanehisa, M., & Goto, S. (2000). KEGG: kyoto encyclopedia of genes and genomes. Nucleic acids research, 28(1), 27-30.
  7. Kanehisa, M. (2019). Toward understanding the origin and evolution of cellular organisms. Protein Science, 28(11), 1947-1951.
  8. Kanehisa, M., Furumichi, M., Sato, Y., Ishiguro-Watanabe, M., & Tanabe, M. (2021). KEGG: integrating viruses and cellular organisms. Nucleic acids research, 49(D1), D545-D551.
  9. Bruschi, M., Boyes, S. J., Sugiarto, H., Nielsen, L. K., & Vickers, C. E. (2012). A transferable sucrose utilization approach for non-sucrose-utilizing Escherichia coli strains. Biotechnology advances, 30(5), 1001-1010.
  10. Wu, P., Wang, G., Wang, G., Børresen, B. T., Liu, H., & Zhang, J. (2016). Butanol production under microaerobic conditions with a symbiotic system of Clostridium acetobutylicum and Bacillus cereus. Microbial cell factories, 15(1), 1-11.
  11. Mahadevan, R., & Schilling, C. H. (2003). The effects of alternate optimal solutions in constraint-based genome-scale metabolic models. Metabolic engineering, 5(4), 264-276.


The co-culture aspect makes this project unique and allows us to bring the properties of cyanobacteria and
E. coli together to form the biomanufacturing unit. Although the organisms have been modelled individually, their behaviour changes in the co-culture setup, affecting the viability and productivity of the whole system. A model that takes into account interactions between the two species can thus offer additional insights and alternative techniques to optimize the production process.
Two techniques have been used to model the co-culture - SteadyCom and Dynamic Modelling. SteadyCom assumes a community steady-state that allows us to predict the stability of the co-culture and long-term interactions between the organisms. The dynamic modelling algorithm, developed by the INSA-UPS Toulouse iGEM Team, allows us to track the development of the co-culture over time. Together, these two models built on slightly different assumptions can offer a wide range of insights.

*Note from the wiki editors: The graphs on this page which include sliders may not instantly load. Please try moving the slider slowly once before viewing the graphs.


Building the Community Model of S. elongatus UTEX 2973 and E. coli

SteadyCom uses a joint community model which is made by putting together the genome-scale models (GSMs) of the individual species in the community. The individual models used to build the community model of our co-culture were the modified versions of iML1515 and iSyu683 that were modified to match KJK01 and the sucrose-producing cyanobacteria (S. elongatus UTEX 2973). All the SteadyCom simulations including the creation of the joint model were performed using the COBRA toolbox on MATLAB. During the process of creating the joint model however, we faced many issues due to lack of documentation on SteadyCom and standardisation of GSM models.

To troubleshoot the difficulties we were facing, we built joint models analogous to the one we needed, but using two copies of a single species’ GSM. Working with such models gave us insights into the intricacies of joint models, and we were eventually able to resolve the issues we faced. This led to us deciding to create a checklist that we believe would greatly help future users of SteadyCom on COBRA navigate their way through creating joint community models and troubleshooting. You can read more about that on our contributions page here : Contributions.

IPTG induction:
Butanol is a secondary metabolite, and is produced at the cost of growth, by diverting carbon flux. Optimizing for growth rate in SteadyCom would thus not lead to any butanol production from E. coli.
In the actual co-culture, the induction by IPTG would force carbon flux through the butanol production pathway. To simulate this we set a lower bound for the butanol exchange reaction which would constrain the model to have flux through the butanol production pathway in E. coli. The particular value for the lower bound was obtained through FBA at different percentages of optimal growth rate - that is, the fraction of the maximum possible growth rate that is exhibited under specific conditions. You can read about the exact method we used to set the lower bound here. In essence, the higher the percentage of optimal growth, the lower the butanol productivity.
The carbon dioxide uptake rate of the cyanobacteria was set at -12.2mmol/gDW/hour. Salt stress was simulated by setting the sucrose produced by cyanobacteria as per the results of FBA at 90% cscB efficiency. The butanol production for E. coli was set at 60% of optimal growth. Oxygen exchange between the two species was allowed but no external oxygen was provided to the culture.

Maximum specific growth rate of the co-culture - 0.21 hr-1
Ratio of E. coli biomass to cyanobacteria biomass - 0.17
Butanol production per gdw of the co-culture- 0.149 mmol/hr/gDW
CO2 uptake by the co-culture - 9.1 mmol/hr/gDW

Identifying Key Metabolites Controlling Interactions in the Co-culture

We knew that sucrose secreted by the cyanobacteria would play an important role in the interaction between the two species of the co-culture as there is no other carbon source available for the E. coli. Apart from sucrose, we ran SteadyCom FVA to identify other metabolites that could play crucial roles in the interactions of the consortium. SteadyCom FVA (Flux Variability Analysis) calculates the maximum and minimum values of flux through a set of reactions while maintaining the optimal growth rate at steady state. By running FVA on the set of transport reactions of metabolites between the individual species and the community space, we identified those metabolites that are required to be taken up or given out by each species to maintain optimal community growth rate.
Below is a table showing the maximum and minimum allowed fluxes of the external metabolites into E. coli and S. elongatus at optimal community growth rate. Negative values denote a net uptake into the organism and positive values denote a net export.

MetaboliteMin Flux Through E coliMax Flux Through E coliMin Flux Through CyanobacteriaMax Flux Through Cyanobacteria
Carbon Dioxide2.0154430682.035662892-16.26443278-16.25798452
H+ (Proton)-81.613113720.466652326300.067976675
Oxygen (O2)-21.97422024-1.45211356621.944778121.97422024
Urea 00.00521225522300
Water (H20)2.65185804343.69752008-13.54056988-13.50658154

Apart from sucrose, two other metabolites that must undergo exchange between the two organisms to grow optimally are carbon dioxide and oxygen. Carbon dioxide is given out by E. coli and taken in by the cyanobacteria and is already known to be an integral part of the production process, as it acts as the main carbon source for the entire process. Cyanobacteria could act as an important source of oxygen for E. coli, who prefer aerobic conditions for growth.
Apart from these, there were other metabolites such as sulfate, magnesium, phosphate, etc that both bacteria require to be supplied in the medium. This would imply that a shortage of these nutrients could lead to competition between the two species and destabilisation of the co-culture.


  1. For the rest of this page:
  2. Sucrose productivity - Sucrose secreted per hour per gDW of S elongatus
  3. Butanol productivity - Butanol secreted per hour per gDW of E. coli
  4. Butanol yield of the co-culture - Butanol produced per hour per gDW of the co-culture

Varying Sucrose Productivity of the Cyanobacteria

Increasing sucrose productivity of cyanobacteria will result in the increase of butanol yield from the co-culture.

Currently, according to the S. elongatus model, cyanobacteria produce about 0.3 mmol/hr/gDW under salt stress. But we decided to vary the sucrose productivity of cyanobacteria and model its effects on the butanol yield of the co-culture. This was done by changing the flux of sucrose given out by cyanobacteria per gDW.

Figure Desc: The above graphs show the results of SteadyCom runs at varying values of sucrose productivity of S. elongatus

As the sucrose productivity of cyanobacteria increases:

  1. The max growth rate of the co-culture comes down.
  2. The relative abundance of the cyanobacteria in the co-culture goes down and that of the E coli goes up.
  3. The butanol produced per gDW of the co-culture initially increases but decreases beyond an optimal value of sucrose productivity.

For current levels of oxygen and CO2 uptake rates, there exists an optimum value of sucrose productivity from cyanobacteria which gives a maximum butanol yield per gDW of the co-culture. This is due to the fact that the sucrose productivity of the cyanobacteria is inversely correlated with its relative abundance in the co-culture. Thus, very high values of sucrose productivity (which is per gDW of the cyanobacteria) correspond to low cyanobacterial abundance in the culture and hence less sucrose for the E. coli to take up and produce butanol. A more detailed explanation of this result is given in the appendix.

Varying butanol productivity of E coli

Increase in the butanol productivity of E. coli will result in an increase of the butanol yield of the co-culture.

Along with the sucrose productivity of cyanobacteria in our analysis is another important factor - the butanol productivity of E. coli. As a consequence of how induction by IPTG is simulated in the model, a higher butanol productivity is correlated with the E. coli growing at a lower percentage of its optimal/maximum growth rate at the given conditions. In reality, increasing butanol productivity could be achieved by a higher IPTG concentration, or by increasing the efficiency of the butanol pathway. This would redirect carbon flux away from growth and towards producing butanol.
The butanol productivity was increased by decreasing the percentage of the optimal specific growth rate the E. coli was constricted to grow at.

Figure Desc: The above graph shows how the butanol yield from the co culture varies with sucrose productivity at different values of butanol productivity (depicted by % of optimal growth of the E. coli)

As butanol productivity of E. coli increases:

  1. The community growth rate of the co-culture remains the same for a particular value of sucrose productivity.
  2. The butanol produced per gDW of the co-culture increases.
  3. The value of optimum sucrose productivity did not vary significantly at different levels of butanol productivity.

The maximum specific growth rate of the co-culture is primarily controlled by the cyanobacteria and will hence majorly depend on the CO2 uptake rate and sucrose productivity. When butanol productivity of the E. coli was increased, its relative abundance in the co-culture decreased. This means that since there were fewer E. coli in the co-culture, they could take up more sucrose per gDW and produce more butanol while still matching the growth rate of the cyanobacteria.
Increasing the butanol productivity of E. coli does increase the butanol yield of the co- culture. This is as each of the E. coli are now diverting more flux into the butanol-producing pathway and hence are converting a higher fraction of the sucrose given out by the cyanobacteria into butanol. However, this comes at the cost of decreasing relative abundance of E. coli in the co- culture. While running SteadyCom, it was not possible for the E. coli to grow below around 50% of its optimal growth rate. From which we can infer that the butanol productivity cannot be arbitrarily increased as after a point the co-culture would not be able to maintain a steady state.

These points can be seen in the two graphs below:

Varying Oxygen Levels to Test Aerobic vs Microaerobic vs Anaerobic

Decreasing oxygen levels will result in an increase of the butanol yield of the co-culture.

As seen in the FVA results above, oxygen was one of the important metabolites being exchanged by the cyanobacteria and E. coli. It is to be noted that cyanobacteria are producing more oxygen than the E. coli need and the co-culture as a whole would have a positive production of oxygen (see appendix for more). This means that our simulations would be in an aerobic environment as far as the E. coli are concerned. However, based on discussions with experts and FBA results for E. coli, we realised that anaerobic or even microaerobic conditions might result in a higher butanol yield. Thus we decided to look at how lowering oxygen levels in the co-culture would affect butanol yield. We did this by setting decreasing bounds to how much oxygen the E. coli could take up per gDW.

Oxygen bound (Maximum allowed rate of oxygen uptake for E. coli)

Move the slider to see how the butanol yield of the co culture varies at different levels of oxygen in terms of the bound on the Oxygen uptake given in mmol/gDW/hr.

Fig Description: Results of SteadyCom when E. coli is constrained to not take up any oxygen. The marker for current sucrose productivity shows that the E. coli does not grow in these conditions as its relative abundance is 0.

Fig Description: The above graph shows how butanol yield of the co culture varies with oxygen levels at a sucrose productivity of 0.4mmol/gDW/hr

As the oxygen levels in the co culture environment is brought down:

  1. The maximum possible butanol yield of the co-culture increases.
  2. The sucrose productivity value for which butanol yield of the co-culture is maximumized increases.

However at very low oxygen levels, close to anaerobic conditions:

  1. The E. coli cannot grow at low sucrose productivity of the cyanobacteria, including the current productivity predicted by the S. elongatus model at a salt stress of 150mM.
  2. For sucrose productivity in the range 0.3 - 0.5 mmol/hr/gDW, the butanol yields do not change much with a further decrease in oxygen levels.

Butanol yield does increase as the oxygen levels are brought down; however, at very low levels of oxygen, the E. coli are not able to sustain growth with the current predicted levels of sucrose productivity from the cyanobacteria. Furthermore, the butanol yield at low oxygen levels does not increase significantly unless sucrose productivity of the cyanobacteria is increased to a very high level.
We are thus in a position to corroborate one of the inputs given to us by Dr. Yazdani - viz., that good butanol yields are best achieved under microaerobic conditions, as the E. coli divert most of their carbon flux towards growth under aerobic conditions but grow too slowly under anaerobic conditions. We can use this to inform a practical implementation of our project, and would recommend experimenting with oxygen quenchers to bring oxygen levels down into the appropriate range.
This is also supported by the fact that at very low oxygen levels, the E. coli cannot grow below 90% of its maximum growth rate. The only way for the E. coli to divert more flux towards butanol would be to further lower its relative abundance to take up more sucrose per gram dry weight.

Increasing CO2 Uptake Rate of Cyanobacteria

Increasing the CO2 uptake efficiency of cyanobacteria will increase the butanol yield of the co-culture.

In a meeting with Dr. Malathy, Dr. Dube and Dr. Lobo, we were told to look into increasing the CO2 uptake efficiency of the cyanobacteria to give our production process a competitive advantage over others. We decided to model the effect of this increase on the butanol yield as well. We expected that if CO2 uptake efficiency increases, cyanobacteria biomass and sucrose would be produced at a higher rate, resulting in butanol being produced at a higher rate as well. This was done by increasing the maximum CO2 allowed for the cyanobacteria to take in per hour.

Carbon Uptake Efficiency

Move the slider to see how the butanol yield of the co-culture varies at different carbon uptake efficiency of S. elongatus given in mmol/gDW/hr.


  1. As the CO2 uptake rate of the cyanobacteria is increased, the butanol yield of the co-culture given in mmol/h/gDW increases.
  2. The co-culture is stable at much higher rates of sucrose productivity of cyanobacteria.

Increasing the CO2 uptake efficiency of cyanobacteria will result not only in more CO2 being captured by the co-culture every hour, it will also increase the butanol being produced per hour by the co-culture. This is because a higher CO2 uptake rate would result in a higher sucrose productivity (as is also predicted from FBA and MOMA) which would naturally result in an increased butanol yield.

Dynamic Modelling

Varying Sucrose Productivity of the Cyanobacteria

Dynamic modeling will generate results that concur with SteadyCom’s core conclusion - that there is an optimum value sucrose productivity at which the butanol yield is maximised.

The dynamic simulation was performed under aerobic conditions for 60% butanol productivity of E. coli. Initially, as sucrose productivity of cyanobacteria was increased, the total butanol yield also increased. However, when increasing sucrose productivity beyond 70%, the butanol yield starts decreasing.

Sucrose Productivity

This can be explained because the total amount of sucrose produced is proportional to both the abundance of cyanobacteria and the sucrose productivity of the cyanobacteria, but increasing the sucrose productivity diverts carbon flux away from growth. This causes the biomass of cyanobacteria to grow slower, but each individual is producing sucrose more efficiently. Since such a trade-off exists, the optimum sucrose yield ends up being around 70% sucrose productivity of cyanobacteria, which in turn gives maximum butanol yield.

Varying Butanol Productivity of E. coli

As butanol productivity of E. coli increases, total butanol yield will increase.

The dynamic simulation was done at 60% sucrose productivity from cyanobacteria, at aerobic conditions. The simulations suggest that as butanol productivity of the E. coli increases, from 10% to 95%, the butanol yield of the co-culture keeps increasing.

Butanol Productivity

Methods to increase butanol productivity will help increase butanol yield. However, it is not biologically realistic to reach 95% butanol productivity as shown in the simulation. This would divert too much flux away from growth of the E. coli, which is not feasible in real life.

Varying Oxygen Levels to Test Aerobic vs Microaerobic vs Anaerobic

SteadyCom suggests that micro aerobic conditions are ideal for butanol productivity. Dynamic modeling is supposed to generate concurrent results.

The simulations were run at 60% productivity of sucrose and butanol aerobically. The oxygen uptake flux of E. coli was already well below the oxygen produced by the cyanobacteria. This is because at the sucrose flux available to the E. coli, high oxygen consumption was not required for optimal growth and productivity.
However, as we decrease the oxygen uptake of E. coli, we notice that butanol productivity initially increases. This could be explained by the fact that E. coli starts diverting some of its carbon flux into a different anaerobic pathway, which results in a lower growth rate. This in turn implies that there is more carbon flux available to produce butanol efficiently.
This trend is not sustained as oxygen uptake is further decreased. This could be explained as follows: as oxygen uptake is decreased, the growth rate of the E. coli goes down significantly and brings down the total butanol yield.

Oxygen Uptake Rate

Microaerobic conditions are the most efficient for butanol production in the given co-culture setting.

Effect of CO2

As CO2 input increases in the bioreactor, the productivity of the entire consortium as a system is improved. Increasing the productivity of CO2 sequestration by cyanobacteria should have similar effects.

As CO2 input is increased, the butanol yield, as well as the growth rates of both the organisms increase.
Similar effects are observed as CO2 uptake rate is increased in the cyanobacteria.

CO2 Uptake Rate

Increasing CO2 uptake can be done in two ways - increasing the flow of input CO2 into the bioreactor and increasing efficiency of the cyanobacteria to take up available CO2. They are both good at increasing the yield of butanol.

Summary of results Co-culture

Although the two models begin with slightly different assumptions, the analysis of the co-culture arrives at similar conclusions. This boosts confidence in the accuracy of the results.

  1. There exists an optimum sucrose productivity of the cyanobacteria for which the butanol yield of the co-culture is maximized.
  2. Increasing the butanol productivity of the E. coli will increase the butanol yield of the co-culture.
  3. The oxygen produced by cyanobacteria is in excess of the oxygen required by the E. coli in the co-culture.
  4. Removing the oxygen produced by the cyanobacteria to provide micro-aerobic conditions for the E. coli to grow increases the butanol yield of the co-culture.
  5. Increasing the CO2 uptake efficiency of the cyanobacteria increases the butanol yield of the co-culture.


Simulating IPTG Induction for the Production of Butanol

Similar to how it was done in FBA, IPTG was induced by setting a bound for E. coli to grow at a certain percentage of its optimal growth rate and diverting flux through the butanol production pathway. However due to the nature of the SteadyCom algorithm, this is not as simple to do as in FBA. One cannot optimize for butanol within SteadyCom as the objective function is set to be the community growth rate, and further the sucrose intake by E. coli, on which the optimal growth rate depends, varies. The method we used to overcome the problem was the following:

  1. SteadyCom is run on the joint model while simulating salt stress but not IPTG. The butanol production in this case is usually 0 as growth is being optimized
  2. Sucrose intake by the E. coli per gDW from the SteadyCom result is used as the lower bound for an FBA with the E. coli model
  3. Growth rate is set to a certain percent of optimal at this level of sucrose intake and butanol is optimized for using FBA
  4. Butanol flux gotten from FBA in this manner is set as a lower bound in the joint model and SteadyCom is run once again
  5. This would lead to the production of butanol even though community growth rate is still being optimized for
  6. The true percent of optimal growth rate the E. coli is growing at can be found by comparing the results from the final SteadyCom run to FBA once again
  7. This can be done at different percentages of optimal growth (relating inversely to butanol productivity) as well as different levels of sucrose productivity, oxygen, etc

Detailed Explanation for Optimal Sucrose Productivity

For high values of sucrose productivity, the maximum specific growth rate values of the co-culture are low. This is as the cyanobacteria cannot grow fast while producing high amounts of sucrose per gDW. In turn, due to the steady state conditions, the E. coli is also constrained to grow at a low growth rate. (In reality, if the E. coli do grow at a higher growth rate, they will quickly deplete the sucrose in the medium.) To maintain this condition, the relative abundance of the E. coli in the co culture must be high, so that per gDW of E. coli, there is not much sucrose available and the E. coli is constrained to grow at a lower growth rate.
This makes sense biologically as well. Initially the E. coli will grow fast and deplete the sucrose until it has to slow its growth rate down, but this results in it having a high abundance at steady state.
As sucrose productivity increases the actual butanol produced per gDW of E. coli is decreasing, but its relative biomass in the co culture is increasing. These two competing effects lead to an optimum of sucrose productivity for the butanol productivity per gDW of the co culture.

Image Desc: Results from SteadyCom at different values of sucrose productivity to illustrate the above explanation

Allowing Oxygen into the Co-culture

The two following graphs show net oxygen flux per gDW of the co culture in two different conditions. The first allows only exchange of oxygen between the two species but extra oxygen is not allowed from the environment. The second allows both exchange of oxygen as well as for the E. coli to take up more oxygen from the environment if it is required as well. As one can see, even though the E. coli is allowed to take up more oxygen, it does not require more than the cyanobacteria is producing to grow and produce butanol optimally. This is why the net flux of oxygen is positive in both cases.

Image Desc: The figure on the left shows net oxygen flux through the co-culture when no external oxygen is allowed in. The figure on the right shows net oxygen flux through the co-culture when unlimited external oxygen is allowed to be taken up.