UTEX 2973 Bioreactor: Media Optimization

To maximize the growth of UTEX 2973 in our bioreactors for optimal production of living building materials, we implemented flux balance analysis modelling to determine what the optimal BG-11 media composition and culture conditions are that maximizes UTEX 2973 growth rate.

Introduction to Flux Balance Analysis

Flux Balance Analysis (FBA) is a simplified, constraint-based mathematical way to stimulate the flow of metabolites through an organism’s metabolic network. This network is often referred to as a genomic-scale metabolic map (GSMM) which is developed by annotating a specie’s genome and identifying the metabolites involved in its biochemical reactions and the genes responsible for enzymatic and transport reactions. Due to two assumptions FBA makes, it relies very little (if at all) on enzyme kinetic metrics and metabolite concentrations, making it a more ideal tool for predicting optimal conditions for a specie’s survival.

The first assumption is the steady-state assumption which says that the concentration of produced metabolites is equal to the concentration of metabolites consumed in a reaction. Essentially, the “flux in” equals the “flux out” where a “flux” is the turnover rate for a metabolite in a reaction.

The steady-state assumption

$$\frac {\mathrm{d} \vec{\mathbf{x}}} {\mathrm{d} t} = S \vec{\mathbf{v}} = \mathbf{0},$$

where $\vec{\mathbf{x}}$ is a vector containing the concentrations of all the metabolites, $S$ is the stoichiometric matrix, which is part of the GSMM, and $\vec{\mathbf{v}}$ is a vector containing the fluxes of all the reactions.

The second assumption is that the specie’s metabolic network has been evolutionarily optimized to perform an objective function (Z). The objective function accomplishes a biological goal (e.g. maximized growth, efficient ATP production) and is represented as a linear equation that sums the metabolic fluxes within a network and their individual contributions to this goal.

Objectives

The overarching objective was to characterize the metabolic requirements and limitations of Synechococcus elongatus UTEX 2973 under different environmental conditions. Through testing CO₂ dependence, carbon source preference, nutrient limitations, and oxygen-related knockouts, the modeling work provides insights into how UTEX 2973 could be engineered and optimized for applications such as biocementation and bioink-based construction on Earth and Mars.

• Verify the CO₂ dependence of growth
  • Test how restricting CO₂ uptake affects biomass accumulation in the model.
• Identify optimal carbon sources
  • Compare different exchange reactions to determine which carbon source supports the fastest growth.
• Explore nutrient interactions via phenotypic phase planes
  • Compute biomass as a function of two simultaneous uptake fluxes to map metabolic dependencies and limiting factors.
• Optimize media composition with Flux Variability Analysis (FVA)
  • Identify the critical ranges of nutrient uptake fluxes (from BG-11 components) that biomass production is most dependent on.
• Simulate knockouts of oxygen-related pathways
  • Evaluate the growth impact of deleting genes or reactions related to oxygen usage, which is important for modeling Martian anaerobic conditions.

Methodology

We implemented these analyses using the COBRA Toolbox in MATLAB and Python. In particular, we employed the recently published genome-scale metabolic model of Synechococcus elongatus UTEX 2973 ([1]). Flux optimization algorithms in the COBRA Toolbox were applied for flux balance analysis (FBA) with the single reaction deletion supplemented to mimic the Martian environment and the flux variability analysis (FVA) for media optimization.

Modelling Results

Phenotypic Phase Planes

CO₂ and light were selected as the primary variables for identifying the optimal growth conditions of Synechococcus elongatus UTEX 2973 because they represent the fundamental inputs driving autotrophic metabolism. UTEX 2973 relies on light energy captured by photosystems I and II to generate reducing power and ATP, which fuel carbon fixation through the Calvin—Benson—Bassham cycle. CO₂ serves as the essential carbon source for biomass formation, and its assimilation directly determines the cell’s ability to synthesize macromolecules and grow. Since both light availability and CO₂ uptake are tightly coupled in photosynthetic organisms, varying these parameters provides a mechanistic way to define the metabolic limits and pinpoint the conditions under which biomass accumulation is maximized.

The biomass accumulation rate objective was simulated for various CO₂ uptake (EX_CO2) and photon uptake (EX_PHO1 = EX_PHO2) rate bounds. The CO₂ uptake was bounded from —500 to 0 mmol/gDW/hr, while photon uptake ranged from —2000 to 0 mmol/gDW/h. The analysis identified the optimal growth conditions at a CO₂ uptake rate of -132 mmol/gDW/h and photon uptake rates of -900 mmol/gDW/hr for both photosystems I and II. Under these conditions, the model predicted a maximum biomass accumulation rate of 3.1386 mmol/gDW/h, representing the theoretical upper limit of biomass accumulation achievable within the defined environmental constraints.

Exploring the tradeoffs between other pairs of component uptake rates, this procedure was repeated for CO2 and ammonia, particularly as Mueller et al. identified amino acids, which contain nitrogen, as a significant component of UTEX 2973 biomass ([1]).

The CO₂ uptake was bounded from —1500 to 0 mmol/gDW/hr, while ammonia uptake ranged from —2000 to 0 mmol/gDW/h. The analysis identified the optimal growth conditions at a CO₂ uptake rate of -30 mmol/gDW/h and ammonia uptake rates -30 mmol/gDW/h. Under these conditions, the model predicted a maximum biomass accumulation rate of 0.4323 mmol/gDW/h, the theoretical upper limit of biomass accumulation achievable within the defined environmental constraints.

In order to verify whether the optimal points in the above phenotypic phase plane analyses found were indeed the optima, a hierarchical grid search that progressively zooms into and refines the most promising subspaces was implemented to facilitate significant extension of the search spaces.

Applied to the CO2 vs light uptake phenotypic phase plane analysis beginning with CO2 uptake rate bounded from -10000 mmol/gDW/h to 30 mmol/gDW/h and photon uptake rate tthrough both photosystems I and II from -10000 $\mathrm{ \mu E \cdot m^{-2} \cdot s^{-1} }$ to 30 $\mathrm{ \mu E \cdot m^{-2} \cdot s^{-1} }$, the analysis identified an optimum biomass accumulation rate of 3.138594 mmol/gDW/h at a CO2 uptake rate of -9191.11111 mmol/gDW/h and photon uptake rates through both photosystems of 10000 $\mathrm{ \mu E \cdot m^{-2} \cdot s^{-1} }$. This suggest the previous analysis found only a local optimum.

Single reaction knockout

Single reaction knockout analysis in FBA provides a powerful tool to probe the metabolic robustness of Synechococcus elongatus UTEX 2973 under constrained environments. The objective of the single reaction knockout is to mimic the hypoxic or anoxic atmosphere of Mars. To test if biomass production can be maintained when external oxygen is eliminated, an oxygen transport or exchange reaction knockout is simulated in the model. By systematically removing oxygen fluxes and monitoring biomass growth, we can model the growth and metabolic activity of UTEX2973 in an oxygen-missing external environment. This approach provides critical insights for engineering robust cyanobacterial strains capable of thriving in extraterrestrial environments.

The single reaction knockout of the oxygen exchange reaction and oxygen transport reaction revealed that UTEX 2973 does not fully rely on external oxygen uptake to sustain its growth respectively. This indicated that UTEX 2973 itself produces oxygen through photosynthesis to support its growth. The simulated biomass yield under both knockout conditions was reduced to 0.110 respectively, which is approximately half of the optimal biomass of 0.253. The BG-11 media does not supply oxygen as a media component, which mimics the Martian environment, therefore, the growth rate of UTEX 2973 would show impairment compared to the maximum biomass. This outcome validates the model’s representation of UTEX 2973 as an autotrophic organism capable of sustaining biomass formation without environmental oxygen, an important consideration when extrapolating its potential to survive on Mars.

Flux variability analysis

Flux variability analysis (FVA) recognizes that often many flux vectors/distributions may be optimal, or at least close to optimal. Using the COBRA Toolbox’s fluxVariability function, the ranges of flux for each exchange reaction corresponding to each BG-11 medium that still achieve the same objective function value - here the biomass accumulation/growth rate - as the default FBA solution from optimizeCbModel, were found.

FVA results of BG-11 medium uptake reactions suggested that the exchange reactions of citrate and CO₂ are the two primary constraints for the biomass objective function. The CO₂ flux has a range between 580 and 1000 mmol/gDW/hr, while citrate flux varies from -600 to -180 mmol/gDW/hr. This suggests that the growth of UTEX2973 is mainly driven by citrate uptake, due to its relatively wide flux range, which can be an alternative carbon source and may play a more essential role in biomass accumulation than CO₂ under modelled conditions.

Single Reaction Knockout Validation

We proceeded with validating the single reaction knockout prediction by culturing a flask of UTEX 2973 in anaerobic conditions. The Hallam Lab has an anaerobic chamber, an airtight space that is temperature controlled and only consists of 10% carbon dioxide, 5% hydrogen, and 85% nitrogen gas. Based on modelling the knockout of the oxygen exchange reaction, it is hypothesized that even if growth of UTEX 2973 is slower in anaerobic conditions, the microbe will still survive and the OD750 will increase over time.

In 2025.09.18 Validating UTEX 2973 Growth in Anaerobic Camber, we cultured a flask of UTEX 2973 in the anaerobic chamber as shown above at 30C. Over 7 days of culture, we took OD750 measurements over four time points (0, 34, 70, and 168 hours), resulting in the graph below.

Here, we show that UTEX 2973 is indeed capable of growing in the absence of environmental oxygen, as we observe a linear increasing trend in OD750. We compared this growth rate to growth curves carried out by our Wet Lab team, where UTEX 2973 was cultured with slight changes in conditions, but in aerobic conditions as shown below.

We can observe that all aerobic conditions have a relatively straight line, indicating an approximate exponential growth. Although more OD measurements should have been taken for the sample in the anaerobic chamber, we can see that the UTEX 2973 culture had a long lag phase until it jumped into exponential growth between days 2 and 4. Between 0-50 hours, the slope of the anaerobic condition is significantly lower than those grown in aerobic conditions. Thus, as expected in our modelling, UTEX 2973 can grow in anaerobic conditions, but it is likely non-ideal as exponential growth is very delayed and slow. However, it is also important to note that the chamber consisted mainly of nitrogen gas, and that the low abundance of carbon dioxide may also play a role in slowing UTEX 2973 growth.

Future Directions

The FBA modelling analyses carried out thus far uncover insights into the effect of different media components on UTEX growth individually and interactions between pairs of components in the case of phenotypic phase planes. However, to optimize the growth medium, interactions between all components must be considered.

A first attempt at exploring the full growth landscape over all BG-11 components simultaneously might be to simply extend the previous grid search used to compute the phenotypic phase planes to 16 dimensions, one for each media component. However, the combinatorial nature of such a search space means this would easily become computationally infeasible.

Thus, we employ Bayesian optimal experimental design, a formal Bayesian framework for determining optimal experimental designs adaptively in active/closed-loop experimentation cycles balancing maximum information gain uncovering more of the landscape” with exploiting parts of the space already known to be likely very optimal ([2]), as implemented by the Ax Python platform ([3]).

First, to explore the full search space, in silico optimal Bayesian experiments are performed, where the “observations” used to determine the experimental designs for the next iteration are the optimal growth rates predicted by the deterministic FBA solve under the media component uptake rate bounds being explored by the current iteration.

Eventually, the results of these in silico predictions will be validated following the high-throughput methods demonstrated in Noonan et al. Integrated with automated liquid handlers, it can parallel screen up to 10,000 s of photosynthetic or photo-responsive microorganisms over multiple microplate formats ( [4]). Shown in Figure 9, each rack can hold a microplate with a customized LED array that can shine a controlled light intensity into each well independently.

Thus, given that the lighting system can allow us to screen up to 19,200 cultures in parallel, we can easily validate our FBA outputs to optimize BG-11 media for UTEX 2973. We can also individually vary the concentration of each component in BG-11 to conduct a single component analysis and determine the impact each component plays on the growth of UTEX 2973. Similar to our Design of Experiments approach proposed in Bioink Model Validation, Noonan et al. conducted a Box-Wilson design (central composite) in response surface modelling for UTEX 2973 BG-11 medium optimization. Validating this model in the lighting system led to 38.4% to 61.6% in total biomass accumulation ([4]).

Once we validated our FBA results and have an optimized media composition, we can validate this optimization at a larger scale, such as in our in-house UTEX 2973 bioreactor. This will be a monument step towards optimizing BG-11 medium compositions for growing UTEX 2973 quickly, healthily, and efficiently in a bioreactor.