Computational modeling analysis of the biotechnological exploitability of cyanobacteria photosynthesis considering the spectral light distribution

RWTH Publications (RWTH Aachen)

Abstract

Cyanobacteria are the originators of oxygenic photosynthesis and still perform a large portion of global CO2 fixation today. Moreover, they grow quickly, produce a wide variety of chemical compounds, and can be genetically engineered. Therefore, they are a promising production platform for biomass and high-value chemicals. However, their dynamic behavior requires further investigation to improve the biotechnological exploitation. Therefore, this thesis employs mathematical modeling to systematically study the dynamics of the photosynthetic mechanism and provides a quantitative platform to interpret various experimental measurements. Chapter 1 introduces the background of cyanobacterial photosynthesis and mathematical modeling. It highlights the need for models designed specifically for cyanobacteria. Chapter 2 then provides a detailed description of the modeling methods used throughout the thesis. These methods include the construction of ordinary differential equation (ODE)-based models and the analysis with ensemble models or Metabolic Control Analysis (MCA). Chapter 3 analyzes the effect of various light conditions on photosynthetic electron transport in cyanobacteria. A dynamic, light-spectrum-dependent model of electron transport was developed that describes the process from light absorption to the production of energy-carrying compounds in detail. The model could reproduce experimental measurements of electron fluxes, O2 turnover rates, and fluorescence. MCA revealed that light intensity and spectrum affect the control over carbon fixation, shifting it between the source and sink reactions for electrons. Using an ensemble modeling approach, four hypothesized mechanisms for state transitions in cyanobacteria were compared, finding no physiological benefit for the phycobilisome detachment model. The electron transport model also suggests that the effects of light wavelengths on producing biotechnologically relevant compounds may depend on their required cofactor composition. Chapter 4 focuses on elucidating the dynamic electron transport around photosystem I. ODE models describing the electron transport with increasing complexity were developed and fitted to time-course measurements of the involved redox carriers. The electron transfer from plastocyanin to photosystem I under red light was well described using mass action kinetics. However, a hypothesized formation of complexes between the electron carrier and photosystem was necessary to reproduce measurements under other light conditions. These models provide experimentally relevant kinetic constants for describing photosynthesis. Chapter 5 explores state-of-the-art machine learning methods for interpreting fluorescence-based photosynthesis measurements in cyanobacteria. Two proof-of-concept neural network models were trained on a library of fluorescence induction measurements to predict photosynthetic inhibition. Using culturing information and fluorescence data, a model with only one dense hidden layer could correctly classify the distinct photosynthetic inhibitions with F1-scores over 0.8. The additional interpretation of the fluorescence signal by a recurrent neural network layer further improved the model’s prediction ability. Chapter 6 presents a biotechnology application of the model developed in Chapter 1. Because fluorescent lamps and LEDs are commonly used in bioreactor lighting, their effects on cyanobacterial growth were compared using experimental measurements and steady-state simulations. Experimental measurements showed no significant differences in growth rate and pigment content between the light sources, and the simulated carbon fixation rates were similar. However, the power consumption of the LED was at least fourfold lower. Extending the simulation analysis, literature examples of LED and fluorescent luminaires revealed that only a subset of LEDs could be expected to improve the yield per power consumption above high-efficiency fluorescent lamps. Finally, the combined results of the thesis are discussed in Chapter 7. The importance of each project and the links between them are highlighted. In summary, this thesis employed ODE and neural network models to investigate cyanobacterial photosynthetic mechanisms, measurement methods, and applications for biotechnological growth. Here, the first spectrum-dependent model of cyanobacterial photosynthesis was developed. This model improves the understanding of how light color affects electron transport and photosynthetic output. Dynamic modeling of electron transport around photosystem I was also used to determine possible reaction mechanisms and important kinetic constants. Furthermore, the built neural networks will enable the rapid and straightforward interpretation of the photosynthetic state in cyanobacterial cultures. The models developed here address key questions in photosynthesis research, supporting the investigation and exploitation of cyanobacteria.

Authors 1

1. Tobias Pfennig corresponding Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen