Integrated Regulatory and Metabolic Networks of the Marine Diatom Phaeodactylum tricornutum Predict the Response to Rising CO2 Levels

Abstract

Diatoms are eukaryotic microalgae that are responsible for up to 40% of the ocean's primary productivity. How diatoms respond to environmental perturbations such as elevated carbon concentrations in the atmosphere is currently poorly understood. We developed a transcriptional regulatory network based on various transcriptome sequencing expression libraries for different environmental responses to gain insight into the marine diatom's metabolic and regulatory interactions and provide a comprehensive framework of responses to increasing atmospheric carbon levels. This transcriptional regulatory network was integrated with a recently published genome-scale metabolic model of Phaeodactylum tricornutum to explore the connectivity of the regulatory network and shared metabolites. The integrated regulatory and metabolic model revealed highly connected modules within carbon and nitrogen metabolism. P. tricornutum's response to rising carbon levels was analyzed by using the recent genome-scale metabolic model with cross comparison to experimental manipulations of carbon dioxide. IMPORTANCE Using a systems biology approach, we studied the response of the marine diatom Phaeodactylum tricornutum to changing atmospheric carbon concentrations on an ocean-wide scale. By integrating an available genome-scale metabolic model and a newly developed transcriptional regulatory network inferred from transcriptome sequencing expression data, we demonstrate that carbon metabolism and nitrogen metabolism are strongly connected and the genes involved are coregulated in this model diatom. These tight regulatory constraints could play a major role during the adaptation of P. tricornutum to increasing carbon levels. The transcriptional regulatory network developed can be further used to study the effects of different environmental perturbations on P. tricornutum's metabolism.

Keywords: Phaeodactylum tricornutum; coregulated genes; genome-scale metabolic network reconstruction; integrated network modeling; regulatory network inference.

FIG 1

Visualization of the metabolic network connecting transcriptional clusters in P. tricornutum. Nodes are transcriptional clusters, while edges show strong connectivity in terms of products in the genome-scale metabolic model (6).

FIG 2

Percentages of reactions showing different fluxes at doubled HCO₃ levels per group as identified by the solution space sampling of the genome-scale metabolic model of P. tricornutum. Pigment metabolism contains most of the reactions with different fluxes under the two conditions, i.e., 43% of the reactions in this group were upregulated at high HCO₃ concentrations, followed by nucleotide metabolism (19% in total; 17% upregulated and 2% downregulated) and amino acid metabolism (10%). Three reactions, namely, ITCY_c, CMP_c, and ITPA_c, belonging to nucleotide metabolism were downregulated at high HCO₃ concentrations.

FIG 3

Percentages of reactions differentially expressed at high versus low CO₂ concentrations per group. Energy metabolism contains most of the differentially expressed reactions; i.e., 40.9% of the reactions in this group were upregulated and 6.8% were downregulated at high CO₂ concentrations, followed by lipid and cofactor metabolism.

FIG 4

Comparison of reactions with different flux patterns. Reactions with different flux patterns (either up- or downregulated) at low and high carbon concentrations identified in the genome-scale metabolic model solution space sampling (blue) and the differential expression (DE, yellow) analysis are compared for each subsystem.

FIG 5

Model predictions at various HCO₃ conditions. With increasing HCO₃ concentrations, NO₃ is limiting and is completely required for biomass production; subsequently, NO₃ storage declines. Excess carbon is released as DMSP, which is the only reaction in the model allowing carbon secretion (A). Panel B shows biomass production and NO₃ storage when NO₃ uptake is increased linearly shortly before NO₃ becomes limiting (at an HCO₃ uptake level of 5 mM). Simulation results when carbon can be stored as chrysolaminarin or TAGs are depicted in panel C. Here, NO₃ uptake is constrained and constant as in panel A. The model predicts that excess carbon is stored in the form of TAGs. (D) Effect of the carbon/nitrogen (C/N) ratio on biomass. Within this simulation, carbon uptake was varied from 0 to 10 mM and nitrogen uptake was kept constant at 0.535 mM. Biomass production increases at C/N ratios under 10.28 and stagnates at ratios over 10.28 because of limiting nitrogen availability. Note that for all of the simulations shown, the available CO₂ was forced to be taken up.