Team:Duesseldorf/Model

Optimization of Synechocystis sp. PCC6803 fatty acid metabolism

Considering the aim of the SynMilk-Project - to create a synthetic alternative to native cow’s milk - we considered many directions of interest. To conduct research in all relevant scientific fields, we designed a wide set of experimental approaches. These approaches began by a series of proof-of-concept experiments, leading to a suitable data set and allowing to estimate the most favorable directions for further investigations. Within this frame, molecular-, synthetic and microbiological as well as biochemical experimental set ups were applied to obtain an overview about the feasibility of the SynMilk-Project.

A first step to SynMilk was the biological modeling of a suitable and component productive system for the production of fatty acids. To this system, all contributing parts, including the ACC-Proteins and the thioesterases, which contribute to the biosynthesis of target fatty acids were considered.

To achieve our goal of synthesizing several different lengths of fatty acids, we modified the fatty acid metabolism of Synechocystis sp. PCC6803, a model organism for photosynthetic bacteria. As a model organism, its metabolism has been constructed as a network which is applicable for use with the COBRA toolbox1. The network used as a basis for our Flux-Balance Analysis is available here2.

The fatty acid metabolism in this model is a process that linearly leads to palmitoyl-ACP, at this point the biosynthesis may branch off into unsaturated or longer saturated fatty acids. Contrary to the actual pathway, the model treats every step in the fatty acid biosynthesis as being catalyzed by a separate enzyme. In the organism, it is a circular process which is catalyzed by the same enzymes for all lengths of fatty acids.
In the organism, thioesterases determine the length at which this process is terminated and incorporated into lipids, which make up the cell membrane. This termination can take place at C14 or longer fatty acids, a fact which the model does not take into account. We added thioesterases into the model, thus, including them leads to the secretion of free fatty acids. The thioesterases terminate the fatty acid biosynthesis at lengths shorter than C16:0 and are not usually present in Synechocystis, but are added to the model nonetheless, since they do not carry any flux when not explicitly optimized.
Optimizing the model for bacterial growth did not yield free fatty acids, as the fatty acids which are incorporated into the cell membrane are part of the biomass reaction.

Fig. 1: Fatty acid metabolism of the cyanobacterium Synechocystis sp. PCC 6803. Overexpressed or newly expressed steps are pictured as green arrows, down-regulated steps as red arrows.
In metabolic engineering, deleting reactions leading away from a product is a common method which can lead to the accumulation or secretion of a targeted product of interest. In this case however, fatty acids are the source of bioactive lipids in the cell membrane, which is an essential part for the viability and function of the cell. The simulation of deletions with the model suggests that any deletion in the fatty acid synthesis pathway leads to a reduction of growth to 0, indicating that they would be lethal for the organism.

Since a deletion in the pathway is not an option, an inducible knock-down using CRISPR-dCas9 is viable to induce fatty acid production after growth to a certain density is achieved. This knock-down is expected to slow down growth after induction.

To identify gene targets for knock-downs, assumptions have been made about the constraints under which fatty acid synthesis is simulated:
  1. The maximum photon assimilation of the organism is equivalent to the lightintensity being used in the incubator at 288 mmol photons*h-1*m-2.
  2. Half the value of the optimized growth rate is being used as a minimum to approximate the slowed growth and the cell’s overall maintenance mechanisms, which still have to be active during induction.
  3. All metabolites in the model can be exported and removed from the model as a way to simulate the theoretical maximum yield of fatty acids, circumventing metabolite accumulations slowing down the synthesis.
Under these assumptions, production envelopes were calculated for several different lengths of fatty acids and different reactions in the fatty acid synthesis. The absolute numbers taken from the simulations should not be taken as quantitative predictions, but rather as roughly estimated guidelines to identify high-impact targets for metabolic engineering.
We started by looking at reactions that are at the front of the pathway and would in this way impact all lengths of fatty acid and improve overall yield.

Acetyl-CoA is a central metabolite in a lot of biochemical pathways, making fatty acid synthesis compete with other reactions and limiting fatty acid yields. Several reactions leading away from Acetyl-CoA were looked at and we identified the reaction Acetyl-CoA C-Acetyltransferase (ACACT1r) as a reaction to target for a knock-down. This gene has already successfully been knocked out in the literature, increasing fatty acid yield3.
Fig. 2: Optimized yields of fatty acids plotted against all possible Acetyl-CoA C-Acetyltransferase activities.
It was successfully confirmed through the applied model, that lower flux through this reaction increases fatty acid yield for all chain lengths. A knock-out of this gene would be possible, but we instead decided to perform a knock-down to not stress the organism too much.

The Acetyl-CoA Carboxylase (ACCOAC) enzyme is often overexpressed in literature to increase fatty acid yield4, 5, 6, 7, 8, creating more malonyl-CoA as a substrate for fatty acid synthesis.
Fig. 3: Optimized yields of fatty acids plotted against all possible Acetyl-CoA carboxylase activities. The activity of this enzyme in the growth-optimized model is 0.23.
The model further confirmed that the applied strategy is a viable method for improving the yield of fatty acids. More flux through the ACCAOC reaction is most beneficial for the production of longer fatty acids, since their optimal yields are reached by even higher fluxes through ACCOAC than those of the shorter fatty acids.

Looking further downstream along the fatty acid synthesis pathway, the 3-oxoacyl-ACP synthase (3OAS) is an attractive target for production of butyrate, as it facilitates the elongation from C4 onwards, but is not involved in the synthesis of butyrate itself.
The synthase enzyme is a separate enzyme for each chain-length in this model, so all synthase enzymes were down-regulated at the same rate, by decreasing the upper bound from 100% to 0% of the maximum flux. The percent of retained synthase activity was plotted against the optimized fatty acid yield.
Fig. 4: Optimized yields of fatty acids plotted against down-regulated 3-oxoacyl synthase activities.
The yields for all lengths of fatty acids decreased with further down-regulation, except for butyrate. Butyrate yield is unaffected by the down-regulations, but does not increase either, since it is already optimized and at its theoretical maximum.

A more realistic look at the effect of the down-regulation on butyrate yield can be achieved by setting the flux of the enzymes leading to fatty acids to a value that is achieved during growth without down-regulations. The model is then optimized for growth with the synthase down-regulations. This process essentially sets the upstream reactions to a normal level without making any assumption towards the efficiency or biomass-yield of the downstream reactions.
In the following simulations, no minimum activities have been assumed for the thioesterase enzyme and the model was optimized for growth instead of the fatty acid yield, so butyrate yields might be underestimated.
The flux through the Acetyl-CoA carboxylase was set to its optimal value for growth and the butyrate yield was plotted against the 3-oxoacyl-ACP synthase activity. Then the Acetyl-CoA carboxylase flux was set to its optimal activity for butyrate production and the two yield ranges were compared. The Acetyl-CoA C-acetyltransferase activity was set to 0, assuming a successful knock-down.
Fig. 5: Butyrate yields of a growth-optimized model assuming a successful knock-down of Acetyl-CoA C-Acetyltransferase. The butyrate yields of growth-optimized ACCOAC activity (blue) and butyrate-optimized ACCOAC activity (red) are plotted against the down-regulated 3OAS activity.
With the synthase enzyme down-regulated, butyrate is inevitably accumulating as a side-product and would be exported, even in growth-optimized conditions. Through the overexpression of the Acetyl-CoA carboxylase, higher enzyme activity can be achieved, increasing the flux through this reaction and increasing maximum as well as significantly increasing minimum butyrate yields.

The simulations show that even when assuming full activity of the synthase enzyme and no minimum activity of the thioesterase, overexpressing the acetyl-CoA-carboxylase enzyme will lead to higher fatty acid yields.
While the values of yields and fluxes are likely incorrect, the relative increases suggest a significant impact of our proposed design on the fatty acid metabolism, increasing yields, which will be quantified experimentally.



References
  1. Ali Ebrahim, Joshua A Lerman, Bernhard O Palsson & Daniel R Hyduke "COBRApy: constraints-based reconstruction and analysis for python." BMC systems biology 7.1 (2013): 74.
  2. Juan Nogales, Steinn Gudmundsson, Eric M. Knight, Bernhard O. Palsson, and Ines Thiele "Detailing the optimality of photosynthesis in cyanobacteria through systems biology analysis." Proceedings of the National Academy of Sciences 109.7 (2012): 2678-2683.
  3. Liu, Xinyao, Jie Sheng, and Roy Curtiss III. "Fatty acid production in genetically modified cyanobacteria." Proceedings of the National Academy of Sciences 108.17 (2011): 6899-6904.
  4. Wenjuan Zhaa, Sheryl B. Rubin-Pitela, Zengyi Shaoa, Huimin Zhao "Improving cellular malonyl-CoA level in Escherichia coli via metabolic engineering." Metabolic engineering 11.3 (2009): 192-198.
  5. Peng Xu, Qin Gu, Wenya Wang, Lynn Wong, Adam G.W. Bower, Cynthia H. Collins & Mattheos A.G. Koffas "Modular optimization of multi-gene pathways for fatty acids production in E. coli." Nature communications 4 (2013): 1409.
  6. Davis, Mark S., José Solbiati, and John E. Cronan. "Overproduction of acetyl-CoA carboxylase activity increases the rate of fatty acid biosynthesis in Escherichia coli." Journal of Biological Chemistry 275.37 (2000): 28593-28598.
  7. Liu, Xinyao, Jie Sheng, and Roy Curtiss III. "Fatty acid production in genetically modified cyanobacteria." Proceedings of the National Academy of Sciences 108.17 (2011): 6899-6904.
  8. Xu, Peng & Ranganathan, Sridhar & Fowler, Zachary & Maranas, Costas & Koffas, Mattheos. (2011). Genome-scale metabolic network modeling results in minimal interventions that cooperatively force carbon flux towards malonyl-CoA. Metabolic engineering. 13. 578-87. 10.1016/j.ymben.2011.06.008.