Trimming gene deletion strategies for growth-coupled production in constraint-based metabolic networks: TrimGdel

[1]

[2]

iMM904, and e_coli_core. GDLS had the minimum elapsed

time, but its success ratio was low. As described in Section 3.2, many GDLS strategies resulted in target metabolite

production without growth or growth-coupled production

when the best PR was evaluated. However, when the worst

PR at GR maximization was evaluated, growth-coupled

production was rarely achieved. As optGene returns the

best solution at that point when the specified computation

time runs out, designating a longer computation time may

increase the success ratio.

TrimGdel, optGene, and GDLS are in descending order

of the average number of deleted genes, which is consistent with the descending order of the success ratio. One

reason for the high success ratio of TrimGdel is the success

in searching for larger gene deletion strategies. The elapsed

time of TrimGdel, which was less than 20 m for every case,

even for genome-scale models, is acceptable for actual use.

The average numbers of deleted genes for TrimGdel were

8.90%, 6.02%, and 9.04% of whole genes, but they were

less than 5% for more than a half of the gene deletion

strategies.

For TrimGdel, the number of gene deletions was between

1% and 5% of whole genes for 66.5%, 50.5%, and 56.8% of

the successful gene deletion strategies for iML1515,

iMM904, and e_coli_core, respectively. On the other hand,

for GDLS, the number of gene deletions was 1% or less of

whole genes for 100%, 100%, 33.3% of the successful gene

deletion strategies for iML1515, iMM904, and e_coli_core,

respectively. These numbers were 100%, 80.1%, and 4% for

optGene, respectively. It was seen that GDLS and optGene

could derive gene deletion strategies whose sizes are 1% or

more in a small model. However, we can conclude that

TrimGdel is much more suitable for deriving gene deletion

strategies whose sizes are between 1% and 5% of whole

genes for genome-scale models.

Biotin Production.

In all cases described in Table 7, the values of biotin synthase and MALCOAMT are identical, which may imply

that malonyl CoA should be produced in abundance in biotin production. The sum of the absolute reaction rates of all

reactions (total flux) is correlated with the number of gene

deletions and the growth rate but not with the biotin production rate. Because more gene deletions lead to more

repressed reactions, it results in a lower total flux and GR.

However, there seem to be other factors that determine

whether growth-coupled production of biotin occurs.

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

W. B. Copeland et al., “Computational tools for metabolic engineering,” Metabolic Eng., vol. 14, no. 3, pp. 270–280, 2012.

A. P. Burgard, P. Pharkya, and C. D. Maranas, “OptKnock:

A bilevel programming framework for identifying gene knockout

strategies for microbial strain optimization,” Biotechnol. Bioeng.,

vol. 84, no. 6, pp. 647–657, 2003.

P. Pharkya, A. P. Burgard, and C. D. Maranas, “OptStrain:

A computational framework for redesign of microbial production

systems,” Genome Res., vol. 14, no. 11, pp. 2367–2376, 2004.

P. Pharkya and C. D. Maranas, “An optimization framework for

identifying reaction activation/inhibition or elimination candidates for overproduction in microbial systems,” Metabolic Eng.,

vol. 8, no. 1, pp. 1–13, 2006.

K. R. Patil, I. Rocha, J. F€

orster, and J. Nielsen, “Evolutionary programming as a platform for in silico metabolic engineering,” BMC

Bioinf., vol. 6, no. 1, 2005, Art. no. 308.

S. Ranganathan, P. F. Suthers, and C. D. Maranas, “OptForce: An

optimization procedure for identifying all genetic manipulations

leading to targeted overproductions,” PLoS One Comput. Biol,

vol. 6, no. 4, 2010, Art. no. e1000744.

I. Rocha et al., “OptFlux: An open-source software platform for in silico

metabolic engineering,” BMC Syst. Biol., vol. 4, no. 1, pp. 1–12, 2010.

Y. Toya and H. Shimizu, “Flux analysis and metabolomics for systematic metabolic engineering of microorganisms,” Biotechnol.

Adv., vol. 31, no. 6, pp. 818–826, 2013.

J. D. Orth, I. Thiele, and B. Palsson, “What is flux balance analysis?,” Nature Biotechnol., vol. 28, no. 3, pp. 245–248, 2010.

D. S. Lun et al., “Large-scale identification of genetic design strategies using local search,” Mol. Syst. Biol., vol. 5, no. 1, 2009,

Art. no. 296.

G. Rockwell, N. J. Guido, and G. M. Church, “Redirector: Designing cell factories by reconstructing the metabolic objective,” PLoS

One Comput. Biol.. vol. 9, no. 1, 2013, Art. no. e1002882.

L. Yang, W. R. Cluett, and R. Mahadevan, “EMILiO: A fast algorithm for genome-scale strain design,” Metabolic Eng., vol. 13,

no. 3, pp. 272–281, 2011.

D. Egen and D. S. Lun, “Truncated branch and bound achieves

efficient constraint-based genetic design,” Bioinformatics, vol. 28,

no. 12, pp. 1619–1623, 2012.

N. E. Lewis et al., “Omic data from evolved E coli are consistent

with computed optimal growth from genome-scale models,” Mol.

Syst. Biol., vol. 6, no. 1, 2010, Art. no. 390.

D. Gu, C. Zhang, S. Zhou, L. Wei, and Q. Hua, “IdealKnock:

A framework for efficiently identifying knockout strategies leading to targeted overproduction,” Comput. Biol. Chem., vol. 61,

pp. 229–237, 2016.

S. Ohno, H. Shimizu, and C. Furusawa, “FastPros: Screening of

reaction knockout strategies for metabolic engineering,” Bioinformatics, vol. 30, no. 7, pp. 981–987, 2014.

T. Tamura, “Grid-based computational methods for the design of

constraint-based parsimonious chemical reaction networks to simulate metabolite production: GridProd,” BMC Bioinf., vol. 19,

no. 1, 2018, Art. no. 325.

A. von Kamp and S. Klamt, “Enumeration of smallest intervention

strategies in genome-scale metabolic networks,” PLoS One Comput. Biol., vol. 10, no. 1, 2014, Art. no. e1003378.

O. H€adicke and S. Klamt, “Computing complex metabolic intervention strategies using constrained minimal cut sets,” Metabolic

Eng., vol. 13, no. 2, pp. 204–213, 2011.

B. J. Harder, K. Bettenbrock, and S. Klamt, “Model-based metabolic engineering enables high yield itaconic acid production by

escherichia coli,” Metabolic Eng., vol. 38, pp. 29–37, 2016.

TAMURA: TRIMMING GENE DELETION STRATEGIES FOR GROWTH-COUPLED PRODUCTION IN CONSTRAINT-BASED METABOLIC

[21] P. Schneider, A. von Kamp, and S. Klamt, “An extended and generalized framework for the calculation of metabolic intervention

strategies based on minimal cut sets,” PLoS One Comput. Biol.,

vol. 16, no. 7, 2020, Art. no. e1008110.

[22] D. Banerjee et al., “Genome-scale metabolic rewiring improves

titers rates and yields of the non-native product indigoidine at

scale,” Nature Commun., vol. 1, no. 1, pp. 1–11. 2020.

[23] D. Machado and M. J. Herrgard, and I. Rocha, “Stoichiometric

representation of gene–protein–reaction associations leverages

constraint-based analysis from reaction to gene-level phenotype

prediction,” PLoS One Comput. Biol., vol. 12, no. 10, 2016,

Art. no. e1005140.

[24] Z. Razaghi-Moghadam and Z. Nikoloski, “GeneReg: A constraintbased approach for design of feasible metabolic engineering strategies

at the gene level,” Bioinformatics, vol. 37, no. 12, pp. 1717–1723, 2021.

[25] T. Tamura, A. Muto-Fujita, Y. Tohsato, and T. Kosaka, “Gene deletion

algorithms for minimum reaction network design by mixed-integer

linear programming for metabolite production in constraint-based

models: gDel_minRN,” 2021, doi: 10.21203/rs.3.rs-611371/v1.

[26] I. Rocha, P. Maia, M. Rocha, and F. C. Ferreira, “OptGene:

A framework for in silico metabolic engineering,” in Proc. 10th

Int. Chem. Biol. Eng. Conf., 2008, pp. 218–219.

[27] J. M. Monk et al., “iML1515, a knowledgebase that computes Escherichia coli traits,” Nature Biotechnol., vol. 35, no. 10,

pp. 904–908, 2017.

[28] M. L. Mo, B. Palsson, and M. J. Herrgard, “Connecting extracellular metabolomic measurements to intracellular flux states in

yeast,” BMC Syst. Biol., vol. 3, no. 1, 2009, Art. no. 37.

[29] J. D. Orth, R. M. Fleming, and B. O. Palsson, “Reconstruction and

use of microbial metabolic networks: The core Escherichia coli

metabolic model as an educational guide,” EcoSal Plus, vol. 4,

pp. 1–48, 2010.

1549

[30] L. Heirendt et al., “Creation and analysis of biochemical constraint-based models using the COBRA Toolbox v. 3.0,” Nature

Protoc., vol. 14, no. 3, pp. 639–702, 2019.

[31] S. Liu, P. Kang, Z. Cui, Z. Wang, and T. Chen, “Increased riboflavin production by knockout of 6-phosphofructokinase I and

blocking the entner–doudoroff pathway in escherichia coli,”

Biotechnol. Lett., vol. 38, no. 8, pp. 1307–1314, 2016.

[32] A. Schrijver, Theory of Linear and Integer Programming. Hoboken,

NJ, USA: Wiley, 1998.

[33] E. Brunk et al., “Recon3D enables a three-dimensional view of

gene variation in human metabolism,” Nature Biotechnol., vol. 36,

no. 3, pp. 272–281, 2018.

Takeyuki Tamura (Member, IEEE) received the

BE, ME, and the PhD degrees in informatics from

Kyoto University, Japan, in 2001, 2003, and 2006,

respectively. He joined Bioinformatics Center,

Institute for Chemical Research, Kyoto University

as a postdoctoral fellow, in 2006. He worked as

an assistant professor from Dec. 2007 to Sep.

2017, and works as an associate professor from

Oct. 2017. His research interests include bioinformatics and the theory of combinatorial optimization for graphs and networks.

" For more information on this or any other computing topic,

please visit our Digital Library at www.computer.org/csdl.

...