To synthesize DHA-CoA, it is necessary to refer to the synthetic pathway in living organisms for DHA-PC, phosphatidylcholine, which is the most abundant lipid in eukaryotic cell membranes. In cells, membrane lipids are formed by combining the de novo synthesis pathway with the remodeling pathway [1].

The de novo synthesis pathway, also known as the Kennedy pathway, is responsible for converting glycerol-3-phosphate to phosphatidic acid. This process is carried out by two enzymes, glycerol-3-phosphate acyltransferase and lysophosphatidic acid acyltransferase. From there, diacylglycerol is generated through a series of dephosphorylation and intermediate conversions, ultimately leading to the synthesis of phosphatidylcholine (PC) [2].

The de novo phospholipid molecules produced by the Kennedy pathway mainly carry short to medium-saturated or monounsaturated acyl chains, which are far from the polyunsaturated long-chain phospholipids that are abundantly present in cell membranes. To achieve the conversion of de novo phospholipids to mature membrane phospholipids, cells edit de novo phospholipids using a phospholipid remodeling pathway known as the Lands cycle, which introduces polyunsaturated long acyl chains into phospholipids by erasing (de-acylation) and re-writing (re-acylation) the acyl groups of phospholipids [3]. In the Lands cycle, PLA enzymes mediate the deacylation of phospholipids to form lysophosphatidylcholine lysophosphatidylcholine (LPC), which is followed by the catalytic re-formation of PCs from LPC with polyunsaturated long-chain acyl-coenzyme A by LPCAT [4][5].

After obtaining the above information, we came up with the idea that if we synthesize DHA-CoA in cells and increase its percentage in the acyl-coenzyme A pool, we will be able to synthesize DHA-PC efficiently in cells by taking advantage of LPCAT's preference for doping polyunsaturated long acyl chains into PC.

But how is acyl-coenzyme A formed? After reviewing the literature, we found that acyl-coenzyme A synthetase acyl-CoA synthetase (ACS) mediates this reaction. Acyl-CoA binds free fatty acids to ATP to form a lipoyl adenosine monophosphate (Acyl-AMP) intermediate, and then coenzyme A (CoA) substitutes for AMP to produce acyl-coenzyme A (Acyl-CoA) [6][7][8][9][10][11][12]. In summary, fatty acids and coenzyme A are synthesized into acyl-coenzyme A.

Schizochytrium sp., also known as Schizosaccharomyces cerevisiae, is a unicellular marine fungus belonging to the family of broken-bladder potybacteria, which is rich in docosahexaenoic acid (DHA).Studies on the lipid composition of Schizosaccharomyces pombe showed that its intracellular DHA content accounted for 50% of the total fatty acids, and it has a typical membrane lipid metabolic pathway that can synthesize DHA substrates intracellularly on its own without exogenous supplementation, making it an excellent chassis for us to modify the LACS pathway [13][14].

We focused on optimizing LACS, a key enzyme in the DHA-PC synthesis pathway, to improve the selectivity and catalytic activity of LACS for DHA to enhance the production of DHA-CoA in Schizosaccharomyces pombe. However, we were unable to culture Schizosaccharomyces pombe under the existing laboratory conditions and did not have the guidance of technicians with experience in algal culture. Since the time available for experimental validation in the competition cycle was relatively short, we chose brewer's yeast and Escherichia coli, which are easier to perform experimental manipulations, as protein expression strains to validate our results. Saccharomyces cerevisiae is a model organism widely used in experimental research. It is a eukaryotic organism with similar cellular structure to Schizosaccharomyces pombe, so it is easy to simulate the Schizosaccharomyces pombe pathway for validation. E. coli has the advantage of simple operation, and easy to express and purify proteins, which is extremely important in the limited time of experiments.

The modeling group will optimize the natural LACS protein from a rational design perspective. At the same time, we will also optimize LACS by sequential evolution using the OrthoRep system.

The evolutionary tree (for details of it, see Cycle 3 of the cycle page) made by the modeling group was used to do a BLAST search for multiple LACS sequences of the LACS of Schizosaccharomyces pombe with close relatives, followed by a literature search to verify the functionality of the target enzymes and the experimental data, from which we selected three Long-chain acyl-CoA synthetases, namely, LACS1 (from Arabidopsis thaliana), CzLACS5 (from Oleaginous Alga Chromochloris zofingiensis), LACS6 (from Schizochytrium sp. A-2) [15][7][13].

For the expression of these three enzymes in Saccharomyces cerevisiae, we chose the Saccharomyces cerevisiae expression vector pYES2/CT plasmid pYES2/CT plasmid is a high-copy plasmid that carries a strongly inducible GAL1 promoter, which is repressed in the presence of glucose and is induced only upon the addition of galactose to the medium, resulting in controlled expression of the target gene. Its carrying of orotidine-5-phosphate decarboxylase, a lactonucleoside-5-phosphate decarboxylase required for uracil biosynthesis, allows yeast with the knockout of the endogenous Ura synthesis gene to be grown on uracil-free selective medium. This provides a convenient selection marker for gene transformation and screening. Moreover, the pYES2 plasmid contains six histidines at the C-terminus of the cloning site to allow our protein to contain a C-terminal His-tag, thus facilitating purification of the expressed recombinant protein using nickel affinity chromatography. Subsequently, we inserted three target genes into the multiple cloning site on pYES2/CT to construct pYES2-LACS1, pYES2-czLACS5, and pYES2-LACS6.

To determine the best starting point for LACS optimization, we introduced pYES2-LACS1, pYES2-czLACS5, pYES2-LACS6, into the yeast strain Invsc1 (MATa his3D1 leu2 trp1-289 ura3-52 MAT his3D1 leu2 trp1- 289 ura3-52) with the endogenous LACS-deficient yeast strain YB525 (ura3-52; leu2-3, 112; his3-200; ade2-101; lys2-801; faa1. : HIS3; faa4::LYS2). After verifying the transformation, we first performed protein expression verification in Invsc1 to confirm induction conditions. We then measured the growth curve of strain YB525 containing pYES2-LACS in a medium supplemented with DHA, and galactose with light blue bacteriocin (to inhibit endogenous fatty acid synthesis). A comparison of growth rates was used to identify the LACS with the highest initial selectivity for DHA.

The modeling group obtained the optimized sequence MLACS1 with single amino acid mutation through molecular docking, see part 1 of the model page), and the experimental group determined the mutated bases according to the codon table, and mutated pYES2-LACS to pYES2-MLACS through fixed-point mutation PCR to obtain the vector containing the mutated sequence. For the E. coli protein expression vector, we chose the pET-21a plasmid, which was induced using efficient and sensitive IPTG. The pET-LACS1 and pET-MLACS1 were constructed by enzymatic ligation of MLACS. pET21a-MLACS was subsequently transformed into E. coli BL21 (DE3) versus C43 (DE3), which is more suited for membrane protein expression, for protein expression, isolation, and purification, respectively, and finally, the enzyme activities before and after optimization were measured and compared.

To better increase the likelihood of successful optimization of target proteins, we attempted to obtain LACS with increased activity towards DHA utilizing continuous directed evolution. The OrthoRep system is a cellular hypermutation system for in vivo sequential evolution, which is mainly used in yeast cells [16][17][18][19][20]. The core principle is to use a special low-fidelity DNA polymerase (TP-DNAP1) to replicate the linear plasmid pGKL1(p1), which contains the Tail protein so that the mutation rate of the target genes (GOIs) encoded on the plasmid reaches 10^-5 substitutions per base (s.p.b.) without increasing the mutation rate of the genome (<10-10 s.p.b.). This DNAP/linear plasmid combination is orthogonal to genome replication, i.e., TP-DNAP does not replicate the genome, and the host DNAP does not replicate the linear plasmid, resulting in mutation specific to the GOI.

In the OrthoRep system, GOIs need to be integrated into the p1 plasmid, and the p2 plasmid encodes an RNAP that recognizes a specific promoter to drive the expression of GOIs on p1. In this way, the OrthoRep system can rapidly evolve target genes without affecting the stability of the host genome.

To use the OrthoRep system, we first inserted wild-type LACS into the pCCL plasmid to construct the pCCL-LACS plasmid. pCCL plasmid contains homologous sequences at both ends of the p1 plasmid, which allows for the insertion of the target gene LACS into the p1 plasmid by homologous recombination. After confirming recombination, we cotransformed the p1 plasmid with pAR-Ec633 expressing low-fidelity DNA polymerase into Saccharomyces cerevisiae BY4741 to obtain a chassis for evolution.

It was a test to create a screening pressure for the targeted evolution of LACS proteins in Saccharomyces cerevisiae. Saccharomyces cerevisiae is constantly synthesizing the long-chain fatty acids it needs, which are typically sixteen- and eighteen-carbon fatty acids, rather than the DHA we were targeting, and we were concerned that yeast endogenous LACS would play a role in the evolutionary process and detract from the screening pressures we were trying to create for our target genes.

Therefore, we envisioned a way to create screening pressure using CRISPR technology. The FAS1 gene controls fatty acid synthesis in Saccharomyces cerevisiae. If the FAS1 gene is knocked out using CRISPR technology, it would render the new strain incapable of synthesizing fatty acids on its own. In this way, we can gain screening pressure in continuous evolution by exogenously adding different ratios of DHA to sixteen-/eighteen-carbon fatty acids in the culture medium. At the same time, we hope that we can use CRISPR to knock down endogenous LACS (FAA1; FAA4) in Saccharomyces cerevisiae and thus focus the screening pressure on our exogenous LACS to improve the efficiency of directed evolution.

We used the pCAS plasmid that can work in yeast cells [24]. The pCAS plasmid can express both Cas9 nuclease and sgRNA, allowing Cas9 to induce homologous recombination repair (HRR) by performing double-stranded DNA cleavage at specific locations in the genome based on the guide sequence of the sgRNA. By co-transformation into yeast cells with linearised repair DNA, the pCAS plasmid can be used to enable label-free, scar-free genome editing. We designed gRNAs targeting FAS1, FAA4, and FAA1 through the benching site.

FAA1 and FAA4 are primarily responsible for fatty acid activation, uptake, and maintenance of endogenous lipoyl coenzyme A pool in yeast cells. When both FAA1 and FAA4 are deleted, cellular uptake and activation of exogenous fatty acids are significantly reduced and fatty acid metabolism is severely affected [23]. To prevent the fatty acid metabolism ability of knockout Saccharomyces cerevisiae from being irreversibly affected, we designed three pCAS plasmids targeting FAS1, FAS1 and FAA4, and FAS1, FAA4, and FAA1, respectively. To improve efficiency and increase the likelihood of success in the experiments, we also designed a plasmid that uses light cyanobacteria in (which inhibits fatty acid synthesis) to treat YB525 ( FAA1, FAA4 double deletion in Saccharomyces cerevisiae) to achieve the same effect as CRISPR knockdown.

---

[1] Christie WW, (2021) in The LipidWeb, Christie WW, Ed. (https://www.lipidmaps.org/resources/lipidweb/lipidweb_html/lipids/complex/pc/index.htm)

[2] W. J. Valentine, T. Shimizu and H. Shindou, Biochimie, 2023, 215, 24–33.

[3] W. E. Lands, J Biol Chem, 1958, 231, 883–888.

[4] R. Gopalam, A. Datey, S. Bijoor, D. Chakravortty and A. W. Tumaney, Mol Biotechnol, 2021, 63, 963–972.

[5] A. Po?ońska, K. Jasieniecka-Gazarkiewicz, L. You, X. Hao, S. Klińska, Y. Gong and A. Bana?, IJMS, 2021, 22, 9056.

[6] S. Yan, X.-F. Yang, H.-L. Liu, N. Fu, Y. Ouyang and K. Qing, World J Gastroenterol, 2015, 21, 3492–3498.

[7] T. Wu, Y. Fu, Y. Shi, Y. Li, Y. Kou, X. Mao and J. Liu, J. Agric. Food Chem., 2020, 68, 4473–4484.

[8] S. Kuroha, Y. Katada, Y. Isobe, H. Uchino, K. Shishikura, T. Nirasawa, K. Tsubota, K. Negishi, T. Kurihara and M. Arita, The FASEB Journal, 2023, 37, e23151.

[9] H. Kuwata, E. Nakatani, S. Shimbara-Matsubayashi, F. Ishikawa, M. Shibanuma, Y. Sasaki, E. Yoda, Y. Nakatani and S. Hara, Biochimica et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids, 2019, 1864, 1606–1618.

[10] J. R. Marszalek, C. Kitidis, C. C. DiRusso and H. F. Lodish, Journal of Biological Chemistry, 2005, 280, 10817–10826.

[11] F. Bai, L. Yu, J. Shi, Y. Li‐Beisson and J. Liu, New Phytologist, 2022, 233, 823–837.

[12] H. Dykstra, C. Fisk, C. LaRose, A. Waldhart, X. Meng, G. Zhao and N. Wu, Archives of Biochemistry and Biophysics, 2021, 700, 108773.

[13] X.-H. Yue, W.-C. Chen, Z.-M. Wang, P.-Y. Liu, X.-Y. Li, C.-B. Lin, S.-H. Lu, F.-H. Huang and X. Wan, J. Agric. Food Chem., 2019, 67, 9683–9693.

[14] SONG Ze, PENG Yong-bo, SONG Yue-fan, WANG Qiu-kuan, HE Yun-hai, LIU Shu. 2019. Analysis of nutrient composition and polysaccharide characteristics of fungus Schizochytrium sp Journal of Dalian Ocean University, 34(2): 247-251. doi: 10.16535/j.cnki.dlhyxb.2019.02.014

[15] X. Lin, S. Kaul, S. Rounsley, T. P. Shea, M.-I. Benito, C. D. Town, C. Y. Fujii, T. Mason, C. L. Bowman, M. Barnstead, T. V. Feldblyum, C. R. Buell, K. A. Ketchum, J. Lee, C. M. Ronning, H. L. Koo, K. S. Moffat, L. A. Cronin, M. Shen, G. Pai, S. Van Aken, L. Umayam, L. J. Tallon, J. E. Gill, M. D. Adams, A. J. Carrera, T. H. Creasy, H. M. Goodman, C. R. Somerville, G. P. Copenhaver, D. Preuss, W. C. Nierman, O. White, J. A. Eisen, S. L. Salzberg, C. M. Fraser and J. C. Venter, Nature, 1999, 402, 761–768.

[16] R. S. Molina, G. Rix, A. A. Mengiste, B. Alvarez, D. Seo, H. Chen, J. Hurtado, Q. Zhang, J. D. García-García, Z. J. Heins, P. J. Almhjell, F. H. Arnold, A. S. Khalil, A. D. Hanson, J. E. Dueber, D. V. Schaffer, F. Chen, S. Kim, L. á. Fernández, M. D. Shoulders and C. C. Liu, .

[17] K. V. Gelder, E. R. Oliveira-Filho, J. D. García-García, Y. Hu, S. D. Bruner and A. D. Hanson, ACS Synth. Biol., 2023, 12, 963–970.

[18] J. D. García-García, K. Van Gelder, J. Joshi, U. Bathe, B. J. Leong, S. D. Bruner, C. C. Liu and A. D. Hanson, Plant Physiology, 2022, 188, 971–983.

[19] G. Rix, E. J. Watkins-Dulaney, P. J. Almhjell, C. E. Boville, F. H. Arnold and C. C. Liu, Nat Commun, 2020, 11, 5644.

[20] A. Ravikumar, A. Arrieta and C. C. Liu, Nat Chem Biol, 2014, 10, 175–177.

[21] File:Mechanismabcd.png - Wikimedia Commons, .

[22] X.-H. Yue, W.-C. Chen, Z.-M. Wang, P.-Y. Liu, X.-Y. Li, C.-B. Lin, S.-H. Lu, F.-H. Huang and X. Wan, Journal of Agricultural and Food Chemistry, 2019, 67, 9683–9693.

[23] N. J. Fergeman, P. N. Black, X. D. Zhao, J. Knudsen and C. C. DiRusso, Journal of Biological Chemistry, 2001, 276, 37051–37059.

[24] O. W. Ryan, S. Poddar and J. H. D. Cate, Cold Spring Harbor Protocols, 2016, 2016, pdb.prot086827.