Part 1: Site-directed mutagenesis of TFD
Cycle 1: Modification of the rare-earth binding site of the TFD protein by adding lanthanide ligand residues to enhance its binding of light rare earths.
Design
The lanthanide-binding protein TFD-EE can rely on reactive oxygen atoms to form ligand bonds with rare-earth ions, thereby trapping the rare-earth ions in the TIM barrel. However, we found that the current TFD-EE is slightly weaker in binding light rare earths with smaller atomic numbers than heavy rare earths with larger atomic numbers, and the binding constants of Ce3+ (atomic number 58) are weaker than those of Eu3+ (atomic number 63) and Yb3+ (atomic number 70) by a 10-fold order of magnitude difference. We therefore wished to improve this by protein engineering of TFD.
Through an in-depth study of the four metal-binding sites EF hands (Figure. 1-1) on the natural lanthanide-binding protein lanmodulin, we found that the oxygen atoms on the glutamic acid, aspartic acid, and asparagine residues provide a suitable coordination environment for lanthanides, which are large trivalent pro-oxidative cations, and that slightly more metal-ligand bonds are required to bind light rare earth (9-10) than heavy rare earth (6-8) [1]. Therefore, in Cycle 1, we attempted to enhance the binding ability of the monomeric protein, TFD-S, for light rare earths by increasing the metal coordination residues within the cavity appropriately by continuing to introduce fixed-point mutations of glutamic acid, aspartic acid, and asparagine near the symmetry axis of the TFD.
Figure 1-1. The structure of lanmodulin from Hansschlegelia quercus (Hans-LanM) [1].A. The four metal binding sites, EF hands 1-4, on Hans-LanM. B. Zoomed-in views of EF hand 1 in La(III)-Hans-LanM (left) and Dy(III)-Hans-LanM (right). C. Zoomed-in views of EF hand 4 in Na(I)-Hans-LanM (left) and Dy(III)-Hans-LanM (right). For B and C, the La(III) ion is shown as blue sphere, the Dy(III) ions are shown as green spheres, the Na(I) ion is shown as purple sphere, solvent molecules are shown as red spheres, and coordination bonds are shown as dashed lines. The numbers near the coordination bonds represent the distances between the atoms, with the unit of Å.
Build
After we mutated the amino acid sequences on PyMol, we submitted the results to AlphaFold3 for prediction, and finally obtained four more probable results (Table 1). What they all have in common is that the mutated amino acid side chains are oriented towards the inner cavity of the TFD, which we believe helps to form rare earth coordination bonds.
Table 1. Mutation condition
Table Note: N56D indicates that the asparagine at position 56 is mutated to aspartate, and so on.
Test
Molecular dynamics simulations of these five systems were performed to predict their ability to adsorb rare earth ions, with terbium ion as the representative ion, for specific reasons referred to Measurement.
The crystal structure of TFD-S is predicted by AlplaFold3, which is a TFD monomeric protein with terbium ion. The remaining mutants were all derived from TFD-S mutations. MD simulation was performed using CHARMM36 protein Force Field 45 in the GROMACS2022.3 package 46. The force constant of the harmonic bias potential is 1000kJ∙mol-1∙nm-2. Set the 10Å cuboid box where the edge is widest from the protein, use the TIP3P water model in it, and add sodium ions to the solution to maintain the neutrality of the system.
The above five systems were simulated at 300K temperature. To begin with, the system is minimized by 50,000 steps using the steepest descent algorithm. In the pre-balancing phase, the system is gradually heated to 300K at 1atm using a 200ps NVT set and 200ps NPT set. A 50ns production run followed to collect the balanced configuration for each 2ps interval. A speed scaling thermostat 49 with a time constant equal to 0.1ps was used to keep the temperature constant throughout the simulation. To maintain the pressure, the Berendsen pressure coupler was used in the NPT pre-balance run and the Parrinello-Rahman pressure coupler was used in the production run, with the pressure time constant and isothermal compression rate set to 2ps and 4.5×10-5 bar-1 respectively. In the whole simulation process, the integral time step of the motion equation is 2fs. The cut-off value for the non-bonding interaction is 12Å. The particle grid Ewald algorithm 53 is used to calculate long-range electrostatic interactions. [2]
Finally, the data results of these five systems were obtained (Table 2). The results showed that the average distance between the 8 active oxygen atoms of TFD-S and terbium ion was 2.41 Å (Figure 1-2), and the distance between active oxygen atoms and terbium ion after mutation of the four mutant proteins was much larger than 2.41 Å (Figure 1-3).
Figure 1-2. Schematic diagram of the binding of eight active oxygen atoms to terbium ions of TFD-S. The Tb(III) ion is shown as green sphere, and coordination bonds are shown as dashed lines. The numbers near the coordination bonds represent the distances between the atoms, with the unit of Å.
Figure 1-3. Schematic diagram of binding of active oxygen atoms to terbium ions in four TFD mutants
Table 2. Molecular dynamics simulation results
Learn
In the coordination structure, the distance between the ligand and the central atom can show the size of the binding force, but if the distance is too far, it is difficult to effectively bind. The molecular dynamics simulation results showed that the distance between the terbium ion and the oxygen atom capable of forming effective coordination bonds was about 2.41 Å, while the oxygen atom of the mutant was remote from the terbium ion, and it was expected that effective coordination bonds could not be formed, indicating that increasing the number of ligands near the original binding site could not enhance the activity of the target protein.
In addition, the protein spatial structure of these mutants is affected by the amino acid mutation, resulting in a larger distance between the oxygen atom of the original binding site and the terbium ion, which weakens the ability to adsorb terbium ion. Based on this, we believe that the method of mutating amino acids near the original binding site to increase the number of oxygen ligands is not advisable. By observing the protein (Figure. 1-4), we believe that a new set of rare earth binding sites can be added at the bottom of the TIM bucket, so that one unit of protein can adsorb two units of rare earth ions under an idealized state. At the same time, the influence on the original binding site will be minimized.
Figure 1-4. TFD-S combined with terbium ion
Cycle 2: A new set of rare earth binding sites is introduced de novo into the protein lumen of TFD-EE, thereby increasing the number of rare earth binding units of proteins.
Design
In this cycle, we hope to enhance the protein's ability to bind rare earth ions by increasing the number of rare earth ions that a single protein can trap.
Build
By analyzing the structure of the TFD protein, we chose to continue searching for new locations on the β-sheet of the protein lumen and mutate them into glutamate, thereby introducing new rare earth binding sites. With the help of PyMol and AlphaFold3, four sites I6, I52, I175, and I221 were finally selected for mutation, which was actually very similar to the original four binding sites. Figure 1-5 is an ideal building block for this cycle.
Figure 1-5. Ideal proteins bind to terbium ions
Test
The representative ions of this cycle are Tb(Ⅲ), La(Ⅲ), and Lu(Ⅲ), with La being the lightest lanthanide metal and Lu the heaviest. The process of molecular dynamics simulation is the same as that of Cycle 1, and the final results are shown in Table 3.
Table 3. Molecular dynamics simulation results
Learn
The data in Table 3 shows that compared with adding new oxygen coordination near the original site, the adsorption of rare earth ions at a new group of binding sites is more stable. However, compared with the data of TFD-S, the oxygen coordination distance of the protein in this cycle of mutation was still slightly larger than that of the central ion, indicating that the change of amino acids still had an impact on the protein structure. Specifically, the side chain became longer after isoleucine became glutamic acid, resulting in a slightly loose structure of the TIM bucket and an increase in distance.
In the engineering cycle of protein modification, we not only deepen the understanding of protein molecular structure and molecular dynamics simulation, but also deepen the understanding of synthetic biology engineering and standardization. In the future, we may continue to carry out protein directed evolution on this basis to create better rare earth binding proteins.
Part 2: Construction of yeast surface display system
Cycle1: Construction of yeast surface display system for Saccharomyces cerevisiae at pYD1 plasmid
Design
pYD1 plasmid is an expression vector designed to express, secrete, and display proteins on the extracellular surface of Saccharomyces cerevisiae. AGA2, polyhistidine (6×His) tags, and other elements have been designed on the plasmid. We first planned to achieve the fusion expression of AGA2 and TFD genes on the pYD1 plasmid, and then transferred into Saccharomyces cerevisiae strains, after protein expression and antibody incubation, the protein surface display was measured by flow cytometry.
Build
We successfully constructed the pYD1-TFD plasmid (Figure 2-1), which was transferred into Saccharomyces cerevisiae EBY100 for protein-induced expression.
Figure 2-1. A. The pYD1-TFD plasmid. B. AGA2-TFD gene fusion expression on pYD1-TFD plasmid. C. The sequencing results showed successful insertion of the TFD gene into the pYD1 plasmid.
Test
His tag mouse monoclonal antibody and FITC-coupled goat anti-mouse IgG were used to label the 6×His label on the protein displayed on the surface. The protein surface display was detected by flow cytometry.
Learn
The first flow cytometry results showed that the surface display efficiency (percentage of cells displayed) of our engineered yeast was low and did not differ significantly from the gated control group EBY100. We thought that the protocols for inducing protein expression and antibody incubation were not suitable, so in cycle 2, we modified the protocols and conducted the experiment again.
In addition, since we first conducted proof-of-concept in model yeast in this module, we hope to migrate to the non-model strain Issatchenkia orientalis in the future. We believe that the expression level of foreign proteins in Saccharomyces cerevisiae may not be high, which will also lead to unsatisfactory experimental results. Therefore, in cycle 2, we introduced the surface display pathway to Pichia pastoris for testing, which is generally considered to have high foreign protein expression capacity.
Cycle 2: Construction of yeast surface display system for Pichia pastoris
Design
We introduced the surface display pathway to Pichia pastoris. Firstly, the plasmid panARS-HZP-TFD was constructed for TFD protein surface display in the Pichia pastoris GS115 chassis strain. After being transferred into Pichia pastoris, the protein surface display was measured by flow cytometry after protein expression and antibody incubation.
In addition, we modified the experimental protocols for inducing protein expression and antibody incubation to re-test the protein surface display of engineered Saccharomyces cerevisiae in cycle 1.
Build
Here, we used a surface display system based on alpha-lectin. TFD-S is fused to the N-terminus of the anchoring protein Agα1, whose C-terminus is used for cell surface anchoring [3].
We used Gibson assembly to obtain Alpha factor secretion signaling peptide, 6×tag, (GGGS)4 linker, and Agα1 gene fragments by PCR, respectively., and then inserted them sequentially on the panARS-HZP plasmid between the pAOX1 promoter and the tAOX2 terminator (Figure 2-2). The constructed panARS-HZP-TFD plasmid was transferred to Pichia Pastoris GS115.
Meanwhile, induced expression of protein was performed on Pichia Pastoris GS115 (panARS-HZP-TFD) and Saccharomyces cerevisiae EBY100 (pYD1-TFD). For optimized protocols, please refer to our Protocols page.
Figure 2-2. A. The panARS-HZP-TFD plasmid. B. Fusion expression of alpha factor secretion signaling peptide-his tag-(GGGS)4 linker-Agα1 gene on the panARS-HZP-TFD plasmid. C. The sequencing results showed that the panARS-HZP-TFD plasmid was successfully constructed.
Test
The two strains were incubated with antibodies respectively, and the optimized protocols can be found on our Protocols page. The protein surface display was then measured by flow cytometry. The results showed that the surface display efficiency (percentage of displayed cells) of Saccharomyces cerevisiae EBY100 (pYD1-TFD) and Pichia pastoris GS115 (panARS-HZP-TFD) was significantly improved after optimizing the protocols of inducing protein expression and antibody incubation, which was close to 10% (Figure 2-3 A and C). The mean fluorescence intensity (Mean) of the cells also showed successful construction of the engineered strains (Figures 2-3 B and D).
Figure 2-3. A and B. Saccharomyces cerevisiae EBY100 cells carrying the pYD1-TFD plasmid were labeled with a 6×His (FITC) -labeled fluorescent group-conjugated antibody and analyzed by flow cytometry. C and D. Pichia pastoris GS115 cells carrying the panARS-HZP-TFD plasmid were labeled with a 6×His (FITC) -labeled fluorescent group-conjugated antibody and analyzed by flow cytometry.
Learn
Tests on the chassis strains of Saccharomyces cerevisiae and Pichia pastoris respectively showed that the protein surface display efficiency was similar, indicating that the influence of the chassis strains was not significant, and the surface display efficiency could be greatly improved by optimizing the protein-induced expression conditions and the incubation conditions of antibodies. Although the display efficiency of our test is still not high (about 10%), it has been proved that the engineered strain construction is successful. In the future, we can improve the display efficiency by further optimizing the experimental conditions.
Cycle 3: Detection of the rare earth adsorption capacity of engineered yeast by tryptophan-sensitized lanthanide luminescence mechanism
Design
After successfully constructing the engineered yeast, we need to further test the rare earth adsorption capacity of the engineered yeast. In our protein engineering module, we detect the rare earth binding capacity of TFD proteins using tryptophan-sensitized lanthanide luminescence mechanisms, as detailed on our Measurement page. In the yeast surface display module, we initially planned to use this strategy for detection when the TFD protein was displayed on the cell surface.
Build
The engineered yeast was incubated with TbCl3 solution, and a time-resolved fluorescence mode was adopted in the microplate reader to give excitation light of 280 nm, and the fluorescence values in the wavelength range of 520-570 nm were detected.
Test
The results showed that the fluorescence emission peak could not be observed at 545 nm for different proportions of the cell-REE incubation system.
Learn
We believe that the reason why fluorescence cannot be detected is that when the TFD protein is displayed on the cell surface, the cell system will bring huge background interference to the detection compared to the protein system. Therefore, here the tryptophan-sensitized lanthanide luminescence mechanism is no longer applicable. In cycle 4, we turned to finding new and convenient methods for rare earth detection.
Cycle 4: Detection of the rare earth adsorption capacity of engineered yeast by the Arsenazo III-based spectrophotometry assay
Design
Due to the high cost of using ICP-MS or ICP-OES, the Arsenazo III-based spectrophotometry assay was used here to detect the residual rare earth ion content in solution after biosorption, as described in our Measurement page.
Build
100 μM TbCl3 was added to the surface of Saccharomyces cerevisiae EBY100 (pYD1-TFD) cell culture solution displaying TFD protein. After 1 day of oscillating incubation, the cells were precipitated by centrifugation, and the concentration of Tb (Ⅲ) still free in the supernatant was detected. The cell culture medium without TbCl3 was used as the blank control.
Test
After the adsorption process, the content of Tb (Ⅲ) in the solution decreased significantly, which was manifested in the obvious color depth difference of the Arsenazo III-REE complex formed by samples before and after adsorption (Figure 2-4. A), which was also reflected by the value of A650. See our results page for more details.
Finally, the experimental results showed that our surface display engineered strain was successfully constructed and had strong rare earth ion adsorption capacity.
Figure 2-4 A. The absorbance at 650 nm of the solution before and after biosorption was determined by the Arsenazo III-based assay, and an obvious color depth difference was observed. B. A650 value of the standard solution of TbCl3 from 0 to 150 μM.
Part 3: Construction of yeast biofilm: overexpression of FLO11
Cycle1: Constructing the FLO11 gene expression cassette driven by PTEF1 promoter and integrating it into the genome of Saccharomyces cerevisiae
Design
FLO11 is an endogenous gene on the genome of Saccharomyces cerevisiae, which endows yeast with surface adhesion properties and is responsible for adhering cells to the surface of the substrate (including agar and plastic). However, in the type strain S. cerevisiae BY4741, the expression of FLO11 is inhibited due to a nonsense mutation in the coding sequence of the transcription regulator Flo8p, thus inhibiting the surface adhesion phenotype of the BY4741 strain.[4]
Figure 3-1. There is a nonsense mutation in the endogenous FLO8 of S. cerevisiae S288C/BY4741.
To overexpress FLO11 and restore the surface adhesion phenotype of Saccharomyces cerevisiae, we planed to first obtain the full length of the endogenous FLO11 gene on the genome of Saccharomyces cerevisiae BY4741 by PCR, then inserted the plasmid to construct the PTEF1-FLO11-TCYC1 gene expression cassette, and finally used the Cre/loxp recombination system to integrate the FLO11 gene expression cassette into the Saccharomyces cerevisiae genome.
TEF1 gene encodes transcription elongation factor 1α, which is a highly conserved protein that promotes GTP-dependent binding of aminoacyl-tRNA to ribosomal A sites in eukaryotes. The promoter of TEF1 has a strong and continuous expression activity, which can be used to efficiently express the target gene.[5]
Build
We obtained the pUMRI-HO-mCherry plasmid from the research group of Associate Professor Lidan Ye of Zhejiang University. The plasmid is a set of toolboxes constructed by the research group for homologous recombination in Saccharomyces cerevisiae[6], genotype: loxp-KanMX-URA3-pbr322ori-loxp, pTEF1-mCherry-tCYC1, HO homologous arm. The pUMRI-HO-mCherry plasmid carried the Cre/loxp recombination system and the homologous arm fragment at the HO site of the S. cerevisiae genome. The plasmid was linearized by the enzyme SfiI and transformed into S. cerevisiae, which could be integrated into the HO locus of S. cerevisiae by homologous recombination.[7]
Therefore, we planed to use Gibson assembly to obtain the FLO11 gene by PCR, inserted the position between the pTEF1 promoter and the tCYC1 terminator on the pUMRI-HO plasmid, and constructed the PTEF1-FLO11-TCYC1 gene expression cassette.
Figure 3-2. pUMRI-HO-pTEF1-FLO11-tCYC1 plasmid was constructed by Gibson Assembly.
Test
We obtained the FLO11 gene and pUMRI-HO plasmid vector skeleton by PCR, respectively. The results of agarose gel electrophoresis (Figure 3-3) showed that the pUMRI-HO plasmid vector skeleton (theoretical band of 5705 bp) and FLO11 gene (theoretical band of 4129 bp) could be successfully obtained. However, it can be seen that the brightness of the FLO11 electrophoresis strip is low, and the DNA concentration of the PCR product is extremely low. Therefore, we continued to purify and recover the PCR product of FLO11 by 8 fold and obtained 8×FLO11 with increased DNA concentration.
Figure 3-3. Electrophoresis results of pUMRI-HO plasmid vector skeleton, FLO11 gene, and 8×FLO11 gene
The pUMRI-HO plasmid vector skeleton and 8×FLO11 gene were assembled by Gibson Assembly. The ligation product was transformed into competent E. coli DH5α, and positive clones were screened on the Kan plate. Six E. coli clones were selected for colony PCR verification of plasmid construction (theoretical band of 297 bp), and electrophoresis results (Figure 3-4) showed non-specific bands.
Figure 3-4. Electrophoresis results of colony PCR of pUMRI-HO-FLO11 plasmid
For further verification, we took the above E. coli monoclonal amplification, extracted the plasmid, and sequenced it. The sequencing results (Figure 3-5) showed that the construction of pUMRI-HO-FLO11 plasmid failed.
Figure 3-5. The sequencing results of the pUMRI-HO-FLO11 plasmid showed that the construction failed.
After that, we conducted two rounds of repeated experiments and still failed to construct.
Learn
The results of electrophoresis experiments showed that no matter we changed the PCR DNA polymerase or changed the primer design, it was difficult to obtain a single FLO11 gene band, and it was often prone to bright primer dimer bands. After 8×FLO11 concentration, the brightness of the band was still low, and it was difficult to obtain correctly assembled positive clones after Gibson Assembly.
We presume that the reason is that the FLO11 gene is longer (full length of 4104 bp) and the CDS region contains a large number of serine and threonine repeat sequences[8]. On the one hand, it is prone to gene self-pairing during PCR amplification. On the other hand, it is also prone to mutations such as displacement during replication [9], which makes the PCR gene bands impure and trailing.
The FLO11 gene mediates the encoding of a GPI-anchored glycoprotein rich in serine and threonine, and its length is also diverse between parents and offspring. Therefore, we planed to use the CRISPR-Cas9 system to directly insert the strong constitutive promoter PTEF1 into the upstream of the endogenous FLO11 gene in the S. cerevisiae genome instead of amplifying the exogenous FLO11 gene expression cassette to achieve overexpression.
Cycle 2: Using the CRISPR-Cas9 system to integrate a constitutive strong promoter PTEF1 into the upstream of endogenous FLO11 gene in Saccharomyces cerevisiae
Design
During this cycle of design, we used the CRISPR-Cas9 system to directly insert the constitutive strong promoter PTEF1 into the upstream of the endogenous FLO11 gene in S. cerevisiae. Using the CRISPR-Cas9 system required us to construct plasmids containing gRNA, plasmids containing Cas9, and donor fragments for gene integration based on homologous recombination.
Figure 3-6. The composition and knockout principle of gene integration fragments.
Build
We obtained the gRNA expression plasmid p426-SpSgH-URA3 and Cas9 expression plasmid p416-Cas9-G418 from Dr. Ye's group at Zhejiang University for the construction of Saccharomyces cerevisiae CRISPR-Cas9 system.
According to the gene sequence of the recombination site, the appropriate gRNA sequence was designed by CRISPR assisted design website (https://benchling.com), and the gRNA expression plasmid was obtained by inserting the p426-SpSgH plasmid with enzyme BsaI. Then, homologous primers were designed to obtain the integrated fragment pTEF1 donor from pUMRI-HO plasmid by PCR, and the homologous arm length was 40 bp. Finally, the two plasmids and donor were introduced into Saccharomyces cerevisiae by lithium acetate transformation to achieve gene integration.
Figure 3-7. CRISPR-Cas9 tool. A. p426-SpSgH-URA3 plasmid. B. p416-Cas9-G418 plasmid. C. The pTEF1 promoter was inserted upstream of the FLO11 gene.
Test
We successfully obtained the pTEF1 donor fragment (theoretical length of 493 bp) and the plasmid p426-SpSgH-URA3-gRNA containing gRNA. The plasmid p416-Cas9-G418 containing Cas9 didn't need to be reconstructed.
Figure 3-8. A. pTEF1 donor electrophoresis results showed that it was successfully obtained. B. The sequencing results of p426-SpSgH-URA3-gRNA plasmid showed that the construction was successful.
The above two plasmids and donor were transferred into Saccharomyces cerevisiae BY4741 by lithium acetate method and coated on SC-URA + G418 plate and cultured at 30 °C for 2 ~ 3 days. Colony PCR was used to verify the gene integration (target band of 729 bp), and two positive clones appeared (Figure 3-9).
Figure 3-9. Colony PCR verified the electrophoresis results of CRISPR-Cas9 gene integration. No. 2 and No. 13 showed positive results (target band of 729 bp).
After verifying the integrated genotype of S. cerevisiae, the p416-Cas9-G418 plasmid and p426-SpSgH-gRNA plasmid introduced by the CRISPR/Cas9 system were removed based on the principle that the plasmid would be naturally lost during passage without screening pressure and SC+FOA plate screening. After sequencing verification, a stable genetic S. cerevisiae BY4741 PTEF1-FLO11 strain was obtained.
Figure 3-10. Sequencing results showed that the pTEF1 promoter was successfully integrated into the upstream of the endogenous FLO11 gene in S. cerevisiae BY4741.
Learn
According to the results of sequencing and marker screening, the TEF1 promoter sequence was successfully integrated into the upstream of the endogenous FLO11 gene of S. cerevisiae BY4741 by the CRISPR-Cas9 system. Finally, S. cerevisiae strain overexpressing the FLO11 gene was obtained, which was able to carry out subsequent biofilm formation and characterization experiments.
In the process of the engineering cycle, we not only deeply understand the methods of recombination of target genes such as enzyme digestion, Gibson Assembly, and CRISPR-Cas9 system, but also iteratively obtain engineering yeasts that meet our own experimental design. It also popularizes a variety of synthetic biology knowledge and thinking paths for us. In the future, we may continue to explore optimization schemes on this basis to achieve more efficient endogenous gene overexpression.
References
[1] Mattocks, Joseph A., et al. Enhanced rare-earth separation with a metal-sensitive lanmodulin dimer. Nature, 2023, 618(7963): 87-93.
[2] Tutorials and Webinars — GROMACS webpage https://www.gromacs.org documentation
[3] Andreu, Cecilia, and Marcel Lí Del Olmo. Yeast arming systems: pros and cons of different protein anchors and other elements required for display. Applied microbiology and biotechnology. 2018, 102: 2543-2561.
[4] Van Mulders, Sebastiaan E., et al. Phenotypic diversity of Flo protein family-mediated adhesion in Saccharomyces cerevisiae. FEMS Yeast Research. 2009, 9(2): 178-190.
[5] Kitamoto, N., Matsui, J., Kawai, Y. et al. Utilization of the TEF1-a gene (TEF1) promoter for expression of polygalacturonase genes, pgaA and pgaB, in Aspergillus oryzae. Appl Microbiol Biotechnol. 1998, 50: 85-92.
[6] Ye, Lidan, Xiaomei Lv, and Hongwei Yu. Assembly of biosynthetic pathways in Saccharomyces cerevisiae using a marker recyclable integrative plasmid toolbox. Frontiers of Chemical Science and Engineering, 2017, 11: 126-132.
[7] Baganz, Frank, et al. Suitability of replacement markers for functional analysis studies in Saccharomyces cerevisiae. Yeast, 1997, 13(16): 1563-1573.
[8] Zara G, Zara S, Pinna C, Marceddu S, Budroni M. FLO11 gene length and transcriptional level affect biofilm-forming ability of wild flor strains of Saccharomyces cerevisiae. Microbiology. 2009 Dec;155(Pt 12):3838-3846.
[9] Watari, Junji, et al. Molecular cloning and analysis of the yeast flocculation gene FLO1. Yeast. 1994, 10(2): 211-225.