2.2 Canavanine plates selecting for CAN1 mutant colonies
The mutation rate of CAN1 (with PT7 targeting sequence) was determined to characterize the efficiency of our mutagenesis system in this work. Cells were grown to saturation for 24 h in liquid SC medium lacking appropriate amino acids depending on the autotrophic markers to maintain plasmids. Cultures were then diluted and adjusted to OD600=1 and added into inducing media. Inducing media contains 0.2% galactose and 1 uM β-estradiol. After incubation at 30℃ for 8~24 h, 30~50 μL of the cultures were plated onto SC-Leu-His-Arg plates with 60mg/L canavanine, and the same volume of culture was gradient diluted and plated onto YPD plates. Colonies on canavanine and YPD plates were counted after 2~3 days to determine the mutation rate.
Mutation rate analysis
The SC-Leu-His-Arg with 60mg/L canavanine plates were incubated at 30℃ for 2~3 days and the colonies were counted. The sample size was based on the number of countable colonies on a single plate (up to 500), and the number of colonies on the YPD plates was used as a control. Statistical analysis was performed using Graphpad Prism.
Mutation diversity analysis
For mutation diversity analysis, the average-sized colonies were randomly selected and the target locus was PCR amplified. The PCR products were analyzed by Sanger sequencing and compared with the reference sequence.
Results
PmCDA1 increased the mutation rate of the target gene
Cytidine deaminase PmCDA1 can catalyze the deamination of cytosines, mutating cytosine (C) to uracil (U), while uracil pairs with adenine (A) in the subsequent DNA repair and replication process to complete the C→T conversion.[28, 29] Uracil glycosylase inhibitor (UGI) is usually used to block the activity of uracil glycosylase (UNG) and inhibit the removal of mismatched uracil, thereby reducing other types of mutations such as C→G and C→A.[30, 31] We hypothesized that in the absence of UGI, the types of base substitutions generated by PmCDA1 might be more diverse.[32] We constructed the mutagenesis plasmids carrying the pGAL-PmCDA1-T7 RNAP expression cassette. Meanwhile, we inserted the T7 promoter sequence upstream of the target CAN1 gene so that PmCDA1-T7 RNAP could be specifically recruited to the target site defined by the T7 promoter (Fig. 1A). The mutation rate was characterized by the frequency ofCAN1 gene inactivation. We performed the assay on yeast strains with and without the mutagenesis plasmids. After induction with galactose for 8~24 h, we plated cells on canavanine plates capable of inhibiting CAN1 + cell growth, and counted colonies on canavanine plates to assess the mutation rates. Compared with the control strain, the mutation frequency of theCAN1 gene in the strain expressing PmCDA1-T7 RNAP was significantly increased (Fig. 1C), indicating that PmCDA1-T7 RNAP can effectively increase the mutation rate of the target gene in S. cerevisiae .
Appropriate extension of the linker length can sometimes expand the targeting scope.[30, 33] Based on this, we further investigated the influence of different linker lengths on the mutation effect of PmCDA1-T7 RNAP. Two linker lengths (32a.a. and 84a.a.) were selected (Fig. 1B), and the mutation rate of CAN1 gene was determined. We observed that the length of linker between PmCDA1 and T7 RNAP had no significant effect on the mutagenic activity of the fusion protein. After 24 h of induction, both PmCDA1-32a.a.-T7 RNAP and PmCDA1-84a.a.-T7 RNAP could increase the mutation frequency up to 1.0x10-3~1.2x10-3(Fig. 1D). We analyzed the mutations generated by PmCDA1-32a.a.-T7 RNAP and PmCDA1-84a.a.-T7 RNAP by sequencing the PT7CAN1 locus. The data demonstrated that the mutation types generated by PmCDA1-32 a.a.-T7 RNAP and PmCDA1-84 a.a.-T7 RNAP were basically the same, with C→T mutations accounting for more than 97% and the remaining 3% being other types of mutations (Fig. 1E), which was also consistent with the mutation characteristics of PmCDA1.[16] The distribution of the mutations in the CAN1 gene was also similar (Fig.1F).Therefore, the length of linker between PmCDA1 and T7 RNAP has no significant effect on the mutation effect. Meanwhile, even without UGI, the mutation types generated by PmCDA1 were really simple, and most of them were C→T mutations.
DNA-modifying enzymes improved mutation effect
When using PmCDA1-T7 RNAP as the mutator, the strong bias towards C→T mutations would reduce the diversity of mutants. In cells, the mismatched U resulting from the deamination of C is excised by DNA-modifying enzymes to form abasic sites. In the subsequent DNA repair process, different bases could be randomly inserted into the abasic sites, resulting in different types of mutations.[34, 35] Thus, we hypothesized that fusing different DNA-modifying enzymes to PmCDA1 would improve the mutation outcome.
We first chose MAG1 as the DNA-modifying enzyme to link to PmCDA1. MAG1 can remove mismatched bases and initiate base excision repair (BER).[36, 37] Overexpression of MAG1 in cells leads to an elevated genomic mutation rate.[38] We assumed that the addition of MAG1 would enhance the excision of mismatched U and create more abasic sites, thereby generating diverse mutation types during the subsequent DNA repair process. Since the relative positions of PmCDA1, T7 RNAP and the DNA-modifying enzymes may influence the mutation outcome, we designed five expression cassettes with different constructions and linkers (Fig. 2A). We compared the mutagenic activity of these fusions with PmCDA1-T7 RNAP and the control strain without mutagenesis fusions. We found that the construction of the fusion proteins significantly affects the mutation outcome. The data indicated that Cons. 3 could raise the mutation frequency up to 1.9x10-3, which was 1.6 to 2 times higher than that of PmCDA1-T7 RNAP and was the highest among these five fusions. The mutation frequencies of the other four constructions were about 3x10-4, which was significantly reduced compared with PmCDA1-T7 RNAP (Fig. 2B). When analyzing the types of mutations, we found that C→T mutations accounted for 64.5% of the mutations produced by Cons. 3, followed by G→A mutations (19.2%), C→G mutations (12.5%), and G→C mutations (3.8%). The proportion of non-C→T mutations is 11-fold higher than that of PmCDA1-T7 RNAP (Fig. 2C). We suspect that the increase in non-C→T mutations may be due to the enhancement of the excision of mismatched bases, thus forming more abasic sites-which are important for BER-and increasing the diversity of mutations. Although the mutation frequencies of the other four candidates were low, the mutation types were diverse and most of them were non-C→T mutations. We speculated that the presence of DNA-modifying enzymes in these constructions affected the activity of PmCDA1, resulting in the mutation effect that was apparently different from that of PmCDA1. Considering the mutation frequency and diversity, we selected the Cons. 3 for further work.
Based on this, we chose 6 other DNA-modifying enzymes and analyzed their mutagenic activity (Fig. 3A).[39-44] Among these candidates, EXO1-PmCDA1-T7 RNAP produced the highest mutation frequency of 2.2x10-3, which was twice that of PmCDA1-T7 RNAP (Fig. 3B). When analyzing the mutations generated by EXO1-PmCDA1-T7 RNAP, we found a strong bias towards C→T mutations, similar to PmCDA1. In 48 randomly selected colonies, C→T mutations accounted for 80.0%, followed by C→G mutations (11.0%) and other types of mutations (8.0%) (Fig. 3C). EXO1 is a key enzyme in DNA double-strand break repair, mismatch repair, and other repair pathways,[43, 45] and we speculated that EXO1 may act synergistically with PmCDA1 to further increase the mutation frequency. In laboratory evolution, a high mutation rate can greatly accelerate the evolution process. The mutation frequency generated by REV3-PmCDA1-T7 RNAP was about 1.26x10-3, which was slightly lower than that of MAG1-PmCDA1-T7 RNAP and EXO1-PmCDA1-T7 RNAP (Fig. 3B), but the mutation types were diverse, of which C→T mutations accounted for 70.4%, followed by C→G mutations (14.1%), G→A mutations (9.86%) and other mutations (5.64%)(Fig. 3C). REV3 involves in DNA translesion synthesis repair, double-strand break repair, and DNA damage-induced mutagenesis.[44] Therefore, we hypothesized that, similar to MAG1, REV3 strengthens the DNA translesion synthesis repair, in which different bases are inserted into abasic sites, resulting in multiple types of mutations. In the process of laboratory evolution, the occurrence of different types of mutations enlarges the mutant spectrum, and allows us to screen a wider range of desired strains. Different from cytidine-bearing mutators,[21, 22, 24] after altering the mutation spectrum by DNA-modifying enzymes, mutations could occur across all four nucleotides, with G→A or C→G mutations being the main mutation types, except for C→T mutations, meaning that our system is able to play a complementary role to the cytidine-based evolutionary tools.
Dual T7 promoters increased mutation frequency
After changing the constrution of the fusion proteins and adding DNA-modifying enzymes to improve the mutation effect, the mutagenic activity of mutators have been improved significantly compared with PmCDA1-T7 RNAP. Based on this, we inserted two reverse T7 promoters on both sides of the CAN1 gene and analyzed the mutation effect of MAG1/EXO1/REV3-PmCDA1-T7 RNAP under this condition (Fig. 4A).[22, 24] We observed that the addition of the second T7 promoter significantly increased the mutation frequency. In the dual T7 promoter system, the mutation rate generated by EXO1-PmCDA1-T7 RNAP could reach 5.13x10-3 after 24 h of induction, which was 1.57-fold higher than that of the single T7 promoter system (Fig. 4B). With dual T7 promoters, the mutation frequencies generated by MAG1-PmCDA1-T7 RNAP and REV3-PmCDA1-T7 RNAP were also significantly increased. After induction for 24 h, the mutation rates of strains expressing MAG1-PmCDA1-T7 RNAP and REV3-PmCDA1-T7 RNAP were 3.72x10-3 and 3.26x10-3, respectively, which were 2~2.5-fold higher than that of the single T7 promoter system (Fig. 4B). We speculated that the dual T7 promoters may increase the probability of T7 RNAP binding to the T7 promoter, so that DNA-modifying enzymes-PmCDA1 have a greater chance of acting on the target gene, leading to higher mutation rates.
When analyzing the mutations produced in the dual T7 promoter system, we found that the introduction of the reverse T7 promoter had no strong effect on the mutation types. In the dual T7 promoter system, C→T mutations generated by MAG1-PmCDA1-T7 RNAP comprised 65.7%, followed by G→A mutations (13.6%), C→G mutations (11.9%), and other types of mutations (8.8%). Although the proportion of G→A mutations was slightly lower than that in the single T7 promoter system, other types of mutations increased, such as some transversion mutations like G→T. The mutation types of EXO1-PmCDA1-T7 RNAP in the dual T7 promoter system were barely changed, among which C→T mutations comprised 82.4%, followed by C→G mutations (8.33%), G→A mutations (5.1%) and other mutations (4.17%). In REV3-PmCDA1-T7 RNAP, C→T mutations comprised 73.6%, followed by C→G mutations (12.2%), G→A mutations (12.12%) and other mutations (2.8%) (Fig. 4C). It can be seen that the dual T7 promoter system had little effect on the mutation characteristics of our mutagenesis tools, but slightly increased the frequency of some transversion mutations (such as G→T, G→C, etc.) and made the mutation types more diverse. Existing deaminase-based evolutionary techniques are difficult to achieve transversion mutations, and most of them are biased towards generating specific types of mutations.[22, 24, 25] Therefore, our mutagenesis tools with dual T7 promoters can further enlarge the mutation libraries, thus promoting the process of evolution.