Identification of functional butanol-tolerant genes from Escherichia coli mutants derived from error-prone PCR-based whole-genome shuffling

Background Butanol is an important biofuel and chemical. The development of butanol-tolerant strains and the identification of functional butanol-tolerant genes is essential for high-yield bio-butanol production due to the toxicity of butanol. Results Escherichia coli BW25113 was subjected for the first time to error-prone PCR-based whole-genome shuffling. The resulting mutants BW1847 and BW1857 were found to tolerate 2% (v/v) butanol and short-chain alcohols, including ethanol, isobutanol, and 1-pentanol. The mutants exhibited good stability under butanol stress, indicating that they are potential host strains for the construction of butanol pathways. BW1847 had better butanol tolerance than BW1857 under 0–0.75% (v/v) butanol stress, but showed a lower tolerance than BW1857 under 1.25–2% (v/v) butanol stress. Genome resequencing and PCR confirmation revealed that BW1847 and BW1857 had nine and seven single nucleotide polymorphisms, respectively, and a common 14-kb deletion. Functional complementation experiments of the SNPs and deleted genes demonstrated that the mutations of acrB and rob gene and the deletion of TqsA increased the tolerance of the two mutants to butanol. Genome-wide site-specific mutated strains DT385 (acrB C1198T) and DT900 (rob AT686–7) also showed significant tolerance to butanol and had higher butanol efflux ability than the control, further demonstrating that their mutations yield an inactive protein that enhances butanol resistance characteristics. Conclusions Stable E. coli mutants with enhanced short alcohols and high concentrations of butanol tolerance were obtained through a rapid and effective method. The key genes of butanol tolerance in the two mutants were identified by comparative functional genomic analysis. Electronic supplementary material The online version of this article (10.1186/s13068-019-1405-z) contains supplementary material, which is available to authorized users.


Error-prone PCR genome shuffling
Genomic DNA of E. coli BW25113 was extracted and used as the template for error-prone PCR (epPCR) amplification. The epPCR was performed using a ramping procedures according to the previous reports with some modifications (Luhe et al. 2011;Ye et al. 2013), which included that 10 mer, 12 mer, 14 mer and 15 mer random primers (synthesized by Sangon, Shanghai) were combinatorically used, and 20-50 ng genomic DNA and 5 U of Taq DNA polymerase was added in a final volume of 50 μL.
The PCR product amplified by primer pair 12-mer (13.24 µM) and 10-mer (20.06 µM) and that by 15-mer (33.33 µM) had higher DNA concentration than that amplified by other primer sets. Optimal primer combination was used for epPCR amplification, and the PCR products were then concentrated 5-10 times through ethanol-precipitated for electroporation. The transformation was performed as previous described (Huang et al 2018).
In the first round of shuffling, competent cultures were plated on LB plates containing 1.3, 1.4, 1.5 1.6, 1.7, 1.8 and 1.9 % (v/v) butanol for selection of transformants. The transformants were picked up from these plates with 1.3-1.8 % of butanol, and sixty one clones and the initial strain BW25113 were transferred into 25-mL screw-capped tubes containing 3 mL LB for overnight growth and tolerance evaluation in batches. Five percent (v/v) of the overnight cultures of above transformants were then subcultured in 3 mL LB medium containing 0.5 % butanol 4 for butanol adapted culture. Then, these transformants and BW25113 were precultured in LB media for overnight, and finally transferred into test-tubes containing 4 ml LB media with 0.85 % butanol of for primary growth evaluation. The optical densities at 600 nm (OD600) were measured at 3 and 8 h to evaluate their butanol tolerance. Control strain BW25113 and five strains showing higher cell densities than control strain, were cultured in 250 mL screw-cap conical flask containing 50 mL LB media with 0.95 % butanol for further growth evaluation. Strain BW184 from the plate containing 1.8 % butanol had 33% higher cell density (OD600=0.48) than BW25113 (OD600=0.36) at 7 h, and it was thus used as an initial strain for next round of genome shuffling.
In the second round of shuffling, transformants were screened on plates with 1.8, 1.9, 2.0 and 2.1% butanol. Ninety eight colonies were picked up from plates containing 1.9-2.1% butanol for growth evaluation in test tubes in batches. Eight strains, showing higher cell densities than BW184 and BW25113, were further cultured in conical flask containing LB media with 0.95% butanol for growth evaluation as described aforementioned. The cell densities (OD600=0.64-0.70) of these strains were increased 180-200 % compared with those of control (OD600=0.22).
And BW1847(OD600=0.70) and BW1857(OD600=0.68) showed relative higher maximum cell densities than other strains, therefore, the two strains were used for further growth evaluation, resequencing and functional identification of butanol tolerant genes.

5
The cassettes of the wildtype and mutated genes were amplified from BW1847 and BW1857 genome using primers listed in  Table S3, and the SNPs between mutants and wildtype were also proved by sequencing, as shown in Table 1.

Construction of deletion mutants 6
Each mutated (Table 1) and deleted gene (Table S1) in BW1847 and BW1857 was knocked out in BW25113 (pKD46) via lambda red recombinant system (Datsenko et al, 2000) for functional complementation experiments. The chloramphenicol resistance cassettes flanked by homologous sequence of target genes were amplified from pKD3 plasmid using serial primers (DF/DR) ( Table S4). The target fragments were then electrotransformed into competent BW25113(pKD46) cells, and the transformants were screened on LB plates with 25 μg/mL chloramphenicol.
The PCR identification of transformants was performed to detect the deletion of target gene as follows: a freshly isolated colony was suspended in 20 μL PCR reaction mixture containing primer pairs TF/ZY-R, ZY-F/TR and TF/TR (Table S4), respectively. PCR program was described as aforementioned. The primers TF and ZY-R were used to identify the junction between left homologous arm and Cm R , and primer pair ZY-F/TR were used to amplify the junction between Cm R and right homologous arm. Primer pair TF/TR was used to verify simultaneous loss of the target gene and gain of a novel DNA fragment (the Cm R ). The deletion mutants obtained were shown in Table S5.

Construction of overexpression vectors
The wild-type and mutated target genes were amplified using serial primer pairs OF/OR (as shown in Table S6), and plasmids obtained by TA cloning (listed in The pBAD30 and overexpression plasmids obtained above (Table 2) were transformed into the competent wild E. coli BW25113 and the corresponding gene deletion strain (Table S5). The overexpression strains were obtained and named as "name of host stain (plasmid)" (listed in Table 2), and used to growth evaluation.
These strains were transferred to 3 mL of LB medium with 100 μg/mL ampicillin antibiotic, 0.5% (v/v) butanol and 0.02% (w/v) L-arabinose for 4-5 h of culture. Then, 15 μL of cultures were added to 15 mL of LB medium as described above without butanol for overnight culture. The cells were harvested by centrifugation (4°C/5000g/2 mins) and resuspended with fresh LB medium. The resuspending culture was inoculated to 50 mL of LB medium contained 100 μg/mL ampicillin antibiotic, 0.02-2% (w/v) L-arabinose and with or without 0.75% (v/v) butanol. The 8 initial OD600 was controlled at 0.1. The cultures were incubated at 37°C with 190 rpm agitation, and the OD600 was measured every 1.5 hours. The 0.02% L-arabinose was found to be the optimal induction concentration.

Construction of multiple-gene deletion strains by CRISPR (Clustered regularly
interspaced short palindromic repeat sequences)/Cas9 system Several deletion strains were performed using CRISPR/Cas9 system (Jiang et al, 2015). Inverse PCR was performed using pTargetF as template to introduce the target sequence of N20 (20-bp complementary region) to the upstream of sgRNA in pTargetF plasmid. The N20 sequence was introduced to the primers shown by Table   S7. The 50 μL reaction mixture contained 2 ng pTargetF plasmid, 1 μL of 10 μM geneF/Target-R (Table S7), 25 μL of 2 x pfu Master Mix (CW Bio, China). "Touch down" procedure as described above was used except that the annealing temperature decreased from 60°C to 50°C. Then, 0.5 μL DMT enzyme (10 U/μL; GD111, TRANSgene, China) was added into the PCR products, and the mixture were incubated at 37 for 1 h to digest methylated template plasmid. The products were ℃ purified and transformed into DH5α cells (MCC001, DingGuo, China). After incubation at 37 for 1 h, the cultures were plated on LB agar plate with 50 μg/mL ℃ spectinomycin. The colonies were confirmed by colony PCR using geneIF/pTarget-IR as primers. The positive pTargetF-geneN20 plasmids (Table S8) were extracted for the subsequent experiment.
The left and right homologous arms of the recombinant fragment were amplified using BW25113 genome as template, geneDLF/geneDLR (left arm) and 9 geneDRF/geneDRR (right arm) as primer sets, respectively (Table S7). The two homologous arms were fused to a recombinant fragment using overlapping PCR. And the overlap PCR was performed using equal molar ratio of left and right homologous fragments (10-20 ng) as template, and geneDLF/geneDRR as primers. The recombinant fragments were then purified and concentrated to 200 ng/μL for electro-transformation.
The pCas plasmid was transformed into BW25113 competent cells, and the transformants were screened on LB agar plate with 50 μg/mL kanamycin. The positive BW25113 (pCas) colonies were picked up and cultured in LB media with 10 mM L-arabinose (for induction expression of Red recombinase) for preparation of competent cells. When the cells were cultured at 30 until the OD ℃ 600 reached 0.375-0.6, the electrically competent cells were prepared. About 50-100 ng of pTargetF-geneN20 plasmid and 500 ng of recombinant fragments were electro-transformed into 40 μL of competent cells. The competent cultures were incubated at 30 for 1.5 h and then plated on LB agar plates with 50 μg/mL ℃ spectinomycin and kanamycin. The colonies were screened at 30 overnight and ℃ confirmed by colony PCR using geneIF/geneDRR as primers.
The IPTG was added to media in order to induce the expression of a sgRNA in pCas plasmid, whose expression product could locate on the replicon pMB1 of the pTargetF-geneN20 plasmid (Table S8). The above positive colonies were inoculated into 2 mL LB media with 50 μg/mL kanamycin and 0.5 mM IPTG for 12-15 h cultivation at 30 with 200 rpm stirring to eliminate the pTargetF ℃ -geneN20 plasmid.
The cultures were then diluted 10 5 -10 7 -fold with LB media, and 0.1 mL of this dilution was plated on LB agar plate containing 50 μg/mL kanamycin and 0.5 mM IPTG. The clones were further picked up from the agar plates after 12-15 h cultivation, and then point inoculated on LB (with 50 μg/mL kanamycin) agar plates with and without 50 μg/mL spectinomycin, respectively. Clones that could grow on the latter plate but not on the former that were picked up for further elimination of pCas. The positive clones were inoculated into LB media for 12-15 h cultivation at 37 ℃ with 200 rpm agitation to eliminate temperature-sensitive plasmid pCas. The cultures were then diluted 10 5 -10 7 -fold in LB media, and 0.1 mL of the dilution was plated on LB agar plates, and then the single clone was point inoculated on LB agar plates with or without 50 μg/mL kanamycin, respectively. The clones that could grow on LB agar plate and could not grow on that with kanamycin were selected for sequencing confirmation of the edited gene.

Genome-wide mutation of rob and acrB gene using site-specific mutagenesis
A strain with AT 686-7 base deletion of rob in genome was constructed by site-specific genomic integration ( Fig. S8 was performed using touchdown procedure as described above. The fragment F1234 was then purified and concentrated to 100 ng/μL. Strain BW25113(pTK-RED) was cultured in 50 mL SI-LB (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl, 100 μg/mL spectinomycin and 2 mM isopropyl-β-D-thiogalactopyran oside (IPTG)) medium at 30 . ℃ When its OD600 reached 0.375 -0.6, the competent cells were then prepared. About 5-6 μL of the purified F1234 fragment was then electro-transformed into 40 μL of BW25113(pTK-RED) competent cells. The competent cultures were incubated at 30℃ for 1 h and then plated on SI-LB agar plates with 100 μg/mL tetracycline. The tetracycline resistant mutants were screened and confirmed by colony PCR.
To induce the expression of meganuclease I-Sce Ⅰ and remove the resistance gene tetA from the genome, the positive colonies were inoculated into 3 mL of SIL -LB media (LB media with 100 μg/mL spectinomycin, 2 mM IPTG and 0.2% w/v L-arabinose) for 12-15 h cultivation. The cultures were then diluted 10 5 -10 7 -fold in LB media, and 0.1 mL of the dilution was plated on SIL-LB agar plates. Positive clones were picked up from the SIL-LB agar plates, and then point inoculated on SI-LB agar plates with or without 100 μg/mL tetracycline, respectively. Clones, which could grow on SI-LB agar plates without tetracycline but not on that with tetracycline, were selected and confirmed by colony PCR used primers UFrob and LR. The PCR-positive clones were further PCR-amplified using 2 X pfu Master Mix for sequencing verification (Sangong Biotech, Shanghai, China). All primers used were listed in Table S9. The confirmed positive strains were inoculated into LB media and cultured at 42℃ with 200 rpm shaking for 12-15 h in order to eliminate plasmid pTK-RED.

Knockout of the 14-kb DNA fragment
Both BW1857 and BW1847 have a 14-kb DNA fragment deletion. We speculated that the 14-kb deletion can be obtained by the lambda-Red recombination system.
Furthermore, the corresponding experiments for knocking out the 14-kb DNA fragment and a control gene (RS18950) were performed using BW25113(pKD46) as a target strain as described in the aforementioned methods. Primers 14kbDF and14kbDR (Table S11) were used to amplify the chloramphenicol resistance cassettes flanked by homologous sequences of the target 14-kb DNA fragment. The target fragments were then electrotransformed into competent BW25113(pKD46) cells, 13 and the transformants were screened on LB plates with 25 μg/mL chloramphenicol.
The PCR identification of transformants was performed to detect the deletion of the 14-kb DNA fragment with primer pairs 14kbTF/ZY-R, ZY-F/395TR and 14kbTF/395TR (Table S11), respectively. Three transformants were picked up for PCR identification using the above three primer sets. The PCR products showed the predicted size bands (Fig. S9), which indicates that the 14-kb DNA fragment was successfully deleted and replaced by a chloromycetin resistance cassette (Fig. S9A).
These experiments, thus, demonstrate that the 14-kb deletion in the two mutants may have resulted from homologous recombination. In this experiment, about 40-bp homologous arm sequences could successfully yield crossover recombination, resulting in the knockout of a 14-kb DNA fragment, which means that random PCR products with coincident left (up-stream) and right (down-stream) homologous arm sequences of 14-kb fragments could also yield a crossover homologous recombination.
Therefore, the 14-kb deletion could have resulted from crossover recombination via Red recombinase, which is produced from pKD46 plasmid used in the epPCR-based genome shuffling experiment.

Calculation of the mutation rate of BW1847 and BW1857
The spontaneous mutation rate of E.coli is reported to be 8.9 × 10 −11 per base-pair per generation (Wielgoss et al., 2011), and E.coli has a generation period of 15-20 mins. The total time for recovery and screening after electroporation was about 18-20 h (~100 generations), and the final spontaneous mutation rate was 8.9×10 −9 per base-pair after 100 generation. The final spontaneous mutation rate of the two mutants 14 (BW1847 and BW1857) was calculated as follows: BW1847 and BW1857 had about 7 and 9 mutations, respectively. The number of mutations (7-9) were divided by 4.6×10 6 bp of the E. coli genome, the final mutation rate obtained, 1.5-2.0×10 -6 per bp, is about 1000-fold higher than the spontaneous mutation rate (8.9 × 10 −9 per base-pair).             The strains were cultured in LB media containing 0.75 or 1.25 % (v/v) butanol, and the growth improvement was calculated according to the formula (1) as described in Method section. The value of growth improvement of each strain was divided by that of control BW25113, and the ratio value was shown as the relative growth improvement.    Table S5.
29 Fig. S6 The putative mechanism of AcrB efflux The butanol efflux mechanism model was speculated according to the previous report (Seeger et al., 2006).   indicate the truncated RS08320 and RS08395 genes, respectively. The 14-kb DNA fragment was deleted and replaced by a chloromycetin resistance cassette, using the lambda-Red recombination system. Three primer sets were used to identify transformants with the 14-kb deletion, and the corresponding DNA fragments 1 (red), 2 (blue) and 3 (black) were amplified using the three primer sets. Panel B shows the agarose gel electrophoresis results for PCR identification of transformants.
Three clones designated as T1, T2 and T3 were selected for PCR identification using the above primers.
A B