lysis gene E counter-selection cassette in S. marcescens
A counter-selection marker would be more attractive if it has a universal application. In the interest of generality, application of lysis gene E counter-selection cassette in other bacteria should be tested.
S. marcescens , a rod-shaped gram-negative bacterium, have the ability to produce a variety of valuable metabolites, such as prodigiosin, chitinase and serratiopeptidase(Emruzi et al., 2018; Pan et al., 2019; Velez-Gomez, Melchor-Moncada, Veloza, & Sepulveda-Arias, 2019; Yip et al., 2019). We select S. marcescens to test the counter-selection effect of lysis gene E cassette. Unfortunately, high-temperature condition like 42 °C is not suitable for the growth ofS. marcescens (W. Chen et al., 2021), We resorted to the AraC/PBAD regulatory system which function well in this species. At first, E-GmR fragment with short homologous arms was amplified and transformed into competent MG1655[pKD46] to substitute the CDS of Red recombinases in plasmid pKD46 to construct the goal plasmid pKD-EG. Therefore, expression of lysis gene E is thoroughly controlled by AraC/PBAD inducible system (Fig. 3A). After identification (Fig. 3B), plasmid pKD-EG was transformed into S. marcescens GY1 for functional testing. GY1[pKD-EG] grew well in LB broth, but it cannot survive in LB broth supplemented with 0.4% arabinose, and the LB broth is very clear even 18h after inoculation (Fig. 3C). This result indicates that expression of lysis gene Ethrough AraC/PBAD regulatory system could effectively kill its host.
In genomic seamless modification, counter-selection marker usually works in single copy. Therefore, the counter-selection effect of lysis geneE inserted in the chromosome must be evaluated. The 2,303 bpPBAD-E-GmR (including AraC/PBAD regulatory system, CDS of lysis gene E and gentamycin resistance geneGm R ) double selection cassette was therefore inserted into pigA CDS of S. marcescens. The successfully constructed strain was named as GY4 (Fig. 3D). Obvious differences can be seen when GY4 grown in LB broth without or with the addition of 0.4% arabinose (Fig. 3E). This result indicates that single copy of counter-selection marker gene E is also efficient. It is further verified by the spot dilution experiment: GY4 grew well on LB plate, while it cannot survive on arabinose added plate and only a faintly layer of dead cells was observed (Fig. 3F). It should be noted thatpigA is an indispensably gene for the synthesis of prodigiosin, a kind of red pigment, therefor GY1 shows red phenotype while pigAgene mutated GY4 is white.
Having verified the lethality of chromosomal expressing lysis geneE in S. marcescens , reverse mutation experiment was finally conducted to test its counter-selection effect in genomic modification. Using ssDNA mediated Red recombination, we tried to repair the insertion inactivation of pigA gene in GY4 (Fig. S1). Numerous red colonies dotted with a few white colonies were observed after transported GY4 with ssDNA, while only sparse white colonies appear in the control group without adding ssDNA (Fig. 4A). Colony in red indicates that the ability of synthesizing prodigiosin was repaired due to successful reverse mutation, which is verified by colony PCR identification (Fig. 4B).
dsDNA mediated homologous recombination was also tested. Partial fragment of T7 RNA polymerase (~ 1,000 bp in length) was amplified and transported into GY4 to substitute thePBAD-E-GmR cassette. 4 out of 10 of randomly selected colonies were true recombinants (Fig. 4C).
All these results suggest that lysis gene E counter-selection cassette can be used in S. marcescens for genomic seamless modification.
Improving selection stringency frequency through combining E and kil
In genomic seamless modification, selection stringency frequency is a decisive factor that influences counter-selection and therefor an indicator for evaluating an excellent counter-selection marker(Khetrapal et al., 2015). The higher the selection stringency frequency is, the lower the background it causes. Low background is very conducive to the selection of recombinants during seamless modification. The selection stringency frequency of lysis E under the control of promoterpL at ack locus in MG1655 is about 2.7×10−7 (Fig. 5A), performing better than thekil counter-selection cassette constructed in our previous work (Wei Chen et al., 2019). However, it is still much lower than the best counter-selection system, inducible toxins system(Khetrapal et al., 2015).
We tried to increase the selection stringency by combing the two counter-selection marker genes. At first, CDS of lysis gene Ewith a consensus RBS sequence (AAGGAGATATACAT) and gentamycin resistance gene GmR were inserted immediately behind the stop codon of kil gene at ack locus of E. coliMG-10. In this constructed strain CWE-3, kil and E were expressed as a bi-cistron under the control of promoter pL and repressor cI857 (Fig. S2). We named this combining counter-selection cassette askil-sd-E . Then the selection stringency of these counter-selection systems was compared. At the ack locus, selection stringency frequency of E is 2.7×10−7, several fold lower than that of kil(8.7×10−7). While selection stringency frequency ofkil-sd-E was significantly decreased to 4.9×10−8 (Fig. 5A), which is very close to the best reported inducible toxins system.
In consideration that insertion sites may influence stringency, another non-essential gene locus araB were selected for the further analysis. Selection stringency frequency of E(2.9×10−6) and kil (2.1×10−6) are almost at the same level. But selection stringency frequency ofkil-sd-E dropped sharply to 3.2×10−8 at this locus, about 65- to 90- fold lower than the above two counter-selection system (Fig. 5B). This result hinds that co-expression of kil andE in the form of bi-cistron can significantly increase their selection stringency in E. coli to the degree to the best reported inducible toxins system.
To explore whether it also works in other bacteria, S. marcescenswas used for the subsequent research. At first,E-GmR and kil-sd-E-GmRwere amplified from plasmid pBBR-E and strain CWE-3 separately and then were used to substitute the CDS of araB to construct the goal strain MG-4A and MG-4B (Fig. S3). Sole expression of E or co-expression of kil and E were both under the control of AraC/PBAD in the two strains. After adding 0.4% arabinose into LB broth, both MG-4A and MG-4B can’t grow normally. In particular, the LB broth was very clear in the MG-4B cultivating tube 18h after inoculation, while slightly turbid cultivated bacteria can be seen in the MG-4A cultivating tube (Fig. 5C). These results indicate that E and kil-sd-E counter-selection marker under the control of AraC/PBAD function well in E. coli . It also hints that co-expression of kil and E is of better lethal effect than the sole expression of E .
AraC/PBAD-kil-sd-E-GmR cassette was then introduced into S. marcescens to insert into the CDS ofpigA . The goal strain was named as GY5. Selection stringency frequency of GY5 (1.4×10−7) is about 10- fold lower than that of GY4 (1.9×10−6) (Fig. 5D). The result hints that counter-selection system in GY5 should perform better than GY4. This was then verified by the ssDNA mediated mutation repair: compared with GY4 (Fig. 4A), no matter in the control group or in the ssDNA added group, fewer white colonies were shown on GY5 (Fig. 5E). When 1,000 bp T7 RNA polymerase gene fragment was used to substitute the AraC/PBAD-kil-sd-E-GmR in GY5, 6 out of 10 randomly selected colonies were correct (Fig. 5G). The ratio of correct recombinants is also higher than that in GY4 (Fig. 4C).
All these results show that co-expression of kil and E can elevate the selection stringency by orders of magnitude in multiple species and decrease the number of escaping colonies during seamless modification.