Restructured Lactococcus lactis strains with emergent properties constructed by a novel highly efficient screening system

Background After 2.83% genome reduction in Lactococcus lactis NZ9000, a good candidate host for proteins production was obtained in our previous work. However, the gene deletion process was time consuming and laborious. Here, we proposed a convenient gene deletion method suitable for large-scale genome reduction in L. lactis NZ9000. Results Plasmid pNZ5417 containing a visually selectable marker PnisZ-lacZ was constructed, which allowed more efficient and convenient screening of gene deletion mutants. Using this plasmid, two large nonessential DNA regions, L-4A and L-5A, accounting for 1.25% of the chromosome were deleted stepwise in L. lactis 9k-3. When compared with the parent strain, the mutant L. lactis 9k-5A showed better growth characteristics, transformability, carbon metabolic capacity, and amino acids biosynthesis. Conclusions Thus, this study provides a convenient and efficient system for large-scale genome deletion in L. lactis through application of visually selectable marker, which could be helpful for rapid genome streamlining and generation of restructured L. lactis strains that can be used as cell factories.


Introduction
With a crucial role in dairy and health industries, Lactococcus lactis is a GRAS (generally regarded as safe) microorganism [1]. Features such as lack of immunogenic lipopolysaccharides and secretion of only one major protein [2] make L. lactis the commonly used microorganism in traditional food fermentation. Nowadays, with the development of whole genome sequencing and functional genomics technology, abundant data, including whole genome sequence and metabolic pathway of L. lactis, are available, which allow using L. lactis as an "efficient cell factory" for recombinant protein production and secretion [2]. However, an efficient genetic engineering system for L. lactis is still missing.

Open Access
Microbial Cell Factories *Correspondence: qiaomq@nankai.edu.cn 1 Key Laboratory of Molecular Microbiology and Technology, College of Life Sciences, Ministry of Education, Nankai University, Room 301, Tianjin, China Full list of author information is available at the end of the article [20] was used to construct knock-out vectors, which was subsequently transformed into L. lactis and replaced the target gene with cassette lox66-P 32 -cat-lox71 through double-crossover recombination. Then, temperaturesensitive plasmid pNZTS-Cre [19] for Cre recombinase expression was transformed into the mutants, which integrated with the lox66-P 32 -cat-lox71 cassette. The Cre recombinase excised the cat gene through the two lox sites (lox66 and lox71) and generated a double-mutant loxP site (lox72), which displayed strongly reduced recognition by Cre recombinase [20], finally, pNZTS-Cre was cured through the shift of temperature [19]. However, screening of the second-crossover recombination through replica plating method is laborious and time-consuming.
In the present study, a convenient gene deletion system was established by replacing plasmid pNZ5319 with a new plasmid containing a selectable marker P nisZ -lacZ. This enabled identification of deletions through visual screening, which facilitated quicker and easier detection of deletion of genes in L. lactis. With this system, two large nonessential DNA regions (L-4A and L-5A) accounting for 1.25% of the genome were selected and deleted stepwise in L. lactis 9K-3 [14] with high efficiency, and ultimately, five large nonessential DNA regions deletion mutant L. lactis 9K-5A with 3.24% genome reduction was constructed. To explore the genetic potential of mutants, the whole genome of L. lactis 9K-5A was sequenced. Comparison of physiological traits and transcriptome analysis revealed that L. lactis 9k-5A outperformed the wild strain in several physiological traits assessed and exhibited much higher expression of genes involved in routine metabolism.

Construction of vector pNZ5417
Plasmid pNZ5417, based on pNZ5319, was constructed by replacing ery with lacZ gene under the control of nisin-inducible promoter P nisZ , which produced blue colony on LB plate containing chloramphenicol and X-gal LB-CX (Fig. 1d).

Evaluation of the new gene deletion system in L. lactis NZ9000
To evaluate the feasibility of pNZ5417 for large-scale gene deletion in Lactococcus strains, two large nonessential DNA regions, L4A and L5A, were successfully deleted stepwise in L. lactis 9k-3. The distribution of L4A and L5A throughout the genome is indicated in Fig. 2a. The genetic organization of the two deleted DNA regions is shown in Fig. 2b. A detailed description of genes included in the L4A and L5A regions is provided in Additional file 1: Table S1. These two DNA regions formed approximately 1.25% of the L. lactis 9k genome, as shown in Table 1.
To prove the higher deletion efficiency of plasmid pNZ5417, L4A was also deleted with plasmid pNZ5319 (Fig. 3). The deletion mutants were generated with both the deletion plasmids through double-crossover recombination and the following three steps were executed to obtain genes deletion mutants: (i) the deletion vector was constructed and the first-crossover recombination was accomplished; (ii) the cells were cultured and secondcrossover recombination was achieved; and (iii) chloramphenicol resistance marker was deleted, and after it, plasmid pNZTS-Cre was cured. When compared with pNZ5319, 65% of the time was saved at the screening of second-crossover recombination, pNZ5417 gene deletion system allowed mutants screening based on color change, which is convenient and time-saving compared to the replica plating method.

Assessment of growth profiles and transformability of mutants
The growth profiles of Lactococcus strains (L. lactis 9k, L. lactis 9k-4A, L. lactis 9k-5A) were monitored by recording the OD 600 in GM17, respectively. The growth curves are shown in Fig. 4a (GM17). The results revealed that both the mutants grew much faster in the exponential phase, and reached the lag growth phase 1 h earlier than the parent strain. The doubling time of the two mutants proved to be much shorter than that of L. lactis 9k in the exponential growth phase ( Table 1). As reported before, genome evolution should promote an enlarged genome size by incorporating non-essential accessory genes [21,22], which might be disadvantageous for the growth fitness, given the additional cost for the replication and expression of the newly acquired sequences. After genome reduction, accumulative loss of dispensable genomic sequences like pseudogenes, phage/IS and unknown function genes contributes to bacterial growth in a dose-dependent manner [23]. Thus, we suppose the losing of unknown nonessential genes cased faster growth of mutants.
The electroporation efficiency of all the strains was measured by electroporating a small supercoiled plasmid, pNZ8048, into the cells. As shown in Fig. 4b, both mutants L. lactis 9k-4A (1.70-fold, P < 0.001) and L. lactis 9k-5A (2.45-fold, P < 0.001) exhibited higher electroporation efficiency than the parent strain (1.10 × 10 6 /μg plasmid DNA). GyöRgy et al. [7] reported that removal of external structures and unknown deoxyribonuclease or restriction system or activation of an unknown DNA uptake factor could affect the recovery of transformants. Considering the results of transcriptome analysis (Additional file 2: Fig. S1), we suppose deletion of unknown genes affected the cellular component including membrane composition, and altered the ability of mutants to receive exogenous DNA.

Assessment of mutants' phenotype
Extensive fermentation phenotype analyses of L. lactis 9k, L. lactis 9k-4A, and L. lactis 9k-5A were conducted using the phenotype microarrays to explore the physiological difference between the wild and mutant Lactococcus strains. All of the substrates that the mutants consumed were significantly different from those of L. lactis 9k. The result shown in Fig. 4c revealed that L. lactis 9k-4A can efficiently metabolize 12 carbon sources, particularly, α-cyclodextrin, β-cyclodextrin, and maltose; while L. lactis 9k-5A effectively metabolized 19 carbon sources, among which metabolism of α-cyclodextrin, β-cyclodextrin, maltose, maltotriose, and adenosine was 4.9-, 10.8-, 5.1-, 3.2-, and 4.6-fold higher than that of wild strain, respectively. We suppose this make it possible for mutants to utilize more carbon sources as the sole carbon Fig. 1 Schematic representation of deletion plasmid pNZ5417 construction. a Scheme of pNZ5319; b color selectable marker gene cassettes (P nisZ -lacZ); c scheme of pNZ5417; d chromogenic reaction of pNZ5417 in E. coli on LB medium containing chloramphenicol and X-gal; e chromogenic reaction of L. lactis NZ9000 harboring pNZ5417Δ L4A in M17 medium containing chloramphenicol, X-gal, and gradient nisin source, especially α-cyclodextrin and β-cyclodextrin, as a member of oligosaccharide, they are much easier to get and cheaper carbon source than glucose. In contrast, both the mutants showed poor capacity to metabolize d-galacturonic acid and 3-methyl-d-glucose, with L. lactis 9k-5A losing its ability to metabolize d-galacturonic acid.

Genome sequencing and analysis of L. lactis 9k-5A
The genome of mutant L. lactis 9k-5A was sequenced by Shanghai Majorbio Bio-pharm Technology Co. (Shanghai, China) using the Illumina MiSeq platform. As shown in Additional file 3: Data S1A, the final assemble consisted of 91 scaffolds with a total size of 23,33,697 bp and 35.61% G+C content, including 66 large scaffolds with the largest scaffold comprising 396,482 bp. Furthermore, 2390 genes with a total length of 1,985,466 bp were predicted and annotated. The putative replication origins of L. lactis 9k-5A were localized by GC skew (Additional file 4: Data S2A). The 66 large scaffolds of L. lactis 9k-5A were arranged (Additional file 5: Data S1B) in the order of genome sequence of L. lactis NZ9000, and then subjected to BLAST (National Center for Biotechnology Information; https ://blast .ncbi.nlm.nih.gov/ Blast .cgi?PROGR AM=blast n&PAGE_TYPE=Blast Searc h&LINK_LOC=Multi Senso r). Sequence alignment indicated that the five nonessential DNA regions L1, L2, L3, L4A, and L5A were successfully deleted (Additional file 6: Data S2B).

Discussion
With the rapid development of metabolic engineering, genome editing in bacteria, including industrial microorganisms such as E. coli [4,24] and B. subtilis [25], has provided significant benefits. Xin et al. [26,27] reported a single-plasmid genome editing system in lactic acid bacteria, which offered convenient and easy-to-use Overall scheme for L4A deletion using two systems in L. lactis 9k. (1) Plasmid pNZ5319ΔL4A or pNZ5417ΔL4A was first loaded into L. lactis 9k-3 cells and the recombinants were selected on M17 plates supplemented with 5 μg/mL chloramphenicol (Cm); for pNZ5417ΔL4A, 40 μg/mL X-gal and 10 IU/mL nisin were also added; (2) recombinants harboring plasmid pNZ5319ΔL4A or pNZ5417ΔL4A were cultured in M17 medium supplemented with 5 μg/mL chloramphenicol for generations, and positive mutants with successful L4A deletion were selected by replica plating method (for "pNZ5319/pNZTS-Cre gene deletion system") or color change (for "new gene deletion system"); (3) deletion of the chloromycetin resistance marker and elimination of temperature-sensitive plasmid pNZTS-Cre [19] genome-editing tool for metabolic engineering in Lactobacillus casei. Guo et al. [17] established a rapid tool for genomic engineering by combining ssDNA recombinants with improved CRISPR/Cas9 counter selection, and achieved seamless genomic DNA deletions (50/100 bp) in L. lactis. However, sequential deletion of multiple genes and large-scale genome in L. lactis is still a time-consuming and laborious process. In the present study, we proposed a convenient system for sequential generation of combinatorial genome deletions in L. lactis.
Counter selection method based on homologous recombination is a convenient and efficient technique for L. lactis genome streaming [28,29]. Nevertheless, the percentage of revertant mutations (60-92%) is much higher in double-crossover mutants [28], which makes the whole process of gene deletion much laborious. In our previous study, a two-plasmid (pNZ5319 and pNZTS-Cre) based gene deletion system was established with 100% correct deletion efficiency [14,19]. However, laborious procedures were still needed to screen the second-crossover recombination through replica plating method (Fig. 3, Additional file 15: Table S4).
In the present study, the screening time decreased by 38% with the developed gene deletion system (Additional file 15: Table S4). Our proposed system comprised a visually selectable marker lacZ, which was under the control of inducible promoter P nisZ [30]. While the chromogenic reaction became more significant with the induction of Fig. 4 Characterization of lactococcal strains. a Growth profile analysis with M17 medium; b comparison of electroporation efficiency. Data showed as mean ± SD and compared by t-test, ***P < 0.001. c Assessment of extensive fermentation phenotype of the mutants nisin, the presence of original constitutive promoter P 32 in "NZ9000 and MG1363 harboring pNZ5417ΔL4A" turned the cells blue in GM17 medium containing chloramphenicol and X-gal without nisin induction (Fig. 1e, Additional file 16: Fig. S5). Therefore, the use of this plasmid is not limited to nisin-controlled gene expression (NICE) system. Two large nonessential DNA regions (L4A and L5A) of the L. lactis NZ9000 genome were deleted sequentially with our developed gene deletion system, and subsequently, a five large nonessential DNA regions (3.24% of the genome) deletion mutant was constructed. When compared with the parent strain, the two mutants, L. lactis 9k-4A and L. lactis 9K-5A, showed some good phenotypic changes, including better growth characteristics and transformability. The capability of the strains in metabolizing 95 carbon sources was compared using GP2 MicroPlate, which revealed that L. lactis 9k-4A and L. lactis 9k-5A had better capacity to metabolize 12 and 19 carbon sources, respectively. The results of transcriptome analysis indicated that 245 genes were upregulated and 93 genes were downregulated in L. lactis 9k-5A. The selected 93 DEGs showed significant enrichment of KEGG pathway, which indicated a much higher expression of malL, malP, malD, mdxF, and malX genes involved in the metabolism of maltose and transformation of maltotriose, suggesting that L. lactis 9k-5A had better ability to metabolize maltose and maltotriose, similar to the results of GP2 MicroPlate. Besides, genes involved in the pathway of histidine, valine, and isoleucine biosynthesis and some other pathways were significantly upregulated in L. lactis 9k-5A, implying that this mutant could be employed as a possible industrial cell factory for the production of these three amino acids.

Conclusion
To the best of our knowledge, this study is the first to introduce inducible visually selectable marker P nisZ -lacZ into L. lactis NZ9000 gene deletion system with improved efficiencies of 38% in achieving gene deletion mutants, which will save much more time in genome reduction. By using this system, two nonessential DNA regions were deleted sequentially in L. lactis. Our main contributions, in addition to the improved gene deletion system, was the final genome-streamlined mutant L. lactis 9k-5A exhibited good phenotypic changes, including better growth characteristics, transformability, carbon metabolic capacity, and biosynthesis of amino acids. The results of this study indicated that further genome refinements and reductions in L. lactis could eventually generate a significantly simplified strain that could contribute to broadening the use of this bacterium.

Bacterial strains, plasmids, and culture conditions
The strains and plasmids used in this study are listed in Table 2. L. lactis was grown at 30 °C under static condition in GM17 medium supplemented with 0.5% (w/v) glucose. E. coli DH5α cells were used as cloning host and grown aerobically at 37 °C in LB medium (1% tryptone, 0.5% yeast extract, and 1% NaCl; the solid medium contained 1.5% agar). Antibiotic selection was used when appropriate: for E. coli (per mL), 150 μg of erythromycin and 15 μg of chloramphenicol were employed and for L. lactis (per mL), 5 μg of erythromycin and 5 μg of chloramphenicol were applied. X-gal was used at a concentration of 80 μg/mL and nisin was utilized at a concentration of 10 IU/mL.

DNA manipulations and chemicals
DNA marker, T4 DNA ligase, restriction enzymes, and DNA gel extraction kit were purchased from Takara (Dalian, China). The PCR product purification kit, firststrand cDNA synthesis kit, and SYBR Green RT-qPCR kit were obtained from Thermo Fisher Scientific (Waltham, USA). The commercial X-gal and nisin were bought from Sigma-Aldrich (St. Louis, USA). L. lactis plasmid DNA, chromosomal DNA, and total RNA were isolated by using Qiaprep spin kit (small scale) following manufacturer's instructions. PCR was performed with Phusion enzyme (Thermo Fisher Scientific, Waltham, USA). Primers were synthesized by BGI (Beijing, China) and the corresponding sequences are listed in Additional file 17: Table S5. PCR products and plasmids were sequenced by GENEWIZ service (Hangzhou, China). The competent E. coli DH5α cells were purchased from Takara (Dalian, China) and transformed by CaCl 2 procedure [31]. Recombinant plasmids were introduced into L. lactis by electroporation according to the method described earlier [32].

Construction of vector pNZ5417
The plasmid pNZ5417 (Fig. 1c), containing lacZ gene under the control of nisin-inducible promoter P nisZ [30], was constructed from pNZ5319. Promoter P nisZ and lacZ gene were obtained from L. lactis N8 with primer pairs P nisZ -F/R and LacZ-F/R, combined by overlap PCR (Fig. 1b), and digested with BglI-SphI, and replaced the ery gene of pNZ5319 to generate pNZ5417.

Feasibility of new gene deletion system in L. lactis NZ9000
To evaluate the new gene deletion system, we deleted the large nonessential DNA region L4A in L. lactis 9k-3 [14] by using pNZ5319/pNZTS-Cre [19] and pNZ5417/ pNZTS-Cre gene deletion system, respectively, and successfully constructed L. lactis 9k-4A. Gene knock-out vectors pNZ5319Δ L4A and pNZ5417Δ L4A were generated with the primer pairs L4A-UP-F/R and L4A DP-F/R. Vector pNZ5319Δ L4A was transformed into L. lactis 9k-3, and the deletion of L4A mutant with pNZ5319/ pNZTS-Cre system was achieved as described earlier [19]. pNZ5417Δ L4A was transformed into L. lactis 9k-3, and single cross-over recombinant was selected at 30 °C on GM17-CXN solid medium containing chloramphenicol, X-gal, and nisin. The single cross-over recombinants were sub-cultured at 30 °C in GM17-CXN liquid medium several times, and the overnight cultures were diluted and plated on GM17-CXN medium at 37 °C until most colonies turned blue. The white colonies were selected and identified by primer pairs L4A Int-F/R and L4A Out-F/R. After single colony isolation, the cat selectable marker was excised as described previously, and the deletion mutants were tested by PCR with appropriate primers. The gene knock-out vector pNZ5417Δ L5A was constructed with primer pairs L5A-UP-F/R and L5A-DP-F/R, and the nonessential DNA region L5A was deleted pNZ8048 Cm r [42] in L. lactis 9k-4A with pNZ5417 gene deletion system to obtain five large nonessential DNA regions deletion mutant L. lactis 9k-5A.

Analysis of growth profiles
Lactococcus lactis 9k, L. lactis 9k-4A, and L. lactis 9k-5A were cultured to OD 600 of 0.8 in GM17 medium and diluted to OD 600 of 0.4. Then, 2 µL of the diluted cultures were reinoculated into 200 µL of GM17 medium in shake-flasks. The growth profiles were monitored by measuring OD 600 for 10 h at 30 °C by using a Bioscreen machine (Lab-systems, Helsinki, Finland) [33]. The experiment was repeated three times.

Measurement of electroporation efficiency
Electrocompetent cells of all the strains were prepared by the method of Holo [32], and 2.5 µg of plasmid pNZ8048 DNA were added to 0.1 mL of competent cells. After electroporation, the cells were cultured in plates containing 15 µg/mL chloramphenicol for the selection of chloramphenicol-resistant transformants. The transformants were enumerated after 2 days of incubation at 30 °C, and the experiment was repeated three times.

Microarray analysis of mutants' phenotype
The metabolism of the wild strain and mutants was examined with GP2 MicroPlate ™ using phenotype microarrays system (Biolog, California, USA). Sample preparation and assays were conducted according to the manufacturer's instructions. In brief, Lactococcus cells on the surface of solid medium were collected using cotton swab and suspended in inoculating fluid (0.40% NaCl, 0.03% Pluronics F-68, and 0.02% Gellan Gum) (Biolog, California, USA). The cell density was equalized, and 150 µL of the cells suspension were pipetted into GP2 plates with various substrates, respectively. Then, the plates were incubated in OmniLog ® instrument (Biolog, California, USA) at 30 °C for 24 h. The data were automatically recorded every 30 min, and were analyzed by OL-OM software (version 3.0) (Biolog, California, USA).

Sequencing and analysis of L. lactis 9k-5A genome
The genomic DNA of L. lactis 9k-5A was extracted and purified, and then quantified using Nanodrop 2000 spectrophotometer (Thermo Scientific, USA). The L. lactis 9k-5A genome was sequenced by Shanghai Majorbio Bio-pharm Technology Co. (Shanghai, China) using Illumina MiSeq platform with a paired-end library. Following trimming and merging, the reads were assembled de novo using SOAP denovo V2.04 [34]. Open reading frames (ORFs) were predicted using Glimmer 3.02 program [35], and annotated by comparison with NCBI-NR and KEGG databases using BLASTp (BLAST 2/2/28+).

Transcriptome analyses of mutants
The total RNA of L. lactis 9k, L. lactis 9k-4A, and L. lactis 9k-5A strains cultured in GM17 to an OD 600 of 0.8 was extracted and purified by TRIzol kit (Promega USA), sequenced on Illumina sequencing platform, and analyzed by Genedenovo Biotechnology Co., Ltd (Guangzhou, China). Each sample was prepared in triplicate. The transcription of genes malD, purF, galE, and galK was measured through quantitative real-time PCR (RT-qPCR) to recheck the transcriptomic data. All RT-qPCR reactions were repeated independently three times. Data analysis was conducted by using comparative CT (2 −∆∆CT ) method with the housekeeping gene rpoB [37] as control. Transcription with more than twofold changes was regarded as significant difference [38].

Statistical analysis
The data obtained are reported as mean ± standard deviation (SD). The difference between two groups was compared by t-test with P < 0.05 considered as significant.