One-plasmid based CRISPR-Cas9 editing for gene deletion in Lactococcus lactis

Lactococcus lactis strains are promising cell factories and delivery vehicles of plasmid DNA and recombinant protein for therapeutic applications. However, the limited yields of recombinant molecules obtained with these bacteria limits their wide applicability. Genome engineering of this host may solve the problem. However, the current genome editing toolbox available for L. lactis is either too laborious or incapable of large edits, limiting the scope of strain editing experiments. In this work, the basis for a one-plasmid CRISPR-Cas9 based genome editing plasmid was developed and tested. The new plasmid (pTCas9dO) adapted from the pKCcas9dO plasmid was used to delete 657 bp of the lactococcal nuclease nth of L. lactis subsp. lactis LMG19460, with the aim of improving yield and quality of plasmid DNA replicated in this strain. Although deletion mutants were successfully generated, plasmid curing was unsuccessful. Thus, further modications are required before the plasmid is truly applicable for genome editing experiments. Unexpectedly, the generated deletion mutants generated a roughly 40% decrease in plasmid yield alongside with a decrease in the quality of produced pDNA. Plasmid copy number of in Lactococcus lactis is increased by modication of the repDE ribosome-binding site.


Introduction
Over the last decades there has been an increasing interest in the use of Lactococcus lactis as cell factories. Their non-pathogenic nature, coupled to their extensive use in the food industry, make these bacteria safer alternatives to produce active biomolecules than the more widely used Escherichia coli. Although E. coli delivers high yields and is easy to use, it produces lipopolysaccharides, requiring costly and timely downstream processes to avoid co-puri cation with the nal product (Wakelin et al. 2006, Wicks et al. 1995, Schneier et al. 2020. The increased safety pro le has led to a particular focus on L. lactis as both producers and delivery vehicles of biopharmaceuticals, with particular focus on mucosal vaccination with plasmid DNA (pDNA) (Asensi et al. 2013, Wells et al. 1993, Pontes et al. 2003, Pereira et al. 2015, Bermúdez-Humarán et al. 2004. The safe use of L. lactis in this context has been demonstrated with a genome edited human IL-10 producing strain that was tested in clinical trials for the therapy of Crohn's disease (Braat et al 2006).
However, the usually disappointing yields of recombinant molecules produced by L. lactis strains undermine their potential applicability. An improved productivity of high-quality pDNA, for example, would make these bacteria better vectors for mucosal vaccination using DNA vaccines, and would also improve yields in the context of recombinant protein production (Duarte et al. 2021). Genome editing might provide a solution, as demonstrated for a few selected genes. For example, the effects of the ybdD (Morello et al. 2012, Nouaille et al. 2004, clpP and hrtA (Cortes-Perez et al. 2006) genes on recombinant protein production were investigated, with knockout strains exhibiting higher protein yields compared to the wild-type. Larger scale genome editing was also successfully accomplished in L. lactis, as illustrated by the development of NZ9000 genome reduced strains (Liu et al. 2019, Zhu et al. 2017, which exhibited an overall increased robustness and substantially enhanced heterologous protein production. Despite this, current genome editing attempts of L. lactis strains are still limited by the tools available. The reduced genome strains mentioned above, for example, were developed using a Cre-loxP genome editing system (Zhu et al. 2015). Although robust and capable of driving knockouts over 20 kb long, the system is laborious and time-consuming, a feature that prompted the development of improved and easier to use versions (Liu et al. 2019). Faster and more e cient genome editing plasmids exist in L. lactis like the pLRecT  dual plasmid system, which allows for highly e cient and expedite editing of Lactococcus. However, it is incapable of knockouts roughly 150 bp in size, which limits its use in larger scale genome editing projects. Additionally, both systems employ the use of multiple plasmids and/or ssDNA donors for recombination of the targeted gene, reducing their effectiveness in strains di cult to transform.
CRISPR-Cas9 based genome editing systems provide quick and effective generation of mutants, with the capacity for large-scale edits, as well. These systems typically function by targeting the Cas9 protein to a particular gene of interest and inserting a double strand DNA break at the targeted site. This damage leads to the recruitment of the host's homology-directed repair (HDR) mechanisms to the targeted site.
Inclusion of the anking regions of the target in trans, typically in a plasmid, allows the host's HDR mechanisms to substitute the affected region with the repair template. This can be applied to drive deletions, insertions and base edits through careful design of the repair template. Additionally, as chromosomally located DNA breaks tend to be fatal if not repaired, Cas9 activity also provides selection for cells carrying the desired mutation (Vercoe et al. 2013). As the Cas9 protein requires only the protein itself and a target speci c single-guide RNA (sgRNA) to work, CRISPR-Cas9 based genome editing systems can be incorporated into a single plasmid, providing a relatively simple system for a relatively quick genome editing without limitations to the scale of possible edits. Applications employing multiplexed single guide RNA (sgRNA) expression were successfully employed in E. coli (Huang et al. 2019) and Streptomyces coelicor (Huang et al. 2015) to drive deletions of over 80 kb in a single genome editing event, for example. However, to the best of our knowledge, CRISPR-Cas9 based genome editing systems for L. lactis based on a single plasmid have not be developed so far. Development of such a system would provide a very effective tool for large-scale genome editing experiments in L. lactis, thus contributing to facilitate the engineering of these bacteria.
In this paper we set out to develop a one-plasmid based CRISPR-Cas9 system for the editing of L. lactis. The new plasmid is derived from the pKCcas9dO plasmid developed previously by Huang et al. (Huang et al. 2015) and modi ed by our group (Duarte 2018) and was used to generate deletion mutants of the lactococcal nuclease nth of L. lactis subsp. lactis LMG19460. The 657 bp nth gene, which codes for a nuclease with a predicted molecular weight of 24.5 kDa, was selected as a target for deletion with the speci c aim of obtaining L. lactis mutants more appropriate for plasmid replication. Its homologue in E. coli codes for Endonuclease III, a repair enzyme with redundant activity (Cunningham et al. 1985, Saito et al. 1995. We reasoned that by deleting nth, the yields and quality of plasmid DNA replicated in L. lactis could potentially be improved by limiting non-speci c digestion.

Methods
Bacterial Strains, Plasmids and Growth Conditions E. coli DH5α cells were used as the host for all plasmids during plasmid construction. L. lactis subsp. lactis IL1403 (Chopin et al. 1984) was used as the source of the P XylT promoter and L. lactis subsp. lactis LMG19460 was used as the nal recipient of the constructed plasmids and the target of the nth gene deletion. All strains used in this work are listed in Table 1.

INRAE, France
The pTRKH3 plasmid has been described before (O'Sullivan et al. 1993) and was used as the new vector backbone for the constructed plasmids while the pKCcas9dO_nth_apra (Duarte 2018) plasmid was used as the source of the Cas9 gene sequence and the nth gene speci c single-guide RNA and repair template. The pKCcas9dO_nth_apra (Duarte 2018) plasmid is derived from the pKCcas9dO plasmid (Huang et al. 2015) and contains a S. coelicor optimized Cas9 (ScoCas9) gene sequence under the control of the thiostrepton inducible P TipA promoter, a constitutive promoter named J23119, downstream of which is an nth gene speci c sgRNA and the minimal replicon of the thermosensitive plasmid pSG5 (Muth et al. 1988, Muth et al. 1989). Additionally, the plasmid also contains an nth speci c repair template, or homologyarms (HAs). This constitutes two regions, 1,028 and 955 bp in size, that share homology to the anking regions of the nth gene sequence in the L. lactis LMG19460 genome. All plasmids used in this work, alongside their relevant characteristics, are listed in Table 2. when cells were transformed with thermosensitive plasmids (pKCcas9dO_nth_apra). When needed, Erythromycin or Apramycin were added at 500 µg/mL. L. lactis subsp. lactis LMG19460 and IL1403 cells were grown in M-17 medium (Sigma-Aldrich) supplemented with either 20 g/L of D(+)-glucose (GM-17) or 20 g/L of D(+)-xylose (XM-17), at 30°C, 100 rpm. On solid media, cells were incubated on either modi ed Lactic Agar (Elliker et al. 1956) (5 g/L lactose, 20 g/L tryptone, 5 g/L yeast extract, 4 g/L sodium chloride, 1.5 g/L sodium acetate, 0.5 g/L Lascorbic acid and 15 g/L agar) or SR medium (Holo et al. 1989) at 30°C. Erythromycin was added at 5 ng/µL, when needed.

General Cloning Procedures
Plasmid DNA was obtained from E. coli and L. lactis cells using the High Pure Plasmid Isolation Kit (Roche) and the Nucleospin Plasmid Puri cation Kit (Macherey-Nagel), respectively. L. lactis strains' genomic DNA (gDNA) was obtained using the Wizard Genomic DNA Puri cation Kit (Promega). The NZYGelpure Kit (NZYTech) was used for extraction of DNA fragments from agarose gels and clean-up of PCR and restriction reaction products. Sanger sequencing was done through Stabvida (Portugal).
All ampli cation reactions for cloning purposes were done using the KOD Hot Start DNA polymerase (Novagen), with the exception of the ampli cation reaction using the pTRKH3:pXylT:ScoCas9_F/R primer pair, which was done using the Supreme NZYLong Colourless Master Mix (NZYTech) polymerase. Ampli cation reactions for con rmation of the nth gene knockout were done using the NovaTaq Hot Start Master Mix (Novagen) enzyme and nth_conf_F/R primer pair. All primers used are listed in Table 3. Underlined bases represent the region of the primer which anneals exclusively to its complementary Gibson Assembly pair. Bases in bold represent the added restriction sites for the BsrGI enzyme while the bases in bold and italics represent the added restriction sites for the NcoI enzyme.
Restriction endonucleases used were DpnI (Thermo Fisher Scienti c), NcoI (Thermo Fisher Scienti c), BsrGI (Thermo Fisher Scienti c), BamHI (Promega) and HindIII (Promega) enzymes. All PCR products intended for ligation were incubated for 3 hours with DpnI enzyme. Ligation reactions were mostly done through the Gibson Assembly® Master Mix (New England Biolabs), except for the construction of the pTScoCas9 intermediary plasmid which was done using a T4 DNA Ligase enzyme (Promega). Ligation products were transformed into competent E. coli DH5α cells and plated onto LB agar with appropriate antibiotic.
All kits and enzymes were used in accordance with manufacturer's guidelines.
Chemically competent E. coli DH5α cells were prepared and transformed as previously described by Chung et al. (1989), with the following modi cations: competent cells were prepared from cultures at an optical density at 600 nm (OD600nm) of 0.95, transformation was done with 100 ng of plasmid per transformation and cells were given a 42°C heat shock, in a water bath, for 1 minute following the 4°C incubation period. Transformants were selected by plating in antibiotic supplemented LB agar.
L. lactis LMG19460 cells were made competent and transformed as previously described by Holo et al. (1989) with slight modi cations. Electroporation was done using a Gene Pulser (Bio-Rad Laboratories) with a Pulse Controller (Bio-Rad Laboratories) and cells were given 2 to 5 shocks of 10.0 kV/cm, 400 Ω and 25 µF. Transformants were selected by incubation in SR medium supplemented with erythromycin.

Induction of Cas9 Expression
Induction of the pTCas9dO plasmid was done as described previously by Miyoshi et al. (2004) with slight modi cations. Transformed L. lactis LMG19460 cells were grown overnight in GM-17 until con uency. These cultures were then pelleted and washed twice with fresh M-17 medium, to remove most traces of glucose, before being inoculated into XM-17, such that the starting OD 600nm of these new cultures was 0.1. After overnight growth, appropriate dilutions of the cultures were prepared and plated onto erythromycin supplemented Lactic Agar plates to isolate possible mutants. The e ciency of the plasmid for deletion of the nth gene was calculated as described previously (Song et al. 2017).
pTCas9dO Plasmid Curing Curing attempts of the pTCas9dO plasmid were done through extended incubation at higher temperatures and protoplast regeneration. For plasmid curing by extended incubation, pTCas9dO transformed Δnth LMG19460 cells were inoculated into erythromycin free GM-17 medium and incubated at 32°C, 37°C and 40°C for up to 120 hours, in the same culture. Appropriate dilutions were then plated onto SR medium for colony counting at 96 and 120 hours of incubation.
Longer incubation times, in both liquid and solid media, were also attempted. pTCas9dO transformed Δnth LMG19460 cells were grown in erythromycin free GM-17 medium for a total of 26 days, with cells being transferred to fresh GM-17 medium every day to ensure the culture's viability. Appropriate dilutions of the cultures were plated onto erythromycin free Lactic Agar plates before every transfer. On solid medium, pTCas9dO carrying deletion mutants were plated onto erythromycin free Lactic Agar plates. Every two days, of the obtained colonies, 6 were streaked onto new plates while another 50 were tested for loss of plasmid.
Curing attempted through protoplast regeneration was done as previously described by Mastrigt et al. .
Following all protocols, screening for loss of plasmid from the obtained colonies was done through replica plating onto Lactic Acid Agar, with and without erythromycin.
Quanti cation of pDNA yield and pDNA quality analysis pTCas9dO transformed wild-type and Δnth L. lactis LMG19460 cells were grown up to 48 hours in M-17 medium supplemented with 0.5% glucose. Samples were taken at late-exponential phase (OD 600nm of 1.6) and at 24 hours following start of incubation. The cell count was standardized for each sample by diluting enough culture to obtain 5 mL samples with an OD 600nm of roughly 1.6. These samples were pelleted and used for pDNA puri cation.
pDNA yield was assessed spectrophotometrically using a NanoDrop One (Thermo Fisher Scienti c) instrument while pDNA quality was assessed by comparing the relative intensity of the plasmid's isoforms in a 0.8% (w/V) agarose gel, loaded with 2 µg of pDNA.

Statistical analysis
Statistically signi cant differences in pDNA yield were determined using an unpaired, two-tailed t-test, performed using GraphPad Prism version 9.0.2 for Windows, GraphPad Software, San Diego, California USA, www.graphpad.com.

Plasmids in silico design
In previous work we used the pKCcas9dO (Huang et al. 2015) plasmid to generate pKCcas9dO_nth_apra, a plasmid that carries an nth gene speci c sgRNA and repair template (Duarte 2018). However, this plasmid was ineffective to knockout L. lactis, due to the inability of the thiostrepton-inducible promoter to successfully induce Cas9 expression. Additionally, the apmR resistance marker from pKCcas9dO plasmid is inappropriate for L. lactis LMG19460 as this strain is highly resistant to apramycin. We thus designed the new plasmid pTCas9dO using pKCcas9dO_nth_apra as a starting point by replacing the promoter for Cas9 expression for the xylose inducible and glucose repressible P XylT promoter (Miyoshi et al. 2004), and by introducing the widely used pTRKH3 plasmid as a new backbone. Key features of this new backbone include an effective erythromycin selection marker, a medium copy number Gram-negative origin of replication (p15A), and a high copy number Gram-positive pAMβ1 origin of replication. The tetracyclin resistance marker (tetR) was also excluded to decrease the size of the nal construct, since the erythromycin resistance marker (eryR) is su cient for effective selection in both E. coli and L. lactis. The nth gene speci c sgRNA and repair template, which target a 657 bp region of nth for deletion, were obtained from the pKCcas9dO_nth_apra plasmid. The desired components of each plasmid were ampli ed using primers designed for Gibson Assembly, except for the construction of the pTScoCas9 plasmid, which was ampli ed with primers containing added restriction sites for the NcoI and BsrGI enzymes and ligated through a T4 DNA ligase reaction. A schematic of the steps involved in the construction of the plasmid can be seen in Supplemental Figure 1, while the nal construct is shown in Figure 1. Figure 1 -Schematics of the nal constructed plasmids, pTCas9dO. Relevant restriction sites, for the BsrGI and NcoI (plasmid construction) and BamHI and HindIII (screening of transformants) enzymes were included. p15A -Gram-negative, medium copy number origin of replication; pAMβ1 -Grampositive, high copy number origin of replication; eryR -Erythromycin selection marker; pXylT Promoter -Xylose inducible P XylT promoter; J23119 -Constitutive promoter; ScoCas9 -S. coelicor codon-optimized Cas9 gene sequence; nth sgRNA -nth gene speci c sgRNA; nth HA 1-2 -nth gene speci c homologyarms or repair template, constituting the 1,028 (nth HA1) and 955 bp (nth HA2) fragments that share homology with the anking regions of the nth gene; Image prepared using SnapGene software (from Insightful Science; available at snapgene.com).

pTCas9dO construction
Firstly, the P XylT promoter was inserted into the new pTRKH3 vector backbone. The promoter was ampli ed from L. lactis IL1403 gDNA using the pXylT_F/R primer pair, while the vector backbone was ampli ed from the pTRKH3 plasmid using the pTRKH3_F/R primer pair. This ampli cation excluded the ribosomal binding site (RBS) of the P XylT promoter, as this region has an unfavourable G+C mol% content for effective ampli cation and Gibson Assembly ligation. The desired ampli cation products were puri ed from an agarose gel and ligated using Gibson assembly to obtain the pTXylT intermediary plasmid. Cells transformed with the ligation products were used for pDNA puri cation and the correct insert was con rmed by ampli cation using the pXylT_F/R primer pair (Figure 2a). The correct plasmid was successfully obtained and con rmed through sequencing.
The pTXylT plasmid was then linearized through ampli cation using the pTRKH3:pXylT_F/R primer pair while the Cas9 gene sequence, with the RBS of the P TipA promoter immediately upstream, was also obtained through ampli cation of the pKCcas9dO_nth_apra plasmid. This primer pair was designed to contain additional restriction sites for the BsrGI and NcoI enzymes, which allowed ligation of the Cas9 gene sequence immediately downstream of the P XylT promoter using a T4 DNA ligase, resulting in the pTScoCas9 intermediary plasmid. Screening of cells transformed with the resulting ligation products was done through a restriction pro le using the BamHI and HindIII enzymes (Figure 2a). Of the screened colonies, one (clone number 4 in Figure 2a) exhibited the expected pattern of 6,639 and 3,882 bp.
Finally, the sgRNA sequence for the nth gene, alongside the nth speci c repair template, were ampli ed from the pKCcas9dO_nth_apra plasmid using the sgRNA+HAs_F/R primer pair. This fragment was inserted into the pTScoCas9 plasmid, which was previously linearized through ampli cation using the pTRKH3:pXylT:ScoCas9_F/R primer pair. Ligation was done through Gibson Assembly, resulting in the pTCas9dO plasmid. The resulting transformants were screened by a restriction pro le using the BsrGI enzyme ( Figure 2b). One clone (clone 1 of Figure 2b) exhibited the expected pattern of 7,471 and 5,172 bp. The plasmid sequence was further veri ed through sequencing.
Figure 2 -1% (w/V) agarose gels of every screening step of pTCas9dO construction. All molecular markers used were NZYDNA Ladder III (NZYTech). (a) Restriction pro le of transformants obtained from pTScoCas9 ligation reaction using the BamHI and HindIII enzymes. Non-restricted samples (N) were ran alongside their restriction products (R). A positive control, composed of a previously obtained pTScoCas9 solution unusable due to mutation (C+), was included. One clone (4) exhibited the correct pattern of 6,639 and 3,882 bp. (b) Restriction pro le of transformants obtained from pTCas9dO ligation reaction using the BsrGI enzyme. A negative control (C-) was included using the pTScoCas9 plasmid. Non-restricted samples (N) were ran alongside their restriction products (R), and one clone (2) exhibited the correct restriction pattern of 7,471 and 5,172 bp.
The nth gene knockout was successfully obtained with the pTCas9dO plasmid The pTCas9dO plasmid was transformed into L. lactis LMG19460 cells and the resulting transformants were induced by incubation in XM-17 medium. The induced cultures exhibited a noticeable decrease of cell density 3 hours after induction (OD 600nm of 0.126±0.006 to 0.083±0.006 in 2 hours), most likely because of Cas9-induced cell death. The cultures maintained a lower cell density (OD 600nm of roughly 0.09) for over 24 hours following induction. Culture samples were plated after the cells were induced overnight. The resulting colonies were screened for the deletion by ampli cation of gDNA puri ed from individual colonies using the nth_conf_F/R primer pair (Figure 3).
Of the 23 colonies tested after pTCas9dO induction, 18 showed an observable amplicon. Among these, 8 exhibited the amplicon characteristic of the nth gene knockout (2,133 bp), while another 6 exhibited the wild-type amplicon (2,790 bp). The remaining 4 colonies appeared mixed as they exhibited both amplicons. An e ciency of 66.7% was calculated for the deletion of the 657 bp nth gene sequence using the pTCas9dO plasmid.
General Curing Procedures Were Incapable of Curing the pTCas9dO Following knockout of the nth gene, attempts were the made to cure the plasmid pTCas9dO from the successful clones. Deletion mutants carrying the pTCas9dO plasmid were subjected to extended incubations at 32°C, 37°C and 40°C in erythromycin free GM-17 medium for a minimum of 96 hours and up to 120 hours. Samples of these cultures were taken at both the 96 and 120 hour-mark for all temperatures and plated onto erythromycin free SR medium. Colonies were only successfully obtained for the plates grown at 32°C, with only one colony obtained after 120 hours of incubation. Plates obtained from cultures grown at 37°C and 40°C did not exhibit any growth when plated undiluted onto SR medium plates. Of the plates obtained from the cultures incubated at 32°C for 96 hours, a total of 150 colonies were patch plated to check for loss of plasmid, alongside the one colony obtained from the cultures incubated at 32°C for 120 hours. Unfortunately, all tested colonies were capable of growing in erythromycin supplemented medium, implying the pTCas9dO was still present.
Curing was also attempted by successively incubating the cells in erythromycin free GM-17. pTCas9dO bearing deletion mutant cells were incubated in GM-17 medium, with the culture being diluted into fresh medium daily. Before each transfer, appropriate dilutions of the culture were plated onto erythromycin free Lactic Agar medium. Fifty colonies for each of the plates were then patch plated onto Lactic Agar plates, with and without erythromycin, to check for loss of plasmid. A similar strategy was attempted with solid medium, wherein six colonies of pTCas9dO carrying deletion mutant cells were successively streaked onto erythromycin free Lactic Agar plates. Every two days, 50 colonies were patch plated onto Lactic Agar plates, with and without erythromycin, to check for loss of plasmid, while another 6 colonies were chosen and transferred to new Lactic Agar plates. The 1,300 screened colonies obtained from incubation in GM-17 medium, and the 750 screened colonies obtained from incubation in Lactic Agar were all capable of growing in erythromycin supplemented medium.
A nal attempt at pTCas9dO curing was done through protoplast regeneration. Plasmid bearing cells were washed twice with 30 mM Tris-HCl buffer, before being incubated at 37°C, for 1 or 3 hours, in lysis buffer. The resulting cell suspension was then plated onto SR medium to allow protoplast regeneration for up to 72 hours. The resulting colonies were then patch plated onto erythromycin supplemented and erythromycin free Lactic Agar. In total, 150 colonies were tested for cells incubated in lysis buffer for 1 hour and another 50 were tested for cells incubated for 3 hours. None of the tested colonies exhibited loss of erythromycin resistance.
Deletion of the nth gene results in a decrease of total pDNA produced with no noticeable differences in plasmid quality Although we were unable to cure pTCas9dO from the nth deletion mutants, we proceeded to analyse the effects of the deletion on pDNA productivity and quality by examining the replication of the pTCas9dO.
To do this, wild-type and Δnth mutant LMG19460 cells carrying the plasmid were grown in GM-17 medium until late-exponential phase (OD 600nm of roughly 1.6) and stationary phase (24 hours of growth) wherein samples of each culture were taken and used to purify pDNA. Monitoring of cell growth throughout this incubation (Supplemental Figure 2) showed that there was a slight increase in both the speci c growth rate and nal biomass of the Δnth mutants when compared to the wild-type cells. Speci cally, the deletion mutants reached a speci c growth rate of roughly 0.69±0.06 and an average nal OD 600nm of 2.82, whereas the wild-type cells reaching a speci c growth rate of 0.56±0.32 and an OD 600nm of 2.45. pDNA isolated from both the wild-type and deletion mutants was quanti ed (Figure 4a).
At the late exponential phase, there was a signi cant decrease in pDNA yield (in µg/L of culture) for the deletion mutants of roughly 40 % when compared to the wild-type cells, while no signi cant difference was observed for the stationary phase. Furthermore, the relative intensity of the bands for the different isoforms of the plasmids appeared slightly fainter for the nth deletion mutants, regardless of growth phase (Figure 4b), implying an overall decrease in pDNA quality and quantity for the nth deletion mutants. Several unexpected bands appeared most notably the heavier of which is probably genomic DNA and a 4 kb apparent size fragment in the wild-type cells. As the strain used is a plasmid-free strain, it is possible that the observed extra bands are the result of undesired recombination events with the pTCas9dO plasmid. This should be an effect of the nth deletion as L. lactis LMG19460 cells transformed with other plasmids (pTRKH3) do not exhibit this behaviour.

Discussion
In this work we adapted a one-plasmid CRISPR-Cas9 based genome editing system, previously developed for use in S. coelicor (Huang et al. 2015), for use in L. lactis LMG19460. To the best of the authors knowledge, this is the rst time such a system has been applied in a single plasmid in L. lactis. Key features of the resulting 12 kbp pTCas9dO plasmid include the erythromycin resistance gene, the pAMβ1 origin and a CRISPR-Cas9 system. As a proof of concept, we designed the pTCas9dO plasmid with and RNA guide directed towards creation of a knockout mutant for the lactococcal nuclease nth. Successful mutant strains are expected to result in improved replication of pDNA both in terms of yield and quality.
The 657 bp nth gene knockout in L. lactis LMG19460 was successfully obtained with the pTCas9dO plasmid, with 66.7% of tested clones registering a gene editing event and 44 % of clones being pure deletion mutants. While these e ciencies were satisfactory, other genome editing systems used in Lactic Acid Bacteria have been shown to be capable of higher e ciencies. The Cas9 nickase based plasmid pLCNICK developed for Lactobacillus casei, for example, yielded deletions of similar size with e ciencies of up to 65 % before subculturing and 100% after sub-culturing (Song et al. 2017), while the previously mentioned pLRecT plasmid reached genome editing e ciencies of over 75% . Improving the expression of the Cas9 protein through codon-optimization or the substitution of the P XylT promoter with stronger ones like the widely used NICE system (Mierau et al. 2015) could lead to similarly high e ciencies in our plasmid. The pTCas9dO plasmid's relative simplicity allows for hassle-free generation of mutants, requiring only transformation of the plasmid, followed by induction of Cas9 expression.
However, the greatest advantage of our system is the use of a fully-functioning Cas9 protein to drive the knockout of the gene. This should allow the plasmid to drive large-scale deletions if two sgRNAs are employed, with each targeting one of the extremes of the desired region (Huang et al. 2015, Huang et al. 2019, allowing for similar deletion sizes as those demonstrated by Cre-loxP systems (Liu et al. 2019, Zhu et al. 2017), but faster, without employing multiple plasmids or requiring laborious screening of generated mutants. The capacity for multiplexed sgRNA's in our plasmid, however, still needs to be veri ed.
While the capacity for genome editing of the plasmid was demonstrated, its curing was unsuccessful. This likely due to the high stability and copy number of the pAMβ1 origin used (O'Sullivan et al. 1993) (45-80 copies). Nevertheless, curing should be possible given that a similar plasmid containing the same origin was removed by successive streaking in solid medium without antibiotic supplementation . Repeating the curing procedures in unbuffered medium could provide a solution as the rapid acidi cation of the medium has been shown to increase plasmid instability (Sinha 1989), but this might make plasmid curing too time-consuming and unreliable. The use of Gram-positive thermosenstive replicons would be more likely to succeed. With this in mind, we are currently modifying the pTCas9dO plasmid by substituting both of its origins of replication with the thermosensitive version of the origin of replication of the lactococcal plasmid pWV01 (Maguin et al. 1992). This origin has been used in plasmids in L. lactis strains for insertional mutagenesis (Maguin et al. 1996) and is readily curable in just 8 hours, which would facilitate the application of the pTCas9dO plasmid for iterative generation of mutants.
Although di cult to ascertain due to the presence of unexpected DNA in the generated deletion mutants, the effects of the nth deletion on plasmid yield and quality were also veri ed, with the generated mutants presenting roughly half the yield of pDNA when compared to those of the wild-type cells. Plasmid quality also appeared to be slightly lower for the deletion mutants as the supercoiled isoform of the plasmids appeared less intense when loaded onto an agarose gel when compared to the wild-type cells, although the difference was very slight. The nth gene is identi ed as an Endonuclease III homologue (through sequence homology), which is a nuclease responsible for excision of bases damaged by oxidation or radiation in E. coli (Cunningham et al. 1985, Matsumoto et al. 2001, Saito et al. 1997. It is possible that the decreased yield and quality of pDNA replicated in the nth deletion mutants is due to the accumulation of unrepaired bases. These results are interesting as there was a slight but noticeable increase in both speci c growth rate and nal biomass of the nth deletion mutants when compared to the wild-type cells. It would be expected that the accumulation of DNA damage due to lack of repair would lead to cell death or loss of growth. It is possible that the increase in growth is due to the lower yield of pDNA caused by the mutation, leading to a smaller metabolic burden on the cells. The likely increase in mutations caused by the lack of a DNA repair enzyme could have also led to unexpected mutations, and this behaviour could be due to a random mutation. With plasmid cured deletion mutants the results could be more accurately assessed. Additionally, a widely used plasmid, such as the pTRKH3 plasmid, could be used to assess the pDNA yield and quality more accurately and ensuring more comparable results between different mutants. This would also allow for more accurate estimation of plasmid yield and quality as it would avoid the presence of the undesired fragments observed in the agarose gels. Other methods should also be used to provide more accurate results. Real-time quantitative PCR targeting the pTRKH3 plasmid could be used to assess plasmid yield (Duarte et al. 2019), while uorescence (Levy et al. 2000) or chromatography (Abdulrahman et al. 2018) based methods could allow for more accurate estimation of the relative abundance of the produced plasmids isoforms.
In closing, we demonstrated that a one-plasmid, CRISPR-Cas9 genome editing plasmid, is viable in L. lactis and although the plasmid is still incomplete, it provides a basis for a quick and robust genome editing system for these bacteria. Programa Operacional Regional de Lisboa 2020 (Project N. 007317). The authors acknowledge Fundação para a Ciência e a Tecnologia (FCT, Portugal) the project grant (PTDC/BTM-SAL/28624/2017).

Authors Contributions
José Santos wrote the manuscript and conducted and designed experiments, Gabriel Monteiro wrote the manuscript and designed experiments, Miguel Prazeres wrote the manuscript and designed experiments and So a Duarte wrote the manuscript and conducted and designed experiments.