Multiplex CRISPRi System Enables the Study of Stage-Specific Biofilm Genetic Requirements in Enterococcus faecalis

Enterococcus faecalis causes multidrug-resistant life-threatening infections and is often coisolated with other pathogenic bacteria from polymicrobial biofilm-associated infections. Genetic tools to dissect complex interactions in mixed microbial communities are largely limited to transposon mutagenesis and traditional time- and labor-intensive allelic-exchange methods. Built upon streptococcal dCas9, we developed an easily modifiable, inducible CRISPRi system for E. faecalis that can efficiently silence single and multiple genes. This system can silence genes involved in biofilm formation and antibiotic resistance and can be used to interrogate gene essentiality. Uniquely, this tool is optimized to study genes important for biofilm initiation, maturation, and maintenance and can be used to perturb preformed biofilms. This system will be valuable to rapidly and efficiently investigate a wide range of aspects of complex enterococcal biology.

formed GFP_g2, encoding an sgRNA that targets the chromosomal gfp gene, together with either the pMSP3545 empty vector control or pMSP3545-dCas9 Str , into EF dCas9 . We monitored the green fluorescent protein (GFP) signal using flow cytometry, and upon nisin induction, we observed that only 1% of cells transformed with p3545-dCas9 Str remained GFP positive, compared to 98% GFP-positive cells in the empty vector control (see Fig. S2 in the supplemental material). Since inactivated dCas9 from E. faecalis does not recognize the streptococcal scaffold to silence gfp in the empty vector control, nor does it interfere with gfp silencing by dCas9 Str , we reason that the native (catalytically active) enterococcal Cas9 would also not bind to the scaffold since only catalytic and not binding residues are mutated. These results demonstrate that dCas9 EF from E. faecalis does not interfere with scaffold recognition, and subsequent gene silencing, by streptococcal dCas9.
CRISPRi silencing of chromosomal genes via template or nontemplate strand targeting. We next tested two parameters that can be potentially optimized for efficient CRISPRi targeting, namely, the GC content of the sgRNA and the guide position within the gene and its promoter region (23,24,34). We also tested template and nontemplate DNA strand targeting to determine whether only nontemplate strand targeting is efficient in E. faecalis, as has been shown in E. coli, Pseudomonas aeruginosa, Mycobacterium tuberculosis, and Caulobacter crescentus (23)(24)(25)35). We designed guides to test the ability of CRISPRi to silence the gfp gene by targeting (i) a promoter region with GFP_p1 (35% GC content), (ii) a nontemplate DNA strand with GFP_g1 (25% GC) and GFP_g2 (20% GC), and (iii) a template DNA strand with GFP_g3F (35% GC) and GFP_g4F (30% GC) ( Fig. 2A). To determine the expression conditions for maximal targeting efficiency, we used a saturating concentration of nisin of 50 ng/ml and compared planktonic bacteria subcultured without (Ϫ) and with (ϩ) nisin for 2 h. We observed only partial silencing (up to 70%) for 4 of the 5 guides when bacteria were induced for 2 h (Fig. 2B). To improve silencing, we presensitized bacteria with nisin induction overnight before subculturing the bacteria into fresh medium again with nisin for 2 h (ϩϩ). Presensitizing the bacteria universally increased the silencing efficiency for all active guides from 70% to 99% (Fig. 2B). In contrast to what has been reported for E. coli, P. aeruginosa, and M. tuberculosis, where maximal efficiency was observed upon targeting the nontemplate DNA strand close to the transcription start site, we observed that 4/5 guides, including the template-targeting GFP_g3F, per- were tested for gfp repression activity by presensitizing bacteria with nisin overnight and subculturing with nisin induction the next day (ϩϩ), or bacteria were grown overnight without nisin and induced (ϩ) or not induced (Ϫ) the following day. After 2.5 h of subculture, cells were washed and analyzed by flow cytometry. The percentage of GFP-expressing cells was determined by proprietary Attune NxT flow cytometry software from 500,000 events using the EF dCas9 p3545-dCas9Str, pGCP123 (EF dCas9 pp) empty vector as a 100% positive control. Statistical analysis was performed by 2-way analysis of variance (ANOVA) test using GraphPad. *, P Ͻ 0.05; ***, P Ͻ 0.001; ****, P Ͻ 0.0001; ns, not significant.
formed similarly under each test condition (Fig. 2B) (23)(24)(25). In contrast, sgGFP_g4F, which targeted the template strand at a distance from the translation start site (TSS), exhibited zero silencing and mimicked the empty vector control regardless of the nisin induction time (Fig. 2B). In conclusion, maximal silencing efficiency is achieved with presensitization, where all nontemplate targeted guides perform similarly regardless of the distance from the TSS or the GC content.
Efficient ebpABC operon and selective nontemplate DNA strand silencing in planktonic and biofilm bacteria. To explore the efficiency of CRISPRi silencing of whole operons, we designed sgRNAs to target the ebpABC operon, which encodes endocarditis-and biofilm-associated pili (Ebp) important for biofilm formation (36,37). The ebpABC operon is comprised of ebpA, ebpB, and ebpC expressed from the same promoter upstream of ebpA (38). EbpC is the major pilin subunit, EbpA is the tip adhesin, and EbpB is found at the base of the polymerized pilus (38). In the absence of EbpA, long pili are still polymerized, in which EbpC comprises the stalk of the pilus, while in the absence of EbpC, only short EbpA-EbpB dimers are formed (39). To test whether silencing the first gene in the operon could effectively silence the entire operon, we designed EbpA_g1 and EbpC_g1, which target the nontemplate proteincoding DNA strand 1,845 and 5,094 nt downstream of the ebpA TSS (Fig. 3A). Since we observed selective efficiency of gfp template DNA strand targeting, to further explore this selectivity, we also designed EbpA_g2F and EbpC_g2F, which target the template DNA strand 1,157 and 6,002 nt downstream of the ebpA TSS, respectively (Fig. 3A). We assessed the efficiency of operon transcriptional silencing by quantifying the amount of polymerized pili by Western blotting and by quantifying pilus function in biofilm formation since pilus-deficient strains are attenuated for biofilm formation (37).
Both guides that target ebpA, the first gene of the ebpABC operon, regardless of the targeted DNA strand, were similarly efficient in silencing the whole operon as measured by the absence of a high-molecular-weight ladder (HMWL) by Western blotting (Fig. 3B, lanes 5 and 8), indicating a lack of pilus polymerization. As expected, the lack of polymerized pili after EbpA_g1-mediated silencing correlated with reduced biofilm formation, similar to an ebpABC-null mutant (Fig. 3C). EbpC_g1 targeting the nontemplate strand efficiently silenced ebpC transcription but did not affect the expression of the upstream ebpA and ebpB genes, resulting in the absence of HMWL Ebp but the presence of EbpAB dimers (Fig. 3B, lane 6). In contrast, EbpC_g2F did not silence ebpC, leaving pilus expression unaffected (Fig. 3B, lane 10). Therefore, taken together with the gfp targeting data described above, these data suggest that the silencing efficiency achieved by targeting the template DNA strand varies and, at least in these two instances, is less effective when the target is farther away from the TSS. In contrast, targeting the nontemplate strand is efficient to silence a single gene or the whole operon and can act at a significant distance from the operon TSS.
While Ebp is important for biofilm formation, EbpA itself can bind host fibrinogen via the amino-terminal domain of EbpA (40,41). Since biofilm formation is a complex multifactorial process, we tested if CRISPRi of ebpA can also prevent adherence of the bacteria to fibrinogen. We precoated biofilm chambers with fibrinogen, silenced ebpA by inducing EbpA_g1, and performed fibrinogen-binding assays as described previously (40). We observed few attached cells in EbpA_g1-induced (EbpA_g1 ϩ ) cells compared to empty vector controls or uninduced (EbpA_g1 Ϫ ) bacteria (Fig. 3D, bottom). OG1RF and OG1RF ΔebpABC on fibrinogen-coated (Fg ϩ ) or uncoated (Fg Ϫ ) chambers served as positive-and negative-control strains, respectively (Fig. 3D, top).
Biofilm formation proceeds via a series of developmental steps, starting with the adhesion of a single cell, aggregate or microcolony formation, maturation into a heterogeneous three-dimensional (3D) structure, and, ultimately, dispersal (42). Singlegene knockouts or a priori gene silencing enables the study of the contribution of gene products to biofilm adhesion or initiation; however, the study of genes specific to postadhesion steps of biofilm formation has been more challenging to achieve. Since ebpABC are important in biofilm formation, we tested if CRISPRi can be used to perturb preformed biofilms to assess pili involvement in E. faecalis biofilm maintenance. We allowed biofilms to form with uninduced bacteria for 2, 16, or 24 h; subsequently swapped the medium to fresh medium with or without nisin; and incubated the biofilms for another 24 h. We observed a significant decrease in biofilm formation when nisin was added to silence pilus gene expression (ϪϪ/ϩ) compared to uninduced biofilms (ϪϪ/Ϫ) in early (2-h) and 16-h pregrown biofilms (Fig. 3D). Constitutively suppressed Ebp (ϩϩ/ϩ) formed less biofilm than biofilm with Ebp suppression after the uninduced biofilm was performed (ϪϪ/ϩ), showing that Ebp is important for both the initiation and maintenance of E. faecalis biofilms. To determine whether nisin can fully penetrate the biofilm, leading to CRISPRi silencing of genes at the base of the biofilm, we constructed an EF dCas9 PnisA-GFP strain that expresses gfp under the control of the nisin-inducible promoter, and the nisRK TCS is encoded on cotransformed pMSP3545. We added nisin to uninduced 16-h preformed biofilms and allowed biofilms to grow for an additional 5 h to express gfp and develop the GFP signal. We observed GFP throughout the biofilm, including the biofilm base (see Fig. 3F), indicating that nisin can penetrate the preformed biofilm. These data demonstrate that inducible CRISPRi can be used to probe stage-specific gene contributions to biofilm maturation and maintenance.
CRISPRi silencing of croR mimics a croR::Tn phenotype in antibiotic sensitivity assays. The CroRS two-component system contributes to E. faecalis antibiotic resis- EF dCas9 pp with two empty plasmids and ΔebpABC are the control strains. Ebp appears as a highmolecular-weight ladder (HMWL) of covalently polymerized pili of different lengths. SecA served as a loading control and appears as a double band at ϳ100 kDa. (C) Crystal violet staining of 24-h biofilms formed on plastic in TSBG medium. EF dCas9 pp and ΔebpABC were used as controls, and the test Ebp_g1 strain was uninduced (Ϫ) or induced with nisin (50 ng/ml) or presensitized overnight and induced the following day (ϩϩ) prior to seeding into biofilm chambers. Statistical analysis was performed by the unpaired t test using GraphPad. ****, P Ͻ 0.0001. (D) Confocal images of bacteria attached to fibrinogencoated biofilm chambers after 2 h of incubation at 37°C. Bacteria were stained with Hoechst DNA dye. EbpA_g1 ϩ, induced; EbpA_g1 Ϫ, uninduced; EF dCas9 pp, empty vector. The OG1RF and ΔebpABC strains on Fg-coated (Fg ϩ ) and uncoated (Fg Ϫ ) chambers served as additional controls. Bars, 10 m. (E) Crystal violet (CV) staining on EbpA_g1 biofilms that were pregrown on plastic without nisin induction for 2, 16, and 24 h, followed by a medium swap with nisin, and left to grow for an additional 24 h (ϪϪ/ϩ) (orange bars). Constitutively induced cultures are indicated as ϩϩ/ϩ (blue bars), and constitutively uninduced cultures are indicated as ϪϪ/Ϫ (green bars), where ϪϪ/Ϫ indicates no nisin in cultures grown overnight, seed cultures, or swap medium; ϪϪ/ϩ indicates no nisin in cultures grown overnight or in seed cultures but with nisin present in the swap medium; and ϩϩ/ϩ indicates that nisin is present in the medium of cultures grown overnight, upon seeding, and in the swap medium. A swap is indicated by "/." Statistical analysis was performed by the unpaired t test using GraphPad. ****, P Ͻ 0.0001; ***, P Ͻ 0.001; *, P Ͻ 0.05; ns, not significant. (F) Confocal images of biofilms pregrown in plastic chambers for 16 h without nisin induction, followed by a medium swap without (Ϫ) or with (ϩ) nisin and subsequent growth for 5 h to allow gfp expression and GFP signal maturation. OG1RF PnisA-GFP carries gfp under the control of PnisA from pGCP123, and nisRK is carried on the pMSP3545 plasmid. EFdCas9 pp constitutively expressing GFP from the chromosome is the control strain. Bacteria were washed and stained with Hoechst dye prior to taking the Z-stack images. Imaris software 3D modeling of the confocal laser scanning microscopy images is on the left, and the corresponding first (bottom) layer of the Z-stack is on the right. Bars, 10 m.
tance, survival within macrophages, stress responses, and growth (43)(44)(45)(46)(47). CroR phosphorylation by the cognate CroS sensor kinase is important for resistance to bacitracin, vancomycin, and ceftriaxone, where ΔcroR cells are no longer resistant to these antibiotics (44). To further validate the CRISPRi system in E. faecalis, we designed CroR_g1 to target croR on the nontemplate DNA strand and assessed the sensitivity of the resulting strain to bacitracin, using croR::Tn as a control. Nisin-induced CroR_g1 bacteria mimicked croR::Tn and showed reduced resistance to bacitracin (MIC, 8 g/ml) compared to the empty vector control (MIC, 32 g/ml) (Table 1). Hence, the CRISPRi system can be used to study genes involved in antibiotic resistance.
Efficient combinatorial gene silencing. We next assessed the ability of the CRISPRi system to simultaneously silence multiple genes. At the same time, we tested if the presence of a second guide with the same nisA promoter and Cas9 handle could be unstable or prone to recombination and interfere with the efficiency with CRISPRi, as has been reported for some systems (48,49). We used combinatorial genetic en masse (CombiGEM) technology (50) to generate GFPEbpA_g1g1 and EbpASrtA_g1g1, enabling the expression of two sgRNAs under the control of independent nisA promoters, to simultaneously silence the gene pair gfp and ebpA or ebpA and srtA. SrtA is an enzyme that covalently attaches polymerized pili to the cell wall, and the deletion of this gene results in the absence of Ebp bound in the cell wall fraction of cells (38). Upon nisin induction of GFPEbpA_g1g1, we observed the simultaneous loss of the EbpA signal by Western blotting (Fig. 4A, lane 8) and reduction of the GFP signal by flow cytometry (Fig. 4B). Similarly, in the induced EbpASrtA_g1g1 strain, no signal was observed for EbpA or SrtA by Western blotting (Fig. 4C, lane 6) compared to the empty vector control (Fig. 4C, lane 1). Thus, the combinatorial plasmids silenced two gene pairs with an efficiency equivalent to that of single-guide gene silencing, where the presence of a second sgRNA construct did not affect the silencing efficiency of the other guide.
Essential gene targeting with presensitizing at subinhibitory nisin concentrations. Finally, we assessed the ability of the inducible CRISPRi system to study essential genes. We chose to target secA, an essential gene of the general secretion pathway (51). We designed SecA_g1 to target the nontemplate DNA strand of secA. Because secA is essential, silencing of this gene is predicted to attenuate the growth of the strain. We performed growth curves at various nisin concentrations to assess growth inhibition after secA targeting. Even without induction, we observed a reduced growth rate compared to the empty vector control, presumably due to a leaky nisA promoter with basal expression of dCas9 Str and SecA_g1 (Fig. 5). Upon induction, we observed similar growth inhibition for nisin concentrations of 2.5 to 50 ng/ml, suggesting that maximal inhibition of SecA without presensitization is achieved at 2.5 ng/ml (Fig. 5B). To increase the degree of inhibition, we presensitized bacteria overnight with 2.5 ng/ml nisin, as we observed no growth for cultures grown overnight at nisin concentrations of 5 to 50 ng/ml (data not shown). When presensitized cultures were further subcultured with 2.5 ng/ml of nisin, we observed a pronounced growth defect, which was most apparent between 4 and 10 h after induction, compared to the empty vector (no-guide) control (Fig. 5C). In summary, we showed that the function of essential genes can be studied using a low-level nisin presensitization and induction protocol.

DISCUSSION
Genetic tools to easily and rapidly study the contribution of single or multiple enterococcal genes in a given biological process are lacking. To address this, we developed a scalable dual-vector nisin-inducible CRISPRi system for E. faecalis. The system is most efficient on presensitized cultures and can be used to study a variety of bacterial phenotypes, including biofilm formation, antimicrobial resistance, and gene essentiality.
Similar to CRISPRi systems developed for other bacterial species, our system is inducible and efficient and can be multiplexed (21,24). We employed a two-plasmid system: one plasmid encodes dCas9 and a nisin-responsive two-component system, and the other encodes the nisin-inducible sgRNA. The second plasmid, pGCP123, is small and easily modifiable for combinatorial targeting. We can introduce sgRNA into a digested plasmid in the form of a gBlock (or reannealed oligonucleotides) through Gibson assembly or In-Fusion reactions (22,50,52). The plasmid can be further modified for the simultaneous targeting of multiple genes by the ligation-digestion reaction of compatible restriction sites using CombiGEM technology (50). Our system uses the streptococcal dCas9 with the handle, a well-characterized tool for CRISPRi (19,23). Although Cas9 Str is orthologous to native enterococcal Cas9 and shares the same PAM, we showed that the streptococcal dCas9 handle is not recognized by native enterococcal dCas9 and can be used in E. faecalis without modifying the endogenous Cas9 (26, 53).
We observed the highest level of dCas9 Str protein expression at 25 ng/ml of nisin. However, nisin is more stable at a lower pH and may partially degrade over 24 h in the culture medium (54). To account for degradation, we used a higher nisin concentration of 50 ng/ml to maintain the maximal strength of the nisA promoter (28,54). At 50 ng/ml of nisin, most Gram-positive and Gram-negative bacteria are able to replicate as the MIC is typically Ͼ1,000 ng/ml, allowing our system to be used in the context of multispecies interactions (55). Since the nisin-controlled gene expression system is functional in a wide range of Gram-positive bacteria, including Lactococcus, Lactobacillus, Leuconostoc, Streptococcus, and Enterococcus, our CRISPRi-Cas9 system therefore can be potentially used in these genera (56).
To design sgRNAs, we used the CHOPCHOP database on the E. faecalis OG1RF genome and selected for the guides with a zero off-target score (57). We tested sgRNAs of various GC contents (20 to 60%), targeting template and nontemplate DNA strands, and at various distances from the translation start site (TSS). The GC content played no role in the efficiency of silencing. All nontemplate DNA strand-targeting sgRNAs were ® efficient, independent of the distance from the TSS, even though in E. coli, the efficiency of silencing was reported to decrease with increasing distances from the transcription start site (19). Surprisingly, we observed that 2 out of the 4 guides designed on the template DNA strand were still efficient in silencing the targeted gene, whereas it has been generally assumed that CRISPRi sgRNA must target the nontemplate DNA strand for efficient silencing in bacteria (19)(20)(21)23). It is possible that template DNA strandtargeting guides, such as GFP_g3F, that bind to the template strand just 7 nt away from the TSS may allow dCas9 to interfere with the assembly of the transcription machinery, preventing transcription initiation. Consistent with this possibility, another guide, GFP_g4F, that targets the template strand 200 nt downstream from the TSS does not silence GFP. However, targeting the template strand of ebpA using EbpA_g2F, which binds 1,157 nt away from the TSS, is efficient in gene silencing, indicating that template strand targeting and silencing are not universally distance dependent. Further work is needed to understand the nature and mechanism of template DNA strand silencing in E. faecalis.
A great challenge in studying gene contributions to different stages of developmental cycles, such as those that occur during biofilm formation, arises when early steps are essential for later steps to occur, necessitating the ability for stage-specific gene silencing. We leveraged the inducibility of our system to trigger ebpA silencing in preformed biofilms to address the role of these pili after biofilm initiation, during the maturation and maintenance stages. We showed that nisin can penetrate the entire biofilm to trigger gene expression from the nisA promoter. We observed a significant reduction in biofilm biomass in the induced cultures compared to uninduced controls, indicating that CRISPRi-Cas9 can be used to perturb the preformed biofilms and to identify and interrogate gene targets in a biofilm stage-specific manner.
To expand the uses of this CRISPRi system, we utilized CombiGEM technology to generate, in one easy step, combinatorial plasmids to target two genes simultaneously (50). We showed that the simultaneous expression of two guides was as efficient in silencing as the expression of a single guide per cell. Despite the presence of the same promoter sequence on different plasmids, we did not observe disruptive recombination of the promoters or between the guides, with stable and consistent repression of both targeted genes. Therefore, this system has the potential to be scaled up for sgRNA library preparation and high-throughput combinatorial studies.
Finally, our system was tested in the study of essential genes, where targeting the essential secA gene with minimal nisin induction significantly impaired bacterial growth, while high concentrations of nisin impeded the growth and killed the bacteria. Genetic tools to study essential genes in E. faecalis are limited to transposon mutant library sequencing approaches and essential gene inactivation with in trans complementation (6,58,59). Therefore, we can employ CRISPRi to study gene essentiality under various nutrient conditions or leverage upon the systems inducibility and probe essentiality of the genes in vivo.
In summary, we have developed and validated an efficient CRISPRi system that can be readily used to study single or a combination of genes involved in different biological processes and can be modified for high-throughput screens, including combinatorial analyses, in E. faecalis. This tool will effectively facilitate the study of E. faecalis pathogenesis and allow the rapid identification of novel targets for future interrogation.

MATERIALS AND METHODS
Bacterial strains and medium conditions. The strains and plasmids used in the study are listed in Table 2. E. faecalis strains were grown statically at 37°C in tryptone soy broth (Oxoid, UK) supplemented with 10 mM glucose (TSBG) for biofilm studies, in Mueller-Hinton broth II (MHB-II; Merck, USA) for bacitracin susceptibility tests, and in brain heart infusion (BHI) medium (Merck, USA) or BHI agar (Merck, USA) for the rest of the experiments. E. coli was grown in Luria-Bertani broth Miller (LB; BD, Difco, USA) at 37°C with shaking at 200 rpm. Erythromycin (100 g/ml) was used to maintain the pMSP3545 plasmid in E. faecalis, and kanamycin (500 g/ml for E. faecalis and 50 g/ml for E. coli) was used to maintain pGCP123 and its derivatives. A nisin (Sigma, USA) stock solution was prepared as 0.1 mg/ml by dilution Multiplex CRISPRi in Enterococcus faecalis ® September/October 2020 Volume 11 Issue 5 e01101-20 mbio.asm.org 11 in deionized water. The nisin solution was then filter sterilized through a 0.22-m filter, aliquoted, and frozen at Ϫ20°C. When needed, an aliquot was thawed and used once. Genetic manipulations. To abolish the endonuclease activity of native enterococcal Cas9 EF , we first aligned csn1 (OG1RF_10404) to S. pyogenes Cas9 (BLASTP, NCBI) and identified conserved catalytic D10 and H852 residues within the RuvC-1 active site and HNH endonuclease domains, respectively (13). Both amino acids are essential for cleaving the template and nontemplate DNA strands (13). To create D10A and H852A substitutions and introduce a 1-kb flanking region into Cas9 EF for subsequent allelic exchange, we amplified 3 fragments from the bacterial chromosome with primers pairs 1/2, 3/4, and 5/6, listed in Table 3, and performed splice by overlap extension (SOE) PCR of the 3 fragments with primer pair 1/6. The resulting product was introduced into the PstI/KpnI-digested temperature-sensitive vector pGCP213 by an In-Fusion reaction (TaKaRa Bio, Japan). The resulting pGCP213-dCas9 EF vector was verified by sequencing and used for allelic exchange to generate SD234::dCas9 as described previously (8). The presence of both mutations D10A and H852A in the csn1 gene encoding Cas9 EF was verified by PCR and sequencing with primer pairs 7/8 and 9/10.
To generate pMSP3545-dCas9 Str , dCas9 Str was amplified from pdCas9-bacteria (Addgene plasmid 44249) using the primer pair 11/12, and the purified product was introduced by an In-Fusion reaction into pMSP3545 (Addgene plasmid 46888) digested with SpeI and NcoI.
To generate sgRNA expression vectors, a 307-bp gBlock (IDT, USA) consisting of the nisA promoter linked directly to a 20-nt sgRNA linked to the dCas9 scaffold (see Fig. S1 in the supplemental material) was introduced by an In-Fusion reaction into pGCP123 digested with BglII and NotI. The gBlock contains 4 restriction sites (BglII, BamHI, EcoRI, and MfeI) for the generation of a pairwise library through restriction-ligation reactions and an 8-nt unique barcode for high-throughput screens coupled with amplicon sequencing (Fig. S1) (50). sgRNA sequences were selected using the CHOPCHOP database with a zero off-target score and minimal self-complementarity (0 to 1) ( Table 4) (57).
To construct pGCP123-PnisA-GFP, we amplified PnisA from pMSP3545 using the primer pair 13/14 and amplified gfp from the SD213 chromosome using the primer pair 15/16. The two fragments were  GFP_R GTCGTTTTACAGATCCTATTTGTATAGTTCATCCATGCCATGTG introduced into BglII-digested pGCP123 by an In-Fusion reaction and validated by sequencing. The pGCP123-PnisA-GFP plasmid was then transformed into OG1RF together with pMSP3545 to generate the OG1RF PnisA-GFP strain. Flow cytometry. A single colony was inoculated and grown overnight with or without the addition of nisin. The following day, cultures were diluted 1:30 in 1,160 l of fresh medium in a 2-ml tube and grown for 3 h statically at 37°C. After incubation, the cells were collected by centrifugation, resuspended in 1 ml phosphate-buffered saline (PBS), and analyzed on an Attune NxT flow cytometer. The percentage of GFP-expressing cells was determined by proprietary Attune NxT flow cytometry software from 500,000 events based on polygonal gating on the EF dCas9 empty vector control.
Western blotting. A single colony was selected from agar plates, inoculated into liquid medium, and grown overnight with or without nisin induction. Cultures grown overnight were then diluted 1:10 in fresh medium in the presence of antibiotics and nisin where appropriate and grown until mid-log phase. Samples were normalized to an optical density at 600 nm (OD 600 ) of 0.6 and pelleted by centrifugation, and the pellet was resuspended in 75 l of 10 mg/ml of lysozyme in lysozyme buffer (10 mM Tris-HCl [pH 8], 50 mM NaCl, 1 mM EDTA, 0.75 M sucrose) and incubated at 37°C for 1 h. After lysozyme treatment, 25 l of 4ϫ lithium dodecyl sulfate (LDS) buffer (Invitrogen, USA) was added to the samples, and the samples were heated at 95°C for 10 min and stored at Ϫ20°C until analysis.
For immunoblot analysis, 10 l of each sample was loaded onto a 4-to-12% gradient NuPAGE Bis-Tris minigel for SecA or SrtA or a 3-to-8% gradient NuPAGE Bis-Tris minigel for EbpA and run in 1ϫ morpholinepropanesulfonic acid (MOPS) (for 4 to 12% gels) or Tris-acetate (for 3 to 8% gels) SDS running buffer, respectively, in an XCellSureLock minicell for 50 min at 200 V. Proteins from the gel were then transferred to a membrane using the iBlot dry blotting system. The membrane was then blocked with 3% bovine serum albumin (BSA) in phosphate-buffered saline with 0.05% Tween 20 (PBS-T) for 1 h, with shaking at room temperature (RT). SecA, SrtA, and EbpA were detected using custom antibodies raised in rabbit, mouse, and guinea pig, respectively (8,60). Cas9 was detected using a monoclonal mouse anti-Cas9 antibody (Abcam). Appropriate IgG secondary horseradish peroxidase (HRP)-conjugated antibodies (all Thermo Scientific, Singapore) were used for detection.
Biofilm assay. Bacterial cultures grown overnight were washed and normalized to an OD 600 of 0.7 as described previously (61). A total of 5,000 CFU/well were inoculated in TSBG in a 96-well flat-bottom transparent microtiter plate (Thermo Scientific, Waltham, MA, USA) and incubated at 37°C under static conditions for 24 h. After the removal of planktonic cells, the adherent biofilm biomass was stained using 0.1% (wt/vol) crystal violet (Sigma-Aldrich, St. Louis, MO, USA) at 4°C for 30 min. The microtiter plate was washed twice with PBS followed by crystal violet solubilization with ethanol-acetone (4:1) for 45 min at room temperature. The quantity of adherent biofilm biomass was measured by absorbance at the OD 595 using a Tecan Infinite 200 Pro spectrophotometer (Tecan Group Ltd., Männedorf, Switzerland).
Fibrinogen-binding assay. Bacterial binding to fibrinogen was assayed as described previously (40), with the following modifications. The strains were grown overnight with or without nisin induction, normalized to an OD 600 of 1, and washed once with PBS. A total of 200 l of the normalized culture was added to 8-well biofilm chambers (Ibidi, Germany) precoated overnight with 10 mg/ml human fibrinogen (catalog number F3879; Sigma) at 37°C. Cells were incubated for 1 h at 37°C, washed once with PBS, and stained with Hoechst 33342 (Thermo Fisher Scientific, USA) for 30 min at RT. Cells were washed once with PBS, and visualized in Ibidi chambers after excitation with a 405-nm laser using a Zeiss LSM780 confocal microscope (Zeiss, Germany) with a 63ϫ NA1.4 objective.
Confocal laser scanning microscopy and 3D reconstruction. Confocal imaging of biofilms was performed as previously described (36), with the following modifications. Cultures grown overnight were normalized to an OD 600 of 0.7, and 8 l was inoculated into each well of an 8-well ibiTreat biofilm chamber (Ibidi, Germany) containing 200 l TSBG medium, with a final concentration of 10 5 CFU/ml. Biofilms were preformed at 37°C for 16 h with or without nisin, after which the culture medium was swapped, fresh medium with or without nisin was added, and the biofilms were incubated for 5 h at 37°C. Supernatants were removed, and biofilms were washed once with PBS and stained with Hoechst 33342 (Thermo Fisher Scientific, USA) for 30 min at RT. Biofilms were imaged directly from the chambers by exciting with 405-nm and 488-nm lasers using a Zeiss LSM780 confocal microscope (Zeiss, Germany) with a 63ϫ NA1.4 objective. Z-stacks were collected through the entire biofilm thickness, and 3D reconstruction was performed with Imaris software (Bitplane, USA). Bacitracin susceptibility determination. The MIC of bacitracin was determined in liquid MHB-II medium in a 96-well plate. Twofold serial dilutions from 128 g/ml to 4 g/ml of bacitracin were prepared in triplicate from a 512-g/ml bacitracin stock. Cultures of bacteria grown overnight were normalized to an OD 600 of 0.7, and 8 l was inoculated into each well containing 200 l TSBG medium, with a final concentration of 10 5 CFU/ml. Plates were incubated at 37°C for 16 h. The next day, the MIC was determined by visually assessing turbidity. The lowest concentration of the antibiotic that prevented growth was recorded as the MIC.
Growth curve assessment. Cultures grown overnight were washed in PBS and normalized to an OD 600 of 0.7. Normalized cultures were inoculated into 200 l BHI medium at a ratio of 1:25. Three biological replicates and four technical replicates were performed for each culture. The 96-well plates were incubated at 37°C for 16 h using the BioTek Synergy 4 plate reader (BioTek, USA). The optical density was taken at 600 nm at 30-min intervals to determine the growth curve of each culture.
Statistical analysis. Statistical analyses were performed using GraphPad Prism software (version 6.05 for Windows). All experiments were performed in at least three biological replicates, and the mean value was calculated. All graphs show the standard deviations from independent experiments. Statistical analysis was performed by the unpaired t test using GraphPad. P values of Ͻ0.05 were deemed significant.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.