CRISPR-Cas immunity is repressed by the LysR-type transcriptional regulator PigU

Abstract Bacteria protect themselves from infection by bacteriophages (phages) using different defence systems, such as CRISPR-Cas. Although CRISPR-Cas provides phage resistance, fitness costs are incurred, such as through autoimmunity. CRISPR-Cas regulation can optimise defence and minimise these costs. We recently developed a genome-wide functional genomics approach (SorTn-seq) for high-throughput discovery of regulators of bacterial gene expression. Here, we applied SorTn-seq to identify loci influencing expression of the two type III-A Serratia CRISPR arrays. Multiple genes affected CRISPR expression, including those involved in outer membrane and lipopolysaccharide synthesis. By comparing loci affecting type III CRISPR arrays and cas operon expression, we identified PigU (LrhA) as a repressor that co-ordinately controls both arrays and cas genes. By repressing type III-A CRISPR-Cas expression, PigU shuts off CRISPR-Cas interference against plasmids and phages. PigU also represses interference and CRISPR adaptation by the type I-F system, which is also present in Serratia. RNA sequencing demonstrated that PigU is a global regulator that controls secondary metabolite production and motility, in addition to CRISPR-Cas immunity. Increased PigU also resulted in elevated expression of three Serratia prophages, indicating their likely induction upon sensing PigU-induced cellular changes. In summary, PigU is a major regulator of CRISPR-Cas immunity in Serratia.


Introduction
In almost every environment bacteria are outnumbered by their viral invaders -bacteriophages (phages).Indeed, there are an estimated 10 31 phages on Earth that cause ∼10 25 infections of bacteria every second ( 1 ,2 ).These phage-bacterium interactions influence bacterial evolution and nutrient cycling (3)(4)(5).To protect themselves, bacteria have evolved many defence strategies, including diverse innate immune mechanisms and CRISPR-Cas adaptive immunity ( 6 ,7 ).CRISPR-Cas systems are small RNA-based defence mechanisms.CRISPR 'memory banks' contain short repeats separated by similarsized 'spacers' that contain genetic memories of past infections.Immunity involves three phases.During adaptation short invader-derived sequences are added as new spacers to the CRISPR array.Next, expression results in a precursor CRISPR RNA (crRNA) and its processing into small guide cr-RNAs by Cas proteins.Finally, Cas protein(s) and the crRNAs form interference complexes that recognise and degrade complementary nucleic acids.Although many CRISPR-Cas systems share similarities, there are two major classes (Class 1 and 2) made up of six major types (I-IV), which are further divided into > 30 sub-types ( 8 ).Class 1 systems are composed of systems encoding multi-subunit interference complexes, such as type I systems, which target DNA, and type III systems, which target both DNA and RNA in a transcriptiondependent manner ( 8 ).Type III systems also encode accessory nucleases which often result in collateral DNA / RNA damage ( 9 ).This diversity provides a bounty of proteins to exploit as biotechnological tools ( 10 ).
Despite the obvious immune benefits to bacteria and archaea that contain CRISPR-Cas systems, there are downsides to harbouring these systems.For example, CRISPR-Cas systems can make mistakes while recognising invaders and may incorrectly generate immunity against their own host genomes, sometimes resulting in self-targeting and cell death (i.e.autoimmunity) (11)(12)(13)(14)(15). Indeed, CRISPR-Cas systems are proposed to incur a fitness cost to cells (16)(17)(18).Therefore, in the absence of invaders, it is vital for bacteria to limit CRISPR-Cas activity and the potential costs.Conversely, when exposed to invaders, an elevated immune response is desirable ( 19 ,20 ).
While there have been considerable advances in deciphering adaptation, processing and interference, regulation of CRISPR-Cas expression is less well understood ( 19 ,20 ).One conserved form of CRISPR-Cas regulation is through quorum sensing (QS), a widespread form of cell-cell communication in which accumulation of extracellular chemical signals can alter population gene expression ( 21 ).QS upregulates adaptive immunity at high cell density when the risk of a phage epidemic is increased ( 22 ,23 ).The progress of identifying mediators of CRISPR-Cas activity had been hindered by the lack of unbiased discovery methods.Recently, to overcome this limitation, we developed SorTn-seq ( 24 ,25 ).SorTn-seq is a high throughput method that couples high-density random transposon mutagenesis with fluorescent reporter genes, fluorescence activated cell sorting (FACS) and transposon insertion deep-sequencing to identify regulators of any gene of interest.
We previously applied SorTn-seq to discover regulators of the type III-A cas ( csm ) operon in Serratia sp.ATCC39006 strain LacA (hereafter Serratia ) ( 24 ,25 ).This Serratia strain is an environmental isolate that encodes type I-E, I-F and III-A CRISPR-Cas systems, in addition to multiple toxin-antitoxin and other defence systems ( 22 ,26 ).Using SorTn-seq, we identified and characterised an Rcs stress response pathway that inversely controls CRISPR-Cas and surface immunity.This approach also allowed a thorough genome-wide identification of the regulators of the type III-A cas operon.However, we did not know whether the CRISPR array promoters associated with the III-A system (CRISPR3 and CRISPR4) had coordinate or separate regulation compared with the cas operon.In addition, no studies have systematically assessed genes that influence CRISPR array expression.Here, we identified numerous genetic loci affecting CRISPR array expression for the type III-A system in Serratia and compared these to genes controlling the type III-A cas operon.We further characterised a key LysR-type transcriptional regulator called PigU (a homologue of Esc heric hia coli LrhA) that co-ordinately repressed type III-A CRISPR array and cas gene expression and hence CRISPR-Cas interference activity.PigU also repressed interference and adaptation by the type I-F system.

Construction of type III-A CRISPR-cas phage JS26 targeting strain
To introduce a type III-A phage JS26-targeting spacer into the Serratia chromosome, the native CRISPR3 array was replaced with a mini-array (repeat-spacer-repeat) using allelic exchange mutagenesis as previously described ( 25 ) with pPF3581 in the WT (LacA) background.The resulting strain (PCF925) contains one spacer targeting the phage JS26 JT354_gp01 gene (encoding a hypothetical protein).

SorTn-seq: transposon mutagenesis
Transposon libraries for SorTnSeq were constructed as previously described ( 24 ,25 ).Briefly, E. coli donor strain ST18 harbouring the Tn 5 transposon delivery plasmid pKRCPN2 was conjugated into the recipient strain PCF396 harbouring either the CRISPR3 or CRISPR4 reporter plasmids (pPF1923 or pPF1924).Following conjugation, cells were inoculated into three flasks (2 L) containing 500 ml of LB (supplemented with Km and Cm) at a starting OD 600 of 0.02 and grown at 30 • C with shaking (180 rpm) for 24 h to select for transposon mutants.After outgrowth, cells were pooled (45 ml in total), centrifuged and resuspended in LB to a final OD 600 of 3 to generate the final Tn library.Aliquots (1 ml) of the library were mixed 1:1 with 50% glycerol and frozen in cryotubes at − 80 • C for future use.

SorTn-seq: fluorescent activated cell sorting (FACS)
Sorting of mutant libraries was performed as previously described ( 24 ,25 ).Briefly, 1 ml of frozen Tn mutant library were subcultured (starting OD 600 = 0.05) into 30 ml of LB with Cm for reporter plasmid selection, Km for transposon selection and IPTG for mCherry induction.Cells were grown for 16 h to allow expression of the CRISPR-eYFP reporter.Cells were then diluted in 15 ml of phosphate-buffered saline (PBS, 1:30) and sorted using a FACSAria Fusion (BD Biosciences).Despite minimal spectral overlap between eYFP and mCherry, single stain controls (mCherry only-PCF396 + pPF1438; eYFP only-PCF396 + pPF1307; and unstained cells-PCF396) were used to establish a compensation matrix in the BD FACSDiva software (v.8, BD Biosciences).Cells were gated on forward scatter (FSC) and side scatter (SSC) parameters (area, height, width) as seen in ( 24 ,25 ).The mCherry + cells were selected and sorted into three bins based on eYFP fluorescence levels.Gates high and low were each set at approximately 5% of the total, whereas the depleted gate was set at 90%.Cells were sorted under 'purity' mode, and approximately 2.0 × 10 7 cells were sorted per experiment.Sorted cells were recovered in 0.2 ml of PBS.Outgrowth and sorting were performed in triplicate on different days, with each experiment yielding one low, one high, and one depleted fraction (total n = 9).

SorTn-seq: DNA extraction, library preparation and sequencing
Preparation of sorted cells for deep-sequencing was performed as previously described ( 24 ,25 ).Sorted cell fractions ( n = 9) were centrifuged and DNA was extracted using the DNeasy Blood and Tissue Kit (QIAGEN) following the manufacturer's instructions.Sequencing libraries were constructed using the NEBNext Ultra II FS DNA Library Prep Kit for Illumina.The protocol was modified to have two rounds of PCR enrichment.Round one used customized PCR enrichment primers: PF3140, which binds the NEB adaptor, and PF3139, a biotinylated primer that binds within the Tn.Biotinylated products were captured using Dynabeads M-270 Streptavidin (Invitrogen) following the manufacturer's instructions, and beads were used as the template in the second-round PCR with a nested Tn primer PF3270 and an indexing primer (NEBNext Multiplex Oligos for Illumina).Library quality was assessed on an Agilent Bioanalyzer 2100 using a High Sensitivity DNA Kit.Libraries were further assessed through quantitative PCR (KAPA Library Quantification Kit, Universal, catalogue no.KK4824) using primers PF3124 / PF3125 to determine molarity of fragments with Illumina P5 / P7 ends (sequences required for flow-cell hybridization).Tn sequencing primer PF2926, along with PF3125, were used to determine the percentage of fragments containing true Tn sequences.Libraries were quantified using a Qubit fluorimeter and dsDNA HS Kit (Thermo Fisher Scientific) and diluted to 10 nM based on Qubit concentration and average fragment size.Libraries were then pooled with 10% PhiX control library and loaded at 1.5 pM (for low diversity) using a MiSeq Reagent Kit v.3 150 cycle kit for Illumina.Libraries were sequenced for 75 cycles (single ended) using custom sequencing primer PF2926 and Illumina Read 1 primer PF3441 (for PhiX library) at the Otago Genomics Fa-cility (OGF).Sequencing with PF2926 generates a 12-nt transposon 'tag' to verify reads originating from Tn junctions.

SorTn-seq: data analysis
SorTn-seq data was analyzed as previously described ( 24 ).Samples were de-multiplexed based on index sequence by OGF using standard Illumina software.FASTQ files were trimmed from the 3 -end to 50 nt using trimmomatic ( 30 ) and mapped to reference sequences using the SorTn-seq pipeline ( 31 ).The following parameters were used to run the bac-teria_tradis script: '-smalt -smalt_k 10 -smalt_s 1 -smalt_y 0.92 -mm 2 -v -f filelist.txt-t T A T AA GA GA CA G -r laca.fasta',where -smalt_ commands specify mapping parameters, -t specifies the transposon tag sequence, -r specifies the reference sequence, -f indicates the files to be processed and -mm indicates the number of mismatches allowed in the Tn tag.Plot files (.insert_site_plot), which tabulate the number of reads at each nucleotide position in the reference sequence (plus and minus strands), were generated by the bacteria_tradis script ( 32 ).Subsequent analysis was performed in R using SorTnseq R scripts ( 31 ).Plot files were used to generate a table of unique Tn insertion sites (for each sample).Feature enrichment was determined using differential expression analysis (exact methods-classic) of unique insertions in edgeR ( 33 ).The depleted samples served as the control group, against which low and high samples were compared.We considered features with an adjusted P < 0.05 (Benjamini-Hochberg false discovery rate (FDR) correction) and a log 2 (fold-change (FC)) > 0.5 as potential regulators.

Flow cytometry expression analysis of cas10 , CRISPR3 and CRISPR4 promoters
Flow cytometric quantification of expression from the cas10 , CRISPR3 and CRISPR4 promoters from the eYFP reporter plasmids (pPF1307, pPF1923 and pPF1924 respectively) was analysed in different Serratia strains as described previously ( 25 ).

Conjugation interference assays
To assess the ability of CRISPR-Cas systems to interfere with conjugation, plasmids harbouring protospacer sequences corresponding to the first spacer in either the CRISPR1 (type I-E: pPF724), CRISPR2 (type I-F: pPF722), CRISPR3 (type III-A: pPF1043 / pPF2841), CRISPR4 (type III-A: pPF3932) arrays, or control plasmids (pPF719, pPF1043, pPF1621) were conjugated into Serratia from E. coli ST18 donor cells.The efficiency of plasmid conjugation was calculated as the ratio of transconjugants per total recipients as described previously ( 34 ,35 ).

Type I-F CRISPR adaptation assays
To assess the effect of pigU expression on CRISPR adaptation, a plasmid that induces I-F priming (pPF3807 -harbouring a sequence complementary to spacer 2 from the native I-F CRISPR2 array with an AGA PAM) or control plasmid (pPF3805) were conjugated into Serratia WT + control expression vector (pQE-80LoriT) and WT + PigU expression vector (pPF1983) from E.coli ST18 donors.Transconjugants were grown in LB with Cm (for priming plasmid maintenance), Ap (for expression plasmid maintenance), IPTG (for PigU induction) and arabinose for mCherry induction.
After 24 h, strains were subcultured (1 / 500) in fresh LB with only Ap (for expression plasmid maintenance) and arabinose (for mCherry induction).Passaging was performed for four days, and each day the density of cultures was measured (OD 600 ) and dilutions of cells were frozen for array expansion screening.Aliquots of cultures from each day were also mixed 1:1 with 50% glycerol and frozen at -80 • C for future use.CRISPR2 array expansion (indicative of adaptation) was assessed on day two of passaging via PCR (20 cycles) with primers PF1888 / PF1990.PCR samples were run at 180 volts on a 2% (w / v) agarose gel with ethidium bromide in sodium borate buffer.

Phage efficiency of plating (EOP) assay
Strains with chromosomal anti-phage spacers for type I-F (PCF525) and III-A (PCF925) with either control (pQE-80LoriT) or pigU overexpression (pPF1983) plasmids were grown overnight.These cultures (100 μl) were used to seed 4 ml of LB top agar (0.35%) containing Ap for plasmids maintenance.The seeded top agar was spread on LB plates to create a bacterial lawn.Phage tenfold dilutions (5 μl) were spotted on these top agar overlays.After incubation of the plates overnight the plaque forming units (PFU) were counted.For plates with no individual plaques (e.g.only lawn clearing or lysis from without visible), the number of plaques was recorded as one in the following dilution without clearing.EOP was calculated by dividing the PFU with the PFU of the control strain (non-targeting, empty vector).

RNA sequencing and analysis
Overnight cultures of Serratia LacA (WT) and HSPIG43 strains were subcultured into 25 ml LB medium in 250 ml flasks (biological triplicates) to a starting OD 600 of 0.05.Next, the cultures were grown for 12 h at 30 • C with shaking at 200 rpm, and 2 ml samples were collected for RNA extraction.The 12 h time point (early stationary phase) was previously established as a point of elevated CRISPR-Cas activity due to the rising density of bacterial populations ( 22 ).Bacteria were harvested by centrifugation and the resulting cell pellets were resuspended in 1 ml RNAlater (ThermoFisher Scientific) and stored at −20 • C until further processing.Next, the Qiagen RNeasy kit was used to extract total cellular RNA.In addition, the cells resuspended in RLT buffer (Qiagen RNeasy kit) and β-mercaptoethanol were subjected to a bead-beating step of 30 s to ensure full cell lysis.Two μl of TurboDNase (Ther-moFisher Scientific) was subsequently added to the samples for 30 min to degrade all the residual genomic DNA (gDNA) and ensure the purity of RNA samples.
The samples were confirmed to be gDNA-free by means of PCR analysis with primers PF796 and PF797 that are designed to amplify the flhDC operon of Serratia .Further quality control of the resulting RNA samples were performed using the Nanodrop (ThermoFisher Nanodrop one) and 2100 Bioanalyzer (Agilent Genomics) using the High Sensitivity RNA chip.RNA samples were sequenced at Vertis Biotechnologie in Freising, Germany.Ribosomal RNA (rRNA) was depleted using a RiboZero kit (Illumina), and the remaining RNA species were fragmented by means of an ultrasound.Synthesis of antisense cDNA was initiated through ligation of a TruSeq adaptor sequence (Illumina) to the 3 OH end of the fragmented RNA.Next, the antisense cDNA was purified, followed by a ligation of a 5 sequencing adaptor to the 3 end of the anti-sense cDNA.The cDNA was then amplified using PCR (the number of PCR cycles was dependent on the amount of starting product) and the resulting products were gel fractionated to satisfy the size requirements for Illumina sequencing.Lastly, cDNA libraries were sequenced with the Illumina NextSeq 500 System to an average depth of approximately 10 million reads per library, generating an output in the form of 75 bp demultiplexed reads in FASTQ format.
Generated reads in FASTQ format were initially processed by removing adaptors and low-quality reads using Trimmomatic ( 30 ).Additionally, quality assessment of the reads was carried out using FASTQC ( 36 ).Bowtie2 ( 37 ) was used with default parameters for mapping reads to the reference genome of Serratia sp.ATCC 39006 (accession number: CP025085.1),followed by a conversion to BAM format for analysis using SAMtools ( 38 ).Statistical analysis was performed using the DESeq2 package in an R environment to identify differentially expressed transcripts with a false discovery rate (FDR) of less than 5% ( 39 ).

SorTn-seq identifies regulators of type III CRISPR expression
Serratia has a type III-A CRISPR-Cas system that contains two independent CRISPR arrays ( 22 ) (Figure 1 A).To identify regulators affecting CRISPR transcription, SorTn-seq experiments were performed.P CRISPR -eYFP reporters were constructed for the promoter regions of both CRISPR3 and CRISPR4 (Figure 1 B and Supplementary Figures S1 a-S1 d).High density mini-Tn 5 transposon libraries were made using Serratia strains carrying either of the reporters.Fluorescence activated cell sorting (FACS) was then used to sort individual transposon mutants based on the level of eYFP fluorescence into ' low' , ' high' and ' depleted' bins.High throughput, Tn 5 insertion site deepsequencing was then performed on each bin, and the Tn 5 insertion locations were mapped to the Serratia genome (Figure 1 c and Supplementary Figure S1 e).The CRISPR3 and CRISPR4 libraries had approximately 278 000 and 262 000 mutations respectively, corresponding to an insertion on average every ∼18 and ∼19 nt ( Supplementary Table S4 ).Approximately 80 000 unique insertions were obtained across the triplicates for the depleted samples for both CRISPR3 and 4 ( Supplementary Figure S1 e and Supplementary Table S4 ).This is comparable to what has previously been obtained for the type III-A cas operon in Serratia ( 25 ).Transposon insertions were identified in the low and high bins for both CRISPR3 and CRISPR4, with a range of ∼14 500-26 000 unique mutations ( Supplementary Figure S1 f and Supplementary Table S4 ).
Loci of interest were features significantly enriched in the high or low bins, relative to their respective depleted samples ( 24 ,25 ).Using this approach, 13 and five features were identified in the high and low pools for CRISPR3, respectively (Figure 1 D and Supplementary Table S5 ).For CRISPR4, 29 and 34 hits were identified in the high and low pools respectively (Figure 1 D and Supplementary Table S6 ).When the CRISPR3  S7 .( B ) Locations of Tn 5 insertions in the forward (red) or re v erse (blue) direction enriched in the low bins for the csm , CRISPR3 and CRISPR4 promoters relative to the genomic region containing the alaA and pigU ( lrhA ) genes.Transcriptional start sites (TSS; green) and promoters (blue) are indicated as well as the transposon insertion site in the isolated mutant (HSPIG43; red triangle) ( 29 ).( C ) Mutations in the alaA-pigU intergenic region led to decreased CRISPR-Cas expression.Fluorescence of eYFP was assessed for the csm (pPF1334), CRISPR3 (pPF1923) and CRISPR4 (pPF1924) promoters in either the WT (PCF396) or the alaA-pigU intergenic transposon mutant (Tn mutant, PCF632).Fluorescence normalized by removing empty vector control (pPF1 439 / pPF1 567).Median Fluorescence Intensity = MFI, arbitrary units.Bars are the means and error bars ± SD and individual biological replicates are shown ( n = 5).Statistical significance was assessed using a t wo-t ailed Student's t -test.and CRISPR4 high pools were compared, many hits were in both, with CRISPR3 having a smaller subset (Figure 2 A and Supplementary Table S7 ).Some genes were excluded from further analysis as their effects were ambiguous, being identified as significantly enriched in both high and low pools (flagella genes, and two transcriptional regulators (RefSeq locus tags: RS20795 / pigP ( 29 ) and RS17770).It is possible that some of these might have genuine effects, but that the precise positioning of Tn insertions has led to inverse effects (e.g. via disruption or upregulation of a gene).Many genes enriched in these CRISPR promoter screens had predicted roles in outer membrane synthesis (both the outer leaflet lipopolysaccharide (LPS) layer and phospholipid inner leaflet) and peptidoglycan synthesis.
Other mutations had predicted roles in phospholipid and peptidoglycan synthesis.For example, both screens identified an inner membrane protein that delivers cardiolipin to the outer membrane to increase hydrophobicity in response to outer membrane stress (RS16360 / pbgA ) ( 41 ).Likewise, mutations in asmA (RS17875), which transfers phospholipids from the inner to outer membrane, were uncovered in both CRISPR promoter screens ( Supplementary Figure S2 d) ( 42 ).The CRISPR4 screen also revealed glutamate racemase (RS03635 / murI ) that generates D-glutamate for peptidoglycan and glycerol-3-phosphate dehydrogenase (RS00755 / glpD ), which is involved in central metabolism and phospholipid synthesis.
Interestingly, in the CRISPR4 low pool, mutations in the Rcs stress response pathway were identified (RS00620 / igaA and RS09790 / rcsA ), supporting our previous work with type III-A cas operon regulation ( 25 ).This pathway responds to periplasmic and outer membrane stress, so it may be involved in sensing perturbations caused by the other mutations described above.In addition, genes within the Rsm / Csr pathway (e.g.pigQ ) -and others which influence it (e.g.crp , ptsG (PTS IIC)) -were identified and previously shown to control type III-A CRISPR -Cas ( 25 , 43 ).Full lists of enriched genes are provided in Supplementary Table S5 and S6 .In summary, expression of both type III-A CRISPR arrays responds to changes in outer membrane and peptidoglycan synthesis or composition, and multiple known stress response / signalling pathways also regulate array expression.

Insertions upstream of PigU decrease type III CRISPR and cas expression
We wanted to identify genes that were co-ordinately influencing both CRISPR arrays and the cas ( csm ) operon of the type III-A system.We exploited the power of SorTn-Seq to compare our CRISPR3, CRISPR4 results with our previous csm dataset ( 25 ).This analysis allowed the identification of regulators that were unique or common between the three dif-ferent promoters (Figure 2 A and Supplementary Table S7 ).A general trend of co-ordinate regulation was observed between CRISPR3 and CRISPR4 as discussed above.Five and eight % of all significant loci were shared between all three promoters ( csm , CRISPR3 and CRISPR4) in the high and low pools, respectively (Figure 2 A).We focused on loci that affected the expression of all three promoters.As mentioned above, ambiguous hits (in both high and low pools) were excluded.Mutations in a DNA helicase ( dinG ) and in a pigU-alaA intergenic region were shared between these three SorTn-seq experiments.We decided to focus on the intergenic region between the pigU and alaA genes, since it was enriched for transposon insertions in the low pools for all three promoters (Figure 2 B) and its role in CRISPR-Cas expression was unexplored.AlaA is a glutamate-pyruvate aminotransferase involved in alanine biosynthesis ( 44 ) and PigU is a DNA-binding LysR-type transcriptional regulator (LTTR), that is homologous to Pectobacterium carotovorum HexA ( 45 ), Dicke y a dadantii PecT ( 46) and E. coli LrhA ( 47 ) ( Supplementary Figure S3 ).These LT-TRs have roles in regulating flagella-based motility, secondary metabolites and plant cell-wall degrading enzymes ( 45 , 46 , 48 ).
Previously, a random transposon mutagenesis had identified an insertion in this same alaA-pigU intergenic region that resulted in decreased red pigment (prodigiosin) and carbapenem antibiotic production in Serratia ( 29 ).The position of this transposon insertion was similar to insertions enriched in our SorTn-seq experiments for the different CRISPR-Cas promoters (Figure 2 B; compare red triangle with red and blue lines ).Therefore, to test the effect of mutation in this region on CRISPR-Cas expression, we made a double mutant of this transposon mutant with the pigmentless ( pigA-O ) strain of Serratia used for fluorescence assays.The eYFP reporter plasmids for the cas operon and CRISPR3 and CRISPR4 were introduced into this strain and expression was assessed by flow cytometry.Expression of the csm promoter ( cas operon) and the CRISPR3 and CRISPR4 promoters was lower in the transposon mutant than in the isogenic control (Figure 2 C).Therefore, insertions in the intergenic region between alaA and pigU decrease type III-A CRISPR-Cas expression.

Overproduction of PigU represses type III CRISPR-Cas interference
Since the alaA-pigU intergenic transposon insertions led to decreased expression of the csm , CRISPR3 and CRISPR4 promoters, we predicted that type III-A CRISPR-Cas interference would be lower.Type III-A CRISPR-Cas systems function through sequence-specific RNA targeting that triggers a series of enzymatic activities involved in invader defence ( 49 ).Cas6 generates crRNAs and the Cas protein complex assembles on these guides.The complex recognises and binds complementary RNAs, such as those produced by invader gene expression.RNA recognition triggers the HD domain in Cas10 to cleave ssDNA (e.g.within the transcription bubble) and the Cas10 palm domain catalyses cyclic oligoadenylate (cOAs) formation.These cOA molecules are bound by accessory proteins that are often sequence non-specific RNases and DNases that cleave phage and host nucleic acids ( 49 ).In Serratia , the accessory nuclease is NucC (Figure 1 A), which responds to cA 3 and cleaves dsDNA (50)(51)(52).The Serratia type III system provides protection against plasmids and phages ( 22 , 25 , 43 , 50 , 51 ).To test the effect of the alaA-pigU intergenic mutation on type III-A activity, we used a plasmid conjugation-based CRISPR-Cas interference assay using an untargeted control plasmid, or a plasmid with sequences expressed from an arabinose inducible promoter to generate RNA targets of spacer 1 of CRISPR3 (Figure 3 A) ( 22 ).The untargeted plasmid was acquired efficiently by conjugation in the WT and intergenic Tn mutant, whereas the targeted plasmid was inhibited > 10 000-fold in the WT background (Figure 3 B).In contrast, type III-A interference was completely abolished by the alaA-pigU intergenic transposon mutation (Figure 3 B).Likewise, protection provided by spacer 1 of the CRISPR4 array was similarly reduced in the alaA-pigU intergenic transposon mutant background ( Supplementary Figure S4 a).These results highlight the power of these SorTn-seq screens to identify regulators of physiological relevance.
Despite the clear impact of this mutation on type III-A CRISPR-Cas activity, whether this was related to AlaA or PigU (or other effects) was unknown.Interestingly, in E. coli, insertion of a transposon in the alaA-lrhA intergenic region resulted in overexpression of lrhA ( 47 ).Since PigU homologues regulate diverse functions in different proteobacteria, we hypothesised that the intergenic mutations also led to increased PigU in Serratia , which in turn repressed CRISPR-Cas.In agreement with this model, deletion of pigU in the intergenic transposon mutant background restored interference to this double mutant (Figure 3 B).This suggested that pigU expression is normally low in these growth conditions and that the intergenic mutation has led to a PigU increase, which is acting as a repressor.Indeed, a single mutation (deletion) of pigU had no effect on interference levels (Figure 3 B), presumably due to its low basal expression in these growth conditions.To confirm that PigU overexpression was responsible for decreased CRISPR-Cas activity, recipient cells containing a plasmid expressing pigU were tested in the conjugation efficiency assay.In agreement, PigU overexpression in the WT background abolished type III-A interference (Figure 3 C) and mimicked the phenotype of the alaA-pigU intergenic mutant.Interestingly, this effect did not require any inducer (IPTG) and demonstrated that leaky expression of PigU was sufficient for these strong phenotypic effects.In summary, mutation of the intergenic region leads to increased PigU, which represses type III CRISPR-Cas interference.

PigU represses type I-F CRISPR-Cas plasmid interference and adaptation
We previously identified type III-A CRISPR-Cas system regulators that also controlled the type I CRISPR-Cas systems in Serratia ( 22 , 25 , 43 ).We hypothesised that PigU may also regulate the DNA-targeting type I CRISPR-Cas systems in Serratia .To test this, we performed conjugation-based CRISPR interference assays using either an untargeted control plasmid, or plasmids with sequences targeted by spacer 1 in CRISPR1 (type I-E) or CRISPR2 (type I-F) (Figure 4 A and Supplementary Figure S4 b).In both type I systems, untargeted plasmids were acquired efficiently by conjugation in all strains and targeted plasmids were strongly inhibited in the WT background (Figure 4 B and Supplementary Figure S4 c).Whereas there was no effect on type I-E interference in the strain with elevated PigU ( Supplementary Figure S4 c), type I-F interference was severely attenuated ( > 1000-fold; Figure 4 B).In agreement with the requirement of pigU for repression of type III-A interference, deletion of pigU in the transposon  mutant restored targeting of the type I-F CRISPR-Cas system to WT levels (Figure 4 B).
Since PigU repressed type I interference, we hypothesised that CRISPR adaptation would also be inhibited.To measure CRISPR adaptation, a priming assay was performed using a plasmid that increases new type I-F spacer acquisition, due to a matching spacer but an imperfect protospacer adjacent motif (PAM).Adaptation assays were performed with or without PigU expression from a plasmid in a WT background, and new spacer acquisition was visualised as CRISPR array expansion in the bacterial population by using PCR (Figure 4 C).No adaptation was detected in the un-primed plasmid control, whereas type I-F CRISPR primed adaptation was reduced by PigU expression (Figure 4 D).In conclusion, in addition to its effects on type III-A activity, PigU also represses type I-F CRISPR-Cas interference and adaptation.

PigU controls type I and III-mediated phage resistance
Since PigU repressed both type I and III CRISPR-Cas interference against plasmid uptake by conjugation, we wanted to test if it also affected the phage resistance response.We used strains of Serratia that contained a single spacer that targeted phage JS26 in either the type I-F (CRISPR2) or type III-A (CRISPR3) chromosomal arrays (Figure 5 A).A control vector or a PigU expression plasmid in the WT and these JS26-targeting strains were infected with phage JS26 and their plaquing examined quantitatively (Figure 5 B).JS26 is a siphovirus with a dsDNA genome of ∼64 kb that is sensitive to type I-F and III-A CRISPR-Cas systems in Serratia ( 28 , 50 , 53 , 54 ).The single chromosomal spacers in either the type I-F or type III-A systems reduced the plaquing of JS26 by almost 10 5 -fold, demonstrating a strong CRISPR-Cas response (Figure 5 B).Expression of PigU reduced the strength of CRISPR-Cas immunity by ∼10 3 and ∼10 2 -fold for the type I-F and III-A systems, respectively.Therefore, PigU represses both type I and III CRISPR-Cas immunity against phages.

PigU is a global regulator of secondary metabolism, motility and CRISPR-Cas
To further investigate the role of PigU, RNA sequencing (RNA-seq) was used to identify changes in RNA abundance  between a WT and the alaA-pigU transposon mutant.Both the WT and alaA-pigU mutant showed similar growth profiles and RNA was extracted at early stationary phase (12 h) for sequencing ( Supplementary Figure S5 a).Using DESeq2 ( 39 ), we identified 960 genes ( ∼21% of the genome) having significant differential expression, with 514 upregulated and 446 downregulated in the transposon mutant compared to the WT (Figure 6 A and Supplementary Table S8 ).First, we examined our hypothesis that the transposon mutation led to increased pigU expression.Consistent with our earlier work suggesting that PigU was elevated in the mutant (Figure 3 ), pigU mRNA was ∼1.6-fold higher than the levels in the WT ( Supplementary Table S8 and Supplementary Figure S5 b).Interestingly, genes immediately upstream ( alaA ) and downstream (RS10445) of pigU were also significantly upregulated ( Supplementary Table S8 ), consistent with LTTRs controlling the expression of adjacent genes ( 55 ).Interestingly, in the transposon mutant, there were more reads mapping upstream of the transposon insertion site-located 825 bp upstream of pigU ( Supplementary Figure S5 c).While there are no annotated features in this region, the finding that insertions there in both Serratia and E. coli lead to pigU / lrhA overexpression, and that full lrhA complementation in Pantoea stewartii required ∼921 bp of the intergenic region ( 56 ) suggests there could be important regulatory elements, such as sRNAs, in this location.Furthermore, the levels of the type III-A cas genes for the interference ( csm ) operon were significantly decreased in the mutant, with the exception of cas6 , the last in the transcript (Figure 6 b).The gene encoding the accessory nuclease NucC, which is expressed convergently with the cas operon under a different promoter, was not altered ( Supplementary Table S8 ).We saw a similar lack of nucC regulation under the Rsm system previously ( 43 ).Expression of both type III-A adaptation genes ( cas1 and cas2 ) was significantly lower in the mutant (Figure 6 B).Surprisingly, no significant differences in the type I-F genes were detected ( Supplementary Table S8 ), suggesting that PigU-mediated control of this system involves post-transcriptional regulation.
This alaA-pigU intergenic transposon mutant of Serratia had previously been shown to result in reduced production of two antimicrobial secondary metabolites: the red tripyrrole pigment prodigiosin and the beta-lactam antibiotic, carbapenem ( 29 ).In agreement, RNA-seq showed that expression of several genes in operons encoding these two antimicrobial molecules (prodigiosin; R S20700-R S20630 and carbapenem; R S09910-R S09950) was significantly lower in the transposon mutant ( Supplementary Table S8 ).Since PigU homologues affect motility ( 45 ,48 ), we also examined the regulators ( flhDC ) and structural genes required for flagella biosynthesis.Most genes for flagella-mediated motility and neighbouring chemotaxis genes were increased in the transposon mutant ( Supplementary Table S8 ), consistent with an lrhA mutant transcriptome in E. coli ( 48 ).Serratia also floats using gas vesicles encoded by R S01290-R S01190 ( 57 ).Gas vesicles give Serratia a white opaque phenotype and are inversely controlled with flagella.Consistent with the effects on flagella, the elevated levels of pigU in the transposon mutant led to a translucent colony colour ( 29 ) and the gas vesicle genes were repressed ( Supplementary Table S8 ).In summary, increased PigU levels led to pleiotropic effects, with decreased transcripts for CRISPR-Cas, secondary metabolism and gas vesicles, whereas expression of flagella genes were increased.
Since PigU controls flagella and mutations in flhDC were also identified during SorTn-seq, we investigated whether changes in flagella alter CRISPR-Cas immunity .Previously , we demonstrated that transposon insertion in flhD or flhC led to small decreases in csm expression levels ( 25 ).Interestingly, deletion of either flhD or flhC did not significantly alter type III-A interference against plasmid conjugation ( Supplementary Figure S6 a).To determine if flagella are required for PigU-mediated CRISPR-Cas repression, we overexpressed pigU and performed a conjugation efficiency assay in the presence or absence of flhDC .Conjugation was still severely attenuated upon deletion of flhDC in a PigU overexpression strain, indicating that flhDC is not required for PigU-mediated CRISPR-Cas ( Supplementary Figure S6 b).In addition, overexpression of flhDC in the pigU mutant had no appreciable effect on conjugation, with interference levels similar to that of WT ( Supplementary Figure S6 c).These results indicate that, while PigU controls expression of flhDC ,  S8 ). ( B ) Downregulation of the type III-A cas genes in the alaA-pigU intergenic transposon mutant.Log2 fold change and statistical significance was assessed via DESeq2 ( Supplementary Table S8 ; not significant (ns)).( C ) Upregulation of three predicted Serratia prophages: SP1(R S07685-R S07960; 54 genes), SP2 (R S14105-R S14425; 62 genes) and SP3 (R S14800-R S14975; 35 genes).
these changes in flagella do not contribute to CRISPR-Cas repression.

PigU influences prophage, to xin-antito xin and tRNA gene expression
Surprisingly, some of the most strongly upregulated genes in the transposon mutant were from three prophages-named SP1, SP2 and SP3 ( 26 ), with ∼93% of their genes upregulated (Figure 6 C and Supplementary Table S8 ).These prophages are active under certain conditions, as excision and phage particle release has been previously demonstrated in response to rsmA deletion in Serratia ( 58 ).Prophage induction is often associated with the SOS response, and while recA expression was upregulated ∼2-fold in the transposon mutant, downstream DNA-repair genes such as uvrA / B were not differentially expressed ( Supplementary Table S8 ).Interestingly, overexpression of rcsA , which encodes a component of the r egulator of c apsule s ynthesis (Rcs) stress response, results in RecAindependent E. coli lambdoid prophage induction ( 59 ,60 ).In the transposon mutant, rcsA (RS09790) was upregulated ∼3fold ( Supplementary Table S8 ).Expression of rcsA is directly controlled by the PigU homolog LrhA in P. stewartii ( 61 ).This might suggest that increased PigU in Serratia induces prophages via RcsA, in a pathway distinct from the canonical SOS-response.
Components of seven toxin-antitoxin (TA) systems were also differentially expressed in the transposon mutant ( Supplementary Tables S8 and S9 ) and multiple TA systems are known to provide phage defence ( 62 ).Previously, we characterised the TA repertoire of Serratia, which contains 32 predicted TA systems ( 26 ).In almost all cases, increased pigU expression led to upregulation of TA systems ( Supplementary Table S9 ).We also observed differential expression of 55 tRNA-related genes (45 tRNAs and 10 tRNArelated genes (e.g.synthases / ligases / methyltransferases).In nearly all cases (51 / 54 genes), expression was upregulated ( Supplementary Table S8 ).Whether tRNAs may be induced in response to TA system activation or prophage induction (or both) is unknown.In summary, the alaA-pigU transposon mutation resulted in increased prophage expression and elevation of TA systems.

PigU controls expression of sRNAs that regulate rpoS translation
Interestingly, upon pigU overexpression, we also observed decreased transcripts for rpoS , which encodes the stationaryphase sigma factor RpoS ( σ S ) ( Supplementary Table S8 ).RpoS levels are regulated at transcriptional, translational, and post-translational levels, and involves non-coding small RNAs (sRNAs) ( 63 ).The sRNAs RprA and ArcZ, assisted by the post-transcriptional chaperone Hfq, bind the rpoS 5 UTR to favour secondary structures that allow translation ( 63 ).In Serratia, pigU overexpression downregulated rprA ( Supplementary Figure S7 a-S7 b, Supplementary Table S8 ), consistent with E. coli ( 64 ).A homology search revealed a putative arcZ homolog in Serratia , which was downregulated by PigU ( Supplementary Figure S7 c-S7 d).While ArcZ controls translation of the PigU homologs PecT and HexA in Dicke y a dadantii and Photorhabdus / Xenorhabdus spp.( 65 ,66 ), our data suggests that the inverse (PigU regulation of arcZ) occurs in Serratia.Interestingly, expression of another sRNA, CyaR, which is negatively regulated by ArcZ ( 67 ), is upregulated upon pigU overexpression in Serratia ( Supplementary Figure S7 e-S7 f).CyaR represses rpoS translation, and the upregulation upon pigU overexpression may be a result of lowered transcription of arcZ ( Supplementary Figure S7 c-S7 d).In E. coli , it is thought that LrhA-mediated control of rpoS translation likely requires another Hfq-dependent sRNA (in addition to RprA ( 64 )), thus ArcZ is a candidate worth investigating.
To test whether rpoS influences CRISPR-Cas immunity, we performed type III-A conjugation interference and phage infection assays with an rpoS mutant ( 68 ).There was a slight ( ∼5-fold) decrease in type III-A plasmid targeting in the rpoS mutant ( Supplementary Figure S7 g), a more subtle phenotype than that of the pigU overexpression mutant (Figure 3 B).For phage infection, there was no detectable difference in protection in a strain harbouring a type III-A anti-phage spacer upon mutation of rpoS ( Supplementary Figure S7 h).Interestingly, the rpoS mutation affected phage JS26 plaque phenotype ( Supplementary Figure S7 i).These results suggest that RpoS levels do not substantially contribute to CRISPR-Cas regulation, and the strong repression observed upon pigU overexpression is due to an alternative pathway.

PigU inhibition of CRISPR-Cas is predominantly independent of known CRISPR-Cas regulators
We investigated if increased PigU was acting through other known CRISPR-Cas regulators to influence adaptive immunity in Serratia ( Supplementary Figure S8 a).RsmA (CsrA) is a post-transcriptional regulator which represses CRISPR-Cas ( 43 ).Levels of free RsmA are modulated by the sRNA RsmB (CsrB) and PigX (CsrD), which degrades RsmB.Upon pigU overexpression, rsmB was elevated ( Supplementary Figure S8 b, c) and csrD was decreased ( Supplementary Table S8 ).These conditions would result in greater RsmA sequestration and more Cas translation due to decreased repression of target cas gene mRNAs ( 43 ).Therefore, the CRISPR-Cas repression mediated by PigU cannot be explained through effects on the Rsm pathway.QS also regulates CRISPR-Cas in Serratia ( 22 ) and the PigU homolog HexA in E. carotovora activates production of the QS signal ( 69 ).However, in Serratia , PigU did not influence transcript levels from the smaIR QS genes ( Supplementary Table S8 ), or affect QS signal levels ( 29 ).Finally, both PigU and the Rcs pathway, a previously characterized regulator of CRISPR-Cas in Serratia ( 43 ), regulate transcription of the small RNA rprA ( 70 ).In Serratia, rprA expression is increased during Rcs activation ( 43 ) and decreased upon pigU overexpression ( Supplementary Figure S7 a, Supplementary Table S8 ).Deletion of rprA or rpoS ( Supplementary Figure S7 e) had no ( 43 ), and very little ( Supplementary Figure S7 g), effect on CRISPR-Cas immunity , respectively .In summary , although various CRISPR-Cas regulators are influenced by PigU upregulation, PigU is likely acting through independent pathways to repress adaptive immunity.

Discussion
Previously, we developed SorTn-seq to systematically identify regulators of bacterial gene expression and used it to identify factors which influence type III-A cas ( csm ) expression in Serratia ( 25 ).Here, we performed SorTn-seq of both type III-A CRISPR arrays (CRISPR3 and CRISPR4) to produce a comprehensive analysis of type III-A CRISPR-Cas regulation in Serratia .We discovered several mutations which affect CRISPR expression, including those which likely alter outer membrane / cell surface structures including LPS, O-antigen, and peptidoglycan.Interestingly, many of these genes were also enriched during cas ( csm ) SorTn-seq.While the exact mechanism of regulation remains to be elucidated, these findings indicate that changes in outer membrane biogenesis can lead to coordinate changes from both CRISPR and cas ( csm ) gene promoters.
Of particular interest were mutations mapping to the intergenic region between alaA and pigU that were enriched in the low expression bins of all three type III-A SorTnseq screens.Overexpression of pigU suppressed both type III-A and type I-F CRISPR-Cas targeting, reducing protection against both plasmids and phage.PigU repressed adaptation by the type I-F system, thus, all stages of CRISPR-Cas immunity-expression, interference and adaptation-are controlled by PigU.Although type III adaptation has not been observed in Serratia , expression of genes encoding the adaptation machinery (type III-A cas1 and cas2 ) also falls under PigU control-suggesting PigU may also influence type III spacer acquisition.Interestingly, type I-E immunity was not affected, indicating that the three CRISPR-Cas systems in Serratia are not always co-ordinately regulated.In agreement with the importance of PigU as a regulator of CRISPR, E. coli LrhA was recently shown to activate type I-E CRISPR-Cas immunity in an hns / leuO double mutant through binding to the cas promoter ( 71 ).
In addition to its role as a CRISPR-Cas repressor, RNA-seq revealed that PigU is also a pleiotropic regulator, controlling genes involved in secondary metabolism and prophage / TA system induction.Interestingly, six of eight genes in a predicted LPS and O-antigen biosynthesis operon that were enriched in the SorTn-seq ( Supplementary Figure S2 c) were also differentially expressed in the transposon mutant ( Supplementary Table S10 ), suggesting some shared regulatory pathways.However, the cellular cues which control pigU expression remain largely unknown.In P. stewartii, the QS master regulator EsaR directly activates lrhA expression ( 72 ,73 ).Therefore, similarly, pigU expression in Serratia may be under QS control, but this currently unknown.In E. coli, mutation of ftsK , encoding a DNA translocase involved in chromosome segregation and cell division ( 74 ) leads to lrhA overexpression, which influences levels of the stationary-phage sigma factor RpoS ( 64 ,75 ).It is likely that some of the global transcriptomic shifts observed upon pigU overexpression are a result of decreased levels of RpoS.However, in Serratia , these changes do not contribute substantially to CRISPR-Cas immunity.Likewise, PigU appears to repress adaptive immunity in Serratia independently of known CRISPR-Cas regulators, although significant regulatory interconnection exists ( Supplementary Figure S8 ).
Overall, we have demonstrated the utility of the SorTn-seq method to identify regulators of non-coding RNA (CRISPR array) expression.We have uncovered a major regulator of type III CRISPR-Cas that co-ordinately controls cas operon and CRISPR array expression to influence CRISPR-Cas interference and adaptation.Importantly, our study further demonstrates that CRISPR-Cas systems are often 'wired' into major conserved regulatory networks that have pleiotropic effects on cellular processes.The type III-A system in Serratia is present in an integrative and conjugative element, and is likely the most recent CRISPR-Cas system acquired by this strain.Type III-A regulation by QS ( 22 ), Rcs ( 25 ), Rsm ( 43 ) and now LrhA (PigU), support the idea that following this horizontal gene transfer event, this defence system has connected into the existing regulatory networks that co-ordinate diverse cellular behaviours.

Data availability
Raw reads from SorTn-seq analyses and RNA-seq have been deposited at the NCBI Sequence Read Archive (SRA) under BioProject accession number PRJNA1020058.Processed RNA-seq data have been deposited at the NCBI Gene Expression Omnibus (GEO) under accession number GSE243790.Data analysis pipeline and scripts used for SorTn-seq analyses are available at http:// doi.org/ 10.5281/ zenodo.4554398( 31 )

Figure 1 .Figure 2 .
Figure 1.SorTn-seq re v eals mutations in genes that significantly impact the Serratia CRISPR3 and CRISPR4 promoters.( A ) Schematic of the type III-A CRISPR-Cas locus in Serratia .Interference genes are indicated in pink, adaptation genes in mau v e and promoters with black arrows.The CRISPR arra y s are depicted with repeats (diamonds) and spacers (coloured rectangles).CRISPR3 and CRISPR4 contain nine and eight spacers, respectively.( B ) Promoter elements ( −35 and −10) and transcriptional start sites (+1) for both CRISPR3 and CRISPR4.Details of the reporter vectors used in the SorTn-seq experiments are shown in Figure S1a-d.( C ) The Serratia genome displaying transposon insertion density for both CRISPR3 and CRISPR4 experiments (outer ring) and genes on the f orw ard (blue) and re v erse (gre y) strands.Numbers of unique insertions per replicate and in the high , low and depleted samples is provided in Figure S1e.(kb = kilobase, Mb = megabase, ins.= insertions).( D ) Significant enrichment of mutants in the low and high samples compared with the depleted controls, as determined via exact test for differential expression using edgeR .Horizontal lines indicate a P value of 0.05 (adjusted for multiple comparisons using Benjamini and Hochberg false discovery rate correction); vertical lines indicate a log 2 fold change of 0.5.Complete data sets of significantly enriched genes in the CRISPR3 and CRISPR4 SorTn-seq screens are provided in Supplementary TablesS5 and S6, respectively.

Figure 3 .
Figure 3. PigU represses type III-A CRISPR-Cas interference.( A ) Schematic of the conjugation efficiency CRISPR-Cas interference assa y s.Plasmids with (targeted) or without (untargeted) a protospacer matching spacer 1 in the particular CRISPR array are conjugated into Serratia strains containing the native CRISPR-Cas systems.Interference is assessed by measuring conjugation efficiency by plating with and without selection for the transconjugants and recipients.( B ) CRISPR conjugation interference assay of an untargeted plasmid (pPF781) or a plasmid targeted (pPF1043) by CRISPR3 spacer 1 in the type III-A system in WT (PCF396), an alaA-pigU intergenic transposon mutant (Tn mutant, PCF632), a pigU mutant ( pigU , PCF708) and an alaA-pigU intergenic, pigU double mutant (Tn mutant pigU , PCF709).Statistical significance of conjugation efficiency (targeted strains) assessed using a one-w a y ANO V A ( P < 0.0 0 01) with Dunnett's multiple comparisons test against the WT (PCF396).( C ) CRISPR conjugation interference assay as in (B) in WT (LacA) either containing a vector control (pQE-80LoriT; Control) or a plasmid expressing PigU (pPF1983; PigU).Statistical significance was assessed using a t wo-t ailed Student's t -test of untargeted versus targeted.In (B) and (C), bars are the means and error bars ± SD.Individual biological replicates are shown for (B) ( n = 6) and (c) ( n = 3).

Figure 4 .
Figure 4. PigU represses type I-F CRISPR-Cas interference and adaptation.( A ) Schematic of the type I-F CRISPR-Cas locus in Serratia .

Figure 5 .
Figure 5. PigU decreases phage immunity of both the type I-F and III-A CRISPR-Cas systems.( A ) Schematic of the experimental setup.Strains of Serratia with either a control (pQE-80LoriT) or PigU expression (pPF1983) plasmid contained no additional spacer (LacA) or an additional single spacer targeting phage JS26 at the leader-end of the CRISPR arra y s f or t ype I-F (CRISPR2; PCF525) or t ype III-A (CRISPR3; PCF925) sy stems.T hese strains were infected with phage JS26 and plaquing assessed.( B ) Representative spot titre results of phage JS26 on the WT control strain and strains with a type I-F or III-A spacer targeting the phage in the presence or absence of PigU expression ( left ).Quantification of the efficiency of plating (EOP) relative to the WT with the control vector.Bars are the means and error bars ± SD and individual biological replicates are shown ( n = 3).Statistical significance was assessed using t wo-t ailed Student's t -tests of strains carrying control plasmid (pQE-80LoriT) versus the PigU expression plasmid (pPF1983).
. Reference sequences and annotations used in this study are available through NCBI: Serratia sp.ATCC 39006 LacA (CP025085.1