A Robust CRISPR Interference Gene Repression System in Pseudomonas

ABSTRACT

Pseudomonas spp. are widely used model organisms in different areas of research. Despite the relevance of Pseudomonas in many applications, the use of protein depletion tools in this host remains limited. Here, we developed the CRISPR interference system for gene repression in Pseudomonas spp. using a nuclease-null Streptococcus pasteurianus Cas9 variant (dead Cas9, or dCas9). We demonstrate a robust and titratable gene depletion system with up to 100-fold repression in β-galactosidase activity in P. aeruginosa and 300-fold repression in pyoverdine production in Pseudomonas putida. This inducible system enables the study of essential genes, as shown by ftsZ depletions in P. aeruginosa, P. putida, and Pseudomonas fluorescens that led to phenotypic changes consistent with depletion of the targeted gene. Additionally, we performed the first in vivo characterization of protospacer adjacent motif (PAM) site preferences of S. pasteurianus dCas9 and identified NNGCGA as a functional PAM site that resulted in repression efficiencies comparable to the consensus NNGTGA sequence. This discovery significantly expands the potential genomic targets of S. pasteurianus dCas9, especially in GC-rich organisms.

IMPORTANCE Pseudomonas spp. are prevalent in a variety of environments, such as the soil, on the surface of plants, and in the human body. Although Pseudomonas spp. are widely used as model organisms in different areas of research, existing tools to deplete a protein of interest in these organisms remain limited. We have developed a robust and inducible gene repression tool in P. aeruginosa, P. putida, and P. fluorescens using the Streptococcus pasteurianus dCas9. This method of protein depletion is superior to existing methods, such as promoter replacements and addition of degradation tags, because it does not involve genomic modifications of the target protein, is titratable, and is capable of repressing multiple genes simultaneously. This gene repression system now enables easy depletion of specific proteins in Pseudomonas, accelerating the study and engineering of this widely used model organism.

INTRODUCTION

Pseudomonas spp. are Gram-negative bacteria that are prevalent in a variety of environments, such as the soil, on the surface of plants, and in the human body. For example, Pseudomonas aeruginosa is an opportunistic pathogen that is often responsible for hospital-acquired infections and is the leading cause of morbidity in cystic fibrosis patients (1). Pseudomonas putida is a soil microbe that has been used for bioremediation of contaminated land as well metabolic engineering for the production of value-added products (2, 3). Pseudomonas fluorescens is found in the rhizosphere and on the surface of plants and has been engineered for industrial production of proteins (4). Even though Pseudomonas spp. are widely used as model organisms for several areas of research, tools to deplete specific proteins in these hosts have not been widely adopted. Three techniques have been used to study the cellular response to protein depletion in these bacteria. First, the promoter of the targeted gene can be replaced with an inducible promoter, either directly on the chromosome at the native location or from plasmid expression and deletion of the native gene (5). For essential genes, the cells must be grown with induction, and then the cells are washed to remove the inducer and deplete the protein. Second, the gene of interest can be tagged with a ClpXP recognition peptide, which targets the protein for ATP-dependent protein degradation by the ClpX machinery. Induction of the sspB gene then degrades the tagged protein and results in decreased protein amounts (6). Finally, transposon mutant libraries exist for both P. aeruginosa and P. putida that have enabled an assessment of essential genes and investigations into the physiology of nonessential genes (7, 8). However, the timing and dynamics of protein depletion cannot be controlled in these strains, limiting the utility of the transposon mutant libraries. More recently, the native type I CRISPR/Cas system has been shown to be able to act as a transcriptional repressor in P. aeruginosa PA14; however, a cas3 deletion background or the expression of anti-CRISPR proteins from prophage is required (9).

The type II CRISPR-Cas9 system has been reengineered for gene repression by creating a nuclease-null Streptococcus pyogenes Cas9 (dead Cas9, or dCas9) through single point mutations in both the RuvC1 and HNH domains (10). A chimeric single-guide RNA (sgRNA) that contains a 20-bp complementary sequence to the target DNA is sufficient to direct dCas9 to the DNA sequence of interest (11). Recruitment of dCas9 blocks transcription initiation when targeted to the promoter region or disrupts transcription elongation when targeted to the noncoding strand of the gene of interest, a method that has been termed CRISPR interference (CRISPRi) (Fig. 1A). CRISPRi with the widely used dCas9 from S. pyogenes provides robust repression in hosts such as Escherichia coli (10), Bacillus subtilis (12), and mammalian cells (13). However, in mycobacteria, repression by the S. pyogenes dCas9 was only 6.8-fold while that by the Streptococcus thermophilus CRISPR1 dCas9 was 166-fold, indicating that the performance of dCas9 variants may be host dependent (14).

FIG 1.

CRISPRi provides robust repression in P. aeruginosa. (A) The CRISPR interference system consists of S. pasteurianus dCas9 that is directed to the target DNA by an sgRNA. Binding of dCas9 to target DNA prevents transcription by blocking RNA polymerase. (B) CRISPRi expression vectors: pBx-Spas-sgRNA for constitutive expression of sgRNA and pUC18-mini-Tn7T-Lac-dCas9 for genomic integration of dCas9. (C) NADP-dependent glutamate dehydrogenase activity of P. aeruginosa cells harboring sgRNA targeting the promoter region of gdhA with its respective PAM site. Relative mRNA levels of gdhA compared to those with an empty sgRNA control of the same cultures are also shown. Fold repression (3× and 5×) is indicated on the graph. (D) Measurements of β-galactosidase activity as a function of IPTG concentrations, reported as fold change compared to the level of an empty sgRNA control. lacZ is expressed from the genome with the promoter of gdhA. Bent arrows indicate transcriptional start sites of promoters.

The Streptococcus pasteurianus Cas9 was first characterized using in vitro DNA cleavage assays in which the consensus protospacer adjacent motif (PAM) sequence was found to be NNGTGA (15). Recently, the S. pasteurianus dCas9 was tested for gene repression in Mycobacterium smegmatis, and a 10-fold gene knockdown was observed, comparable to that of S. pyogenes dCas9 in the same host (14). Here, we developed a system for CRISPRi gene repression in Pseudomonas using S. pasteurianus dCas9. We demonstrate robust gene repression of up to 100-fold in β-galactosidase activity in P. aeruginosa and 300-fold in pyoverdine production in P. putida. Repression of essential genes in P. aeruginosa, P. putida, and P. fluorescens resulted in dynamic inhibition of growth and phenotypes consistent with the targeted depletion. In addition, we confirmed that the noncanonical PAM site NNGCGA was functional in vivo, significantly expanding the potential genomic targets of S. pasteurianus dCas9.

RESULTS

Construction of the CRISPRi system.

To develop a system for CRISPRi in Pseudomonas aeruginosa, we utilized the type II cas9 ortholog from S. pasteurianus, which was previously found to function as a transcriptional repressor in M. smegmatis and was less toxic than the canonical S. pyogenes dCas9 (14). To enable dynamic control of dcas9, the gene was placed under the control of the Ptet promoter on the pUC18-mini-Tn7 vector. This vector carries the mini-Tn7 element that integrates into the genome of Gram-negative bacteria at a defined locus, attTn7 (16). Genomic integration of dcas9 was preferred as recombinant expression from a single chromosomal integration enables tighter transcriptional control and increased strain stability (17). Initial testing of this construct showed that it was capable of targeted gene repression: a reduced swimming radius of P. aeruginosa was observed when the motility gene flgB was targeted. However, in the absence of the inducer anhydrotetracycline (aTc), the target gene remained repressed, indicating leaky expression from the Ptet promoter in P. aeruginosa. This was confirmed by Western blotting against the hemagglutinin (HA) tag on the N terminus of dCas9 (see Fig. S1 in the supplemental material). Testing of this same construct in P. putida also showed similar leaky expression of dcas9 (Fig. S2).

To obtain tighter control of dcas9 expression, the TetR-Ptet repressor-promoter pair was replaced with the LacI-Ptac repressor-promoter by cloning S. pasteurianus dcas9 into the pUC18-mini-Tn7T-Lac vector (16) (Fig. 1B). In contrast to the Ptet promoter, this construct was capable of dynamic gene repression when it was used to target the essential gene ftsZ in P. putida and P. fluorescens. In P. aeruginosa, however, dcas9 expression from the Ptac promoter remained leaky since targeting ftsZ resulted in growth inhibition even in the absence of the inducer isopropyl-β-d-1-thiogalactopyranoside (IPTG). By testing variants of the lac promoter that have been shown to exhibit tighter repression (18) (Table S1) with the same qualitative assay of observed swimming radius, we found Plac to provide the tightest transcriptional control of dcas9 expression in P. aeruginosa. Interestingly, Ptac still provided a higher dynamic range than Plac in P. putida and P. fluorescens, indicating that differences in transcriptional control between these closely related bacteria are not always generalizable. The S. pasteurianus sgRNA is expressed from a Ptet promoter on a broad-host-range pBBR1 plasmid, pBx-Spas-sgRNA (Fig. 1B). The sgRNA construct was designed with BbsI sites flanking the 20-bp target sequence to enable easy cloning using annealed oligonucleotides by Golden Gate assembly (see Materials and Methods) (19). We also tested different lac promoter variants for sgRNA expression to further improve the range of CRISPRi repression (Table S1); however, only Ptet resulted in targeted gene repression.

CRISPRi targeting of nonessential genes in P. aeruginosa.

To characterize the efficiency of the CRISPRi system, we targeted endogenous genes that were not in an operon to prevent pleiotropic effects from polar mutations that could affect cell physiology. First, the NADP-dependent glutamate dehydrogenase gene in P. aeruginosa, gdhA, was targeted for depletion at three different locations with PAM sites of NNGTGA or NNGCGA (Fig. 1C). After 5 h of dcas9 induction, enzyme assays were performed on crude cell extracts of P. aeruginosa to measure GdhA activity. Up to 5-fold repression in glutamate dehydrogenase activity was obtained between the induced and uninduced state with sgRNA g3, which targets the region between −10 and −35, with an NNGCGA PAM site. Compared to an empty sgRNA control, which possessed the sgRNA handle and BbsI recognition sites in place of the 20-bp target sequence, however, sgRNA g3 also resulted in reduced glutamate dehydrogenase activity when it was uninduced due to leaky expression of dcas9 from the Plac promoter. Comparing sgRNA g1 and g2 that target both NNGTGA PAM sites showed that sgRNA g2, which targets between the transcriptional start site (TSS) and start codon, resulted in 3-fold repression while sgRNA g1, which targets between the −10 position and TSS, resulted in negligible repression. The difference between the repression efficiencies of the three sgRNAs alludes to the effects of PAM site and target location on the efficiency of CRISPRi repression, which we address in later sections. Since the CRISPRi depletion system functions at the level of transcription, we used reverse transcription quantitative PCR (RT-qPCR) to quantify the mRNA abundance after depletion (Fig. 1C). Quantification of gdhA mRNA showed a 10- to 100-fold reduction in mRNA with sgRNA g2 and g3 compared to the level with an empty sgRNA control, indicating that the decreased enzyme activity was due to reduced transcription of the gdhA gene. The discrepancy in the fold repression observed between the mRNA and enzyme activity levels may be attributed to posttranscriptional regulation of glutamate dehydrogenase. In M. smegmatis, for example, regulation of the NADP-dependent glutamate dehydrogenase activity under nitrogen availability was also not reflected at the level of gene transcription, implicating posttranscriptional regulation mechanisms (20).

We next sought to build a more sensitive reporter assay that still represented targeting of endogenous genes. We designed a lacZ construct that was transcribed from the native gdhA promoter and integrated it into the genome at the ϕCTX attachment site using the mini-CTX1 vector (21) (Fig. 1D). An sgRNA targeting the region between the −10 and −35 positions of the gdhA promoter with an NNGCGA PAM site was designed, and P. aeruginosa cells harboring this sgRNA were induced with several different IPTG concentrations. After 5 h of CRISPRi induction, β-galactosidase activity assays were performed on crude cell extracts, and the fold change in activity relative to the level of an empty sgRNA control was calculated (Fig. 1D). The CRISPRi system is titratable, with up to a 100-fold decrease in activity observed with the highest IPTG concentration of 50 mM. This notably high concentration of IPTG, however, had a minimal effect on the growth of P. aeruginosa (Fig. S3). The prevalence of efflux pumps in Pseudomonas (22) and possible limitations in the uptake of IPTG (23) may explain the requirement for high IPTG concentrations for maximum repression. However, IPTG has been shown to inhibit o-nitrophenyl-β-d-galactopyranoside (ONPG) hydrolysis, which is the readout for β-galactosidase activity (24). To confirm that carryover of IPTG from the cell cultures to the enzyme assay was not responsible for this apparent decrease in β-galactosidase activity, we cultured P. aeruginosa cells harboring an empty sgRNA control at increasing IPTG concentrations of up to 50 mM. Assays from crude cell extracts showed only small decreases in β-galactosidase activities, indicating that carryover of IPTG was not responsible for the decreased activity observed with the on-target sgRNA (Fig. S4).

sgRNA design rules for transcriptional repression.

The ability to repress the transcription of a gene of interest with the CRISPRi system is dependent upon the availability of PAM sites within the gene and promoter region. The consensus PAM sequence for S. pasteurianus Cas9 was reported to be NNGTGA based on an in vitro DNA cleavage assay (15). By taking the top two bases identified in the DNA cleavage assay at positions 4 to 6, we determined a degenerate PAM site of NNG(T/C)(G/A)(A/T); however, there has been no comprehensive comparison of the different PAM sites with respect to the degree of CRISPRi repression. To investigate the effects of the PAM site and its location within the promoter on the degree of CRISPRi repression, we modified the endogenous gdhA promoter from P. aeruginosa such that the four PAM sites NNGTGA, NNGCGA, NNGTAA, and NNGTGT are present at three different regions within the promoter: (i) upstream of −35, (ii) between −10 and −35, and (iii) between −10 and the TSS (Fig. 2A). The −10 and −35 elements annotated here were predicted using the tool BPROM (Softberry) (25); however, we later found reports on the −10 and −35 elements that are slightly upstream of the prediction (26). Despite this difference, the −10 and −35 hexamers of the modified promoter were the same as both the predicted and reported sequences, including the spacing between these two regions (Fig. S5). This modified promoter was then used to drive expression of lacZ and was integrated into the genome at the ϕCTX attachment site using a mini-CTX1 vector. Twelve sgRNAs targeting each PAM site at each location were designed, and β-galactosidase activity was measured with and without IPTG induction (Fig. S6 and S7).

FIG 2.

Characterization of the effects of PAM site variants and target locations on S. pasteurianus CRISPRi repression efficiency in P. aeruginosa. (A) Design of β-galactosidase reporter in which all four PAM variants, NNGTGA, NNGCGA, NNGTAA, and NNGTGT, are present at all three locations, A, B, and C, within the gdhA promoter. (B) Fold change of β-galactosidase activity when induced with 1 mM IPTG compared to that when uninduced in P. aeruginosa cells harboring sgRNAs targeting the specific PAM site and location. (C) Fold change of β-galactosidase activity compared to that with an empty sgRNA control in P. aeruginosa cells harboring sgRNAs targeting inside the ORF of lacZ. sgRNA g1 and g4 target the nontemplate strand (NT), while sgRNA g2 and g3 target the template (T) strand. Expression of lacZ is driven by the same modified gdhA promoter as described for panel A.

Comparing the fold change in activity of the induced condition to that of the uninduced condition for the 12 sgRNAs showed a few trends (Fig. 2B). First, the consensus PAM sequence NNGTGA provided the highest fold repression when locations A and B within the promoter were targeted. The PAM site NNGCGA resulted in repression comparable to that obtained with the consensus PAM sequence when location A was targeted, the first in vivo demonstration that this PAM site is functional. This finding is valuable because high-GC-content bacteria are more likely to possess this PAM site than the consensus sequence. The PAM site NNGTAA resulted in modest repression when locations B and C were targeted, while the PAM site NNGTGT was nonfunctional. In terms of target location, locations A (upstream of −35 sequence) and B (between the −10 and −35 sequences) resulted in more effective repression than location C (between −10 and the TSS). Therefore, in designing sgRNA targets, we recommend a PAM site preference of NNGTGA > NNGCGA > NNGTAA and targeting locations upstream of the −35 or between the −10 and −35 regions.

Gene depletion by targeting inside ORFs in P. aeruginosa.

In addition to preventing transcriptional initiation by targeting dCas9 to a promoter, CRISPRi was shown to cause repression by preventing transcription elongation in bacteria. Targeting the nontemplate strand inside the open reading frame (ORF) of a gene provided 10- to 300-fold repression in E. coli with S. pyogenes dCas9, with higher repression when a region close to the start codon was targeted (10). In M. smegmatis, up to 100-fold repression was observed with S. thermophilus dCas9 when a region up to bp +150 inside the ORF was targeted (14). To investigate the effectiveness of the S. pasteurianus dCas9 in gene repression when regions inside ORFs of genes in P. aeruginosa were targeted, we utilized the lacZ construct described above, where a modified gdhA promoter was used to drive lacZ expression from the genome. When targeting the PAM site NNGTGA on the nontemplate strand with sgRNA g1 and g4, we observed 3- to 5-fold repression in β-galactosidase activity (Fig. 2C). Blocking transcription elongation closer to the start codon with sgRNA g1 resulted in higher fold repression, consistent with results shown by other investigators (10, 14). Additionally, both sgRNA g2 and g3 that targeted the template strand did not provide significant repression compared to an empty sgRNA control. The maximum fold repression obtained when a region within an ORF was targeted was comparable to that when the promoter was targeted, albeit slightly lower (6- to 7-fold in the experiment shown in Fig. 2B compared to 3- to 5-fold in the experiment shown in Fig. 2C, both with 1 mM IPTG; the 100-fold repression in the experiment shown in Fig. 1D was attained with 50 mM IPTG). Interestingly, leaky repression under the uninduced condition when the region within the ORF was targeted was generally lower than that when the promoter region was targeted.

Depletion of essential gene targets in P. aeruginosa, P. putida, and P. fluorescens.

Next, we sought to extend the application of this system to other species in the Pseudomonas genus. To demonstrate that this system was capable of depleting gene expression in an inducible manner not achievable by conventional knockouts, we chose to target the essential cell division protein ftsZ in P. aeruginosa, P. putida, and P. fluorescens (Fig. 3A). The ftsZ gene is on a single transcriptional unit in P. putida and P. fluorescens; however, the gene is first in a three-gene operon in P. aeruginosa. The sgRNAs designed to target ftsZ and the corresponding PAM sites are shown in Fig. 3A. For all three Pseudomonas species, spotting cells on plates with IPTG to deplete ftsZ resulted in no growth, while spotting on uninduced plates resulted in greater than three orders of magnitude more colonies (Fig. 3B). To confirm the ftsZ depletion phenotype, we cultured cells harboring ftsZ sgRNA under uninduced conditions and then induced gene depletion with IPTG for 2 h. Differential interference contrast microscopy of these cells showed filamentous growth with much longer cells than those with an empty sgRNA control (Fig. 3C), consistent with the established ftsZ depletion phenotype (27). Taken together, we demonstrate that this system enables the inducible depletion of essential genes in each of the three species of Pseudomonas.

FIG 3.

CRISPRi targeting of essential genes in P. aeruginosa, P. putida, and P. fluorescens. (A) sgRNAs designed to target the promoter or inside the ORF of ftsZ genes, with the corresponding PAM sites. The −10 and −35 sequences of ftsZ in P. fluorescens are unknown. (B) Growth plates of Pseudomonas strains harboring the sgRNA targeting ftsZ or an empty control sgRNA, spotted as 10-fold dilution series on plates with and without 1 mM IPTG. (C) Differential interference contrast microscopy of Pseudomonas strains harboring the sgRNA targeting ftsZ or an empty control sgRNA after induction of gene depletion with 1 mM IPTG for 2 h.

To demonstrate that this CRISPRi system also enables the study of genes with poorly understood functions, we targeted ftsJ, which was identified as essential based on the absence of mutants observed in transposon sequencing (Tn-seq) (28). However, the E. coli ortholog of ftsJ, which has been characterized as a 23S rRNA methyltransferase gene, is not essential although mutant cells grow slowly (29). Consistent with the predicted essentiality, no growth was observed when ftsJ was targeted with CRISPRi and spotted onto plates with IPTG (Fig. S8). Differential interference contrast microscopy of cells with CRISPRi induced for 2 h in liquid culture showed cells linked with constricted membranes that are morphologically different from the filamentous cells with the ftsZ depletion (Fig. S8). Similarly, CRISPRi depletion of the cell division protein encoded by zipA and a putative DNA replication initiation factor encoded by PA0947 in P. aeruginosa resulted in no colony growth (data not shown), consistent with the determination that these are essential genes in P. aeruginosa (28).

Robust repression of single and multiple genes in P. putida.

Next, we sought to multiplex the CRISPRi system and demonstrate that two nonessential genes could be simultaneously depleted in P. putida. First, we characterized the depletion of the endogenous gene pvdH, which is an aminotransferase involved in pyoverdine synthesis. Pyoverdine is a fluorescent siderophore secreted by many species of Pseudomonas and is known to be involved in the regulation of virulence factors in P. aeruginosa (30). Pyoverdine concentrations can be estimated by fluorescence measurements, making it a quantitative and physiologically relevant target for depletion with the CRISPRi system. Two sgRNAs targeting the promoter region and ORF of pvdH, with the PAM site NNGCGA, were designed and individually tested (Fig. 4A). Up to 300-fold repression in pyoverdine production was achieved with sgRNA g1 that targets between the −10 region and the TSS. Targeting inside the ORF with sgRNA g2 resulted in 16-fold repression. Similar to our observation with the CRISPRi system in P. aeruginosa, low-level repression in the uninduced state was observed, likely due to leaky dcas9 expression from the Ptac promoter.

FIG 4.

Robust CRISPRi repression of single and multiple genes in P. putida. (A) Pyoverdine production, measured as absorbance at 405 nm normalized to the OD₆₀₀ of P. putida cells harboring sgRNAs targeting the promoter and inside the ORF of pvdH. (B) Pyoverdine production and motility plates of P. putida harboring sgRNAs targeting both pvdH and flgB compared to that with an empty sgRNA control. Cultures were induced with 2 mM IPTG.

For the simultaneous repression of two genes, sgRNAs were expressed from separate Ptet promoters on the same pBx-Spas-sgRNA plasmid. We chose to simultaneously deplete flgB, which encodes the flagellar basal body rod protein, and pvdH in P. putida as shown in the schematic in Fig. 4B. Cells with the double flgB-pvdH sgRNAs showed reduced swimming radius when plated on soft agar plates with 1 mM IPTG, indicating reduced motility that is consistent with an flgB depletion phenotype. A liquid culture of the double depletion showed a 90-fold reduction in pyoverdine production between the induced and uninduced conditions in contrast to the 300-fold repression observed in the experiment shown in Fig. 4A when only sgRNA g1 was expressed. The difference in repression levels could be attributed to the decreased transcription of the pvdH sgRNA resulting from its being the downstream sgRNA in the double flgB-pvdH construct. Nonetheless, the CRISPRi system enables the downregulation of multiple genes simultaneously, broadening its applicability as a gene repression tool and facilitating gene-gene interaction studies.

DISCUSSION

In Pseudomonas spp., existing methods to deplete specific proteins of interest have involved expressing the targeted gene with an inducible promoter (5) or adding recognition peptides for ClpXP-dependent protein degradation (6). Compared to replacing the native promoter of the targeted gene with an inducible promoter, the ClpXP degradation method has the advantage that the gene remains under the control of its native promoter, thereby retaining native expression levels until sspB expression is induced. However, an ΔsspB mutant needs to first be generated for the degradation system to be inducible. In contrast, the CRISPRi system allows depletion of specific proteins once dCas9 is integrated into the Tn7 site, without editing the genome for each target protein, which can be a laborious process in Pseudomonas. Since neither the promoter nor the gene is changed, the expression level, activity, and stability of the targeted gene's transcript and protein remain unaffected, allowing depletion studies under physiologically relevant conditions. In P. aeruginosa PA14, the native type I CRISPR/Cas system was able to act as a transcriptional repressor when a CRISPR RNA (crRNA) targeting the promoter was expressed in a cas3 deletion mutant or in strains that expressed anti-CRISPR protein AcrF3 (9). Here, we characterized a transcriptional repression system using a heterologous type II CRISPR/Cas system that does require a mutant strain background and is both inducible and titratable, which enables the temporal and dynamic depletion of the protein of interest. In addition, we demonstrated the ability to repress two genes simultaneously, increasing the breadth of applications for this system. The sgRNA plasmid design, which contains flanking BbsI restriction sites around the 20-bp target sequence, facilitates easy and efficient cloning that is readily scalable for the construction of sgRNA libraries for system-level gene targeting. Unlike the expression of S. pyogenes dcas9 in E. coli (31) and M. smegmatis (14), growth of the three Pseudomonas species was not affected by the expression of S. pasteurianus dcas9.

One limitation of the CRISPRi system is the availability of a PAM site at the desired target location. The S. pasteurianus consensus PAM site of NNGTGA (15) is found 31,817 times in the genome of P. aeruginosa PAO1, averaging approximately one PAM site for every 200 bp. In addition to the consensus sequence, we demonstrated that two additional PAM sites, NNGCGA and NNGTAA, also provide robust repression. The addition of the GC-rich PAM site NNGCGA is valuable for the use of S. pasteurianus dCas9 in GC-rich organisms. In PAO1, there are 109,509 NNGCGA sites, averaging approximately one site per 57 bp, significantly expanding the number of potential target sites. Another limitation of the CRISPRi system is that depletions are polar, which is a problem with all protein depletion systems that function at the transcriptional level. The depletion of individual proteins in bacteria is intrinsically difficult because they are often encoded in multigene operons carried on a single transcript. Thus, disrupting transcription of an upstream gene will cause depletion of cotranscribed downstream genes as well.

Off-target effects, which result from CRISPR/Cas9 interactions with genome locations that have a similar sequence as the intended target site, have been well characterized for S. pyogenes Cas9 in human cells (32). In the experiments presented here, there were no growth defects with S. pasteurianus dCas9 expressed in any of the three Pseudomonas spp., which confirms that dCas9 expressed from the chromosome in these cells is not toxic and suggests that there are no significant off-target effects. If off-targeting were a significant problem, it is likely that general toxicity would be observed because some essential genes would inevitably be repressed. However, this result does not definitively rule out the possibility of off-target effects in Pseudomonas, and target sites should be judiciously chosen to ensure that they are unique and not highly similar to other sites in the genome. In general, off-targeting has not been observed in bacteria, in part because the small genome size of bacteria limits the potential for sites with only one or two mismatches relative to the sequence of the intended target (14, 33).

Basal expression of dcas9 due to leakiness from the repressed promoter, which caused up to 50% repression compared to the level with an empty sgRNA control, may prove to be problematic, depending on the intended application. To tune dcas9 expression, we tested combinations of several promoter systems (TetR-Ptet, LacI-Plac, AraC-Para), addition of degradation tags to dCas9 (AAV and LVA tags) (see Fig. S1 in the supplemental material), and varying ribosome binding site strengths of dcas9. The results of these system variants showed that controlling transcription, rather than translation or degradation, of dCas9 in the uninduced state was most effective in preventing leaky expression, and the system with the best dynamic range was presented here. We report a robust, inducible, and titratable CRISPRi gene repression system in Pseudomonas using the S. pasteurianus dCas9. We demonstrate efficient depletion of endogenous genes, including essential genes leading to phenotypic changes consistent with established phenotypes. In addition, we show that NNGCGA and NNGTAA are two additional PAM sites that provide robust repression, significantly expanding the number of possible targets with S. pasteurianus dCas9.

MATERIALS AND METHODS

Bacterial strains and cultivation.

E. coli DH5α was used for plasmid cloning and maintenance. E. coli strains were cultured in LB medium at 37°C with antibiotic supplements where necessary: 5 mg/ml gentamicin and 50 mg/ml kanamycin. Pseudomonas strains used in this study were P. aeruginosa PAO1, P. putida KT2440, and P. fluorescens SBW25. Pseudomonas strains were cultured in LB medium with antibiotic supplements where necessary: 30 mg/ml gentamicin and 50 mg/ml kanamycin. P. aeruginosa was cultured at 37°C, while P. putida and P. fluorescens were cultured at 30°C. For CRISPRi repression, cultures were induced with 1 mM IPTG unless otherwise noted.

Pseudomonas transformations were performed by electroporation according to Choi et al. (34). Briefly, 3-ml overnight cultures were washed twice with 1 ml of 300 mM sucrose and resuspended in 300 μl of 300 mM sucrose. Aliquots of 100 μl were used per reaction, and cells were electroporated in a 2-mm cuvette using a Bio-Rad MicroPulsor electroporator with the Ec2 preset program setting of 2.5 kV (Bio-Rad, Hercules, CA). Cells were recovered in LB medium for 1 h before being plated on selection plates.

Plasmid construction and cloning.

The CRISPRi system consists of two vectors: an sgRNA plasmid, pBx-Spas-sgRNA, and an integrating dCas9 plasmid, pUC18-mini-Tn7T-Lac-dCas9 (Fig. 1B). The integrating dCas9 plasmid was constructed by amplifying S. pasteurianus dCas9, codon optimized for expression in mycobacteria (a kind gift from the Fortune lab, Harvard T. H. Chan School of Public Health, Boston, MA) (14), with primers containing BamHI and SacI sites (see Table S1 in the supplemental material), followed by ligation into a pUC18-mini-Tn7T-Gm-Lac (16) integrating vector at the BamHI/SacI site. For expression in P. putida and P. fluorescens, dCas9 was expressed from the existing Ptac promoter on pUC18-mini-Tn7T-Gm-Lac. For expression in P. aeruginosa, dCas9 was expressed from Plac by mutating the Ptac promoter sequence by PCR (Table S1). The dCas9 construct was then integrated into Pseudomonas strains by transforming the Tn7 vector via electroporation as described above and subsequent selection on 30 mg/ml gentamicin. The gentamicin marker was then removed by transforming pFLP3, as described previously (16).

The sgRNA plasmid pBx-Spas-sgRNA was constructed by amplifying the S. pasteurianus sgRNA handle and placing it under Ptet control on a broad-host-range pBBR1 plasmid. The 20-bp target sequence of the sgRNA was flanked with BbsI sites to facilitate Golden Gate cloning. All other BbsI sites on the plasmid were removed by introducing point mutations on oligonucleotides used for PCR amplification, followed by circular polymerase extension cloning (CPEC) (35). Target sgRNAs were cloned by first annealing complementary 24-mer oligonucleotides consisting of 20 bp of target sequence and 4 bp of overhangs to the sgRNA plasmid (Table S1). The sgRNA plasmid was digested with BbsI at 37°C and then ligated with the annealed oligonucleotides with T4 DNA ligase at room temperature and with the addition of more BbsI. We found that this method using predigestion followed by ligation resulted in higher cloning efficiency than Golden Gate PCR methods that cycle between digestion and ligation steps. Plasmids and sequences from this study are available via Addgene (Table S1).

The lacZ construct used for the characterization of PAM sites was constructed by amplifying the native gdhA promoter from genomic DNA of P. aeruginosa PAO1 and cloned upstream of lacZ on the mini-CTX-lacZ plasmid (21). The −10 and −35 sequences of the gdhA promoter were predicted using BPROM (Softberry) (25). All four PAM sites (NNGTGA, NNGCGA, NNGTAA, and NNGTGT) were inserted into three different regions in the promoter (upstream of −35 [location A], between −10 and −35 [location B], between −10 and TSS [location C]) by PCR in a manner that retained the −10 and −35 sequences and minimized changes to the native promoter sequence (Fig. S4). This modified lacZ construct was then integrated into P. aeruginosa at the ϕCTX attachment site with the mini-CTX1 vector.

NADP-dependent glutamate dehydrogenase activity assay.

P. aeruginosa cultures were grown in LB medium, with and without 1 mM IPTG. Ten-milliliter cultures were harvested at an optical density at 600 nm (OD₆₀₀) of ∼ 0.8 to 1.0 after 5 h of CRISPRi gene depletion, and cells were lysed using B-PER bacterial protein extraction reagent (ThermoFisher Scientific, Waltham, MA). NADP-dependent glutamate dehydrogenase activity was measured using the procedure for the enzymatic assay of l-glutamic dehydrogenase (NADP) (EC 1.4.1.4) from Sigma-Aldrich. Briefly, the rate of conversion of β-NADPH to β-NADP was determined by monitoring A₃₄₀, with 7.6 mM α-ketoglutarate and 0.22 M NH₄Cl as substrates in 100 mM Tris-HCl buffer, pH 8.3. Total protein concentration was measured using a Pierce bicinchoninic acid (BCA) protein assay kit according to the manufacturer's instructions (ThermoFisher Scientific, Waltham, MA). One unit of activity is defined as the reduction of 1 μmol of α-ketoglutarate to l-glutamate per minute at pH 8.3 at 30°C in the presence of ammonium ions and β-NADPH. Reported activity level is estimated as the unit of activity normalized by total protein content.

RT-qPCR.

P. aeruginosa cultures were grown in LB medium, with and without 1 mM IPTG. Four-milliliter cultures were harvested at an OD₆₀₀ of ∼ 0.8 to 1.0 after 5 h of CRISPRi gene depletion. Total RNA was extracted using hot phenol-chloroform (36). Total RNA samples were then treated with Turbo DNase (Life Technologies, Grand Island, NY), and equal amounts of RNA were reverse transcribed to cDNA using a QuantiTect reverse transcription kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. Quantitative PCR of the gdhA gene was then performed using primers gdhA-qpcr-F and gdhA-qpcr-R (Table S3) and Brilliant II Sybr green High ROX QPCR mix (Agilent Technologies, Santa Clara, CA) on an ABI 7300 real-time PCR system (Applied Biosystems, Beverly, MA). Reported transcript levels are the averages of biological triplicates measured in technical duplicates.

β-Galactosidase assay.

P. aeruginosa cultures were grown in LB medium, with and without 1 mM IPTG. Two-milliliter cultures were harvested at an OD₆₀₀ of ∼ 0.8 to 1.0 after 5 h of CRISPRi gene depletion. Cells were lysed with beads in 0.1 M phosphate buffer, pH 7.5. The initial rate of β-galactosidase activity was measured by monitoring the change in A₄₂₀ in the presence of ortho-nitrophenyl-β-galactoside (ONPG) as a substrate in 0.1 M phosphate buffer, pH 7.5. Total protein concentration was measured using a Pierce BCA protein assay kit according to the manufacturer's instructions (ThermoFisher Scientific, Waltham, MA). β-Galactosidase activity was estimated as the rate of change of A₄₂₀ normalized by total protein content.

Growth plates for essential gene depletion.

Cells transformed with sgRNA plasmids were recovered in LB medium and plated in a series of 10-fold dilutions on LB agar with and without 1 mM IPTG. Plates were then incubated at 37°C for P. aeruginosa and at 30°C for P. putida and P. fluorescens.

Differential interference contrast microscopy.

Pseudomonas cultures were grown in LB medium to an OD₆₀₀ of ∼ 0.3. IPTG (1 mM) was then added to induce CRISPRi depletion for 2 h. Cells were diluted and mounted on LB agar pads and viewed using a Nikon A1R Ultra-Fast spectral scanning confocal microscope with a 60× objective lens. Pictures were taken with an Andor Clara camera.

Pyoverdine measurements.

P. putida cultures were grown in M9 minimal medium, supplemented with 2 mM magnesium sulfate, 0.1 mM calcium chloride, and 20 mM succinate as the sole carbon source, with a final pH of 7.5. Cultures were grown with and without 2 mM IPTG for CRISPRi induction. After 48 h, cells were spun down, and pyoverdine concentrations were estimated by measuring the A₄₀₅ of the culture supernatant. Pyoverdine concentrations were normalized by cell density to the OD₆₀₀.

Motility assay.

A motility assay was performed as described previously (37). Briefly, a colony of P. putida grown on a plate was first diluted into 200 μl of sterile water, and 0.5 μl of this solution was then stabbed onto soft LB agar plates (0.5% agar), with and without 1 mM IPTG, using a pipette tip and incubated at 30°C. After 15 h of incubation, the diameter of the bacterial zone was measured.