Quorum sensing controls the Pseudomonas aeruginosa CRISPR-Cas adaptive immune system

Significance

The cell–cell communication process, called quorum sensing, activates all three key aspects of the prokaryotic adaptive immune system (termed CRISPR-Cas): expression, activity, and adaptation in the pathogen Pseudomonas aeruginosa. We show that pro- and antiquorum-sensing compounds activate and repress CRISPR-Cas, respectively, suggesting the exciting possibility of a combination quorum-sensing–inhibition-phage therapy cocktail. In P. aeruginosa, quorum-sensing inhibitors repress virulence, making P. aeruginosa more susceptible to elimination by the human immune system, while simultaneously making P. aeruginosa more prone to killing by phage therapy through inhibition of the CRISPR-Cas defense mechanism. Finally, because we show that quorum sensing activates adaptation by the CRISPR-Cas immune system, a quorum-sensing inhibitor should also reduce acquisition of resistance against the administered phage.

Abstract

CRISPR-Cas are prokaryotic adaptive immune systems that provide protection against bacteriophage (phage) and other parasites. Little is known about how CRISPR-Cas systems are regulated, preventing prediction of phage dynamics in nature and manipulation of phage resistance in clinical settings. Here, we show that the bacterium Pseudomonas aeruginosa PA14 uses the cell–cell communication process, called quorum sensing, to activate cas gene expression, to increase CRISPR-Cas targeting of foreign DNA, and to promote CRISPR adaptation, all at high cell density. This regulatory mechanism ensures maximum CRISPR-Cas function when bacterial populations are at highest risk for phage infection. We demonstrate that CRISPR-Cas activity and acquisition of resistance can be modulated by administration of pro- and antiquorum-sensing compounds. We propose that quorum-sensing inhibitors could be used to suppress the CRISPR-Cas adaptive immune system to enhance medical applications, including phage therapies.

Many bacteria and almost all Archaea carry CRISPR-Cas (clustered regularly interspaced short palindromic repeats; CRISPR-associated) adaptive immune systems, which provide sequence-specific immunity against previously encountered viruses and plasmids (1, 2). Genomic CRISPR arrays are composed of repetitive sequences alternating with spacers derived from parasitic genomes (viruses, plasmids, transposons). The process of spacer acquisition, known as adaptation, results in heritable immunization (1). Upon reinfection, processed CRISPR RNAs (crRNAs) guide Cas proteins to cleave complementary parasite genomes, which provides the bacterium with immunity (3, 4). Thus, CRISPR-Cas, by patrolling the cell, combats viral attacks and also enables the cell to avoid acquisition of foreign plasmids to which its ancestors have been exposed (5, 6).

Expression of CRISPR-Cas adaptive immune systems is costly (7, 8), possibly because of autoimmunity (9, 10) and deployment of resources that could otherwise be invested in growth. To limit fitness costs, some CRISPR-Cas systems are induced upon infection (11–13). Other environmental cues regulate CRISPR-Cas. In Escherichia coli, membrane stress activates CRISPR-cas (14) and CRISPR-cas is repressed by the DNA binding protein H-NS (histone-like nucleoid structuring protein). H-NS–mediated repression is relieved by the transcription factor LeuO (15, 16). CRISPR-cas is repressed by glucose and activated by cAMP receptor protein–cAMP in Pectobacterium atrosepticum (17). In addition to these mechanisms, theory and data suggest phage proliferation—and therefore risk of infection—increases with increasing bacterial cell density (18, 19). Bacteria monitor cell density using a cell–cell communication mechanism known as quorum sensing (QS). QS involves the production, release, and detection of extracellular signal molecules, called autoinducers (AI). QS controls behaviors that require cells to act in synchrony to achieve effective outcomes (20).

Results

We explored the idea that bacteria could use high AI levels at high cell density as an indicator of high risk of phage infection. To do this, we investigated whether QS controls CRISPR-Cas in the pathogen Pseudomonas aeruginosa (21). We used P. aeruginosa UCBPP-PA14 (denoted PA14), which has a type I-F CRISPR-Cas system (22) that provides phage resistance (8, 23, 24). In PA14, two CRISPR regions flank the cas genes: cas1, cas3, and csy1–4. Csy1–4 form a complex with a mature crRNA (22). Cas3, which is a nuclease and a helicase, cleaves DNA bound by the Csy1–4 complex. The two primary QS AI-receptor pairs in PA14 are called LasIR and RhlIR. LasI produces the AI 3-oxo-C12-homoserine lactone (3OC12-HSL), which is bound by LasR. The LasR–3OC12-HSL complex activates target genes, including lasI, resulting in autoinduction, as well as genes required for virulence (25–27). LasR–3OC12-HSL also activates rhlI and rhlR (28). RhlI synthesizes the AI C4-homoserine lactone (C4-HSL) that, in conjunction with RhlR, activates a second wave of QS genes (20, 29).

As a readout of CRISPR-Cas, we followed expression of cas3, encoding the nuclease that cleaves target DNA. cas3 expression tracks with cell density: minimal cas3 expression could be detected at low cell density, and activation of expression occurred in exponential phase (Fig. 1A) (P = 0.00005). With respect to QS control, single ΔlasR, ΔrhlR, ΔlasI, and ΔrhlI mutants and the double ΔlasR ΔrhlR mutant showed no change or a modest reduction in cas3 expression compared with the WT (Fig. 1B). The ΔlasI ΔrhlI double-synthase mutant, however, exhibited pronounced reductions in expression of cas1 and -3 and csy1–4 relative to WT (Fig. 1B and Fig. S1) (P = 0.0004). Addition of AIs to the ΔlasI ΔrhlI mutant restored expression to WT levels (Fig. 1B). It is possible that the double-AI synthase mutant showed a more dramatic reduction in cas gene expression compared with the single mutants and the mutant lacking both receptor genes because of compensatory effects from the orphan QS receptor QscR, which promiscuously binds AIs (30). Another possibility is that the AIs regulate cas and csy expression through a pathway that operates independently of LasR, RhlR, and QscR, as has been discovered for a few genes in P. aeruginosa PAO1 (31).

Fig. 1.

QS activates cas3 expression. (A) Relative cas3 expression normalized to 5S RNA measured by qRT-PCR in PA14 at low and high cell density (OD₆₀₀ = 0.1 and 1.0, respectively). (B) Relative cas3 expression at high cell density measured as in A for PA14 (WT) and the designated QS mutants. AI indicates 2 µM 3OC12-HSL + 10 µM C4-HSL. Error bars represent SD from n = 3 replicates (A) and n = 6 replicates (B).

Fig. S1.

QS activates cas and csy expression. Relative expression of cas1, cas3, and csy1–4. Relative expression of all cas and csy genes normalized to 5S RNA measured by qRT-PCR in WT PA14 (black) and the ∆lasI ∆rhlI double-AI synthase mutant (white). Strains were grown to OD₆₀₀ = 1. Error bars represent SD from n = 3 replicates.

To examine the consequences of QS on CRISPR-Cas activity, which is also called interference, we assayed the effectiveness of CRISPR-Cas in eliminating the CRISPR-targeted plasmid, called pCR2SP1 (23). This plasmid contains a protospacer targeted by CRISPR 2 spacer 1 flanked by a protospacer-adjacent motif (PAM) that is required for CRISPR interference (32). We used a plasmid rather than a phage because QS regulates phage adsorption in P. aeruginosa and may also affect other aspects of phage–host dynamics, complicating the analysis (33). We quantified retention of the control plasmid pHERD30T and the CRISPR-targeted plasmid pCR2SP1 over time in WT PA14 and in the ΔlasI ΔrhlI double-QS AI synthase mutant. No loss of the control plasmid occurred over the course of the experiment in either strain (Fig. 2A). With respect to the CRISPR-targeted plasmid, no loss occurred in either strain during low cell-density growth, and addition of AI had no effect (Fig. 2B) (0–3 h). However, after 5 h of growth, conditions under which QS has initiated, plasmid loss occurred in WT cells. In contrast, at T = 5 h, minimal loss occurred in the ΔlasI ΔrhlI double-QS AI synthase mutant. Addition of AI restored plasmid loss to WT levels in the ΔlasI ΔrhlI mutant (Fig. 2B). At 6.5 h, when QS is highly induced, although modest plasmid loss occurred in the ΔlasI ΔrhlI mutant, over 20-fold more of the ΔlasI ΔrhlI mutant cells retained the plasmid than did WT cells or ΔlasI ΔrhlI mutant cells supplemented with AI (Fig. 2B). This result shows that QS is required to potently induce CRISPR-Cas activity in PA14. There is residual CRISPR-Cas activity in the ΔlasI ΔrhlI mutant, suggesting a LasI RhlI-independent CRISPR-Cas activation mechanism exists in PA14.

Fig. 2.

QS regulates CRISPR-Cas activity. Retention of the control plasmid pHERD30T (A) and the CRISPR-targeted plasmid pCR2SP1 (B) in WT and in the ∆lasI ∆rhlI mutant during growth (100% denotes no plasmid loss). (C) EOT of WT PA14 and designated mutants at high cell density (OD₆₀₀ = 1) quantified as the percentage transformation by the CRISPR-targeted plasmid pCR2SP1 compared with that of the parent vector pHERD30T lacking the targeted sequence (23). Here, 100% denotes an EOT ratio of 1 for the two plasmids. In all panels, AI indicates 2 µM 3OC12-HSL + 10 µM C4-HSL. Error bars denote SD from n = 3 replicates.

The above results show that QS enhances plasmid loss during growth; that is, when the plasmid has already generated copies of itself (Fig. 2B). We next examined the influence of QS on the ability of CRISPR-Cas to eliminate a single incoming genetic element, which would resemble an attack by a single phage. For this assessment, we measured efficiency of transformation (EOT) at high cell density in WT PA14, the ΔlasI ΔrhlI mutant, and the ΔlasI ΔrhlI mutant supplemented with AI. In the ΔCRISPR Δcas mutant, lacking both CRISPR arrays and all cas and csy genes, the EOT of the pCR2SP1 plasmid compared with the control plasmid pHERD30T was 100% because the ∆CRISPR ∆cas mutant is incapable of targeting either plasmid (Fig. 2C). In contrast, in WT PA14 the EOT was 2% because CRISPR-Cas is functional and efficiently cleaves the targeted plasmid DNA (Fig. 2C). The EOT was 14% in the ΔlasI ΔrhlI double-QS mutant, showing that—in the absence of QS—the CRISPR-Cas immune system is sevenfold less effective than in the WT (P = 0.0006). Addition of AIs to the ΔlasI ΔrhlI double-mutant restored CRISPR-Cas activity, reducing the EOT to 4%. Thus, QS regulation of CRISPR-Cas activity in PA14 is essential for high-level CRISPR-Cas–dependent immunity against infecting elements.

Our results reveal that 2% of CRISPR-Cas–proficient PA14 colonies survived antibiotic selection despite CRISPR-targeting of the plasmid conferring antibiotic resistance. It is unlikely that this result could be a consequence of spontaneous loss of CRISPR-Cas activity, because loss occurs with a frequency of ∼0.001% (34). One possibility is that, in these colonies, plasmid mutations had been acquired that prevented CRISPR targeting. To test this idea, we sequenced the pCR2SP1 plasmids in 10 WT and 10 ΔlasI ΔrhlI colonies that had retained pCR2SP1. No mutations were present in the protospacer or PAM sequences of any of the 20 plasmids (Fig. S2). This result suggests that QS regulation of CRISPR-Cas activity at the point of transformation is responsible for the altered EOT shown in Fig. 2C. We suggest that, if the CRISPR-Cas system fails to rapidly eliminate an incoming plasmid, plasmid replication occurs. Only one of the 35 CRISPR spacers in the two PA14 CRISPR arrays targets the protospacer in our plasmid. Replicating plasmids could titrate out all available CRISPR-Cas complexes possessing the matching crRNA, possibly explaining how a CRISPR-targeted plasmid persists despite an active CRISPR-Cas system.

Fig. S2.

Sequencing of the CRISPR-targeted protospacer and PAM in pCR2SP1. WT and ∆lasI ∆rhlI mutant cells were transformed with the CRISPR-targeted plasmid pCR2SP1 at OD₆₀₀ = 1, and colonies were allowed to form on selective medium. The region in pCR2SP1 encoding the targeted protospacer and PAM was sequenced in 10 WT and 10 ∆lasI ∆rhlI colonies. No mutations were discovered in the protospacer or PAM in any of the 20 plasmids. A representative sequencing result from a WT colony is shown. The protospacer sequence targeted by CRISPR 2 spacer 1 is underlined and the GG PAM sequence is highlighted in bold.

Population-level CRISPR spacer diversity is crucial for bacteria to survive phage attack because phage cannot readily acquire point mutations, enabling them to simultaneously escape multiple crRNA spacers (24). Synchronized QS-mediated activation of CRISPR-Cas could boost population-wide acquisition of diverse spacers. The frequency of spacer acquisition is higher when a bacterium is challenged with a protospacer to which it already has immunity, a process called primed adaptation (35). We introduced a plasmid harboring a protospacer with an adaptation-priming seed mutation into the WT, the ∆cas3 mutant, and the ∆lasI ∆rhlI mutant, and assayed individual colonies for expansion of the CRISPR2 locus. We investigated CRISPR2 because higher frequency adaptation occurs to the CRISPR2 locus than to the CRISPR1 locus (8). In the absence of Cas3, which is required for adaptation, incorporation of new spacers into the CRISPR2 array did not occur (Fig. 3). In contrast, 26.9% of the WT cells had incorporated one or two spacers. Significantly fewer ∆lasI ∆rhlI mutant cells underwent adaptation and acquired spacers (11.4%). Addition of AI to the ∆lasI ∆rhlI mutant increased the total fraction of cells that acquired new spacers to 26% (i.e., to the level of the WT) (Fig. 3). The QS inhibitor Baicalein (36) prevents production of the virulence factor pyocyanin, the canonical QS-readout in PA14 (Fig. S3A). Baicalein also blocked the positive effect of AI on cas3 expression (Fig. S3B) and inhibited the AI-mediated enhancement of CRISPR adaptation in WT PA14 and in the ∆lasI ∆rhlI mutant supplemented with AI (Fig. 3).

Fig. 3.

QS controls CRISPR-Cas–mediated immunity by increasing spacer acquisition. Integration of new CRISPR spacers into the CRISPR2 locus was measured by PCR of single colonies of WT PA14, the ∆cas3 mutant, and the ∆lasI ∆rhlI mutant. Each of the strains harbored the CRISPR-targeted plasmid, pCR2SP1 seed, containing a seed mutation to promote adaptation. Each adaptation event results in acquisition of a new spacer and CRISPR repeat, which is exhibited by a 60-bp expansion of the CRISPR locus. Quantitation of the spacer population present in each colony is shown below each lane of the gel. Data are shown for representative colonies. AI indicates 2 µM 3OC12-HSL + 10 µM C4-HSL and inhibitor indicates 100 µM Baicalein (36).

Fig. S3.

The QS inhibitor Baicalein represses pyocyanin production and cas3 expression in PA14. (A) Relative pyocyanin production measured at OD₆₉₅ in PA14 treated with the indicated concentrations of the QS inhibitor Baicalein normalized to the pyocyanin levels from PA14 treated with DMSO. (B) Relative cas3 expression normalized to 5S RNA measured by qRT-PCR in the ∆lasI ∆rhlI mutant at high cell density (OD₆₀₀ = 1.0). Additions: DMSO (control), 2 µM 3OC12-HSL + 10 µM C4-HSL (AI), or 2 µM 3OC12-HSL + 10 µM C4-HSL + 100 µM Baicalein (AI + inhibitor) (36). Error bars denote SD from n = 3 replicates.

Our results demonstrate that P. aeruginosa uses QS to activate expression, activity, and adaptation of its CRISPR-Cas immune defense system. Consistent with our findings, microarrays and quantitative RT-PCR (qRT-PCR) in Burkholderia glumae PG1 showed altered expression of cas genes in a QS mutant, suggesting that QS regulation of CRISPR-Cas activity may be a general phenomenon for bacterial species harboring both QS and CRISPR-Cas systems (37). Our discovery of QS-mediated activation of CRISPR-Cas aligns well with previous studies showing QS-mediated phage defense via down-regulation of phage receptors, reducing infection rates (38, 39). QS repression of phage surface receptors at high cell density could be the first line of defense, effectively enabling bacteria to prevent infection. If this initial strategy fails, the second line of defense becomes crucial: QS activation of CRISPR-Cas immune defense enables pursuit of phages that make it into the cytoplasm. Although we have not yet shown both defense mechanisms operate in a single bacterial species, the capacity to do so certainly exists.

High cell-density induction of cas gene expression could, in addition to providing optimal phage defense, minimize the danger of autoimmunity, especially at low cell density when autoimmunity is expected to be particularly deleterious to a bacterial population. In Streptococcus thermophilus, Cas9 is constitutively produced and it confers a fitness cost that is most pronounced during lag phase and at low cell density, when the risk of phage infection is low (7, 13, 19). Experiments show that CRISPR-Cas–mediated autoimmunity results in deleterious effects, ranging from genome rearrangements and cell filamentation to suicide (9). Integration of spacers against the bacterial genome occurs at high frequencies in S. thermophilus, but these events are eliminated through autoimmunity (10). Consistent with this finding, computational analysis of CRISPR arrays in 330 bacterial and archaeal species found that only 0.4% of all spacers are self-targeting, suggesting strong selection against these events in nature (40). In E. coli, a mechanism exists to ensure preferential incorporation of spacers targeting foreign DNA (41), and in P. atrosepticum, selection against self-targeting occurs (42). We suggest that CRISPR-Cas systems may require tight regulation to properly balance the danger of autoimmunity with the risk of phage infection. We propose that, because QS correlates with a high probability of phage infection, placing the CRISPR-Cas system under QS control allows efficient activation of the system when the relative risk is high, while additionally enhancing the benefit-to-cost ratio for maintaining a functional CRISPR-Cas system by minimizing autoimmunity at low cell density when infection risk is low. Whether QS regulates cas expression directly, or whether the effect of QS on cas expression occurs via intermediates, awaits further study.

P. aeruginosa is a major pathogen that affects cystic fibrosis sufferers and causes hospital-acquired infections (21). The heavy use of antibiotics for P. aeruginosa control has led to widespread antimicrobial resistance (43). Thus, alternatives to conventional treatments for P. aeruginosa infection are urgently needed. QS inhibitors reduce P. aeruginosa virulence (44). Similarly, phage therapy targeting clinical P. aeruginosa isolates enhances survival of mice (45). Our findings suggest the exciting possibility of synergistic efficacy through a combination QS–inhibition-phage therapy mixture. QS inhibitors would repress virulence, making P. aeruginosa more susceptible to elimination by the host immune system, while simultaneously making P. aeruginosa more prone to killing by phage therapy through inhibition of the CRISPR-Cas defense mechanism. Finally, because QS activates CRISPR-dependent adaptation, a QS inhibitor should also reduce acquisition of resistance against the administered phage.

Materials and Methods

Bacterial Strains and Plasmids.

Strains and plasmids used in this study are listed in Table S1. To construct chromosomal deletions in P. aeruginosa PA14, DNA fragments flanking the gene of interest were amplified, stitched together by Gibson assembly, and cloned into pEXG2 (a generous gift from Joseph Mougous, University of Washington, Seattle) (46, 47). The resulting plasmids were used to transform E. coli SM10, and subsequently mobilized into PA14 via biparental mating. Exconjugants were selected on LB (Luria–Bertani) containing gentamicin (30 μg/mL) and irgasan (100 μg/mL), followed by recovery of deletion mutants on M9 medium containing 5% (wt/vol) sucrose. Candidate mutants were confirmed by PCR. The pCR2SP1 seed plasmid was constructed by inserting a protospacer, targeted by CRISPR 2 spacer 1 containing a single base mutation in the seed region, between the HindIII and EcoRI sites in pHERD30T. PCR primers are listed in Table S2.

Table S1.

Bacterial strains and plasmids

Strain or plasmid	Description	Source
PA14	WT, generous gift from George O’Toole, Geisel School of Medicine at Dartmouth, Hanover, NH	Laboratory stock
SM67	PA14 ∆lasR	Present study
SM32	PA14 ∆rhlR	Present study
SM73	PA14 ∆lasR ∆rhlR	Present study
SM51	PA14 ∆lasI	Present study
SM52	PA14 ∆rhlI	Present study
SM53	PA14 ∆lasI ∆rhlI	Present study
SMC4279	PA14 ΔCRISPR Δcas	(48)
SMC4268	PA14 Δcas3	(48)
pEXG2	Allelic exchange vector with pBR origin, gentamicin resistance, sacB, generous gift from Joseph Mougous, University of Washington, Seattle	(47)
pHERD30T	Empty plasmid, gentamicin resistance	(23)
pCR2SP1	pHERD30T containing the protospacer to CRISPR2 spacer 1	(23)
pCR2SP1 seed	pHERD30T containing the protospacer to CRISPR2 spacer 1 with a one base seed mutation	Present study

Table S2.

Primers used in this study

Primer	Sequence 5′-3′	Source
rhlR UP F	AGGAGGAAGCTTACGTGCTGCAGCGCGCCTAC	Present study
rhlR UP R	AACTCGAGCCGCAAGCATGCTGAATCCGTCATTCCTCATTGCAG	Present study
rhlR DN F	TTCAGCATGCTTGCGGCTCGAGTTCTGGGCCTCATCTGAAGCGT	Present study
rhlR DN R	CCTCCTTCTAGAGTAGCGCGAAAGCTCCCAGA	Present study
rhlI UP F	AGGAGGAAGCTTTTTCCGTGGCGCGCGACCAG	Present study
rhlI UP R	TTCAGCATGCTTGCGGCTCGAGTTGAGCAATTCGATCATGACCAAG	Present study
rhlI DN F	AACTCGAGCCGCAAGCATGCTGAATCGATGGCGGTGTGAGGTCG	Present study
rhlI DN R	CCTCCTTCTAGAGGTTGATCGAGATGCCGCTC	Present study
lasI UP F	AGGAGGAAGCTTCGGCGAGCTGGCGATCGGTA	Present study
lasI UP R	TTCAGCATGCTTGCGGCTCGAGTTAATTTGTACGATCATCTTCACTTC	Present study
lasI DN F	AACTCGAGCCGCAAGCATGCTGAACTGGCGGTTTCATGACGGG	Present study
lasI DN R	CCTCCTTCTAGACCTTACCCATCTGGGGCTGG	Present study
lasR UP F	AGGAGGAAGCTTGCCCGTGCGCCGCGCACAG	Present study
lasR UP R	TTCAGCATGCTTGCGGCTCGAGTTGTCAACCAAGGCCATAGCGC	Present study
lasR DN F	AACTCGAGCCGCAAGCATGCTGAACTTATTACTCTCTGATCTTGCC	Present study
lasR DN R	CCTCCTGGATCCTTCACTTCCTCCAAATAGGAAG	Present study
5S qPCR F	GAACCACCTGATCCCTTCCC	Present study
5S qPCR R	TAGGAGCTTGACGATGACCT	Present study
cas1 qPCR F	TCAAGGACTCGCTGATCCTG	Present study
cas1 qPCR R	GATCATGAAGTCCAGGGCCT	Present study
cas3-intF	GGTTGATCGTCAGCCATCAT	(49)
cas3-intR	GGCCTTTTCTTTTGCGTCT	(49)
csy1 qPCR F	TCTTCGAGCATGACTTCGGA	Present study
csy1 qPCR R	TGGCGAGGTTGTTATGGACT	Present study
csy2 qPCR F	CGTCCGAAGAAGAAGCATCG	Present study
csy2 qPCR R	CGCAGCGGTGTTTCTCTATC	Present study
csy3-intF	AAGACCAAGGACCGTGACC	(49)
csy3-intR	AGCCCTGATCGTTCACGTAG	(49)
csy4 qPCR R	GATCTGAGTGAGGAGGAGGC	Present study
csy4 qPCR F	TGAATCCTCCTTCCTCTGCC	Present study
pHERD30T verify F	TGCAAGGCGATTAAGTTGGG	Present study
pHERD30T verify R	CGCAACTCTCTACTGTTTCTCC	Present study
CR2SP check F	GAGGGTTTCTGGCGGGAA	Present study
CR2SP check R	GTCCAGAAGTCACCACCCG	Present study
CR2SP1seed F	AGCTTCCATCAGGCGGACGTTGTAGTAGTCGAGCGCGGTG	Present study
CR2SP1seed R	AATTCACCGCGCTCGACTACTACAACGTCCGCCTGATGGA	Present study

Growth Conditions.

PA14 and mutants were grown overnight at 37 °C with shaking in LB broth. Cultures were back-diluted 1:1,000 and grown to OD₆₀₀ = 0.1 for low cell density, and back-diluted 1:100 and grown to OD₆₀₀ = 1.0 for high cell density in the presence or absence of the solvent control DMSO, 3OC12-HSL, C4-HSL (Sigma), or Baicalein (Cayman Chemical) at the specified concentrations. LB was supplemented with 50 µg/mL gentamicin where appropriate.

qRT-PCR.

Cells were harvested at the indicated OD₆₀₀. RNA was purified using TRIzol (Ambion), and subsequently, DNase-treated (TURBO DNA-free, Thermo Fisher). cDNA was synthesized using SuperScript III Reverse Transcriptase (Invitrogen) and quantified using PerfeCTa SYBR Green FastMix Low ROX (Quanta Biociences).

Plasmid Retention Assay.

PA14 or the ∆lasI ∆rhlI mutant was transformed with the CRISPR-targeted plasmid pCR2SP1 or the parent vector pHERD30T lacking the targeted sequence, as described below. Single colonies were suspended in LB and grown at 37 °C with shaking for 6.5 h. Colony forming units were enumerated on LB agar with and without appropriate antibiotics at time 0 and 3 h (low cell density) and at 5 and 6.5 h (high cell density). The percentage of plasmid retention was calculated.

EOT Assay.

PA14 was grown to the appropriate OD₆₀₀, washed twice at room temperature in 300 mM sucrose, and electroporated with 1 µg pHERD30T or pCR2SP1 plasmid DNA. One milliliter of LB was added, and the cells were grown for 1 h at 37 °C with shaking, after which they were plated on LB medium containing 50 µg/mL gentamicin and incubated overnight at 37 °C. Colonies were counted using Image Quant Las 4000 and Image Quant TL software (GE Healthcare). Colony forming units per milliliter were quantified and the EOT was calculated as the percentage colonies transformed by pCR2SP1 compared with those transformed by pHERD30T.

Sequencing.

WT and ∆lasI ∆rhlI mutant cells were transformed with the CRISPR-targeted pCR2SP1 plasmid at OD₆₀₀ = 1, as described above. Ten WT and 10 ∆lasI ∆rhlI colonies were chosen for colony PCR. The region in pCR2SP1 containing the targeted protospacer and PAM was amplified as described below, using primers designed to anneal upstream and downstream of this region. The PCR fragments were separated by agarose gel electrophoresis and purified. Sequencing was performed by Genewiz.

Adaptation Assay.

WT PA14, the ∆cas3 mutant, and the ∆lasI ∆rhlI mutant were transformed with pCR2SP1 seed as described above. Single colonies were restreaked on LB medium containing 50 µg/mL gentamicin and either DMSO (control), 2 µM 3OC12-HSL + 10 µM C4-HSL (AI), or 100 µM Baicalein (inhibitor) and incubated at 37 °C overnight. Single colonies were tested for population-wide integration of new immunity spacers against the CRISPR-targeted plasmid by PCR using DreamTaq Green PCR Master Mix (Thermo Fisher), with primers designed to anneal upstream of the CRISPR2 array and in the second spacer, which enabled detection of expansion of this array. The PCR products were subjected to agarose gel electrophoresis and band intensities were analyzed using Image Quant TL software (GE Healthcare).

Statistical Analysis.

P values were calculated using a Student t test, except for the data in Fig. 2B, which were analyzed using one-way ANOVA for multiple comparisons.

SI Materials and Methods

For pyocyanin assays, Pseudomonas aeruginosa PA14 was grown overnight at 37 °C in liquid LB medium with shaking in the presence of Baicalein or DMSO. The cultures were back-diluted 1:1,000 into fresh medium containing Baicalein or DMSO and grown for 5 h. The cultures were back-diluted 1:50 into fresh medium containing Baicalein or DMSO and grown for 18 h. Cells were collected by centrifugation and the cell-free culture fluids were passed through 0.22-μm filters. Pyocyanin was quantified by measuring OD₆₉₅.