Quorum Sensing Controls Adaptive Immunity through the Regulation of Multiple CRISPR-Cas Systems

1In nature, bacteria persist in myriad environments, from sparse populations to localized communities of high cell density, including cell chains, microcolonies, and biofilms. These bacterial populations can provide collective advantages, but a trade-off may be an increased susceptibility to bacteriophage (phage) infection and invasion by mobile genetic elements. Thus, it has been theorized that formation of microbial groups is only advantageous in times of low phage abundance, or if the threat is attenuated through elevated bacterial defenses. It is well established that groups of bacteria regulate their behavior in response to cell density through QS, which is a widespread form of population-level communication. As cell density increases, QS mediates accumulation of extracellular chemical signals, which are sensed by nearby bacteria, resulting in altered gene expression.

2In response to viral invasion and potentially deleterious impacts of HGT, bacteria possess an arsenal of defense systems. The CRISPR-Cas (clustered regularly interspaced short palindromic repeats [CRISPR] and their CRISPR-associated [Cas] proteins) systems provide adaptive sequence-specific immunity against foreign elements, such as phages and plasmids. Immunity is first generated during adaptation when short invader-derived sequences (spacers) are integrated into CRISPR arrays. Second, the CRISPR arrays are transcribed and processed by Cas proteins, and in some cases host proteins, into short non-coding CRISPR RNAs (crRNAs). Finally, the crRNAs are assembled with Cas proteins into complexes that identify complementary invading nucleic acids and mediate their destruction—a process termed interference. The evolutionary success of CRISPR-Cas systems is evident from their broad distribution within bacteria and archaea. However, although CRISPR-Cas systems confer tangible benefits, there are associated fitness costs. Hence, multiple systems are transcriptionally regulated, which might enable physiological responsiveness to a changing environment and, thereby, a net cost-benefit balance. Since bacterial defensive requirements are predicted to change relative to population density, we hypothesized that CRISPR-Cas immunity could be integrated into the host QS circuit, allowing increased defense at higher cell densities.

3To test the role of QS in CRISPR-Cas regulation, we used Serratia sp. ATCC39006, which possesses a LuxIR-type QS system and three CRISPR-Cas systems (type I-E, I-F, and III-A), each with at least one CRISPR array (Figure 1A). Quorum sensing in Gram-negative bacteria typically utilizes LuxI family proteins to generate N-acyl homoserine lactone (AHL) signals, which are sensed by LuxR-type transcriptional regulators. In Serratia, the luxIR homologs, smaI and smaR, control secondary metabolite production and motility, and SmaI produces predominantly N-butanoyl-l-homoserine lactone (C4-HSL). Under our experimental conditions, the transcription of both smaI and AHL levels rose as cell densities increased, peaking at late exponential growth as cultures transitioned into stationary phase (Figure S1). To examine the effects of QS on CRISPR-Cas, we assessed cas operon and CRISPR expression in the wild-type (WT) and a signal-deficient smaI mutant throughout growth (Figures 1B and S1). Remarkably, expression of cas operons for all three CRISPR-Cas systems, as well as CRISPR1 (type I-E) and CRISPR2 (type I-F), was significantly reduced in the absence of AHL signal production (Figure 1B). The CRISPR arrays associated with the type III-A system (CRISPR3 and CRISPR4) exhibited low expression in the WT and were not regulated by QS since no further reduction was detected in the smaI mutant (Figures 1B and S1). We were able to fully complement the smaI mutant throughout growth by the addition of chemically synthesized C4-HSL, thereby confirming that the decreased cas and CRISPR expression in the smaI mutant resulted from the lack of AHL production (Figure S1). In agreement with previous work examining QS controlled secondary metabolite production in Serratia, addition of C4-HSL did not induce precocious induction of gene expression in the WT. Overall, expression of one or both core components (cas genes or CRISPRs) from all three CRISPR-Cas systems was subject to QS control.

4In the absence of the AHLs, the SmaR transcriptional regulator acts as a DNA-binding repressor. At increased cell density, AHLs accumulate and bind SmaR, thereby inhibiting its DNA binding activity, resulting in elevated gene expression through a de-repression mechanism. Mutation of smaR alone had no effect on cas and CRISPR expression throughout growth (Figures 2 and S2). The lack of enhanced expression in the smaR mutant is well established for genes previously shown to be controlled by QS in Serratia and is likely to be due to other required physiological and regulatory inputs. Deletion of smaR in the smaI mutant restored expression of the cas operons and CRISPR arrays throughout growth (Figures 2 and S2), demonstrating that, in the absence of AHL production, SmaR acts as a repressor of CRISPR and cas gene expression. In agreement, plasmid-encoded SmaR caused significantly reduced expression from each of the QS-regulated CRISPR and cas promoters but not from a non-QS regulated control promoter (Figure S3). The SmaR-mediated repression observed using this system was similar to the reduction in CRISPR and cas expression upon deletion of smaI in Serratia. Therefore, these results demonstrate that SmaR represses CRISPR-Cas expression in the absence of the QS signaling molecules.

5Evidence that the CRISPR-Cas modules were regulated by the host QS circuit supported our hypothesis that defense against invaders would be elevated at high cell densities. To determine whether the transcriptional changes correlated with modulation of immunity, we exposed Serratia cells growing in high-density populations to donor bacteria that transfer, via conjugation, plasmids that mimicked invaders that were encountered previously. These plasmids contained sequences complementary to the first spacer present in CRISPR1, CRISPR2, or CRISPR3 for the type I-E, I-F, and III-A systems, respectively (Table S2). These target sequences are termed protospacers and, for the type I-E and I-F systems, included canonical protospacer adjacent motif (PAM) sequences that are necessary to evoke direct interference. In the WT populations, all three CRISPR-Cas systems were capable of robust interference of the respective target plasmids but not of untargeted control plasmids (Figure 3), demonstrating that each native system is functional. The conjugation efficiencies of untargeted (naive) control plasmids for the smaI mutant were comparable to the WT, demonstrating that there were no CRISPR-Cas-independent effects in this background. In contrast, the interference capability was significantly reduced in signaling-deficient populations (the smaI mutant) by ∼20-fold for type I-E, ∼500-fold for type I-F, and ∼240-fold for type III-A targeting (Figure 3). Unexpectedly, the type I-E system showed the weakest interference response to QS, despite having the strongest effect on the cas8e promoter (Figure 1). It is likely that the activity of other type I-E components might form a bottleneck for the overall level of interference, which is the case for cas3 in the E. coli type I-E system. The impaired interference in all three CRISPR-Cas systems could be rescued via the addition of exogenous QS signal (Figure S4). Despite the reduced levels of interference in the smaI mutant, we still observed relatively efficient recognition and destruction of the targeted plasmids by each of the CRISPR-Cas systems. Together, these results demonstrate that QS signaling modulates the efficiency of interference and is necessary to allow enhanced defense at high cell densities.

6Adaptation is a critical function of CRISPR-Cas systems, allowing generation of new immunity through spacer acquisition. Therefore, we asked whether this aspect of CRISPR-Cas was also regulated by QS. Two adaptation modes are known, naive and primed. During naive adaptation, spacers are acquired from elements to which no previous immunity exists, whereas primed adaptation, observed in type I systems, enhances acquisition of spacers from elements resembling those previously encountered. To examine adaptation in the WT and the smaI mutant, we tested their abilities to acquire spacers from either a “naive” plasmid, representing an unrecognized invader, or “primed” plasmids, representing escape mutants from targets of the type I-E or type I-F systems. Primed plasmids contained non-consensus PAMs to trigger primed acquisition of additional spacers (Table S2). For WT cells containing the naive plasmid, repeated passage to high cell density in the absence of antibiotic selection yielded no detectable naive spacer acquisition (Figure 4). In contrast, primed spacer acquisition from the escape plasmids was readily observed in the WT for both the type I-E and type I-F systems. Adaptation in the smaI mutant was reduced by ∼75% and ∼80% for the type I-E and type I-F systems, respectively (Figure 4). The impaired adaptation in both CRISPR-Cas systems was rescued via the addition of exogenous QS signal (Figure S4). In summary, QS-mediated elevation of CRISPR-Cas activity enhances the generation of immunological memory within high-density populations by promoting increased spacer acquisition.

7Here we demonstrate that in a single strain, the expression of three different CRISPR-Cas systems (including types I and III) is regulated by the host QS circuit to significantly modulate immunity, including both interference and adaptation. The QS effect on adaptation is highly relevant with respect to bacterial population fitness because increased diversity of CRISPR spacers within communities restricts the success of phage escape mutants. We show that QS-defective populations generate fewer new spacers during adaptation, hence less diversity, thereby highlighting the importance of cell-cell communication in stimulating population-level CRISPR-Cas resistance. This nascent immunity is further reinforced by the elevated interference invoked by spacers from the three different systems.

8Our results demonstrate the importance of cell signaling in coordinating adaptive immunity when microbial groups are at high cell densities. Since HGT frequency, or phage spread, is less likely at low population densities, defense does not need to be high, but it is still necessary. In agreement, we still observed efficient, albeit significantly reduced, CRISPR-Cas immunity under conditions mimicking low cell densities (i.e., the smaI mutant). Since SmaR is a repressor that is inactivated by AHLs, this particular QS system can be viewed as dampening down immunity at low cell density. The burden of CRISPR-Cas systems, such as lethal auto-immunity caused by self-targeting, might have provided selective pressure to evolve this “suppress when least required” mechanism. In contrast, successful phage infection of high cell density bacterial communities results in high localized viral loads that might overwhelm basal level CRISPR-Cas defenses. Thus, upregulation of CRISPR-Cas via QS facilitates transition to a heightened defensive state that is better suited to cope with high multiplicity of infection phage attacks.

9As well as increasing general defense against invading elements, upregulation of CRISPR-Cas activity might also allow for an enhanced response to the stimulation of HGT or prophage release that can be triggered by QS. For example, diverse mobile elements coordinate their dissemination via QS, including AHL-based control of conjugative Ti plasmid transfer in Agrobacterium tumefaciens. Furthermore, many QS signals, including AHLs, Pseudomonas quinolone signal (PQS), and AI-2, can induce prophage induction in Gram-positive and Gram-negative bacteria. It is salient that we observed QS-dependent regulation of type III-A activity. Type III are the only known CRISPR-Cas systems to target DNA in a transcription-dependent manner, which is thought to protect bacteria from active or induced prophages, while minimizing self-targeting of integrated prophages. Therefore, the QS-dependent response of the type III-A system might not only protect from phage infection, but also restrict the proliferation of viral progeny during prophage induction. Interestingly, a recent global metagenomic study highlighted the relevance of temperate phages in ecosystems with high bacterial abundances, implying that the role of QS in prophage induction and CRISPR-Cas regulation could be ecologically significant.

10The broad distribution of both CRISPR-Cas and QS systems within diverse bacteria suggests that QS-dependent regulation of immunity should be widespread. In support of this, we performed an analysis of published microarray data from Pectobacterium atrosepticum and discovered a significant reduction in type I-F cas gene expression in an AHL synthase mutant (expI) that was most pronounced at high cell density. Likewise, in a Burkholderia glumae transcriptomic study, mutation of luxI homologs resulted in decreased expression of type I-F cas genes. Quorum sensing mechanisms used by bacteria are diverse, with peptide pheromones common in Gram-positives and the “universal” AI-2 signal produced by LuxS in many disparate bacteria. Furthermore, multiple CRISPR-Cas types (e.g., I-E, I-F, and III-A in Serratia) can be connected to QS circuits. Therefore, we predict that the control of adaptive immunity by QS is likely to be widespread across diverse bacteria and CRISPR-Cas types, irrespective of the precise signaling mechanism.

11Fittingly, other phage defense systems operate under QS control. For example, an AHL-dependent reduction in receptors on E. coli limits infection by λ and χ phages. Since E. coli does not produce AHLs, but has a LuxR sensor (SdiA), this might provide protection against broad host-range phages preying on neighboring bacteria. Within heterogeneous populations, analogous cross-species QS signaling could boost the CRISPR-Cas defenses of minority species, reducing the risk these individuals pose as vectors or reservoirs for phage spread. Other defense strategies provide population-level protection, such as abortive infection systems, which typically result in the “altruistic” suicide of infected cells. One such system, mazEF from E. coli, is regulated by a QS pentapeptide and limits phage P1. These suicidal defenses are most successful when bacteria are growing with spatial structure. It is interesting that both abortive infection and CRISPR-Cas are most effective in populations at high cell density.

12Because QS is an important regulator of CRISPR-Cas, invaders are likely to have evolved evasion mechanisms. In agreement, some phages encode acylases, enzymes that degrade AHLs, and others encode their own QS systems. These phages might block or interfere with QS to improve their reproductive success in the face of CRISPR-Cas competition—akin to anti-CRISPR proteins. Indeed, phages have been engineered to express AHL-degrading enzymes that enhance their ability to disrupt biofilms by eliciting cell death and by inhibiting QS. Our findings suggest that phage therapies that are combined with anti-QS strategies (e.g., engineered phages or anti-QS molecules) might assist in the evasion of CRISPR-Cas defense during treatment.

13Tables S1 and S2 list all strains and plasmids used in this study, respectively, and Table S3 lists the oligonucleotides used. Details of strain and plasmid constructions are provided in the Supplemental Experimental Procedures. Unless otherwise stated, Serratia sp. ATCC39006 strains were grown at 30°C and E. coli strains at 37°C in Lysogeny Broth (LB), minimal medium agar (0.1% w/v (NH4)2SO4, 0.41 mM MgSO4, 0.2% w/v glucose, 40 mM K2HPO4, 14.7 mM KH2PO4 [pH 6.9–7.1], 1.5% w/v agar) or on LB-agar (LBA) plates containing 1.5% (w/v) agar. When required, media were supplemented with antibiotics as follows: ampicillin (Ap; 100 μg/mL), chloramphenicol (Cm; 25 μg/mL), kanamycin (Km; 50 μg/mL), spectinomycin (Sp; 50 μg/mL), and tetracycline (Tc; 10 μg/mL). 5-aminolevulinic acid (Ala; 50 μg/mL) was added for growth of ST18. Bacterial growth was measured in a Jenway 6300 Spectrophotometer at 600 nm (OD600), except when grown in 96-well microtiter plates, where it was measured in a Varioskan Flash Multimode Reader (Thermo Fisher Scientific) at 600 nm. All experiments were repeated in at least three biological replicates.

14AHL production was assessed using bioassay plates as previously described. Briefly, the bioassay plates were prepared by seeding 100 mL of molten 0.75% LBA overlay with 1 mL of an ISTSO4 overnight culture, which was then poured over the surface of 1.5% LBA in a 25 cm × 25 cm square petri dish. Once set, holes were punched in the agar using a sterile cork borer. Samples for assay of AHL production were taken from cultures of WT (LacA) Serratia grown in multiple 1 mL aliquots of LB media in a 96 square deep-well plate (Labcon) incubated at 1,200 rpm at 30°C in a microplate shaker (BioProducts Incumix). Samples (1 mL) were pelleted by centrifugation at 13,000 rpm for 4 min, and the supernatants were sterilized using 0.22 μM syringe filters (Millipore). Supernatant samples for each time point were then used to fill the wells in the bioassay plate. The plate was subsequently incubated at 30°C for 24 hr, and the area of pigmentation surrounding each well was measured and reported as arbitrary units (a.u.). Supernatants from smaI mutant cultures were included at each time point and did not produce any detectable AHL.

15Growth of bacterial strains containing the lacZ reporters and the β-galactosidase assays were performed as previously described. The reporter strains contained a single chromosomal integration of the lacZ reporter fused to the ATG start codon of the various cas genes in the native genetic context, or to the start of the different CRISPR arrays. Therefore, they report on the expression from the various genes or arrays from their natural promoter positions within the chromosome (for a schematic see Figure S1 in ). Briefly, bacteria were grown in 1 mL of LB with Tc in 96 square deep-well plates (Labcon) and incubated in a microplate shaker (BioProducts Incumix) at 1,200 rpm and 30°C. β-galactosidase assays were performed in a Varioskan Flash Multimode Reader (Thermo Fisher Scientific) as described previously using the fluorogenic substrate (4-Methylumbelliferyl β-D-galactoside; MUG). Relative fluorescent units (RFUs) per second were calculated using the linear increase in fluorescence, which was normalized to the OD600 of the sample (RFU/s/OD600). RFU/s/OD600 measurements are depicted as arbitrary units in the axis labels of relevant figures.

16N-butanoyl-l-homoserine lactone (C4-HSL) was synthesized as described previously and the chemical nature confirmed. C4-HSL was stored in dimethyl sulfoxide (DMSO) and was added at a final concentration of 0.5 μM at the start of growth in the chemical complementation experiments. In the control samples, an equivalent volume of DMSO was added as a solvent control. Growth and β-galactosidase assays were performed as described earlier. For complementation of interference and adaptation experiments, assays were performed as described later, but all plates and bacterial cultures were supplemented with 0.5 μM C4-HSL or DMSO (control).

17Conjugation efficiency was assessed in a similar manner to that described previously. E. coli ST18 were used as donors for the conjugation of control (pPF719) and type I-E (pPF724) or type I-F (pPF722) targeted plasmids, or of control (pPF781) and type III-A (pPF1043) targeted plasmids. Plasmids pPF724, pPF722, and pPF1043 each contain a protospacer targeted by spacer 1 from either CRISPR1 (type I-E), CRISPR2 (type I-F), or CRISPR3 (type III-A), respectively. Recipient strains were WT (LacA) and smaI (LIS). Strains were grown overnight in LB with appropriate antibiotics, the OD600 adjusted to 1 and washed twice with LB. Donors and recipients were mixed in a 1:1 ratio, and 5 μL was spotted on 0.2 μm filters (Millipore) on LBA + Ala and incubated for 24 hr. Next, the filters were added to 2 mL PBS, the bacteria were resuspended, and dilution series were plated onto LB for recipient counts or with the addition of antibiotics for selection of transconjugant counts. For the type III-A experiments either 20 mM glucose or 0.02% arabinose was included in the plates for the filter matings or transconjugant selection, respectively. In all cases, conjugation efficiency was calculated as transconjugants per recipients.

18Plasmids pPF719 (non-targeted “naïve” control), pPF1048 (“primed” type I-E), and pPF1032 (“primed” type I-F) were transferred from E. coli ST18 by conjugation into WT (LacA) and smaI (LIS) strains and plated on LBA + Tc. After PCR confirmation of the transconjugants, overnight cultures of each strain grown in the presence of Tc were used to inoculate fresh 5 mL cultures in LB without antibiotics in 20 mL universals. These were then incubated at 30°C with shaking and passaged for 6 days by daily transfer of 10 μL to 5 mL of fresh LB. CRISPR expansion (indicative of spacer acquisition) was determined by PCR directly on cells from passaged cultures (DreamTaq, Thermo Fisher Scientific) using primers PF1887 + PF1989 for CRISPR1 and PF1888 + PF1990 for CRISPR2. PCR products were separated by 3% agarose gel electrophoresis and stained with ethidium bromide, and spacer acquisition was quantified using ImageJ.

19Conceptualization, A.G.P., S.A.J., R.P., R.H.J.S., and P.C.F. Investigation, A.G.P., S.A.J., C.T., R.P., and R.H.J.S. Resources, G.B.E. and G.P.C.S. Formal Analysis, A.G.P., S.A.J., G.P.C.S., R.P., R.H.J.S., and P.C.F. Writing – Original Draft, A.G.P., S.A.J., R.H.J.S., and P.C.F. Writing – Review & Editing, A.G.P., S.A.J., G.B.E., G.P.C.S., R.P., R.H.J.S., and P.C.F. Funding Acquisition and Supervision, P.C.F.