SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

The recently discovered prokaryotic CRISPR/Cas defence system provides immunity against viral infections and plasmid conjugation. It has been demonstrated that in Escherichia coli transcription of the Cascade genes (casABCDE) and to some extent the CRISPR array is repressed by heat-stable nucleoid-structuring (H-NS) protein, a global transcriptional repressor. Here we elaborate on the control of the E. coli CRISPR/Cas system, and study the effect on CRISPR-based anti-viral immunity. Transformation of wild-type E. coli K12 with CRISPR spacers that are complementary to phage Lambda does not lead to detectable protection against Lambda infection. However, when an H-NS mutant of E. coli K12 is transformed with the same anti-Lambda CRISPR, this does result in reduced sensitivity to phage infection. In addition, it is demonstrated that LeuO, a LysR-type transcription factor, binds to two sites flanking the casA promoter and the H-NS nucleation site, resulting in derepression of casABCDE12 transcription. Overexpression of LeuO in E. coli K12 containing an anti-Lambda CRISPR leads to an enhanced protection against phage infection. This study demonstrates that in E. coli H-NS and LeuO are antagonistic regulators of CRISPR-based immunity.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Invasions by viruses and conjugative plasmids pose a threat to microbial cells. To neutralize selfish DNA elements, bacteria and archaea have developed several defence strategies, such as receptor masking, restriction/modification and abortive infection (Hyman and Abedon, 2010; Labrie et al., 2010). Recently it was discovered that Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) (Ishino et al., 1987) and CRISPR-associated (cas) genes (Jansen et al., 2002) form a sophisticated immune system that uses small RNAs to target mobile genetic elements, as reviewed previously (van der Oost et al., 2009; Horvath and Barrangou, 2010; Karginov and Hannon, 2010; Marraffini and Sontheimer, 2010). CRISPRs consist of repeating sequences of approximately 30 nucleotides that are separated by unique sequences of similar size, called spacers (Mojica et al., 2000). The spacer sequences are commonly derived from phages and plasmids (Bolotin et al., 2005; Mojica et al., 2005; Pourcel et al., 2005), and new spacers can be added to the existing CRISPR array, expanding the invader repertoire (Barrangou et al., 2007), in a process known as CRISPR adaptation. The presence of a spacer matching a viral or plasmid sequence confers resistance to invasion by these elements (Barrangou et al., 2007; Brouns et al., 2008; Marraffini and Sontheimer, 2008). The biochemical pathways underlying CRISPR defence are partially known and involve transcription of the array into a long precursor CRISPR RNA. This precursor is cleaved in the repeat sequences by a Cas endonuclease [CasE in Escherichia coli (Brouns et al., 2008), Cas6 in Pyrococcus furiosus (Carte et al., 2008)], releasing small crRNAs that serve to guide the defence.

The cas genes encode the protein machinery that carries out the various steps of CRISPR defence. Approximately 45 families of cas genes have been identified (Haft et al., 2005) that are classified in eight typical combinations or subtypes named after a representative organism, e.g. type E after E. coli (Haft et al., 2005).

The type E CRISPR/Cas immune system in E. coli K12 is composed of eight cas genes (cas1, cas2, cas3 and casABCDE) and a downstream CRISPR locus with type 2 repeats (Kunin et al., 2007) containing 12 spacer-repeat units (CRISPR I) (Fig. 1A). Two additional CRISPRs with type 2 repeats (CRISPR II and CRISPR III, containing six and two spacers, respectively), as well as a CRISPR with type 4 repeats (Kunin et al., 2007) containing one spacer repeat unit (CRISPR IV) are located elsewhere on the genome (Diez-Villasenor et al., 2010). In addition to a CRISPR containing an anti-invader sequence, only Cas3 and CasABCDE, forming the protein complex Cascade (CRISPR-associated complex for antiviral defence), are required for CRISPR interference (Brouns et al., 2008). A recent study has demonstrated that in E. coli K12 transcription from the casA and CRISPR I promoters is repressed by heat-stable nucleoid-structuring protein (H-NS) (Pul et al., 2010), a global repressor of transcription in many Gram-negative bacteria. A microarray study indicates that transcription of casABC and cas2 is elevated in an E. coli K12Δhns strain compared with wt E. coli K12 (Hommais et al., 2001). In addition, H-NS was shown to possess high binding affinity for the intergenic region between cas3 and casA (Oshima et al., 2006; Pul et al., 2010). H-NS has a preference for binding AT-rich DNA sequences (Navarre et al., 2006). After initial binding of H-NS to high-affinity nucleation sites (Bouffartigues et al., 2007; Lang et al., 2007) repression of transcription is mediated by cooperative spreading along the DNA [defined as DNA stiffening (Liu et al., 2010)] and by creating looped structures through formation of DNA–protein–DNA bridges (Dame et al., 2005). Moreover, H-NS acts as a DNA–structuring protein (Stoebel et al., 2008; Liu et al., 2010).

image

Figure 1. LeuO and H-NS regulate cas gene expression. A. Schematic illustration of the CRISPR/Cas locus in E. coli K12 that consists of eight cas genes [cas3 (ygcB), casA (ygcL or cse1), casB (ygcK or cse2), casC (ygcJ or cse4) casD (ygcI or cas5e), casE (ygcH or cse3), cas1 (ygbT) and cas2 (ygbF)] and a downstream CRISPR locus containing 12 spacers and 13 repeats (CRISPR I). The cas3, anti-cas3 (anti-Pcas), casA (Pcas) and CRISPR I promoter are indicated with an arrow (Pul et al., 2010). B. qPCR analysis of cas gene transcript abundance in E. coliΔhns and E. coli W3110 expressing leuO (induced or non-uninduced). Fold changes are given as compared with wt E. coli W3110 expression levels. Error bars indicate one standard deviation.

Download figure to PowerPoint

Overcoming H-NS-mediated repression of cas gene transcription may be a key requirement for CRISPR/Cas functionality. Generally, H-NS repression can be relieved by a number of proteins, such as SlyA, VirB and others (Stoebel et al., 2008). One of these proteins is the regulator LeuO (Chen and Wu, 2005; De la Cruz et al., 2007), which belongs to the LysR family of transcription factors (Stoebel et al., 2008) and is found in all proteobacteria, except the δ subdivision (Maddocks and Oyston, 2008). The leuO gene maps next to the leuABCD operon (Hertzberg et al., 1980; Chen and Wu, 2005; Chen et al., 2005), whose gene products are required for leucine synthesis (Vartak et al., 1991). Recent data indicate that LeuO is involved in regulating transcription of many genes, often as an H-NS antagonist (Stoebel et al., 2008; Shimada et al., 2009). However, since under laboratory growth conditions the genomic leuO gene itself is repressed by H-NS (Klauck et al., 1997; Majumder et al., 2001) all LeuO regulation studies make use of plasmid encoded leuO under control of constitutive or inducible promoters. In the host environment leuO is likely to be induced under certain conditions as for example virulence of Salmonella enterica leuO mutants is attenuated (Lawley et al., 2006).

A genomic screen for LeuO-binding DNA fragments in E. coli K12 revealed 12 gene clusters, including the casA-cas2 operon (ygcL, ygcK, ygcJ, ygcI, ygcH, ygbT, ygbF) (Shimada et al., 2009). When LeuO was overexpressed, increased expression of casA and cas2 was observed in E. coli (Shimada et al., 2009), and of casA (STY3070) in S. enterica serovar Typhi (Hernandez-Lucas et al., 2008). We therefore investigated whether LeuO can mediate H-NS derepression of cas gene and CRISPR transcription. In this study we demonstrate that LeuO counteracts H-NS-dependent repression of the casA promoter by reorganizing the DNA protein contacts within the transcription initiation region. This results in increased transcription of the Cascade genes, the limiting factor for CRISPR-based defence against phage infection in E. coli K12.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

LeuO activates cas gene expression

To study the effect of LeuO on cas gene expression, transcript levels of the E. coli K12 cas genes in mid-exponential growth phase were examined using a DNA microarray approach. RNA samples isolated from a wt E. coli K12 strain containing a leuO-encoding plasmid were compared with RNA isolated from a strain containing the empty vector. In addition, RNA isolated from a ΔleuO mutant carrying the empty vector was analysed. Comparison of cas gene transcription levels between the LeuO-expressing strain and the control strain revealed a significant upregulation of transcription of casABCDE, cas1 and cas2, showing a gradual decrease from casA (65-fold) to cas2 (fivefold) (Table 1). No change in the transcription level of cas3 was detected. These results are consistent with a polycistronic transcription of the casABCDE and possibly the cas1 and cas2 genes, with polar effects for the transcription of the more downstream genes. However, we did not observe significant differences in cas gene transcription in the ΔleuO mutant compared with the wild-type strain (Table 1), indicating that leuO is not expressed under the growth conditions used here.

Table 1.  Microarray analysis of activation of cas genes by LeuO.
GeneMicroarray analysis of activation of cas genes by LeuO
pLeuO/wtΔleuO/wt
Fold changeaP-valueFold changeaP-value
  • a.

    The fold change of cas genes expression was determined by microarray analysis. pLeuO/wt indicates the ratio of cas transcripts detected upon overexpression of LeuO (using plasmid pKEDR13) as compared with wild-type E. coli K12 (transformed with the empty vector plasmid pKESK22). ΔleuO/wt indicates the ratio of cas transcripts detected in a ΔleuOFRT mutant as compared with wild type.

  • n.s., not significant.

cas31.0n.s. 0.931.1n.s. 0.76
casA65.4< 0.05−1.2n.s. 0.08
casB30.0< 0.05−1.1n.s. 0.20
casC24.8< 0.051.0n.s. 0.81
casD17.5< 0.051.1n.s. 0.68
casE15.4< 0.05−1.2n.s. 0.31
cas18.8< 0.051.2< 0.05
cas25.4< 0.051.1n.s. 0.34

To verify the observed increase in cas gene expression levels, quantitative PCR (qPCR) was performed on total RNA isolated from three strains during mid-exponential phase: wt E. coli, a Δhns strain and a wt strain expressing leuO from a plasmid. This analysis showed that casABCDE displayed increased transcription in both hns knockout and leuO-expressing strains (Fig. 1B). While the increase in casABCDE transcripts was modest in hns knockout strains, on average fivefold, the effect of introducing leuO was more dramatic, with an average increase of 236-fold after induction of leuO expression. An increase in casABCDE transcripts was also observed when leuO expression was not induced, due to leakage from the PT5 lac-1 promoter. The cas1 and cas2 genes also displayed increased transcript abundance in leuO-expressing strains, although at lower levels than casABCDE. Consistent with the microarray data, a trend of transcript fold change was observed, with polar effects downstream of casA, again suggesting a polycistronic mRNA of the casABCDE12 operon. Compared with the effect on the other cas genes, only a small increase of cas3 transcription was observed in leuO-expressing strains.

To further evaluate the effects of H-NS and LeuO on transcription from the casA promoter [known as Pcas (Pul et al., 2010)], RNA samples from wt strains expressing leuO from a plasmid and strains lacking hns were compared by primer extension analysis. No cas transcripts were detected in wt cells containing an empty expression vector. Transcripts directed from Pcas were only detected in cells expressing leuO from a plasmid or in hns knockout strains (Fig. S1), indicating that transcription of the casABCDE12 operon is tightly controlled by H-NS and LeuO.

LeuO causes increased crRNA abundance

The CRISPR I locus is transcribed in E. coli K12 and the transcript is cleaved by the CasE subunit of Cascade into small crRNAs that subsequently remain bound by Cascade (Brouns et al., 2008; Pul et al., 2010). In K12 small crRNAs were virtually undetectable by Northern blot analysis (Brouns et al., 2008 and Fig. 2A). To investigate whether this was due to too low transcription levels of casABCDE, the wt strain was transformed with a plasmid encoding the Cascade protein components under control of an arabinose-inducible promoter. In the wt strain expressing casABCDE from a plasmid, crRNAs with a length of about 60 nt could be detected. The requirement for plasmid-encoded synthesis of Cascade for detection of small crRNAs indicates that the level of Cascade in wt E. coli is insufficient for generating and stabilizing mature crRNAs. Furthermore we analysed the levels of crRNAs in an hns knockout strain and in the wt strain expressing leuO constitutively from a plasmid. Both deletion of hns and overexpression of leuO caused significant crRNA accumulation, due to enhanced expression of Cascade in these two strains (Fig. 2B and C). The CasA knockout strain (JW2730) serves as a control and marker for the mature crRNA. It was previously shown that a CasA knockout strain generates elevated levels of mature crRNA (Brouns et al., 2008), due to read-through of the downstream cas genes from the kanamycin resistance marker containing recombination cassette by which the casA gene is replaced (Pougach et al., 2010).

image

Figure 2. Formation of mature crRNA. A. Northern analysis of total RNA from wt E. coli K12 and wt E. coli K12 + casABCDE (pNH6) using the single-stranded spacer sequence NH30 (Table S2) as a probe. B and C. (B) Northern analysis as in (A) of total RNA from E. coli K12 Δhns and (C) wt E. coli K12 expressing leuO from a plasmid (pNH41) with an OD600 of 0.5, 1 and 3.0. M, size marker (pUC19/MspI ladder). E. coli K12 ΔcasA (JW2730) serves as a control and marker for mature crRNA.

Download figure to PowerPoint

Binding of LeuO and H-NS to the DNA sequence upstream of casA

The casA-cas3 intergenic region (here denoted IGLB) contains Pcas, for which H-NS has strong binding affinity as well as the divergently oriented anti-cas3 (known as anti-Pcas) promoter, that is located 80 bp upstream of Pcas and gives rise to an antisense transcript of unknown function (Figs 1A and 3C) (Pul et al., 2010).

image

Figure 3. H-NS and LeuO binding to the DNA region upstream of casA (the IGLB fragment). A. Electrophoretic mobility shift assay (EMSA) of the IGLB fragment with LeuO and H-NS, either alone (lanes 5–7) or with pre-bound LeuO and subsequent addition of H-NS (lanes 8–11) or pre-bound H-NS and subsequent addition of LeuO (lanes 12–15). B and C. (B) DNase I footprint of IGLB in the presence of either H-NS or LeuO or both. LeuO was pre-incubated with the reaction mixture. The two main regions protected by LeuO are indicated. The ladder indicates the IGLB coordinates relative to the casA start codon. (C) The sequence and coordinates of IGLB relative to the casA start codon. The H-NS and LeuO binding sites are indicated with boxes. See also Fig. 4A for footprints showing the LeuO binding region I.

Download figure to PowerPoint

Both LeuO and H-NS bind the IGLB fragment, as determined by electrophoretic mobility shift assay (EMSA) (Fig. 3A, lanes 2–4 and 5–7). Pre-bound LeuO impedes cooperative binding of H-NS to the IGLB fragment (Fig. 3A, lanes 9–11). In line with this, pre-bound H-NS is partly released from the DNA when LeuO is added to the complex (Fig. 3A, lanes 12–15). In order to map the binding region of LeuO within the IGLB fragment, DNase I footprint analysis was performed. Upon limited DNase I hydrolysis of the IGLB DNA fragment, H-NS causes an extended footprint (Fig. 3B), as shown before (Pul et al., 2010). In addition, LeuO protects two sites (site 1 and site 2) within the IGLB fragment that flank the high-affinity H-NS nucleation site (Fig. 3B and C). LeuO site 1 is located 20 bp downstream of Pcas and LeuO site 2 spans the divergent anti-Pcas (Fig. 3B and C). Interestingly, in the presence of LeuO the extended protection by H-NS is no longer visible (Fig. 3B, compare lanes 2 and 4), indicating that due to LeuO binding the DNA region containing the H-NS high-affinity binding site is no longer protected from DNase I cleavage, in agreement with decreased cooperative binding (Fig. 3A).

In order to analyse the effect of LeuO on RNA polymerase (RNAP) binding to the promoter sites, DNase I footprints were performed in the presence of RNAP and LeuO. Moreover, the effect on transcription initiation and RNAP open complex formation was analysed by KMnO4 footprints of stable initiation complexes. RNAP binds to the two promoters (Pcas and anti-Pcas) [Fig. 4A, lanes 3 and 3′, indicated I (Pcas) and II (anti-Pcas)] (Pul et al., 2010). Addition of the DNA binding proteins LeuO or H-NS alone does not cause changes in the KMnO4 reactivity (Fig. 4A, lanes 2′, 4′ and 5′). Binding of LeuO abolishes the spreading of H-NS along the DNA, resulting in a lack of protection by H-NS in the region between positions −160 to −240 (Fig. 4A, compare lanes 2 and 5), as observed before (Fig. 3B). When RNAP binding was studied in the presence of both transcription factors it turned out that the order of addition to the DNA is crucial for the resulting footprint. RNAP binding was only affected when H-NS and/or LeuO were added to the DNA prior to RNAP. While prior binding of H-NS to the IGLB DNA fragment completely abolished RNAP–promoter interaction and open complex formation (Fig. 4A and B, compare lanes 3′ and 6′), prior binding of LeuO had a repressive effect only on RNAP binding at anti-Pcas (Fig. 4A and B, compare lanes 3′ and 7′; Fig. 4E, lane 5′). This can also be seen on the retardation gels (Fig. 4C and D), where the DNA/RNAP complex II is lost in the presence of LeuO (Fig. 4C, lane 7). This complex remains stable when H-NS is added last (Fig. 4C, lane 9), in contrast to a sample with only H-NS or where H-NS is added before LeuO (Fig. 4C, lanes 6 or 8 respectively). Moreover, the change in nucleotide reactivities indicates that LeuO binding alters the architecture of the transcription initiation complex at the Pcas promoter (compare Fig. 4B, lanes 7′ with Fig. 4A, line 3′ and Fig. 4E, lane 3′ with lane 5′). Altogether these data indicate that LeuO plays an important role in the regulation of casABCDE12 gene expression by antagonizing H-NS-dependent repression of Pcas.

image

Figure 4. Effects of LeuO and H-NS on RNA polymerase open complex formation at the IGLB fragment. A and B. Either DNase I (lanes 1–14) or KMnO4 (lanes 1′–14′) footprint analyses of the non-template strand of IGLB after binding of RNA polymerase, H-NS or LeuO, either individually or in combination. The numbers above the lanes indicate the order of addition of the respective proteins. The casA promoter (Pcas) (I) and the anti-cas3 promoter (anti-Pcas) (II) are indicated. The arrowheads indicate the nucleotides within open complexes at the promoter sites. C and D. The binding of RNA polymerase, H-NS or LeuO, either individually or in combination. (C) shows electrophoretic mobility shift assays of the samples in (A) and (B). The numbers above the lanes indicate the order of addition of the respective proteins. (D) shows electrophoretic mobility shift assays of the samples in (E). The positions of the free DNA and complexed DNA are given on the left. The retardation gels (C) and (D) are not standard retardation gels but controls for the footprint complexes to see if no major decomposition has occurred. Because complexes are kept at 30°C for some time under RNAP binding conditions these gels are sometimes different from standard retardation experiments (e.g. Fig. 3A). E. A similar analysis as in (A) and (B) for the template strand of IGLB (DNase I footprint lanes 1–8; KMnO4 footprint lanes 1′–8′). The protein concentrations used are H-NS (1 µM), LeuO (1 µM) and RNA polymerase (50 nM).

Download figure to PowerPoint

H-NS and LeuO regulate CRISPR-based immunity against phage infection

The effect of H-NS on CRISPR-based defence against phage infection was analysed in wt and Δhns E. coli strains, grown in Luria–Bertani broth. Since none of the spacers of E. coli K12 target known bacteriophages, an artificial seven spacer CRISPR (J3) with the native promoter was designed containing one spacer that targets the template strand of the gene encoding the phage Lambda tail protein (J). A non-targeting (N) CRISPR (Brouns et al., 2008) served as a negative control. Introducing the J3 CRISPR reduced the sensitivity to virulent phage Lambda (λvir) infection fourfold in the Δhns but not in the wt strain (Fig. 5A). Complementation of the Δhns strain reversed the reduction in phage sensitivity, demonstrating that CRISPR-based defence is negatively regulated by H-NS (Fig. 5A). When cells were grown in richer media (2YT) until stationary growth phase, higher resistance levels were observed, up to a 10-fold reduced sensitivity compared with a non-targeting strain (data not shown). Moreover, plaques were much smaller in the hns knockout strains equipped with the J3 CRISPR when using 2YT.

image

Figure 5. Effect of H-NS and LeuO on CRISPR-based resistance in vivo. A. A synthetic CRISPR with one spacer (J3) targeting phage Lambda on a plasmid (pWUR564) is introduced in E. coli K12Δhns, E. coli K12, and complemented E. coli K12Δhns expressing hns from the pHOP11 plasmid. Phage resistance is monitored by determining the efficiency of plaquing (EOP) after challenge with virulent Lambda phage. B. The effects of leuO (pKEDR13) and cas3 (pWUR608) expression on phage resistance is monitored in E. coli K12ΔCRISPRI::CRISPRJ3 (indicated as J3) C. The effect of introducing the casABCDE12 (pWUR607), cas3 or a CRISPR on phage resistance is monitored in E. coli K12ΔCRISPRI::CRISPRJ3 (indicated as J3).

Download figure to PowerPoint

Further evidence that H-NS controls CRISPR-based immunity was obtained using genetically engineered strains (Table S4) in which the genomic CRISPR I locus was replaced by the J3 or a non-targeting CRISPR. E. coli K12ΔCRISPRI::CRISPRJ3 (E. coli J3) was fully sensitive to infection by phage Lambda, despite the presence of a genomic J3 spacer (Fig. S2). However, when the dominant negative hnsG113D mutant was expressed from a plasmid, the sensitivity of E. coli J3 to phage λvir infection was reduced 3.6-fold (Fig. S2). This mutant still forms heterodimers with wt H-NS, but does not bind DNA and therefore interferes with H-NS-mediated transcriptional repression resulting from the formation of higher-order DNA–protein complexes (Ueguchi et al., 1996). This observation is consistent with the reported finding that expression of hns mutant G113D induces transcription from Pcas in wt E. coli (Pul et al., 2010).

When plaque assays were performed in E. coli J3 overexpressing leuO from a plasmid, an approximately sixfold reduced sensitivity to phage λvir infection was observed (Fig. 5B), demonstrating that LeuO activates CRISPR-based defence in E. coli. A threefold reduced sensitivity was observed when leuO expression was not induced, probably due to leakage from the Ptac promoter. When E. coli J3 cells were grown to stationary phase in rich 2YT medium, an increased resistance level was observed with turbid and very small plaques in the leuO-overexpressing strains containing a targeting CRISPR (data not shown), whereas plaques in the same strain containing a non-targeting CRISPR were clear and of normal size. Although cas3 gene expression was not strongly induced when LeuO was expressed from a plasmid (Table 1 and Fig. 1B), the expression of cas3 was not a limiting factor for resistance, since introduction of a cas3 expression plasmid into E. coli J3 expressing leuO did not lead to elevated resistance levels (Fig. 5B).

When a plasmid expressing casABCDE12 was introduced in E. coli J3, a 2.5-fold reduced sensitivity to phage infection was observed (Fig. 5C), which was not observed when a CRISPR expression vector containing spacer J3 or a plasmid encoding cas3 was introduced (Fig. 5C), indicating that expression of the genes encoding Cascade (casABCDE) is limiting for CRISPR-based defence in wt E. coli.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

The type E CRISPR/Cas system [Cse (Haft et al., 2005)] is present in many proteobacteria and in some actinobacteria, firmicutes and methanogenic archaea. A recent analysis of a collection of natural isolates shows that this CRISPR/Cas subtype occurs in approximately 60% of the E. coli strains (Diez-Villasenor et al., 2010). The study presented here provides experimental evidence for regulation of the type E CRISPR/Cas system in E. coli K12 by the antagonists H-NS and LeuO. These antagonistic DNA-binding proteins regulate the expression of several genes in E. coli, such as the bgl operon (utilization of β-glucosides) (Ueguchi et al., 1998), the yjjQ-bglJ operon (virulence factor and activator of bgl respectively) (Stratmann et al., 2008) and the S. enterica serovar Typhi ompS1 gene (outer membrane protein and pathogenicity determinant) (De la Cruz et al., 2007).

We demonstrate that relieving H-NS-mediated repression of cas gene transcription is required for CRISPR-based immunity and that derepression is mediated by LeuO through direct binding of DNA sequences upstream of casA. The EMSA and footprint results (Fig. 3) support the finding that elevated amounts of LeuO counteract H-NS-mediated repression of casABCDE12 in vivo. Moreover, these data indicate that LeuO-induced activation of transcription from the casA promoter (Pcas) does not simply result from a displacement of bound H-NS, since LeuO cannot facilitate the binding of RNA polymerase when H-NS is pre-bound. Instead, LeuO abrogates the cooperative spreading of H-NS upon binding to the casA promoter region.

Interestingly, the transcript levels of the casABCDE12 operon were higher in cells expressing leuO than in hns knockout strains (Fig. 1B), suggesting that either LeuO functions as an enhancer of casABCDE12 transcription, or that derepression in K12Δhns is incomplete. The latter could be due to additional repressors involved in silencing casABCDE12, or due to functional redundancy between suppressors of gene transcription. In particular, StpA has been reported to possess high binding affinity for Pcas (Pul et al., 2010). Although a K12ΔstpA strain showed similar cas gene transcript levels as a wt K12 strain (Pul et al., 2010) it cannot be excluded that StpA-mediated repression of cas gene transcription takes place in the absence of H-NS.

Cells expressing leuO showed higher resistance levels compared with hns knockout strains, due to the higher expression of the cas genes and higher abundance of mature crRNA (Figs 1B and 2B and C). Compared with the CRISPR-based resistance levels to phage infection observed in Streptococcus thermophilus (Barrangou et al., 2007) or E. coli BL21-AI overexpressing the cas genes and the CRISPR (Brouns et al., 2008), the resistance levels of wt E. coli overexpressing leuO are relatively low. However, at present it is unknown whether a similar level of CRISPR-based immunity can be reached by wt E. coli, and if it can, under what conditions.

Although we were able to show that CRISPR-based immunity is activated by overproducing LeuO, the natural growth conditions that induce CRISPR-based defence are still unknown. Our experiments show that a genomic anti-Lambda spacer alone does not provide resistance to phage infection in wt E. coli due to the absence of Cascade. We speculate that leuO expression levels under laboratory growth conditions are too low to induce derepression Pcas, and that phage exposure itself does not activate CRISPR defence. Unaltered expression of leuO and the cas genes was also observed during infection with bacteriophage PRD1 (Poranen et al., 2006).

Since H-NS is known to bind DNA of incoming phage or plasmid directly (Navarre et al., 2006; Navarre et al., 2007) this might result in redistribution of H-NS (Doyle et al., 2007; Dillon et al., 2010), allowing expression of the Cascade genes due to decreased local concentrations of the repressor. As leuO expression is negatively regulated by H-NS and positively by LeuO itself (Hommais et al., 2001; Chen et al., 2005), this would further amplify the activating signal for cas gene transcription.

Interestingly, leuO expression levels are induced by the alarmone guanosine tetraphosphate (ppGpp) (Fang et al., 2000; Chen et al., 2001; Majumder et al., 2001). ppGpp is involved in stress signalling cascades leading to the stringent response under nutrient limiting conditions. Since these conditions slow down phage proliferation dramatically (Schrader et al., 1997), bacterial cells may then stand a better chance of surviving phage encounters, hence inducing CRISPR-based defence may be more beneficial. However, induction of the stringent response by amino acid starvation, e.g. by serine hydroxamate (Tosa and Pizer, 1971) neither increased the transcription from Pcas nor the formation of mature crRNA (data not shown). Although under laboratory conditions CRISPR-based defence is suppressed, the diversity in spacer content in natural isolates of E. coli strongly suggests that the CRISPR/Cas system as a whole is active and functional in natural ecosystems (Diez-Villasenor et al., 2010).

In an independent parallel study, it has been shown that an E. coli hns knockout strain containing an anti-Lambda spacer is less sensitive to phage infection (Pougach et al., 2010), in agreement with the data presented here. It seems that the Δhns strain containing the T3 spacer used in (Pougach et al., 2010), shows higher levels of resistance than the Δhns strain containing the J3 spacer that was used in this study. The T3 spacer has originally been described (Brouns et al., 2008) as the spacer that confers the highest level of immunity of eight different spacers tested. In BL21-AI overexpressing the casABCDE and cas3 genes together with either the T3 CRISPR or the J3 CRISPR showed that the T3 CRISPR provides 10-fold more resistance (data not shown), indicating that the observed difference in immunity between (Pougach et al., 2010) and this study is most likely resulting from a difference in the efficiency of the spacers used.

Although a number of studies involving H-NS and LeuO have been carried out in E. coli and S. enterica (Hommais et al., 2001; Lucchini et al., 2006; Navarre et al., 2006; Hernandez-Lucas et al., 2008; Shimada et al., 2009), the outcome of these studies has never been interpreted in the light of CRISPR-based defence. Based on these genome-wide analyses we propose that the expression of the type E (Cse) cas genes from S. enterica are likely to be regulated by H-NS and LeuO as well. For instance, in S. enterica serovar Typhi transcription of casA (STY3070) appears to be affected by H-NS and LeuO (Hernandez-Lucas et al., 2008), despite the poor conservation of the intergenic region between the divergently oriented cas3 and casA genes in this strain. In S. enterica serovar Typhimurium strain LT2 H-NS binding sites were found encompassing the translation start site of the cas3 gene (Lucchini et al., 2006). Another study showed that in this strain the transcription of cas3, casB, casC and casD is elevated in the absence of H-NS (Navarre et al., 2006). Perhaps the cas genes are controlled by a single promoter in this strain, since the intergenic region between cas3 and casA is only 12 nucleotides in length.

Altogether, this study provides evidence that the type E CRISPR/Cas system in E. coli is regulated by the antagonists H-NS and LeuO, and we propose that this regulatory mechanism is conserved in S. enterica as well. The upcoming challenge will be to identify conditions that activate this sophisticated defence system to allow defence against invasion by foreign DNA.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Strains

The wild-type E. coli K12 W3110 (BW25113) strain and the E. coli K12 W3110 derivative Δhns (JW1225) and ΔcasA (JW2730) from the KEIO collection (Baba et al., 2008), supplied by the American Type Culture Collection (ATCC), and E. coli K12 MC4100 (Peters et al., 2003) derivative Δhns (PD32) (Dersch et al., 1993) were used throughout the study.

Gene cloning and recombination

A synthetic recombination cassette was designed corresponding to 400 bp flanking regions on each side of the CRISPR I locus separated by a kanamycin resistance gene flanked by FRT sites (GAAGTTCCTATACTTTCTAGAGAATAGGAACTTC). The construct contained a PstI site followed by 400 nt of the CRISPR I upstream region of the E. coli K12 W3110 genome (2875875–2876274), followed by a NcoI site, then the sequence AAACAAAGAATT, a KpnI site, followed by an FRT site, a SphI site, a kanamycin resistance gene with a sequence corresponding to pJJDuet30 (2186-1276), a XhoI site, an FRT site, a NotI site, and 395 nt of the CRISPR I downstream region of the E. coli K12 W3110 genome (2877225–2877619) (GeneArt AG, Regensburg, Germany). A synthetic CRISPR sequence including leader sequence containing seven spacers and eight repeats was used (Table S1) (Geneart AG, Regensburg, Germany). This synthetic CRISPR was cloned between the flanking regions using the NcoI and KpnI sites (Fig. S3). The NcoI and EcoRI sites in the leader and second spacer were used to exchange the first spacer sequence of the CRISPR; the constructs created were named J3 and R44 (Table S1). The other spacers in the CRISPR were sequences with no homology to phage Lambda. These constructs were used as recombination cassettes to replace the existing CRISPR I locus in the E. coli K12 W3110 genome, following a protocol described elsewhere (Datsenko and Wanner, 2000), with minor modifications. For recombination, the sequences were PCR-amplified using primers BG3017 and BG3019 (Table S2) with high-fidelity pfu-turbo polymerase and transformed by electroporation into E. coli K12 W3110 containing pKD46, kindly provided by the ATCC. Transformants were grown at 30°C and plated on LB agar supplemented with kanamycin (50 µg ml−1). The pKD46 plasmid has a temperature-sensitive origin of replication, and was removed through growth at 37°C (Datsenko and Wanner, 2000). Recombination was validated by PCR and sequencing. The antibiotic resistance cassette was removed using Flp recombinase encoded on plasmid pCP20, and subsequent growth at 37°C, as described (Datsenko and Wanner, 2000). The ΔleuO mutant was constructed with the λred-gam system using oligonucleotides T209 and T210 (Table S2), as described (Datsenko and Wanner, 2000). After deletion of leuO the resistance cassette used for selection was removed using Flp recombinase encoded on plasmid pCP20 (Datsenko and Wanner, 2000).

Plasmids and vectors

Plasmid pWUR607 (TetR) contains the casABCDE12 operon, which was PCR-amplified from E. coli K12 MG1655 genomic DNA using primers BG2173 and BG2174 (Table S2), and cloned into vector pACYC184 using the restriction sites EcoRI and NcoI. Plasmid pWUR608 (CamR) was constructed by cloning a cas3 amplicon generated with primers BG2171 and BG2172 (Table S2) into pACYC184 using the restriction enzymes BamHI and SphI. In the experiments where a CRISPR was introduced on a plasmid, the pACYCduet-1 vector (CamR) (Novagen) was used, using the NcoI and Acc65I restriction sites. pWUR477 containing the non-targeting CRISPR (N) was described previously (Brouns et al., 2008). Expression of the CRISPR from this plasmid in K12 was under control of the leader sequence that contains the CRISPR I promoter (Pul et al., 2010). pWUR564 is a derivative of pWUR477 that has the NcoI–EcoRI fragment (containing the leader sequence up to half of the second spacer) replaced with the NcoI–EcoRI fragment of construct J3 (Table S1). For expression of wt hns and hnsG113D the previously described pHOP11 and pHM52 plasmids were used respectively (Pul et al., 2010). The pCA24N plasmid from ASKA(−) clone JW0075 encodes leuO behind a PT5 lac-1 promoter (IPTG inducible). pKEDR13, encoding leuO behind a Ptac promoter (IPTG inducible), and the control vector pKESK22 were described earlier (Madhusudan et al., 2005; Stratmann et al., 2008). The IPTG-inducible leuO expression plasmid pNH41 was constructed by cloning the leuO amplicon, generated using primers NH329 and NH330 (Table S2), into the 2.2 kb XbaI fragment of pZE12-luc, following a previously published protocol (Urban and Vogel, 2007). Plasmid pNH6 contains the casABCDE operon [PCR amplified with pre-phosphorylated primer NH193 and primer NH194 (Table S2)] inserted by blunt end and EcoRI cloning into vector pCU01 (pBAD-TOPO vector derivative), as described (Unoson and Wagner, 2008). Plasmid pUC18-IGLB was described before (Pul et al., 2010).

Microarray

Escherichia coli K12 MG1655 was transformed with plasmid pKEDR13 (KanRlacIq Ptac leuO) (Stratmann et al., 2008) for expression of LeuO or with control vector pKESK22 (KanRlacIq Ptac). Exponential cultures were inoculated from fresh overnight cultures to an OD600 of 0.1 in LB supplemented with 25 µg ml−1 kanamycin. IPTG was added after 30 min of growth to a final concentration of 1 mM. After additional 60 min the bacteria were harvested using Qiagen RNAprotect and used for RNA isolation using the Qiagen RNeasy MiniKit system. In brief, 1 ml of each culture (OD600 between 0.5 and 0.6) was used and processed according to the manufacturer's instructions including an on-column DNase I treatment. RNA quality was assayed by denaturing urea-PAGE and by measuring the ratio of absorption at 260/280 nm in a GeneQuant II spectrophotometer (Amersham). RNA concentration was determined by measuring UV light absorption at 260 nm. The procedure was carried out four times with independent clones.

Synthesis of cDNA (and cRNA) and hybridization of Affymetrix GeneChip®E. coli Genome 2.0 microarrays was carried out according to the manufacturer's instructions. In total, four independent RNA samples of each group (wild-type, leuO-expressing and leuO-deficient strains) were used. Data analysis was performed using Affymetrix Software. Fluorescence values were normalized to the GeneChip standard reference probes. Differential expression values were calculated as fold change of leuO-expressing samples compared with samples of LeuO-deficient control strains.

qPCR analysis of gene expression

qPCR analysis of cas gene transcript abundance was performed on cDNA synthesized using High Capacity Reverse Transcription Kit (Applied Biosystems) from RNA extracted by the hot-phenol method (Blomberg et al., 1990) and DNase I-treated using Turbo DNA-free kit (Ambion). Ten millilitres of samples for RNA extraction were taken at OD600 ∼ 0.5 from E. coli W3110, E. coli W3110 carrying pCA24N (leuO) and E. coliΔhns (JW1225-2). When LeuO expression was induced, samples were taken 30 min after addition of 0.5 mM IPTG. The qPCR reactions were performed using Power SYBR green PCR master mix (Applied Biosystems) according to manufacturer's instructions, and primers were designed using Primer Express 3.0 (Applied Biosystems). For the complete list of primers used see Table S3. As an internal control two primer pairs were designed against casA. The PCR reactions were performed on a 7300 Real Time PCR System (Applied Biosystems) and analysed using 7300 System SDS Software 1.3 (Applied Biosystems). Fold change of cas gene transcription was calculated using the relative quantification method with tmRNA as endogenous control and E. coli W3110 cas gene transcript abundance as calibrators. All PCR reactions were performed in six replicates. Control PCRs without template or without cDNA (produced by standard cDNA synthesis but excluding reverse transcriptase) were performed to monitor general contamination levels and genomic DNA contamination of RNA extracts respectively.

Northern blotting

Total RNA was extracted at the OD600 indicated using TRIZOL reagent (Invitrogen) according to the manufacturer's protocol. Expression of casABCDE from pNH6 was induced at an OD600 of 0.5 by adding 0.2% of arabinose for 15 min. Ten micrograms of total RNA was denatured at 95°C with an equal volume of formamide loading dye, FD (90% formamide, 15 mM EDTA, 0.05% bromophenol blue and 0.05% xylene cyanol), and subsequently separated on an 8% denaturing polyacrylamide gel. A 32P-labelled pUC19 DNA/MspI ladder (Fermentas) was used as size marker. The RNA was electrotransferred to Nylon N+ membranes (GE Healthcare) at 10 V for 15 h. Transfer was performed in a BIORAD blotting chamber in 1× TBE buffer at 4°C followed by drying of the membrane and UV-cross-linking. Pre-hybridization was carried out for 2–4 h at 42°C in 15 ml of pre-hybridization buffer (5× SSC, 5× Dernhardt, 0.05 M sodium phosphate pH 6.7, 1% dextran sulphate, 0.1% SDS) together with 75 µl of herring sperm DNA (20 mg ml−1). Hybridization was carried out overnight at 42°C in the same buffer lacking herring sperm DNA but containing [γ-32P]-ATP-labelled oligonucleotide probe NH30 (Table S2) specific for spacer 2 of the CRISPR1 locus. The probe was labelled with [γ-32P]-ATP [40 pmol of DNA, 10× kinase buffer, T4 polynucleotide kinase (PNK, Ambion), [γ-32P]-ATP] by incubation at 37°C for 45 min. Prior to hybridization, the probe was purified over a G-50 column (GE Healthcare). Membranes were washed once for 20 min at 60°C in 2× SSC, 0.5% SDS and once for 20 min in 0.5× SSC, 0.5% SDS. Signals were quantified in a Molecular Dynamics PhosphorImager model 400S with ImageQuant software version 4.2a (Molecular Dynamics).

Electrophoretic mobility shift assay

Escherichia coli RNAP, LeuO and H-NS were purified according to published procedures (Stratmann et al., 2008; Pul et al., 2010). The IGLB fragment [position −1 to −414, relative to the first nucleotide of the casA (ygcL) start codon] was obtained by EcoRI/HincII or BamHI/SacI digestion of plasmid pUC18-IGLB. Purified DNA fragments were end-labelled by Klenow (Promega) and [α-32P]-dATP. Binding reactions with the indicated amounts of protein were performed in 50 mM Tris-HCl, pH 7.4, 70 mM KCl, 15 mM NaCl, 1 mM EDTA, 10 mM β-mercaptoethanol at a final heparin concentration of 20 ng µl−1. Complexes were separated on native 5% (w/v) polyacrylamide gels and visualized by autoradiography as described (Pul et al., 2010).

Footprint analyses

DNase I footprinting of free DNA and DNA–protein complexes was performed as described (Pul et al., 2010). Formation of open RNAP–promoter complexes was analysed by KMnO4 modification of single-stranded nucleotides within the transcription bubble. Forty microlitres of RNAP–DNA complexes were treated with 160 mM KMnO4 for 2 min at 30°C. The reaction was stopped by addition of 5.3 µl of β-mercaptoethanol and 5.3 µl of 500 mM EDTA and the samples were ethanol precipitated after phenol/chloroform extraction. Pellets were dissolved in 10% piperidine and incubated at 90°C for 30 min. After two rounds of washing with distilled water followed by lyophilizing, the pellets were dissolved in 50 µl of distilled water and precipitated with ethanol. Cleavage products were separated on 10% denaturing polyacrylamid gels and visualized by autoradiography. The following protocol was used in footprint experiments with more than one protein: LeuO or the protein-free buffer, and H-NS or the respective buffer, were incubated with the template DNA for 5 min at 30°C. Next RNAP or the RNAP dilution buffer was added and incubated for another 10 min. Finally heparin was added to a final concentration of 200 ng µl−1 with a further incubation at 30°C for 5 min. An aliquot of this solution was loaded on a native gel to verify complex formation and the remaining solution was used for the different footprint experiments.

Primer extension analysis

Primer extension reactions with 25 µg of total RNA hybridized to a radiolabelled specific cas primer oligonucleotide (5′-ATACAATTAATCTATACATATATTAAGATG-3′) were performed with AMV reverse transcriptase (Promega) as described (Afflerbach et al., 1998).

Phage Lambda infection studies

Host sensitivity to phage infection was tested using a virulent phage Lambda (λvir), as before (Brouns et al., 2008). The host strains for infection were either wt E. coli K12 W3110, E. coli K12 W3110 Δhns, or the engineered E. coli K12 W3110 strains (E. coli K12ΔCRISPRI::CRISPRJ3 and E. coli K12ΔCRISPRI::CRISPRR44) (Table S4). The sensitivity of the host to infection was calculated as the efficiency of plaquing (the plaque count ratio of a strain containing an anti-Lambda CRISPR to that of the strain containing an non-targeting CRISPR) (Brouns et al., 2008).

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

This work was financially supported by a Veni grant to S.J.J.B. (863.08.014) and a Vici grant to J.v.d.O. (865.05.001) from the Dutch Organization for Scientific Research (Nederlandse Organisatie voor Wetenschappelijk Onderzoek). E.R.W. was financially supported by Spinoza resources awarded to Willem M. de Vos. Ü.P. was supported by the Deutsche Forschungsgemeinschaft PU435/1-1. N.H. was supported by a postdoc fellowship from the Swedish Research Council (Vetenskapsrådet). M.L. was supported by the Wenner-Gren Foundations. T.S. and K.S. were supported by the Deutsche Forschungsgemeinschaft Schn 371/10-1. A.R. and E.G.H.W. were supported by the Swedish Research Council.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
MMI_7315_sm_FigS1-S3_TabS1-S4.pdf244KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.