Imipenem represses CRISPR-Cas interference of DNA acquisition through H-NS stimulation in Klebsiella pneumoniae

Analysis of the genome of Klebsiella pneumoniae NTUH-K2044 strain revealed the presence of two clustered regularly interspaced short palindromic repeats (CRISPR) arrays separated with CRISPR-associated (cas) genes. Carbapenem-resistant K. pneumoniae isolates were observed to be less likely to have CRISPR-Cas than sensitive strains (5/85 vs. 22/132). Removal of the transcriptional repressor, H-NS, was shown to prevent the transformation of plasmids carrying a spacer and putative proto-spacer adjacent motif (PAM). The CRISPR-Cas system also decreased pUC-4K plasmid stability, resulting in plasmid loss from the bacteria with acquisition of new spacers. Analysis of the acquired proto-spacers in pUC-4K indicated that 5′-TTN-3′ was the preferred PAM in K. pneumoniae. Treatment of cells by imipenem induced hns expression, thereby decreasing cas3 expression and consequently repressed CRISPR-Cas activity resulted in increase of plasmid stability. In conclusion, NTUH-K2044 CRISPR-Cas contributes to decrease of plasmid transformation and stability. Through repression of CRISPR-Cas activity by induced H-NS, bacteria might be more able to acquire DNA to confront the challenge of imipenem.


Prevalence of CRISPR-Cas among drug resistant and sensitive strains. CRISPR-Cas was
demonstrated to interfere with the transformation and stability of plasmids that often carry drug resistant genes [5][6][7] . Accordingly, we assessed whether presence of CRISPR-Cas in clinical strains is related to their drug resistance. PCR using cas1, CRISPR1, and CRISPR2 primers found that all of the cas1 PCR positive strains also had at least one CRISPR array (CRISPR1 or CRISPR2). The prevalence of CRISPR-Cas system was 5/85 in carbapenem-resistant strains and 22/132 in drug-sensitive intestinal strains, showing a highly significant inverse correlation between prevalence and resistance (P = 0.0205, chi-square test). The low frequency of CRISPR-Cas system in drug resistant K. pneumoniae implied that CRISPR-Cas may play a role in preventing acquisition of drug resistance genes.
The function of NTUH-K2044 CRISPR-Cas in plasmid transformation. The CRISPR-Cas system was reported to be involved in resistance to plasmid transformation [5][6][7] . Recent studies reveal that the recognition of proto-spacer adjacent motif (PAM) is essential for the targeting of CRISPR-Cas 10 . Therefore, an engineered proto-spacer containing spacer sequences identical to spacer 2 ( Fig. 1) in the NTUH-K2044 CRISPR2 and the predicted PAM sequence (5′-CTT-3′) of type I-E CRISPR-Cas 11,12 were cloned into a pUC-4K plasmid. Then, the transformation efficiencies of plasmids with or without engineered proto-spacer sequences were compared. Inserting proto-spacer sequences into a pUC-4K plasmid significantly decreased the transformation efficiencies in wild type strain by approximately 50% (Fig. 2).
A histone-like DNA-binding protein that can modulate gene expression globally, H-NS has been shown to repress the expression of CRISPR-Cas in E. coli [13][14][15] . Therefore, we examined whether H-NS represses the CRISPR-Cas in K. pneumoniae NTUH-K2044 strain. Deletion of hns revealed a tenfold increase in the absolute transformation efficiency with a pUC-4K plasmid (data not shown). The transformation efficiency of a plasmid with an added proto-spacer sequence was significantly decreased to approximately 3% of that of a control plasmid in NTUH-K2044 hns deletion (∆hns) mutant (Fig. 2). The cas3 gene encoding a DNA nuclease/helicase responsible for degradation of targets in CRISPR interference 1,2 was further deleted in ∆hns mutant to determine whether CRISPR-Cas interfered with transformation in the ∆hns mutant. Deletion of cas3 significantly Figure 1. The structure of CRISPR-Cas system in K. pneumoniae NTUH-K2044 strain. A cas locus located between two CRISPR arrays (CRISPR1 and CRISPR2) in K. pneumoniae NTUH-K2044 strain was revealed. CRISPR1 array composed of four repeated sequences (solid diamond) and three spacer sequences (open square) was located at nucleotides 3011356 to 3011567. CRISPR2 array composed of twenty-three repeated sequences and twenty-two spacer sequences was located at nucleotides 3001234 to 3002603. Sequences of spacer 2 flanked by repeat sequences in CRISPR2 and engineered proto-spacer 2 (sequences matching spacer 2 and putative PAM) are shown. restored the transformation efficiency in ∆hns mutant, of the plasmids carrying proto-spacer sequences (Fig. 2). Therefore, in the absence of the CRISPR-Cas repressor, H-NS, CRISPR-Cas in NTUH-K2044 could prevent the transformation of plasmids carrying spacer sequences identical to those of CRISPR.
The hns complementation strain was also generated to further confirm the repression effect of H-NS on CRISPR-Cas activity. No interference with transformation of plasmids carrying sequences identical to those of CRISPR spacers was apparent in the hns complementation strain (Fig. 2).
H-NS repressed CRISPR-Cas RNA expression. The expression of cas3 RNA in the NTUH-K2044 wild type, ∆hns mutant, and hns complementation strains was determined by quantitative RT-PCR. The expression of cas3 RNA was significantly increased by deletion in the ∆hns mutant and restored to wild-type levels in the hns complementation strain (Fig. 3). Our results in line with a previous study 16 confirmed that H-NS could repress the expression of CRISPR-Cas.
The function of NTUH-K2044 CRISPR-Cas system in phage resistance. The CRISPR-Cas system was reported to be involved in phage resistance [3][4][5] . In our previous study, we isolated a bacteriophage, Figure 2. The transformation efficiencies of pUC-4K and pUC-4K with engineered proto-spacer 2 identical with spacer 2 of CRISPR2 in NTUH-K2044 wild type, hns deletion (∆hns), hns and cas3 double deletion (∆hns∆cas3) and hns complementation strains. The transformation efficiency of pUC-4K was set as 100% and that of proto-spacer2(CTT)::pUC-4K was calculated accordingly. Data are presented as means ± SEM from three independent experiments. NTUH-K2044-K1-1, that infects K. pneumoniae NTUH-K2044 17 . In order to test the function of NTUH-K2044 CRISPR-Cas system in phage resistance, clones of NTUH-K2044 ∆hns mutant (with de-repressed expression of CRISPR-Cas) showing resistance to phage NTUH-K2044-K1-1 infection were selected after overnight co-culture of bacteria and phage. Consistent with the finding that capsules are essential for infection by this phage, the majority of the resistant clones were found to be non-encapsulated. Therefore, the non-encapsulated variants are not expected to have acquired new spacers and were not tested further. We identified 427 encapsulated clones after screening of approximately one hundred thousand resistant-clones and tested those for new spacers. However, no new spacers were found. Therefore, CRISPR-Cas seemed not to be involved in resistance to phage NTUH-K2044-K1-1.
The function of NTUH-K2044 CRISPR-Cas in plasmid stability. The CRISPR-Cas system was reported to decrease plasmid stability 5 . The stability of a pUC-4K plasmid (with kanamycin resistance) transformed in NTUH-K2044 ∆hns mutant (having de-repressed expression of CRISPR-Cas) was examined. Ten out of two hundred clones became sensitive to kanamycin and were confirmed to have lost the pUC-4K plasmid, after six passages of the bacteria in LB broth without kanamycin. New spacers in the CRISPR2 array were found to have been acquired by three clones that had lost the plasmid (Fig. 4A). Sequences of new spacers were identical to the sequences of pUC-4K plasmid. After six passages of ∆hns∆cas3 mutant carrying a pUC-4K plasmid in LB broth, none of 200 clones was sensitive to kanamycin. These results demonstrated that CRISPR-Cas expression could decrease plasmid stability in K. pneumoniae. PAM sequences analysis. PAM sequences in K. pneumoniae were identified by first determining the sequences of newly acquired spacers. The CRISPR2 sequences in 7 clones out of another 39 ∆hns mutant clones that had lost the pUC-4K plasmid were observed to be elongated (Fig. 4A). The sequences of newly acquired spacers in ten clones (three clones described above and seven clones here) were analyzed. Each clone had acquired one to three spacers, and the new spacers of two clones (6-2 and 6-5) were identical. In all, 14 new spacers had been identified. Their proto-spacers and adjacent sequences on the pUC-4K plasmid were aligned and analyzed. The only PAM sequence identified in K. pneumoniae NTUH-K2044 (12/14, 86%) was 5′-TTN-3′ (Fig. 4B).
The function of CRISPR1 array on efficiency of plasmid transformation. As shown in Fig. 4A, naturally acquired new spacers were all integrated into the CRISPR2 array. Therefore, CRISPR1 seems to be non-functional for adaptation. To examine the function of CRISPR1 array in interference of plasmid transformation, proto-spacer sequences identical with spacer 2 in the CRISPR1 flanked by PAM sequence (5′-TTT-3′) were further cloned into a pUC-4K plasmid. Then, the transformation efficiencies of plasmids with or without proto-spacer sequences were compared both in wild type and ∆hns mutant strain (Fig. 5). In contrast to significant interference conferred by proto-spacer2 which was targeted by crRNA transcribed from CRISPR2 array (as described in Fig. 4C), no interference was detected in transformation of plasmids bearing proto-spacers [CRISPR1-spacer2(TTT) and CRISPR1-spacer2′(TTT)] which were targeted by two-directional crRNA transcribed from CRISPR1 (Fig. 5). Therefore, CRISPR1 seems to be also non-functional for interference.

Expression of hns and CRISPR-Cas system under imipenem treatment. CRISPR-Cas activity in
K. pneumoniae NTUH-K2044 strain was demonstrated to be repressed by H-NS. H-NS is considered a global regulator of gene expression in response to environmental stimuli; hence, whether imipenem acts through regulation of H-NS to alter the expression of CRISPR-Cas was further examined. The RNA expressions of hns and cas3 were determined by quantitative RT-PCR after treatment with different concentrations of imipenem (0, 0.125, 0.25 and 0.5 μg/ml) for 3 hours. The bacterial growth curves under different concentrations of imipenem shown in Fig. 6A revealed that the bacterial growth was suppressed after treatment with 0.25 and 0.5 μg/ml of imipenem for 3 hours. The bacterial morphology under microscopic examination was not significantly affected after treatment with imipenem for 3 hours (data not shown). Imipenem induced hns RNA expression and inhibited cas3 RNA expression in NTUH-K2044 wild type strain (Fig. 6B), whereas the cas3 RNA expression was not altered under imipenem treatment in ∆hns mutant (Fig. 6C). These results suggested that imipenem increases hns expression and that the induced H-NS subsequently decrease cas3 expression.
The imipenem effect on hns expression was further examined in an imipenem resistant strain N308. The RNA expressions of hns and cas3 were determined by quantitative RT-PCR after treatment with different concentrations of imipenem (0, 2, 4 and 8 μg/ml) for 3 hours. The expression of hns was also induced and expression of cas3 was inhibited in N308 strain after treatment with 4 and 8 μg/ml of imipenem which resulted in suppression of bacterial growth. (Fig. 6D).

CRISPR-Cas activity under imipenem treatment.
To study whether imipenem treatment indeed represses the activity of CRISPR-Cas, we first determined the basal activity of CRISPR-Cas in NTUH-K2044.
To confirm the observation of regulation of hns RNA and cas3 RNA expression by imipenem treatment, pUC-4K plasmid stabilities in NTUH-K2044 and NTUH-K2044 (6-1 spacer) was examined with or without imipenem treatment. The pUC-4K plasmid was lost in 63/300 (21%) NTUH-K2044 (6-1 spacer) clones subcultured in LB broth for 8 hours. The loss was significantly decreased (14/300, 4.7%) by the addition of 0.5 μg/ml of imipenem to the subculture, whereas all 300 clones of NTUH-K2044 with or without imipenem treatment were still plasmid-containing. These results indicated the CRISPR-Cas-mediated interference is repressed by imipenem and suggest that exposure of K. pneumoniae to imipenem inhibits CRISPR-Cas activity thereby giving K. pneumoniae a greater opportunity to acquire resistant genes.   (Fig. 6A). The RNA levels of hns and cas3 treated with different imipenem concentrations in NTUH-K2044 (Fig. 6B), ∆hns strain (Fig. 6C) and imipenem resistant N308 strain (Fig. 6D). The RNA expressions of hns and cas3 were determined by quantitative RT-PCR after treatment with different concentrations of imipenem for 3 hours; meanwhile the bacterial growth was determined by absorbance measurement under 600 nm. The RNA level without treatment was set as 1, and those with imipenem treatment were calculated accordingly. Data are presented as means ± SEM from three independent experiments. *P-values of <0.05 were considered significant (Student's t-test).
Scientific RepoRts | 6:31644 | DOI: 10.1038/srep31644 Discussion A recent study surveyed CRISPR-Cas systems in K. pneumoniae genomes 18 . CRISPR-Cas systems were detected using bioinformatics tools in only 6 out of 52 complete and draft genomes of K. pneumoniae; therefore, the CRISPR-Cas system is not widely distributed in K. pneumoniae. Blast search analysis showed that 33% (38/116) of spacer sequences were very similar to plasmid, phage, or bacterial genome sequences. In this study, anti-plasmid immunity in K. pneumoniae NTUH-K2044 strain was directly attributable to CRISPR-Cas activity. Moreover, the regulation of CRISPR-Cas systems and their correlation with antibiotic resistance in K. pneumoniae were also revealed.
H-NS has been shown to repress the CRISPR-Cas system in E. coli through direct binding to the promoter of the cas operon 14 . In contrast to endogenous expression of CRISPR-Cas in E. coli carrying an engineered spacer identical to the corresponding phage lambda sequence, which provided only weak protection against phage infection, disruption of the hns gene conferred a high level of protection 13 . These findings were in line with our results in K. pneumoniae. As shown in Figs 2 and 4C, the activity of CRISPR-Cas in K. pneumoniae NTUH-K2044 strain is not cryptic, but interference with plasmid transformation was prominent in ∆hns mutant. The repression of cas3 expression by hns was also confirmed by quantitative RT-PCR. Besides, the acquisition of new spacers (adaptation stage of CRISPR-Cas) was also observed only in those ∆hns mutant clones that had lost their plasmids under non-selective conditions. Therefore, the regulation of CRISPR-Cas activity seems to be conserved between E. coli and K. pneumoniae.
The role of CRISPR-Cas in phage resistance was first described in Streptococcus thermophilus 3 . Despite numerous attempts, we failed to isolate ∆hns mutant clones resistant to phage NTUH-K2044-K1-1 due to the acquisition of new spacers and their incorporation into CRISPR arrays. Our previous study indicated that components of the capsule may be receptors of phage NTUH-K2044-K1-1 17 , which would explain why most resistant clones were non-encapsulated and lacked detectable CRISPR-Cas activity. Therefore, the function of CRISPR-Cas in the phage resistance of K. pneumoniae should be investigated further using other phage strains.
Even though two CRISPR arrays (CRISPR1 and CRISPR2) were detected in K. pneumoniae NTUH-K2044 strain, naturally acquired new spacers were all integrated into the CRISPR2 array. Moreover, there was no decrease detected in plasmid transformation conferred by CRISPR1. Therefore, CRISPR1 appears to be non-functional for both adaptation and interference. There were five variable base-pairs revealed in the repeat sequences of CRISPR1 array, which might be the reason contributed to the defect of CRISPR1.
In our study, 5′-TTN-3′ was identified as the preferred PAM sequence in K. pneumoniae and differed from the preferred PAM sequence (5′-CTT-3′) in the best-studied type I-E CRISPR-Cas system of E. coli 11,12 . Two non-consensus PAM sequences were observed (5′-ATT-3′ and 5′-GTT-3′), but the ∆hns (6-10 spacer) strain carrying a spacer with 5′-GTT-3′ also had another spacer bearing the consensus PAM sequence. A previous study reported multiple spacers integrated into a single E. coli clone, all targeting the same DNA strand 11 . However, an exception was observed in our study, that is, a ∆hns (6-10 spacer) strain carrying two spacers targeting different strands. Therefore, the mechanism of adaptation still awaits further study.
Inverse correlation between the presence of CRISPR-Cas loci and carbapenem resistance was also revealed in K. pneumoniae in this study. The function of K. pneumoniae CRISPR-Cas in adaptation and interference was fully demonstrated in the absence of H-NS, confirming activated CRISPR-Cas indeed could prevent the assimilation of foreign DNA such as antibiotic resistance genes. Researchers hypothesized that H-NS will bind to invading AT-rich DNA, then consequently free the cas promoter, thus bolstering the defense against foreign DNA by CRISPR-Cas 19 . Otherwise, the exchange of DNA facilitates the rapid adaptation of bacteria to environmental change. The gaining of new genetic material needs to be delicately balanced against the limiting of horizontal gene transfer. In this study, we observed that imipenem treatment caused decreased CRISPR-Cas activity by induced H-NS expression. Through this regulatory mechanism, bacteria might become more able to acquire resistance genes to confront the challenge of antibiotics.
The imipenem effect on hns expression level was observed in both imipenem sensitive and resistant strains when encountered suppression of growth. Previous study also demonstrated the expression of hns was induced under cold-shock stress 20 . Therefore, we suggested that unrelated antibiotic or stress might have the same effect on hns expression level. We also examined the imipenem effect on transformation efficiency. However, electroporation caused approximately 100-fold bacterial deaths in imipenem-treated bacteria than in untreated bacteria. The transformation efficiency should be better analyzed by natural transformation rather than by electroporation. However, ability of natural transformation was not observed in our strain.
Besides involvement in defense against foreign DNA, CRISPR-Cas has been correlated with bacterial virulence 21 . A recent study indicated that the type II CRISPR-Cas system of Legionella pneumophila has an important role in intracellular survival and replication in amoebae 22 . The type I-F CRISPR-Cas system of Pseudomonas aeruginosa has been demonstrated to be involved in biofilm formation and swarming 23 . Cas9 of the type II-B CRISPR-Cas system from Francisella novicida has been shown to repress the expression of bacterial lipoprotein (BLP) and thereby contributes to immune avoidance during infection 24 . Therefore, whether the CRISPR-Cas of K. pneumoniae has other roles such as bacterial virulence needs further study.
In conclusion, NTUH-K2044 CRISPR-Cas contributes to decrease of plasmid transformation and plasmid stability. The basal activity of CRISPR-Cas can be repressed through induction of H-NS by imipenem.

Methods
Bacterial strains. The K. pneumoniae NTUH-K2044 strain causing pyogenic liver abscess and meningitis 25 , and 85 carbapenem-resistant K. pneumoniae (CRKP) strains obtained from four hospitals located in north or south Taiwan as described elsewhere were used in this study 26  The stool specimens were collected in fecal occult blood test tubes, stored at 4 °C, cultured on EMB agar plates, and identified as K. pneumoniae using the Enterotube system (BD, NJ, USA). This study protocol was approved by the Institutional Review Board of National Taiwan University Hospital (IRB approval number: 9561701018). The methods were carried out in accordance with the approved guidelines and written informed consent was obtained from each participant.
Both K. pneumoniae and E. coli were grown in Luria-Bertani (LB) broth or agar at 37 °C, except as noted below. Where appropriate, medium was supplemented with kanamycin (50 μg/mL) or sucrose (5%).

Construction of hns complementation strain.
The hns and its putative promoter were PCR amplified (5′-GGTCGACTTACCTGCATTC-3′ and 5′-TTAGATCAGGAAATCGTCCAG-3′) and cloned into the intergenic region of the two open reading frames, pgpA and yajO, in a pKO3-Km-pgpAyajO recombinant vector 28 . The resulting construct was electroporated into a ∆hns mutant strain. The complementation strain was selected as previously described 28 and confirmed by PCR as well as sequencing with appropriate primers.

Transformation efficiency.
Aliquots of approximately 1 × 10 9 cfu of K. pneumoniae bacteria in 100 μl of 10% glycerol were mixed with 1 μg of plasmid DNA, and then shocked by using an electroporator (BTX ECM630, MA, USA) with settings as below (Voltage = 2500 V, Resistance = 200 ohms, Capacitance = 50 μf). After recovery in LB broth for one hour, the total and transformed bacterial numbers were enumerated by plating after serial dilutions on LB and LB supplemented with kanamycin plates, respectively. The transformation efficiency was calculated accordingly.
Plasmid stability. NTUH-K2044 and ∆hns mutant transformed with pUC-4K plasmid was used to inoculate 5 ml of LB broth. Fifty microliters of the previous culture was inoculated into 5 ml of fresh LB medium every morning (37 °C) and night (30 °C) for six passages. For each culture, bacteria were plated on LB plates and kanamycin-sensitive colonies were screened after replica on LB plates with kanamycin. The CRISPR1 and CRISPR2 of the kanamycin-sensitive clones were detected by PCR (CRISPR1-F and CRISPR1-R for CRISPR1; CRISPR2-F and CRISPR2-S2R 5′-GTGATAGAGATCGTGGTG-3′ for CRISPR2) and sequencing.