A total of 697 consecutive, non-duplicate clinical K. pneumoniae isolates were obtained from 11 hospitals in eight provinces in China between July 2018 and January 2019 (Supplementary Table S1). All of the isolates were initially identified as K. pneumoniae using a Vitek II system (bioMerieux, Marcy-l’Etoile, France) according to the manufacturer’s recommendations. K. pneumoniae was grown on Luria-Bertani (LB) agar or broth at 37°C. The isolates were stored at −80°C in LB broth containing 30% glycerol (v/v) until used.

DNA was extracted using the boiling method. The collected clinical isolates were tested for the presence of CRISPR-Cas systems (including type I-E^* and type I-E CRISPR systems) by polymerase chain reaction (PCR) using primers from a previous study (Li et al., 2018). PCRs were prepared using 2 × Hieff™ PCR Master Mix (with dye) (Yeasen) according to the manufacturer’s instructions (12.5 μL of Master Mix, 10 μM of each primer, and 1 μL DNA template, with a total volume of 25 μL).

Multi-locus sequence typing (MLST) was performed on type I-E^* and type I-E CRISPR-containing isolates by amplifying seven housekeeping genes, rpoB, gapA, mdh, pgi, phoE, infB, and tonB. STs were assigned by querying the MLST database of K. pneumoniae.¹

All of the clinical isolates were tested for susceptibility to 16 antimicrobial agents, including ertapenem, imipenem, and meropenem. Minimum inhibitory concentrations (MICs) of antimicrobial agents were determined using the agar dilution method of the Clinical and Laboratory Standards Institute (CLSI), while the broth dilution method was used for colistin and tigecycline (CLSI, 2019). For colistin, the breakpoints defined by the European Committee on Antimicrobial Susceptibility Testing (EUCAST)² were applied, whereas the breakpoints of tigecycline were interpreted according to the US Food and Drug Administration (FDA). E. coli ATCC 25922 was used as a quality control strain.

A total of 21 CRKP isolates carrying CRISPR-Cas systems were sequenced using a HiSeq X10 Sequencer (Illumina, San Diego, CA, United States), with 150 bp paired-end short reads and 200X coverage. Of the 21 isolates, 13 isolates were also subjected to long-read sequencing using a MinION Sequencer (Nanopore; Oxford, United Kingdom). Both short and long reads were utilized to generate a de novo hybrid assembly using Unicycler (Wick et al., 2017) and Pilon (Walker et al., 2014). CRISPRCasFinder³ was used to detect CRISPR arrays (repeats and spacers) and cas genes in the genomes. Acquired antimicrobial resistance genes and IS were identified by ResFinder 3.2 and ISfinder (Siguier et al., 2006) using BLAST (Zankari et al., 2012). tra gene cluster was identified by ICEfinder (Liu et al., 2019). Replicon STs of plasmids were determined using the pMLST database.⁴ Comparisons of the sequences were plotted using BRIG (Alikhan et al., 2011).

Fisher’s exact test was used for comparisons of the presence of CRISPR-Cas systems between the susceptible and resistant isolates. Chi-square tests were used to test the difference between carbapenem-susceptibility and CRISPR-Cas systems. Statistical significance was assessed using Prism 8 (GraphPad Prism) software. p-values of <0.05 were considered to be significant.

Overall, 23.5% (164/697) of clinical K. pneumoniae isolates had CRISPR-Cas systems, and type I-E^* was more common (15.9%, 111/697) than type I-E (7.7%, 54/697). One isolate had both type I-E^* and type I-E systems, which has not been reported previously in K. pneumoniae.

The 164 CRISPR-positive isolates comprised 41 different MLST types (23 isolates were untypable). Type I-E^* CRISPR-Cas was found in 18 STs, while type I-E CRISPR-Cas was present in 23 STs (Figure 1). The most prevalent ST among the isolates carrying type I-E^* CRISPR was ST23 (n = 51, 45.9%), followed by ST15 (n = 21, 18.9%). The most common ST among type I-E isolates was ST45 (14.8%), and the second was ST592 (11.1%). These results suggested that the distribution of CRISPR-Cas types was associated with particular STs among clinical isolates of K. pneumoniae. Among the 164 CRISPR-positive isolates, only one isolate (BJKP38) was ST11, indicating that CRISPR-Cas loci are rare in ST11 K. pneumoniae.

The CRISPR-positive isolates (harboring either the type I-E^* or the type I-E CRISPR-Cas system) were significantly more susceptible to 10 antimicrobials tested, including amikacin, aztreonam, cefotaxime, cefoxitin, ceftazidime, ciprofloxacin, ertapenem, fosfomycin, imipenem, and meropenem, compared with the CRISPR-negative isolates (Table 1). The rate of carbapenem-resistance of 697 K. pneumoniae was 37.1% (259/697), and the rates of carbapenem-resistance of CRISPR-positive and -negative isolates were 12.8% (21/164) and 44.7% (238/533), respectively (Table 2). Of the 21 CRKP isolates, 17 carried type I-E^* and four carried type I-E CRISPR-Cas systems.

Although most isolates carrying CRISPR-Cas systems were susceptible to carbapenems (87.2%, 143/164), there were still 21 isolates that showed resistance to carbapenems. These 21 isolates were subjected to whole-genome sequencing and genome analysis. More than half of the isolates (n = 13) harbored the carbapenemase-encoding gene bla_KPC-2 and type I-E^* CRISPR-Cas systems, and they were resistant to all three carbapenems tested. Eight other isolates, including all four isolates carrying type I-E CRISPR-Cas systems, did not produce any carbapenemase, and six of the isolates were resistant to ertapenem only, and the remaining two isolates were also resistant to the three carbapenems tested (Table 3).

The spacer sequences of CRISPR-Cas systems in these 21 isolates were analyzed and compared the difference between the two groups (13 isolates vs. 8 isolates, Supplementary Table S2). The 13 isolates with bla_KPC-2-bearing plasmids mainly carried 18 spacers, of which ten spacers were present in 12 isolates, six in 11 isolates, and two in seven isolates. The eight isolates without KPC-2 contained 166 spacers, and each one existed in one to three isolates.

MLST analysis showed that among the 13 isolates carrying both type I-E^* CRISPR-Cas systems and bla_KPC-2, 12 belonged to ST15 and one was ST23 (Table 3), while only eight (5.6%, 8/143) isolates belonged to ST15 in carbapenem-susceptible K. pneumoniae carrying CRISPR-Cas systems. The 12 ST15 isolates were collected from three hospitals, including Huashan Hospital (n = 8) and Ruijin Hospital (n = 3) in Shanghai, and the First Affiliated Hospital of Guangzhou Medical University (n = 1). The isolation dates of the 12 isolates were from July 2018 to January 2019.

The bla_KPC-2 genes were located on plasmids in all 13 isolates. Of note, bla_KPC-2-carrying plasmids of nine isolates belonged to IncFII_K34 (Table 3), and the sequences of the nine plasmids were highly conserved (Figure 2). The IncFII_K34 replicon was first identified by our group during a study on ST23 CR-HvKP in China and has been submitted to the plasmid MLST database. The IncFII_K34 plasmid was recognized by comparing the sequence of the copA region of the plasmid with the other IncFII_K alleles in the database. All the nine IncFII_K34 plasmids harbored bla_KPC-2, meanwhile, pHSKP33 harbored dfrA14 and qnrS1 and pHSKP39 harbored aac(3)-IId, bla_SHV-12 and bla_TEM-1B additionally. Besides, tra gene cluster that encodes proteins needed for conjugation was identified on all the IncFII_K plasmids, while was absent on IncFII(PHN7A8) plasmids. The bla_KPC-2-carrying plasmid types of HSKP1 and RJKP41 were IncFII(PHN7A8), the most common type among ST11 KPC-producing K. pneumoniae, and the plasmid types of the remaining two isolates were IncFIA(HI1) and IncR.

Among the 13 isolates that carried both type I-E^* CRISPR-Cas systems and bla_KPC-2-bearing plasmids, 11 isolates had self-targeting spacers matching their respective protospacers on bla_KPC-2-bearing plasmids, of which two isolates (HSKP1 and RJKP41) had point mutations. The sequence of the self-targeting spacer was 5’-CCGCCGTTT A ATCGCGGTGATGATATCCGGCA-3′ (the point mutation is shown in bold and underlined). The remaining two isolates (GZKP13 and HSKP39) had no spacer matching their bla_KPC-2 plasmids. The sequences of hns gene encoding transcriptional repressor H-NS, which showed to prevent the plasmids transformation (Lin et al., 2016), from 13 isolates were identical to that from K. pneumoniae NTUH-K2044. In addition, no mutation of CRISPR-Cas loci was identified on 13 genomes compared to these sequences from CRISPR-Cas finder database.