Efficient Suppression of Natural Plasmid-Borne Gene Expression in Carbapenem-Resistant Klebsiella pneumoniae Using a Compact CRISPR Interference System

ABSTRACT

There is an urgent need for efficient tools for genetic manipulation to assess plasmid function in clinical drug-resistant bacterial strains. To address this need, we developed an all-in-one CRISPR interference (CRISPRi) system that easily inhibited the gene expression of a natural multidrug-resistant plasmid in an sequence type 23 (ST23) Klebsiella pneumoniae isolate. We established an integrative CRISPRi system plasmid, pdCas9gRNA, harboring a dcas9 gene and a single guide RNA (sgRNA) unit under the control of anhydrotetracycline-induced and J23119 promoters, respectively, using a one-step cloning method. This system can repress the single resistance gene bla_NDM-1, with a >1,000-fold reduction in the meropenem MIC, or simultaneously silence the resistance genes bla_NDM-1 and bla_SHV-12, with a 16-fold and 8-fold respective reduction in the meropenem and aztreonam MIC on a large natural multidrug-resistant pNK01067-NDM-1 plasmid in an ST23 K. pneumoniae isolate. Furthermore, an sgRNA targeting the bla_NDM-1 promoter region can silence the entire bla_NDM-1-ble_MBL-trpF operon, confirming the existence of the operon. We also used this tool to knock down the multicopy resistance gene bla_KPC-2 in pathogenic Escherichia coli, increasing the susceptibility to meropenem. In a word, the all-in-one CRISPRi system can be used for efficient interrogation of indigenous plasmid-borne gene functions, providing a rapid, easy genetic manipulation tool for clinical K. pneumoniae isolates.

INTRODUCTION

Carbapenem-resistant Klebsiella pneumoniae (CRKP) is a major threat to human health and has recently aroused increasing attention (1–3). The resistance of CRKP is mainly derived from large plasmids containing multiple resistance genes, such as bla_NDM-1, bla_KPC-2, or extended-spectrum β-lactamase (ESBL)-coding genes (4–7). These plasmids also have other important functional genes, including virulence, host-related regulation, and conjugation genes (8). Understanding the function of these genes on plasmids is a key step toward preventing and treating infections caused by CRKP. Nevertheless, until now, there have been few efficient genetic manipulation tools for investigating functional genes on large plasmids in clinical K. pneumoniae hosts.

Traditional gene knockout strategies in K. pneumoniae, such as allelic exchange and transposon insertion inactivation, are time-consuming and laborious (9–11). In the last decade, clustered regularly interspaced short palindromic repeat (CRISPR) systems have been applied in eukaryotic cells and bacterial genome engineering (12, 13). The most commonly used CRISPR-Cas9 tool produces DNA double-stranded breaks (DSBs) and performs knockout via homology-directed repair (HDR) templates (13). CRISPR-Cas9 has been used to efficiently delete chromosomal genes in K. pneumoniae (14, 15). However, inactivating the large plasmid-borne gene with this system resulted in plasmid loss from K. pneumoniae (16), which impeded functional research on plasmid-borne genes. Furthermore, the low homologous recombination efficiency in K. pneumoniae ultimately prevented researchers from obtaining desirable gene mutants, even when combined with the exogenous λ-Red recombination system. In addition, cytidine base editors (CBEs), which combine catalytically impaired nuclease Cas9 nickase (nCas9) and a deaminase, create C-to-T base conversions in target genes on natural plasmids, but the editing efficiency is largely dependent on the active window (15). A proactive genetics (Pro-AG) gene-drive system inactivates the ampicillin resistance determinant on a high-copy-number plasmid in Escherichia coli MG1655. However, this system needs at least two established plasmids and usually uses the exogenous recombinase λ-Red system to improve the repair ability (17), which increases the growth burden of the bacteria and plasmid construction steps.

The CRISPR interference (CRISPRi) system, composed of a catalytically inactive Cas9 (dCas9) nuclease and a single guide RNA (sgRNA), is used to inhibit gene expression, ideally suitable for investigation of essential genes or plasmid-derived genes (18, 19). CRISPRi technology, which does not produce DSB or use HDR, has been widely used to investigate gene function (e.g., the discovery of drug-resistant targets [19], explorations of pathogenicity and virulence [20], analyses of bacterial cell growth phenotypes [21, 22], the enhancement of bacterial metabolites [23, 24], and studies of genetic interactions [19, 25]). In addition, CRISPRi can be integrated into the K. pneumoniae chromosome and block expression of the chromosomal gene folA, increasing susceptibility to trimethoprim (26). However, an easy-to-use and efficient system for the genetic manipulation of natural plasmids is lacking for clinical K. pneumoniae isolates.

In this study, we established an all-in-one CRISPRi tool for CRKP. This system not only efficiently turned off the expression of a single resistance gene but also simultaneously knocked down multiple resistance genes on a large multidrug-resistant (MDR) plasmid. Furthermore, it was used to verify the existence of a potential operon. Using this system, we were able to easily prevent the transcription of multicopy genes on large plasmids in clinically pathogenic E. coli. In conclusion, this new CRISPRi tool provides a paradigm for the rapid investigation of large plasmid-borne gene function in clinical, complex isolates.

RESULTS AND DISCUSSION

Establishment of an all-in-one CRISPRi system.

To study the functions of genes on large plasmids of clinical MDR K. pneumoniae, we developed an all-in-one CRISPRi system with an anhydrotetracycline (aTc)-inducible dCas9 expression cassette, a programmed sgRNA unit, a chloramphenicol selectable marker, and a high-copy-number ColE1 origin of replication (Fig. 1A). Construction of this platform was based on the bacterial dCas9 expression vector (Addgene number 44249) and the sgRNA expression vector (Addgene number 44251) (27). The dCas9 protein binds to the sgRNA and forms a protein-RNA complex. Through sgRNA matching with the protospacer of target gene regions, the dCas9-sgRNA complex can bind to the corresponding loci and block transcript elongation by RNA polymerase, leading to transcriptional repression (Fig. 1B). Using the CRISPRi platform, phenotypic identification of plasmid-borne genes takes only 3 days, including the construction of sgRNA plasmids via one-step cloning (day 1) (Fig. 1C), the transformation of sgRNA plasmids into clinical strains by electroporation (day 2), and a gene phenotypic assay with quantitative real-time PCR (qRT-PCR) and MIC testing (day 3). The tool saves time compared to conventional double-exchange knockout approaches (28). Its detailed construction is given in the methods.

FIG 1.

Schematic of the all-in-one CRISPRi tool. (A) Construction of the CRISPRi system (i.e., plasmid pdCas9gRNA). The plasmid contains an aTc-inducible ptet promoter, a Cm^R, a high-copy-number ColE1 origin, and a strong constitutive J23119 promoter with an sgRNA unit. The sgRNA unit consists of three parts: a base pairing region (BPR), a dCas9 handle region (Handle), and transcription terminator (Termi). (B) Overviews of CRISPRi mechanisms. sgRNA-directed dCas9 blocks RNA polymerase binding and elongation. (C) The one-step cloning procedure for the all-in-one CRISPRi system. N20, 20-nucleotide BPR.

Effectively turning off the expression of bla_NDM-1 on a natural large plasmid in a CRKP isolate using CRISPRi.

The MDR plasmid pNK01067-NDM-1 (52 kb), derived from an sequence type 23 (ST23) CRKP, carries bla_NDM-1 metallo-beta-lactamase, conferring high-level resistance to carbapenems. To verify the effects of our CRISPRi system, we selected plasmid-borne bla_NDM-1 as a target gene. Nine sgRNAs were designed for different bla_NDM-1 regions. Of these, gNDM3, 8, and 9 target the bla_NDM-1 promoter region, and the remainder (gNDM1, 2, and 4 to 7) target coding regions (Fig. 2A). The qRT-PCR assay showed that gNDM3 and gNDM8, targeting the nontemplate strand in the promoter region, resulted in respective 900- and 80-fold decreases in mRNA expression compared to wild-type NK01067 (control) with aTc induction (Fig. 2B). We also detected the expression levels of the control and empty vector without aTc induction. There were no statistical differences between them (see Fig. S1 in the supplemental material), indicating that aTc does not affect the expression of the target genes.

FIG 2.

Highly efficient silencing of the bla_NDM-1 gene on the nontemplate DNA strand. (A) Nine sgRNAs targeted the promoter region on nontemplate (gNDM3 and gNDM8) and template (gNDM9) DNA strands and the coding region on nontemplate (gNDM1, gNDM6, and gNDM7) and template (gNDM2, gNDM4, and gNDM5) DNA strands. (B) Nine sgRNAs and an empty vector were tested for relative mRNA transcription levels. Wild-type strain K. pneumoniae NK01067 was used as a control. (C) Meropenem MIC values of different knockdown strains. The number above the column is the MIC value. Statistical analysis was performed by a one-way analysis of variance (ANOVA), followed by Dunnett’s multicomparison test, using GraphPad. ns, not significant; **, P < 0.01; ****, P < 0.0001.

gNDM9, targeting the template strand of the promoter region, only decreased it by 11-fold. Targeting the nontemplate strand of the coding region, gNDM1, 6, and 7 caused 2- to 16-fold reductions. Targeting the template strand of the coding region, gNDM2, 4, and 5 did not produce any changes. These experiments indicated that an sgRNA targeting the nontemplate strand is more efficient than one targeting the template strand in both the promoter and coding regions, consistent with previous reports (19, 27, 29). We also found that inhibiting transcription initiation by targeting the promoter region is more efficient than transcription elongation, probably because the protein-sgRNA complex does not form a hard roadblock in the coding region and still allows for some passage of RNA polymerase. Interestingly, sgRNAs matching different gene regions or DNA strands lead to differential expression levels, which can be applied to control the transcription strength.

We tested the meropenem resistance levels of the target strains (Fig. 2C; Fig. S2). Compared with the control or empty vector, these target strains had different resistance levels (from 32 to 0.03 μg/mL) in the presence of aTc induction. Among them, strain gNDM3 showed the most dramatic changes (>1,000-fold), which matched the mRNA expression levels. To investigate whether dCas9 can express in the absence of aTc, we tested the mRNA expression of dCas9 with and without aTc (Fig. S3). The result indicated that dCas9 can be slightly expressed without aTc. This leaky expression was also reported in previous studies (30, 31). However, the level of expression was significantly lower than that achieved with aTc. The MIC levels for these target strains were consistent with those in wild-type or empty vector controls, except in gNDM3- and gSHV12-related strains (Table S1). This is likely due to the strong repression in gNDM3 and gSHV12.

To test whether the CRISPRi system influences plasmid replication, we determined the copy number of bla_NDM-1 and another plasmid-borne gene, repA. The results clearly showed that both target genes (bla_NDM-1 and repA) were stable, with no significant difference in copy number compared with the control (Fig. S4). This indicates that the CRISPRi system does not impede plasmid replication.

The CRISPRi system can inhibit bla_SHV-12 and bla_NDM-1 expression simultaneously.

Plasmid pNK01067-NDM-1 also contains bla_SHV-12 (aztreonam resistance), in addition to bla_NDM-1. The chromosome of K. pneumoniae NK01067 also encodes bla_SHV-11 (Fig. 3A), which was reported as an intrinsic bla_SHV that is not expressed normally (32). To determine whether CRISPRi can repress the expression of multiple resistance genes, we designed dual sgRNAs to target bla_SHV-12 and bla_NDM-1. The sgRNA of bla_SHV-12, gSHV12, was designed to bind only the promoter region of bla_SHV-12, not that of bla_SHV-11, due to the high degree of similarity between bla_SHV-11and bla_SHV-12 (Fig. 3A). The sgRNA of bla_NDM-1 was still gNDM3. The two sgRNAs were together constructed in pdCas9gRNA. The mRNA expression levels of the gNDM3 plus gSHV12 combination showed significant suppression, decreasing by at least 100-fold, similar to that for the single gNDM3 or gSHV12 knockdown strains (Fig. 3B). The corresponding meropenem resistance levels of the combination fell sharply from 32 to 2 μg/mL. The aztreonam resistance decreased by 8-fold (from 256 to 32 μg/mL) (Table 1). We also verified the ESBL activity of bla_SHV-11 and bla_SHV-12 using pUC57 in E. coli. The results showed that bla_SHV-11 does not display aztreonam resistance, and bla_SHV-12 has significant aztreonam resistance (Table S2), which was consistent with the previous report (32).

FIG 3.

Knockdown of both bla_NDM-1 and bla_SHV-12 using multiplexed CRISPRi. (A) The locations of three antibiotic resistance genes; bla_NDM-1 and bla_SHV-12 are located on the plasmid, and bla_SHV-11 is located on the chromosome. gNDM3 targeting bla_NDM-1, and gSHV12 targeting bla_SHV-12. (B) qRT-PCR was used to detect the mRNA levels in the knockdown strains gNDM3, gSHV12, and gNDM3+gSHV12. mRNA expression was determined as the level of the sgRNAs relative to no sgRNA. All experiments consisted of at least three replicates. The error bars indicate the standard deviation. A statistical analysis was performed using a pairwise two-way ANOVA with Sidak’s test, using GraphPad. ns, not significant; ****, P < 0.0001.

TABLE 1.

MIC for diverse mutants and wild-type strains in the study^a

Strain	MIC of:
	MEM (μg/mL)	ATM (μg/mL)
NK01067	32	256
Empty vector	32	256
gNDM3	0.03	256
gSHV12	32	1
gNDM3+gSHV12	2	32

^aMEM, meropenem; ATM, aztreonam.

We concluded that the CRISPRi system can rapidly downregulate the expression of both bla_SHV-12 and bla_NDM-1 resistance genes on natural MDR plasmids.

Our CRISPRi tool can quickly validate an intact operon.

A previous study showed that bla_NDM-1 and ble_MBL constitute an operon (33). Through analysis of the plasmid pNK01067-NDM-1 sequences, we found that the expression direction of the trpF gene closing on ble_MBL was consistent with that of the upstream bla_NDM-1-ble_MBL operon (Fig. 3A). Currently, no experimental studies have demonstrated that trpF shares a promoter with the bla_NDM-1-ble_MBL operon. To confirm this, we simultaneously quantified the expression levels of ble_MBL and trpF in strain gNDM3. The result indicated that the transcription levels of ble_MBL and trpF were decreased by 890- and 610-fold, respectively, compared with the control or empty vector (Fig. 4), which was consistent with targeting bla_NDM-1. Hence, we identified that bla_NDM-1, ble_MBL, and trpF are located on the same operon, tentatively named bla_NDM-1-ble_MBL-trpF. These results indicated that our CRISPRi system can effectively silence an intact operon by targeting only its promoter region. However, some limitations also exist. For example, the polar effect on the operon may affect the expression of peripheral nontarget genes. In addition, when a target gene overlaps the promoter region of its downstream gene, it will increase the difficulty of sgRNA design, especially for a small target gene.

FIG 4.

Efficient inhibition of the entire bla_NDM-1-ble_MBL-trpF operon by CRISPRi through targeting of the bla_NDM-1 promoter region. The transcription of ble_MBL and trpF was suppressed by inhibiting bla_NDM-1 expression. A statistical analysis was performed using a two-way ANOVA with Sidak’s multicomparison test, using GraphPad. ns, not significant; ****, P < 0.0001.

CRISPRi-mediated silencing of a multicopy bla_KPC-2 gene in pathogenic E. coli.

CRISPRi inhibited the expression of green fluorescent protein (GFP) in a multicopy plasmid in Pseudomonas putida and E. coli (18, 27). However, inhibiting the expression of multicopy genes on natural plasmids in clinical strains has uncertainties due to the complicated genetic background. We thus investigated whether our CRISPRi system could control multicopy genes on natural plasmids in clinical E. coli strains. We selected both copies of the bla_KPC-2 (encoding carbapenemase) gene located on the MDR plasmid pE0171_KPC in clinical E. coli strain E0171 (Fig. 5A) (34). We designed an sgRNA targeting the nontemplate strand of the bla_KPC-2-coding region. qRT-PCR detection showed that the expression of bla_KPC-2 decreased significantly, by about 600-fold, compared to that in the control or empty vector (Fig. 5B). The meropenem MIC of the target strain decreased from 8 to 0.125 μg/mL (Fig. 5C). The corresponding dilution spotting plate assay also yielded similar results (Fig. S5). Furthermore, we found no significant changes in either the plasmid or target gene copy numbers (Fig. S6), consistent with the above result in K. pneumoniae. These results suggest that CRISPRi can be used to repress multicopy resistance genes in clinical E. coli strains.

FIG 5.

CRISPRi inhibits the expression of MDR plasmid-borne multicopy bla_KPC-2. (A) bla_KPC-2 is distributed on the IS26 element of plasmid pE0171_KPC (GenBank accession number MK370988). (B) Relative expression of bla_KPC-2 in cultures expressing an sgRNA targeting bla_KPC-2. Escherichia coli E0171 was used as the control; the results are the mean of three independent biological replicates. (C) Meropenem MIC values of the gKPC knockdown strains. The numbers above the columns indicate the MIC. A statistical analysis was performed using a one-way ANOVA, followed by Dunnett’s multicomparison test, using GraphPad for the transcription data. ns, not significant; ***, P < 0.001.

In summary, our optimized CRISPRi tool efficiently silenced target genes on MDR plasmids in clinical K. pneumoniae and E. coli strains. It also inhibited multiple genes simultaneously. Our CRISPRi tool opens the door to understanding the genetic background of clinical MDR K. pneumoniae and may enable high-throughput screening for determinants of K. pneumoniae virulence and mucoid phenotype in the future. However, a limitation of the CRISPRi tool is the selection of antibiotic resistance markers when an MDR clinical strain occurs, which also is an issue for the gene editing field.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The strains used in this study are listed in Table S3 in the supplemental material. Klebsiella pneumoniae NK01067 and E. coli E0171 were isolated in a hospital. The bacterial strains were grown in lysogeny broth (LB) with shaking at 200 rpm or on LB agar plates at 37°C. Chloramphenicol (Cm) was added to the medium with a final concentration of 30 μg/mL for screening transformants. The concentration of anhydrotetracycline (aTc) used in culture for induction of dCas9 expression was 2 μM.

Construction of plasmid pdCas9gRNA.

Tables S3 and S4 list all of the plasmids and primers used in this study. To construct plasmid pdCas9gRNA, we used plasmids pdCas9-bacteria (Addgene number 44249) and pgRNA-bacteria (Addgene number 44251) as the construction framework (27). The base pairing region (BPR), dCas9 handle region (Handle), and transcription terminator (Termi) constitute the sgRNA unit. The unit, with a strong constitutive promoter (BBa-J23119) and high-copy-number replicon fragment (ColE1), was amplified with the primer pair sgRNA_F/R using plasmid pgRNA-bacterial as the template. The Streptococcus pyogenes-derived dcas9 gene, with its inducible tetR promoter, and a Cm resistance marker (Cm^R) were amplified with the primer pair dCas9_F/R, using pdCas9-bacterial as the template. Next, the two fragments were assembled using an In-Fusion cloning kit (TaKaRa Bio Inc., Shiga, Japan), resulting in the final plasmid pdCas9gRNA.

New sgRNA editing plasmids were then constructed using inverse PCR (iPCR). A 20-nucleotide BPR (N20) was introduced at the 5′ end of the forward primer (5′-N₂₀GTTTTAGAGCTAGAAATAGCAAG-3′). The reverse primer (5′-ACTAGTATTATACCTAGGACTGAGCTAGCTGTC-3′; J23P_R) was designed based on the J23119 promoter sequence. The iPCR was performed using pdCas9gRNA as the template, removing the background plasmid with DpnI endonuclease. After phosphorylation (T4 polynucleotide kinase [PNK]; NEB), the PCR fragment was self-ligated using T4 DNA ligase (NEB), and the ligation product was transformed into E. coli Top10. Transformants were screened on agar plates containing Cm. The correct clone was verified by colony PCR and sequencing with the primer pair yzgRNA_F/R. The recombinant plasmid was introduced into K. pneumoniae NK01067 or E. coli E0171 by electroporation. The correct clones were induced using aTc for phenotypic study. The protocol requires only PCR and blunt-end ligation. Following this procedure, we quickly generated the plasmids pdCas9gNDM (1 to 9), pdCas9gSHV12, and pdCas9gKPC.

To construct the bla_NDM-1 and bla_SHV-12 dual-target plasmid, the entire 7.2-kb plasmid backbone of pdCas9gNDM3 and the 200-bp fragment containing the J23119 promoter and the sgRNA unit of pdCas9gSHV12 were obtained by PCR amplification using the primer pairs gN3_F/R and gSHV200_F/R, respectively. The products were then gel extracted. The 7.2-kb and 200-bp fragments were ligated using Gibson assembly and seamless cloning. Finally, the 200-bp fragment of pdCas9gSHV12 was inserted downstream of the terminator of pdCas9gNDM3, generating a tandem sgRNA plasmid structure, pdCas9gNDM3+SHV12.

To construct ESBL bla_SHV-11 and bla_SHV-12 gene expression plasmids, we amplified the vector fragment with the primer pair pUC_F/R, using pUC57-Kan plasmid as the template. Next, the bla_SHV-11 and bla_SHV-12 gene fragments were amplified with the primers SHV11_F/R and SHV12_F/R, respectively, using the K. pneumoniae NK01067 genome as the template. Finally, plasmids pUC57-Kan-bla_SHV-11 and pUC57-Kan-bla_SHV-12 were produced by Gibson assembly.

sgRNA design.

Table S5 lists all sgRNAs used in this study. The sgRNA sequences were composed of 20 complementary base pairs (N20) and were designed based on an efficiency score using the sgRNA design website of the Zhang Feng lab (https://zlab.bio/guide-design-resources). To enable highly efficient transcriptional repression, we chose target sequences in the 200-bp region ranging from −100 to +100 bp relative to the translation start site of the coding region.

Antibiotic susceptibility assays.

The MICs of the antibiotics were determined using the broth microdilution method of the Clinical and Laboratory Standards Institute (35). Briefly, 200-μL aliquots of Mueller-Hinton medium with Ca²⁺ and Mg²⁺ containing serial dilutions of antibiotics were inoculated with 10⁵ CFU of the studied strain and incubated at 37°C for 16 h. An MIC corresponds to the lowest concentration of an antibiotic at which bacterial growth is fully inhibited. All assays were performed in triplicate.

Dilution spotting assay.

Overnight bacterial cultures were serially diluted 10-fold in sterile phosphate-buffered saline. From each diluted culture, 5 μL was spotted onto LB agar plates supplemented with 2 μM aTc and an appropriate concentration of meropenem. The plates were incubated at 37°C for 16 h.

qRT-PCR and qPCR.

Overnight cultures were subcultured in LB liquid medium containing 2 μM aTc inducer and incubated at 37°C to an optical density at 600 nm (OD₆₀₀) of 1.0. The cultures were harvested by centrifugation. Total RNA was extracted using a FastPure cell/tissue total RNA isolation kit v2 (Nanjing Vazyme Biotech Co., Ltd., Nanjing, China) according to the manufacturer’s instructions. Reverse transcription, using 1 μg of RNA as the template in a 20-μL reaction volume, was performed using HiScript II qRT SuperMix (Nanjing Vazyme Biotech Co., Ltd.) as per the supplied instructions. Transcripts were quantified using a real-time PCR instrument with ChamQ universal SYBR qPCR master mix (Nanjing Vazyme Biotech Co., Ltd.). Relative abundances were determined using the relative standard curve 2^-ΔΔCt method, with gyrB as the reference gene for normalization of the total RNA levels (10).

Using genomic DNA as the template, the changes in gene copy numbers on the large plasmid were calculated through qPCR using the 2^-ΔΔCt method to determine copy numbers against the reference gene gyrB.

Statistical analysis.

Statistical analyses were performed using GraphPad Prism v8. Values are given as the mean ± standard deviation of independent experiments with at least three replicates. A one- or two-way analysis of variance (ANOVA) with a posttest was used to determine significance to obtain all pairwise comparisons.

Data availability.

The genomic and plasmid sequences were deposited at GenBank. The data on the K. pneumoniae NK01067 genome and plasmid pNK01067-NDM-1 can be found under accession numbers CP097651 and CP097653, respectively.