CRISPR/Cas9-Mediated Re-Sensitization of Antibiotic-Resistant Escherichia coli Harboring Extended-Spectrum β-Lactamases

Abstract

Recently, the clustered regularly interspaced short palindromic repeats (CRISPR)-associated protein 9 (CRISPR/Cas9) system, a genome editing technology, was shown to be versatile in treating several antibiotic-resistant bacteria. In the present study, we applied the CRISPR/Cas9 technology to kill extended-spectrum beta-lactamase (ESBL)-producing Escherichia coli. ESBL bacteria are mostly multidrug resistant (MDR), and have plasmid-mediated antibiotic resistance genes that can be easily transferred to other members of the bacterial community by horizontal gene transfer. To restore sensitivity to antibiotics in these bacteria, we searched for a CRISPR/Cas9 target sequence that was conserved among >1,000 ESBL mutants. There was only one target sequence for each TEM- and SHV-type ESBL, with each of these sequences found in ~200 ESBL strains of each type. Furthermore, we showed that these target sequences can be exploited to re-sensitize MDR cells in which resistance is mediated by genes that are not the target of the CRISPR/Cas9 system, but by genes that are present on the same plasmid as target genes. We believe our Re-Sensitization to Antibiotics from Resistance (ReSAFR) technology, which enhances the practical value of the CRISPR/Cas9 system, will be an effective method of treatment against plasmid-carrying MDR bacteria.

Introduction

Intensive use of antibiotics has dramatically accelerated the emergence of antibiotic-resistant bacteria. The increasing frequency of drug resistance in human pathogens is threatening, as it limits therapies used previously for treatable infections and increases the risk of fatal outcomes [13,22,23]. However, t he r ate of d evelopment o f new antibiotics is lower than the rate of emergence of drugresistant strains [20]. Only two systemic antibiotics, linezolid and daptomycin, have been approved by the Food and Drug Administration in the past 5 years [4].

Within 10 years of the discovery of penicillin by Alexander Fleming [7] and even before penicillin had entered clinical use, the first resistant bacteria against penicillin were detected [1]. In the early 1980s, when the third-generation antibiotics cephalosporins (extended-spectrum β-lactam antibiotics) were introduced into clinical practice, they were considered a breakthrough in the treatment of β-lactamase-mediated resistant bacteria (e.g., TEM-1, TEM-2, and SHV-1). However, soon after cephalosporins were introduced clinically in the mid-1980s, a new group of enzymes, extended-spectrum β-lactamases (ESBLs), were detected [12,17].

ESBLs are a rapidly emerging group of plasmid-mediated β-lactamases that can hydrolyze major therapeutic antimicrobial drugs. The ESBLs are categorized into several types, including SHV, TEM, CTX-M, OXA, PER, and others [15], as most ESBLs are derived from these original β-lactamase genes. Although their original genes (e.g., SHV-1, TEM-1, and TEM-2) are not ESBLs, because they cannot hydrolyze extended-spectrum β-lactam antibiotics, their derivatives are ESBLs. Most ESBLs have only a few amino acid changes from their original form, due to point mutations. Because ESBL-containing plasmids can be transferred horizontally to other cells, ESBLs are found frequently in clinical isolates and can easily spread antibiotic resistance to other organisms in hospitals. Furthermore, ESBL-containing plasmids frequently carry other types of resistance genes to other antimicrobial agents [14,16]. Strains containing multiple ESBL genes are hard to treat because they are multidrug resistant.

Recently, three studies reported the use of CRISPR/Cas9 (clustered regularly interspaced short palindromic repeats/CRISPR-associated protein 9) technology for the creation of sequence-specific antimicrobials that target only antibiotic-resistant bacteria [3,5,24]. Cas9, a double-stranded DNA (dsDNA) nuclease, can be programmed to cleave almost any desired DNA sequence [9,19]. Bikard et al. [3] and Citorik et al. [5] transformed Staphylococcus aureus and Escherichia coli, respectively, with plasmids that coded for Cas9 and guided RNAs to specifically cleave resistance genes. The results demonstrate that programmed Cas9 mediates the cytotoxicity of resistant cells through sequence-specific targeting. These results suggest that antibiotic-resistant bacteria might be induced to revert to antibiotic-sensitive cells by specifically cleaving resistance genes using the CRISPR/Cas9 system.

In practice, however, there are so many mutations in resistance genes that the target sequences of the CRISPR/Cas9 system in one mutant DNA might be different from those in another mutant. For example, the total number of ESBL genes is over 1,000, and SHV-, TEM-, and OXA-type ESBLs have over 200 mutants per type. These are detailed in the ESBL nomenclature website hosted by George Jacoby and Karen Bush (http://www.lahey.org/studies/). If one guide RNA is required for each ESBL mutant, more than 1,000 guide RNAs will be necessary, which is impossible to develop practically for therapeutic purposes. To overcome this problem, we aimed to find a representative sequence common to all mutants, especially in TEM- and SHV-type ESBLs, that could be targeted by the CRISPR/Cas9 system. After testing the capability of the system to clear the plasmid-carrying resistance genes, we showed that the sequence-targeting TEM mutants could also be utilized to disarm other resistance genes such as CTX. We name the strategy Re-Sensitization to Antibiotics from Resistance (ReSAFR).

 

Materials and Methods

Reagents and Bacterial Strains

Escherichia coli K12 BW25113 was purchased from Open Biosystems (Thermo Fisher Scientific, Waltham, MA, USA). Klebsiella pneumoniae was isolated from a patient (K01-Bact-08-03094; Samsung Medical Center, Korea). Antibiotics were added at the following concentrations: 100 μg/ml of ampicillin (Amp), 34 μg/ml of chloramphenicol (Clm), 12.5 μg/ml of tetracycline (Tet), and 1 μg/ml of ceftazidime (Cef).

Target Sequences Representing ESBLs and bla Gene

All nucleotide sequences of ESBL mutants were obtained from http://www.lahey.org/studies/ and the National Center for Biotechnology Information (NCBI). Sequences of each type of ESBL mutant were aligned using a multiple alignment tool (ClustalW2) to find conserved sequences. We searched for PAM sequences (NGG) on both the positive and negative strands, and only when there was no mutation within the 20 nucleotides upstream of the PAM sequence was it selected as the target for TEM- or SHV-type ESBLs. For the ReSAFR system to target the bla gene, 20 nucleotides followed by the PAM sequence were selected from the bla gene in pUC19.

Plasmids Construction

The plasmid pRESAFRESBL was constructed based on pCAS9 purchased from Addgene (Plasmid No. 42876). crRNAs targeting TEM- and SHV-type ESBLs were cloned into pCAS9 following the protocol that uses BsaI [7]. To test the ReSAFR system targeting the bla gene, three plasmids were constructed: pBADCAS9 as a negative control vector, which expresses only Cas9 protein; psgRNAbla, from which sgRNA is transcribed; and pRESAFRbla, which expresses both Cas9 protein and sgRNA. Plasmids pUC19, pET21b, and pBR322 were employed as model plasmids that conferred antibiotic resistance when pRESAFRbla was tested.

Re-Sensitization of E. coli ESBLs by ReSAFR

To apply ReSAFR to ESBL-producing bacteria, an E. coli strain that is resistant to both Amp and Cef was manufactured by conjugation. We used K. pneumoniae K01-Bact-08-03094 isolated from a patient as a plasmid donor and E. coli BW25113 as a plasmid recipient. Each of the bacteria was inoculated and pre-cultured in Luria-Bertani (LB) medium at 37℃ with shaking at 250 rpm for 16 h. LB medium was inoculated with 1% (w/v) pre-cultured cells and the cells were cultured until the OD600 value was 1. Donor cell cultures and recipient cell cultures were mixed at a ratio of 1:10 or 1:100 into 3 ml of fresh LB medium and incubated for 3 h. The cells were plated onto MacConkey medium containing Amp, and E. coli containing ESBLs were selected. pRESAFRESBL and pCAS9 were transformed into E. coli carrying the ESBL plasmid. Transformants were serially diluted and plated onto LB agar plates containing Clm alone or both Clm and Amp. The re-sensitization ratio was calculated from colony-forming units (CFU) after overnight incubation at 37℃.

Disk Diffusion Assay

Colonies selected from LB agar plate with Clm were inoculated into fresh LB medium and cultured at 37℃ with shaking at 250 rpm for 16 h. Bacterial cultures were mixed with 3 ml of 0.75% agar and poured above the LB agar plates. Paper disks placed on the plates were loaded with 34 μg Clm, 1 μg Cef, or 10 μg Amp, because these amounts of antibiotics were enough to form clear halos around the disks in the preliminary experiments. The plates were incubated overnight and the diameters of the inhibition zones were measured.

Measurement of the Fraction of Antibiotic-Resistant Cells Due to the bla Gene

E. coli containing pRESAFRbla and pUC19 or pBR322 was pre-cultured in LB broth supplemented with 2 g/l glucose, 100 μg/ml of Amp, and 34 μg/ml of Clm. The pre-cultured cells (0.1% (v/v)) were inoculated into 3 ml of fresh LB medium containing 0.1 mM arabinose, and were cultured at 37℃ with shaking at 250 rpm for 12 h. Then, the cells were washed twice with phosphate buffered saline and plated onto LB agar plates containing an antibiotic (Amp or Tet). Colony-forming units were only calculated from the plates containing 50–250 colonies. For time-dependent measurements, CFUs were determined at the indicated time period after inoculation of pre-cultured cells into fresh LB medium with arabinose.

Detection of Clearance of Resistance Gene-Carrying Plasmids

Cells re-sensitized by ReSAFR were collected, and the presence of the plasmid was confirmed by colony PCR. Cell cultures were pelleted by centrifugation, resuspended with Tris-EDTA buffer, and incubated at 95℃ for 10 min. Using Taq polymerase, colony PCR was performed under the following conditions: denaturing (95℃, 10 sec), annealing (55℃ to 60℃, 5 sec), and extension (72℃, 5 sec), with a total of 30 cycles. All primers that were employed to confirm plasmid stability are described in Table S1.

 

Results

Design of ReSAFR and Determination of the Target Sequence

Cas9 cleaves specific target sequences guided by RNA. Two necessary components are (i) short CRISPR-derived RNAs, crRNA, and tracrRNA, and (ii) the double-stranded DNA endonuclease, Cas9 protein [2,21]. The crRNA/tracrRNA/Cas9 complex is recruited to the target sequence by base-pairing between the crRNA sequence and the target DNA sequence (Fig. 1A). For successful binding of Cas9, the target sequence must contain the correct protospacer adjacent motif (PAM) sequence immediately after the 20 nucleotide recognition sequence. The PAM sequence for Streptococcus pyogenes Cas9 is NGG. Target sequences (20 nucleotides + PAM) can be on either strand of a gene. Generally, we find many such sequences in a gene that can be targeted by the CRISPR/Cas9 system. However, we did not know whether a gene with several hundred point mutations would have a common target sequence that could be used with hundreds of mutants.

Fig. 1.Schematic diagram of ReSAFR for ESBLs. (A) Schematic of ReSAFR and the plasmid map of pRESAFRESBL. When a cell is transformed with all ReSAFR components (Cas9 protein, tracrRNA, and crRNAs for ESBL genes), they combine and cleave the target ESBL gene. We used pRESAFRESBL to deliver the ReSAFR components. Specific bacteriophages will be a good delivery system of the ReSAFR components. crRNAs targeting TEM- and SHV-type ESBLs were transcribed as a single, long RNA. After forming a complex with tracrRNA and Cas9 protein, crRNA is spontaneously cleaved by host RNase III into each CRISPR module that works independently. (B) crRNA sequence for TEM-type ESBLs. (C) crRNA sequence for SHV-type ESBLs.

The ESBL mutation database was downloaded from http://www.lahey.org/studies/. Mutant nucleotide sequences of each type were obtained from NCBI. All obtained sequences were analyzed using the multiple sequence alignment program ClustalW2. Sequences that did not have any mutations within the 20 nucleotides upstream of the NGG PAM sequence were screened, and from >200 TEM mutants, only one target sequence on the negative strand was found to be conserved (Fig. 1B). SHV mutants commonly contained one target sequence on the positive strand (Fig. 1C). We could not find any common target sequence for OXA and CTX-M ESBLs. We constructed a plasmid, pRESAFRESBL, that produced Cas9 protein under the strong and constitutive synthetic promoter BBa_J23102, tracrRNA, and both crRNAs for the target sequences (Table 1).

Table 1.Component sequences targeting each antibiotic resistance gene.

Re-Sensitization of Antibiotic-Resistant Cells

Re-sensitization of antibiotic-resistant cells by pRESAFR was tested. E. coli BW25113 cells were transformed with the plasmid pESBL isolated from K. pneumoniae K01-Bact-08-03094 through conjugation. This antibiotic-resistant E. coli (designated as ESBL strain) was transformed with either pRESAFRESBL or pCas9 that expressed only Cas9. The ESBL strain containing pRESAFRESBL could be killed upon treatment with Amp, whereas the strain with pCas9 grew in the presence of 100 μg/ml of Amp (Fig. 2A). We found that more than 99% of resistant cells with pRESAFRESBL were killed, suggesting that the plasmid rendered the ESBL strain sensitive to Amp by cleaving the target sequence in pESBL.

Fig. 2.Re-sensitization to antibiotics of E. coli carrying ESBL plasmids by the ReSAFR system. (A) pRESAFRESBL was transformed into E. coli carrying ESBL plasmids and the antibiotic-resistant cell fraction was measured by counting the colony-forming units (CFUs). The antibiotic-resistant cell fraction was obtained by comparing the CFU after antibiotic treatment with the CFU before treatment. Thus, antibiotic-resistant cells showed a value of more than 1, because they grew even in the presence of antibiotics. The cells transformed with pCas9 grew in the presence of Amp, whereas only 0.3% cells survived when the cell was transformed with pRESAFRESBL. (B) Disk diffusion assay of antibiotic resistance of the ESBL strain. E. coli BW25113 became resistant to Cef and Amp by acquiring pESBL from K. pneumoniae through conjugation. This ESBL E. coli could be made sensitive to Amp and Cef by transforming the ESBL E. coli with pRESAFRESBL, but not through the transformation of pCas9. (C) Diameter of the inhibition zone. (D) pESBL was cleared from the cell during re-sensitization. Primers and PCR conditions for plasmid corfirmation are shown in Table S1.

Next, we tested the ESBL strain against several antibiotics. The ESBL strain was resistant to both Amp and Cef, but was sensitive to Clm (Fig. 2B). The disk diffusion assay confirmed that the ESBL strain had become sensitive to Amp specifically in the presence of pRESAFRESBL. Surprisingly, we found that the ESBL strain also became sensitive to Cef with pRESAFRESBL (Figs. 2B and 2C). Resistance to Cef is mediated by CTX-M β-lactamases that do not contain the same target sequence as pRESAFRESBL. This result suggests that pRESAFRESBL may have disarmed other resistance genes that made the strain multidrug resistant. This may have occurred because multiple ESBL genes are frequently carried together in a single plasmid [10,18] and the double-stranded break in the target sequence of TEM and SHV induced clearance of the whole plasmid (Fig. 2D).

CRISPR/Cas9-Mediated Double-Strand Break Induces Clearance of Whole Plasmid

We thought that TEM-targeting pRESAFRESBL induced re-sensitization to Cef, mediated by CTX-M, through the clearance of the entire plasmid carrying both resistance genes. To test the clearance of the plasmid through the CRISPR/Cas9-mediated double-strand break, we used commercially available plasmids, pUC19 and pET21, as the substrates of the CRISPR/Cas9 system because they contain multiple well-known resistance genes. In addition, we used single guide RNA (sgRNA), which combines the targeting specificity of the crRNA with the scaffolding properties of the tracrRNA into a single transcript (Figs. 3A and 3B). We screened several PAM sequences in the bla gene and selected a 25 nt sequence upstream of PAM that does not have a homologous sequence within the E. coli genome (Fig. 3B). sgRNAbla was produced by combining tracrRNA and crRNA using overlap extension PCR [6,9]. The sequence coding for sgRNAbla was under the control of a synthetic promoter that constitutively transcribes sgRNA (Fig. 3B). The Cas9 gene was under the control of an arabinose-inducible promoter with a strong synthetic ribosomal binding site. Three plasmids, pBADCAS9, psgRNAbla, and pRESAFRbla, were constructed based on pBAD33 (Fig. 3C).

Fig. 3.Design and construction of the ReSAFR plasmid targeting the bla gene. A ReSAFR system against the bla gene was constructed to confirm plasmid clearance mediated by this system. (A, B) The sgRNA, a combined form of crRNA and tracrRNA, was designed and constructed by overlapping PCRs. The crRNA sequence for the bla gene was confirmed so that there was no homologous region within the E. coli genome. (C) Plasmids constructed for Cas9 protein expression, sgRNA for bla, and both of these.

First, pUC19-mediated resistance against Amp was tested. The resistant E. coli containing pUC19 was co-transformed with one of the three plasmids, pBADCAS9, psgRNAbla, or pRESAFRbla, incubated with arabinose, diluted, and spotted onto LB agar containing Amp. The pUC19-mediated resistance was abolished by pRESAFRbla, whereas pBADCAS9 did not make the cell sensitive to Amp (Fig. 4A). More than 99.9% of resistant cells became sensitive to Amp in the presence of pRESAFRbla (Fig. 4B). The re-sensitization showed a several hour lag period, perhaps because sufficient amounts of the CRISPR/Cas9 system to outcompete hundred copies of plasmid had accumulated inside cells only after a certain period of time (Fig. 4C).

Fig. 4.Plasmid clearance mediated by ReSAFR. (A, B) Replica and colony-forming unit (CFU) data for re-sensitization of E. coli containing pUC19. These data showed that the antibiotic-resistant cell fraction was decreased about ~1,000 times by pRESAFRbla. (C) Time-dependent analysis of the re-sensitization of E. coli containing pBR322. The CFU was measured at 0, 1, 2, 3, 6, and 12 h after induction. After a 3-h induction, the resistant cell fraction decreased exponentially. (D) Cleavage of the target sequence in the bla gene made the cell sensitive not only to Amp but also to Tet. (E) Colony PCR to confirm plasmid clearance. The pBR322 plasmid was not detected in the re-sensitized cells. Note that the weak band intensity of pMB1 origin of the second lane does not stand for plasmid loss but merely the result of picked colony size.

The plasmid pBR322 contains tet (tetracycline resistance gene) as well as the bla gene. As we observed from pESBL-derived multidrug resistance, pBR322-mediated resistance against both Amp and Tet was abrogated by pRESAFRbla, which causes a double-stranded break specifically within the bla gene (Fig. 4D). This indicates that CRISPR/Cas9-mediated cutting of a plasmid does not allow rejoining of genetic material. Clearance of pBR322 by pRESAFRbla was confirmed through PCR amplification of the origins of replication from colonies. P15A ori of pRESAFRbla was amplified (using the primers shown in Table S1) in both Amp-resistant cells and Amp-sensitive cells (Fig. 4E). Re-sensitized cells that did not grow on LB Amp plates but formed colonies on LB plates did not contain the pMB1 ori of pBR322, indicating clearance of the plasmid. The pMB1 ori could be amplified only from the colonies formed on the LB plates containing Amp. This result showed that CRISPR/Cas9-mediated cleavage of the plasmid does not merely inactivate the target sequence, but leads to complete elimination of the plasmids in the cell.

 

Discussion

Bacteria that produce ESBLs are resistant to many penicillin and cephalosporin antibiotics, and often to other types of antibiotics. E. coli and Klebsiella species are two main bacteria that produce ESBLs. A recent report of the Centers for Disease Control and Prevention in the USA indicates that nearly 26,000 healthcare-associated Enterobacteriaceae infections are caused by ESBL-producing Enterobacteriaceae each year. Patients with bloodstream infections caused by ESBL-producing Enterobacteriaceae are about 57% more likely to die than those with bloodstream infections caused by a non ESBL-producing strain.

The CRISPR/Cas9 system provides new opportunities to eradicate ESBL strains, as this RNA-guided DNA nuclease can specifically cleave bacterial genes, leading to re-sensitization of the antibiotic-resistant cell. However, there are so many mutations in the sequences of ESBL genes that one target sequence for a single mutant will be of limited clinical value. For example, there are over 200 mutants in the TEM-type ESBLs alone. Thus, we searched for a conserved sequence among mutants that can be targeted by the CRISPR/Cas9 system. Although CTX-M mutants did not have such a sequence, we found that TEM- and SHV-type ESBLs had a conserved sequence without mutations that could be targeted by CRISPR/Cas9. The CRISPR/Cas9 system that targeted this sequence successfully cleaved the ESBL plasmid of a clinical isolate.

The pRESAFRESBL not only re-sensitized the bacterial cells with the targeted resistance gene but also disarmed resistance against other antibiotics. In the presence of pRESAFRESBL, resistance against Cef was also eliminated, due to clearance of pESBL that carries both TEM and CTX-M. CTX-M enzyme is the ESBL that E. coli most often produces. Thus, we expect that the target sequence found in this study will provide opportunities to fight against various plasmidmediated multidrug-resistant bacteria, as resistance genes are frequently transferred together in a plasmid [11].