A pentaplex real-time PCR assay for rapid identification of major beta-lactamase genes KPC, NDM, CTX, CMY, and OXA-48 directly from bacteria in blood

Abstract

Introduction

Antibiotic resistance, particularly in cases of sepsis, has emerged as a growing global public health concern and economic burden. Current methods of blood culture and antimicrobial susceptibility testing of agents involved in sepsis can take as long as 3–5 days. It is vital to rapidly identify which antimicrobials can be used to effectively treat sepsis cases on an individual basis. Here, we present a pentaplex, real-time PCR-based assay that can quickly identify the most common beta-lactamase genes ( Klebsiella pneumoniae carbapenemase (KPC); New Delhi metallo-beta-lactamase (NDM); cefotaximase-Munich (CTX-M); cephamycin AmpC beta-lactamases (CMY); and Oxacillinase-48 (OXA-48)) from pathogens derived directly from the blood of patients presenting with bacterial septicemia.

Aim

To develop an assay which can rapidly identify the most common beta-lactamase genes in Carbapenem-resistant Enterobacteriaceae bacteria (CREs) from the United States.

Hypothesis/Gap statement

Septicemia caused by carbapenem-resistant bacteria has a death rate of 40–60 %. Rapid diagnosis of antibiotic susceptibility directly from bacteria in blood by identification of beta-lactamase genes will greatly improve survival rates. In this work, we develop an assay capable of concurrently identifying the five most common beta-lactamase and carbapenemase genes.

Methodology

Primers and probes were created which can identify all subtypes of Klebsiella pneumoniae carbapenemase (KPC); New Delhi metallo-beta-lactamase (NDM); cefotaximase-Munich (CTX); cephamycin AmpC beta-lactamase (CMY); and oxacillinase-48 (OXA-48). The assay was validated using 13 isolates containing various PCR targets from the Centre for Disease Control Antimicrobial Resistance Isolate Bank Enterobacterales Carbapenemase Diversity Panel. Blood obtained from volunteers was spiked with CREs and bacteria were separated, lysed, and subjected to analysis via the pentaplex assay.

Results

This pentaplex assay successfully identified beta-lactamase genes derived from bacteria separated from blood at concentrations of 4–8 c.f.u. ml⁻¹.

Conclusion

This assay will improve patient outcomes by supplying physicians with critical drug resistance information within 2 h of septicemia onset, allowing them to prescribe effective antimicrobials corresponding to the resistance gene(s) present in the pathogen. In addition, information supplied by this assay will lessen the inappropriate use of broad-spectrum antimicrobials and prevent the evolution of further antibiotic resistance.

Introduction

Considered a pillar of modern medicine, antimicrobials have transformed the treatment of bacterial infections that were once fatal. However, as early as the 1940s, reports of microbial resistance to commonly used antibiotics surfaced, marking the rise of what is feared to become the post-antibiotic era. Today, 70 % of bacterial infections are resistant to at least one commonly used antibiotic [1–3].

Carbapenem-resistant Enterobacteriaceae (CRE) infections are notoriously challenging to treat, as these bacteria are generally resistant to most of the common antibiotics, including those of last resort such as the carbapenems and colistin. Septicemia from CREs has rapidly emerged and contributed to severe infections and deaths worldwide. In one study, survival rates were shown to decrease by 9 % every hour that effective antimicrobial treatment is delayed; in another study, even when onset of active antimicrobial therapy for bacteremia resulting from CRE infection occurred within 47 h, sepsis caused by CRE organisms resulted in a 49 % mortality rate within 30 days [4]. Another study concluded that patients with CRE-induced septicemia had a mortality rate of 63.8%, as compared to a non-CRE septicemia mortality rate of 33.4 % [5]. Encouragingly, it has been observed that the rapid identification of resistant organisms (ideally within 1–3 h of diagnosis of sepsis or septic shock) and instigation of effective antimicrobial treatment in sepsis patients can result in survival rates of 80 % [6, 7].

The current method of determining antimicrobial susceptibility for organisms causing sepsis can require up to 3–5 days, during which blood is drawn from the patient, the organism is cultured, and then is exposed to a panel of antimicrobials to determine the organism’s resistance profile [8]. This culture-based methodology is susceptible to contamination by common skin flora, cannot be used to identify uncultivatable organisms, and is limited by low bacterial counts in blood samples [9, 10]. While waiting for results of the susceptibility testing, practitioners generally prescribe broad-spectrum antibiotics and only deescalate to targeted narrow-spectrum antibiotics once speciation and antibiotic susceptibilities are known. This strategy of employing empiric antibiotics often results in use of overly broad antibiotic exposure, which in turn contributes to resistance in microbial populations and decreased likelihood of successful treatment in the future. In sum, the proper identification of beta-lactamase genes by rapid multiplex PCR will provide opportunity to reduce inappropriate usage of last-line antibiotics in sepsis treatment when the pathogen may be susceptible to other antimicrobials; it will also help improve survival rates by preventing prescription of drugs which cannot control the pathogen [6, 11–14].

Additional factors that contribute to the difficultly of timely appropriate treatment of sepsis caused by CREs are the low bacterial numbers present in blood samples, which are generally between 1–10 c.f.u. ml⁻¹, and interference from components of formed elements present in whole blood such as haemoglobin [10, 15]. Using PCR-based methods to rapidly identify bacteria from whole blood has historically been difficult due to sensitivity loss in multiplex assays and formed element interference [15, 16]. Although there are multiple assays commercially available meant to quickly diagnose bloodstream infections, there are limitations to the usefulness of these assays in a clinical setting.

For example, the Biofire BCID2 Panel requires a positive blood culture before the panel can be run; it also does not test for the presence of AmpC beta-lactamases [17]. The T2Resistance Panel RUO requires 5 h for sample identification, while the recommended time for sample identification is 3 h or less [6, 7, 18]. Other methods such as the EDTA-modified carbapenem inactivation method (eCIM) and the modified Hodge test require bacterial culture and may result in misclassifying strains that express both a carbapenemase and an AmpC or extended-spectrum beta-lactamase (ESBL) [19]. Commercially available multiplex immunochromatographic assays do not test for ESBL or AmpC beta-lactamases [20].

ESBL, AmpC beta-lactamases, and other carbapenem-hydrolysing enzymes such as oxacillinases are also not detected by most currently available assays, thereby overlooking common resistance genes involved in many antibiotic-resistant bloodstream infections [21]. Additionally, each carbapenemase, ESBL, AmpC, and oxacillinase enzyme provides resistance to specific antimicrobials, due to differences in the enzyme active sites, necessitating the need to know which beta-lactamase(s) are present, thus allowing for more targeted and effective treatments [22, 23]. It is also quite common for a single CRE to carry multiple drug resistance genes, which necessitates treatment by a combination of antimicrobials specific to the individual enzymes present in an organism [21, 24].

In the United States, most carbapenem-resistant blood infections are caused by organisms carrying either KPC or NDM; these organisms also commonly carry multiple ESBL, AmpC, or oxacillinase (OXA) genes [25]. The most common ESBL gene present in resistant pathogens in the United States is Cefotaximase-Munich (CTX), while the most common AmpC gene is CMY [26–29]. Oxacillinase-48 is a less common enzyme but is present in many antibiotic resistant isolates, and is capable of hydrolysing carbapenems and other commonly used antibiotics [21].

Here we present a real-time PCR-based assay that, when coupled with existing bacteria-blood separation technology, can rapidly identify genes present in a multi-drug resistant bacterial sample at physiologically significant levels (<10 c.f.u. ml⁻¹). The bacteria-blood separation technology removes interfering elements from the blood, which allows the pentaplex qPCR assay to detect low levels of bacterial DNA targets typically present in blood samples from patients with early bacteremias. Additionally, this assay returns results within the desired 1 to 3 h range for septicemia identification and is effective in a single-tube format without losing sensitivity due to multiplexing, a common problem for many multiplex PCR assays [16, 30]. Using the results from this assay in conjunction with existing methods for bacterial identification and antibiotic susceptibility will help inform increasingly effective treatment options for patients presenting with septicemia.

Methods

Bacterial strains and culture conditions

The isolates listed in Table 1 are from the Centre for Disease Control Antibiotic Resistance Isolate Bank Enterobacterales Carbapenemase Diversity Panel, and were used to develop and validate this assay.

Table 1.

Isolates used to validate assay and resistance genes present in each isolate

Isolate Number	Beta-lactamase gene*
	KPC	NDM	CTX	CMY	OXA-48
0112	+
0113	+
0114	+
0115	+
0116	+			+
0118		+		+
0119		+	+	+
0120	+
0127		+		+
0137		+	+	+
0149		+		+
0150		+		+
0160					+

*A (+) indicates the presence of a beta-lactamase gene in the given isolate.

Stock cultures of each isolate were grown on Luria broth (LB) agar plates containing 8 µg ml⁻¹ imipenem, then incubated at 37 °C for 24 h. Plates were stored at 4 °C.

DNA preparation

A single colony from a stock plate of each isolate was used to inoculate 10 ml LB containing 8 µg ml⁻¹ imipenem at 37 °C in a non-shaking incubator. Bacteria were allowed to grow for 12 h to an OD reading of 1, roughly 2×10⁹ c.f.u. ml⁻¹. Total genomic DNA was extracted from each isolate from 1 ml of culture suspension using the DNeasy UltraClean Microbial Kit (Qiagen). DNA concentration was measured using an ND-1000 spectrophotometer (Nanodrop Technologies). The DNA concentration before dilution averaged 4800 µg ml⁻¹.

Primer and probe design and testing

Quantitative PCR assays for each beta-lactamase gene using 5′-hydrolysis Taqman probes were designed using Clustal Omega [31] and software available from Integrated DNA Technologies. Fluorophores were chosen based on emission spectra of each fluorophore. Eighty sequences for each beta-lactamase gene subtype were obtained from NCBI and aligned using the Clustal Omega software. The blastn programme at the National Centre for Biotechnology Information (NCBI) and the Clustal Omega software confirmed that the PCR target regions were specific to each gene and covered the most common single-nucleotide polymorphism subtypes for each gene (i.e. the PCR target regions for KPC target KPC-2, KPC-3, etc.), but did not target other genes such as CTX. Blind tests were conducted using a random numbering system of the target DNA from each isolate listed in Table 1. These tests were used to test sensitivity and detect false positives, false negatives, and ensure accurate identification of each target gene. No false positives or false negatives were found, and each target was accurately identified each time. The assay indicated similar sensitivity, regardless of the isolate from which the DNA was obtained. Replicates for each gene were repeated 3–6 times. Difficulties commonly faced when designing multiplex PCR assays, such as differing primer annealing temperatures and inter-region differences in GC content were overcome by careful design of primers and analysis of the surrounding regions of the primers and probes [32]. The primer and probe sequences used are listed in Table 2. Primers and probes were purchased from Integrated DNA Technologies.

Table 2.

Pentaplex primer and probe sequences

Target	Primer/Probe	Sequence (5′→3′)	Product (bp)
KPC	KPC_F*	CCGTCTAGTTCTGCTGTCTTGTCTCT	109
	KPC_R	GCCAAAGTCCTGTTCGAGTTTAGCG
	KPC_Pro	FAM-GCTGGCTTTTCTGCCACCGCGCTGACCAA-BHQ1
NDM	NDM_F	GGTTTGATCGTCAGGGATGGCG	107
	NDM_R	GGCAGGTTGATCTCCTGCTTGAT
	NDM_Pro	Cy5-TGCTGGTGGTCGATACCGCCTGGACCGATGAC-IBRQ
CTX	CTX_F	GTGTGCCGCTGTATGCGC	127
	CTX_R	GCACGATAAAGTATTTGCGAATTATCTGCTGTG
	CTX_Pro	Cy3-GCCGAATTAGAGCGGCAGTCGGGAGGCAGA-IBRQ
CMY	CMY_F	AGCGACTTTACGCTAACTCCAGCA	91
	CMY_R	CGTCTGGTCATTGCCTCTTCGTAACTC
	CMY_Pro	JOE-TGGCGCGCTGGCGGTGAAACCC-BHQ1
OXA-48	OXA_F	GATATCGCCACTTGGAATCGCGATC	134
	OXA_R	CCATAATCGAAAGCATGTAGCATCTTGCTC
	OXA_Pro	TEX615-TTGCCCGCCAAATTGGCGAGGCACGT-BHQ2

*_F, forward primer; _R, reverse primer; _Pro, probe; BHQ1, Black Hole Quencher 1; BHQ2, Black Hole Quencher 2; IBRQ, Iowa Black R Quencher.

PCR cycling conditions

Quantitative PCR assays were performed on a QuantStudio five using Integrated DNA Technology’s PrimeTime Gene Expression Master Mix. Individual assay conditions were 500 nM forward primer, 500 nM reverse primer, 250 nM probe, 50 ng target DNA, and HPLC-grade H₂O for a final assay volume of 20 µl. Pentaplex reaction conditions were the same for each target gene, except no water was added. The final sample volume for pentaplex reactions was 25.2 µl. Thermal cycling conditions were 5 min at 95 °C, followed by 40 cycles of 5 s at 95 °C and 30 s at 70 °C and a post-read cycle of 30 s at 60 °C. A positive signal was determined by QuantStudio five threshold software before cycle 22; cycle 22 was chosen because it was the latest cycle that the positive control samples amplified the target gene. The baseline and threshold settings used were the standard default values on the QuantStudio five platform.

Separating bacteria from blood and bacterial DNA isolation

Whole human blood was obtained from volunteers (BYU IRB No.: X18-340) by venipuncture into EDTA tubes and spiked with a dilution of the bacterial cultures prepared as described above. The final concentration of bacteria in the spiked blood samples was about 10 c.f.u. ml⁻¹. Bacteria were separated from human blood using the spinning disc method previously described by Buchanan et al. This spinning disc method utilizes novel separation technology that exploits the subtle size and density differences between erythrocytes and bacterial cells, which conventional centrifugation does not [33].

Following plasma collection, samples were added to microcentrifuge tubes and lysozyme (0.5 mg ml⁻¹, Sigma-Aldrich, chicken egg white #L6876) was added to each tube, vortexed for 5 s, then incubated for 10 min at room temperature. One hundred microlitres each of 6 M guanidine HCl (Promega, #H5381) solution and 1 % sodium dodecyl sulphate (SDS, USB #18220) solution was then added to each microcentrifuge tube and vortexed for 30 s. Following lysis, 200 µl of isopropanol and 25 µl of Spherotech magnetic beads (#SIM-05–10 h) were added to each microcentrifuge tube, which were vortexed for 30 s and placed on a magnet holder. Beads were allowed to aggregate on the side of the tube wall adjacent to the magnet for two min, at which time the liquid was removed. Then 450 µl of Wash 1 solution (6 M guanidine isothiocyanate, 20 mM TRis-HCl, 40 % v/v isopropanol, balance water) was added to each microcentrifuge tube and the tubes were vortexed for 30 s before being placed back in the magnet holder. Beads were aggregated and after 2 min, Wash 1 solution was removed and Wash 2 solution (0.1 M NaCl, 10 mM Tris-HCl, 70 % v/v ethanol, balance water) was added. The tubes were vortexed again for 30 s, placed back in the magnet holder, and beads allowed to aggregate for 2 min. Wash 2 solution was removed and the tubes were air-dried for 5 min. Fifteen microlitres of EDTA elution buffer solution (1 mM EDTA in water) was added and tubes were vortexed for 2 min to allow the DNA to detach from the beads for final resuspension in the elution buffer solution. The microcentrifuge tubes were placed back in the magnet holder for 1 min, after which the solutions were placed in clean microcentrifuge tubes. DNA was then serially diluted for further analysis via qPCR.

The bacterial separation and DNA extraction take approximately 25 min, while the time to prepare and run the qPCR is about an hour. Combined, this process requires almost 1.5 h from time of blood sample collection to reading the assay results.

Results and discussion

Development of a pentaplex assay detecting major beta-lactamase genes

The goal of this project was to identify the major beta-lactamase genes present in bacteria isolated directly from blood. This quantitative PCR assay was able to concurrently identify the five most common beta-lactamase genes (KPC, NDM, CTX, CMY, and OXA-48) in a single-tube reaction format with no loss of sensitivity due to multiplexing. The results below indicate this assay is capable of rapid, precise identification of the major beta-lactamase genes present in bacteria taken directly from blood, requiring no blood culturing. These results can be available in under 2 h from blood having bacterial concentrations down to 4 c.f.u. ml⁻¹.

Singleplex detection sensitivity

Serial tenfold dilutions of whole genomic DNA from clinical CRE isolates were used in singleplex reactions to determine the detection limit for each gene. See Table 3 for the r² values and sensitivities for each gene. The strains used for Figs 1–3 were (a) 0112; (b), (c), (d) 0119; (e) 0160, although all the strains mentioned in the Methods section were tested during the validation of this assay. The assay detected similar sensitivities with each isolate.

Table 3.

r² values and sensitivities of each gene from culture in singleplex and pentaplex, or pentaplex from blood

	Beta-lactamase gene
	KPC	NDM	CTX	CMY	OXA-48
Source
Culture, singleplex	r2=0.999	r2=1.000	r2=0.994	r2=1.000	r2=0.999
Culture, pentaplex	r2=0.999	r2=0.998	r2=0.999	r2=0.998	r2=0.999
Blood, pentaplex; all target genes present	r2=0.994	r2=0.993	r2=1.000	r2=1.000	r2=0.993
Sensitivity	4 c.f.u. ml−1	8 c.f.u. ml−1	8 c.f.u. ml−1	8 c.f.u. ml−1	4 c.f.u. ml−1

Fig. 1.

Sensitivities of individual target assays. Serial tenfold dilutions of CRE DNA containing the desired target genes were tested for the presence of: (a) KPC, from strain 0112; (b) NDM, strain 0119; (c) CTX, strain 0119; (d) CMY, strain 0119; and (e) OXA-48, strain 0160. Individual runs with different DNA concentrations provided Ct values which were plotted against the log of the concentration of DNA to obtain the standard curve. DNA quantity was adjusted to 481 µg ml⁻¹ for each starting concentration.

Fig. 2.

Sensitivities of the pentaplex assay from bacteria in culture. Serial tenfold dilutions of CRE DNA containing the desired target gene(s) were tested in a single tube format as a pentaplex reaction. Target genes were: (a) KPC, from strain 0112; (b) NDM, strain 0119; (c) CTX, strain 0119; (d) CMY, strain 0119; and (e) OXA-48, strain 0160. Individual runs with different DNA concentrations provided Ct values which were plotted against the log of the concentration of DNA to obtain the standard curve. DNA quantity was adjusted to 481 µg ml⁻¹ for each starting concentration.

Fig. 3.

Sensitivities of pentaplex assay in bacteria derived from blood. Serial tenfold dilutions of CRE DNA containing the desired target gene(s) were tested in a single tube format as a pentaplex reaction. Targeted genes were: (a) KPC, from strain 0112; (b) NDM, strain 0119; (c) CTX, strain 0119; (d) CMY, strain 0119; and (e) OXA-48, strain 0160. Individual runs with different DNA concentrations provided Ct values which were plotted against the log of the concentration of DNA to obtain the standard curve. DNA quantity was adjusted to 481 µg ml⁻¹ for each starting concentration.

Pentaplex detection sensitivity in bacteria derived from pure culture and from blood

Serial tenfold dilutions of whole genomic DNA from clinical CRE isolates were used for optimization of primer and probe sets. Many iterations were performed to arrive at a successful combination of the five individual assays into a single tube, pentaplex reaction. Fig. 2 shows the sensitivity of the pentaplex reaction using DNA from pure cultures. Fig. 3 shows the performance of the pentaplex reaction using DNA extracted from bacteria in blood. The sensitivities of the individual assays did not decrease when combined into the final single reaction, whether the DNA was from pure cultures (Fig. 2), or from bacteria present in blood (Fig. 3).

To prove that the concurrent presence of all target genes in a pentaplex reaction did not decrease the sensitivities of any of the individual assays, all five genes were combined into a pentaplex reaction in a single tube. Gene targets were present in three different bacteria which were spiked into human blood. Following separation, extraction, and serial tenfold dilution, each target gene amplified in the pentaplex reaction at the same rate as when run in a single reaction. The assay’s successful identification of all five targets concurrently also confirms no fluorophore interference (Fig. 4). See Table 3 for r² values and sensitivities of each gene in pentaplex with each gene target present.

Fig. 4.

Amplification plot showing simultaneous amplification of all five beta-lactamase gene targets in the pentaplex format. Red represents KPC (strain 0112); green represents CMY (strain 0119); blue represents NDM (strain 0119); yellow represents CTX (strain0119); grey represents OXA-48 (strain 0160).

Conclusion

Currently, there are multiple assays commercially available to test antibiotic susceptibility. However, none can interrogate organisms directly from blood and simultaneously detect the most common ESBL and carbapenemase genes. For example, available tests either require isolation of an organism from a blood culture for testing, do not specify which resistance genes are present, are lacking the sensitivity of this assay, or do not have the range of testing for common beta-lactamase genes [13, 34–37].

We have successfully developed an assay that can be completed in under 2 h using DNA from bacteria taken directly from blood, that identifies the most common beta-lactamase genes present in bacteria that cause bloodstream infections in the United States. Detection limits are as low as four genome copies, i.e. 4 c.f.u. ml⁻¹, per blood sample. Following established clinical validations, it is hoped that the information provided by this assay can improve patient care and outcomes by allowing for early CRE detection and characterization. In combination with other technologies such as tests for vancomycin resistance, early CRE information provided by this assay will promote the utilization of antibiotics which can successfully treat multi-drug resistant infections, improving the current bleak survival rates of bacteremias caused by these organisms.