Figures
Abstract
Molecular methods that enable the detection of antimicrobial resistance determinants are critical surveillance tools that are necessary to aid in curbing the spread of antibiotic resistance. In this study, we describe the use of the Antimicrobial Resistance Determinant Microarray (ARDM) that targets 239 unique genes that confer resistance to 12 classes of antimicrobial compounds, quaternary amines and streptothricin for the determination of multidrug resistance (MDR) gene profiles. Fourteen reference MDR strains, which either were genome, sequenced or possessed well characterized drug resistance profiles were used to optimize detection algorithms and threshold criteria to ensure the microarray's effectiveness for unbiased characterization of antimicrobial resistance determinants in MDR strains. The subsequent testing of Acinetobacter baumannii, Escherichia coli and Klebsiella pneumoniae hospital isolates revealed the presence of several antibiotic resistance genes [e.g. belonging to TEM, SHV, OXA and CTX-M classes (and OXA and CTX-M subfamilies) of β-lactamases] and their assemblages which were confirmed by PCR and DNA sequence analysis. When combined with results from the reference strains, ∼25% of the ARDM content was confirmed as effective for representing allelic content from both Gram-positive and –negative species. Taken together, the ARDM identified MDR assemblages containing six to 18 unique resistance genes in each strain tested, demonstrating its utility as a powerful tool for molecular epidemiological investigations of antimicrobial resistance in clinically relevant bacterial pathogens.
Citation: Leski TA, Vora GJ, Barrows BR, Pimentel G, House BL, Nicklasson M, et al. (2013) Molecular Characterization of Multidrug Resistant Hospital Isolates Using the Antimicrobial Resistance Determinant Microarray. PLoS ONE 8(7): e69507. https://doi.org/10.1371/journal.pone.0069507
Editor: Shamala Devi Sekaran, University of Malaya, Malaysia
Received: April 14, 2013; Accepted: June 4, 2013; Published: July 25, 2013
This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Funding: This work was supported by a Defense Medical Research and Development Program Intramural Applied Research and Advanced Technology Development Award. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
The evolution, increasing prevalence and dissemination of pathogenic bacteria resistant to multiple antimicrobial agents is currently recognized as one of the most important problems in global public health [1]. The rapid spread of antibiotic resistance genes, facilitated by mobile genetic elements such as plasmids and transposons, has led to the emergence of multidrug resistant (MDR) strains of many clinically important species that now frequently leave clinicians out of therapeutic options [2], [3].
Traditional phenotypic methods currently used to determine antimicrobial resistance profiles (e.g., disk diffusion, broth microdilution) remain critical in guiding appropriate treatment options. However, techniques such as these are unable to determine the actual molecular mechanisms of resistance, and are especially lacking in situations where the observed phenotype is a result of the interaction of multiple gene products with overlapping activities [4]. Molecular techniques, such as PCR and DNA sequencing, have recently been employed to mitigate some of these deficiencies by identifying genes and genetic assemblages responsible for antibiotic resistance and MDR, monitoring the spread of resistance determinants, and elucidating the genetic elements responsible for the dissemination of resistance determinants.
The use of DNA microarrays is another promising technology for the identification of antimicrobial resistance determinants in any number of species. Due to their inability to determine whether resistance determinants are expressed or gene products are functional, DNA microarrays are not intended to replace standard phenotypic testing. Rather, microarrays provide a powerful platform for molecular epidemiology and broad-based screening. This contention has been supported by a number of recent reports that have successfully applied a variety of microarray platforms for both limited [5], [6], [7], [8] and broad spectrum [9], [10], [11] detection of antibiotic resistance genes. Furthermore, as they allow for the simultaneous detection of a large number of genes in a single assay, microarrays can be used to track determinants directed against multiple classes of antibiotics that are frequently found clustered in mobile genetic elements [12]. The rapid spread of such assemblages not only makes the successful treatment of infections increasingly difficult but also complicates epidemiological investigations aimed at elucidating the spread of antimicrobial resistance genes and the underlying genetic structures that produce MDR phenotypes.
In this study, we describe the Antimicrobial Resistance Determinant Microarray (ARDM), an electrochemically interrogated DNA microarray platform containing 2,241 oligonucleotide probes targeting 239 genes that confer resistance to 12 classes of antimicrobial compounds, quaternary ammonium compounds and streptothricin, and its accompanying sample processing and analysis methods that were developed for broad-range MDR gene detection. Using well-characterized and/or sequenced reference strains, we confirmed ∼25% of the ARDM content and successfully tested Acinetobacter baumannii, Escherichia coli and Klebsiella pneumoniae hospital isolates, demonstrating the applicability of this technology for the detection of antibiotic resistance gene assemblages in clinically relevant bacterial pathogens.
Materials and Methods
Bacterial strains
Genomic DNA preparations from reference strains of A. baumannii, E. coli, K. pneumoniae, Staphylococcus aureus, S. epidermidis and Enterococcus faecalis were purchased from American Type Culture Collection (ATCC), Manassas, VA (Table 1). Four A. baumannii clinical strains with published sequences [13] were obtained from the US Department of Defense Multidrug Resistance Surveillance Network (MRSN, Walter Reed Army Institute of Research, Bethesda, MD). Fifteen additional clinical strains with high levels of β-lactam resistance were isolated from blood or urinary tract infections from six hospitals located in Egypt from June 2001 to November 2007 by the Naval Medical Research Unit No. 3 (NAMRU-3), Cairo, Egypt (Table 2). Isolation of these clinical strains was approved by the Institutional Review Board (IRB) at NAMRU-3. As all identifiers were previously stripped from these strains, the described study was deemed exempt from consideration as Human Subjects research by IRBs at both NAMRU-3 and the Naval Research Laboratory (NRL).
Antibiotic susceptibility testing of clinical isolates from Egyptian hospitals
The sensitivity of clinical strains to 20 antibiotics and antibiotic combinations: ampicillin, ampicillin+sulbactam, ticarcillin+clavulanic acid, piperacillin, carbenicillin, cephalothin, cefotaxime, ceftazidime, cefopodoxime, ceftriaxone, cefepime, aztreonam, imipenem, gentamycin, amikacin, tetracycline, chloramphenicol, sulfamethoxazole+trimethoprim, ciprofloxacin and nalidixic acid was determined using disk diffusion assays according to CLSI guidelines [14]. Not all isolates were tested using all of these antibiotics (see Table S1). The extended spectrum β-lactamase (ESBL) phenotypes of all E. coli and K. pneumoniae isolates were determined using double disk diffusion and the metallo-β-lactamase phenotypes of A. baumannii were determined using E-test Metallo-β-Lactamase Strips which contained a combination of imipenem and imipenem/EDTA (BioMerieux, Marcy l'Etoile, France) according to the manufacturer's instructions.
ARDM
The ARDM version 1 is a custom designed microarray for use with the ElectraSense platform [15] (CustomArray, Bothell, WA). Electrochemical interrogation of biotinylated amplicon fragments is accomplished using a horseradish peroxidase-streptavidin conjugate with the redox mediator, 3,3′,5,5′-tetramethylbenzidine (TMB). Each 4×2K microarray contains four identical sub-arrays of 2,240 individually addressable microelectrodes functionalized with DNA probe sequences of 30 to 35 nucleotides in length. The probes (Table S2) were derived from 239 genes that confer resistance to quaternary ammonium compounds, streptothricin, and 12 different classes of antimicrobials: aminoglycosides (n = 29), ansamycins (n = 1), β-lactams (n = 48), chloramphenicol (n = 12), diaminopyrimidines (n = 29), glycopeptides (n = 14), lincosamides (n = 8), macrolides (n = 37), quinolones (n = 5), streptogramins (n = 17), sulfonamides (n = 3) and tetracyclines (n = 38) (Figure 1); some determinants (e.g., multidrug efflux pumps) confer resistance to multiple families of antimicrobials. Probe sequences were optimized for uniform melting temperatures and to avoid target cross-hybridization to the extent possible. Each of the 239 genes is represented on the ARDM by a range of 6 to 10 probes, with the most frequent representation comprising four or five pairs of duplicate probes per gene. Chip content includes alleles derived from a wide variety of bacteria: Firmicutes (including bacilli, clostridia, lactobacilli; 38% total chip content), Bacteroidetes (2% total content), Actinobacteria (9% total content), and α-, β-, ε-, and γ-Proteobacteria (Enterobacteriales, Aeromonadales, Pseudomonadales, Vibrionales, and Pasteurellales; 49% total content).
Two hundred thirty-nine different resistance determinants are represented by probes on each sub-array.
DNA purification and processing
Genomic DNA was extracted from isolate cultures grown overnight on MacConkey agar using the MagNA Pure LC system (Roche Diagnostics Corp., Indianapolis, IN) or MasterPure DNA and RNA Complete Purification Kit (Epicentre Biotechnologies, Madison, WI) and quantified using the Qubit fluorometer (Invitrogen/Life Technologies, Grand Island, NY) and BR dsDNA QuantIT DNA quantification reagent set (Invitrogen). Ten ng of each DNA sample was subjected to whole genome amplification using the GenomiPhi V2 reagent kit (GE Healthcare, Piscataway, NJ) according to manufacturer's instructions, with the reaction mixture incubated for 90 min at 30°C and subsequently for 10 min at 65°C to stop the reaction. Two µg of the resulting high molecular weight amplicons were fragmented for 1 min at 37°C using Fragmentation Reagent (DNase I, 0.045 units/µg DNA) and Fragmentation Buffer (GeneChip Resequencing Assay Kit, Affymetrix, Sanata Clara, CA) in a total reaction volume of 60 µl. The enzyme was subsequently inactivated by incubation at 95°C for 10 min. The amplified and fragmented DNA was then purified using DNA Clean & Concentrator −5 kit (Zymo Research, Irvine, CA) and labeled using the ULS Platinum Bright Biotin nucleic acid labeling kit (Kreatech Diagnostics, Durham, NC) according to the manufacturer's instructions (total reaction volume of 10 µl). Following purification of the labeled fragments using the kit's Kreapure columns, the samples were hybridized to the ARDM.
ARDM hybridization and data analysis
Hybridization of the biotinylated amplicon fragments to the ARDMs was performed according to the manufacturer's instructions, except that pre-hybridization and hybridization temperatures were increased to 60°C (from 50°C recommended by the manufacturer). Four samples were analyzed on each 4×2K microarray slide (one sample for each sub-array). Following overnight hybridization, each array was processed using CustomArray's ElectraSense Detection Kit and interrogated on the ElectraSense reader while substituting a streptavidin-peroxidase conjugate (S104PHRP, Fitzgerald Industries, North Acton, MA) for the Biotin Labeling Solution provided, as this highly polymerized conjugate was previously shown to improve detection sensitivity in electrochemically interrogated assays [16]. All data obtained from the ElectraSense reader were analyzed with custom developed Perl scripts using the following criteria for positive/negative signal determinations: individual probes were designated positive when their signal exceeded a threshold defined as the average signal of all microarray probes (excluding the top 5%) plus three standard deviations (SD). Each gene was considered present if 30 or 40% of its probes provided positive signal determinations (depending on the thresholding algorithm used).
PCR
Clinical isolates from Egyptian hospitals were tested for the presence of alleles from the blaSHV, blaTEM, blaCTX-M and blaOXA families of ESBLs using a multiplex PCR assay [17]. As this assay was not designed to detect blaOXA-9, an additional PCR was required to detect this gene using the following amplification conditions: 95°C for 2 min, followed by 30 cycles of 95°C for 30 s, 53°C for 30 s, and 72 for 1 min [18]. See Methods S1 for information on methods used for PCR confirmation of tetracycline resistance genes. PCR amplicons were size-confirmed by electrophoresis using 1.2% FlashGel DNA cassettes (Lonza, Walkersville, MD). The identities of the detected amplicons were further verified by DNA sequencing (Eurofins MWG Operon, Huntsville, AL).
Results and Discussion
The overall efficacy of the ARDM v.1 was first determined using a series of reference strains. Once an appropriate thresholding algorithm was determined for the desired levels of sensitivity and specificity, the 15 clinical isolates obtained from six hospitals in Egypt were subjected to ARDM analysis. As these isolates were originally selected based on their high levels of resistance to β-lactams and/or expression of ESBL phenotypes, results for hybridization to β-lactam/ESBL-specific probes on the ARDM were subsequently confirmed by PCR and DNA sequencing. ARDM detection of tetracycline resistance genes was also confirmed using the same molecular techniques.
Determination of optimal algorithm(s) for positive determination and confirmation of probed content
Based on results for each of the 2,241 target-specific probes, background signals from microelectrodes with non-hybridized probes could be very clearly distinguished from signals resulting from those with positive probes (Figure 2 and Figure S1). For each microarray, a “probe threshold” was calculated as the mean signal of the 2,138 probes with the lowest signals (95% total probe content on each array) plus three standard deviations. Thus, the probe threshold calculation accounts for differences in overall microarray sensitivity, as well as the variability between negative probes. Once the positive probes were determined for each array, each allele (represented by 6–10 probes) was subjected to another thresholding algorithm, which deemed the allele positive or negative based on the percentage of its probes that were above the “probe threshold.” Fourteen reference strains (Table 1) were used to confirm the efficacy of ARDM probe content and to optimize the threshold algorithm for determination of positive alleles. This panel included MDR strains representing both Gram-negative and Gram-positive bacteria and containing a variety of antibiotic resistance markers. It also included some wild type strains with few or no antibiotic resistance determinants.
Each data point represents one probe; data are sorted according to alleles represented by each probe. In each panel, the horizontal dotted line indicates the “probe threshold” used to determine whether probes were considered positive (mean of lowest 95% probes+3 SD). For panel A, peaks correspond to the following alleles: a: tet(G) [9/9 probes positive]; b: cat4 [5/7 probes positive]; c: dfrA1 [9/9 probes positive]; d: catA1 [6/8 probes positive]; e: arr-3 [8/8 probes positive]; f: aph6/str(B) [8/10 probes positive], aph3″/str(A) [10/10 probes positive], aadA1b [10/10 probes positive], aadB [10/10 probes positive]; g: qacEΔ1 [10/10 probes positive]; h: aadA2 [2/9 probes positive - allele deemed negative], aadA1 [7/7 probes positive]; i: blaVEB [9/9 probes positive]; j: blaOXA-10 [7/7 probes positive]; k: dfrA10 [8/10 probes positive]; l: blaPSE-1/CARB [2/10 probes positive – allele deemed negative]; m: aacC1 [9/9 probes positive], aphA1 [10/10 probes positive]; n: ant(2′)-Ia [10/10 probes positive]; o: sulI [10/10 probes positive]; p: tet(A) [9/9 probes positive]. For panel B, the sulII allele (peak indicated) was deemed positive with 7/7 probes positive; other alleles had one or two positive probes, but were deemed negative.
Initial algorithm testing using commercially prepared DNA.
DNA preparations from six strains with sequenced, annotated genomes were initially used to narrow the threshold conditions for positive/negative allele determinations for the desired range of array sensitivity and specificity. Comparing the positive alleles determined under each threshold (10–100% probes positive for each allele) with their presence within the published gene sequences, sensitivity and specificity were calculated and a receiver operating characteristic (ROC) curve was generated (Fig. 3). The desired sensitivity (>80%) and specificity (>99%) were obtained when positive determinations were based on 30–70% of the allele's representative probes meeting “probe threshold” requirements. Under these conditions, no false-positive alleles were detected, and between nine and 13 false-negative alleles were encountered, depending on the stringency of the algorithm.
Specificity and sensitivity data for commercial DNA preparations of sequenced strains (six strains; phenol-chloroform-extracted) are indicated in black. Analogous data for DNA preparations extracted in our laboratories are presented in gray (four strains).
Several non-sequenced reference strains with confirmed allelic content (E. coli 25922, K. pneumoniae strains 70603, BAA-1705, BAA-2146) were similarly analyzed using the same set of algorithms. In E. coli 25922, an antimicrobial-sensitive strain, only the intrinsic ABC transporter genes, macA and macB, were detected under any but the least stringent threshold algorithms. In the blaSHV-18-producing strain, K. pneumoniae 700603, 100% of the probes representing blaSHV-1 and blaSHV-5 on ARDM were considered positive, confirming the high identity of these sequences and the blaSHV-18 gene (99% identical). Similarly, a number of alleles from the blaNDM-1-containing reference strain, K. pneumoniae BAA-2146, were also strongly positive, with 100% of their probes deemed above “probe threshold” values (blaFEC, blaCTX-M-3, blaSHV-1/PIT-1, blaLAT-1/CMY, aac(6′)-ib, aph3″/str(A), aph6/str(B), qacEΔ1, sulI, sulII, dfrA14), indicating their strong identity to PCR- and sequence-confirmed alleles [19]. Interestingly, however, other alleles on the ARDM that were closely related to this strain's confirmed content were not as strongly positive (e.g., blaCTX-M-12, blaCTX-M-32, blaTEM-1, blaSHV-5, aac(6)-ib, aadA1b, aadA2, aadA1, mph(A), mph(K), blaTEM-10, aac(3)-III, tet(A)). These results are likely due to sequence divergence between strains, and illustrate the presence of highly conserved and less conserved sequences among the ARDM content.
The trio of alleles representing the aad(A1/A2) family of aminoglycoside adenyltransferases exemplify the importance of having an appropriate threshold algorithm to reflect both conserved and non-conserved sequences. Since K. pneumoniae strain BAA-2146 possesses aadA2, it was not surprising that 8/9 probes for aadA2 were positive. However, only 2/7 probes for aadA1 and 4/10 probes for aadA1b were deemed positive, in spite of the 89% sequence identity of aadA1 and aadA1b with aadA2. In contrast, A. baumannii BAA-1710, which possesses two copies of aadA1, had 100% positive probes for both aadA1 and aadA1b, but only 2/9 probes positive for aadA2. Thus, each strain showed its appropriate pattern of positive probes from these alleles, but positive determination for the aadA1/aadA2 family of determinants would be highly dependent on which gene is represented on the microarray and the threshold algorithm used. These results also underscore the need for the careful selection of probes for the detection of families of closely related resistance genes. In practice, due to finite space available on the microarrays, tradeoffs must be made between the breadth of the microarray content and the depth of resolution for particular gene families.
Algorithm testing with laboratory-extracted DNA.
To determine whether the same algorithms for positive allelic determination was appropriate for DNA prepared in our laboratory, a similar ROC analysis was performed with four genome-sequenced A. baumannii strains (Figure 3, gray curve). As evident from the ROC curve, the ARDM is a still statistically valid test for correct determination of the covered resistance genotypes, albeit somewhat less accurate for lab-extracted strains than those prepared commercially using phenol-chloroform extraction. This latter set of ROC analyses also assisted us in selecting the thresholding conditions giving the desired specificity and sensitivity for samples prepared in our laboratories. Under the lowest stringency algorithm previously determined to provide acceptable sensitivity and specificity (30% positive probes for each allele), 15 false positives were observed. However, by requiring 40% positive probes for each allele, the number of false-positives was decreased by 60%, while maintaining the specificity and sensitivity desired. We therefore decided to use this algorithm (40% of allele-specific probes positive) for the analyses of unknown clinical isolates. Table 3 shows the results from ARDM analyses for all 14 reference strains tested using the selected threshold algorithm. The efficacy of hybridization for over 50 of the 239 alleles represented on the ARDM was confirmed by PCR, DNA sequencing, or both. Alleles confirmed positive included those from both Gram-negative and Gram–positive species and conferring resistance to β-lactams, aminoglycosides, macrolides, tetracyclines, sulfonamides, chloramphenicol, quaternary amine compounds, glycopeptides, rifampin and lincosamides. Of the sequenced reference strains tested, A. baumannii AYE (BAA-1710) and K. pneumoniae 700721 each had at least 19 positive alleles with the potential to render these strains resistant to eight and six different classes of antimicrobials, respectively. Of the non-sequenced strains, K. pneumoniae BAA-2146 (blaNDM-1-positive) had 21 positive alleles conferring resistance to six classes of antimicrobials; in addition to PCR confirmation of these alleles [19], antibiograms provided by ATCC further support the phenotypic traits provided by these alleles (http://www.atcc.org/attachments/19037.pdf).
Several false-negative alleles were observed in our analyses of the reference strains. In all but two cases, the genes represented on the chip (dfrA15, floR, aadA2, blaLEN, blaPER, tet(S)) were less than 90% identical to the sequence of the reference strains tested, and therefore might not be expected to hybridize to the short probes used in this type of microarray. As a result, probes representing more distantly related alleles will be incorporated into future ARDM versions. Of the two false-negative alleles on the ARDM that were >90% identical to the reference strains, all of the tet(M) probes had at least three mismatches and all but two of the blaOXA-7 probes had at least two mismatches from the reference strain sequences. Given the overall high identity between the sequences used to construct the ARDM and the reference strains, it is clear that redesigned probes will need to have more conserved regions of each gene represented in future versions of the ARDM.
The source for the false-negatives observed with K. pneumoniae BAA-2146 is not known, as this strain's genome has not yet been fully sequenced. Therefore, while the presence of a tet(A) allele was previously confirmed by PCR, sequence non-homology within the probed regions may be responsible for its poor hybridization on the ARDM. Somewhat more puzzling is the false-negative determination for blaTEM-10. Although sequences for blaTEM-1 and blaTEM-10 are 99% identical, K. pneumoniae BAA-2146 hybridized strongly to blaTEM-1 sequences (8/9 probes positive), but poorly to blaTEM-10 probes (2/6 probes). While redesign of these probes may improve ARDM performance, representation of two such highly homologous alleles on the ARDM (discussed below) is not required unless a higher degree of redundancy is desired.
Presumptive false-positive results were observed with two of the four A. baumannii strains with published sequence contigs. While each of these alleles met the criteria for positive detection (40% probes positive), in all cases a maximum of four probes were deemed positive. Furthermore, several sets of the tet(30), norA, and vcaM probes gave positive signals in many other samples, including reference strains for which there were no analogous sequences found. The redesign of probes targeting these alleles in subsequent microarray designs (e.g., ARDM v.2) has already begun to address the shortcomings of the first generation ARDM to reduce both false-negative and false-positive identifications.
Analysis of clinical isolates from Egyptian hospitals
Recent surveillance efforts in the Middle East have indicated that the majority of healthcare-acquired, MDR infections are due to Gram-negative organisms [20]. A panel of A. baumannii, E. coli and K. pneumoniae blood and urine infection isolates (Table 2) was selected for characterization of their underlying genotypes. These three species represent a significant proportion (in some studies, over half) of the Gram-negative organisms associated with nosocomial infections throughout the Middle East [20], [21], [22], [23]. Using the selected threshold algorithm, ARDM analyses revealed the presence of five to 17 determinants (Tables 4 and 5) in each of the clinical isolates. In each case, alleles conferring resistance to at least three classes of antimicrobials were identified.
ARDM detection of β-lactam resistance genes and confirmation by PCR and sequencing
A significant proportion of the ARDM v.1 was devoted to genes conferring resistance to β-lactam antibiotics due to their importance in clinical settings. The most important mechanism of β-lactam resistance in Gram-negative bacteria relies on the production of β-lactamases. Due to the rise of broad-spectrum β-lactamases such as ESBLs and carbapenemases among nosocomial isolates in the Middle East [20], [24], we sought to confirm the presence of these genes in the Egyptian isolates by PCR and DNA sequencing to document in detail the prevalence and spread of specific alleles in the analyzed collection.
In parallel with ARDM analyses, DNA from each of the 15 clinical isolates was subjected to multiplex and individual PCR for the blaSHV, blaTEM, blaCTX-M and blaOXA β-lactamase families (Table 4). Multiplex and individual PCRs were performed using the same, previously published primers [17]. In each case where PCR generated the appropriately sized product, the amplicons were purified and sequenced to determine the specific allele present.
The ARDM v.1 content covers genes for ampC cephalosporinases and the blaTEM, blaSHV, blaOXA and blaGES families of β-lactamases (some of which are ESBLs and carbapenemases), as well as other families of β-lactamase determinants encoding exclusively ESBLs (e.g., blaACC, blaCTX-M, blaVEB, blaPER, and blaLAT-1/CMY) or carbapenemases (blaKPC, blaSME, blaIMP, blaVIM, blaSPM, blaGIM, and blaSIM). Although several of the reference strains were ARDM-positive for blaVEB, blaGES, and blaLAT-1/CMY (in agreement with their published gene contents [25], [26]), only alleles belonging to the ampC, blaTEM, blaSHV, blaOXA and blaCTX-M families were detected in the Egyptian isolates. In all of the Enterobacteriaceae isolates tested, alleles specific for multiple families were detected; the presence of multiple β-lactamases/ESBLs from different groups has been widely observed in other studies of Egyptian Enterobacteriaceae strains [27], [28], [29].
None of the carbapenemases represented on the ARDM v.1 were detected in any of the clinical isolates tested. The absence of these alleles in the E. coli and K. pneumoniae isolates was supported by their sensitivity to imipenem. In contrast, the Acinetobacter isolates were phenotypically resistant to imipenem but were negative for hybridization to any of the carbapenemase probes. It is very likely that the A. baumannii isolates harbored one or more carbapenem-hydrolyzing class D β-lactamases, which are commonly associated with Acinetobacter. Studies with the four (carbapenem-resistant) A. baumannii reference strains obtained from MRSN, for which genome sequences are available, indicate the presence of blaOXA-51 in all four, and blaOXA-23 in two (data not shown). As a result, probes for blaOXA-51, blaOXA-23 and other carbapenem hydrolyzing OXAs will be incorporated into future versions of the ARDM.
BlaTEM was the most frequently detected family of β-lactamase genes by ARDM analyses. All 12 ARDM-positive results were confirmed by PCR, as were blaTEM-negative results. BlaTEM genes were found in two of the A. baumannii isolates, in five of the six E. coli strains tested, and in all of the K. pneumoniae strains tested. Similar distributions of blaTEM have been described in several other studies of Enterobacteriaceae isolates from Egypt [28], [29], but typically, blaTEM is only rarely encountered in Acinetobacter [30], [31]. In spite of the presence of two blaTEM alleles represented on the ARDM, neither ARDM nor sequence analysis of PCR amplicons was able to distinguish between the broad-spectrum blaTEM determinants (blaTEM-1, blaTEM-2) and their ESBL derivatives. Although in several cases, sequences of the PCR amplicons could confirm which family of blaTEM was present (blaTEM-1 group or blaTEM-2 group; samples N24 and N25, respectively), an unequivocal identification of a specific allele could not be made for any of the clinical or reference samples tested.
BlaSHV β-lactamases were detected in five K. pneumoniae strains by ARDM analysis, with all five confirmed by PCR. As with the blaTEM alleles, no definitive identification of any specific allele could be made based on the sequence data. At least eight different alleles were identified as equivalent matches for the detected sequences, among them blaSHV-12 and blaSHV-5, which have been previously observed in Klebsiella from Egypt [29], [32]. No blaSHV alleles were detected in the A. baumannii or E. coli samples tested.
The blaOXA family of β-lactamases was represented on the ARDM v.1 by probes specific for four alleles: blaOXA-1 and blaOXA-7 (blaOXA-1 family), blaOXA-9, and blaOXA-10. Genes from the blaOXA-1 group were detected in both E. coli and K. pneumoniae, but not in A. baumannii. In each case, results were confirmed by PCR and DNA sequencing, and it was determined that the allele present in each was blaOXA-1. BlaOXA-9 was present in four of the five clinical K. pneumoniae isolates and these results were confirmed as well using a blaOXA-9-specific PCR. Two of these K. pneumoniae samples (N25, N29) possessed genes representing both OXA-1 and OXA-9 families. None of the 15 clinical isolates tested was ARDM-positive for genes from the blaOXA-10 family, which includes ESBLs such as blaOXA-11, blaOXA-14, blaOXA-16 (also known as expanded-spectrum OXAs) and the carbapenem-hydrolyzing class D β-lactamase, blaOXA-128. While information on the presence and spread of OXA-type enzymes in Egypt is limited to a single case report describing blaOXA-1 and blaOXA-10 containing Pseudomonas aeruginosa strains [33], our results suggest that both OXA-1 and OXA-9 type enzymes may be common among E. coli and K. pneumoniae strains isolated from hospital environments in Egypt.
Genes belonging to CTX-M family of ESBLs were detected in the E. coli and K. pneumoniae isolates, but not in any of the A. baumannii samples. BlaCTX-M-1 group alleles (which include blaCTX-M-3, blaCTX-M-12, blaCTX-M-32, blaFEC-1) were found in four E. coli isolates and in two of the five K. pneumoniae isolates. Additionally, alleles representing the blaCTX-M-9 family (which include blaCTX-M-9, blaCTX-M-13) were found in half of the E. coli isolates and in one K. pneumoniae isolate. One isolate (E. coli N24) contained alleles from both blaCTX-M-1 and blaCTX-M-9 groups. To our knowledge, this isolate is the first Egyptian strain documented to harbor two blaCTX-M genes belonging to distinct families. None of the clinical isolates were positive for genes from the blaCTX-M-2 and blaCTX-M-8 groups. All ARDM detection results for CTX-M family genes were confirmed by PCR and DNA sequencing determinations further indicated that the ARDM analyses correctly classified the detected alleles within the appropriate families. In five samples (E. coli N7, N16, N24 and K. pneumoniae N11, N29), it was possible to identify the individual allele present based on the sequences obtained. In each instance, the gene identified was either blaCTX-M-15 (blaCTX-M-1 group) or blaCTX-M-14b (blaCTX-M-9 group). These results agree with other studies documenting the widespread distribution of blaCTX-M-15 in Egypt [34], [35], [36], and also suggest a high prevalence of blaCTX-M-14 in the hospitals from which the strains originated. Although blaCTX-M-14 was previously reported from this region, its prevalence was reported to be relatively low [28], [37], [38], [39].
ARDM detection of genes conferring resistance to other antibiotic classes
Aminoglycoside resistance.
ARDM content covers 27 alleles conferring resistance to aminoglycosides, and in all of the clinical isolates tested, at least one aminoglycoside resistance determinant was present. Eight different alleles were detected, representing all three categories of aminoglycoside-modifying enzymes (acetyltransferases, adenylyltransferases, phosphotransferases). Isolates N2, N9, and N16 possessed alleles for all three types of enzymes, with as many as five different alleles detected in the same sample (N16) (Table 5). The most commonly encountered genes were those belonging to the aadA1/aadA2 family (nine isolates); only one of these isolates harbored an allele more closely homologous to aadA2 than aadA1/aadA1b based on the percentage of probes positive for the three alleles. The aph3/str(A) and aph6/str(B) genes were co-localized in seven isolates, while a single sample (N1) harbored only aph3/str(A). Acetyltransferase genes aac(3)-iii and aac(6)-ib/aac(6′)-ib were detected in 50–60% percent of the K. pneumoniae and E. coli isolates, often together. Similar profiles for aac(3)-III, aac(6)-ib/aac(6′)-ib, aph3/str(A), aph6/str(B), and aadA1/aadA1b/aadA2 in isolates of E. coli and K. pneumoniae from other Cairo hospitals have been described [21], [39]. Two of the A. baumannii isolates exhibited strong hybridization to ant(2″)-Ia/aadB probes (80–90% positive probes) and two to aacC1 probes (100% positive probes). However, neither of these alleles was detected in the E. coli or K. pneumoniae isolates.
Macrolide resistance.
A number of macrolide resistance determinants were detected in the clinical isolates. Only one of the K. pneumoniae isolates (N11) was positive for a resistance determinant, ere(A2), which demonstrated strong hybridization to the ARDM probes. Genes for phosphotransferases mph(A) and mph(K) – 99% identical in sequence – were also detected in two of the E. coli samples (N21, N23), whereas mph(B) was identified in E. coli N16. The intrinsic ABC macrolide transporter genes, macA and macB, were detected in all E. coli isolates except for sample N28, where only macB was detected.
Tetracyline resistance.
The presence of ARDM-detected tetracycline resistance alleles appeared to be mostly species-specific. Tet(39) was detected only in A. baumannii (samples N1, N2, N3), tet(D) in K. pneumoniae (N11, N25, N29) and tet(B) mostly in E. coli (N21, N24, N28). On the other hand, tet(A) was observed in one A. baumannii (N2) and one E. coli (N16). Individual PCRs were used to confirm these results and revealed that the ARDM-detected tet(B), tet(38) and tet(30) alleles detected in A. baumannii N3 were false-positives. Although the potential for the tet(30) and tet(38) probes on the ARDM giving rise to false-positive results in A. baumannii has been recognized (data not shown), the presence of three false-positive results for this single sample may indicate that contamination occurred during preparation for ARDM analysis; no other confirmed false-positives were encountered during the study of these clinical samples.
Two phenotypically tetracycline resistant samples were found to be negative for all of the 39 tetracycline resistance determinants represented on the ARDM. In one case (E. coli N23) the presence of tet(A) was confirmed by PCR. This false-negative result was due to unusually high variability in probe signals throughout the array; this high variability increased the “probe threshold” (mean of lowest 95% probes+3 SD), leading to a large number of the tet(A) probes erroneously being deemed negative. When analyzed using lower stringency algorithms for calculating the “probe threshold” (e.g., mean of lowest 90% probes+3 SD, mean of lowest 95% probes+2 SD), eight of the nine tet(A) probes were deemed positive. The probe signal variability may have also affected detection of the sulI allele in this sample (see below).
Chloramphenicol resistance.
Chloramphenicol resistance is represented by 12 alleles present on the ARDM. Resistance determinants of the catA1/cat4 family were detected in five of the six E. coli clinical isolates, as well as in single K. pneumoniae and A. baumannii isolates. While the presence of these alleles corresponded to phenotypic chloramphenicol resistance, several strains (N3, N9, N11) also demonstrated a resistant phenotype but were negative for any of the determinants represented on the ARDM. This resistance may be due to catB or cmlA genes (not present on this ARDM version), which have been detected in other E. coli isolated from Egyptian hospitals [40]. The floR allele, conferring resistance to florfenicol, was not detected in this study.
Diaminopyrimidine resistance.
Resistance to diaminopyrimidines was represented by probes directed against 29 alleles of dihydrofolate reductase (dfr). Although no dfr alleles were detected in the A. baumannii isolates, five E. coli and K. pneumoniae isolates were positive for the presence of at least one dfr gene; isolates N23 and N26 each harbored two different dfr genes. The alleles, dfrA5, dfrA7, and dfr12 - observed at high frequency in other studies of MDR isolates from this region [22], [41], [42], [43], [44] – were not found in the clinical strains tested here. However the other dfr alleles (dfrA1, dfrA15, dfrA17, and dfrA19) detected in our isolates have been well documented by others in various E. coli, K. pneumoniae, and Enterobacteriaceae isolates from this region [22], [41], [42], [43], [44], [45], [46], [47].
Interestingly, the trimethoprim resistance determinant most frequently encountered in the current study was dfrA14; to our knowledge, this allele has never been observed in any Enterobacteriaceae isolates of Egyptian origin. The high frequency at which it was detected in the current study (33% of isolates) may be indicative of a widely disseminated genetic element within the Cairo area which has not yet been unequivocally identified. Anantham and Hall recently described a plasmid, pCERC1, possessing dfrA14 within a disrupted aph3/str(A) gene and postulated its close relationship to other small, globally distributed plasmids with similar gene sequences of sulII→str(A)→dfrA14→str(A)→str(B) [48]; several of these potentially related plasmids have been recovered from African isolates [49], [50]. In the present study, two E. coli isolates and one K. pneumoniae isolate possessing dfrA14 (N21, N23, N26) also harbored sequences that hybridized strongly to aph3/str(A), aph6/str(B), and sulII probes on the ARDM. While we have not ascertained clustering between the four genes detected in these samples, it is possible that sequences homologous to those found in pCERC1 or its related plasmids are present in these Egyptian strains.
Determinants directed against other classes of antimicrobials.
Resistance to quinolones was represented on the ARDM v.1 by probes directed against norA and qnrA. Although 12 of the clinical isolates were phenotypically resistant to quinolones, only A. baumannii N3 was deemed positive for hybridization to norA, and none were positive for qnrA. The mechanism for quinolone resistance for the remaining 11 resistant isolates was not determined, but could potentially be due to mutations within the DNA gyrase or DNA topoisomerase IV genes, which would not be detectable using this technology. Six of the clinical isolates that were quinolone-resistant were deemed positive for aac(6′)-Ib, which potentially indicated the presence of its quinolone resistance-conferring variant, aac(6′)-Ib-cr. However, all six isolates were also resistant to nalidixic acid, indicating the absence of the aac(6′)-Ib-cr variant, which is highly specific for quinolones possessing piperazinyl substituents [51]. Alternatively, these isolates may possess other plasmid-borne genes conferring resistance to fluoroquinolones (e.g., qep, qnrB, qnrC, qnrD, qnrS) not represented on the current version of the ARDM but which will be incorporated into future versions.
Sulfonamide resistance genes were detected in 10 isolates across all analyzed species. SulI was found in seven isolates (N2, N9, N11, N16, N19, N26, N28) and was also most likely present in sample N23, which had an unusually high probe signal variability (resulting in false-negative detection of tet(A), see “tetracycline resistance” above). When less stringent criteria (the same as in case of tet(A)) were used for “probe threshold” calculations, sulI became positive with 7/10 probes. SulII was detected in six samples; in three of these samples, it was co-localized with sulI (N16, N11, N26). SulIII was not detected.
Resistance gene clusters.
The identification of often-linked groups of characteristic genes allowed us to deduce the likely presence of a number of antimicrobial determinant-harboring genetic structures such as integrons and resistance islands (in addition to the previously mentioned pCERC1 plasmid). In seven of the eight strains where the qacEΔ1 gene (encoding for resistance to quaternary ammonium compounds) was detected, sulI was also present, which may potentially indicate the presence of a class 1 integron. Five of the qacEΔ1+/sulI+ samples were also positive for the presence of genes from the aadA1/aadA2 family, and two harbored the trimethoprim resistance genes dfrA1 and dfrA12; aadA1 has been described as the most frequently integrated gene cassette in class 1 integrons [52], [53]. As the ARDM content did not include probes targeting sat2 or intI2, it is unclear whether the coexistence of aadA1 and dfrA1 in samples N16 and N23 might also indicate the presence of class 2 integrons (the most widespread of which carries conserved dfrA1, sat2, and aadA1 gene cassettes) [54]. The ARDM assay did confirm the presence of resistance island AbaR1 in the A. baumannii BAA-1710 reference strain, the strain in which it was first identified [55], as all 16 AbaR1 genes represented on the microarray were hybridization-positive. We also observed specific clusters of resistance determinants in clinical A. baumannii isolates N2 and N9 that indicated the potential presence of resistance islands AbaR3 (seven genes: tet(A), catA1, blaTEM, aphA1b, aacC1, aadA1, sulI) and AbaR6/R7 (four genes: aphA1b, aacC1, aadA1, sulI) [56], respectively.
Conclusions
Taken together, the results demonstrated that the ARDM is capable of detecting antimicrobial resistance genes and genetic assemblages that are known to confer resistance to first and second line antibiotics that are preferred for the treatment of Gram-negative infections. While the findings of clinical strain analysis were largely in line with previous studies [27], [35], [57], [58], [59], [60], we also detected new alleles (e.g. dfrA14, blaOXA-9) that were clearly circulating in their region of origin and likely missed due to the narrow spectrum of most molecular biology tools. In addition to its strength as a broad screening tool which is complemented by front-end random nucleic acid amplification, the ARDM offers a number of advantages over previously developed tools due to its use of 30–35mer probes and empirically derived detection thresholds (increased specificity), electrochemical hybridization signal detection (increased sensitivity), and a small scanner without any optical systems that can be prone to misalignment in transport (portability). Combined, these attributes make the ARDM a powerful tool for providing molecular epidemiological surveillance information on emerging resistance trends, the identification of resistance gene reservoirs, and potentially providing supporting data to better inform therapeutic decisions in uncharacterized or poorly characterized environments.
Supporting Information
Figure S1.
For top panel (24522), peaks correspond to the following alleles: a: mac(A) [5/9] and mac(B) [10/10]; b: blaACC-2 [1/8] and mefA [2/9]; c: van(C3) [2/10]; d: norA [1/9]; e: aph(3′)-III [2/10]; f: blaPSE-1/CARB-1 [2/9]; g: blaCTX-M-2 [1/10]. All alleles except mac(A) and mac(B) were deemed negative using the optimized threshold algorithm. For panel (700721), peaks correspond to the following alleles: a: mecA [1/10 probes positive; allele deemed negative]; b: blaSHV-1 [10/10]; c: blaTEM-1 [9/9] and blaSHV-5 [6/8]; d: sulII [4\6] and cat4 [3/7]; e: catA1[6/8]; f: (cluster) aph6/str(B) [6/10], aph3″/str(A) [7/9], aac(6)-Ib [5/8], and aad(A1b) [6/10]; g: aadB [4/10], qacEΔ1 [10/10], and aad(A1) [7/7]; h: aac(6′)-Ib [3/6]; i: blaOXA-9 [6/9]; j: blaTEM-10 [3/6] and aphA1 [8/10]; k: sulI [8/10]; l: tet(D) [8/8].
https://doi.org/10.1371/journal.pone.0069507.s002
(PDF)
Table S1.
Antibiotic resistance profiles of the clinical isolates1. 1 R - resistant, S - sensitive, I - intermediate resistant (according to CLSI criteria). Blank cells - test was not done.
https://doi.org/10.1371/journal.pone.0069507.s003
(PDF)
Acknowledgments
We would like to thank COL E. Lesho, CPT B. Kirkup, and Dr. D. Zurawski for providing the genome-sequenced MRSN strains. We also thank Dr. J. Klena for his assistance with the E. coli strain N7 (E450) and Drs. B. Lin and Z. Wang for their critical evaluation of this manuscript. B.B. was a National Research Council postdoctoral fellow at CBMSE/NRL. The opinions and assertions contained herein are those of the authors and are not to be construed as those of the U.S. Navy, military service at large or U.S. Government.
Some of the authors are employed by the US Government and this work was prepared as part of their official duties. Title 17, USC, section 105 provides that ‘Copyright protection under this title is not available for any work of the US Government.’ Title 17, USC, section 101 defines a U.S. Government work as a prepared by a military service member or employee of the US Government as part of that person's official duties.
Author Contributions
Conceived and designed the experiments: TAL GJV CRT GP. Performed the experiments: TAL GJV CRT BRB BLH MN MW MAM. Analyzed the data: TAL GJV CRT GP. Contributed reagents/materials/analysis tools: TAL GJV CRT BRB BLH MN MW MAM. Wrote the paper: TAL GJV CRT. Designed the software used in analysis: TAL.
References
- 1. Bush K (2010) Alarming β-lactamase-mediated resistance in multidrug-resistant Enterobacteriaceae. Curr Opin Microbiol 13: 558–564.
- 2. Hawkey PM, Jones AM (2009) The changing epidemiology of resistance. J Antimicrob Chemother 64 Suppl 1: i3–10.
- 3. Livermore DM (2009) Has the era of untreatable infections arrived? J Antimicrob Chemother 64 Suppl 1: i29–36.
- 4. Hanson ND (2010) Molecular diagnostics could help in coping with hidden β-lactamases. Microbe 5: 333–339.
- 5. Lascols C, Hackel M, Hujer AM, Marshall SH, Bouchillon SK, et al. (2012) Using nucleic acid microarrays to perform molecular epidemiology and detect novel β-lactamases: A snapshot of ESBLs throughout the world. J Clin Microbiol 50: 1632–1639.
- 6. Garneau P, Labrecque O, Maynard C, Messier S, Masson L, et al. (2010) Use of a bacterial antimicrobial resistance gene microarray for the identification of resistant Staphylococcus aureus. Zoonoses Public Health 57 Suppl 1: 94–99.
- 7. Cassone M, D'Andrea MM, Iannelli F, Oggioni MR, Rossolini GM, et al. (2006) DNA microarray for detection of macrolide resistance genes. Antimicrob Agents Chemother 50: 2038–2041.
- 8. Naas T, Cuzon G, Bogaerts P, Glupczynski Y, Nordmann P (2011) Evaluation of a DNA microarray (Check-MDR CT102) for rapid detection of TEM, SHV, and CTX-M extended-spectrum β-lactamases and of KPC, OXA-48, VIM, IMP, and NDM-1 carbapenemases. J Clin Microbiol 49: 1608–1613.
- 9. Batchelor M, Hopkins KL, Liebana E, Slickers P, Ehricht R, et al. (2008) Development of a miniaturised microarray-based assay for the rapid identification of antimicrobial resistance genes in Gram-negative bacteria. Int J Antimicrob Agents 31: 440–451.
- 10. Frye JG, Lindsey RL, Rondeau G, Porwollik S, Long F, et al. (2010) Development of a DNA microarray to detect antimicrobial resistance genes identified in the National Center for Biotechnology Information database. Microb Drug Resist 16: 9–19.
- 11. Perreten V, Vorlet-Fawer L, Slickers P, Ehricht R, Kuhnert P, et al. (2005) Microarray-based detection of 90 antibiotic resistance genes of Gram-positive bacteria. J Clin Microbiol 43: 2291–2302.
- 12. Stokes HW, Gillings MR (2011) Gene flow, mobile genetic elements and the recruitment of antibiotic resistance genes into Gram-negative pathogens. FEMS Microbiol Rev 35: 790–819.
- 13. Zurawski DV, Thompson MG, McQueary CN, Matalka MN, Sahl JW, et al. (2012) Genome sequences of four divergent multidrug-resistant Acinetobacter baumannii strains isolated from patients with sepsis or osteomyelitis. J Bacteriol 194: 1619–1620.
- 14.
Clinical and Laboratory Standards Institute (2009) Performance standards for antimicrobial susceptibility testing: nineteenth informational supplement M100-S19.
- 15. Roth KM, Peyvan K, Schwarzkopf KR, Ghindilis A (2006) Electrochemical detection of short DNA oligomer hybridization using the CombiMatrix ElectraSense microarray reader. Electroanalysis 18: 1982–1988.
- 16. Wojciechowski J, Chase-Baldwin K, Wasieloski LP Jr, Padilla S, Vora GJ, et al. (2010) Enhancement of deoxyribonucleic acid microarray performance using post-hybridization signal amplification. Anal Chim Acta 679: 85–90.
- 17. Fang H, Ataker F, Hedin G, Dornbusch K (2008) Molecular epidemiology of extended-spectrum β-lactamases among Escherichia coli isolates collected in a Swedish hospital and its associated health care facilities from 2001 to 2006. J Clin Microbiol 46: 707–712.
- 18. Bogaerts P, Montesinos I, Rodriguez-Villalobos H, Blairon L, Deplano A, et al. (2010) Emergence of clonally related Klebsiella pneumoniae isolates of sequence type 258 producing KPC-2 carbapenemase in Belgium. J Antimicrob Chemother 65: 361–362.
- 19. Leski T, Vora G, Taitt CR (2012) Multidrug resistance determinants from NDM-1 producing Klebsiella pneumoniae, United States of America. Int J Antimicrob Agents 40: 282–284.
- 20. Meyer W, Pavlin J, Hospenthal D, Murray C, Jerke K, et al. (2011) Antimicrobial resistance surveillance in the AFHSC-GEIS network. BMC Public Health 11: S8.
- 21. El Kholy A, Baseem H, Hall GS, Procop GW, Longworth DL (2003) Antimicrobial resistance in Cairo, Egypt 1999–2000: A survey of five hospitals. J Antimicrob Chemother 51: 625–630.
- 22. Hussein A, Ahmed A, Sato M, Shimamoto T (2009) Characterization of integrons and antimicrobial resistance genes in clinical isolates of Gram-negative bacteria from Palestinian hospitals. Microbiol Immunol 53: 595–602.
- 23. Saied T, Elkholy A, Hafez SF, Basim H, Wasfy MO, et al. (2011) Antimicrobial resistance in pathogens causing nosocomial bloodstream infections in university hospitals in Egypt. Am J Infect Control 39: e61–e65.
- 24. Poirel L, Abdelaziz M, Bernabeu S, Nordmann P (2013) Occurrence of OXA-48 and VIM-1 carbapenemase-producing Enterobacteriaceae in Egypt. Int J Antimicrob Agents 41: 90–91.
- 25. Poirel L, Menuteau O, Agoli N, Cattoen C, Nordmann P (2003) Outbreak of extended-spectrum β-lactamase VEB-1-producing isolates of Acinetobacter baumannii in a French hospital. J Clin Microbiol 41: 3542–3547.
- 26. Pasteran F, Mendez T, Rapoport M, Guerriero L, Corso A (2010) Controlling false-positive results obtained with the Hodge and Masuda assays for detection of class a carbapenemase in species of Enterobacteriaceae by incorporating boronic acid. J Clin Microbiol 48: 1323–1332.
- 27. Ahmed SH, Daef EA, Badary MS, Mahmoud MA, Abd-Elsayed AA (2009) Nosocomial blood stream infection in intensive care units at Assiut University Hospitals (Upper Egypt) with special reference to extended spectrum β-lactamase producing organisms. BMC Res Notes 2: 76.
- 28. Khalaf NG, Eletreby MM, Hanson ND (2009) Characterization of CTX-M ESBLs in Enterobacter cloacae, Escherichia coli and Klebsiella pneumoniae clinical isolates from Cairo, Egypt. BMC Infect Dis 9: 84.
- 29. Al-Agamy M (2012) Genetic basis of cefotaxime resistant isolates of Klebsiella pneumoniae from Cairo. African J Microbiol Res 6: 20–27.
- 30.
Poirel L, Nordmann P (2008) Acinetobacter baumannii: Mechanisms of resistance, multiple β-lactamases. In: Bergogne-Berezin E, Friedman H, Bendinelli M, editors. Acinetobacter Biology and Pathogenesis. New York: Springer. pp. 129–144.
- 31. Naiemi NA, Duim B, Savelkoul PH, Spanjaard L, de Jonge E, et al. (2005) Widespread transfer of resistance genes between bacterial species in an intensive care unit: implications for hospital epidemiology. J Clin Microbiol 43: 4862–4864.
- 32.
Fouad S, House B, Husain T, Newire E, Pimentel G, et al.. (2011) First detection of SHV-12, and SHV-5 and SHV-2a variants of ESBLs producing Klebsiella pneumoniae in Egypt. Proc ICAAC Sept 2011.
- 33. Szabo D, Szentandrassy J, Juhasz Z, Katona K, Nagy K, et al. (2008) Imported PER-1 producing Pseudomonas aeruginosa, PER-1 producing Acinetobacter baumannii and VIM-2-producing Pseudomonas aeruginosa strains in Hungary. Ann Clin Microbiol Antimicrob 7: 12.
- 34. Fam N (2008) El-Damarawy (2008) CTX-M-15 extended-spectrum β-lactamases detected from intensive care unit of an Egyptian medical research institute. Res J Med Med Sci 3: 84–91.
- 35. Fam N, Leflon-Guibout V, Fouad S, Aboul-Fadl L, Marcon E, et al. (2011) CTX-M-15-producing Escherichia coli clinical isolates in Cairo (Egypt), including isolates of clonal complex ST10 and clones ST131, ST73, and ST405 in both community and hospital settings. Microb Drug Resist 17: 67–73.
- 36. Poirel L, Rotimi V, Bernabeu S, Jamal W, Nordmann P (2007) P1029 Explosive emergence of CTX-M-15 extended-spectrum β-lactamase in Enterobacteriaceae in Kuwait. Int J Antimicrob Agents 29, Suppl 2: S276.
- 37. Abdel Ghani S, Moland E, Black J, Hanson N, Goering R, et al. (2010) First report of CTX-M-14-producing clinical isolates of Salmonella serovar Typhimurium from Egypt. J Infect Dev Ctries 4: 058–060.
- 38. Al Hashem G, Al Sweih N, Jamal W, Rotimi V (2011) Sequence analysis of blaCTX-M genes carried by clinically significant Escherichia coli isolates in Kuwait hospitals. Med Princip Pract 20: 213–219.
- 39. Al-Agamy M, Ashour M, Wiegand I (2006) First description of CTX-M β-lactamase-producing clinical Escherichia coli isolates from Egypt. Int J Antimicrob Agents 27: 545–548.
- 40. Al-Agamy M, El-Ghareeb K, Ashour M, Wiedemann B, Weigard I (2006) Detection of chloramphenicol resistance genes in broad spectrum β-lactamase-producing Escherichia coli strains. N Egypt J Microbiol 13: 17–28.
- 41. Ahmed AM, Shimamoto T (2011) Molecular characterization of antimicrobial resistance in Gram-negative bacteria isolated from bovine mastitis in Egypt. Microbiol Immunol 55: 318–327.
- 42. Ahmed AM, Shimamoto T (2012) Genetic analysis of multiple antimicrobial resistance in Salmonella isolated from diseased broilers in Egypt. Microbiol Immunol 56: 254–261.
- 43. Ahmed AM, Younis EEA, Osman SA, Ishida Y, El-Khodery SA, et al. (2009) Genetic analysis of antimicrobial resistance in Escherichia coli isolated from diarrheic neonatal calves. Vet Microbiol 136: 397–402.
- 44. Hammad AM, Ishida Y, Shimamoto T (2009) Prevalence and molecular characterization of ampicillin-resistant Enterobacteriaceae isolated from traditional Egyptian Domiati cheese. J Food Protect 72: 624–630.
- 45. Ahmed AM, Younis EEA, Ishida Y, Shimamoto T (2009) Genetic basis of multidrug resistance in Salmonella enterica serovars Enteritidis and Typhimurium isolated from diarrheic calves in Egypt. Acta Tropica 111: 144–149.
- 46. Ishida Y, Ahmed AM, Mahfouz NB, Kimura T, El-Khodery SA, et al. (2010) Molecular analysis of antimicrobial resistance in Gram-negative bacteria isolated from fish farms in Egypt. J Vet Med Sci 72: 727–734.
- 47. Veldman K, Dierikx C, van Essen-Zanbergen A, van Pelt W, Mevius D (2010) Characterizatin of multidrug-resistant, qnrB2-positive and extended-spectrum-β-lactamase-producing Salmonella Concord and Salmonella Senftenberg isolates. J Antimicrob Chemother 65: 872–875.
- 48. Anantham S, Hall R (2012) pCERC1, a small, globally disseminated plasmid carrying the dfrA14 cassette in the strA gene of the sul2-strA-strB gene cluster. Microb Drug Resist 18: 364–371.
- 49. Kikuvi G, Schwarz S, Ombui J, Mitema E, Kehrenberg C (1983) Streptomycin and chloramphenicol resistance genes in Escherichia coli isolates from cattle, pigs, and chicken in Kenya. Microb Drug Resist 13: 62–68.
- 50. Ojo K, Kehrenberg C, Schwarz S, Odelola H (2002) Identification of a complete dfrA14 gene cassette integrated at a secondary site in a resistance plasmid of uropathogenic Escherichia coli from Nigeria. Antimicrob Agents Chemother 46: 2054–2055.
- 51. Robicsek A, Strahilevitz J, Jacoby G, Macielag M, Abbanat D, et al. (2006) Fluoroquinolone-modifying enzyme: A new adaptation of a common aminoglyocide acetyltransferase. Nat Med 12: 83–88.
- 52. Levesque C, Piche L, Larose C, Roy P (1995) PCR mapping of integrons reveals several novel combinations of resistance genes. Antimicrob Agents Chemother 39: 185–191.
- 53. Sallen B, Rajoharison A, Desvarenne S, Mabilat C (1995) Molecular epidemiology of integron-associated antibiotic-resistance genes in clinical isolates of Enterobacteriaceae. Microb Drug Resist 1: 195–202.
- 54. Hansson K, Sundström L, Pelletier A, Roy PH (2002) IntI2 integron integrase in Tn7. J Bacteriol 184: 1712–1721.
- 55. Fournier PE, Vallenet D, Barbe V, Audic S, Ogata H, et al. (2006) Comparative genomics of multidrug resistance in Acinetobacter baumannii. PLoS Genet 2: e7.
- 56. Post V, White P, Hall R (2010) Evolution of AbaR-type genomic resistance islands in multiply antibiotic-resistant Acinetobacter baumannii. J Antimicrob Chemother 65: 1162–1170.
- 57. Salem M, Muharram M, Alhosiny I (2010) Distribution of classes 1 and 2 integrons among multi drug resistant E. coli isolated from hospitalized patients with urinary tract infection in Cairo, Egypt. Aust J Basic Appl Sci 4: 398–407.
- 58. Mohamed Al-Agamy MH, El-Din Ashour MS, Wiegand I (2006) First description of CTX-M β-lactamase-producing clinical Escherichia coli isolates from Egypt. Int J Antimicrob Agents 27: 545–548.
- 59. Bouchillon SK, Johnson BM, Hoban DJ, Johnson JL, Dowzicky MJ, et al. (2004) Determining incidence of extended spectrum beta-lactamase producing Enterobacteriaceae, vancomycin-resistant Enterococcus faecium and methicillin-resistant Staphylococcus aureus in 38 centres from 17 countries: The PEARLS study 2001–2002. Int J Antimicrob Agents 24: 119–124.
- 60. Tham J, Odenholt I, Walder M, Brolund A, Ahl J, et al. (2010) Extended-spectrum β-lactamase-producing Escherichia coli in patients with travellers' diarrhoea. Scand J Infect Dis 42: 275–280.
- 61. Smith MG, Gianoulis TA, Pukatzki S, Mekalanos JJ, Ornston LN, et al. (2007) New insights into Acinetobacter baumannii pathogenesis revealed by high-density pyrosequencing and transposon mutagenesis. Genes Dev 21: 601–614.
- 62. Tracz DM, Boyd DA, Bryden L, Hizon R, Giercke S, et al. (2005) Increase in ampC promoter strength due to mutations and deletion of the attenuator in a clinical isolate of cefoxitin-resistant Escherichia coli as determined by RT-PCR. J Antimicrob Chemother 55: 768–772.
- 63. Rasheed JK, Anderson GJ, Yigit H, Queenan AM, Domenech-Sanchez A, et al. (2000) Characterization of the extended-spectrum β-lactamase reference strain, Klebsiella pneumoniae K6 (ATCC 700603), which produces the novel enzyme SHV-18. Antimicrob Agents Chemother 44: 2382–2388.
- 64. Ogawa W, Li DW, Yu P, Begum A, Mizushima T, et al. (2005) Multidrug resistance in Klebsiella pneumoniae MGH78578 and cloning of genes responsible for the resistance. Biol Pharm Bull 28: 1505–1508.
- 65. Carvalhaes CG, Picao RC, Nicoletti AG, Xavier DE, Gales AC (2010) Cloverleaf test (modified Hodge test) for detecting carbapenemase production in Klebsiella pneumoniae: be aware of false positive results. J Antimicrob Chemother 65: 249–251.
- 66. Centers for Disease Control and Prevention (2010) Detection of Enterobacteriaceae isolates carrying metallo-β-lactamase - United States, 2010. MMWR Morb Mortal Wkly Rep 59: 750.
- 67. Kuroda M, Ohta T, Uchiyama I, Baba T, Yuzawa H, et al. (2001) Whole genome sequencing of methicillin-resistant Staphylococcus aureus. Lancet 357: 1225–1240.
- 68. Gill SR, Fouts DE, Archer GL, Mongodin EF, Deboy RT, et al. (2005) Insights on evolution of virulence and resistance from the complete genome analysis of an early methicillin-resistant Staphylococcus aureus strain and a biofilm-producing methicillin-resistant Staphylococcus epidermidis strain. J Bacteriol 187: 2426–2438.
- 69. Paulsen IT, Banerjei L, Myers GS, Nelson KE, Seshadri R, et al. (2003) Role of mobile DNA in the evolution of vancomycin-resistant Enterococcus faecalis. Science 299: 2071–2074.