Prevalence of ESBL-producing Pseudomonas aeruginosa isolates in Warsaw, Poland, detected by various phenotypic and genotypic methods

Knowledge of the prevalence of ESBL enzymes among P. aeruginosa strains compared to the Enterobacteraiceae family is limited. The phenotypic tests recommended by EUCAST for the detection of ESBL-producing Enterobacteriaceae are not always suited for P. aeruginosa strains. This is mainly due to the presence of other families of ESBLs in P. aeruginosa isolates more often than in Enterobacteriaceae, production of natural AmpC cephalosporinase and its overexpression, and co-production of metallo-β-lactamases. The aim of this study was to determine the occurrence of ESBLs in P. aeruginosa isolated from patients from hospitals in Warsaw, to evaluate the ESBL production of these isolates using currently available phenotypic tests, their modifications, multiplex PCR and molecular typing of ESBL-positive isolates by PFGE. Clinical isolates of P. aeruginosa were collected in 2000-2014 from four Warsaw hospitals. Based on the data obtained in this study, we suggest using three DDST methods with inhibitors, such as clavulanic acid, sulbactam and imipenem, to detect ESBL-producing P. aeruginosa strains. Depending on the appearance of the plates, we suggest a reduction in the distance between discs with antibiotics to 15 mm and the addition of boronic acid at 0.4 mg per disc. The analysed isolates carried genes encoding ESBL from the families VEB (69 isolates with VEB-9), GES (6 with GES-1, 1 GES-5, 5 GES-13 and 2 with GES-15), OXA-2 (12 with OXA-15, 1 OXA-141, 1 OXA-210, 1 OXA-543 and 1 with OXA-544) and OXA-10 (5 isolates with OXA-74 and one with OXA-142). The most important result of this study was the discovery of three new genes, blaGES-15, blaOXA-141 and blaOXA-142; their nucleotide sequences have been submitted to the NCBI GenBank. It is also very important to note that this is the first report on the epidemiological problem of VEB-9-producing bacterial strains, not only in Poland but also worldwide.

Introduction standard P. aeruginosa strains showed that only 12 out of 54 the Spanish laboratories performed the ESBL-detection complementary tests after routine automated susceptibility testing for PER-1 or OXA-161 producing strains, and 8 out of 54 in the case of GES-1-positive strain [19]. Consequently the correct interpretation of β-lactam resistant phenotypes of P. aeruginosa strains was obtained by only 12, 8 and 2 out of 54 laboratories in the case of strains producing PER-1, OXA-161 or GES-1 enzymes.
The masking effect of the presence of AmpC on the phenotypic detection of ESBLs and the spread of MBL-positive strains are the main causes of no detection of the ESBL-producing P. aeruginosa isolates. At the same time there is no recommendation to demonstrate the possible presence of ESBL in P. aeruginosa strains. Additionally, it is considered that there is a lack of phenotypic tests allowing for rapid and easy detection of ES-OXAs. ESBL detection tests for Enterobacteriaceae are not suitable for the detection of the all ESBL-type P. aeruginosa enzymes. This is why so little is known about the prevalence of ESBL enzymes among P. aeruginosa strains compared to Enterobacteriaceae family isolates. Particularly little is known about the occurrence of such strains in Poland. To date, only one outbreak of P. aeruginosa infection with PER-1 and OXA-74 in a Warsaw hospital has been published [20].
The aim of this study was to determine the occurrence of ESBLs in P. aeruginosa clinical isolates from patients at Warsaw hospitals and to evaluate the ESBL production of these isolates using currently available phenotypic tests, their modifications, and by multiplex PCR. Moreover, the phenotypic test results were compared with the presence of particular bla genes encoding ESBL-type enzymes. Three new nucleotide sequences of ESBLs were discovered and were submitted to GenBank at NCBI.

Bacterial isolates
Nine hundred clinical isolates of P. aeruginosa obtained from 4 Warsaw hospital microbiological laboratories were examined in this study. These strains were isolated from samples of various clinical materials derived from different patients according to the principle of one isolate from one patient. The isolates were collected in the period 2000-2014 and were initially identified as ceftazidime or/and cefepime resistant P. aeruginosa strains in microbiological laboratories by routine microbial methods, including API tests (bioMérieux). All strains were stored at -80˚C until analysis. Prior to testing, each strain was subcultured twice on TSA (bioMérieux) medium for 24 to 48 h at 30˚C to ensure viability.
During the determination of ESBL production, two different AmpC inhibitors were used: cloxacillin (Sigma) and boronic acid (purchased as phenylboronic acid from Sigma).

Determination of antibiotic susceptibility
The antibiotic susceptibility of the clinical isolates against ATM, CAZ-10, and FEP was determined on Mueller-Hinton II (MH II) agar plates (Becton Dickinson) using the disc diffusion method, according to the EUCAST guidelines [21]. The assay was validated by susceptibility testing of the reference strain P. aeruginosa ATCC 27853 against studied antimicrobial agents. The diameters of the growth inhibition zones were interpreted according to EUCAST guidelines [22].

Phenotypic detection of ESBL production
Detection of the ESBL production by P. aeruginosa strains was performed on Mueller-Hinton II (MH II) agar plates (Becton Dickinson) with and without cloxacillin (the concentration in the medium was 250 mg/L) [15] using three disc diffusion tests: 1. Double-disc synergy test (DDST) by EUCAST [15]  3. Two variants of a combination disc test (CDT) that is the extended CLSI confirmatory disc diffusion test using discs with ceftazidime or cefepime alone and discs with these agents plus clavulanate (CAZ/CL test and FEP/CL test, respectively) [15]. An isolate was recognized as ESBL-positive when a difference of at least 5 mm in the inhibition zones of CAZ-CL versus CAZ-30 or FEP-CL versus FEP was obtained [15].
All DDS tests were interpreted as positive, indicating ESBL-positive strains, when clean extension of the inhibition zone of antibiotics towards the discs with AMC, SAM, TZP or IPM was observed. In the case of an isolate where there were no visible inhibition zones around the discs with β-lactams or the result was inconclusive, boronic acid (0.4 mg per disc) [12] was added to all discs with antibiotics and the DDS tests were carried out again.

Genotypic detection of ESBL genes
The total DNA from P. aeruginosa isolates was extracted using a commercial Genomic Mini Kit (A&A Biotechnology). The specific primers used in this study are shown in Table 1. Detection of β-lactamases was carried out in three multiplex PCR reactions: the first detected bla VEB , bla GES , bla OXA-10group and bla CTX-M2group genes and the second detected bla OXA-2group , bla SHV and bla TEM genes, while the third detected bla OXA-18 , bla CTX-M-1group , bla PER , bla OXA-1group and bla BEL genes. Amplification was performed using Maxima Hot Start Taq DNA Polymerase (Thermo Scientific). A positive result in the multiplex PCR reaction was confirmed by amplification with a single primer pair. Control strains harbouring bla VEB-1 , bla GES-1 , bla PER-1 , bla BEL-1 , and bla SHV-2a (kindly provided by Prof. M.A. Toleman), bla OXA-1 , bla CTX-M2 (kindly provided by Prof. M. Gniadkowski) and bla OXA-2 and bla OXA-10 (isolated in our laboratory and confirmed by sequencing) were used. Moreover, in the isolates carrying the new genes, the presence of class 1 integrons by PCR and location analysis of bla ESBL genes within the variable part of integrons were determined.
Sequencing and nucleotide sequence analysis DNA sequencing was performed at the Laboratory of DNA Sequencing and Oligonucleotides Synthesis, Institute of Biochemistry and Biophysics, Polish Academy of Science, Warsaw, Poland.
For the preparation of DNA templates and sequencing, primers that generated the whole reading frame of the bla ESBL gene family were used. DNA and amino acid sequence analyses were carried out using the Vector NTI Advance TM 11 program (Invitrogen). The obtained sequences were compared to sequences deposited in GenBank at NCBI.

Pulsed-field gel electrophoresis (PFGE)
All isolates producing ESBL detected by phenotype tests (ceftazidime and/or cefepime-resistant isolates) were typed by PFGE according to the Seifert et al. 2005 protocol [26] with modifications, using the SpeI enzyme (Thermo Scientific). Electrophoresis was performed by using a CHEF DR II system (BioRad). The migration conditions were as follows: switch angle 120, voltage 6 V/cm, temperature 12˚C, and a two-blocks program with a total run time of 26 h (the first-block 1-20 s for 20 h and the second-block 15-40 s for an additional 6 h). The genomic DNA of Salmonella serotype Braenderup strain (H9812) was digested with XbaI (Thermo Scientific) [27] and the Lambda-DNA Ladder PFG Marker (New England BioLabs) were used as the DNA molecular-weight markers. PFGE patterns were analysed using GelCompar II software (Applied Maths, Belgium) with the Dice coefficient and clustering by UPGMA with 1% tolerance. According to the recommendation of Tenover et al. [28], the isolates were clustered in the PFGE pulsotypes (PTs).

Phenotypic detection of ESBL-type enzymes
Among the 900 studied P. aeruginosa isolates initially described as resistant to third-generation cephalosporins, in the case of 720 isolates (80%), resistance to ceftazidime or/and cefepime was confirmed by the disc diffusion method. In 180 out of 900 (20%) isolates, the inhibition zones around the discs with cephalosporins were similar to the zone diameter breakpoints. ESBL-type enzyme production was detected in 110 out of 720 isolates (15%) in at least one of the phenotypic assays. The largest number of ESBL-positive isolates was found using the DDS-SAM test (92 out of 110 isolates) and then by using CDT with ceftazidime, DDS-AMC and DDS-IMP (77, 76 and 75 isolates, respectively). The information describing the detection of ESBL-positive P. aeruginosa isolates by different phenotypic tests in relation to the resistance profiles of the studied strains is shown in Table 2. The largest group, 79 out of 110 ESBL-positive isolates, were resistant to three antibiotics, CAZ, FEP and ATM. The second group, 21 out of 110, consisted of isolates that were resistant to CAZ and FEP and at the same time sensitive to ATM. The largest percentage of ESBL-positive strains among these two groups were detected using the DDS-SAM test, 93.7% (74 out of 79) and 71.4% (15 out of 21) isolates. Interestingly, for some of the first group of strains resistant to CAZ, FEP and ATM, it was necessary to conduct tests in the presence of boronic acid, especially in the case of DDS-AMC (31.6%, 18 out of 57 ESBL-positive isolates) and CDT with cefepime (56.3%, 18 out of 32). Generally, for the majority of ESBL-positive isolates, clean extension of the inhibition zone of ceftazidime as well as cefepime towards discs with AMC, SAM, TZP or IPM was observed. However, unlike cefepim, when using discs with ceftazidime in the extended CLSI confirmatory disc diffusion test, a large percentage of ESBL-positive isolates, 82.3% (65 out of 79) in the first group and 52.4% (11 out of 21) in the second group, was observed.  Occurrence of bla genes encoding ESBL-type enzymes Among 110 phenotypically ESBL-positive isolates, 5 groups of bla genes encoding ESBLtype enzymes (VEB, GES, OXA-2, OXA-10 and TEM) were detected. Sequence analysis of all PCR bla genes detected in multiplex allowed the definition of the produced enzymes ( Table 3). The most important observation is that four P. aeruginosa isolates carried the novel bla genes. In 2 out of these 4 isolates, a novel bla GES variant close to the bla GES-5 gene was identified and submitted as bla  . The new GES-15 had one substitution, Pro167Ser, compared to GES-5, and two amino acid differences, Pro167Ser and Gly170Ser, compared to GES-1. Furthermore, the isolate carried a novel bla The occurrence of all detected bla genes encoding ESBL-type enzymes with regard to the resistance profiles of P. aeruginosa isolates is presented in Table 3. Among 110 ESBL-positive isolates in phenotypic tests, the presence of bla genes encoding ESBL was shown in 99 isolates, including bla VEB-9 alone or with bla OXA-10, as well as with bla GES-13 in 69 isolates, bla GES (bla GES-1 , bla GES-15 ) in 8, bla OXA-2 -type (bla OXA-15 , bla OXA-141 , bla OXA-210 , bla OXA-543 and bla OXA-544 ) in 16 and bla OXA-10 -type (bla OXA-74 and bla OXA-142 ) in 6 isolates. Moreover, in one isolate, the bla GES-5 gene encoding carbapenemase was detected. On the other hand, the sequencing analysis showed the presence of only the bla TEM-1f gene encoding TEM-1 β-lactamase in three isolates. In the remaining seven ESBL-positive isolates, the genes encoding enzymes of the families VEB, GES, PER, OXA, TEM, SHV, BEL and CTX-M were not detected.

Comparison of phenotypic tests results with the presence of bla genes encoding ESBL-type enzymes
In Table 4, the comparison of the results obtained from phenotypic tests and multiplex PCR detection is presented. In the case of the presence of the bla VEB-9 gene in the genomes of 69 studied isolates, the DDS-SAM test allowed for the detection of 100% of such ESBL-positive isolates. However, for one out of 10 isolates carrying only the bla VEB-9 gene, this test was positive only when the medium was supplemented with cloxacillin. Furthermore, it was necessary to use boronic acid in 3 out of 5 isolates producing additional GES-13 and OXA-10, and in 2 out of 54 isolates when only VEB-9 and OXA-10 were produced. Interestingly, the DDS-IMP test was the only phenotypic method allowing the detection of isolates producing GES-1, GES-5, OXA-74 and OXA-142, whereas the DDS-AMC test identified the strains carrying bla OXA-543. Unlike the DDS tests, the two variants of a combination disk test enabled the identification of only 78% (78 out of 100 isolates) of ESBL-producing isolates, and only 35% of CDT with cefepime (35 out of 100). Finally, the use of all three DDS tests (DDS-AMC, DDS-SAM and DDSA-IMP) allowed for phenotypic detection of 100% of studied isolates that produced ESBLs (Table 4). However, the detection of single isolates producing VEB-9, GES-1, OXA-74, OXA-543 or OXA-141 enzymes was possible due to the determination of ESBL-production by three DDS tests on the medium with cloxacillin. Carrying out these DDS tests in the presence of boronic acid, the next nine Table 4. Comparison of phenotypic test results with the presence of bla genes encoding ESBL-type enzymes.  ESBL-positive isolates (two isolates carrying bla GES-1 , two other with bla VEB-9 and bla OXA-10 , and also three carrying bla GES-13 , bla VEB-9 and bla OXA-10 ) were detected.

Molecular typing of ESBL-positive isolates
All 110 ESBL-positive clinical isolates of P. aeruginosa as determined in phenotypic tests were genotyped by PFGE. Twenty-two different pulsotypes were obtained, from which 18 with a similarity of over 50% are presented in Fig 1. Isolates were considered as a cluster if the similarity was over 80%. The PFGE molecular typing showed the relatedness between all 69 VEB-9-positive P. aeruginosa isolates obtained from different patients from one Warsaw hospital (Hospital A) during the years 2010-2014. Among these isolates belonging to PT M, two subclusters with greater than 85% similarity could be distinguished, PT Ma (isolates carried the bla VEB-9 and bla OXA-10 genes and these isolates additionally with the bla GES-13 gene) and PT Mb (carried only the bla VEB-9 gene).

Discussion
A wide panel of phenotypic methods applied in this study allowed for the detection of P. aeruginosa isolates producing various ESBL-type enzymes. Clinical isolates from patients from Warsaw hospitals carried genes encoding ESBL from the following families: VEB (69 isolates with VEB-9), GES (6 with GES-1, 5 GES-13 and 2 with GES-15), OXA-2 (12 with OXA-15, 1 OXA-141, 1 OXA-210, 1 OXA-543 and 1 with OXA-544) and OXA-10 (5 isolates with OXA-74 and one with OXA-142). The most important achievement of this study was the discovery of three new genes, named bla GES-15 , bla OXA-141 and bla OXA-142 , whose nucleotide sequences have been submitted to GenBank (accession number GU208678, EF552405 and EU358785, respectively). Among P. aeruginosa strains, the OXAs were the most common and the most frequently isolated enzymes worldwide. Besides most of the ES-OXAs have been identified in P. aeruginosa strains [2]. Therefore, it is not surprising that the same ES-OXAs were found at the same time in a different part of the world. Outside Poland, the presence of OXA-142 was also detected in P. aeruginosa strains isolated in Bulgaria in 2008-2009 (3 isolates) [10] and Korea (in 4 isolates) [3]. In the case of the Polish strain, the variable part of the class 1 integron contained the gene cassette array aac(6)-Ib-bla OXA-142 -aadA6/aadA10 (NCBI GenBank acc. no EU358785). Four isolates producing OXA-142 in Korea had a class 1 integron with the order of gene cassettes (aacA4bla OXA-142 -aadA2) identical to the previously submitted sequence of a plasmid-located class 1 integron from another Korean P. aeruginosa strain (NCBI GenBank acc. no KF257849). Our studies are the first to describe an OXA-141-producing bacteria. To date, in only one P. aeruginosa strain, isolated in the year 2000, the bla OXA-141 gene encoding a new ESBL enzyme was observed. Conting 67 of the whole genome shotgun sequence of P. aeruginosa strain WH-SGI-V-07247, which also contains this gene (accession no. LLTH01000109), has been recently submitted to GenBank. Our work is also the first in which two additional new ESBL-type enzymes from the OXA-2 family, OXA-543 and OXA-544, were found in P. aeruginosa clinical strains. Unfortunately, only the nucleotide sequences of contings from whole genome shotgun sequences of P. aeruginosa strains carrying these genes were submitted to GenBank, without specifying the range of activity of these new enzymes (acc. no LLTZ01000057 and LLMY01000078, respectively). Our research showed that P. aeruginosa isolates carrying these two new enzymes, OXA-543 and OXA-544, have demonstrated typical results for ESBL-positive strains in phenotypic tests. Moreover, among the studied isolates, the presence of genes encoding the OXA-15, OXA-74 and OXA-210 enzymes was demonstrated. The OXA-15 β-lactamase was identified inside a class 1 integron detected in P. aeruginosa strain from Turkey [29]. The presence of OXA-74 in the P. aeruginosa clinical Phenotypic and genotypic methods of detection of ESBL-producing Pseudomonas aeruginosa isolates isolates was described for the first time in 2007 in Poland [20], and then only in 2008 in Hungary (2 isolates) [30] and in 2013 in Bulgaria (2 isolates) [10], whereas the presence of the gene encoding OXA-210 was demonstrated so far in only one P. aeruginosa strain isolated in Korea [3]. So, this research is the second reported case of isolating an OXA-210-producing strain in another part of the world-Poland. An equally important observation of this study is the first report of the epidemiological problem with VEB-9-producing bacterial strains not only in Poland but also worldwide. The PFGE molecular typing showed the relatedness between all 69 VEB-9-positive P. aeruginosa isolates obtained from different patients from one Warsaw hospital during 2010-2014. Only 5 of 59 isolates grouped in the subcluster PT Ma with greater than 85% similarity also carried the bla GES-13 gene. Perhaps there has been an acquisition of the bla GES-13 gene by strains originally possessing only bla VEB-9 and bla OXA-10 genes. At first, the VEB-9 enzyme was observed in two P. aeruginosa isolates from Kuwait in 2001 [31]. To date, there are no other publications reporting the detection of VEB-9-producing strains worldwide. Recently, the whole genome shotgun sequences of three strains, P. aeruginosa AR_0092 scaffold00001 (acc. no. MPBS01000001), Acinetobacter baumannii ABBL 142 contig-1000047 (acc. no LLIK01000044) and K. pneumoniae OC511 AOT23_contig000137 (acc. no LOEI01000136), which also contain the bla VEB-9 gene, were submitted to GenBank. Among the P. aeruginosa clinical worldwide isolates, the VEB-1 enzyme was detected the most frequently [2,6,10,32].
Other important data obtained in this research include a large variety of enzymes from the GES family (GES-1, -5, -13 and GES-15) detected in P. aeruginosa isolates from such a small area as the city of Warsaw. Unlike the other detected GES-type enzymes, GES-5 possesses carbapenemase activity [33,34]. GES-1 is the most common of the ESBL-type enzymes of the GES group produced by P. aeruginosa [2,9] as well as by Enterobacteriaceae strains [16] in the world. The presence of GES-13-producing P. aeruginosa strains was described only twice, for the first time in the case of a MDR strain isolated in 2008 in Greece [35] and the second time in a P. aeruginosa strain isolated in Canada in 2012 [36]. In Polish isolates, the presence of bla GES-13 as well as bla VEB-9 and bla OXA-10 genes in the same isolate is unusual. This work is the first report of GES-15-producing bacteria in the world. The bla GES-15 gene was detected in one P. aeruginosa strain within the variable part of class 1 integron located in the plasmid.
The detection of the various ESBL-type of enzymes in P. aeruginosa isolates from patients in Warsaw hospitals was possible due to the use of a wide panel of phenotypic methods. We modified and extended a double-disk synergy test, mainly by the incorporation of other clavulanic acid β-lactamase inhibitors, such as sulbactam, tazobactam and imipenem. The ESBLtype enzymes produced by P. aeruginosa strains are not always inhibited by clavulanic acid [5,18]. The most difficult was to detect the isolates harbouring the genes encoding the GEStype [18,36] or OXA-type [5] of ESBLs. As previously described, synergy between imipenem and ceftazidime was observed in the case of GES-1-producing P. aeruginosa strains in the discdiffusion method [18]. Among ES-OXAs, for example, OXA-19 can be inhibited by imipenem [5]. In addition, tazobactam was a more potent inhibitor of GES-type ESBLs (GES-1, GES-13) than other inhibitors, such as clavulanic acid, imipenem and sulbactam [34,35]. Sulbactam was only sporadically used in the CDT test to detect the ESBL-positive co-produced AmpC strains [37]. According to the EUCAST recommendations [15], for Enterobacteriaceae with inducible chromosomal AmpC, for the phenotypic detection of ESBL in this work, cefepime and the agar plates containing cloxacillin were included. It described that cefepime is less rapidly inactivated by AmpC cephalosporinases than by ESBLs [38] or even that it is stable to AmpC hydrolysis [15]. Performing the DDST on agar supplemented with cloxacillin strengthened the ability of this test to detect bacterial strains carrying genes encoding ESBLs among P. aeruginosa isolates [17,23,38]. The greatest effect improving the detection of the ESBL-positive isolates in this work was the reduction in the distance between the discs in DDS tests. In the case of P. aeruginosa isolates for which the diameter zones of inhibition around the discs was not observed or was very small, the distance between the discs in DDST was reduced from 20 to 15 mm. Generally, performing conventional DDST and the above-described modifications, in 105 of the 110 ESBL-positive P. aeruginosa strains, the ability to produce these types of enzymes was observed. However, it was necessary to reduce the distance between discs with antibiotics to 15 mm in the case of 56 out of 105 ESBL-positive isolates. However, the phenotypic detection of two isolates carrying the bla GES-1 gene, two isolates carrying bla VEB-9 and bla OXA-10 genes and three isolates carrying bla GES-13 , bla VEB-9 and bla OXA-10 genes required not only a reduction in the distance between discs but also the use of boronic acid. Previously, the DDST carried out with the addition of 0.4 mg of boronic acid per disc allowed the detection of BEL-1-positive P. aeruginosa strains in Belgian hospitals [12]. The data obtained in this study showed that boronic acid has a greater ability than cloxacillin to inhibit cephalosporinase overproduction.
Finally, comparing the occurrence of bla genes encoding ESBLs in relation to the results obtained in this research concerning ESBL detection by phenotypic tests, it is concluded that in the case of P. aeruginosa, modified DDS tests should be used instead of tests dedicated for Enterobacteriaceae strains. Based on the data obtained in this study, we propose to use three methods (DDS-AMC, DDS-SAM and DDS-IMP) to detect ESBL-producing P. aeruginosa strains. Depending on the images of the plates, we suggest a reduction in the distance between discs with antibiotics to 15 mm and the addition of boronic acid to the discs. This study is also a good example to demonstrate the effectiveness of the proposed methods in the detection of the various ESBLs, including enzymes from the VEB, GES, OXA-2 and OXA-10 families, which allowed us to determine the occurrence of ESBLs in P. aeruginosa isolates from clinical materials from patients in Warsaw hospitals.