Abstract
After the Elizabethkingia anophelis outbreak in Wisconsin, USA, an active search for the detection of the microorganism in hospital water systems from Central and Northern Greece was performed from June to December 2016. In total, 457 water samples from 11 hospitals were analyzed. Elizabethkingia spp. was detected in three samples collected from two hospitals, both of which are located in Northern Greece. Two of the three isolated strains were identified as Elizabethkingia anophelis. No cases of Elizabethkingia infection were reported in either hospital during 2016. This is the first reported isolation of the pathogen in water supply systems in Greece.
INTRODUCTION
Members of the genus Elizabethkingia are Gram-negative, non-motile, non-fermenting, aerobic bacteria. The genus belongs to the phylum Bacteroidetes and the family Flavobacteriacae and initially, comprised four species: Elizabethkingia meningoseptica, Elizabethkingia anophelis, Elizabethkingia miricola, and Elizabethkingia endophytica (Kim et al. 2005; Kämpfer et al. 2015). Both E. meningoseptica and E. miricola were previously classified to the genus Chryseobacterium (Vandamme et al. 1994; Kim et al. 2005). Recently, using whole genome sequencing (WGS) E. endophytica was proved to be a subjective synonym of E. anophelis (Doijad et al. 2016). Elizabethkingia is a microorganism with currently vague pathophysiology, transmission, and reservoir (Centers for Disease Control and Prevention 2016). All species of the genus are commonly found in the environment (soil, water, and plants). In particular, E. anophelis is abundant in the midgut of the mosquito Anopheles gambiae (Lau et al. 2016). The microorganism may colonize hospital environment, is highly persistent to decontamination measures, thus contaminating medical solutions and devices (Moore et al. 2016). Recent studies have proposed that hospital water supply systems possibly act as a reservoir, being responsible for long-term transmission of the microorganism in the hospital environment (Lau et al. 2015; Moore et al. 2016).
E. meningoseptica mainly causes healthcare-associated infections in immunocompromised patients as well as neonatal meningitis and sepsis (Lau et al. 2015, 2016; Tai et al. 2016). Infections caused by E. meningoseptica are often very severe, displaying high death rates. The existing co-morbidities and immunosuppression of these patients in combination with the multidrug-resistant profile of the microorganism (Lau et al. 2016) contribute to the fatal outcome of the infection.
E. anophelis has been widely known since the outbreak in Wisconsin, USA, that was attributed to the microorganism, which accounted for 67 cases and 18 deaths from 01.11.2015 to 11.01.2017 (Centers for Disease Control and Prevention 2016; Wisconsin Department of Health Services 2017). E. anophelis usually causes pneumonia and bacteremia with high rates of mortality (Lau et al. 2016). Recently published data indicate that the outbreak was caused by a single, highly pathogenic strain with 13 characteristic genomic regions (Perrin et al. 2017). Consequently, Elizabethkingia can be considered an emerging opportunistic pathogen and although the majority of the strains are primarily identified as E. meningoseptica, they are in fact E. anophelis, as 16S rRNA gene sequencing and WGS indicates (Breurec et al. 2016; Nicholson et al. 2016; Eriksen et al. 2017).
To the best of our knowledge until presently, the study of Arvanitidou et al. (2003) is the only one supporting the presence of the species in hemodialysis water and dialysate in renal units in Greece. The study denoted the presence of Chryseobacterium meningosepticum (presently E. meningoseptica) in hemodialysis water and dialysate at a frequency of 14.9%.
Urged by the outbreak of E. anophelis in Wisconsin, USA, the source and transmission of which is still unknown (Centers for Disease Control and Prevention 2016; Navon et al. 2016), and by lately published research that traced E. anophelis in the hospital water supply system (Breurec et al. 2016), we decided to analyze all water samples that we received from hospitals from June to December 2016 for the presence of Elizabethkingia spp. This is the first report for the presence of E. anophelis in hospital water supply systems in Greece.
METHODS
A total of 457 tap water samples from 11 hospitals of Northern and Central Greece were analyzed at the Laboratory of Hygiene and Epidemiology of the University of Thessaly, Greece. All samples were collected without disinfecting the tap before sample collection. Water samples were collected and transported to the laboratory according to the national standards that are based on ISO 5667-5. A total of 100 mL of water was filtered by using 0.45 μm filter membranes (EZ-Pak Filters 0.45 μm, 47 mm white gridded, Merck, Millipore), and the membranes were incubated on nutrient and MacConkey agar, for each sample, in ambient air at 37°C for 48 hours. Samples where Gram-negative, oxidase-positive (Bactident-Oxidase strips, Merck), catalase-positive colonies were isolated were subcultured onto nutrient agar and identification was conducted using matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF MS) (Moore et al. 2016). In total, 243 isolates were identified using MALDI-TOF MS.
The measurements were performed with MALDI Microflex LT (Bruker Daltonics, Bremen, Germany). Protein profiles were acquired using linear-positive mode analysis with laser frequency at 20 Hz. Raw spectra were automatically acquired with AutoXecute control software (Flex control 3.4; Bruker Daltonics, Bremen, Germany) and were recorded within the range of 2,000 to 20,000 Da. The identification was performed using the MALDI BIOTYPER Software, version 3.1, with default parameters and the acquired spectra were compared with the mass-spectrum library (6.093 MSPs). The method was calibrated using the Bruker Bacterial Test Standard (BTS), a manufactured extract of Escherichia coli DH5 alpha spiked with two additional proteins (RNAase A and myoglobin) that extend the upper boundary of the mass range covered by BTS. Isolates identified as either Elizabethkingia spp. or Chryseobacterium spp. were kept frozen in glycerol stock solutions at −80°C until further use in the study.
The laboratory characteristics and biochemical profile of the isolates were examined after subculture on nutrient agar and 48 h incubation at 36°C.
To validate the initial identification of the strains by the standard MALDI Biotyper library, the 16S rRNA gene was amplified and sequenced. The bacterial isolates were cultured on nutrient agar and were incubated for 48 h at 36°C. Total DNA was extracted manually using the QIAamp DNA Mini kit (Qiagen). Total nucleic acid extracts (5 μL) were used for 16S rRNA gene amplification with Platinum Taq DNA Polymerase (Invitrogen) in a validated Eppendorf Mastercycler Gradient System. Master mixes (45 μL) with primer concentrations of 10 nmol/μL were prepared and amplified at 94°C for 5 min, 34 cycles at 94°C for 30 sec, 55°C for 30 sec, and 72°C for 30 sec, followed by 72°C for 10 min. The primers used were 533F (5′-GTG CCA GCA GCC GCG GTA A-3′) and P1033R (5′-TGC GGG ACT TAA CCC AAC A-3′). Amplicons were separated, purified, and cycle sequenced with 533F and P1033R primers. Sequences were analyzed on ABI 3730XL genetic analyzer (Applied Biosystems) and were compared to known gene sequences in GenBank and in BLASTSearch of MicrobeNet CDC Reference Laboratory. Furthermore, we submitted the acquired spectra to the mass spectrometry tool of MicrobeNet CDC Reference Laboratory using the Bruker-CDC merged library (7,737 MSPs).
Finally, the Bruker library was expanded by adding the standard mass spectra (MSPs) of the 16S rRNA sequenced strains to the existing database and an MSP dendrogram of the strains was constructed. Twenty-four raw spectra from the 16S rRNA sequenced strains were acquired using the protein extraction protocol according to the manufacturer's instructions. Each of the 24 raw spectra was submitted to baseline subtraction and smoothing procedure and afterwards processed using the MALDI Biotyper Offline Classification 3.1 software with the default parameters for MSP creation. The created MSPs were added to the existing database. The MSP dendrogram was constructed using the MALDI Biotyper Offline Classification 3.1 software with the default parameters for MSP dendrogram construction.
The antimicrobial susceptibility of the strains was tested by using MIC test strips (Liofilchem Diagnostics MIC Test Strip) following the recommendations of the European Committee on Antimicrobial Susceptibility Testing (EUCAST), version 7.1 (Breurec et al. 2016; Lau et al. 2016; Eriksen et al. 2017; Perrin et al. 2017).
RESULTS
Only three out of 243 isolates were identified as Elizabethkingia spp. by MALDI-TOF. This corresponds to three samples out of 457, collected from two hospitals out of eleven. The first isolate was classified as E. meningoseptica with a score value of 2.093, the second, originating from the same hospital was identified as E. miricola with a score value of 2.306, while the third strain was also identified as E. meningoseptica and demonstrated a score value of 2.394. The concentrations of the microorganism in the water samples were 2 cfu/100 mL, 16 cfu/100 mL, and 5,120 cfu/100 mL, respectively. Unfortunately, the two Elizabethkingia strains from the first hospital were not available for further analysis following their identification by MALDI BIOTYPER.
Subculture of the third strain onto nutrient agar gave white-yellow, shiny with entire edges colonies after 48 h of incubation at 36°C. No growth of the third strain on MacConkey agar was observed after 48 h incubation at 36°C. The results of the biochemical tests performed are demonstrated in Table 1.
Biochemical characteristics of Elizabethkingia isolate
Characteristics . | Elizabethkingia strain . |
---|---|
Indole production | + |
H2S production | − |
Citrate utilization | − |
Malonate utilization | − |
Acid production from: | |
Arabinose | − |
Sucrose | − |
Salicin | − |
Lactose | + |
Cellobiose | + |
Mannitol | + |
Trehalose | + |
Growth on MacConkey agar | − |
Hydrolysis of urea | − |
ONPG | + |
Characteristics . | Elizabethkingia strain . |
---|---|
Indole production | + |
H2S production | − |
Citrate utilization | − |
Malonate utilization | − |
Acid production from: | |
Arabinose | − |
Sucrose | − |
Salicin | − |
Lactose | + |
Cellobiose | + |
Mannitol | + |
Trehalose | + |
Growth on MacConkey agar | − |
Hydrolysis of urea | − |
ONPG | + |
16S rRNA gene sequence identified the isolate as E. anophelis (99% nucleotide identity to E. anophelis type strain R26, GenBank accession number MF615392 and 99.08% nucleotide identity with CSID_3015183678_outbreak strain after sequence analysis to CDC MicrobeNet BLASTSearch). E. anophelis is not included in the current Bruker reference library.
Submission of the peak list files of the isolate to the mass spectrometry tool of MicrobeNet CDC Reference Laboratory confirmed our results with identification score of the 16S rRNA sequenced strain 2.432 as E. anophelis. Using the protein profile of the 16S rRNA sequenced strain we expanded the custom MALDI-TOF database with the addition of E. anophelis MSP. Using the MALDI Biotyper Offline Classification 3.1 software, we repeated the identification of the saved protein profiles of the three isolates. It was shown that two of them were misidentified as E. meningoseptica, whereas the new classification with the expanded database correctly identified them as E. anophelis with score values of 2.358 and 2.816, respectively (Figure 1(a)). The E. miricola strain was correctly identified by MALDI-TOF. The MSP dendrogram of Elizabethkingia isolates clearly showed that there are two branches, one for E. anophelis strains and one for E. miricola (Figure 1(b)).
Results of MALDI-TOF MS identification of the three Elizabethkingia isolates. (a) MALDI-TOF MS spectra of two Elizabethkingia species (E. anopheles and E. miricola) are shown. (b) Dendrogram generated from hierarchical clustering of MALDI-TOF MS spectra of the three Elizabethkingia isolates using ClinProTools 3.0 (Bruker Daltonics, Germany). Distances are displayed in relative units.
Results of MALDI-TOF MS identification of the three Elizabethkingia isolates. (a) MALDI-TOF MS spectra of two Elizabethkingia species (E. anopheles and E. miricola) are shown. (b) Dendrogram generated from hierarchical clustering of MALDI-TOF MS spectra of the three Elizabethkingia isolates using ClinProTools 3.0 (Bruker Daltonics, Germany). Distances are displayed in relative units.
Antibiotic susceptibility testing of the E. anophelis strain showed that the strain was susceptible to ciprofloxacin, trimethoprime-sulfamethoxazole, piperacillin-tazobactam, and tigecycline, while resistant to amoxicillin, amoxicillin/clavulanic acid, ceftazidime, meropenem, imipenem, gentamicin, amikacin, tombamycin, and colistin. The isolate demonstrated intermediate in vitro susceptibility to vancomycin. The antibiotic susceptibility profile was consistent with the usual profile of the microorganism. Results are demonstrated in Table 2.
Antimicrobial susceptibilities of Elizabethkingia isolate determined by MIC strips
Antimicrobial agent . | MIC (μg/mL) . | Interpretation . |
---|---|---|
Ciprofloxacin | 0.25 | S |
Amicacin | >256 | R |
Amoxicillin | >256 | R |
Trimethoprim-sulfamethoxazole | 0.25 | S |
Gentamicin | >256 | R |
Aztreonam | >64 | R |
Ceftazidime | >256 | R |
Tobramycin | >256 | R |
Amoxicillin-clavulanic acid | >256 | R |
Piperacillin-Tazobactam | 8 | S |
Meropenem | >32 | R |
Imipenem | >32 | R |
Colistin | >256 | R |
Tigercyclin | 0.25 | S |
Vancomycin | 4 | I |
Antimicrobial agent . | MIC (μg/mL) . | Interpretation . |
---|---|---|
Ciprofloxacin | 0.25 | S |
Amicacin | >256 | R |
Amoxicillin | >256 | R |
Trimethoprim-sulfamethoxazole | 0.25 | S |
Gentamicin | >256 | R |
Aztreonam | >64 | R |
Ceftazidime | >256 | R |
Tobramycin | >256 | R |
Amoxicillin-clavulanic acid | >256 | R |
Piperacillin-Tazobactam | 8 | S |
Meropenem | >32 | R |
Imipenem | >32 | R |
Colistin | >256 | R |
Tigercyclin | 0.25 | S |
Vancomycin | 4 | I |
Submitted isolate to Genbank (accession no. MF615392). The results were interpreted according to the recommendations of the European Committee on Antimicrobial Susceptibility Testing, version 7.1. for Staphylococcus spp. (vancomycin) and Pseudomonas aeruginosa (other antibiotics except for co-trimoxazole in which Stenotrophomonas maltophilia breakpoints were used), since there are no criteria for Elizabethkingia.
DISCUSSION
These are the first reported strains of E. anophelis isolated in hospital water systems, not only in Greece, but in Europe also, since despite our extensive research of the current literature, we did not find any other studies supporting the presence of the bacterium in hospital water supply systems. A probable explanation could be that E. anophelis is a new species, and isolates which were previously identified as E. meningoseptica could, in fact, be E. anophelis (Breurec et al. 2016; Lau et al. 2016; Janda & Lopez 2017). Although only one strain was used in our study in order to construct the E. anophelis MSP, the offline classification of the strains demonstrated satisfactory score values (2.358 and 2.816, respectively) and was in full concordance with the MicrobeNet CDC Mass Spectrometry classification. We will continue the research in order to enrich the reference database with more E. anophelis strains.
Even though our findings are consistent with those of researchers who suggested 16S rRNA gene sequencing as the gold standard method for the identification of Elizabethkingia spp. at the moment, since E. anophelis is not included in the Bruker MALDI-TOF reference library (Lau et al. 2015; Breurec et al. 2016; Eriksen et al. 2017), it should be taken into serious consideration that recent research indicates that Elizabethkingia spp. are very similar, when using 16S rRNA sequencing and there are still uncertainties regarding Elizabethkingia taxonomy (Doijad et al. 2016; Eriksen et al. 2017). Advanced molecular identification techniques are required for definite species identification (Janda & Lopez 2017). Nevertheless, the addition of E. anophelis to the MALDI-TOF MSP reference library will contribute to the rapid and reliable identification of this specific strain.
Recent studies have suggested that the hospital environment acts as a reservoir for the microorganism (Breurec et al. 2016), and that the microorganism persists in the hospital water microbiome despite the control measures (Moore et al. 2016). Although the strains detected three months apart in the water system of the same hospital belonged to two different Elizabethkingia species, our results could be considered supportive of previous findings.
In both hospitals, Elizabethkingia spp. infections were not recorded, but two cases of Elizabethkingia-like species (Chryseobacterium indologenes) infections were recorded in the first hospital during 2016. One possible reason could be that in Greece, Elizabethkingia infections are probably underdiagnosed, since automated systems or biochemical tests, which are the most prevalent used methods for identification in Greek hospitals, may misidentify the microorganism (Centers for Disease Control and Prevention 2016). Further studies for the virulence of the strains isolated should be performed since strains that could lead to an outbreak seem to have several characteristic pathogenic genomic regions (Perrin et al. 2017). Since there are no recorded cases of Elizabethkingia infections in Greece, the Greek medical community should be aware about the clinical significance of the bacterium in order to be alert.
CONCLUSIONS
Given the severity of the infection in hospitalized patients (Lau et al. 2016), the multidrug-resistant profile (Lau et al. 2016), the persistence of the microorganism in the environment (Breurec et al. 2016), and the unknown way of transmission and pathogenesis, we should consider that E. anophelis may be a cause of nosocomial infection in Greece. In that respect, infectious diseases specialists, microbiologists, and epidemiologists should actively search for the pathogen both in clinical and environmental samples.
ACKNOWLEDGEMENTS
The authors wish to thank the Public Health personnel of the hospitals for supporting the collection of the samples. The research did not receive any grants from funding agencies in the public, commercial or no profit sector.