Escherichia coli from community-acquired urinary tract infections resistant to fluoroquinolones and extended-spectrum beta-lactams

Background: Uropathogenic Escherichia coli are increasingly becoming resistant to flouroquinolones and to other commonly available antimicrobials. We sought to investigate the genetic basis for fluoroquinolone and extended spectrum beta-lactam (ESBL) resistance in 17 fluoroquinolone-resistant (MIC of levofloxacin and ciprofloxacin >32 μg/ml) E. coli isolated from patients with urinary tract infections (UTIs). Methods: We applied PCR and Pulsed Field Gel Electrophoresis (PFGE) to characterize resistance genes and to determine clonal relatedness of strains, respectively. Results: Twelve of the 17 E. coli were resistant to multiple drugs, including ampicillin, co-amoxyclav, cefotaxime, ceftriaxone, ceftazidime and gentamicin and nalidixic acid and produced plasmid-mediated CTX-M-15 type ESBLs and CMY-2 AmpC type enzymes. The other 5 E. coli that were non-ESBL-producing were multiply resistant to ampicillin, nitrofurantoin, cefoxitin, nalidixic acid. Resistance to fluoroquinolones resulted from a combination of the presence of qnrA, qnrB, ciprofloxacin acetylating enzyme designated aac(6’)-1b-cr, and mutations in the two amino acid substitutions; 83 Serine (TCG) to Leucine (TTG) and 87 Aspartic acid (GAC) to Asparagine (AAC). Conclusion: Antibiogram patterns and PFGE of E. coli showed that these were community acquired UTI caused by pockets of clonally-related and some discreet strain types. Plasmid-mediated CTX-M-15 beta-lactamases and CMY-2 AmpC enzymes and fluoroquinolone resistant E. coli are becoming increasingly prevalent in hospitals in Kenya, posing a major challenge in the management of UTIs.


Introduction
Urinary tract infections (UTIs) including cystitis and pyelonephritis are common both in community and hospital settings [1][2][3] with Escherichia coli being the predominant pathogen [4].However, other organisms gain a greater foothold in patients with complicated UTI [5].Although uncomplicated UTIs caused by E. coli and other Enterobacteriaceae should be treated empirically with co-trimoxazole [6,7] the rates of communityacquired strains with decreasing susceptibility to first-line agents such as ampicillin, nitrofurantoin, and co-trimoxazole have been on the increase.More recently resistance to fluoroquinolones such as ciprofloxacin and levofloxacin has been on the increase [1,[8][9][10].It is therefore important that antibiotic resistance profiles at the local level must be known in order to support an empiric approach to the management of these infections.We present data on antimicrobial susceptibility and the genetic basis of resistance in multidrug resistant E. coli isolates from community-acquired UTIs in patients attending two hospitals in Nairobi, Kenya.

Patients
Patients with symptoms of urinary tract infection (UTI) were studied between June 2004 and August 2005 at The Aga Khan University Hospital, Kenya.During this period the respective laboratories isolated a total of 64 multidrugresistant (MDR) E. coli isolates (resistant to at least ampicillin and cotrimoxazole) representing 36% of all E. coli isolations from routine urine cultures.A total of 17 of the MDR E. coli were initially resistant to ciprofloxacin and levofloxacin by routine disk diffusion methods and these isolates were selected for further study.They were isolated from 15 adults and two children (a 6month-old baby boy and 12-year-old girl) with UTIs.Eleven of the adults and the 2 children were treated as outpatients; one adult each was admitted to the surgical and medical wards, respectively.All identifying clinical information was removed from the clinic data base and permission to further test the E. coli strains after isolation during routine clinical care was obtained from the KEMRI/National Ethical Review Board.

Antibiotic susceptibility testing
Initially all isolates were tested for susceptibility to ampicillin, co-amoxyclav, ceftazidime, cefotaxime, ceftriaxone, co-trimoxazole, gentamicin, cefoxitin, cefepime, nalidixic acid, ciprofloxacin, imipenem, meropenem, tetracycline and nitrofurantoin (Oxoid Ltd., Basingstoke, United Kingdom) using E-Test MIC strips (E-test; AB BioDisk, Solna, Sweden) on Mueller-Hinton agar (Oxoid) as described by the Clinical Laboratory Standards Institute [11].In addition, testing of susceptibility to cefotaxime, cefepime and to ceftazidime alone (E-test; AB BioDisk) and in combination with clavulanic acid (E-test for ESBLs; AB BioDisk) was performed on donors and E. coli K-12 transconjugants.Screening to detect phenotypic production of AmpC β-lactamases was done by direct susceptibility testing with a cefoxitin 30-µg disk and the three-dimensional enzyme extract test [12].Escherichia coli ATCC® 25922 and ATCC® 35218 were included in each test to control for growth and potency in tests for MICs and for β-lactam/β-lactamase inhibitor combinations, respectively.

Pulsed-field gel electrophoresis
Initially E. coli cultures were grown overnight in Nutrient broth (Oxoid), and the bacterial cells were then embedded in agarose plugs and chromosomal DNA obtained by digestion with Lysozyme (25 mg/mL) (Sigma, Poole, United Kingdom).Chromosomal DNA in agarose plugs was digested using 35 U of XbaI and 25 U of SpeI (Life Technologies, Paisley, United Kingdom) using a previously developed protocol [13].PFGE of agarose plug inserts was then performed on a CHEF-DR II system (Bio-Rad Laboratories, Richmond, Calif.) on a horizontal 1% agarose gel for 22 h at 120 V, with a pulse time of 1 to 40 s at 14°C.A lambda DNA digest consisting of a ladder (ca.22 fragments) of increasing size from 48 kb to approximately 1,000 kb was included as a DNA size standard.The gel was stained with ethidium bromide and photographed on a UV transilluminator (UVP Inc., San Gabriel, Calif.).The restriction endonuclease (RE) digest patterns were compared, and their similarities were scored by the method of Tenover et al. (14).By these criteria isolates that gave PFGE banding patterns that were indistinguishable were assumed to be from a single outbreak strain.Isolates that gave banding patterns showing differences in fewer than four bands were assumed to be closely related, as they may represent isolates differing by a single genetic event.In addition, isolates with differences in four to six bands may also be part of an outbreak.However, isolates showing a difference in more than seven bands of their banding patterns may represent more than three genetic events, in which case they were considered epidemiologically unrelated.

In-vitro conjugation and plasmid analysis
Each of the 17 MDR E. coli were conjugated with an azide resistant strain of E. coli J53, according to a previously described procedure [12] and plated on Mueller-Hinton agar containing ampicillin (32 µg/ml) and Sodium azide (200 µg/ml).The putative transconjugants were tested for susceptibility to all 15 antimicrobial agents by a disk-diffusion method to identify transferable resistance.Plasmid DNA extraction from both donor and transconjugants was performed using a Plasmid Mini Prep Kit (Qiagen Ltd., West Sussex, United Kingdom) according to manufacturer's instructions.

Extraction of DNA template for PCR
Cells from 1.5 ml of the overnight culture were harvested by centrifugation at 13,000 × g for 5 min.After the supernatant was decanted, the pellet was resuspended in 5% Chelex-100 resin slurry (Bio-Rad) in injection grade water followed by boiling for 10 min.The supernatant, 2 µl (1/250 volume) of the total sample, was used as the source of template DNA for amplification.

Sequence determination
Sequence determination was performed using the PCR primers for both strands of the amplicons with a dideoxy-chain termination method using an automated DNA sequencer ABI PRISM 377 (Perkin-Elmer, Warrington, United Kingdom) and was analyzed using commercial software (Lasergene; DNAStar Inc., Madison, Wis.).

Results
Twelve out of the 17 E. coli isolates were resistant to ampicillin, ceftazidime, cefotaxime, ceftriaxone, tetracycline, nalidixic acid (MICs >256 µg/ml), gentamicin (MIC 24-32 µg/ml), ciprofloxacin and levofloxacin (MICs >32 µg/ml).They showed intermediate resistance to coamoxyclav (MIC= 8-16 µg/ml) and cefoxitin (MIC 12-16 µg/ml) but were fully resistant to nitrofurantoin (MIC >128 µg/ml).Using ESBL Etest strips, these 12 E. coli also expressed cefotaximase and cefepimase activities, with correspondingly high MICs of the respective cephalosporins (MIC > 256 µg/ml).The 5 other E. coli that were non-ESBL producers were fully sensitive to gentamicin, cefotaxime, ceftriaxone, ceftazidime and cefepime, and one isolate to cotrimoxazole, but they were resistant to ampicillin, co-amoxiclav, nalidixic acid, ciprofloxacin, levofloxacin, nitrofurantoin and cefoxitin.Imipenem and meropenem were the only antimicrobials to which each of the 17  were 3-5 band differences between them.These isolates were from 3 different outpatient clinics and were obtained over a one-year period, and hence were likely to represent distinct pockets of community-acquired clonally-related strain types.In contrast all 6 isolates from Nairobi Hospital produced different PFGE digest patterns indicating that these were isolates from distinct and unrelated sources of infection in the respective communities.All the 12 ESBL-producing E. coli contained a 42 kb plasmid in addition to two smaller plasmids of 4 and 2.5 kb.The non-ESBL producers all contained a 3 kb plasmid.In vitro conjugation tests showed that the 42 kb plasmid co-transferred resistance to ampicillin, cefotaxime, ceftriaxone, ceftazidime, cefepime, co-trimoxazole and tetracycline.Each of the 17 MDR E. coli isolates contained two amino acid substitutions in the QRDR, i.e. 83 Serine (TCG) to Leucine (TTG) and 87 Aspartic acid (GAC) to Asparagine (AAC), and two silent mutations in amino acids 156 Proline (CCC to CCG) and 182 Iso-leucine (ATC to ATT).In addition, all the MDR E. coli contained the aac(6')-1b-cr2 gene while 6/17 isolates also carried the qnrA and qnrB genes.The 5 cefoxitin resistant non-ESBL producing isolates were also positive for the AmpC gene, with amplicons of 462 bp, indicating that the genes probably arose from Citrobacter freundii.These isolates all carried a 3kb plasmid which was not transferred to E. coli J53 in vitro.The AmpC sequence showed high identity to CMY-2 but to none of the other AmpC genes tested.These 5 E. coli isolates were from patients seen at 2 separate outpatient clinics and one surgical ward at the Aga Khan Hospital.They are likely to be clonal as their XbaI and SpeI PFGE patterns were indistinguishable.PCR amplification of the entire coding sequence of the blaCTX-M gene revealed complete sequence identity for each of the 12 cefotaxime-resistant E. coli as well as from their E. coli J53 transconjugants with CTX-M-15.

Discussion
Worldwide E. coli is the major cause of UTIs and accounts for 75% to 90% of UTI isolates [2,17].
The increasing prevalence of infections caused by antibiotic-resistant bacteria makes the empirical treatment of UTIs difficult and outcome unpredictable [6].
In the current study we examined 64 MDR E. coli -which represent 36% of all UTI isolates from two hospitals in Nairobi.A total of 17 E. coli were also resistant to fluoroquinolones and 12 of these were also resistant to each of the four extended spectrum beta-lactams, cefotaxime, ceftriaxone, ceftazidime and cefepime that form the mainstay of reserve antibiotics for treatment of severe infections in these hospitals.In Kenyan private clinics and hospitals, ciprofloxacin is the most frequently prescribed fluoroquinolone for UTIs because of its availability in oral and intravenous formulations.In poor-resource settings where the availability of alternative effective antibiotics is limited, serious problems are anticipated in the treatment of multidrug-and fluoroquinolone-resistant E. coli.Many studies worldwide have also reported a sharp increase in ciprofloxacin resistant E. coli isolates from UTIs.For example, in China, from 1998 to 2002, the prevalence of ciprofloxacin resistance has increased steadily from 46.6% to 59.4% [6], and in Bangladesh the prevalence was 26% [8].Although some studies have identified plasmid-mediated fluoroquinolone resistance in clinical bacterial isolates that showed reduced susceptibility to fluoroquinolones [12,16] it appears that full resistance to fluoroquinolones (MICs > 32 ug/ml) in our E. coli isolates was due to point mutations in the QRDR, namely 83 Serine (TCG) to Leucine (TTG) and 87 Aspartic acid (GAC) to Asparagine (AAC), and qnrA, qnrB and aac(6')-1b-cr2 genes.However, we cannot exclude the possibility that the presence of the qnrS gene might have also contributed to the mutations in the QRDR.Each of the ESBL producing E. coli contained a plasmid mediated cefotaximehydrolysing CTX-M-15 enzyme encoded on a 42 kb plasmid which was also associated with transferable resistance to ampicillin, cefotaxime, ceftriaxone, ceftazidime, cefepime, co-trimoxazole and tetracycline.To our knowledge this is the first report of CTX-M-15 enzyme to be found in E. coli isolates from the region.Previously we identified a novel CTX-M-12 enzyme in a nosocomial isolate of Klebsiella pneumoniae from a local hospital [18].Although we were unable to categorically determine prior usage of fluoroquinolones by individual patients, these antibiotics are the current drug of choice for treatment of UTIs and it is plausible that patients who had previous episodes of UTIs may already have been exposed to these antibiotics.Other studies in USA [19] and Tunisia [20] have also identified urinary E. coli that produced the plasmid-mediated AmpC enzyme beta-lactamase CMY-2 and CTX-M15 ESBLs, and these were resistant to ceftazidime, ceftriaxone, and cefepime but sensitive to meropenem.In Taiwan, E. coli isolates recovered from food animal feces, retail ground meats, and urinary E. coli isolates from outpatients were shown to carry a CMY-2 -lactamase [21].
The PFGE patterns of E. coli isolates from Aga Khan Hospital indicate that UTI infections in the outpatient group were likely to be from pockets of a clonally-related community-acquired strain while those that caused UTIs at the Nairobi Hospital may have originated from different sources.As these isolates were collected over a period of one year and were mainly from outpatients, it is plausible that the UTIs may have several sources of infection in the community as even isolates from inpatients showed distinct PFGE patterns.Previous studies in Canada [22] and USA [23] have also described the occurrence of several E. coli clonal groups in a community suggesting that a proportion of community-acquired UTIs may be caused by E. coli disseminated from one or more point sources.
In conclusion, antimicrobial resistance patterns and PFGE showed that these were community acquired E. coli UTIs caused by pockets of both clonally related and some discreet strains.These findings also show that plasmid-mediated CTX-M-15 and CMY-2 AmpC -lactamase, and fluoroquinolone resistance among urinary E. coli are prevalent, posing a major challenge in the management of community-acquired UTIs.

Figure 1 .
Figure 1.PFGE patterns of 11 XbaI-digested cefotaxime and ciprofloxacin resistant E. coli from cases of urinary tract infections attending a hospital in Nairobi.Lanes 1-11 are PFGE patterns of E. coli from Aga Khan Hospital, while lanes 12-17 shows PFGE patterns of E. coli from Nairobi Hospital.Lane M, 50 kb molecular size standard; Lanes 1 to 7, E. coli from main PFGE pattern 1 (all from outpatients); Lanes 9 to11, E. coli from PFGE pattern 2 (one from outpatient, one from surgical ward and 1 from medical ward); Lanes 12-17 E. coli from Nairobi Hospital (from outpatients) showing 5 distinct PFGE patterns.

Table 1 .
Primers used for PCR amplification and sequencing of the CMY gene