Journal of Antimicrobial Chemotherapy (1992) 30, 429-447

Survey of the prevalence of pMactamases amongst 1000 Gram-negative bacilli isolated consecutively at the Royal London Hospital P. Y. F. Lint, Deniz Gar}, Ludnda M. C. Hall and D. M. Iirennore* Department of Medical Microbiology, The London Hospital Medical College, Turner Street, London El 2AD, UK /J-Lactamase expression was examined in 1000 consecutive Gram-negative bacilli isolated from urine, wound swab, sputum or blood specimens received at the Microbiology Laboratory of the Royal London Hospital. This survey, performed between January and April, 1991, followed a similar study undertaken in early 1982. The distribution of species was similar in the two surveys, except that the proportion of Pseudomonas aeruginosa isolates had increased from 11% in 1982 to 17-5% in the present study. This increase was balanced by a decreased proportion of enterobacteria. Amongst plasmid-mediated 0-lactamases, TEM-1 (especially), TEM-2, SHV-l and OXA types continued to predominate in enterobacteria. Their frequency in Escherichia coli was unchanged (46% in 1991 compared with 43% in 1982), but had increased from 5 to 22% amongst Proteus mirabilis isolates. An apparent decrease in their frequency amongst Enterobacter cloacae isolates, from 48% in 1982 to 17% in 1991, probably reflected changes to strain prevalence rather than enzyme prevalence. Plasmid type /?-lactamases were present in fewer than 2% of P. aeruginosa isolates in both surveys. In the present study, chromosomal /J-lactamase derepression (constitutive hyperproduction) was detected in 10/76 isolates of E. cloacae, Enterobacter aerogenes, Citrobacter freundii, Serratia spp. and Morganella morganii, and in 2/170 P. aeruginosa isolates. These proportions were increased, compared with those seen the 1982 survey, though the significance was borderline (P a 0O5; j? test). Extended-spectrum plasmid mediated /Mactamases, unknown in 1982, were found in 11/70 Klebsiellae pneumoniae isolates in the present study. Ten of these organisms, representing at least five distinct strains, produced TEM-10 enzyme, encoded by a plasmid of c. 90 kb; the remaining organism had an unidentified SHV-derived enzyme.

Introduction

0-Lactamase production is the commonest mechanism of bacterial resistance to /Mactam antibiotics. Chromosomal /Mactamases are ubiquitous amongst Gram-negative bacteria, but their ability to give resistance depends on their amount and mode of expression (Livermore, 1987). In addition, plasmids that determine further /Mactamases have been reported in Gram-negative bacteria since the mid-1960s, with over 50 types now recognized (Medeiros, 1984; Jacoby & Medeiros, 1991). Matthew (1979)reviewedthe distribution of the plasmid-mediated /Mactamases then known, and 'Corresponding author. Present addresser t Department of Infectious Disease*, Takhung Veterans Hospital, Taichung, Taiwan, ROC; {Section o f infectious Diseases, Department of Internal Medicine, Hacettepe University School of Medicine, Ankara 06100, Turkey. 0305-7453/92/100429+ 19 J08.00/0

429 © 1992 The British Society for Antimicrobial Chemotherapy

430

P. Y. F.UaetaL

found that the TEM-1 type was the most frequent. This observation was confirmed by subsequent surveys at the Royal London Hospital (RLH, then the London Hospital) (Whitaker, Hajipieris & Williams, 1983) and elsewhere in the UK (Simpson, Harper & O'Callaghan 1980), as well as in the Netherlands (Stobberingh, Houben & Van Boven, 1982), Spain (Roy et al., 1983, 1985), Germany, Amman, Mexico and Colombia (Simpson et al., 1986). Other common types include TEM-2, SHV-1, OXA-1 and, in Pseudomonas aeruginosa, PSE-1 and -4. Many other plasmid-mediated /Mactamase types have been found in only a few isolates. In the period since these surveys were undertaken, the use of second- and thirdgeneration cephalosporins (particularly cefuroxime in the UK), monobactams and carbapenems has increased relative to that of anti-Gram-negative penicillins, thereby exerting selection pressure for new types of resistance. Many new plasmid-mediated /Mactamases have been discovered, most notably 'extended-spectrum' variants of TEM- and SHV-enzymes. Unlike their parent enzymes, these confer resistance to newer-generation cephalosporins, though not to cephamycins or carbapenems (Philippon, Labia & Jacoby, 1989; Jacoby & Medeiros, 1991). Mutants that produce high levels of Class I chromosomal /Mactamases constitutively have also emerged, particularly in Enterobacter cloacae, but also in Citrobacter, Pseudomonas, Morganella and Serratia spp. (Livermore, 1987; Sanders & Sanders, 1987). Such mutants are resistant to most newer cephalosporins and monobactams, but not to carbapenems. The 'escape' of Class I /Mactamase genes to plasmids, long suggested (Bobrowski et al., 1976), has recently been confirmed unequivocally (Papanicolaou, Medeiros & Jacoby, 1990). The present study, performed nine years after the previous survey of /Mactamase production by Gram-negative bacilli isolated at the RLH, aimed to re-assess the prevalence of established and newer forms of /Mactamase-mediated resistance amongst these organisms. Methods and materials

Bacterial isolates and their identification One thousand Gram-negative bacilli were collected from consecutive clinical specimens of urine, blood, sputum and tissue fluids received at the Microbiology Laboratory of the RLH between January and April 1991. Multiple isolates of the same species from the same patient were excluded. Oxidase-negative isolates were identified by the API-20E System (BioMerieux, La Balme Les Grottes, France). Oxidase-positive isolates were tested for pyocyanin production on Pseudomonas Agar-P (Difco Laboratories, Detroit, USA). Isolates producing this pigment were identified as P. aeruginosa. Pyocyanin-negative, oxidase-positive isolates were tested by the API-20NE system (BioM6rieux), as were those identified by API-20E tests as belonging to non-fermentative genera. Reference strains producing known /Mactamase types were from our strain collections, or were kindly provided by Glaxo Group Research (Greenford, UK). Antibiotics Antibiotic discs were obtained from Oxoid (Basingstoke, UK) and included: ampicillin (25/ig), co-amoxiclav (20 fig amoxycillin + 10 /ig clavulanate), cefotaxime (30 ^g),

p-Lactamases in Gram-aegatire bacilli

431

ceftazidime (30 fig), cefoxitin (30 /ig), ccfuroximc (30 /ig), imipcnem (10 /ig), ticarcillin (75 /ig), ticarcillin/clavulanate (75+ 10 /ig), tcmocillin (30 /ig) and pipcracillin (75 /ig). Antibiotic powders were provided as follows: ampicillin sodium, carbenicillin disodium and clavulanate lithium from SmithKline Beecham (Brockham Park, UK); benzylpenicillin sodium, cephaloridine and ceftazidimc from Glaxo Group Research; cefotaxime from Roussel Laboratories (Wembley, UK); ceftriaxone from Roche (Welwyn Garden City, UK); imipcnem and cefoxitin from Merck, Sharp & Dohme (Hoddesdon, UK); piperacillin sodium and tazobactam from Lederle (Gosport, UK) and sulbactam from Pfizer (Sandwich, UK). Nitrocefin was purchased from BBL Laboratories (Cockeysville, USA). Antibiotic solutions were prepared on the day of use in 10 DIM phosphate buffer, pH 70. Susceptibility tests Disc diffusion tests were performed by the Differential Inoculum Method (Working Party of the British Society for Antimicrobial Chemotherapy, 1991). Organisms were grown overnight on Nutrient Agar (Oxoid), then resuspended in peptone water 1%. The turbidity of the suspensions was determined at 450 nm and then adjusted with reference to a dilution schedule (Working Party of the British Society for Antimicrobial Chemotherapy, 1991), such that semi-confluent growth was obtained following inoculation of 0-1 mL of the dilution on to IsoSensitest agar (Oxoid; CM471). Antibiotic discs were applied before incubation for 18 h in air at 37°C. The zones of inhibition were measured to ±0-1 mm with calipers. MICs were determined on IsoSensitest agar containing doubling dilutions of antibiotics. The inocula contained 104 cfu, taken from overnight Nutrient Broth (Oxoid) cultures. Detection and electrofocusing of fl-lactamases

Organisms were grown overnight, with continuous shaking, at 37°C in 10 mL volumes of Nutrient Broth (Oxoid; CM1). One millilitre amounts were then added to 10 mL volumes of fresh warm (37°C) broth. Incubation was continued for 4 h to allow the cultures to reach exponential phase. Subsequently, the cells were harvested at 6000 g and 37°C, resuspended in 1 mL of 10 mM phosphate buffer, pH 70, and then disrupted by sonication (MSE Ultrasonic Disintegrator; MSE, Crawley, UK). Debris was removed by centrifugation for 15 min at 10,000 g. The supernatants were then subjected to isoelectric focusing at 10 W for 60-90 min on polyacrylamide gels, as described by Matthew et al. (1975). Extracts were screened first on gels containing broad pH range (3-5-9-5) ampholytes (Ampholine; LKB Biotechnology, Bromma, Sweden), then re-run on narrow range gels (pH 5-0-8-0 or 4-0-6-5). The gels were stained for /J-lactamase activity with nitrocefin. Identification of enzymes was by comparison, on the same gel, with known standards (including TEM-1, -2, -6, -9 and TEM-10, SHV-1, OXA-1-7, PSE-1-4). Extraction of extended-spectrum fl-lactamases from ceftazidime-resistant Klebsiella pneumoniae isolates for fi-lactamase assays Cultures were grown overnight at 37°C, with shaking, in Nutrient Broth No. 2 (Oxoid), then diluted into ten-fold larger volumes of the same broth, pre-warmed to 37°C. After incubation for 4 h, the cells were harvested at 6000 g and 37°C, washed once in 10 mM

432

P. Y. F. Lin et aL

phosphate buffer, pH 6-2, and rcsuspended in 10 mL of the same buffer. /?-Lactamases were released by sonication and separated from debris at 100,000 g. The centrifugation supernatants were subjected to cation-exchange chromatography on a column (1-6 cm diameter x 40 cm high) of Carboxymethyl-Sephadex C-50 (Pharmacia, Uppsala, Sweden) which had been equilibriated in 10 mM phosphate buffer, pH 6-2. Elution was with the same buffer containing a linear (0-10 M) NaCl gradient. Enzymes with pi values s$ 6-2 eluted with the loading buffer, those with higher pi values eluted with the gradient. fl-Lactamase assays Enzyme activity was assayed by UV spectrophotometry in 10 mM phosphate buffer, pH 70, at 37°C. The wavelengths used were: benzylpenicillin, ampicillin and carbenicillin, 235 nm; cephaloridine, 295 nm; cefoxitin, 260 nm; cefotaxime, ceftriaxone and ceftazidime, 257 nm; and imipenem, 297 nm. Relative hydrolysis rates of each antibiotic (0-5 mM) were determined with benzylpenicillin as the reference substrate. The light path was 1 mm or 1 cm, as appropriate. Since imipenem is chemically unstable, its breakdown in the absence of added enzyme was subtracted from the hydrolysis rate in the presence of enzyme. Typing of cephalosporin-resistant klebsiellae For serological typing, the bacteria were grown overnight at 37°C on Worfel-Ferguson agar (Ewing, 1986), and then suspended in sterile saline, 0-9%. These suspensions were reacted in countercurrent immunoelectrophoresis experiments with antisera raised against 77 distinct capsular (K) antigens in rabbits (Palfreyman, 1978). The capsular swelling-Quellung test was performed for isolates that reacted with more than one antiserum (Orskov & Orskov, 1984). For phage typing, the bacteria were grown in Tryptone Soya Broth (Oxoid). The surfaces of Phage Typing Agar plates (Gaston, Ayling-Smith & Pitt, 1987) were flooded with these cultures. The plates were dried and a set of 15 phages (Ayling-Smith & Pitt, 1990) was applied with a multiloop applicator. Lytic reactions were recorded after overnight incubation at 32°C. Genetic studies on cephalosporin-resistant klebsiellae Transfer of ceftazidime resistance from Klebsiella pneumoniae isolates to Escherichia coli K12 J62-1 (lac, pro, his, trp, nal1) (Coetzee, Datta & Hedges, 1972) was attempted by a standard plate mating procedure. Briefly, logarithmic phase cultures of the donor and recipient strains were harvested from 10 mL volumes of nutrient broth culture. The pellets were resuspended in 1 mL of the same broth, mixed together, and spread on nutrient agar plates. These were incubated overnight at 37°C, then washed with 5 mL of saline 0.9%. The washings were streaked on to DST agar (Oxoid) plates containing ceftazidime 25 mg/L plus nalidixic acid 100 mg/L. Transconjugant colonies were picked off after 18 h at 37°C, confirmed as lactose-negative on MacConkey agar (Oxoid), and examined for 0-lactamase production by isoelectric focusing. DNA extraction and probing Plasmid DNA was extracted by the method of Kado & Liu (1981). Total DNA extraction was based on the method of Pitcher, Saunders & Owen (1989), modified as

P-Lactamases in Gram-negative badffi

433

described previously (Hall et al., 1992). Restriction endonuclease digestion of DNA was with Hincll or BamHl (Promega, Southampton, UK), and was performed with the buffers and conditions advocated by the supplier. DNA fragments were separated by electrophoresis in agarose 0-9% in TBE buffer (0-9 M Tris-borate, 0-004 M EDTA), as described by Sambrook, Fritsch & Maniatis (1989). Southern blots were made with a Posiblot Apparatus (Stratagene, Cambridge, UK) on to Hybond-N filters (Amersham International, Amersham, UK). The TEM probe consisted of a 493 bp restriction fragment from pUC19 (Seetulsingh, Hall & Livermore, 1991). It was labelled with a DIG DNA Labelling Kit (Boehringer, Lewes, UK). The SHV probe consisted of an oligonucleotide corresponding to positions 712-736 in the sequence reported by Mercier & Levesque (1990). The sequence is conserved in SHV types and OHIO-1, but is absent from TEM and other known /Mactamase sequences (Swiss-Prot database, August 1991; PC-Gene Software supplied by Intelligenetics Inc., California, USA). The oligonucleotide was labelled with the DIG Oligonucleotide 3'-End Labelling Kit (Boehringer). Probes were hybridized to Southern blots at 65°C, and washed to a maximum stringency of 0-2xSSC, 0-1% SDS and 65°C by standard procedures (Sambrook et al., 1989). Hybridization was detected with the DIG Non-Radioactive Detection Kit (Boehringer), used in accordance with the manufacturer's directions. Results Strain collection

Of the 1000 isolates collected, 971 survived and were identified. These comprised: E. coli (419), K. pneumoniae (sensu latu) (70), Klebsiella oxytoca (27), Citrobacter freundii (8), Citrobacter diversus (17), E. cloacae (36), Enterobacter aerogenes (11), other Enterobacter spp. (4), Proteus mirabilis (113), Morganella morganii (12), Proteus vulgaris (5), Proteus penneri (2), Serratia marcesens (5), Serratia liquefaciens (4), other Serratia spp. (1), Salmonella spp. (4), Shigella sonnei (1), Hafnia alvei (1), Kluvia spp. (1) Cedecea spp. (1), P. aeruginosa (170), other identifiable Pseudomonas spp. (5), Xanthomonas maltophilia (5), Acinetobacter calcoaceticus var. anitratus (22), Acinetobacter calcoaceticus var. Iwoffi (12), Achromobacter spp. (7), Alkaligenes spp. (2) Flavobacterium spp. (3), Moraxella spp. (3). Susceptibility and fl-lactamase production

/7-Lactamase production was examined, by electrofocusing of extracts, for all the isolates that gave zones of ^ 20 mm diameter in response to ampicillin 25 /ig discs or, in the case of P. aeruginosa, where the zone given by a ticarcillin 75 fig disc was < 23 mm. Additionally, extracts of all the klebsiellae were subjected to electrofocusing, since these organisms were almost always (96/97 cases) more susceptible to co-amoxiclav than to ampicillin alone (see below), suggesting /Mactamase function even when substantial zones ( > 20 mm diameter) to ampicillin discs were observed. Greater susceptibility to co-amoxiclav than to ampicillin, also was seen for some 'ampicillin susceptible' (zone > 20 mm) C. diversus isolates, but these were not subjected to electrofocusing as their behaviour was thought to reflect the function of the clavulanate-inhibited chromosomal /Mactamase typical of this species (Amicosante et al., 1987). Except for these organisms, and some Acinetobacter spp., greater susceptibility

434

P. Y. F.UuetaL

to co-amoxiclav than to ampicillin was the prerogative of ampicillin-resistant organisms (zones < 20 mm). /?-Lactamase identification was primarily by isoelectric focusing but the identifica-. tions (Table I) were accepted only when the resistance pattern conformed to that typical of the enzyme type. Otherwise, prominent enzyme bands are listed in the 'uncertain identification' column of this table. Excellent agreement was noted between electrofocusing and antibiogram data for the great majority of isolates with plasmid type /Mactamases (see below), but chromosomal /Mactamases presented a greater problem. Such enzymes, identified by their high pi values (generally > 8) and failure to align with known plasmid-mediated types in electrofocusing, are indicated in Table I only when they gave a particularly strong reaction on the focusing gel and when their presence was associated with resistance to cephalosporins (see below). Trace levels of /Mactamases that were inferred to be chromosomal were noted in many other isolates, but were not associated with resistance (see Discussion). The relationships between the /Mactamase types detected and the inhibition zone diameters observed for major groups of organisms are shown in Tables II and III. Ampicillin resistance in E. coli and P. mirabilis was, almost always, associated with plasmid-mediated /Mactamase types. TEM-1 was, by far, the commonest enzyme type amongst E. coli isolates, whereas TEM-1 and TEM-2 were almost equally frequent in P. mirabilis isolates. TEM-2, SHV-1 and OX A /Mactamases were found in a few E. coli isolates, being observed either alone or in combination with TEM-1 enzyme. Production of any of these enzymes in these species was associated with ampicillin resistance, without cross-resistance to cefotaxime, ceftazidime, cefoxitin, cefuroxime and imipenem (Table II). The inhibition zones given by co-amoxiclav discs for the TEM- SHV- and OXA-producers generally were smaller than those for non-producers, although some overlap occurred (Table II). Two E. coli and one P. mirabilis contained high pi /Mactamases ( > 80) that failed to align with known plasmid-mediated /Mactamase types. The two E. coli isolates gave reduced zones to all the /Mactams tested except imipenem (Table II). This resistance pattern conforms to that associated normally with high-level production of a Class I chromosomal /Mactamase (Livermore, 1987), and this mechanism of resistance was inferred. The identity of the high pi enzyme in the P. mirabilis isolate, which was resistant only to ampicillin, remains doubtful. Klebsiellae presented a more complex picture than E. coli and P. mirabilis. Except for one K. pneumoniae, all the isolates belonging to this genus, even those that gave zones of > 20 mm to ampicillin discs, exhibited substantially larger zones to discs containing co-amoxiclav than those containing ampicillin. Electrofocusing revealed /Mactamases, identified as TEM, SHV, OXA or chromosomal types, in all the isolates apart from the one that was as susceptible to ampicillin as to co-amoxiclav (Table II). Combinations of TEM and SHV enzymes predominated amongst the most ampicillinresistant isolates, whilst those that gave substantial zones to ampicillin discs ( > 20 mm) generally only had /Mactamases that aligned with SHV-1. It seems likely that the lack of clear resistance to ampicillin in these latter isolates reflected low-level /?lactamase production, but this hypothesis remains to be confirmed by quantitative assay. Most of the K. pneumoniae isolates remained fully susceptible to newer cephalosporins (Table II), suggesting that their TEM or SHV enzymes corresponded to the classical forms of these /Mactamases. However, ten K. pneumoniae isolates had /?lactamases that focused almost identically to the SHV-1 and TEM-2 standards, (pis 7-6

15

6 2 9

172

10

TEM-2

E. coli K. pnewnoniae K. oxytoca C. freundii E. cloacae A. calcoaceticus var. anitratus

2 42 8

PSE-1

10

1 (pi 80)

1 (Pi 7-8)

2 (pi > 9 ) 5 (pi 6-7 and 7-8)

Uncertain I/D

1 4 2

TEM-2

OXA-1

SHV-1

Chromosomal hyperprodcuction +

Chromosomal hyperproduction

TEM-1

PSE-2

OXA-3 + pi 81

Number of isolates with: SHV-l+pI56

12

Number of isolates with: SHV-1 OXA-1 OXA-3

TEM-1 + SHV-1 TEM-1 + OXA-1

B. Isolates in which multiple enzyme types were detected

E.coli K. pnewnoniae K. oxytoca C. freundii C. diversus E. cloacae E. aerogenes Other Enterobacter spp. P. mirabilis M. morganii P. vulgarls Other Proteus spp. Serratia spp. Other fermenters P. aeruginosa X. maltophilia A. calcoaceticus var. anitratus A. calcoaceticus var. hvoffii Other non-fermenters

TEM-1

TaMe I. Distribution of /Mactamases in the isolates examined A. Isolates where only a single enzyme type was identified

Notes a-c are explained on Table III.

E. coli TEM, OXA, SHV + (19I) Chr+ + + »(2) ampicillin susceptible' (226) K. pneumoniae TEM, OXA, SHV + (57) SHV-l+pI5-6(10) none detected (1) K. oxytoca TEM, OXA, SHV + (15) Chr+ + + ±TEM, OXA, or SHV (4) P. mirabilis TEM+ (25) ampicillin susceptible (87) C. diversus TEM+ (10) none detected (7)

Species and enzyme(s) (no. of isolates)

27-913-9 18111-8 35-0 30-311-9 9-8-14-7 25-015-8 32-312-4 29-811-6 31-312-4

151 ±5-9 5-5 ± 0 35-0 15-5 ±3-9 7-5 ±3-8 33-4 ±2-3 12-3 ±2-7 18-7±5-8

5-5

20013-6 10-5-23-9 26-712-5

co-amoxiclav

5-5±0-2 5-5-10-9 27-5 ±2-7

ampicillin

41-3111 43-812-1

42-812-8 43-5130

41-611-3 29-9-38-0

40-1131 27-612-8 43-6

38-312-9 30-1-31-0 38-412-8

31-5111 33112-5

28-812-0 29-512-4

39-4121 39-413-0 38-813-6 37-612-8

31-312-1 29-1-29-9

29-212-2 27113-6 29-7

29411-9 16-7-20-1 29-412-1

38-412-0 35-3-38-9

35-612-8 5-510 39-5

35-212-6 251-304 35112-3

Zone diameter (mm)* cefotaxime ceftazidime cefoxitin

29-3131 30-212-3 37-612-5 39-6120

26-411-2 29-5111

350121 32-9-37-9

33-713-0 33-011-3 33-9

33-812-3 33-7-341 33-612-4

imipenem

31-711-7 32-312-5

29-411-8 7-5-14-9

28-8131 22-512-5 30-9

26013-0 15-8-17-4 26-712-6

cefuroxime

Table II. Inhibition zones for isolates of E. coli. klebsiellae, C. diversus and P. mirabilis, grouped according to /Mactamases detected

a

F

y «;

amptcillin

co-amoxidav cefotaxime

5-5 16-1 3l-6±fr7

A. ealeoaerllats var. hroffi TEM-2onry(l) Chr + + + (l) unclassified/none detected (10) 21-9 32-8 ±5-1

32-8 ±5-1

12-2-17-5 1*3 2O6±4-3

11-0

13-7-17-8 17-5 24-5 ±4-3

32-9 ±4-3

201 11-5 27O±54

ll-0±5-l

2«±4-l 21-9 2O9

5-5-8-7

6-7 ±3-5 18-3 ±9-8 16-1 ±7-3

221 1O6 28-0±fr6

5-5-11-6 14-1 1M±6O

5-8 ±O9 2I-9±1O6 2O8±6-2

Zone diameter* cefoxitin cefuroxfane

2O0-21-0 22-5

27-3 16-7-27-8 33-3 ±2-3

l2-3±7-O 37-4 ±1-7 36-2 ±3-2

ceftazidune

'Mean zone diameters±standard deviation are shown lor groups of > 5 organisms, otherwise the range in indicated. 'Chromosomal 0-lactamase hyperproduction. 'Zone > 20 mm to an impiaTlin 25 pg disc

5-5-13-5 5-5 19-1 ±5-2

A. ealeooclellcHS var. anllralus Chr + + +±TEM-I(2) TEM-I (1) unclassified/none detected (18)

none detected (167)

P. otntginosa PSE-I (1)

£. cloocxtt, E. atrogenet, C.frrundU, M. morgaidl and Serratia spp. 8-3±5-9 105±frl Chr + + + ±*TEM, OXA, SHV (10) 5-5±0 5-5±0 12-6±frO 40O±l2 TEM/SHV/OXA + (5) l9O±7-6 l4-4±7-2 38-3 ±3-7 none detected (61)

Species and enzyme(s) (no. of isolates)

33H) 32-6 42-2 ±5-1

32-7 ±32-3 35H) 38-4±4-l

36-8 28-7-31-4 3l-7±>3

32-6 ±5O 31-3±l-2 32-8 ±4-6

5-5 2>9 36-8±3-4

5-5-25* 5-5

5-5 5-5-154 26-6±4-l

81 ±5-2 103±ll-7

nnipcncm Ucardltin

21-5 5-5

5-5±9* 81 IO9±6-3

3O4±3-4

252 260 39-1 ±3-6

I7-4±25O I9-2-2O4 3O3±3-9

101 14-5-16-0 28-0±3-9

29-2±4-8 34-9±3-6

tkarcinra/ davulanate

TaUe HI. Inhibition zones for isolates of more resistant genera of cnterobacteria and non-fermenters, grouped according to /Mactamase* detected

5-5 17-0-24-7 33 25 mm for all other K. oxytoca isolates), without any cross-insusceptibility to ceftazidime, cefoxitin or imipenem (Table II). These isolates had only slightly reduced susceptibility to cefotaxime. Electrofocusing showed that they had strongly reactive /Mactamases of pi 6-5-6-7; additionally, three of these isolates had SHV-1, TEM-2 or OXA-1 enzymes. It seems likely the pi 6-5-6-7 enzymes were the Class IV (Kl) chromosomal /Mactamase typical of K. oxytoca, but which is expressed normally at a low level. Hyperproduction of this enzyme has been associated with resistance to cefuroxime and aztreonam, but not other newer /Mactams (Arakawa et al., 1989). Only TEM-1 and TEM-2 enzymes were found amongst the 17 C. diversus isolates, being present in ten of these organisms. Once again, isolates with these enzymes had increased resistance to ampicillin, but not to the other /Mactams tested (Table II). Six of the seven isolates without TEM enzymes were less susceptible to ampicillin than to co-amoxiclav, supporting the view that typically, and unlike C.freundii, C. diversus produces a clavulanate-inhibited chromosomal /Mactamase (Amicosante et al., 1987). SHV, TEM or OXA /Mactamases were detected in only three of seven C. freundii isolates, six of 36 E. cloacae, one of ten Serratia spp., and in none of the 11 E. aerogenes and 12 M. morganii isolates (Table I). Five isolates of these species had such enzymes without showing, from electrofocusing or antibiogram, any suggestion of concomitant chromosomal /Mactamase hypcrproduction ('stable derepression'). These isolates were highly-resistant to ticarcillin and ampicillin, less so ticarcillin-clavulanate, but remained as susceptible as non-producers to cefotaxime and ceftazidime (Table III). This pattern agrees well with the recognized hydrolytic activity of TEM and SHV enzymes, supporting the /Mactamase identifications. Six of 36 E. cloacae, two of 11 E. aerogenes, one of seven C.freundii and one of 12 M. morganii isolates gave zones of < 20 mm to discs containing cefotaxime, compared with mean zones of c. 40 mm for these species. Ceftazidime, cefuroxime, ticarcillin and ticarcillin-clavulanate inhibition zones for these isolates also were reduced compared with those for typical isolates (Table III). These ten organisms all showed prominent high pi ( > 8-0) /Mactamases on electrofocusing and were inferred to be hyperproducers of chromosomal Class I /Mactamases, which are well-known to have high pi values and to confer this resistance pattern (Livermore, 1987). In addition, five of the ten organisms had TEM-1 /Mactamase. All the chromosomal /Mactamase hyperproducers, whether or not they additionally had TEM-1 enzyme, remained fully-susceptible to imipenem and had only slightly reduced susceptibility to temocillin (zones generally c. 5 mm smaller than those typical for these species). High pi chromosomal type /Mactamases were detected also in many other Enterobacter, Citrobacter and Morganella isolates by isoelectric focusing, but the reactions observed were slower than

P-Lactamases in Gram-negatire badllj

439

for the resistant isolates. Detection of these enzymes was not associated with increased resistance and is not detailed in Table I. Of the 170 P. aeruginosa isolates collected, 26 gave zones of < 23 mm to discs containing 75 //g of ticarcillin. Extracts of these isolates were examined by electrofocusing (Table III). Only one had a recognized plasmid /Mactamase type, namely PSE-1. This organism gave small inhibition zones in response to discs containing ticarcillin/clavulanate and piperacillin, but retained full susceptibility to imipenem. Its susceptibility to ceftazidime was reduced slightly compared with that of typical isolates. Chromosomal type, high pi, /Mactamases were detected in several of the other isolates on which electrofocusing was performed, and the two isolates that gave the strongest reactions had relatively small zones to discs containing piperacillin, ceftazidime and ticarcillin/clavulanate, though remaining fully susceptible to imipenem. This behaviour, coupled with their antibiogram, suggested chromosomal /Mactamase derepression. Inhibition zone distributions for Acinetobacter spp. were amongst the most complex and variable observed, with wide ranges and little clear relation to the detection of /Mactamases (Table III). Even those A. calcodceticus var. anitratus isolates that gave substantial zones to ampicillin discs ( > 20 mm) commonly gave 4-5 mm larger zones to discs containing co-amoxiclav, suggesting /Mactamase function. However, electrofocusing resolved clear /Mactamase bands in only a few of the isolates. The difficulties of focusing the chromosomal enzymes of this genus have been recognized previously (Hood & Amyes, 1991), and may explain the present observations. Alternatively, the co-amoxiclav results may reflect the inherent susceptibility of many isolates of Acinetobacter spp. to clavulanate. TEM-1 /Mactamase was found in three isolates of A. calcoaceticus var. anitratus, two of which yielded additional prominent high pi enzymes, which were inferred to be chromosomally-mediated. AU three isolates gave smaller zones to ticarcillin and ampicillin and, less so, to ticarcillin/clavulanate and co-amoxiclav, than did isolates without TEM-1 enzyme. The two organisms with the high pi enzymes had reduced susceptibility to cephalosporins and imipenem, compared with isolates in which no, or only trace levels of, such enzymes were detected. An A. calcoaceticus var. Iwoffi isolate in which a high pi enzyme was detected had reduced susceptibility, compared with typical isolates, to all /Mactams, including imipenem. So, more surprisingly, did one in which only a TEM-2 enzyme was found. In general, where no specific /Mactamase was detected, A. calcoaceticus var. anitratus isolates were less susceptible than A. calcoaceticus var. Iwoffi isolates to all the /Mactams tested except imipenem. Additional studies on resistant klebsiellae As noted earlier, ten isolates of K. pneumoniae were resistant to ceftazidime and, less so, to cefotaxime. These had /Mactamases with pis of 5-6 and 7-6. One further cephalosporin resistant isolate (No. 886) had enzymes with pis of 81 and 71 MIC determinations confirmed the resistance pattern indicated by the disc tests (Table IV). The organisms with the pi 5-6 and 7-6 enzymes had been isolated from patients on at least five different wards over a two-month period, and represented at least five different strains, as judged from serotyping data (Table IV). The enzymes from one representative of each of these distinct strains were extracted and separated by ion exchange chromatography. In every case, activity against third-generation cephalosporins was associated with the pi 5-6 enzyme (Table V), whereas the pi 7-6 /Macta-

32/36 32/36 32/36 32/36 NT NT 27/46 NT NT 22/67 12/13

•erotrpe NT" NT NT NT 4,48 " 4,48 NT NT NT 44.1 1

> > > > > > > > > > >

1024 1024 1024 1024 1024 1024 1024 1024 1024 1024 1024 > > > > > > > > > > >

1024 1024 1024 1024 1024 1024 1024 1024 1024 1024 1024

phige-type arooirdllin ampicillin > > > > > > > > > >

128 128 128 128 128 128 128 128 128 128 32 2 4

2

ccftaztdtme cefotaxime

*Piperacillin +tazobactam, 4 mg/L. *Coamoxiclav, tested with a fixed clavulanate concentration of 4 mg/L. r Amptcillin + sulbactam, 4 mg/L. 'NT, Not typeable.

369 104 372 138 309 266 415 802 360 803 886

Itotate 4 4 4 4 2 2 2 4 2 2 4 8 8 4 4 4 4 4 4 4 8 8

MIC (mg/L) ceflriuone ceAiroxime

i

I

8 4 4

oefoxltin i 0-25 0-5 0-5 0-5 0-23 0-5 0-25 023 023 05

inupcncro

256 236 256 256 256 > 512 236 256 256 256 32

4 2 2 8 2 2 2 2 2 8 2

16 16 16 16 16 32 16 16 16 16 2

piperacUlln pip"+ taxo vnox*+ day

TaMe IV. MIC» of intibtotia from K. pnewnoniat isolates with extended-tpectnim /Mactamasa

32 8 8 16 8 > 1024 16 16 8 16 8

«mp*+ nil

132 149 145 145 129 142 134 134

ampicillin

40 53 63 50 42 15 58 20

carbenicillin

121 129 133 115 122 87 98 123

cephaloridine

27 22 18 23 30 < 0-1 35 78

8-3 8-2 6-7 6-9 7-2 < 01 6-2 28

Relative rate of hydrolysis' of: ceftriaxone cefotaxime

'Relative to benzylpenicillin, at a substrate concentration of 0-S mM, in phosphate buffer, pH 70, at 37°C.

369 (pi 5-6) 309 (pi 5-6) 415 (pi 5-6) 802 (pi 5-6) 803 (pi 5-6) T E M - I (control) T E M - 1 0 (control) 886 (pi 8 1)

Isolate and enzyme

81 86 80 95 127 < 0-1 134 35

ceftazidime

cefoxitin

Table V. Substrate profiles of extended-spectrum /Mactamases from ceftazidime-resistant K. pneumonia* isolates

imipencm

442

P. Y. F. Lta et aL

mase had activity typical of SHV-1 enzyme (not shown). The pi 5-6 enzymes from the different strains were identical in hydrolytic activity (Table V), and closely resembled the TEM-10 enzyme from K. pneumoniae isolates described by Quinn et al. (1989). Agarose gel electrophoresis detected at least one plasmid in each of the isolates with the pi 5-6 and 7-6 enzymes. A plasmid of c. 90 kb was present in all ten isolates, and hybridized with a TEM gene probe. One isolate, No. 226, additionally had a larger plasmid that hybridized with the TEM probe. Total DNA was extracted from the ten isolates and digested with HincW. When a Southern blot of the digest was hybridized with the TEM probe, a band of 1-65 kb was detected in all cases, and an additional band of 0-65 kb detected in extracts from Isolate 226. Isolate 226 was much more resistant than the others to ampicillin/sulbactam (Table IV), and it seems likely that it had another form of TEM enzyme (perhaps TEM-2) as well as the putative TEM-10, and that these activities were not separated in electrofocusing. Purification of the enzyme activities present in this organism was not, however, undertaken. Southern blots of total DNA from the ten isolates were also hybridized with the SHV probe and, in all cases, a 3-7 kb BamHI band was detected. Ceftazidime-resistant E. coli K12 J62-1 transconjugants were obtained in matings with each of the ten isolates. MICs of ceftazidime for the transconjugants were 128 mg/L, compared with 0-12 mg/L for the plasmid-free recipient. The transconjugants acquired the pi 5-6 enzyme, but not the pi 7-6 enzyme, and hybridized with the TEM probe, but not the SHV probe. They acquired a plasmid of c. 90 kb which co-migrated with that found in the donor isolates. Taken together, these data show that ceftazidime- resistance depended on the pi 5-6 enzyme, and that this was TEM-related, probably corresponding to TEM-10. The pi 81 and 71 enzymes from Isolate 886 also were separated by ion exchange chromatography. The pi 71 type, which aligned with OXA-3 in electrofocusing, had negligible activity against ceftazidime, whereas the pi 81 enzyme was active (Table V). Relative hydrolysis rates of /Mactams were different from those with TEM-10. When total DNA from Isolate 886 was digested with BamHI and hybridized with the SHV probe, a band of 3-7 kb was detectable on the Southern blots. This band co-migrated with the band found in the ten isolates with the pi 5-6 and 7-6 enzymes. However, an additional band of 7-0 kb, unique to Isolate 886, was also detected. DNA from Isolate 886 did not hybridize with the TEM gene probe under the conditions used. E. coli K12 J62-1 transconjugants of the isolate acquired the pi 81 enzyme only, and required ceftazidime 64 mg/L for inhibition of growth. Ba/nHI-digested DNA extracts from these transconjugants yielded only the 7 kb fragment that hybridized with the SHV probe, and lacked the 3-7 kb fragment. No plasmid was detected in Isolate 886 or its transconjugants, though the reason for this failure is uncertain. Thus, whilst it is clear that the resistance of Isolate 886 was associated with production of an SHV-derivative, the exact identity this enzyme and of the element that encoded it remain uncertain. Discussion This study examined the extent and nature of 0-lactamase-mediated resistance amongst 1000 consecutively-isolated Gram-negative bacilli obtained from clinical specimens at the RLH during early 1991. Data for 971 fully identified isolates were assessable. The study followed a similar investigation undertaken nine years previously (Whitaker et al., 1983). In both studies, /7-lactamase identification was primarily by isoelectric focusing. This method carries a risk of confusing those enzymes that focus closely to

P-Lactamases in Gram-negative bacilli

443

one another, and this difficulty has been exacerbated as ever more enzyme types have been recognized. Moreover, when an organism has two /Mactamases which focus closely together, these may not be resolved. Unless immense time and resources are available to isolate, purify and assay every enzyme found, these problems cannot be overcome entirely. However, there are many well-established relationships between /?lactamase activity and antibiogram and, in the present study, isolates were screened for susceptibility to a battery of /Mactam antibiotics, chosen to facilitate the recognition of the resistance patterns associated with particular /Mactamase types. This strategy does not prevent possible confusion of (say) different OXA type enzymes, which focus closely to one another and give similar resistance patterns, but does prevent confusion of OXA enzymes with hyperproduced chromosomal /Mactamases, as these give quite different resistance profiles. Moreover (unlike both electrofocusing and the use of DNA probes), it allows discrimination between the parental TEM-1 and -2 and SHV-1 /?lactamases, and their extended-spectrum derivatives. In general, there was good agreement between the /Mactamase identifications inferred from electrofocusing and the observed resistance patterns. Thus, enzymes that were identified by electrofocusing as TEM-1, TEM-2, SHV-1 and OXA types gave resistance to ampicillin and ticarcillin, but not to newer cephalosporins or imipenem, corresponding to the known hydrolytic activity of these enzymes (Bush, 1989). Anomalies were Acinetobacter spp., where antibiogram and /Mactamase profile correlated only poorly, and klebsiellae, where many isolates gave surprisingly large inhibition zones to ampicillin discs, despite /Mactamase production. Chromosomal /Mactamases in enterobacteria and P. aeruginosa presented a further problem. These were detected by electrofocusing in many isolates, few of which exhibited any untoward resistance. It is well known that such enzymes are virtually ubiquitous in Gram-negative bacteria (Matthew & Harris, 1976; Medeiros, 1984). Their intensity on electrofocusing gels tended to be greatest for those isolates in which the antibiogram suggested /Mactamase hyperproduction, whether of Class I enzymes in P. aeruginosa and most enterobacteria, or of Class IV enzymes in K. oxytoca. However, it should be cautioned that the assessment of band intensity is subjective, and depends on the period allowed for colour development on the gel and on the amount of enzyme material loaded, as well as on the specific activity of the cell extract. Therefore, an identification of chromosomal /Mactamase hyperproduction was accepted only where both the electrofocusing results and the antibiogram supported this conclusion. The overall distributions of species were similar in the 1982 and 1991 surveys, except that the proportion of P. aeruginosa isolates had increased from 11 to 17-5%, and the proportion of enterobacteria had decreased commcnsurately (P < 0-01; x2 test). The prevalence of plasmid type /Mactamase production was unchanged amongst E. coli isolates (198/461 in 1982 cf. 191/419 in 1991; P < 0-05; y? test), but had increased amongst P. mirabilis isolates (6/123 in 1982 cf. 25/113 in 1991; P