Corresponding author and reprint requests: I. Stock, Institut für Medizinische Mikrobiologie und Immunologie, Pharmazeutische Mikrobiologie, Universität Bonn, Meckenheimer Allee 168, D-53115 Bonn, Germany Tel: +49 228732114 Fax: +49 228735267 E-mail: firstname.lastname@example.org
Objective To investigate the natural susceptibility to 69 antimicrobial agents of 107 Enterobacter strains comprising E. amnigenus (n = 18), E. cancerogenus (n = 26), E. gergoviae (n = 28) and E. sakazakii (n = 35).
Methods Minimal inhibitory concentrations (MICs) were determined with a microdilution procedure in Isosensitest broth and cation-adjusted Mueller–Hinton broth.
Results All the species were naturally sensitive or intermediate to tetracyclines, amino-glycosides, numerous β-lactams (acylureidopenicillins, ticarcillin, ampicillin/sulbactam, several cephalosporins, carbapenems, aztreonam), quinolones, antifolates, chloramphenicol and nitrofurantoin. Natural resistance was found to penicillin G, oxacillin, several macrolides, lincosamides, streptogramins, glycopeptides, rifampicin and fusidic acid. Species-related differences in natural susceptibility were found to some β-lactams, azithromycin and fosfomycin. Whereas E. gergoviae was the most susceptible species to azithromycin, E. cancerogenus was most susceptible to fosfomycin and was the only species showing natural resistance to amoxicillin, amoxicillin/clavulanic acid, cefaclor, cefazoline, loracarbef and cefoxitin. There were only minor medium-dependent differences in susceptibility to most antibiotics.
Conclusions The present study establishes a database concerning the natural susceptibility of recently established Enterobacter species to a wide range of antibiotics, which can be applied for the validation of routine susceptibility test results. β-Lactam susceptibility patterns indicate the expression of species-specific β-lactamases expressed at high or low levels in all the species except E. sakazakii.
Enterobacter species are common causes of nosocomial infections in humans. Apart from E. hormaechei, E. asburiae and E. aerogenes, which represent the most frequently encountered Enterobacter species in clinical specimens, there are several further Enterobacter taxa associated with human disease. The most common is E. sakazakii, a yellow-pigmented and biochemically heterogeneous species with an unusual thermotolerance [1,2]. Enterobacter sakazakii, first described in 1980 , is an important agent of life-threatening neonatal sepsis and meningitis, with mortality rates of 40–80% [4,5]. In some cases these diseases are complicated by the development of brain abscesses and other secondary disorders . Enterobacter sakazakii was also shown to cause an outbreak of severe necrotizing enterocolitis in a neonatal intensive care unit in Belgium . Although the guts of insects were shown to contain the species  and dried infant formula milk has been implicated in both outbreaks and sporadic cases of E. sakazakii meningitis , its natural habitat remains obscure [3,4].
In 1981, Izard et al. described Enterobacter strains previously called Enterobacter H3, and called them E. amnigenus. These bacteria have been isolated from water and from several clinical specimens, e.g. respiratory tract, wound, blood and feces , and were shown to cause sepsis and other infections in man [12–14]. Strains of E. cancerogenus have been isolated predominantly from human blood and spinal fluid  and were found to be etiological agents of wound and urinary tract infections, sepsis and osteomyelitis [16–21]. Enterobacter taylorae, described in 1985 by Farmer et al. , was shown to be a junior synonym of E. cancerogenus. A further, widely distributed Enterobacter species is E. gergoviae. It has been isolated from environmental samples and clinical specimens and is likely to cause sepsis and urinary tract infections [5,23].
Although there is some information on antimicrobial susceptibility patterns of recently established Enterobacter species, in most cases only a small number of strains/species and/or a few antibiotics have been examined. Data on natural sensitivities or resistances are exceptional.
The aim of the present study was to establish a database concerning the natural susceptibility to a wide range of antibiotics of strains of E. amnigenus, E. cancerogenus, E. gergoviae and E. sakazakii, examining isolates from different sources and areas. In particular, it was investigated whether there are species-related differences in natural susceptibility. The data from 107 strains tested with 69 antibiotics can be valuable for the validation of routine susceptibility test results.
Materials and methods
A total of 107 Enterobacter strains was examined. Most strains were taken from clinical specimens, but environmental strains were also included. Data on the strains studied are summarized in Table 1. Enterobacter amnigenus ATCC 33072T, E. cancerogenus ATCC 33241T, E. gergoviae ATCC 33028T, E. sakazakii ATCC 29544T and Escherichia coli ATCC 25922 were used as control strains for susceptibility testing.
Table 1. Enterobacter strains tested
Strain (additional designation)
Institute of Medical Microbiology, University of Zürich.
Gärtner and Colleagues Laboratory, Weingarten.
d BCCMTM/LMG Bacteria collection, University of Gent.
Department of Clinical Bacteriology, Culture Collection, University of Göteborg.
T, Type strain; T1, formerly Type strain of E. taylorae.
The identification of the strains was confirmed with a commercial identification system for Enterobacteriaceae (Micronaut-E, Merlin-Diagnostika, Bornheim, Germany). This identification system contains biochemical key reactions for most clinically significant Enterobacteriaceae species. The inoculum for the identification tests was a suspension from an overnight culture on Isosensitest agar (Oxoid, Basingstoke, UK) in physiological saline solution at 106 colony-forming units (CFU)/mL; the incubation times were 24 h at 37 °C.
Antibiotics and antibiotic susceptibility testing
Antibiotic susceptibility was tested with a microdilution procedure in Isosensitest broth (Oxoid). Seven strains of each species were also tested in cation-adjusted Mueller–Hinton broth (CAMHB; Difco Laboratories, Detroit, MI, USA). After inoculation of antibiotic-containing microtitration plates (Merlin-Diagnostika) with 100 µL of appropriate bacterial suspension (3 × 105 to 7 × 105 CFU/mL) and incubation for 22 h at 37 °C, the minimum inhibitory concentration (MIC) values were determined with a photometer for microtitre plates (Labystems Multiscan Multisoft, Helsinki, Finland). MIC data were evaluated using Excel (Microsoft). All antibiotics were kindly provided to Merlin-Diagnostika by their manufacturers.
Evaluation of natural antibiotic susceptibility
Plotting the MIC of a particular antibiotic for one species against the number of strains found with the respective MIC usually results in a bimodal distribution. One peak with relatively low MICs represents the natural population and one peak with higher MICs represents the strains with acquired (secondary) resistance. Analysis of the MIC distribution of all strains of one species for each antibiotic permitted the determination of the biological thresholds, which limit the natural population at high MICs, but not those strains with secondary resistance. The MIC values of the natural population were investigated to determine whether they were above or below the breakpoints of the standards, which assess the clinical susceptibility. When the natural population was sensitive or intermediate according to the cited standard, it was described as naturally sensitive or naturally intermediate, respectively. When the natural population was clinically resistant, it was described as naturally (intrinsically) resistant. The method has been described in detail previously [24–26]. Clinical breakpoints for apramycin, lividomycin A and ribostamycin were defined as proposed recently .
There were no, or only small, species-related differences in natural susceptibility to most antimicrobial agents. All the species were naturally sensitive or intermediate to tetracyclines, aminoglycosides, acylureidopenicillins, ticarcillin, ampicillin/sulbactam, several cephalosporins, carbapenems, aztreonam, quinolones, antifolates, chloramphenicol and nitrofurantoin. Natural resistance was found to benzylpenicillin, oxacillin, the macrolides except azithromycin, lincosamides, streptogramins, glycopeptides, rifampicin and fusidic acid. Significant differences in natural susceptibility to some β-lactams were found among the species, azithromycin and fosfomycin. Whereas E. gergoviae was the most susceptible species to azithromycin, E. cancerogenus was most susceptible to fosfomycin. Twenty-two of the 26 strains of the latter species were uniformally resistant or intermediate to amoxicillin, amoxicillin/clavulanate, cefaclor, cefazoline, loracarbef and cefoxitin, showing the natural resistance of E. cancerogenus to these agents (see Discussion). This was in contrast to the remaining Enterobacter species tested, which were naturally sensitive or intermediate to all β-lactams except benzylpenicillin, oxacillin and cefoxitin. Apart from E. cancerogenus, E. gergoviae was the species least susceptible to cefoxitin. Sixty-four per cent of the strains belonging to the natural population of E. gergoviae were intermediate or resistant to this cephalosporin. Although not resistant to any cephalosporins, E. amnigenus showed susceptibility patterns similar to that of E. cancerogenus to some cephalosporins and was less susceptible than E. gergoviae and E. sakazakii to loracarbef, cefotiam, cefixim, cefpodoxime, cefdinir and ceftibutene. An overview of the antibiotic susceptibilities of the tested Enterobacter strains is shown in Table 2. The natural antibiotic sensitivities and resistances are summarized in Table 3.
Table 3. Natural antibiotic susceptibility (+) of E. amnigenus, E. cancerogenus, E. gergoviae and E. sakazakii, classified as sensitive, intermediate and resistant according to the standards mentioned in Table 2
a Data only valid in the presence of Isosensitest broth (see Results).
Note: If ≤20% of the strains belonging to a natural population were attributed to one of the clinical categories, these percentages were disregarded.
For most of the tested antibiotics, there were no, or minor, differences in susceptibility, dependent on the medium. For all the species, the MICs of tetracyclines were generally two doubling dilution steps higher in Isosensitest broth than in CAMHB (data not shown). This led to the absence of strains of intermediate susceptibility to tetracyclines in the presence of CAMHB. Medium-dependent differences in susceptibility were also obtained for the macrolides tested. The MICs of macrolides were one to two doubling dilution steps higher in Isosensitest broth. These differences affected the clinical assessment of the MICs for the natural populations of all species for azithromycin (Table 2).
In both media the MIC data of all antibiotics were reproducible for the control strains examined. The MIC values for E. coli ATCC 25922 in Isosensitest broth and CAMHB were within the control limits for susceptibility testing according to DIN criteria (data not shown).
In the literature, information on antimicrobial susceptibilities of recently established Enterobacter species is usually restricted to a small number of strains and/or a few antibiotics. In most cases data on natural sensitivities or resistances only refer to well-established Enterobacter taxa. The present study was conducted to establish a database concerning the natural susceptibility to a range of antibiotics for some ‘new’Enterobacter species. These data can be valuable for the validation of routine antibiotic susceptibility test results and give insight into the respective mechanisms of natural resistance.
Natural susceptibility to β-lactams
The most extensive studies published on ‘new’Enterobacter species were performed by Farmer et al.  and Muytjens et al. . Both studies included susceptibility data for E. cancerogenus, which was the least susceptible species to β-lactams in the present study. Our data showing natural resistance to aminopenicillins and several ‘older’ cephalosporins are in accordance with Muytjens et al., who found all strains examined (n = 10) to be highly resistant to ampicillin and cefazolin, applying an agar-dilution procedure . Farmer et al., however, showed that only 53% of the total population (n = 32) were resistant to ampicillin . These results are not surprising, in view of the fact that Farmer and co-workers used an agar-diffusion technique, whose results in some instances disagree with the results obtained by dilution procedures [29–31]. In particular, in a study published recently, it was demonstrated that there are method-associated dependencies in the results of susceptibility tests for Enterobacter species . Further reports on aminopenicillin-sensitive E. cancerogenus strains were published by Martinez et al., who found one strain among two ampicillin-resistant strains to be aminopenicillin sensitive , and by Westblom and Coggins, who isolated one ampicillin-sensitive strain from a wound infection . Whereas, in the latter case, it cannot be excluded that the result was due to the method applied (which was not stated), there is evidence that in both cases the result was attributed to strains with acquired sensitivity, as documented in the present study according to four of the strains tested (Table 2). (The phenomenon of ‘acquired sensitivity’ describes a relatively rare phenotype but it occurs for some antibiotics in nearly all Enterobacteriaceae species; see ref.  for an overview). The phenotype of E. cancerogenus, showing natural resistance to amoxicillin and amoxicillin/clavulanic acid as well as to the cephalosporin derivatives cefaclor, cefazoline, loracarbef and cefoxitin is similar to the natural β-lactam phenotypes of well-established Enterobacter species, and indicates the presence of chromosomally encoded AmpC β-lactamases in this species. Class C enzymes are responsible for the respective phenotypes in strains of the E. cloacae complex and of E. aerogenes[30,33]. In agreement with our thesis, Pitout et al. recently found inducible AmpC β-lactamases in all E. cancerogenus strains examined (n = 6) (referred from the authors as Bush group 1) with isoelectric point (pI) values >9, indicating enzymes similar to those found in the same study in wild-type strains of the E. cloacae complex (pIs 8.0 to >9) and of E. aerogenes (pIs 8.4–8.8) .
Apart from natural resistances to aminopenicillins, amoxicillin/clavulanic acid and narrow-spectrum cephalosporins, the intrinsic resistance to cefoxitin is a common feature of numerous Enterobacter species. According to our data, natural high-level cefoxitin resistance in Enterobacter species is present in strains of E. cancerogenus (as presented here), E. aerogenes, E. cloacae, E. hormaechei and E. asburiae (unpublished data). It is of particular interest that in this study the majority of the strains of the natural population of E. gergoviae were resistant or intermediate to cefoxitin (Table 2). This indicates that E. gergoviae—which was sensitive or intermediate to amoxicillin and to all remaining cephalosporins tested (Tables 2 and 3)—expresses a distinct naturally occurring β-lactamase at low levels and that the enzyme assumed shows a more or less specific ‘cefoxitinase’ activity. Recently, a low-level activity of a not-inducible AmpC-β-lactamase was demonstrated in three E. gergoviae strains .
As for E. cancerogenus and E. gergoviae, chromosomally encoded β-lactamases have to be postulated for E. amnigenus strains. Although there was no natural resistance to cephalosporins, including cefoxitin, a decreased susceptibility was found to cefixime, cefpodoxime and ceftibutene, compared to the natural populations of E. gergoviae and E. sakazakii (Table 2). This species-associated phenotype indicates the presence of at least one naturally occurring enzyme, expressed at low levels. β-Lactamase studies have not been published for strains of E. amnigenus.
In contrast to the species mentioned above, the susceptibility testing of natural populations of E. sakazakii to β-lactams gave no evidence for the expression of any β-lactamases. In agreement with the data of Muytjens and van de Ros-van de Repe , the data of this study show that E. sakazakii is the Enterobacter species most susceptible to β-lactam antibiotics examined so far. Reports indicating a naturally occurring high susceptibility of E. sakazakii to numerous β-lactams have appeared repeatedly, but in all but two studies [1,28] only a few strains were included. However, the absence of β-lactamases in E. sakazakii was refuted in a recent study by Pitout et al. . It was shown that E. sakazakii strains sensitive to all cephalosporins including cefoxitin produced Bush group 1 β-lactamases with pIs of 7.4–8.0 at low levels. Enterobacteriaceae species which naturally express their β-lactamases at negligible or low levels are not uncommon and include several species, e.g. Escherichia coli and Shigella spp. , Proteus mirabilis and Edwardsiella tarda. In general, β-lactam MICs for these species do not significantly differ from the MICs seen for some β-lactamase-negative Enterobacteriaceae. In the case of E. sakazakii, the MIC values of β-lactams were nearly identical to the β-lactam MICs seen for the natural populations of Salmonella enterica, which is known to produce no naturally occurring β-lactamase .
Natural susceptibility to other antibiotics
The natural resistance of all species to rifampicin, lincosamides, glycopeptides, streptogramins and fusidic acid was expected, since resistance to these agents is a typical feature of nearly all Enterobacteriaceae and has been largely attributed to the outer membrane of these bacteria, which represents a more or less strong barrier to these agents . Significant differences in susceptibility to non-β-lactams, leading to different categorizations by the standards applied, were only found for fosfomycin and azithromycin (Tables 2 and 3). The molecular background of the species-related differences in fosfomycin susceptibility is not understood and awaits further study. Because there was no naturally occurring high-level fosfomycin resistance in E. amnigenus and E. gergoviae, it is likely that a reduced permeability of the cell membrane in the presence of this antibiotic, rather than a specific fosfomycin:gluthathione-S-transferase, contributed to the phenotype observed (Table 2). Alternatively, a transferase with low affinity to fosfomycin or a low-level enzyme expression would be possible . It should be noted that results in fosfomycin susceptibility testing are highly dependent on the constituents of the media and on several further factors . However, according to the data there were no differences in fosfomycin susceptibility dependent on the medium (data not shown). The natural resistance of Enterobacter species to the macrolides, excepting azithromycin is probably due to the barrier function of the outer membrane . The higher susceptibility of the species tested to azithromycin—resulting in some clinically sensitive E. gergoviae strains—was not unexpected since azithromycin diffuses significantly better than other macrolides through several porin channels, and is known to interact with the acidic lipopolysaccharide, penetrating the outer membrane by virtue of this interaction (see  for an overview). Species-dependent differences in macrolide susceptibility might be due to variations of the membrane architecture resulting in a cell envelope with species-specific hydrophobicity.
The technical assistance of Thomas Grüger is gratefully acknowledged. We thank Alexander von Graevenitz (Zürich, Switzerland) and Gerda Stempfel (Weingarten, Germany) for providing numerous isolates. This study was supported by Merlin-Diagnostika, Bornheim, Germany.