ABSTRACT
Enterobacter cloacae is not a primary human pathogen but has been considered to be an important cause of nosocomial infections. Even so, there are almost no reports on its ability to produce recognized virulence-associated properties. In this study, we show that most of the E. cloacae strains examined were resistant to serum bactericidal activity and were able to produce aerobactin and mannose-sensitive hemagglutinin, and all of them could adhere to and invade HEp-2 cells. Since E. cloacae is part of the normal intestinal floras of many individuals, we believe that infectious disease due to endogenous E. cloacae might be a result of both host predisposing factors and the bacterial virulence determinants that we have detected in this survey.
Enterobacter cloacae is part of the normal flora of the gastrointestinal tract of 40 to 80% of people and is widely distributed in the environment (15, 19, 39). Like most members of the familyEnterobacteriaceae, these organisms are capable of causing opportunistic infections in hospitalized or debilitated patients (18, 19). They were recognized as a minor cause of hospital infection in a survey published in 1981 (31). Since then, clinical awareness of the potential of E. cloacae strains to cause disease has been reflected in the increasing number of epidemiologic studies of these microorganisms showing that they could be a serious cause of nosocomial gram-negative bacteremia (9, 17-19, 23).
Many studies have demonstrated that the environment and personnel usually are not the source for most infections (8, 18, 19). In fact, it is accepted that these infections often arise from the patient’s endogenous microflora, particularly those of the gastrointestinal tract (17, 18).
E. cloacae has an intrinsic resistance to ampicillin and narrow-spectrum cephalosporins and exhibits a high frequency of mutation to resistance to expanded-spectrum and broad-spectrum cephalosporins (37, 41). These characteristics, associated with the frequent endogenous carriage of E. cloacae, may result in abnormally high levels of this organism in the bowels of hospitalized patients, especially those who have received cephalosporins (17, 18). Intestinal overgrowth of a given species, usually due to antibiotic therapy, may precede bacterial translocation to mesenteric lymph nodes or to the bloodstream (4, 5). Several authors have suggested that the mechanism of hospital infections caused by E. cloacae usually corresponds to endogenous translocation from the digestive tract (4, 5, 17, 18). This idea was proven to be true when in 1992 Lambert-Zechovsky et al. (26) reported for the first time results of molecular analysis strongly supporting the endogenous nature of systemic bacteremia and meningitis due to E. cloacae.
Enterobacteria involved in extraintestinal infections, (mainlyEscherichia coli) are known to possess virulence-associated characteristics that distinguish them from random fecal isolates. A number of studies have elucidated the epidemiology and significance of these virulence-associated properties, including somatic antigens, adhesins, serum resistance, and production of enterotoxins, colicins, siderophores, and hemolysin (reviewed in reference16). Moreover, penetration of the epithelial layer of the intestinal mucosa is a key virulence mechanism of several enteric pathogens, such as Salmonella, Shigella,Escherichia, and Yersinia species, which can be assayed in vitro with human epithelial cell lines, such as HEp-2 cells, an epidermoid carcinoma line derived from human larynx (2, 14, 27).
The fact that E. cloacae is still considered a typical commensal and that, so far, there has been no evidence to suggest that its clinical significance extends beyond opportunistic infections may explain why there are, with rare exceptions, no studies related to its ability to produce the virulence-associated properties mentioned above for other microorganisms. To provide an overview of this subject, we have studied 54 strains of E. cloacae isolated from various sources.
MATERIALS AND METHODS
Bacterial strains.A total of 54 E. cloacaestrains were included in this study: 28 from clinical specimens (blood, urine, and secretions) from hospitalized patients, 5 from catheters, 2 from the scalp, and 19 from the feces of healthy volunteers. All of the strains had been isolated recently in the diagnostic laboratory of a Public Health Hospital or at the Discipline of Microbiology, Federal University of São Paulo, São Paulo, Brazil. The strains were identified as recommended by Brenner (10).
Resistance to serum bactericidal activity.The resistance of the bacterial strains to serum bactericidal activity was tested by the turbidimetric assay performed in 96-well flat-bottom microplates (Corning, Inc., Corning, N.Y.) according to the methodology described by Pelkonen and Finne (33). Briefly, an overnight bacterial growth in Luria-Bertani broth (Difco Laboratories, Detroit, Mich.) was diluted (1/100) in fresh medium and grown until the exponential phase. Bacterial cells were harvested and resuspended in 0.1 M phosphate-buffered-saline (PBS [pH 7.4]) at 4°C to a concentration of ∼107 cells/ml. Seventy-five microliters of each bacterial suspension was distributed in the microplate wells. An equal volume of a pool of sera prepared from blood obtained from healthy human volunteers was added to each well to achieve a final concentration of 50%. Each strain was tested in triplicate with both normal serum and heat-inactivated serum (56°C, 30 min). Microplates were incubated at 37°C, and the growth ability of each strain was assessed by measurement of the optical density at 620 nm (OD620) (Titertek-Multiskan MCC/340 MK-2; Flow Laboratories, Inc., McLean, Va.) for 3 h at 30-min intervals.E. coli K-12 strain 711 was used as the sensitive control. The strains were considered serum resistant when they were able to grow equally in both fresh serum (complement active) and heat-inactivated serum (complement inactive).
Aerobactin production.Aerobactin production was assayed by the method of Carbonetti and Williams (11). Briefly,E. coli indicator strain LG1522 cells (11) were grown to 5 × 107 bacteria/ml in 3 ml of M9 broth supplemented with 0.4% glucose, 50 μg of thiamine per ml, and 200 μM α,α′-dipyridyl (all from Sigma Chemical Company, St. Louis, Mo.). Bacteria were pour plated (107 cells) on M9 agar medium. Testing strains grown in the same broth medium were spotted onto the indicator strain, and plates were incubated for 48 h at 37°C. E. coli LG1315 (43) and K-12 strains were used as positive and negative controls, respectively.
Hemagglutination activity.In general, the media and methods used to test hemagglutination were those described by Adegbola and Old (1). Bacteria were grown statically in 10 ml of nutrient broth (Difco) at 30 and 37°C and subcultured six times at 3-day intervals. The cultures were then centrifuged at 2,000 ×g for 15 min and concentrated by resuspension in 1 ml of saline (NaCl, 0.85% [wt/vol]) to approximately 5 × 109 cells/ml. Fresh chicken, guinea pig, horse, human (group O only), pig, sheep, and ox erythrocytes and tannic acid (Sigma)-treated ox erythrocytes were used at a concentration of 3% in saline. The tests were performed in glass slides by mixing equal volumes (30 μl) of erythrocytes and bacterial suspensions.
Interaction with HEp-2-cultured cells.The methods used were described by Maurelli et al. (30) with modifications. HEp-2 cells (∼4 × 105 cells) were grown over tissue culture coverslips in flat-bottom centrifuge tubes with 19-mm-diameter (Vidrolabor; Ind. e Com. de Vidros para Laboratórios, São Paulo, Brazil) containing Eagle’s modified minimal essential medium (DMEM) (Sigma) with 10% (vol/vol) fetal calf serum (FCS) (Cultilab, Campinas, SP, Brazil) for 48 h at 37°C in an atmosphere of 5% CO2. After being washed with prewarmed PBS, the monolayer was infected with ∼108 bacteria (determined by OD and with a growth curve (OD660 × number of viable bacterial cells) from an overnight growth in tryptic soy broth (Difco), resuspended in DMEM supplemented with 2% FCS and 2%d-mannose (Sigma), and centrifuged in a swinging rotor at 600 × g for 10 min at room temperature. After an incubation period of 2 h at 37°C (infection period), the monolayer was washed with PBS and then further incubated in fresh DMEM–2% FCS–2% d-mannose for an additional 4 h (multiplication period). The monolayer was washed with PBS, and associated bacteria were released by treatment with 0.1% Triton X-100 (E. Merck, Darmstadt, Germany) in PBS for 5 min at room temperature and then were quantitated by plate counting. The number of internalized bacteria was determined as described above, except that 100 μg of gentamicin (Sigma) per ml was added to kill the extracellular bacteria during the multiplication period. The results analyzed represent the mean value of triplicate assays.
To check for adherence pattern and bacterial internalization, the assays were performed as described above, except that eucaryotic cell monolayers were grown over tissue culture coverslips. At the end of the multiplication period, the monolayers were fixed with methanol, stained with Giemsa (E. Merck), and observed under a light microscope. For electron microscopy examination, the monolayers were washed in PBS and fixed with 2.5% glutaraldehyde in 0.05 M sodium cacodylate buffer (pH 7.2) (Sigma). The cells were then gently scraped, washed in 9% sucrose (Sigma) in the same buffer, and prepared for transmission electron microscopy as described by Reynolds (34). Nonadherent and noninvasive laboratory strain E. coli HB101 and enteroinvasive E. coli strain O28ac 9/82 were used as controls.
Cytotoxic activity.Hemolytic activity was determined by the presence of a clear halo around bacterial colonies after overnight incubation at 37°C on blood agar base (Difco) with 5% sheep erythrocytes with or without 10 mM CaCl2 (6). The cytotoxic effect on HeLa cells was assayed as described by Gentry (20).
Colony hybridization assays.The search for DNA homology with adherence and invasion genes already described for enterobacteria was performed by colony hybridization according to the method described by Maas (29). The DNA probes used were labeled by nick translation (35) with [α-32P]dATP (Amersham International plc, Buckinghamshire, United Kingdom) and are listed below: aggregative adherence, 1,000-bp XbaI-SmaI fragment from plasmid pCVD432 (3); diffuse adherence, 350-bpPstI fragment from plasmid pSLM852 (7); attaching-effacing, 1,000-bp SalI-KpnI fragment from plasmid pCVD434 (22); and invasion, 2,500-bpHindIII fragment from plasmid pCVD419 (38).E. coli K-12 strains carrying the same plasmids used as the sources of the probe fragments and E. coli K-12 strains carrying plasmids pBR322 and pUC8 were used as positive and negative controls, respectively.
RESULTS
Serum resistance, aerobactin production, and hemagglutination activity.As shown in Table 1, serum resistance (the ability of a bacterial strain to grow in a pool of sera not depleted of complement as well as in the absence of complement [achieved by heating]) was a typical feature of most E. cloacae strains in this study. Curiously, the only four sensitive strains were blood isolates. Aerobactin production was also a very common characteristic among the strains studied, mainly in those isolated from blood, secretions, and the scalp (Table 1). A summary of the hemagglutination abilities of the 54 strains is presented in Table1. All of the strains could agglutinate guinea pig erythrocytes in the absence of d-mannose (mannose-sensitive hemagglutination [MSHA]), except for two strains isolated from feces, which agglutinated only chicken or tannic acid-treated ox erythrocytes. In regard to the other erythrocyte species, the most frequently agglutinated were chicken (87.0%), horse and tannic acid-treated ox (72.2%), and human (68.5%). It is worth mentioning that about 65% of the strains were able to agglutinate erythrocytes of at least four different species. On the contrary, mannose-resistant hemagglutination (MRHA) was observed with only one strain, which was isolated from urine (Table 1).
Detection of serum resistance, aerobactin production, and hemagglutination activity among E. cloacae isolates from different sources
Association with HEp-2 cells.All of the 54 E. cloacae strains were able to associate with HEp-2 cells in either a diffuse pattern or in clusters (Fig.1), regardless of the origin of the strain. Transmission electron microscopy revealed a very intimate contact between the bacterium and cell membrane or microvilli (Fig.2). Quantification of associated bacteria was performed by counting the CFU per milliliter, and the results are presented in Table 2. The values were higher than 106 CFU/ml for 88.9% (48 of 54) of the strains. Compared to the nonpathogenic E. coli strain HB101, the great majority of the strains showed a statistically higher efficiency of association with HEp-2 cells (P < 0.001 [compared by two-tailed Student’s t test]).
Association of E. cloacae with HEp-2 epithelial cells. Bacterial cells can be seen associated with the eucaryotic cell membrane in clusters (strain 05/91 [A]) or in a diffuse adherence pattern (strain 04/91 [B]). Magnification, ×1,000.
Transmission electron micrograph showing adherence ofE. cloacae to HEp-2 cells. Bacterial cells appear in close contact with the cell membrane and microvilli.
Association of E. cloacae with HEp-2 cells according to clinical source of isolate
Since gentamicin does not penetrate the eucaryotic cell membrane, it is accepted that the survival of HEp-2 cell-associated sensitive bacteria after gentamicin treatment means that they are internalized. According to this assumption, although with variable efficiency, all of the 44 gentamicin-sensitive E. cloacae strains tested were able to internalize into HEp-2 cells (Table 3). It is important to mention that feces isolates could be as invasive as blood isolates. It is worth mentioning that the percentage of associated bacteria that were internalized was usually very low: less than 2% (0.1 to 1.9%) for 23 strains, between 2 and 8.3% for 16 strains, and between 10.3 and 59.2% for 5 strains. On the other hand, high association counts were not always accompanied by high invasion counts. Electron microscopy of HEp-2 cells infected with E. cloacae confirmed its invasion ability (Fig.3).
HEp-2 cell invasion of E. cloacae according to clinical source of isolate
Transmission electron micrograph showing E. cloacae inside the cytoplasm of HEp-2 cells. The arrow points to a bacterial cell enclosed by a membrane-bound vacuole.
Hemolytic activity, cytotoxin production, and hybridization with DNA probes.None of the 54 E. cloacae strains tested were hemolytic, produced cytotoxin in the biological assay, or hybridized with the adherence and invasion DNA probes used.
DISCUSSION
It seems likely that the patient’s own flora, particularly from the gastrointestinal tract, may be responsible for the nosocomial infections caused by E. cloacae in debilitated patients. Since the large intestine is a reservoir of possibly thousands of types of commensal bacteria, it would be reasonable to expect that those which are involved in endogenous extraintestinal infections are apparently more virulent than and possess virulence-associated characteristics that distinguish them from random fecal isolates.
The serum resistance presented by most of the E. cloacaestrains in this study is a common feature of a variety of other gram-negative bacteria that invade and survive in the human bloodstream (36, 40). As in the other examples, this finding suggests that serum resistance may be an important determinant of virulence inE. cloacae. The only four E. cloacae strains that we have found to be sensitive to complement-mediated killing were isolated from blood samples. This is not a controversial result, since it is already known that serum from some bacteremic patients may be ineffective in bactericidal assays utilizing the infecting, homologous organism, even though the strain is fully susceptible to serum from healthy individuals (36).
We have shown that E. cloacae is able to produce aerobactin. The production of this siderophore in one strain ofEnterobacter (Aerobacter) aerogenes(21) and one strain of E. cloacae (42) has already been reported. There are no other studies, to our knowledge, of the production of aerobactin in clinical isolates ofE. cloacae. Der Vartanian et al. (13) have suggested that aerobactin influences either the extent of bacterial translocation from the intestinal tract, the extent of bacterial multiplication in tissues following translocation, or both. Thus, aerobactin secretion in vivo could be an important step in the stages of the infection cycle during which intestine-populating opportunistic bacteria effectively colonize the gut, penetrate the mucous layer covering the intestinal villi, translocate out of intestinal lumen through the epithelial cells, and finally spread to organs within which they may survive.
The search for the production of hemagglutinins among the E. cloacae strains has demonstrated that all but one produced a mannose-sensitive hemagglutinin compatible with the presence of type 1 fimbriae. On the other hand, MRHA was not found, except for in one strain isolated from urine. These results are in agreement with those of Adegbola and Old (1), who found MSHA but not MRHA in most of the E. cloacae strains studied.
Using the HEp-2 cell-line as a model, it was verified that E. cloacae has the ability to adhere to and invade eucaryotic cells. There was no correlation between adherence and the presence of hemagglutinins, since the adhesion assays were carried out in the presence of 2% d-mannose, all of the strains were adherent, and all but one were only MSHA positive. Very recently, Livrelli et al. (28), studying the adherence of 14 E. cloacae clinical isolates to the human cell line Intestine 407, showed that 10 strains were weakly adherent, while 2 others (one involved in bloodstream infection) were strongly adherent. In agreement with our findings, Livrelli et al. have found no relationship between the adhesive pattern and the eventual production of specific fimbriae. Further studies will be carried out in order to determine the adhesin(s) involved in the interaction with HEp-2 cells. Our results have also shown that E. cloacae is in general much less invasive to HEp-2 cells than the enteroinvasive E. colistrain used as control. De Kort et al (12) demonstrated that a strain of E. cloacae isolated from a patient with septicemia was able to invade and survive within rabbit gut enterocytes in situ, although with much lower efficiency than that of an invasive strain of Salmonella typhimurium. We are not aware of any other reports of the ability of E. cloacae to invade eucaryotic cells either in vitro or in vivo.
The 54 strains analyzed in this study were unable to produce hemolysin or cytotoxin. Reports about the production of toxins by E. cloacae are very scarce. Recently, Paton and Paton (32) isolated a strain that produced Shiga-like toxin II-related cytotoxin from a patient with hemolytic-uremic syndrome. Klipstein and Engert (24, 25) have isolated strains capable of producing heat-stable and heat-labile enterotoxins from patients with tropical sprue. In our study, the production of enterotoxins was not evaluated, since none of the strains was isolated from patients with gastrointestinal disease.
Under normal conditions, E. cloacae is no more than a harmless gut commensal organism. However, in patients receiving antibiotic therapy, E. cloacae strains may be selected and may excessively grow in the gastrointestinal tract. This work has shown that E. cloacae strains, including those isolated from feces, possess some virulence properties recognized as important in the onset of extraintestinal infections. Since they have the ability to adhere to and invade eucaryotic cells, it is clear that they would have the possibility of becoming systemic after intestinal translocation. Once outside the gastrointestinal tract, they would take advantage of being serum resistant and being able to chelate iron to survive and spread within the host. There is no doubt about the opportunistic nature of E. cloacae infections, but the presence of specific virulence determinants may play a definitive role in allowing them to happen.
ACKNOWLEDGMENTS
We are grateful to Tania A. T. Gomes for kindly providing the fragment DNA probes and to Mônica A. M. Vieira for technical assistance with the hybridization procedures. We also thank E. Freymuller from the Centro de Microscopia Eletrônica, Escola Paulista de Medicina, Universidade Federal de São Paulo, for providing the conditions under which we could perform the electron microscopy analysis. We are especially grateful to Tania Mara I. Vaz from the Section of Bacteriology of Instituto Adolpho Lutz, São Paulo, Brazil, for confirming the biotyping of the strains. Finally, our special thanks goes to Lilian R. M. Marques for critical review of the manuscript.
This work was supported by Centro de Aperfeiçoamento de Pessoal de Nı́vel Superior, CAPES.
FOOTNOTES
- Received 14 July 1997.
- Returned for modification 23 September 1997.
- Accepted 25 November 1997.
- Copyright © 1998 American Society for Microbiology
