ABSTRACT
Over a 2½-month period in 1999, 37 ceftazidime-resistant nonrepetitive enterobacterial isolates were collected from 37 patients in a Bangkok hospital, Thailand. Eighty-one percent of these strains expressed a clavulanic acid-inhibited extended-cephalosporin resistance profile. An identical extended-spectrum β-lactamase (ESBL), VEB-1, was found in 16 unrelated enterobacterial isolates (Escherichia coli, n = 10; Enterobacter cloacae, n = 2;Enterobacter sakazakii, n= 1; and Klebsiella pneumoniae, n = 3) and in two clonally related E. cloacae isolates. Thebla VEB-1 gene was located on mostly self-conjugative plasmids (ca. 24 to 200 kb) that conferred additional non-β-lactam antibiotic resistance patterns. Additionally, thebla VEB-1 gene cassette was part of class 1 integrons varying in size and structure. Thebla VEB-1-containing integrons were mostly associated with bla OXA-10-like andarr-2-like gene cassettes, the latter conferring resistance to rifampin. These data indicated the spread ofbla VEB-1 in Bangkok due to frequent transfer of different plasmids and class 1 integrons and rarely to clonally related strains. Plasmid- and integron-mediated resistance to rifampin was also found in enterobacterial isolates.
Plasmid-mediated extended-spectrum β-lactamases (ESBLs) were first identified in a Klebsiella pneumoniae isolate in Germany in 1983 (15). Since then, the infections caused by ESBL-producing members of the family Enterobacteriaceaehave rapidly increased (20). These enzymes confer variable degrees of protection against expanded-spectrum cephalosporins such as cefotaxime, ceftazidime, and the monobactam aztreonam. Their activity is inhibited by clavulanic acid in vitro (13,15). Most of the ESBLs that are disseminated worldwide are derivatives of narrow-spectrum TEM- and SHV-type β-lactamases, with one or more amino acid substitutions surrounding their active site, thus explaining the extension of their hydrolytic profile (20). In addition to these ESBLs, non-TEM, non-SHV derivatives with weak structural relationships have been detected with specific geographical distributions, such as CTX-M derivatives in Europe and South America, TOHO-1 and TOHO-2 in Japan, PER-1 in Turkey, and PER-2 in South America (2, 3, 7, 8, 11, 19, 25, 40,41).
Strains expressing the ESBL VEB-1 (also named CEF-1) have rarely been identified, i.e., one Escherichia coli and one K. pneumoniae isolate from the same Vietnamese patient and twoPseudomonas aeruginosa isolates from two patients hospitalized in Thailand (23, 30, 38). Genetic analysis ofbla VEB-1 revealed either its chromosome or its plasmid location and always its integration within class 1 integrons of variable structure. Integrons contain a site-specific recombination system able to capture and express genes as gene cassettes (4, 9). The essential components of class 1 integrons are the 5′ conserved segment (5′-CS) that includes an integrase gene, intI, which encodes a site-specific recombinase, an adjacent site, attI, that is recognized by the integrase and acts as a receptor for gene cassettes, and a common promoter region(s), Pant (P1) and/or P2, from which integrated gene cassettes are expressed (5, 32). The 3′ conserved segment (3′-CS), located downstream of the integrated gene cassettes, usually contains a combination of the three genes qacE1 (antiseptic resistance), sulI (resistance to sulfonamides), and an open reading frame (orf5) of unknown function (27). Each gene cassette is associated with a site-specific recombination site designated the 59-base element (59-be) and located downstream of the gene. Among the cassette-integrated β-lactamase genes, most of them encode β-lactamases of Ambler class D (oxacillin-hydrolyzing β-lactamases) and only rarely of class B (such as IMP-1, VIM-1, and VIM-2 carbapenem-hydrolyzing β-lactamases) or of class A (carbenicillin-hydrolyzing β-lactamases) (10, 22, 31,32). The bla VEB-1 andbla GES-1 gene cassettes are the only class A ESBL gene cassettes so far known as part of class 1 integrons (29, 30).
The aim of the present study was to evaluate the prevalence of nosocomial Enterobacteriaceae isolates that produced VEB-1 among nonrepetitive ceftazidime-resistantEnterobacteriaceae isolates over a 2½-month period in 1999 from patients hospitalized in a hospital in Bangkok, Thailand. To determine whether the enterobacterial isolates were epidemiologically related, they were compared for their β-lactamase content, their plasmid profile, and their genotype using arbitrary primer PCR analyses. Conjugation experiments (and/or electrotransformation) were performed to analyze the self-conjugative property of the plasmids. Finally, class 1 integrons were searched and analyzed in the VEB-1-positive isolates. This is, to our knowledge, the first work designed to study the distribution of an Ambler class A ESBL in this part of the world and the first study on the molecular epidemiology of an integron-located class A ESBL gene.
MATERIALS AND METHODS
Bacterial isolates.Thirty-seven nonrepetitive ceftazidime-resistant entero-bacterial isolates were consecutively collected in the bacteriology laboratory in the Department for Microbiology at Siriraj Hospital, Bangkok (the biggest hospital facility in Thailand) from June to August 1999. The total number of enterobacterial isolates in the studied period was as follows: E. coli (n = 266), K. pneumoniae (n = 62), andEnterobacter sp. (n = 46). Isolates were identified by using the API20E system (bioMérieux SA, Marcy-l'Etoile, France). Electrocompetent E. coli DH10B (GIBCO BRL, Life Technologies, Cergy Pontoise, France) was used as a recipient strain in transformation experiments. Nalidixic acid-resistant E. coli JM109 was used as host in conjugation experiments (30). E. coli NCTC 50192, harboring 154-, 66-, 38-, and 7-kb plasmids, was used as a plasmid-containing reference strain (30). E. coli DH10B harboring recombinant plasmid pRLT1 that carries thebla VEB-1 gene was used as a VEB-1-producing reference strain, as described previously (30).
Susceptibility testing and screening for ESBL-producing strains.The antibiotic susceptibility ofEnterobacteriaceae clinical isolates and their E. coli transformants was first determined by the disk diffusion method on Mueller-Hinton (MH) agar plates with β-lactam and non-β-lactam antibiotic-containing disks (Sanofi Diagnostics Pasteur, Marnes-La-Coquette, France), according to the guidelines of the antibiogram committee of the French Society for Microbiology (30). Since in some cases transconjugants were obtained, plasmid DNAs of the transconjugants were electroporated into E. coli DH10B in order to make valid comparisons of the MICs for all transformants in an identical E. coli genetic background. The double-disk synergy test was performed with cefotaxime, ceftazidime, aztreonam, and amoxicillin-clavulanic acid disks on MH agar plates, and the results were interpreted as described previously (13).
MICs were determined for selected β-lactams by an agar dilution technique on MH agar with an inoculum of 104 CFU per spot, as described previously (28). MICs of some β-lactams were determined alone or in combination with a fixed concentration of either clavulanic acid (2 μg/ml) or tazobactam (4 μg/ml). MIC results were interpreted according to the guidelines of the National Committee for Clinical Laboratory Standards (24).
Conjugation, electroporation, and plasmid DNA content analysis.Conjugation experiments were performed between clinical isolates and E. coli JM109 in solid and liquid media at 37°C as reported previously (28). Transconjugants were selected on Trypticase soy (TS) agar plates containing 100 μg of nalidixic acid per ml and 2 μg of ceftazidime per ml. Non-self-conjugative plasmid DNAs of the enterobacterial isolates and plasmid DNAs of the transconjugants were extracted as described previously (28). They were electroporated into E. coli DH10B, and recombinant strains were selected on ceftazidime-containing (2 μg/ml) TS agar plates. Plasmid DNAs of these transformants were analyzed as described previously (28).
IEF analysis.Cultures of the E. colielectroporants were grown overnight at 37°C in 10 ml of TS broth containing 2 μg of ceftazidime per ml. Then, 1 ml of each overnight culture was further grown for 3 h at 37°C in 10 ml of TS broth with ceftazidime. Analytical isoelectric focusing (IEF) was performed using an ampholine polyacrylamide gel as described previously (28).
PCR amplification for detection of ESBL genes, analysis of class 1 integrons, and sequencing.Under standard PCR conditions (33), a series of primers was designed for detection of Ambler class A β-lactamase genes. Detection was performed for genes encoding TEM (TEM-A and TEM-B [29, 35]), SHV (SHV-F and SHV-B [21]), PER-1/PER-2 (PER-A and PER-B [25,29]), VEB-1 (VEB-1A and VEB-1B [23]), CTX-M (CTXM-2A and CTXM2B [8]), TOHO-1/TOHO-2 (TOHO-A and TOHO-B [11, 29]), and GES-1 (GES-1A and GES-1B [29]) (Table 1). For each reaction, 0.5 μg of whole-cell DNA of the ESBL-possessingEnterobacteriaceae isolates or 0.5 μg of plasmid DNA from E. coli DH10B electroporants was used. Primers for the detection of class 1 integrons were located in the 5′-CS and in the 3′-CS regions (primers 5′-CS and 3′-CS [16]). A combination of 5′-CS or 3′-CS andbla VEB-1 primers was also used for the determination of the genetic content of the class 1 integrons (primers 5′-CS and VEB-B, or 3′-CS and VEB-F [16, 23]). Additionally, since oxa-10 and arr-2 genes had been found to be associated with thebla VEB-1 gene (23, 38), their detections by PCR amplification were performed (primers OXA-10casF and -casB [Table 1] and ARR-2-F and ARR-2-B [37]). Their position relative to the bla VEB-1 gene was determined by PCR using primers for bla VEB-1(primers VEB-1A and VEB-1B [23] and VEB-INV4F and VEB-INFV3B [Table 1]), for bla OXA-10(primers OXA-10promB [Table 1] and OPR-1 and OPR-2 [42]), and for arr-2 (primers ARR-2-F and ARR-2-B).
Unpublished PCR primers used in this study
For direct DNA sequencing, PCR products using external primers for thebla VEB-1 gene cassette (primers VEBcas-F and VEBcas-B [23]) were purified with PCR purification columns (Qiagen). Sequencing reactions were performed using the samebla VEB-1 specific primers and an automated sequencer (ABI 377; Applied Biosystems, Foster City, Calif.). The nucleotide and deduced protein sequences were analyzed with software available over the Internet (29).
RAPD fingerprinting.Random amplified polymorphism DNA (RAPD) analysis was performed and interpreted as described by Williams et al. (43) with some modifications as reported previously (28). The RAPD primers were ERIC-2 and 628 (28) and AP1 and AP4 (1).
Hybridizations.DNA-DNA hybridizations were performed as described by Sambrook et al. with a Southern transfer of an agarose gel containing plasmid DNA from E. coli DH10B electroporants as a template (33). The probe consisted of a 650-bp PCR fragment generated from recombinant plasmid pRLT-1 and internal tobla VEB-1 (30). Labeling of the probe and signal detection were carried out using a nonradioactive labeling and detection kit according to the manufacturer's instructions (Amersham Pharmacia Biotech).
RESULTS
Epidemiology and preliminary PCR detection of β-lactamase genes.A total of 37 nonrepetitive ceftazidime-resistant enterobacterial isolates were collected from hospitalized patients in a 2,000-bed hospital in Bangkok from June to August 1999. Ceftazidime resistance accounted for 4.5% (12 of 266) of E. coli, 41% (19 of 46) of Enterobacter sp., and 9.6% (6 of 62) of K. pneumoniae isolates. Preliminary antibiotic susceptibility testing by disk diffusion showed synergy between ceftazidime- and clavulanate-containing disks for 30 of these 37 isolates. The seven isolates that did not show synergy image wereEnterobacter cloacae. They likely had a derepressed expression of their cephalosporinase since the diameter around the ceftazidime disk increased when antibiograms were performed on oxacillin-containing MH plates (200 μg/ml) (data not shown).
Most of the synergy-positive isolates were not K. pneumoniaebut Enterobacter sp. and E. coli(24 of 30 isolates). PCR experiments using primers specific forbla VEB-1, bla TEM,bla SHV,bla GES-1,bla PER-1, andbla CTX-M-1 gavebla VEB-1-positive results for 18 of these 30 strains (Table 2). Among the 12bla VEB-1-negative strains, all E. cloacae isolates (n = 6) werebla SHV positive and 4 of these strains were also bla TEM positive, theEnterobacter sakazakii isolates werebla SHV positive and one of them was additionally bla TEM positive, the E. coli isolate was bla TEM andbla SHV positive, and the naturallybla SHV-positive K. pneumoniaestrain was also bla TEM positive (data not shown). Thus, the 12 bla VEB-1-negative isolates that showed a positive synergy test possessed either abla TEM- and/or abla SHV-derivative gene that may explain their ESBL phenotype.
Clinical features of the ceftazidime-resistantbla VEB-1-positive enterobacterial isolatesa
The study was then focused on the 18bla VEB-1 PCR-positive isolates. Random PCR showed that these isolates were not clonally related within a given enterobacterial species except E. cloacae 11 and 12 from two patients hospitalized in the same surgery department at different times (Table 2).
MICs of β-lactam antibiotics and IEF analysis.In all cases, the MICs of ceftazidime and aztreonam were higher than those of cefotaxime, which is consistent with the presence of VEB-1 β-lactamase (Table 3) (30). The addition of clavulanic acid consistently decreased the MICs of ceftazidime, cefotaxime, and aztreonam (Table 3). The MICs of β-lactams for transformants were very similar to those found for clinical isolates (Table 3). IEF analysis showed that the transformants expressed a β-lactamase pI value of 7.4 that corresponded to VEB-1 in all cases (Table4). An additional β-lactamase with a pI value of 6.3 was found in all cases except for three transformants (E. cloacae strains 11, 12, and 14) (Table 4). In one case (transformant E. cloacae 14) a β-lactamase with a pI value of 6.5 instead of 6.3 was found. Additionally, a β-lactamase with a pI value of 5.4 that may correspond to a TEM derivative was found in four transformants (E. coli 1, E. cloacae13, E. cloacae 15, and K. pneumoniae18), all of them being bla TEM PCR positive (Table 4).
MICs of β-lactams forbla VEB-1-positive clinical isolates, their transformants in E. coli DH10B, and the E. coli DH10B reference strain
Genotypic characterization of enterobacterial isolates and their transformants, β-lactamase isoelectric points and plasmid DNA sizes of transformants, and phenotypic characterization of non-β-lactam antibiotic resistance markers
Transfer of resistance and plasmid analysis.The transmissibility of the ceftazidime resistance marker was tested by conjugation. The transfer of resistance was obtained for 8 of 18 strains at a high frequency (10−3 to 10−4). For the remaining 10 strains, transfer of the ceftazidime resistance marker was obtained by electroporation using plasmid extracts from the clinical isolates. Plasmid DNAs of the transconjugants were extracted and retransformed into E. coli DH10B in order to make a valid comparison between transconjugants and transformants. All transformants were checked to be bla VEB-1positive by hybridization (data not shown). The plasmid sizes of the transformants ranged from ca. 24 to 210 kb (Table 4). A correlation was established between plasmids of similar size and non-β-lactam antibiotic resistance markers of the transformants (Table 4). This was the case for transformants of E. coli strains 2, 3, 4, 6, 7, and 9 on the one hand and transformants of E. coli 1 and K. pneumoniae 18 on the other hand. It is likely that, in some cases, plasmid transfer may explain interspecies transfer ofbla VEB-1. Additionally, E. cloacae 11 and 12 were clonally related and harbored identically sized plasmid conferring identical non-β-lactam antibiotic resistance phenotypes in their transformants (Table 4).
Identification of bla VEB-1 and integron structure determination.From each of the 18bla VEB-1-positive strains, cassette-located external primers for the bla VEB-1 gene were used to PCR amplify and sequence the entirebla VEB-1 genes. An identicalbla VEB-1 gene was identified in all cases, showing the widespread distribution of this ESBL gene and underlining the genetic stability of this coding sequence and its cassette (data not shown). Since thebla VEB-1 gene cassette had been reported as integron located, PCR amplifications were performed to identify its precise location. Using 5′-CS, 3′-CS, andbla VEB-1 primers, PCR fragments of various sizes were obtained (Table 4), indicating thatbla VEB-1 was part of class 1 integrons. Taking into account the size of the amplified fragments, it could be deduced that the surrounding sequences upstream ofbla VEB-1 varied from 0.1 to 4.2 kb. Since the MICs of β-lactams were similar for each of the transformants (except E. cloacae 14 [Table 4]), the expression of VEB-1 was not related to the position of thebla VEB-1 cassette relative to that of the 5′-CS where the integron promoter sequences are located.
Using primers located in the 3′-CS and in thebla VEB-1 gene, amplimers were obtained ranging from 0.8 to 8.2 kb. Thus, the sizes of thebla VEB-1-containing integrons were on the average around 10 kb, a size that fits with that previously described for other bla VEB-1-containing integrons (23, 29, 38). Interestingly, in some cases (transformantsE. coli 1 and 3 and K. pneumoniae 18), two PCR fragments were obtained using primers forbla VEB-1 and 3′-CS. This result may indicate the presence of two differentbla VEB-1-containing integrons within the same plasmid or spontaneous deletion of the gene cassette within integrons.
Since bla OXA-10 is associated withbla VEB-1 in P. aeruginosa JES isolates (23), PCR experiments usingbla OXA-10 primers were performed. In all cases but three (E. cloacae 11, 12, and 14 and their transformants), a bla OXA-10-like gene was identified (Table 4). A correlation was established between expression of a β-lactamase with a pI value of 6.3 andbla OXA-10-like positive PCR results (Table3). These bla OXA-10-like genes likely coded for the narrow-spectrum clavulanic acid-resistant β-lactamase OXA-10 because (i) the MICs of ceftazidime, cefotaxime, and aztreonam for the transformants were decreased by the addition of clavulanic acid (almost at the level for wild-type E. coli DH10B) and (ii) the pI value of 6.3 corresponded to OXA-10. In the case of transformantE. cloacae 14, the β-lactamase with a pI value of 6.5 did not correspond to any of the class A ESBL genes tested nor tooxa-1, oxa-2, and pse-1 derivatives (data not shown).
PCR detection of the arr-2 gene that conferred resistance to rifampin was found for all clinical isolates and transformants, thus indicating a plasmid location for this gene. Expression of rifampin resistance was found for all transformants except E. coli 8 and E. cloacae 11, 12 and 14. These transformants and transformant E. cloacae 13 gave a weakened PCR signal compared to the other transformants (data not shown).
By performing PCR experiments with primers forbla VEB-1,bla OXA-10, and arr-2 genes and comparing the results with those obtained with primers forbla VEB-1, 5′-CS, and 3′-CS, the structures of integrons were deduced. The most common structure of the integrons was the downstream location of the arr-2 gene frombla VEB-1 followed by abla OXA-10-like gene (transformants E. coli 1, 2, 3, 4, 5, 6, 7, 9, and 10, E. cloacae 15, andK. pneumoniae 16, 17, and 18) (Table 4 and Fig.1). In one case (transformant E. cloacae 13), a bla OXA-10-like gene was located outside the 3′-CS sequence (Table 4). For transformantsE. coli 8 and E. cloacae 11, 12, and 13, which gave a weak PCR signal for the arr-2 gene, nobla VEB-1–arr-2 fragment was amplified by PCR. In these cases, an arr2-like gene may be involved in conferring either no or a low level of resistance to rifampin (Table 4). Although the sul1 gene is associated with class 1 integrons, expression of resistance to sulfonamides was not found in several transformants (E. coli 10, E. cloacae 11 and 12, and E. sakazakii 15 [Table 4]). In these cases, the sul1 gene is either lacking or not expressed, as reported previously (4, 5, 32).
Schematic representation of the most common class 1 integron that carries the bla VEB-1 gene cassette as found in transformants of E. coli 5, 6, 7, 9, and 10 and K. pneumoniae 16. The coding regions are represented as boxes, with arrows indicating their transcription direction. Dashed lines represent undetermined nucleotide sequences but are proportional to the gene distance.
Different groups of Enterobacteriaceaewith integrons of identical structure were found: (i) transformantsE. coli 1 and K. pneumoniae 18 and (ii) E. coli 5, 6, 7, 9, and 10 and K. pneumoniae16 (Fig. 1). The presence of a similar-sized plasmid conferring identical non-β-lactam antibiotic-resistance markers in the transformants and similar integrons was found for E. coli 1 and K. pneumoniae 18 on the one hand and forE. coli 6, 7, and 9 on the other hand (Table 4).
Thus, the spread of the bla VEB-1 gene among these nosocomial isolates may be explained rarely by the spread of clonally related strains and often by similar plasmids and integrons.
DISCUSSION
Taking into account the total number of enterobacterial isolates in the studied period, the prevalence of ESBL-producing organisms was 4.5% for E. coli, 9.6% for K. pneumoniae, and 26% for Enterobactersp. These values ranged within those reported for isolates of North American and European hospitals (12, 18, 20, 36, 46). In a multicenter study performed in Thailand, the prevalence of ceftazidime-resistant E. coli strains ranged from 0 to 62.5% (37). In this same study, the prevalence of ceftazidime-resistant K. pneumoniae was 45% on average, and that is higher than the value reported here. In Korea, the prevalence of ESBL-producing E. coli was 4.8% with mostly TEM- and SHV-type ESBLs and was 22% in K. pneumoniae isolates (14). In Taiwan, the prevalence of ESBL-producing K. pneumoniae was quite high (30%), mostly involving TEM-type ESBLs (17, 26,45). On the contrary, in Japan ESBL-producing organisms are rarely encountered (<0.001%) and the ESBLs are mostly of the TOHO-2 type and rarely of the TEM and SHV type (44). In China, ESBLs have been reported but their prevalence is unknown (34).
In Bangkok, as reported previously in other countries (39), Enterobacter sp. but notK. pneumoniae may represent the main reservoir of ESBL-producing enteric isolates. This may result from either an outbreak or a situation in which such organisms are endemic. Thebla VEB-1 gene seemed to be highly prevalent in these ceftazidime-resistant Thai isolates since it accounted for 60% of the ESBL-possessing isolates. However, frequent identifications of bla TEM andbla SHV derivatives may correspond also to additional ESBL genes in the same clinical isolates. The epidemiological analysis at the strain and plasmid levels indicated that bla VEB-1 had spread among various enterobacterial species. Its dissemination was not due to a single strain or a single plasmid type. Most of thebla VEB-1-positive isolates harbored a self-conjugative plasmid of large size, as found for E. coli MG-1 from Vietnam (30).
The present work established the dissemination of an Ambler class A ESBL gene via various structures of class 1 integrons. A dissemination of an unusual Ambler class A ESBL gene has been reported for bla PER-1 in Turkey (40,41). However, in this case the plasmid location but not the integron location of bla PER-1 is known. In the case of the Ambler class B IMP-1 β-lactamase, its integron-located gene has been reported extensively in Japan (10). However, as opposed tobla IMP-1 (10), thebla VEB-1 location on a class 1 integron is always followed by its expression.
Structure analysis of thebla VEB-1-containing integrons showed their variability compared to the three other knownbla VEB-1-containing integrons, which are from E. coli MG-1 (from Vietnam) and P. aeruginosa JES-1 and Thl-1 (from Thailand) (23, 30,38). Various structures of class 1 integrons have been detected in Vibrio cholerae isolates from Thailand carrying a carbenicillinase gene not related tobla VEB-1 (6). Spread of another class A β-lactamase (bla PSE-1) has also been reported to be related to integron location inSalmonella enterica serotype Typhimurium DT104 (28).
In the present study, most of the integrons containing thebla VEB-1 gene cassette possessedarr-2 and bla OXA-10 gene cassettes. This result raises the question of how these genes have evolved and if they have been transferred among soil organisms such as mycobacteria, Enterobacteriaceae, andPseudomonas species. Indeed, arr-2 shares a structural relationship with arr-1 from Mycobacterium smegmatis (38) andbla OXA-10, and their extended-spectrum derivative genes are mostly found in P. aeruginosa isolates (22). Once located on integrons, these resistance genes may have been transferred in block through a transposition-related mechanism. It would be interesting to test ifbla VEB-1-carrying integrons are located on transposons such as Tn21 derivatives, as described for some class 1 integrons (4).
Our results also raise the question of whether antibiotic selective pressure in hospitals in Thailand may have led to dissemination of this integron-located gene. β-Lactam as well as aminoglycoside, sulfonamide, and disinfectant resistance genes are associated with class 1 integrons that may enhance the in vivo dissemination of these integrons. Restricted clinical use of broad-spectrum antibiotics and rigorous hygiene measures are the most important means to prevent the spread of this ESBL gene as much as possible. Sincebla VEB-1 has been detected also in twoP. aeruginosa isolates from Thailand, a further study would evaluate its spread among other gram-negative rods.
The important prevalence of bla VEB-1 among enterobacterial isolates may lead to heavy use of carbapenems in this Bangkok hospital, which in turn may favor the selection of, at least, carbapenem-resistant P. aeruginosa isolates.
ACKNOWLEDGMENTS
We thank Venus Tolun for help in preliminary experiments.
This work was financed by a grant from the Ministères de l'Education Nationale et de la Recherche (grant UPRES, JE-2227), Université Paris XI, France.
FOOTNOTES
- Received 20 July 2000.
- Returned for modification 4 September 2000.
- Accepted 31 October 2000.
- Copyright © 2001 American Society for Microbiology
