49

Evidence of Extensive Interspecies Transfer of Integron-Mediated

Antimicrobial Resistance Genes among Multidrug-Resistant Enterobacteriaceae
in a Clinical Setting
Maurine A. Leverstein–van Hall, Adrienne T. A. Box, Eijkman-Winkler Institute for Microbiology, Infectious
Hetty E. M. Blok, Armand Paauw, Ad C. Fluit, Diseases, and Inflammation, University Medical Center
and Jan Verhoef Utrecht, Utrecht, The Netherlands

Multidrug resistance in gram-negative bacteria appears to be primarily the result of the

acquisition of resistance genes by horizontal transfer. To what extent horizontal transfer may
be responsible for the emergence of multidrug resistance in a clinical setting, however, has
rarely been investigated. Therefore, the integron contents of isolates collected during a no-
socomial outbreak of genotypically unrelated multidrug-resistant Enterobacteriaceae were
characterized. The integron was chosen as a marker of transfer because of its association with

Downloaded from jid.oxfordjournals.org by guest on January 31, 2011

multiresistance. Some genotypically identical isolates harbored different integrons. Grouping
patients carrying the same integron yielded 6 epidemiologically linked clusters, with each
cluster representing a different integron. Several patients carried multiple species harboring
the same integron. Conjugation experiments with these strains resulted in the transfer of
complete resistance patterns at high frequencies (10⫺2 to 10⫺4). These findings provide strong
evidence that the horizontal transfer of resistance genes contributed largely to the emergence
of multidrug-resistant Enterobacteriaceae in this clinical setting.

Multidrug-resistant Enterobacteriaceae (MRE) are being iso- (namely, transposons or defective transposon derivatives) that
lated at an increasing rate in hospital settings and are having a comprise a site-specific recombination system capable of inte-
significant impact on clinical practice and overall treatment costs grating and expressing the genes contained in cassette struc-
[1–5]. This multiresistance may be mediated by chromosomally tures. The 3 essential components of an integron, located within
located resistance determinants or mutations in a resident gene; the 5
conserved segment (CS) of the element, include (1) an
however, it may also develop through the acquisition of resistance integrase gene, Int1, which encodes a site-specific recombinase;
genes or an array of resistance genes by horizontal transfer. This (2) an adjacent att1 site, which is recognized by the integrase
latter phenomenon is currently thought to play an increasing role and acts as a receptor for gene cassettes; and (3) a promoter
in the development of multidrug resistance in Enterobacteriaceae region, Pant. There are 4 distinct classes of integrons, each en-
[6]. Plasmids and transposons are known to be involved in the coding a distinct integrase gene. Class 1 integrons comprise
transfer of resistance genes from one cell to another. These ele- most integrons found in clinical isolates and are strongly as-
ments can hunt as a pack by interacting with each other in a sociated with the multiresistance seen in the hospital environ-
variety of ways that enhance their collective ability to transfer ment [9]. In total, ∼50 different class 1 integrons and 1 60 dif-
resistance genes. Some of the plasmids that carry multiple resis- ferent gene cassettes have been described to date. The currently
tance genes also have transfer systems that enable them to trans- known gene cassettes confer resistance to aminoglycosides, pen-
icillins, cephalosporins, carbapenems, trimethoprim, chloram-
fer DNA between unrelated species (promiscuous or broad-host-
phenicol, rifampin, erythromycin, and quaternary ammonium
range plasmids) [7].
compounds (disinfectants and antiseptics) [10]. Often, class 1
In recent years, it has been shown that a substantial portion
integrons also contain an additional resistance gene in the 3

of the resistance genes present on plasmids and transposons in
CS, downstream from the gene cassettes, namely, sulI. This gene
gram-negative bacilli is integrated in DNA elements called in-
confers resistance to sulfamethoxazole [8].
tegrons [8]. These integrons are potentially mobile elements
In 1996, our hospital was confronted with a sudden increase
in the incidence of MRE on the neurology and neurosurgery
Received 24 September 2001; revised 25 February 2002; electronically wards [5]. Genotyping showed that 17 (28%) of the 61 patients
published 10 June 2002.
Reprints or correspondence: Dr. Maurine A. Leverstein–van Hall, Eijk- involved were either infected with or colonized by a single mul-
man-Winkler Institute for Microbiology, Infectious Diseases, and Inflam- tidrug-resistant Klebsiella oxytoca strain. The isolates recovered
mation, University Medical Center Utrecht, Rm. G04.614, PO Box 85000,
from the other patients comprised 8 different bacterial species,
3508 GA Utrecht, The Netherlands (m.leversteinvhall@lab.azu.nl).
and subsequent genotyping yielded a great variety of strains. The
The Journal of Infectious Diseases 2002; 186:49–56
䉷 2002 by the Infectious Diseases Society of America. All rights reserved.
vast majority of these strains were resistant to at least 3 classes
0022-1899/2002/18601-0007$15.00 of antimicrobial agents, including trimethoprim-sulfamethoxa-
50 Leverstein–van Hall et al. JID 2002;186 (1 July)

zole. The transfer of class 1 integron-mediated antimicrobial re- specimens sent in by physicians for clinical reasons (clinical sam-
sistance genes, therefore, probably played an important role dur- ples). Except for isolates from samples from 2 wounds, all isolates
ing this outbreak. Few attempts have been made to determine were recovered from urine and sputum samples, representing either
to what extent gene transfer occurs in nature [11, 12]. colonization or infection. No deaths were attributable to MRE.
The remaining 34 patients were identified as carriers of MRE on
The objective of the present study was to investigate whether
the basis of the results of an infection-control screening program
and to what extent the horizontal transfer of resistance genes
to detect intestinal colonization (screening samples). Subsequent
contributed to the emergence of MRE on the neurology and clinical samples from only 2 patients (6%) yielded GRE during
neurosurgery wards. The integron was chosen as a marker of these patients’ remaining hospital stays (median, 24 days). These
transfer because it is a well-defined transferable genetic deter- samples were sent for screening 2 and 7 days after collection of the
minant, and the variety of the contents of the gene cassettes first positive screening samples. This low number was possibly the
made one specific integron an unlikely common trait for all of result of the lack of selective antibiotic pressure on these strains
the isolates involved. The hypothesis was that maximal circum- because of the implemented restrictive antibiotic policy followed
stantial evidence for horizontal transfer would be obtained if at the Neurodivision at that time [5]. Environmental cultures were
all of the following observations could be made: (1) Isolates also obtained as part of the infection-control program. Multi-
with an identical genotype harbored different (combinations resistant Klebsiella pneumoniae and K. oxytoca, respectively, were

Downloaded from jid.oxfordjournals.org by guest on January 31, 2011

cultured from a hand-directed soap dispenser in the staff refresh-
of) integrons; (2) different species or strains collected from the
ment room and from a brush used for cleaning the measuring glass
same patient harbored the same integron; and (3) different iso-
needed for parenteral feeding. Eighty-one staff members practicing
lates of epidemiologically linked patients carried the same in- bedside care at the Neurodivision were screened for intestinal car-
tegron. Thus, the presence of integrons was determined among riage. One neurologist was identified as a GRE carrier.
the outbreak strains, the contents of the integrons were char- Bacterial strains. Isolates from 56 of 61 patients involved in
acterized, and the extent to which integron-mediated transfer the outbreak were available for study. Eighty-six of the isolates (29
of antimicrobial resistance genes had occurred was assessed. Escherichia coli, 21 K. oxytoca, 20 K. pneumoniae, 10 Citrobacter
freundii, 3 Enterobacter cloacae, 2 Proteus mirabilis, and 1 Entero-
bacter aerogenes) were selected. Criteria for selection were as fol-
Patients, Materials, and Methods lows: (1) at least 1 isolate per patient, (2) at least 1 isolate of each
Setting and patients. A sudden increase in the incidence of species per patient if multiple species were isolated from the patient,
MRE in general and of multidrug-resistant K. oxytoca in particular and (3) at least 1 isolate of each genotype per species per patient
was observed in March 1996 in the Neurodivision of the University if multiple isolates of the same species were obtained from the
Hospital Utrecht, Utrecht, The Netherlands, an 858-bed teaching patient. Also studied were 3 gentamicin-susceptible but sulfa-
and referral hospital [5]. The Neurodivision comprises the De- methoxazole-trimethoprim–resistant isolates from 3 patients with
partments of Neurology and Neurosurgery, which are located next GRE, the 2 environmental isolates, and the isolate from the
to one another. The Department of Neurosurgery has a 7-bed me- neurologist.
dium-care area, 4 low-care private rooms, and 2 wards, one with Identification and antimicrobial susceptibility. Identification
21 beds (3 private rooms, 3 2-bed rooms, and 3 4-bed rooms) and and susceptibility testing were done using an automated system
one with 19 beds (3 private rooms, 2 2-bed rooms, and 3 4-bed (VITEK) and AMS R09.1 software (both supplied by bioMérieux).
rooms). The Department of Neurology has an 8-bed medium-care Additional testing for susceptibility to sulfamethoxazole, strepto-
area and 2 wards, one with 20 beds (4 private rooms, 4 2-bed rooms, mycin, aztreonam, erythromycin, and chloramphenicol was done
and 2 4-bed rooms) and one with 31 beds (3 private rooms, 4 2- using the agar diffusion method. The break points were those rec-
bed rooms, and 5 4-bed rooms). The private rooms are provided ommended by the National Committee for Clinical Laboratory
with negative air pressure in relation to their anteroom. During Standards [13]. The production of extended-spectrum b-lactamases
the 1-year period of study (week 46, 1995, through week 45, 1996), (ESBLs) was determined with the ESBL Etest (AB BIODISK)
2604 patients were admitted to the Neurodivision, with a mean according to the guidelines of the manufacturer.
length of stay of 17 days. Typing by random amplified polymorphic DNA (RAPD) analysis
All patients who had had MRE isolated from clinical samples and pulsed-field gel electrophoresis (PFGE). To determine
or screening samples during the 1-year period were included in the whether the high prevalence of MRE was the result of an epidemic
study (n p 61). Their mean length of stay in the hospital was 70 spread of a few strains or merely an increase in unrelated MRE,
days (median length of stay, 50 days). Resistance to gentamicin all isolates of the same species (except for Proteus species) were
was the sole criterion for MRE, because 95% of all gentamicin- typed by RAPD analysis, as described elsewhere [5]. If isolates of
resistant Enterobacteriaceae (GRE) are multidrug resistant (i.e., the same species were shown to contain identical integrons, they
resistant to ⭓1 agent belonging to ⭓3 of the following antibiotic were also typed by PFGE analysis. In brief, after the clinical isolates
agent classes: aminoglycosides, broad-spectrum penicillins, cepha- were grown overnight in Luria Bertani (LB) broth, the bacterial
losporins, quinolones, and trimethoprim-sulfamethoxazole) and pellets were washed and suspended in 100 mM Tris-HCl (pH 8.0)
because no MRE were susceptible to gentamicin. The first and plus 200 mM EDTA. Bacterial plugs were then prepared in solu-
subsequent phenotypically different GRE isolates from almost all tions with a final concentration of 50 mM Tris-HCl (pH 8.0) and
patients were stored at ⫺20⬚C. 100 mM EDTA (TE), 0.6% pulsed-field certified agarose (BioRad),
The first isolate from 27 of the 61 patients was obtained from 0.5% SDS, and 0.5 g/L proteinase K (Merck). After a 30-min in-
JID 2002;186 (1 July) Transfer of Integron-Mediated Resistance Genes 51

cubation at 37⬚C in 50 mM Tris-HCl, 50 mM EDTA, 1% sarcosine, the already sequenced CS-PCR product, the 2 integrons were con-
and 0.1 g/L proteinase K, the plugs were washed sequentially in sidered to be identical. If the CS-PCR product contained a different
aqua bidest and TE and were stored at 4⬚C until use. The bacterial RFLP pattern, the new CS-PCR product was sequenced as well.
DNA was digested with restriction endonuclease XbaI (Roche) Amplicon sequencing reactions were performed on purified PCR
according to the guidelines of the manufacturer. Restriction frag- products, using a PCR-cycle sequencing kit (Perkin-Elmer) and an
ments were then separated by PFGE in a 1% pulsed-field certified automated sequencer (ABI 377; Applied Biosystems). Purification
agarose gel (BioRad) in 45 mM Tris-borate (pH 8.3) and 1 mM of single CS-PCR products was attained using MicroSpin G-25
EDTA with a CHEF-DRII drive module (Bio-Rad). The initial columns (Pharmacia). If a strain contained multiple CS-PCR prod-
pulse time of 2.2 s was increased linearly to 54.2 s in 20 h at 6 V/ ucts, each product was cut from the agarose gel and purified, using
cm at 14⬚C. The gels were then stained with ethidium bromide, and the QIAquick Gel Extraction kit (Qiagen). For RFLP analysis, the
the restriction fragments were visualized under UV light. The band- purified CS-PCR amplicons were then digested. At least 2 different
ing patterns were compared visually and classified according to restriction endonucleases were chosen for each RFLP assay on the
previously described criteria [14]. The strains were considered basis of the content of the sequenced product. Digestions were
unique if the combination of the RAPD type and the PFGE type performed according to the manufacturer’s instructions.
was not seen in any of the other isolates. Conjugation experiments. To obtain further evidence for in
Detection of integrons by polymerase chain reaction (PCR) am- vivo horizontal transfer of resistance genes, conjugation experi-

Downloaded from jid.oxfordjournals.org by guest on January 31, 2011

plification. To identify the presence of an integron and to deter- ments were done with at least 1 strain of clinical isolates from each
mine the size of any inserted gene cassette, a CS-PCR was per- patient who carried multiple species with the same (combination
formed according to the method of Lévesque et al. [15], with the of) integrons and with a sulfamethoxazole-susceptible (Smz S), ri-
exception that template DNA was isolated by use of Nucleobond fampicin-resistant (Rif R) E. coli K12 recipient strain. Liquid mating
AX columns (Machery-Nagel). Because the primers used in this was performed in LB broth and incubated overnight at 37⬚C. Trans-
PCR anneal specifically in the 5
and 3
CS of class 1 integrons, conjugants were selected on LB agar plates containing both rifam-
the amplification product contained inserted gene cassettes flanked picin (50 mg/mL) and sulfamethoxazole (512 mg/mL) or rifampicin
on both sides with small parts of the CSs. (50 mg/mL) alone for counter selection. Five transconjugants were
Characterization of integrons by sequencing and restriction analyzed for each conjugation experiment. Donor strains and trans-
fragment–length polymorphism (RFLP) typing. To determine conjugants were tested, using the VITEK (bioMérieux) automated
whether different isolates carried identical integrons, the integron system, for their susceptibility pattern and, depending on the in-
content of every isolate was characterized. That is, each CS-PCR tegron contents, for sulfamethoxazole, streptomycin, aztreonam,
amplification product that had a unique size (number of base pairs) erythromycin, and chloramphenicol, using the agar diffusion
was sequenced. The size of the CS-PCR product of each new strain method. Tests for the presence and contents of the integrons were
was compared with the sizes of the already sequenced products. If done as described above. Transfer frequencies (Tfreq) were expressed
a CS-PCR product of the same size had been sequenced before, as the ratio of the number of transconjugants to the number of
both PCR products were compared by RFLP typing to discover recipients: Tfreq p [N(Smz RRif R)]/[N(SmzR⫹S Rif R )].
whether they contained the same integron content. If the CS-PCR Detection of b-lactamases by PCR amplification of blaSHV and
product of the unknown strain yielded the same RFLP pattern as blaTEM genes. TEM and SHV b-lactamases are the most com-

Table 1. Characteristics of integrons detected in isolates obtained during a hospital outbreak in 1996.
c
CS-PCR RFLP code
a b d
length, bp Gene cassettes Resistance phenotype Endonucleases of integron Source [reference]
1000 aadA2 Stm-Spm BclI, HindIII, HpaII I Cattle, swine [18]
1450 aadB/catB3 Gm-Km-Tm, Chl HpaII, NciI, Sau96I II Human [19]
2500 aacA7/oxa2a/aadA8 Amik-Net-Tm, b-lactams, EcoRI, PvuI, XhoII III
Stm-Spm
1000 aadA1a Stm-Spm BclI, HindIII, HpaII V Human, poultry, swine [15, 20–22]
2200 dfrA5/ereA2 TMP, Em EcoRI, MboII, XhoII VI
1550 dfrA1/aadA1a TMP, Stm-Spm HpaII, Sau96I VII Human, swine [15, 20]
1800 dfrA12/orfF/aadA2 TMP, unknown, Not done VIII Human [23–25]
Stm-Spm
2200 Unknown DraI, EcoRI, XhoII IX
2500 Unknown EcoRI, PvuI, XhoII X
2200 orfD/aacA4/catB8 Unknown, Amik- DraI, EcoRI, MboII, XhoII XI
Net-Tm, Chl

NOTE. Amik, amikacin; Chl, chloramphenicol; CS-PCR, conserved-segment polymerase chain reaction; Em, erythromycin; Gm, gentamicin; Km,
kanamycin; Net, netilmicin; RFLP, restriction fragment–length polymorphism; Spm, spectinomycin; Stm, streptomycin; Tm, tobramycin; TMP,
trimethoprim.
a
The order of genes in the variable region reads from the 5
CS to the 3
CS of the integron. GenBank accession nos. for genes per RFLP code are
as follows: I, X68227; II, U13880/U13880; III, U13880/M95287/AF326210; V, X12870; VI, X12868/AF326209; VII, X00926/X12870; VIII, Z21672/
X68227; and XI, X68227/M55547/AF227506.
b
Endonucleases used for RFLP typing.
c
Each unique restriction pattern obtained by RFLP typing is marked with a different Roman numeral.
d
Sources from which the integrons were previously isolated, based on the alignment of nucleotide sequences.
52 Leverstein–van Hall et al. JID 2002;186 (1 July)

mon plasmid-determined b-lactamases found in the Enterobacter- automated sequencer (Applied Biosystems); and was aligned with
iaceae. To assess whether and to what extent plasmid-determined the gene sequences available from GenBank.
b-lactamases were cotransferred with integrons, we determined the
presence of blaSHV and blaTEM genes in all the strains included
in the conjugation experiments. Whole-cell DNA was prepared for Results
amplification by boiling the cells for 5 min. The primers SHV-Nhe-
Presence of integrons. Eighty-six isolates were included in
F (5
-GAGCGAAAGATCCACTATCG-3
) and SHV-Nhe-R (5
-
the study. Twenty-six of the 56 patients colonized or infected
GTATCCCGCAGATAAATCA-3
) were designed for the ampli-
fication of the 526-bp blaSHV gene–specific fragment (GenBank with MRE carried 46 isolates harboring at least 1 integron.
accession no. AF148850 [SHV1], bp 267–287 and bp 790–771, re- These included 19 isolates of E. coli (17 genotypes), 12 K. pneu-
spectively). The PCR conditions were as follows: 35 cycles of 1 moniae (5 genotypes), 1 K. oxytoca, 9 C. freundii (3 genotypes),
min at 94⬚C, 1 min at 55⬚C, and 30 s at 72⬚C. PCR amplification 2 P. mirabilis, 2 E. cloacae, and 1 E. aerogenes. The E. coli
of the blaTEM gene was done as described by Arlet et al. [16]. isolated from the staff member and the K. pneumoniae isolated
Identification of blaSHV and blaTEM genes. The blaSHV and from the soap dispenser also contained an integron. The re-
blaTEM genes were identified to determine whether they encoded maining 40 isolates did not yield a CS-PCR product. These
for broad-spectrum b-lactamases or ESBLs. Tests for the identi- included isolates of 10 E. coli (10 genotypes), 8 K. pneumoniae

Downloaded from jid.oxfordjournals.org by guest on January 31, 2011

fication of blaSHV and blaTEM genes were performed on all strains (3 genotypes), 1 C. freundii, 1 E. cloacae, and 20 K. oxytoca.
included in the conjugation experiments, except for the blaTEM
Of the 20 K. oxytoca isolates, 19 were found to have the same
genes of the transconjugants. To identify the SHV-encoded b-lac-
genotype as that identified on the brush used to clean the mea-
tamases, PCR amplicons were digested with restriction endonucle-
ase NheI. SHV genes containing an NheI restriction site encode suring glass needed for parenteral feeding.
for ESBLs; the site is absent in chromosomally located SHV-1 genes Susceptibility testing of strains. Susceptibility testing of the
not encoding ESBLs [17]. To identify the blaTEM gene, the am- 46 isolates containing an integron gave the following results:
plified DNA was purified, using MicroSpin G-25 columns (Phar- 96% of the isolates were resistant or intermediately susceptible
macia); was sequenced on both DNA strands, using an ABI 377 to gentamicin, 83% to tobramycin, 87% to cotrimoxazole, 93%

Table 2. Distribution of integrons among multiresistant gram-negative bacilli isolated during a hospital outbreak in 1996.
Patient Escherichia Klebsiella Klebsiella Citrobacter Enterobacter Enterobacter Proteus
or staff Source coli pneumoniae oxytoca freundii aerogenes cloacae mirabilis

1 Rectum II
2 Rectum IIIa
3 Urine IIId
— Soap dispenser IIId
4 Rectum IIIf
5 Urine IIIf
6 Rectum/throat IIIf
7 Sputum IIIf
8 Urine V
9 Rectum V
10 Rectum V
Staff Rectum V
11 Urine V
12 Rectum V V
13 Urine/sputum VI VI VI
14 Rectum VII
15 Throat VIII
16 Rectum/throat I ⫹ II
17 Urine II ⫹ XI
18 Urine/rectum IIb, I ⫹ IIb
19 Rectum I ⫹ IIc III
20 Urine/rectum IIIa IIId, Id
21 Sputum/throat IIId IIIf, IX ⫹ Xg
22 Urine VII IX ⫹ Xg
23 Sputum VII, II ⫹ XIc
24 Sputum I ⫹ II ⫹ XIc I ⫹ II ⫹ XIe
25 Rectum I ⫹ II, I ⫹ II II ⫹ XI
26 Rectum III IIId, I ⫹ II ⫹ XIe, IIIf
III, II ⫹ XIe, II ⫹ III ⫹
XIe

NOTE. Roman numerals represent the unique restriction patterns obtained by restriction fragment–length polymorphism (RFLP) typing. GenBank
accession nos. for genes per RFLP code are as follows: I, X68227; II, U13880/U13880; III, U13880/M95287/AF326210; V, X12870; VI, X12868/AF326209;
VII, X00926/X12870; VIII, Z21672/X68227; and XI, X68227/M55547/AF227506. Isolates with identical genotypes are marked with the same superscript
letter (a–g).
JID 2002;186 (1 July) Transfer of Integron-Mediated Resistance Genes 53

patients (patients 12, 13, 20, 21, and 24–26) were colonized or
infected with 1 1 species of GRE sharing identical (combina-
tions of) integrons. Patients 25 and 26 also harbored strains of
the same species with different PFGE types but with identical
(combinations of) integrons. The strains isolated from patients
13, 20, and 24 were obtained from the same clinical site (urine
obtained from a Foley catheter and sputum). The CS-PCR
products of 8 consecutive isolates obtained from patient 26
during a 3.5-month hospital stay are shown in figure 1 as an
example of horizontal transfer.
The third observation was that different isolates of epide-
miologically linked patients carry the same integron. Integron
VI was restricted to the strains isolated from patient 13, inte-
Figure 1. Evidence of horizontal transfer, based on conserved-seg-
ment polymerase chain reaction amplification products of 8 consecutive gron VIII was restricted to patient 15, and 1 integron combi-
isolates obtained from patient 26 during a 3.5-month hospital stay in nation (IX ⫹ X) was detected in a unique strain shared by 2

Downloaded from jid.oxfordjournals.org by guest on January 31, 2011

1996 (also see table 2). Roman numerals represent the unique restriction patients (patients 21 and 22). Analysis of the distribution of
patterns obtained by restriction fragment–length polymorphism the remaining 6 integrons revealed 6 clusters of epidemiologi-
(RFLP) typing. GenBank accession nos. for genes per RFLP code are
cally linked patients. Integrons III and XI, respectively, were
as follows: I, X68227; II, U13880/U13880; III, U13880/M95287/
AF326210; and XI, X68227/M55547/AF227506. Lanes: 1, Citrobacter found in 6 and 4 unique strains distributed over 3 and 4 species
freundii f (collected 27 July); 2, Klebsiella pneumoniae d (collected 27 collected from 10 and 5 patients. Integron III was also isolated
July); 3, Escherichia coli (collected 8 August); 4, susceptible E. coli from a soap dispenser in the staff refreshment room. For both
(collected 19 September); 5, K. pneumoniae e (collected 27 September); integrons, all patients could be linked to each other by their
6, K. pneumoniae (collected 15 October); 7, K. pneumoniae e (collected
overlapping dates of hospitalization on the same wards. Inte-
11 November); 8, K. pneumoniae e (collected 11 November); and 9, 1-
kb DNA marker. Isolates with identical genotypes are marked with grons I and II, respectively, were found in 4 and 7 unique strains
the same superscript letter. Letters correspond to those in table 2. distributed over 3 and 4 species collected from 6 and 8 patients.
All of these patients had overlapping dates of hospitalization,
to ampicillin, 56% to piperacillin, 80% to amoxicillin-clavulanic and most of them had been in the same ward. Integron V was
acid, 49% to cephalotin, 49% to cefuroxime, 23% to ceftriaxone, found in 6 unique strains distributed over 4 species collected
33% to ceftazidime, and 2% to ciprofloxacin. Moreover, 43% from 5 patients and the staff member. These isolates were ob-
(9/21 strains tested) of the E. coli and Klebsiella species pro- tained within a 4-month period in 3 different wards. Although
duced ESBLs. Susceptibility testing of the 40 isolates without these patients had overlapping dates of hospitalization, only 2
a CS-PCR product gave similar results. had been on the same ward at the same time. Integron VII was
Characterization of integrons. Analysis of the amplification found in 3 different E. coli strains collected from 3 patients
products revealed 10 different integrons. The content and order within a 2-month period. These patients stayed in different
of the resistance genes inserted in the integrons are presented wards, but 2 shared a 6-week period of hospitalization.
in table 1 [15, 18–25]. Fifteen different gene cassettes, encoding Resistance transfer and susceptibility testing of transconju-
resistance to aminoglycosides, b-lactams, chloramphenicol, tri- gants. Eleven strains sharing identical (combinations of) in-
methoprim, and erythromycin, were detected. Two pairs of in- tegrons, obtained from 7 patients colonized or infected with 11
tegrons shared identical gene cassettes: integrons VIII and I species of GRE, were selected for the conjugation experiments.
shared aadA2, and integrons V and VII shared aadA1a. Two All of the strains included yielded transconjugants (table 3).
CS-PCR amplification products were not sequenced. RFLP The transfer frequencies varied from 1 to 3 ⫻ 10⫺2 transcon-
typing of these products, however, showed clear discriminatory jugants per recipient cell for E. coli, K. oxytoca, and E. cloacae;
patterns (RFLP codes IX and X). from 4 to 8 ⫻ 10⫺3 transconjugants per recipient cell for C.
Evidence of horizontal transfer and distribution of integrons freundii; and from 10⫺2 to 10⫺4 transconjugants per recipient
among clinical isolates. The hypothesis was that maximal cir- cell for K. pneumoniae. Resistance to sulfamethoxazole, am-
cumstantial evidence for horizontal transfer would be obtained picillin, cotrimoxazole, gentamicin, tobramycin, and strepto-
if 3 observations could be made. The first observation was that mycin was transferred en bloc from all donor strains. Integrons
isolates with an identical genotype harbor different (combi- were transferred in all but 2 experiments. In these 2 experiments,
nations of) integrons. Four unique strains harboring different part of the transconjugants missed one of the multiple integrons
(combinations of) integrons were found: E. coli b⫹c and K. present in the donor (transconjugants 11 and 13).
pneumoniae d⫹e (table 2). TEM genes were detected in 4 donor strains obtained from
The second observation was that different species or strains patients 13, 24, and 25. The TEM gene present in the E. coli
collected from the same patient harbor the same integron. Seven from patient 13 was identified as a TEM-1 gene encoding broad-
54 Leverstein–van Hall et al. JID 2002;186 (1 July)

Table 3. Transfer frequency of integrons and the resistance patterns of donor strains and transconjugants in a study of multidrug-resistant
Enterobacteriaceae in a clinical setting.
a
Drug
Integron
RFLP Transfer Amox/ ESBL
b
Patient, strain code frequency Smz Cmz Gm Tm Stm Amp Clv Cfur Ctri Czid Atm Em Chl Etest TEM/SHV

13
Escherichia coli VI R R R R R R I S I R NI TEM1
Transconjugant 1 VI 3 ⫻ 10⫺2 R R R R R R I I I NI TEM
21
Citrobacter freundii f III R R R R R R R R S R S ND Neither
Transconjugant 2 III 4 ⫻ 10⫺3 R R R R I R S S S ⫺ Neither
20
Klebsiella pneumoniaed III R R R R R R S S S S S ⫺ Neither
Transconjugant 3 III 1 ⫻ 10⫺2 R R R R I R
26
K. pneumoniaed III R R R R R R R S S S I ⫹ SHVESBL
Transconjugant 4 III 1 ⫻ 10⫺4 R R R R I R S S ⫺ Neither
C. freundiif III R R R R R R R I R I I ND SHVESBL

Downloaded from jid.oxfordjournals.org by guest on January 31, 2011

Transconjugant 5 III 8 ⫻ 10⫺3 R R R R R R R I S S S ⫹ SHVESBL
E. coli III R R R R R R R I S S I NI SHVESBL
c
Transconjugant 6 III 2 ⫻ 10⫺2 R R R I I R S S S ⫺ Neither
Transconjugant 7 — — R R R R R R S I ⫹ SHVESBL
12
Enterobacter cloacae V R S R S R R S R S S ND Neither
Transconjugant 8 V 3 ⫻ 10⫺2 R R I S S ND ND
K. pneumoniae V R R R R R R R S S S ⫺ SHVnon-ESBL
Transconjugant 9 V 1 ⫻ 10⫺3 R R R R I R I R ⫺ Neither
24
d
K. pneumoniae I ⫹ II ⫹ XI R R R R R R R R S S S ⫹ TEMESBL /
SHVnon-ESBL
c
Transconjugant 10 I ⫹ II ⫹ XI 1 ⫻ 10⫺3 R R R R I R I R I ⫹ TEM
Transconjugant 11 II ⫹ XI — R R R R S R I S S ⫺ TEM
25 R
d
Klebsiella oxytoca II ⫹ XI R S R R I R R R S S S ⫹ TEMESBL
c
Transconjugant 12 II ⫹ XI 1 ⫻ 10⫺2 R I R S R I R I ⫹ TEM
Transconjugant 13 II — R R R S R I R I ⫹ TEM
d
E. coli I ⫹ II R R R R R R I R S S I ⫹ TEMESBL /
SHVESBL
Transconjugant 14 I ⫹ II 2 ⫻ 10⫺2 R R R R R R I R I S ⫹ TEM/SHVESBL

NOTE. The GenBank accession nos. for genes per restriction patterns obtained by restriction fragment–length polymorphism (RFLP) typing are as follows: I,
X68227; II, U13880/U13880; III, U13880/M95287/AF326210; V, X12870; VI, X12868/AF326209; VII, X00926/X12870; VIII, Z21672/X68227; and XI, X68227/M55547/
AF227506. Isolates with identical genotypes are marked with the same superscript letter (d or f). TEM and SHV are the plasmid-determined b-lactamases most
commonly found in Enterobacteriaceae. ESBL, extended-spectrum b-lactamase; ND, not done; Neither, neither TEM nor SHV; NI, not interpretable.
a
Amox/Clv, amoxicillin/clavulanate; Amp, ampicillin; Atm, aztreonam; Cfur, cefuroxime; Chl, chloramphenicol; Cmz, cotrimoxazole; Ctri, ceftriaxone; Czid,
ceftazidime; Em, erythromycin; Gm, gentamicin; Smz, sulfamethoxazole; Stm, streptomycin; Tm, tobramycin. I, intermediately susceptible; R, resistant; S, susceptible.
Underlined resistance profiles (R) are encoded by the integron carried by the donor strain.
b
Minus (⫺), negative test result; plus (⫹), positive test result.
c
Combined transfer frequencies of transconjugants 6 and 7, 10 and 11, and 12 and 13.
d
Tentative identification.

spectrum b-lactamases. The TEM genes present in the other fer contributed extensively to the high endemic level of MRE
strains, however, have not been reported. Further investigation seen in the neurology and neurosurgery wards of our hospital.
is needed to determine the character of the b-lactamases they Isolates with an identical genotype were found to harbor dif-
encode. TEM genes were transferred in all of the conjugation ferent (combinations of) integrons at different collection times,
experiments. SHV-ESBL genes were detected in 4 donor strains indicating the acquisition or loss of integrons. Furthermore, 7
obtained from patients 25 and 26 and were transferred in nearly patients carried multiple species harboring identical integrons,
all the tests. Only transconjugants 4 and 6 missed the SHV-ESBL very likely the result of the interspecies transfer of integrons
gene present in the donor strain. The non-transferred SHV genes within the patient. Of interest, the strains from 3 of the patients
detected in the K. pneumoniae isolates of patients 12 and 24 were had been collected from the same clinical site (sputum or cath-
probably chromosomally located SHV-1 genes [26]. eter urine), suggesting that the transfer of resistance genes may
not take place just in the intestinal tract, as recently reported
Discussion [11]. Furthermore, analysis of the distribution of the integrons
The results of this study provide strong circumstantial evi- revealed 6 clusters of patients carrying different strains with
dence in support of the assumption that horizontal gene trans- the same integron. The spread of integrons III and XI was very
JID 2002;186 (1 July) Transfer of Integron-Mediated Resistance Genes 55

likely the result of clonal spread and horizontal transfer within the persistence of these genes (e.g., the administration of these
the hospital. Patients harboring these integrons had overlap- antimicrobial agents to food-producing animals) [20, 28].
ping dates of hospitalization in the same wards, and they often The results of this study also illustrate the limitations of
shared identical strains. In addition, neither integron has ever conventional genotyping in the management of an outbreak of
been isolated outside our hospital. With regard to the remaining MRE. The results of RAPD typing in 1996 and the subsequent
4 clusters, however, it cannot be totally excluded that some of PFGE typing performed in this study indicate that most isolates
the patients were carriers of the integrons prior to hospital were unrelated. Analysis of the integron contents of the same
admission. Because these latter integrons have been isolated strains, however, suggests the opposite in many cases. Fur-
previously from humans [4, 19, 23–25] and food-producing an- thermore, as part of the strategy to control the high level of
imals [18, 20–22], their presence in this study may reflect their MRE in 1996, rectal swab samples from the medical staff were
prevalence in the community. cultured in search of environmental sources. The culture from
The results of the conjugation experiments are highly sup- a neurologist yielded a gentamicin-resistant E. coli. Not much
portive of the hypothesis that horizontal transfer of resistance attention was paid to that strain at the time, because it pos-
genes takes place in vivo. For example, resistance patterns that sessed a unique RAPD type not seen in any of the patients.
were very likely the result of horizontal transfer in vivo were Later analysis of the integron content of the strain, however,

Downloaded from jid.oxfordjournals.org by guest on January 31, 2011

shown to be transferred at very high frequency rates in vitro. identified this person as a potential source for one cluster of
Of interest, the complete resistance patterns (except for known integrons.
chromosomally located resistance determinants), including In conclusion, this study has shown that the horizontal trans-
those encoded by integrons and TEMs or SHVs, were trans- fer of antimicrobial resistance genes can occur very efficiently
ferred in their entirety in most of the conjugation experiments, and at a high rate among Enterobacteriaceae in a nosocomial
suggesting a very efficient mechanism of transferring packages setting. This study has also demonstrated that integron typing
of resistance genes. The results from some experiments also can be a useful tool for studying the dissemination of resistance
show that part of the transconjugants missed one of the multiple genes among gram-negative bacteria. Further studies are
integrons present in the donor. This implies that the various needed, however, to determine whether antibiotic policies or
integrons of the donor strain were located on different mobile other measures can halt or lower the amount of horizontal
elements. transfer that occurs in a hospital.
An unexpectedly high number (10) of different integrons were
involved during this outbreak. Two pairs of integrons shared
References
identical gene cassettes: integrons VIII and I shared aadA2, and
integrons V and VII shared aadA1a. It is possible, therefore, 1. Cross AS, Opal S, Kopecko DJ. Progressive increase in antibiotic resistance
that integrons VIII and I or V and VII evolved from one an- of gram-negative bacterial isolates. Walter Reed Hospital, 1976 to 1980:
specific analysis of gentamicin, tobramycin, and amikacin resistance. Arch
other by the acquisition or deletion of genes. However, since
Intern Med 1983; 143:2075–80.
all 4 integrons have been described before, these mutations may 2. Holmberg S, Solomon S, Blake P. Health and economic impact of anti-
have occurred some time prior to our study. These findings microbial resistance. Rev Infect Dis 1987; 9:1065–78.
support recent views that integrons are rather stable structures 3. Cohen ML. Epidemiology of drug resistance: implications for a post anti-
[27]. The predominant means of acquiring resistance, therefore, microbial era. Science 1992; 257:1050–5.
4. Hobson R, MacKenzie FM, Gould IM. An outbreak of multiply-resistant
may be the acquisition of new integrons, often in association Klebsiella pneumoniae in the Grampian region of Scotland. J Hosp Infect
with other transferable resistance genes, rather than the ac- 1996; 33:249–62.
quisition of new resistance genes by resident integrons. This is 5. Leverstein-van Hall MA, Fluit AC, Blok HEM, et al. Control of nosocomial
supported by our observation that, over time, the integron con- multiresistant Enterobacteriaceae using a temporary restrictive antibiotic
agent policy. Eur J Clin Microbiol Infect Dis 2001; 20:785–91.
tent of K. pneumoniaee from patient 26 changed from II ⫹ XI
6. Rowe-Magnus DA, Mazel D. Resistance gene capture. Curr Opin Microbiol
to II ⫹ III ⫹ XI, and the integron content of an E. coli strain 1999; 2:483–8.
from patient 18 changed from II to I ⫹ II. 7. Hall RM. Mobile gene cassettes and integrons: moving antibiotic resistance
Seven of 8 integrons harbored genes encoding resistance to genes in gram-negative bacteria. Antibiotic resistance: origins, evolution,
either chloramphenicol or streptomycin and spectinomycin, an- selection and spread. Wiley, Chichester, UK: Ciba Foundation Symposium
1997; 207:192–205.
tibiotics that were rarely, if ever, prescribed in our hospital in
8. Hall RM, Collis CM. Antibiotic resistance in gram-negative bacteria: the
the last 2 decades. This finding supports the recent insight that role of gene cassettes and integrons. Drug Resist Updat 1998; 1:109–19.
withdrawal of a certain antimicrobial drug does not automat- 9. Martinez-Freijo P, Fluit AC, Schmitz FJ, Grek VSC, Verhoef J, Jones ME.
ically lead to the disappearance of resistance to that agent [28, Class I integrons in gram-negative isolates from different European hos-
pitals and association with decreased susceptibility to multiple antibiotic
29]. The persistence of these genes is most likely the result of
compounds. J Antimicrob Chemother 1998; 42:689–96.
their structural association with other resistance genes, such as 10. Fluit AC, Schmitz FJ. Class 1 integrons, gene cassettes, mobility, and epi-
the sulI gene, for which selective pressure has continued over demiology. Eur J Clin Microbiol Infect Dis 1999; 18:761–70.
the past years. Other factors, however, may also contribute to 11. Shoemaker NB, Vlamakis H, Hayes K, Salyers AA. Evidence for extensive
56 Leverstein–van Hall et al. JID 2002;186 (1 July)

resistance gene transfer among Bacteroides spp. and among Bacteroides 21. Bass L, Liebert CA, Lee MD, et al. Incidence and characterization of inte-
and other genera in the human colon. Appl Environ Microbiol 2001; 67: grons, genetic elements mediating multiple-drug resistance, in avian Es-
561–8. cherichia coli. Antimicrob Agents Chemother 1999; 43:2925–9.
12. Wiener J, Quinn JP, Bradford PA, et al. Multiple antibiotic–resistant Kleb- 22. Baggesen DL, Sandvang D, Aarestrup FM. Characterization of Salmonella
siella and Escherichia coli in nursing homes. JAMA 1999; 281:517–23. enterica serovar typhimurium DT104 isolated from Denmark and com-
13. National Committee for Clinical Laboratory Standards. Methods for dilution parison with isolates from Europe and the United States. J Clin Microbiol
antimicrobial susceptibility tests for bacteria that grow aerobically: ap- 2000; 38:1581–6.
proved standard. In: NCCLS document M. 2d ed. (7-A3). Villanova, PA: 23. Dalsgaard A, Forslund A, Petersen A, et al. Class I integron-borne, multiple-
National Committee for Clinical Laboratory Standards, 1999. antibiotic resistance encoded by a 150-kilobase conjugative plasmid in
14. Tenover FC, Arbeit RD, Goering RV, et al. Interpreting chromosomal DNA epidemic Vibrio cholerae O1 strains isolated in Guinea-Bissau. J Clin Mi-
restriction patterns produced by PFGE: criteria for bacterial strain typing. crobiol 2000; 38:3774–9.
J Clin Microbiol 1995; 33:2233–9. 24. Guerra B, Soto SM, Arguelles JM, Mendoza MC. Multidrug resistance is
15. Lévesque C, Piché L, Larose C, Roy PH. PCR mapping of integrons reveals mediated by large plasmids carrying a class 1 integron in the emergent
several novel combinations of resistance genes. Antimicrob Agents Che- Salmonella enterica serotype (4.5,12:i:-). Antimicrob Agents Chemother
mother 1995; 39:185–91. 2001; 45:1305–8.
16. Arlet G, Brami G, Decré D, et al. Molecular characterisation by PCR– 25. Heikkila E, Skurnik M, Sundstrom L, Huovinen P. A novel dihydrofolate
restriction fragment length polymorphism of TEM ß-lactamases. FEMS reductase cassette inserted in an integron borne on a Tn21-like element.
Microbiol Lett 1995; 134:203–8. Antimicrob Agents Chemother 1993; 37:1297–304.

Downloaded from jid.oxfordjournals.org by guest on January 31, 2011

17. Nuësch-Inderbinen MT, Hächler H, Kayser, FH. Detection of genes coding 26. Leung M, Shannon K, French G. Rarity of transferable ß-lactamase pro-
for extended-spectrum SHV beta-lactamases in clinical isolates by a mo- duction by Klebsiella species. J Antimicrob Chemother 1997; 39:737–45.
lecular genetic method, and comparison with the E test. Eur J Clin Mi- 27. Martinez-Freijo P, Fluit AC, Schmitz FJ, Verhoef J, Jones ME. Many class
crobiol Infect Dis 1996; 15:398–402. I integrons comprise distinct and stable structures occurring in different
18. Sandvang D, Aarestrup FM. Characterization of aminoglycoside resistance species of Enterobacteriaceae isolated from widespread geographic regions
genes and class 1 integrons in porcine and bovine gentamicin-resistant in Europe. Antimicrob Agents Chemother 1999; 43:686–9.
Escherichia coli. Microb Drug Resist 2000; 6:19–27. 28. Chiew F, Hall LMC, Livermore DM. Can susceptibility to an antimicrobial
19. Tosini F, Visca P, Luzzi I, et al. Class 1 integron-borne multiple-antibiotic be restored by halting its use? The case of streptomycin versus Entero-
resistance carried by IncFI and IncL/M plasmids in Salmonella enterica bacteriaceae. J Antimicrob Chemother 1998; 41:247–51.
serotype typhimurium. Antimicrob Agents Chemother 1998; 42:3053–8. 29. Enne VI, Livermore DM, Stephens P, Hall LMC. Persistence of sulpho-
20. Sunde M, Sørum H. Characterization of integrons in Escherichia coli of the namide resistance in Escherichia coli in the UK despite national prescribing
normal intestinal flora of swine. Microb Drug Resist 1999; 5:279–87. restriction. Lancet 2001; 357:1325–8.