Characterization of Genetic Diversity of Carbapenem-Resistant Acinetobacter baumannii Clinical Strains Collected from 2004 to 2007

Abstract

Genotypes of carbapenem-resistant Acinetobacter baumannii collected by the clinical microbiology laboratory of a university hospital in Chicago, IL, between 2004 and 2007 were analyzed by pulsed-field gel electrophoresis. A single genotype established predominance after being introduced in 2005. Analysis of carbapenemases by PCR revealed that imipenem resistance but not meropenem resistance was associated with the presence of bla_OXA-23 and bla_OXA-40 genes.

Carbapenems are a mainstay treatment for infections caused by multidrug-resistant Acinetobacter baumannii (2, 12, 13). Of concern, a dramatic increase of A. baumannii strains with decreased susceptibility to carbapenems has been observed in recent years both nationally (16) and locally (M. H. Scheetz, T. R. Zembower, M. K. Bolon, M. Malczynski, S. Hayes, J. Warren, and G. A. Noskin, presented at the 2006 meeting of the Infectious Diseases Society of America, Toronto, Ontario, Canada). This phenotypic observation is likely associated with changes in porin proteins, development of efflux pumps, modification of penicillin-binding proteins, and production of β-lactamase (15). Among the four Ambler classes of β-lactamases, class B enzymes (metallo-β-lactamases [MBLs]) and several class D enzymes (carbapenem-hydrolyzing oxacillinases [CHDLs]) hydrolyze carbapenems and produce resistance in the organism (3). Three groups of MBLs, IMP, VIM, and SIM, have been identified in A. baumannii (19). IMP and VIM β-lactamases show much stronger hydrolytic activity against carbapenems than do CHDLs while SIM confers only a moderate level of carbapenem resistance. Clinical A. baumannii strains with MBLs have been identified in Europe, Asia, and South America but not in the United States (15). Three groups of CHDLs have been reported in A. baumannii. The OXA-23-like group is comprised of OXA-23, OXA-27, and OXA-49. OXA-24, OXA-25, OXA-26, and OXA-40 constitute the second group, which shares 60% homology with the OXA-23-like group (15). OXA-58, sharing less than 50% identity with the other two groups, is the only member that has been identified in the third group. Contrasting the genetic heterogeneity among different groups of the OXA carbapenemases, members in the same group share more than 95% homology (3). All the variants of CHDLs were initially identified in carbapenem-resistant clinical A. baumannii isolates. The direct involvement of oxacillinases in the carbapenem resistance of A. baumannii was further demonstrated in the experimental systems. Introduction of the naturally occurring plasmids carrying the bla_OXA-23 and bla_OXA-58 genes into a carbapenem-susceptible A. baumannii reference strain resulted in imipenem MICs being increased between three- and fivefold (9). Additionally, a carbapenem-resistant A. baumannii strain carrying OXA-40 became susceptible to both imipenem and meropenem when the bla_OXA-40 gene was disrupted (9).

Outbreaks associated with carbapenem-resistant A. baumannii have been reported worldwide (8). From April 2005 to December 2005, 15 patients in our hospital were infected with a single carbapenem-resistant A. baumannii strain. Genotype analysis of the carbapenem-resistant A. baumannii strains collected before and after the outbreak revealed a major change in the genetic diversity of this resistant organism.

Strain collection, strain identification, and susceptibility testing.

One hundred sixteen clinical strains (from 116 patients) of A. baumannii with an imipenem or meropenem MIC of ≥16 μg/ml collected by the clinical microbiology laboratory of Northwestern Memorial Hospital in Chicago, IL, between January 2004 and July 2007 were selected for the study. The strains were recovered from blood, urine, respiratory specimens, and various body fluids. Fifteen of the strains were recovered from outpatients, and the rest of the strains were from inpatients. Fifty-nine (51%) isolates were recovered from patients hospitalized in intensive care units. Species identification was performed with the Vitek 2 system (bioMerieux, France). Manual biochemical reactions were used when identification by the Vitek 2 system failed. Antimicrobial susceptibility was initially determined by the Vitek 2 system using the AST-GN09 card according to the manufacturer's instructions, and isolates were qualitatively reported as susceptible, intermediate, or resistant according to Clinical and Laboratory Standards Institute (CLSI) guidelines (5). Susceptibility results for imipenem and meropenem were confirmed with agar dilution in accordance with CLSI standards using imipenem and meropenem obtained from United States Pharmacopoeia (Rockville, MD) (4).

Molecular typing by PFGE.

Preparation of bacterial gel plugs and cell lysis were performed as previously described (18). Whole-cell DNA was digested in a 1% agarose gel with ApaI enzyme (New England Biolabs, Beverly, MA) at 25°C for 4 h. Electrophoresis was performed with a CHEF-DRII system (Bio-Rad Laboratories, Hercules, CA) over 18.5 h at 14°C with 5 to 13 s of linear ramping at 200 V. Pulsed-field gel electrophoresis (PFGE) patterns were interpreted visually by using the criteria proposed by Tenover et al. (17). Each PFGE type included strains with zero to six bands different.

Detection of bla_OXA-23, bla_OXA-40, bla_OXA-58, bla_IMP, bla_VIM, and bla_SIM genes.

Amplification of the genes encoding OXA-23-like, OXA-24-like, and OXA-58-like class D β-lactamases was carried out with PCR using primers OXA23-F (5′-GATCGGATTGGAGAACCAGA-3′), OXA23-R (5′-ATTCTGACCGCATTTCCAT-3′), OXA24-F (5′-GGTTAGTTGGCCCCCTTAAA-3′), OXA24-R (5′-AGTTGAGCGAAAAGGGGATT-3′), OXA58-F (5′-AAGTATTGGGGCTTGTGCTG-3′), and OXA58-R (5′-CCCCTCTGCGCTCTACATAC-3′) as previously described (20). The sizes of the corresponding PCR products were 501 bp, 246 bp, and 599 bp. Primers OXA23-IIF (5′-GATGTGTCATAGTATTCGTCG-3′), OXA23-IIR (5′-TCACAACAACTAAAAGCACTG-3′), OXA24-IIF (5′-GTACTAATCAAAGTTGTGAA-3′), and OXA24-IIR (5′-TTCCCCTAACATGAATTTGT-3′) were used to amplify the full-length gene for sequencing following the previously described conditions (1). An attempt to detect MBLs was performed by PCR amplification using primers IMP1-F (5′-CATGGTTTGGTGGTTCTTGT-3′), IMP1-R (5′-ATAATTTGGCGGACTTTGGC-3′), VIM2-F (5′-ATGTTCAAACTTTTGAGTAAG-3′), VIM2-R (5′-CTACTCAACGACTGAGCG-3′), SIM1-F (5′-TACAAGGGATTCGGCATCG-3′), and SIM1-R (5′-TAATGGCCTGTTCCCATGTG-3′) following previously described conditions (14).

Findings.

A total of 24 distinct genotypic patterns were identified (Fig. 1). In 2004, five different PFGE genotypes were detected in 10 strains of carbapenem-resistant A. baumannii. None of the five genotypes appeared in 2005. The carbapenem-resistant strains collected in 2005 exhibited diverse PFGE patterns. Among 42 strains of A. baumannii recovered, 13 different PFGE types were detected, with 21 (50%) strains being PFGE type A. After 2005, a dramatic increase in the predominance of PFGE type A was observed. In 2006, 32 out of 36 strains were PFGE type A. Twenty-four out of 28 carbapenem-resistant strains recovered in 2007 had genotypes of PFGE type A. Compared to 13 unique genotypes carried by carbapenem-resistant A. baumannii strains in 2005, only three non-type A PFGE types were recovered in 2006 and 2007 (Table 1).

FIG. 1.

Dendrogram (using the unweighted-pair group method using average linkages) for the Acinetobacter strains with distinct PFGE types (shown at right). Computer-assisted analysis of PFGE data was performed with the Advanced Analysis (version 5.10) and Database (version 1.13) modules of the Phoretix gel analysis software (Nonlinear USA, Inc., Durham, NC). Genetic similarity was calculated by the Dice coefficients. Values at the bottom of the figure represent percentages of genetic similarity.

TABLE 1.

Distribution of carbapenem-resistant A. baumannii strains from 2004 to 2007

Yr (January to December)	No. of strains	No. of PFGE types	% of strains
			PFGE type A	Resistant to imipenem
2004	10	5	0	0
2005	42	13	50	52
2006	36	4	89	92
2007	28	4	86	96

All the genotypes of carbapenem-resistant A. baumannii strains recovered in 2004 were resistant to meropenem but susceptible to imipenem. After the appearance of strains resistant to meropenem and imipenem in 2005, the rate of imipenem resistance dramatically increased from 52% in 2005 to 96% in 2007 (Table 1). This change corresponded to the spread of PFGE type A strains. Of the 82 strains displaying resistance to both meropenem and imipenem, 94% were PFGE type A (Table 2).

TABLE 2.

Association of imipenem resistance with the presence of bla_OXA-23 and bla_OXA-40 genes^a

PFGE type	Yr of recovery	No. of strains	Susceptibility to drug:		blaOXA gene
			Meropenem	Imipenem
A	2005 to 2007	77	R	R	blaOXA-40
B	2004	2	R	S	N
C	2004	4	R	S	N
D	2004	2	R	S	N
E	2004	1	R	S	N
F	2004	1	R	I	N
G	2005	4	R	S	N
H	2005	2	R	I	N
I	2005	1	R	S	N
J	2005	1	R	S	N
K	2005	1	R	R	blaOXA-23
L	2005	1	R	S	N
M	2005	1	R	S	N
N	2005	6	R	S	N
O	2005	1	R	S	N
P	2005	1	R	S	N
Q	2005	1	R	S	N
R	2005	1	R	S	N
S	2006	1	R	I	N
T	2006	2	R	I	N
U	2006	1	R	R	blaOXA-40
V	2007	1	R	I	N
W	2007	2	R	R	blaOXA-40
X	2007	1	R	R	blaOXA-23

^aAbbreviations: S, susceptible; I, intermediate; R, resistant; N, not present.

The presence of MBL and CHDL genes in the carbapenem-resistant A. baumannii strains was examined by PCR with primers for bla_OXA-23-like, bla_OXA-24-like, and bla_OXA-58-like oxacillinases and IMP-type, VIM-type, and SIM-type MBLs. MBLs and CHDLs including bla_OXA-23-like, bla_OXA-24-like, and bla_OXA-58-like were absent in all A. baumannii strains of different PFGE genotypes that were resistant to meropenem but susceptible to imipenem (Table 2). A bla_OXA-23-like gene was detected in the imipenem-resistant strains with PFGE types K and X, and the bla_OXA-24-like gene was detected in the imipenem-resistant strains with PFGE types A, U, and W. Sequencing the full-length gene PCR product identified them as bla_OXA-23 and bla_OXA-40.

Conclusions.

A major change in the genotype diversity of carbapenem-resistant A. baumannii was detected since the appearance of PFGE type A strains in 2005. These genetic changes explain the increase in qualitative resistance previously identified (M. H. Scheetz, T. R. Zembower, M. K. Bolon, M. Malczynski, S. Hayes, J. Warren, and G. A. Noskin, presented at the 2006 meeting of the Infectious Diseases Society of America, Toronto, Ontario, Canada). In contrast to carbapenem-resistant strains collected in 2004 and 2005 showing diverse PFGE types, more than 85% of carbapenem-resistant A. baumannii strains carried PFGE type A in 2006 and 2007 (Table 1). In 2004 and 2005, most strains were resistant to meropenem but susceptible to imipenem. A transition to the type A strains led to high-level resistance to both meropenem and imipenem. The meropenem-resistant, imipenem-susceptible strains lacked MBLs and CHDLs, but all the imipenem-resistant strains with different PFGE genotypes carried the bla_OXA-23 or bla_OXA-40 gene.

Analysis of the molecular epidemiology of carbapenem-resistant A. baumanni in the various parts of the world indicated a considerable degree of geographic specificity in the spread of various CHDLs (15). The bla_OXA-40 gene has been identified in the outbreak strains recovered in several regions of Europe (6, 7, 11) from 1999 to 2004 and lately in the United States (10). The dramatic increase in carbapenem-resistant A. baumannii strains in our hospital likely resulted from a local epidemic. Our study showed the bla_OXA-40 gene present in the dominant clonal pattern of meropenem- and imipenem-resistant A. baumannii strains as well as in two other sporadic genetically unrelated strains. Characterization of the sequences surrounding the bla_OXA-40 gene in a previous study suggested that the gene was localized on the chromosome and was not associated with any transferable elements (15). However in a recent study, Lolans et al. demonstrated that a bla_OXA-40-specific probe was able to hybridize to the plasmid extracted from a carbapenem-resistant A. baumannii strain, raising the possibility that the bla_OXA-40 gene may localize on a plasmid (10). Further study is required to determine if dissemination of the bla_OXA-40 gene is the result of the strains coming from the same lineage or the result of strains exchanging a transposable genetic element.

The clinical emergence of imipenem resistance in the presence of CHDL bla_OXA-23 and bla_OXA-40 genes provided further evidence for the direct contribution of OXA-23 and OXA-40 to the development of imipenem resistance in A. baumannii clinical isolates. The absence of MBLs and CHDLs in the meropenem-resistant and imipenem-susceptible strains suggested that meropenem resistance was mediated by a different mechanism in the analyzed isolates. In addition to β-lactamases, lost or modulated outer membrane porin proteins and modifying penicillin-binding proteins were reported to be involved in the development of carbapenem resistance (15). It is possible that these mechanisms could have conferred differential resistance to meropenem and imipenem.