Acinetobacter radioresistens as a Silent Source of Carbapenem Resistance for Acinetobacter spp.

Abstract

Carbapenem resistance results mostly from the expression of acquired carbapenem-hydrolyzing oxacillinases in Acinetobacter baumannii. The bla_OXA-23 oxacillinase gene is increasingly reported worldwide and may represent an emerging threat. Our goal was to identify the progenitor of that carbapenemase gene. A collection of 50 Acinetobacter sp. strains corresponding to several Acinetobacter species was screened for bla_OXA-23-like genes by PCR and hybridization techniques. Five Acinetobacter radioresistens isolates that were susceptible to carbapenems harbored chromosomally encoded bla_OXA-23-like genes. A similar plasmid backbone was identified in several bla_OXA-23-positive A. baumannii and A. radioresistens isolates, further strengthening the vectors of exchanges for these bla_OXA-23-like genes. Therefore, A. radioresistens, a commensal bacterial species which is identified on the skin of hospitalized and healthy patients (a property shared with A. baumannii), was identified as the source of the bla_OXA-23 gene.

Carbapenem resistance in Acinetobacter baumannii is increasingly reported and leads to difficult-to-treat nosocomial infections (25, 28, 33, 38). Carbapenem-hydrolyzing class D β-lactamases (CHDLs) represent the main mechanism of resistance to carbapenems in Acinetobacter spp. (4, 25). Three main groups of acquired CHDLs have been identified in A. baumannii and are of the OXA-23, OXA-40, and OXA-58 types. These oxacillinases confer reduced susceptibility or resistance to carbapenems once they are expressed in A. baumannii (13). Several studies have reported that those oxacillinase genes are encoded by transferable plasmids and likely possess restricted host ranges since attempts to transfer the oxacillinase genes from A. baumannii to Escherichia coli as the recipient strain failed (3, 13, 32). The β-lactamase OXA-23 (formerly ARI-1) was identified first in Scotland and was found to be a source of transferable resistance to imipenem in A. baumannii (10). Since then, OXA-23 producers have been identified as sources of nosocomial outbreaks worldwide, including Brazil, Colombia, the United Kingdom, Korea, Tahiti, and China (6, 9, 19, 23, 40, 42, 44). In addition, OXA-23-producing and carbapenem-resistant Acinetobacter sp. isolates have been isolated from soldiers with infections returning from Iraq and are a possible source of further spread in the United States and the United Kingdom (15, 20). The origin (reservoir) of those CHDL genes remains unknown.

Recently, we identified acquired bla_OXA-23 genes located in peculiar transposon structures, namely, Tn2006 (ISAba1 linked) and Tn2007 (ISAba4 linked) from carbapenem-resistant A. baumannii isolates (7). The Tn2006 structure consists of two copies of the same ISAba1 insertion sequence bracketing the bla_OXA-23 gene together with an ATPase-encoding gene (Fig. 1) (7). Transposon Tn2007 contains a single copy of ISAba4 associated with the same ATPase-encoding gene located downstream of the bla_OXA-23 gene (7). This ATPase-encoding gene has been identified in 13 of 13 bla_OXA-23-positive A. baumannii isolates studied (7).

FIG. 1.

Schematic maps of composite transposon Tn2006 carrying the bla_OXA-23 gene identified on a plasmid in A. baumannii isolates (7) (A) and the sequences identified in the five different A. radioresistens isolates (B). The sequences identical between Tn2006, Tn2007, and the A. radioresistens isolates are indicated by two vertical lines. ATPase is the gene encoding the putative ATPase, and sulf is the gene for a plasmid-borne sulfonamide resistance gene truncated by the insertion of Tn2006. orf1 is the gene encoding a putative protein truncated by the insertion of Tn2007, and mobA is the gene encoding a putative mobilization protein. Sequences indicated by dashes are unknown.

Therefore, considering that the bla_OXA-23 gene (i) has mostly been identified in Acinetobacter spp., (ii) has a GC content of 38% (which fits with that of the Acinetobacter sp. genes [12, 37]), (iii) was associated with a gene encoding an AAA ATPase that shared a high degree of identity with that of another related species (Acinetobacter baylyi [1]), and (iv) was likely mobilized by the ISAba1 insertion sequence that is widespread in Acinetobacter spp. (14, 35), we hypothesized that the bla_OXA-23 gene may originate from a species possibly belonging to the genus Acinetobacter. In addition, we hypothesized that the donor of the gene may share the same reservoir as the recipient A. baumannii isolate, i.e., the human skin. This prompted us to search for the progenitor of this emerging carbapenemase gene.

MATERIALS AND METHODS

Bacterial strains and patients.

Our screening panel included 50 Acinetobacter sp. strains that belonged to 14 different Acinetobacter species, including A. junii; A. johnsonii; A. haemolyticus; A. baylyi; A. lwoffii; A. radioresistens; A. schindleri; A. ursingii; and Acinetobacter genomospecies 3, 10, 13, 15, 16, and 17. Those strains were identified to the species level by using molecular techniques, as described by Dortet et al. (11). Reference strain A. radioresistens CIP103788 was from a cotton plant from Argentina (24). Four additional A. radioresistens strains were from skin and urinary tract specimens of patients from the Hôpital de Bicêtre (data not shown). bla_OXA-23-positive A. baumannii strain Ab13 was used as a control for the molecular experiments and the biochemical assays (7).

Molecular techniques.

A PCR-based screening for the bla_OXA-23 gene was performed with primers OXA-IMP1 and OXA-IMP2 (Table 1) (13). The AAA ATPase-encoding gene was amplified with primer ATPaseB3 (Table 1) in combination with primer OXA-IMP1 (13). The screening for insertion sequences ISAba1 and ISAba4 was performed with primers ISAba1A and ISAba1B and primers ISAba4A and ISAba4B, respectively (Table 1). The chromosomal location of the bla_OXA-23 gene was demonstrated by using the endonuclease I-Ceu-I technique, as described previously (21). Briefly, analysis of the I-Ceu-I-restricted fragments of whole-cell DNAs of the A. radioresistens isolates was performed by pulsed-field gel electrophoresis (PFGE) and gave five DNA fragments in each case. Then, transfer and Southern hybridization were performed as described previously (29, 34) with DNA probes specific for rRNA, consisting of a 1,504-bp PCR fragment specific for 16S rRNA genes (22) and a 840-bp internal PCR fragment specific for the bla_OXA-23 gene.

TABLE 1.

Sequences of primers used in the study

Primer name	Primer sequence (5′ to 3′)	Location, direction
PreOXA-23A	TTTCTATTSATCTGGTGTTTA	blaOXA-23 gene, forward external primer
PreOXA-23B	TTAGAGGTTTCTGTCAAGCTC	blaOXA-23 gene, reverse external primer
ATPaseB3	GCTTCATCCAGAAGCGTCCGG	ATPase gene, reverse primer
ISAba1A	ATGCAGCGCTTCTTTGCCAGG	tnpA gene of ISAba1, forward primer
ISAba1B	AATGATTGGTGACAATGAAG	tnpA gene of ISAba1, reverse primer
ISAba4A	ATTTGAACCCATCTATTGGC	tnpA gene of ISAba4, forward primer
ISAba4B	ACTCTCATATTTTTTCTT	tnpA gene of ISAba1, reverse primer
RepAFW	GAGAGATTTAGTTGTAAAGGACAATGC	repAci1 gene, forward primer
RepARV	CGACTCATAACATTTCGGATATTCCCATTA	repAci1 gene, reverse primer
OXA-IMP1	GCAAATAMAGAATATGTSCCa	blaOXA-23 and blaOXA-40 genes, forward internal primer
OXA-IMP2	CTCMACCCARCCRGTCAACCb	blaOXA-23 and blaOXA-40 genes, reverse internal primer

^aM is A or C, and S is G or C.

^bM is A or C, and R is A or G.

Cloning of the entire bla_OXA-23-like genes (amplified with primers PreOXA-23A and PreOXA-23B [Table 1]) was performed by using kanamycin-resistant plasmid pCR-BluntII-TOPO (Invitrogen, Life Technologies, Cergy-Pontoise, France), and the recombinant plasmids were transferred into E. coli TOP10 (Invitrogen). The selection of recombinant clones was done with plates containing kanamycin (30 μg/ml) and amoxicillin (30 μg/ml). The sequences of the cloned fragments were confirmed by double-strand sequencing.

Culture media and susceptibility testing.

The A. baumannii and A. radioresistens strains were grown on Mueller-Hinton agar plates and incubated overnight at 37°C. The MICs of the antibiotics were determined by Etest (AB Biodisk, Solna, Sweden) on Mueller-Hinton agar plates at 37°C. Carbapenemase activities were assayed by UV spectrophotometry with culture extracts of the A. radioresistens isolates obtained as described and with 100 μM benzylpenicillin or imipenem as the substrate (31). One unit of activity was defined as the amount of enzyme hydrolyzing 1 μmol of substrate per min per mg of protein.

Plasmid analysis.

Since the repAci1 gene encoding a novel replicase was recently identified from a plasmid carrying the gene encoding another CHDL (OXA-58) from an A. baumannii isolate, PCR was used to search for it among the bla_OXA-23-positive A. baumannii isolates and A. radioresistens strains by using primers RepAFW and RepARV (Table 1) (3). The specificity of the repAci1 PCR assay was confirmed by subsequent sequencing of the amplicons obtained from the A. baumannii and the A. radioresistens strains. The aim of that experiment was to determine whether plasmids circulating in A. baumannii could be also identified in A. radioresistens and whether they could thus be possible vehicles for gene exchanges between both species.

Nucleotide sequence accession number.

The nucleotide sequences of the bla_OXA-23-like genes from A. radioresistens have been deposited in the GenBank database under accession number EU131372.

RESULTS

Identification of progenitor.

A preliminary PCR screening gave a bla_OXA-23-positive result only for A. radioresistens strain CIP103788 (renamed strain 1 in this study), as it was the single representative of that species among the 50 isolates corresponding to the 14 Acinetobacter species tested. Additional PCRs were subsequently performed with four A. radioresistens clinical isolates (strains 2 to 5). These isolates also gave positive results; and sequencing identified an identical bla_OXA-23 gene in two cases (strains 3 and 5) and very closely related genes, namely, bla_OXA-103, bla_OXA-102, and bla_OXA-105, in strains 1, 2, and 4, respectively. Those variants had up to six amino acid substitutions compared to the sequence of β-lactamase OXA-23 (Fig. 2).

FIG. 2.

Comparison of the amino acid sequences of the OXA-23-like determinants. OXA-23 is from A. radioresistens isolates 3 and 5, OXA-102 is from isolate 1, OXA-103 is from isolate 2, and OXA-105 is from isolate 4. Dashes indicate identical amino acid residues, and the critical motifs for CHDLs are shaded in gray. The numbering is according to the nomenclature for CHDLs (8).

Analysis of I-CeuI-restricted fragments by PFGE followed by their hybridization showed that the A. radioresistens strains gave a single fragment that cohybridized with both 16S RNA- and bla_OXA-23-specific probes, demonstrating that the bla_OXA-23-like genes were located on the chromosome (Fig. 3). PCR mapping indicated that the gene coding for a putative ATPase and located downstream of the acquired bla_OXA-23 gene (and part of Tn2006 and Tn2007) in A. baumannii was also identified at the exact same downstream position in all five A. radioresistens strains. The corresponding proteins also displayed high degrees of amino acid identity (more than 98% amino acid identity, with no more than four nucleotide substitutions in the gene). A PCR screening of the five A. radioresistens strains for ISAba1- or ISAba4-like elements gave negative results, indicating that A. radioresistens was not a reservoir for these insertion sequence elements involved in the mobilization of bla_OXA-23.

FIG. 3.

PFGE profiles of I-Ceu-I digested whole-cell DNAs of A. radioresistens strains. Lanes 1, A. radioresistens reference strain 1; lanes 2, A. radioresistens clinical isolate 2; lanes 3, A. baumannii Ab13 isolate harboring a plasmid-located bla_OXA-23 gene. Southern hybridization was performed with an internal probe specific for the bla_OXA-23 gene (A) and with a probe specific for the 16S-23S rRNA gene (B). Horizontal arrows indicate the positions of hybridization with the bla_OXA-23 probe. By comparison of the two hybridization patterns, cohybridizations were obtained for strains 1 and 2 (chromosomal location of bla_OXA-23), whereas the bla_OXA-23-positive signal obtained for A. baumannii Ab13 has no corresponding band with the 16S-23S rRNA-specific probe.

Susceptibility patterns.

Antibiotic susceptibility testing showed that the five A. radioresistens were fully susceptible to all antibiotics tested, including penicillins and carbapenems. It is therefore likely that the bla_OXA-23-like genes were not expressed (or were expressed at a very low level) in their host (data not shown).

Catalytic properties of OXA-23-like oxacillinases.

In order to assess whether the A. radioresistens isolates expressed any carbapenemase activity, the β-lactamase activities were determined by using crude enzyme extracts of the cultures of each isolate. No hydrolysis was detected with any of these extracts obtained from the A. radioresistens strains with either benzylpenicillin or imipenem as the substrate, whereas the hydrolysis rates for A. baumannii Ab13 producing OXA-23 obtained with those two substrates were 5 and 0.2 U/mg of protein, respectively. This result further indicated a very weak expression of those naturally occurring bla_OXA-23-like genes in A. radioresistens. In order to evaluate the hydrolysis profile of the newly identified OXA-23-like β-lactamases, the bla_OXA-23, bla_OXA-102, bla_OXA-103, and bla_OXA-105 genes were cloned into the same plasmid vector and expressed in E. coli under the control of the same promoter. The MICs of the β-lactams were very similar, with a slight variability of the MICs of the carbapenems (Table 2). Three- to fourfold increases in the MICs of ertapenem and meropenem were detected for OXA-102, OXA-103, and OXA-105 producers compared to the MICs for the OXA-23 producers. Further detailed kinetic analysis would be necessary to analyze those catalytic properties.

TABLE 2.

MICs of carbapenems for the E. coli TOP10 recombinant strains expressing OXA-23, OXA-102, OXA-103, and OXA-105 and reference strain E. coli TOP10

Antibiotic	MIC (μg/ml)
	E. coli TOP10 (pOXA-23)	E. coli TOP10 (pOXA-102)	E. coli TOP10 (pOXA-103)	E. coli TOP10 (pOXA-105)	E. coli TOP10
Imipenem	1	1	1	1	0.125
Meropenem	0.047	0.094	0.094	0.094	0.047
Ertapenem	0.012	0.125	0.125	0.125	0.004

Plasmid analysis.

The repAci1 gene, which encodes a replicase, was recently identified as part of the replication control system of plasmids carrying bla_OXA-58 (another acquired CHDL-encoding gene from A. baumannii) (3). The bla_OXA-23-positive A. baumannii and A. radioresistens strains were evaluated to determine whether they carry plasmids with the repAci1 gene. These repAci1-positive plasmids may have a restricted host range since they did not replicate in members of the family Enterobacteriaceae (3, 13, 32). Our hypothesis was that this plasmid type could be also the vehicle for the dissemination of bla_OXA-23 and may thus be present in A. baumannii as the recipient and possibly in A. radioresistens as the donor. A preliminary screening of 15 bla_OXA-23-positive A. baumannii isolates collected worldwide showed that the repAci1 plasmids were identified in eight of these isolates, which in some cases also harbored the bla_OXA-23 gene (unpublished data). A similar screening performed with the A. radioresistens strains analyzed in this study identified repAci1 plasmids in a single strain. Thus, plasmids with an identical origin of replication were identified in A. radioresistens and A. baumannii.

DISCUSSION

We found in the present study that A. radioresistens is the progenitor of the bla_OXA-23-like genes currently emerging as the sources of carbapenem resistance in A. baumannii worldwide. This statement is based on the identification on the A. radioresistens chromosome of genes encoding both OXA-23-like and ATPase-like enzymes, with those genes being located near each other. The way in which the bla_OXA-23 gene has been mobilized from A. radioresistens and has disseminated to A. baumannii might correspond to the following model. A plasmid-mediated ISAba1 element originating from A. baumannii may enter A. radioresistens and then transpose and target the regions upstream and downstream of the chromosomal bla_OXA-23-like gene, thus forming a transposon-like structure and also enhancing the expression of the bla_OXA-23-like gene. This structure may then transpose and target a plasmid inside the A. radioresistens genome, and finally, this plasmid conjugates into A. baumannii, thus spreading the resistance determinant in the latter species.

A. radioresistens is a commensal species of the skin of healthy individuals (2, 36) and hospitalized patients (36, 43). The virulence role of that bacterium may be limited, since only one case of bacteremia has been reported so far and occurred in a human immunodeficiency virus-positive immunodepressed patient (41). However, the precise identification of Acinetobacter sp. isolates at the species level is not easy, and the use of molecular tools is required for unambiguous identification. It could therefore be possible that the true pathogenic trait of A. radioresistens might be underestimated, with many Acinetobacter sp.-related infections actually being due to A. radioresistens.

The A. radioresistens isolates studied were fully susceptible to β-lactams, indicating that the bla_OXA-23-like genes are likely poorly expressed from their original hosts. This indicates, as underlined by Iredell and Sintchenko (17), that the control of the reservoir of resistance genes might be more complex than expected. The source of the resistance gene may remain hidden, since it is located in a bacterial species that is not searched for in hospital settings and that is known to be highly resistant to desiccation and radiation (5, 18).

It is possible that genetic exchange between the progenitor (A. radioresistens) and its recipient of clinical relevance (A. baumannii), which leads to carbapenem resistance in the latter species, may have occurred in humans. Taking into account the fact that both A. baumannii and A. radioresistens are identified on the human skin, especially in hospitalized patients, it is possible that bla_OXA-23 gene exchange may occur at that location (2, 36).

A. radioresistens may be more prevalent than expected in the hospital environment, since it has been identified as the most common Acinetobacter species in hospital environmental samples (43). The identification of same plasmid types in A. radioresistens and A. baumannii further strengthens the possibility of gene exchange between those two species.

This study identified the source of an acquired and clinically relevant resistance gene. The unambiguous identification of the reservoir (origin) of an acquired resistance gene, such as bla_SHV from Klebsiella pneumoniae (39), the emerging bla_CTX-M-like genes from Kluyvera ascorbata and Kluyvera georgiana (16, 27), the plasmid-mediated cephalosporinase genes from several gram-negative species (26), and the plasmid-mediated qnrA gene from Shewanella algae (30), has very rarely been reported. Our findings further emphasize the possible role of the hospital environment as a reservoir of antibiotic resistance genes and a place where gene exchange may occur. The future control of multidrug resistance may necessitate identification of not only the multidrug-resistant isolates but also their reservoirs by molecular-based techniques.