Bacterial Identification, Clinical Significance, and Antimicrobial Susceptibilities of Acinetobacter ursingii and Acinetobacter schindleri, Two Frequently Misidentified Opportunistic Pathogens

Abstract

The species belonging to the Acinetobacter genus are currently reported as opportunistic pathogens in hospitalized patients with underlying predispositions. However, except for the Acinetobacter calcoaceticus-Acinetobacter baumannii complex, the identification of other species is frequently unreliable, especially for Acinetobacter ursingii and Acinetobacter schindleri, newly described in 2001. Thus, the clinical significance, phenotypic features, and antimicrobial susceptibilities of these two misidentified species remain unclear. Of 456 Acinetobacter sp. clinical strains isolated from 2002 to 2005 in Henri Mondor Hospital, 15 isolates (10 A. ursingii and 5 A. schindleri isolates) were studied. They were characterized using a phenotypic approach (API 20 NE and VITEK 2 systems), 16S rRNA gene sequencing, and susceptibility to antimicrobial agents with evaluation of impact in clinical relevance. The two corresponding type strains were also included for comparison. All isolates were identified to the species level using molecular tools, whereas the phenotypic methods remained unreliable due to the absence of these two species in the manufacturers' databases. However, the API 20 NE system appeared to be a reasonably reliable phenotypic alternative for the identification of A. ursingii when the numerical code 0000071 was found. Conversely, no discriminative phenotypic alternative existed for A. schindleri isolates. Concerning antimicrobial susceptibility, A. ursingii strains appeared to be more resistant to antibiotics than A. schindleri strains, which could imply therapeutic consequences. Finally, the prevalence of infections caused by A. ursingii and A. schindleri (representing 9.7% and 4.8% of non-A. calcoaceticus-A. baumannii complex strains, respectively) seems to be underestimated.

The species belonging to the Acinetobacter genus are widely distributed in nature since they are found frequently in soil, water, and dry environments (2). Due to their ability for long-term survival on inanimate surfaces, they are commonly isolated from the hospital environment (17) and are associated with skin colonization of inpatients and hospital personnel (25). Despite their low pathogenic potential, Acinetobacter species have increasingly been recognized as opportunistic pathogens mainly in immunocompromised patients and patients hospitalized in intensive care units (ICUs) (12). Their contribution to nosocomial infections has increased over the past 3 decades, and many outbreaks involving these microorganisms have been reported worldwide (9). They particularly represent an important cause of ventilator-associated pneumonia (10, 13) and catheter-related bacteremia (18, 20, 21, 24, 29). They are regularly recovered from urinary tract and wound infections, and some sporadic cases of peritoneal dialysis peritonitis, endocarditis, meningitis, osteomyelitis, arthritis, pancreatic and liver abscesses, and eye infections have also been reported (1, 2, 15).

According to the standard DNA-DNA hybridization method (26), the taxonomy of Acinetobacter has undergone extensive revision since 1986 (16), and there are actually 32 genomic species, including 17 with a validated name (http://www.bacterio.cict.fr/). Only 10 nomenspecies have been isolated in human specimens (A. baumannii, A. calcoaceticus, A. haemolyticus, A. johnsonii, A. junii, A. lwoffii, A. parvus, A. radioresistens, A. schindleri, and A. ursingii), and 7 newly described species were isolated from activated sludge plants (A. baylyi, A. bouvetii, A. gerneri, A. grimontii, A. tandoii, A. tjernbergiae, and A. towneri) (6). Moreover, the association of some unnamed species with human clinical samples has also been reported, especially genomic species 3, 13TU, 10, and 11 (2). The most frequently isolated species from humans is A. baumannii, which belongs to the A. calcoaceticus-A. baumannii complex, including three other species: A. calcoaceticus (a soil organism) and genomic species 3 and 13TU (both commonly isolated from hospitalized patients). Even if these four species are genetically highly related (7) and difficult to differentiate from each other phenotypically (3, 11), the A. calcoaceticus-A. baumannii complex is easy to identify in medical practice (4). On the contrary, phenotypic differentiation of other species remains frequently unreliable, particularly for nonglucidolytic species (3, 11). So, alternative genotypic methods, including sequencing of the 16S rRNA (rrs) gene (14) and several housekeeping genes such as gyrB and rpoB (19, 30), constitute a rapid helpful tool for the precise identification of uncommon Acinetobacter species in clinical microbiology.

Among these species, A. ursingii and A. schindleri, recently described in 2001 (23), could represent nonnegligible opportunistic pathogens because their routine identification is not possible by a phenotypic approach, due to their absence in the databases of all commercial biochemical kits. Thus, only identification by using a molecular approach could estimate their incidence in human specimens and allow an accurate evaluation of clinical relevance. Indeed, apart from the taxonomic description of these two new species (23), only one case of bacteremia caused by A. ursingii (21) and no cases of A. schindleri infection were reported.

In this study, we retrospectively investigated the clinical significance, phenotypic and genotypic identifications, and antimicrobial susceptibilities of A. ursingii and A. schindleri isolates collected from 2002 to 2005 in a French teaching hospital.

MATERIALS AND METHODS

Bacterial isolates.

From January 2002 to December 2005, 456 clinical strains of Acinetobacter were isolated in Henri Mondor Hospital (Créteil, France). Most A. calcoaceticus-A. baumannii complex strains (n = 353) were unambiguously identified by phenotypic methods (“very good” or “excellent” identification ratings with the API 20 NE system). On the other hand, all other strains (n = 103) were identified by using molecular identification (see below): A. lwoffii/genomic species 9 (n = 26), A. calcoaceticus-A. baumannii complex (24) (including A. baumannii [14], genomic species 3 [8], and genomic species 13TU [2]), A. haemolyticus (14), A. ursingii (10), A. junii (10), A. schindleri (5), A. johnsonnii (4), A. radioresistens (3), genomic species 15TU (4), genomic species 17 (1), genomic species 16BJ (1), and genomic species 10 (1). Indeed, A. ursingii and A. schindleri represented nonnegligible species among clinical isolates and 15 strains were then available for phenotypic identification, molecular characterization, and susceptibility testing.

Clinical data.

Clinical data regarding the referral department, the age and gender of the patient, clinical presentation, the site of isolation, predisposing conditions, and concomitant flora were obtained. To define infections, we used the CDC definitions of nosocomial infections (http://www.cdc.gov/ncidod/dhqp/pdf/nnis/NosInfDefinitions.pdf). An isolate was classified as (i) a definite pathogen if the patient had symptoms and signs of infection at the site of isolation and no other pathogen was isolated from that site; (ii) a probable pathogen if the patient had symptoms and signs of infection at the site of isolation but the culture yielded polymicrobial growth or if A. ursingii/A. schindleri was isolated in pure culture but the signs and symptoms of infection were not definitely related to the site of isolation; (iii) a possible pathogen if the signs and symptoms of infection were evident but not clearly related to the site of isolation; (iv) a nonpathogen if there was no evidence of infection at the time of isolation.

Phenotypic identification.

The 15 strains were cultured at 30°C, 37°C, 41°C, and 44°C aerobically on Trypticase soy agar (bioMérieux, Marcy l'Etoile, France) and on selective Drigalski agar (bioMérieux). Horse blood (5%) agar plates (bioMérieux) were used to detect hemolysis. Biochemical characterization was achieved by performing the oxidase reaction (Bio-Rad Laboratories, Marnes-la-Coquette, France) and by using the API 20 NE strip and VITEK 2 with an ID-GNB card (bioMérieux) in accordance with the manufacturers' instructions. The phenotypic characteristics of clinical isolates were compared with those of the two type strains provided by the Pasteur Institute Collection: A. ursingii LUH 3792 and A. schindleri LUH 5832.

Sequencing of the rrs gene and phylogenetic analysis.

Bacterial genomic DNA was extracted using a QIAmp DNA mini kit (QIAGEN, Courtaboeuf, France). PCR was performed to amplify the rrs gene using two sets of primers (synthesized by Proligo France SAS): PB, 5′-TAACACATGCAAGTCGAACG-3′; and BAK2, 5′-GGACTAC(A/C/T)AGGGTATCTAT-3′ (corresponding to positions 49 to 806 in Escherichia coli K-12 numbering); and UNI14, 5′-GTGCCAGCAGCCGCGGTAAT-3′; and P13B, 5′-CGGGATCCCAGGCCCGGGAAC-3′ (corresponding to positions 515 to 1399 in E. coli K-12 numbering). Typical reaction mixtures (50 μl) contained 1× reaction buffer containing 1.5 mM of MgCl₂, 50 μM of each deoxynucleoside triphosphate, 1 μM of each primer, 1.25 U of Taq polymerase (Q-Biogene, Illkirch, France) and ca. 150 ng of DNA template. PCR amplifications were performed using an iCycler thermal cycler (Bio-Rad Laboratories) as follows: (i) initial denaturation step of 5 min at 94°C, (ii) 35 cycles of PCR, with 1 cycle consisting of 30 s at 94°C, 30 s at 47°C, and 30 s at 72°C, and (iii) a final extension step of 7 min at 72°C. After purification with Montage PCR centrifugal filter devices (Millipore, Molsheim, France), PCR products were then directly sequenced using an ABI PRISM BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems, Courtaboeuf, France) with the same sets of primers on an ABI PRISM 3100 automatic sequencer (Applied Biosystems).

The nucleotide sequences were analyzed with SeqScape v2.0 software (Applied Biosystems) and assembled in order to obtain rrs contigs of ca. 1,200 bp. These sequences were then compared to deposited sequences available from GenBank using the BLAST program. Phylogenetic analysis was performed by using the neighbor-joining algorithm with ClustalX software (version 1.83), and the resulting tree was displayed with TreeView software (version 1.6.6). All published Acinetobacter genomic sequences, obtained from GenBank, were used to confirm the different relationships between clinical isolates and type strains as previously described (27).

Antimicrobial susceptibility.

Antimicrobial susceptibility of the clinical isolates and type strains was determined by the disk diffusion method on Mueller-Hinton agar according to the recommendations of the Comité de l'Antibiogramme de la Société Française de Microbiologie (http//:www.sfm.asso.fr). The disks were supplied by Bio-Rad Laboratories, and the following antibiotics were tested: amoxicillin, ticarcillin, amoxicillin-clavulanate, ticarcillin-clavulanate, piperacillin, piperacillin-tazobactam, cephalothin, cefuroxime, cefoxitin, cefotaxime, ceftazidime, aztreonam, moxalactam, cefepime, cefixime, imipenem, kanamycin, tobramycin, gentamicin, netilmicin, amikacin, nalidixic acid, ofloxacin, ciprofloxacin, chloramphenicol, tetracycline, fosfomycin, rifampin, sulfonamide, trimethoprim, furans, and colistin. MICs were determined for several antibiotics (see Table 4) by using the Etest method (AB Biodisk, Solna, Sweden) according to the manufacturer's recommendations.

TABLE 4.

Antimicrobial susceptibility profiles of the 10 A. ursingii isolates and the 5 A. schindleri isolates of this study compared to those of the respective type strains

Antibiotica	A. ursingiib					A. schindlerib
	Profile of LUH 3792T (MIC)	No. of clinical isolates with indicated profile			MIC range for clinical isolates	Profile of LUH 5832T (MIC)	No. of clinical isolates with indicated profile			MIC range for clinical isolates
		S	I	R			S	I	R
Amoxicillin	S (4)	6	4	0	2-12	S (0.25)	5	0	0	0.19-1.5
Amoxicillin + clavulanate	S (3)	10	0	0	1.5-4	S (0.19)	5	0	0	0.125-0.75
Ticarcillin	S	7	3	0	NDc	S	5	0	0	ND
Ticarcillin + clavulanate	S (0.016)	10	0	0	≤0.016-0.032	S (≤0.016)	5	0	0	≤0.016
Piperacillin	S (12)	2	8	0	8-32	S (8)	5	0	0	0.25-16
Piperacillin + tazobactam	S (4)	10	0	0	1-12	S (≤0.016)	5	0	0	≤0.016
Cephalothin	R (>256)	0	0	10	>256	I (24)	1	1	3	1-192
Cefuroxime	I (32)	0	4	6	16-64	S (3)	5	0	0	0.094-6
Cefoxitin	R	0	0	10	ND	I	1	2	2	ND
Cefotaxime	R (>32)	0	1	9	16->32	S (3)	2	3	0	0.094-8
Ceftazidime	I (32)	0	7	3	6-64	S (1.5)	5	0	0	0.38-4
Aztreonam	R (>256)	0	2	8	16->256	S (2)	2	3	0	0.5-16
Moxalactam	R	0	0	10	ND	S	2	0	3	ND
Cefepime	R (>32)	2	6	2	3->32	S (1.5)	5	0	0	0.094-1
Cefixime	R	0	0	10	ND	S	4	0	1	ND
Imipenem	S (0.064)	10	0	0	0.094-0.19	S (0.094)	5	0	0	0.064-0.094
Kanamycin	S	10	0	0	ND	S	5	0	0	ND
Tobramycin	S (0.38)	10	0	0	0.25-0.5	S (0.38)	5	0	0	0.38-0.5
Gentamicin	S (0.75)	10	0	0	0.125-0.38	S (0.125)	5	0	0	0.125-0.25
Netilmicin	I (3)	10	0	0	0.19-0.5	S (0.19)	5	0	0	0.125-0.25
Amikacin	S (1)	10	0	0	1-2	S (0.75)	5	0	0	0.75-1
Nalidixic acid	S (3)	9	1	0	1.5-12	S (2)	5	0	0	1-8
Ofloxacin	S (0.25)	9	1	0	0.25-1	S (0.19)	5	0	0	0.047-0.5
Ciprofloxacin	S (0.125)	10	0	0	0.094-0.38	S (0.047)	5	0	0	0.008-0.19
Chloramphenicol	R (128)	2	1	7	6->256	S (1.5)	5	0	0	0.75-6
Tetracycline	S (3)	8	2	0	1.5-8	S (0.75)	5	0	0	0.25-1.5
Fosfomycin	R (384)	2	0	8	32-128	R (48)	1	0	4	24-256
Rifampin	S (1.5)	6	4	0	3-6	S (4)	3	2	0	1.5-8
Sulfonamide	S	10	0	0	ND	R	3	0	2	ND
Trimethoprim	R	0	3	7	ND	R	0	1	4	ND
Nitrofurans	R	0	0	10	ND	S	2	3	0	ND
Colistin	S (0.25)	10	0	0	0.125-0.25	S (0.38)	5	0	0	0.38-0.5

^aMICs were determined using Etest strips for all antibiotics except for ticarcillin, cefoxitin, moxalactam, kanamycin, sulfonamide, trimethoprim, and nitrofurans. MICs are given in mg/liter.

^bS, susceptible; I, intermediate; R, resistant.

^cND, not determined.

RESULTS

Clinical data.

From 1 January 2003 through 30 December 2005, 10 A. ursingii and 5 A. schindleri clinical isolates were recovered in our laboratory. Clinical data and predisposing conditions are summarized in Table 1. All 15 clinical isolates were isolated after more than 72 h of stay in the hospital, suggesting a hospital-acquired colonization and/or infection. Even if there was no apparent epidemiological link between the isolates of each species (different dates and places of isolation), molecular typing experiments were performed for some strains and confirmed no epidemiological relationship (data not shown). Using clinical criteria, A. ursingii was classified as a definite or probable pathogen in six cases (60%), while A. schindleri was classified as a definite or probable pathogen in four cases (80%). Finally, A. ursingii and A. schindleri were isolated from polymicrobial cultures in 30% and 60% of the cases, respectively.

TABLE 1.

Patient characteristics and clinical syndromes

Organism and patient	Age (yr)	Sexa	Isolation		Clinical presentationb	Specimenc	Concomitant organism(s)d	Underlying comorbidity(ies)e	Type of pathogen
			Date (mo/day/yr)	Ward
A. ursingii
1	84	M	03/13/02	Geriatric	UTI	Urine (2)	None	LL, diabetes mellitus, CRF, PH, urinary catheter	Definite
2	45	F	04/02/09	Neurosurgery	Postneurosurgical meningitis	CSF (wound) (1)	CNS, Acinetobacter lwoffii, Aerococcus viridans	Recent case of an acoustic neurinoma	Possible
3	31	F	05/06/08	Visceral surgery	Intestinal occlusion	Central catheter (1)	CNS	Obesity, gastric bypass surgery	Probable
4	99	M	04/07/21	Emergency	Cholangitis	Blood (1)	None	IHD	Possible
5	67	F	05/05/04	Cardiology	Septic shock	Blood (3)	None	CPH, mitral valvular prosthesis	Definite
6	34	F	05/07/05	Immunology	Cervical adenopathy	Ganglion biopsy (1)	Mycobacterium tuberculosis, CNS	Human immunodeficiency virus infection, tuberculosis	Nonpathogen
7	84	M	03/07/28	Dermatology	UTI	Urine (1)	None	CRF, urinary catheter	Definite
8	69	M	04/10/01	Neurosurgery	Asymptomatic	CSF (1)	None	LL, IHD, PH	Nonpathogen
9	38	F	05/09/21	Oncology	Bacteremia	Blood (1)	None	Gastric adenocarcinoma, recent chemotherapy, CC	Definite
10	52	F	05/11/14	Oncology	UTI	Urine (1)	None	Breast adenocarcinoma, recent chemotherapy, urinary catheter	Definite
A. schindleri
1	51	F	03/08/23	Hematology	Bacteremia	Blood (1)	Acinetobacter lwoffii, CNS	ML, recent chemotherapy, BMA, CC	Probable
2	36	M	03/09/14	Hematology	Bacteremia	Blood (1)	Citrobacter freundii, Pseudomonas aeruginosa, Chryseobacterium meningosepticum	BL, recent chemotherapy, BMA, CC	Probable
3	64	F	04/02/12	Hematology	Bacteremia	Blood (1)	Pseudomonas fluorescens	Myelodysplasia, homograft, BMA, CC	Probable
4	58	F	04/12/06	Hematology	Bacteremia	Blood (2)	None	Myelodysplasia, BMA, Crohn's disease, CC	Definite
5	77	F	04/12/13	Cardiovascular surgery	Asymptomatic	Central catheter (1)	None	Recent substitution of the aortic valve, total hip prosthesis, recent UTI	Nonpathogen

^aM, male; F, female.

^bUTI, urinary tract infection.

^cThe number of positive specimens whose culture yielded A. ursingii or A. schindleri is shown in parentheses. CSF, cerebrospinal fluid.

^dCNS, coagulase-negative staphylococci.

^eLL, lymphatic leukemia; CRF, chronic renal failure; IHD, ischemic heart disease; PH, prostatic hypertrophy; CPH, chronic pulmonary heart; ML, monocytic leukemia; BL, Burkitt's lymphoma; BMA, bone marrow aplasia; CC, central catheter.

In the six symptomatic patients, A. ursingii isolates were recovered in blood (n = 3) and in urine (n = 3). A. ursingii was the only pathogen isolated from blood cultures in all three patients with severe underlying comorbidities and central catheters. It was also recovered in pure culture from urine (≥10⁶ CFU/ml) in two male patients and one female patient with indwelling urinary catheters. In the four symptomatic patients, all A. schindleri isolates were recovered from blood cultures of patients hospitalized in the hematology unit. All of these patients were neutropenic at the moment of infection and had central venous catheters that were not removed.

Phenotypic analysis.

Growth characteristics and biochemical features are summarized in Table 2. All 15 isolates had the properties of the genus Acinetobacter: they were nonmotile, strictly aerobic, and gram-negative coccobacilli with positive catalase and negative oxidase reactions. All strains grew on Trypticase soy agar and on Drigalski agar (colorless colonies) and were nonhemolytic on 5% horse blood agar. A. ursingii isolates were able to grow at 30°C and 37°C but not at 41°C and 44°C, whereas A. schindleri isolates grew at 30°C, 37°C, and 41°C but not at 44°C.

TABLE 2.

Characterization of A. ursingii and A. schindleri compared to other Acinetobacter spp. by the API 20 NE system

Characteristic	Result fora:
	A. ursingii	A. schindleri	A. junii/ A. johnsonii	A. lwoffii	A. radioresistens	A. baumannii/ A. calcoaceticus	A. haemolyticus
Growth at:
30°C	+	+	+	+	+	+	+
37°C	+	+	±b	+	+	+	+
41°C	−	+	−	−	−	±c	−
44°C	−	−	−	−	−	±c	−
β-Hemolysis	−	−	−	−	+	−	+
Nitrate reduction	−	−	−	−	−	−	−
Indole production	−	−	−	−	−	−	−
Fermentation of glucose	−	−	−	−	−	−	−
Arginine dihydrolase	−	−	−	−	−	−	−
Urease activity	−	−	−	−	−	−	−
Esculin hydrolysis	−	−	−	−	−	−	−
Gelatin hydrolysis	−	−	−	−	−	−	+
β-Galactosidase	−	−	−	−	−	−	−
Assimilation of:							−
Glucose	−	−	V (24)	V (11)	V (19)	V (67)	−
Arabinose	−	−	−	−	−	V (70)	−
Mannose	−	−	−	−	−	−	−
Mannitol	−	−	−	−	−	−	−
N-Acetylglucosamine	−	−	−	−	−	−	−
Maltose	−	−	−	−	−	−	−
Gluconate	−	−	−	−	−	V (20)	−
Caprate	+	−	+	V (70)	+	+	+
Adipate	+	−	−	V (20)	+	+	−
Malate	+	+	+	V (46)	−	+	+
Citrate	+	V (60)	−	−	−	+	+
Phenylacetate	−	−	−	V (36)	+	+	−

^aAll species were catalase-positive, oxidase-negative, and strictly aerobic gram-negative rods. Except for differential temperature growth and β-hemolysis detection, all tests were performed using the API 20 NE system. The profiles for 12 isolates of A. ursingii (our 10 isolates plus the type strain LUH 3792 and 1 strain described in the literature [28]) and for 6 isolates of A. schindleri (our 5 isolates plus the type strain LUH 5832) were compared with those in the manufacturer's database for Acinetobacter spp. +, ≥90% of strains are positive; −, ≤10% of strains are negative; V, variable. The values in parentheses are the percentages of strains with a positive reaction.

^bGrowth occurs at 37°C for A. junii but not for A. johnsonii.

^cGrowth occurs at 41°C and 44°C for A. baumannii but not for A. calcoaceticus.

According to the API 20 NE system database, all 15 isolates were identified as having doubtful or unacceptable profiles for species-level identification (Table 3). For the 10 A. ursingii isolates, only two different numerical codes were obtained: 0000070 (1/10) and 0000071 (9/10). These profiles were classified as A. lwoffii or A. junii/A. johnsonii (0000070) or as A. junii/A. johnsonii or A. baumannii/A. calcoaceticus (0000071), with “good or acceptable identification to the genus,” respectively (Table 3). These patterns corresponded to the unique assimilation of caprate, adipate, and malate or caprate, adipate, malate, and citrate, respectively (Table 2). Note that the A. ursingii type strain LUH 3792 also gave the numerical profile 0000071. For the five isolates of A. schindleri, two API 20 NE numerical profiles were found, 0000040 (three of five) and 0000050 (two of five), corresponding to the unique assimilation of malate and malate plus caprate, respectively, whereas the A. schindleri type strain LUH 5832 gave the numerical code 0000051, corresponding to a further assimilation of citrate (Table 2). These profiles were classified as A. lwoffii (0000040), with “acceptable identification,” as A. junii/A. johnsonii or A. lwoffii (0000050), and as A. junii/A. johnsonii or A. haemolyticus (0000051), with “good identification to the genus” (Table 3).

TABLE 3.

Misidentifications obtained with the API 20 NE and VITEK 2 systems for the clinical isolates of A. ursingii (n = 10) and A. schindleri (n = 5) and for the corresponding type strains

Strain and origin	Result(s) obtained with:
	API 20 NE system (quality of identification [%])	VITEK 2 ID-GNB card (quality of identification [%])
A. ursingii
Type strain	A. junii/A. johnsonii (63.5), A. baumannii/A. calcoaceticus (26.2)	B. bronchiseptica (99.0)
Patient 1	A. lwoffii (78.4), A. baumannii/A. calcoaceticus (18.4)	B. bronchiseptica (99.0)
Patient 2	A. junii/A. johnsonii (63.5), A. baumannii/A. calcoaceticus (26.2)	B. bronchiseptica (99.0)
Patient 3	A. junii/A. johnsonii (63.5), A. baumannii/A. calcoaceticus (26.2)	B. bronchiseptica (99.0)
Patient 4	A. junii/A. johnsonii (63.5), A. baumannii/A. calcoaceticus (26.2)	B. bronchiseptica (99.0)
Patient 5	A. junii/A. johnsonii (63.5), A. baumannii/A. calcoaceticus (26.2)	B. bronchiseptica (91.4)
Patient 6	A. junii/A. johnsonii (63.5), A. baumannii/A. calcoaceticus (26.2)	B. bronchiseptica (91.1)
Patient 7	A. junii/A. johnsonii (63.5), A. baumannii/A. calcoaceticus (26.2)	B. bronchiseptica (95.0)
Patient 8	A. junii/A. johnsonii (63.5), A. baumannii/A. calcoaceticus (26.2)	B. bronchiseptica (95.0)
Patient 9	A. junii/A. johnsonii (63.5), A. baumannii/A. calcoaceticus (26.2)	A. junii (97.9)
Patient 10	A. junii/A. johnsonii (63.5), A. baumannii/A. calcoaceticus (26.2)	Inconclusive identification
A. schindleri
Type strain	A. junii/A. johnsonii (92.0), A. haemolyticus (7.0)	A. lwoffii (98.4)
Patient 1	A. junii/A. johnsonii (56.5), A. lwoffii (40.9)	A. lwoffii (97.7)
Patient 2	A. junii/A. johnsonii (56.5), A. lwoffii (40.9)	A. lwoffii (99.0)
Patient 3	A. lwoffii (82.7)	A. lwoffii (98.5)
Patient 4	A. junii/A. johnsonii (56.5), A. lwoffii (40.9)	B. bronchiseptica (99.0)
Patient 5	A. junii/A. johnsonii (56.5), A. lwoffii (40.9)	A. lwoffii (50), Moraxella spp. (50)

Using the ID-GNB card of the VITEK 2 system, only a few of the 47 characteristics tested were positive for A. ursingii and A. schindleri isolates. For all A. ursingii isolates, 11 tests were positive but only 6 were commonly positive: l-lactate alkalinization, 11/11; glutamyl-arylamidase pNA, 9/11; succinate alkalinization, 11/11; tyrosine-arylamidase, 9/11; citrate, 8/11; and urease, 9/11. The nine clinical isolates and the type strain LUH 3792 were misidentified by the VITEK 2 system as Bordetella bronchiseptica (9/11), with “very good identification” or “excellent identification,” or Acinetobacter junii (1/11), with “excellent identification,” and one clinical isolate was not identified (Table 3). For all A. schindleri isolates, six tests were positive but only four were commonly positive: glutamyl-arylamidase pNA, four of six; tyrosine-arylamidase, five of six; urease, four of six; and lactate assimilation, five of six. Four clinical isolates and the type strain LUH 5832 were classified as A. lwoffii, whereas one strain (patient 4) was misidentified as B. bronchiseptica, with “excellent identification” (Table 3).

Genotypic identification and phylogenetic analysis.

Sequences of a 1,200-bp fragment of the rrs gene for all 15 strains were queried to the GenBank database for the best matches after BLAST analysis. This permitted clear identification of all clinical isolates to the species level. Identities ranged from 99.5 to 100% and from 99.3 to 99.6% compared to each type strain for clinical strains of A. ursingii and A. schindleri, respectively. The intraspecies sequence divergences were only 0.4% and 0.2% among 10 A. ursingii and 5 A. schindleri isolates, respectively. Phylogenetic analysis based on the comparison of the sequences of the rrs gene confirmed that both A. ursingii and A. schindleri groups clustered with other members of the genus Acinetobacter and constituted independent monophyletic clades with a low intraspecies variability (data not shown).

Antibiotic susceptibility.

The results of in vitro susceptibility testing of the 15 clinical isolates and the two type strains to 32 antimicrobial agents by the disk diffusion and Etest methods are shown in Table 4. For the A. ursingii species, all isolates, including the type strain, were fully resistant to cephalothin, cefoxitin, moxalactam, cefixime, and furans and susceptible to amoxicillin-clavulanate, ticarcillin-clavulanate, piperacillin-tazobactam, imipenem, aminoglycosides, ciprofloxacin, tetracycline, sulfamide, and colistin. Amoxicillin, ticarcillin, piperacillin, cefepime, rifampin, and fosfomycin exhibited a variable activity. Note that none was categorized as susceptible to cefuroxime, cefotaxime, ceftazidime, aztreonam, and trimethoprim. Most of the isolates were resistant to chloramphenicol (70%) and susceptible to nalidixic acid and ofloxacin (90%).

The strains of A. schindleri were susceptible to penicillin, cefuroxime, ceftazidime, cefepime, and imipenem, whereas other β-lactams exhibited a variable activity. They were uniformly susceptible to aminoglycosides, quinolones, chloramphenicol, tetracycline, and colistin. Almost all strains were resistant to fosfomycin (80%) and trimethoprim (70%), while they were variably susceptible to rifampin, sulfonamide, and furans.

DISCUSSION

In recent years, there has been an increasing incidence of opportunistic infections due to Acinetobacter spp. most common among patients in ICUs (15). Even if A. baumannii is the most frequently encountered, other species have also been reported, but their identifications remain difficult (3). Indeed, in many studies, species identification remains unreliable because of the use of names that are no longer valid or inadequate phenotypic methods. The true clinical significance of Acinetobacter spp. could be better understood if accurate identification to the species level was performed (15). In particular, two newly described human species, A. ursingii and A. schindleri, need further characterization to provide a better understanding of their pathogenic potential.

Between 2002 and 2005, 456 strains of Acinetobacter spp. were identified in Henri Mondor Hospital and 77.4% (353/456) were easily identified as species belonging to the A. calcoaceticus-A. baumannii complex. A. ursingii and A. schindleri represented 9.9% and 4.9% of isolates among the non-A. calcoaceticus-A. baumannii complex strains, respectively. In the first report of these two species in 2000, which were provisionally termed phenon 1 (A. ursingii) and phenon 2 (A. schindleri), of 700 prospective human isolates from the Czech Republic, 549 were identified as belonging to the A. calcoaceticus-A. baumannii complex, and among the remaining 147 isolates, phenon 1 and phenon 2 represented 17 (11.6%) and 15 (10.2%) isolates, respectively (22). The taxonomic status of the A. ursingii and A. schindleri species, investigated in 2001, was based on 29 and 22 isolates, respectively (23). In those previous works, the majority of the A. schindleri strains were isolated from nonsterile body sites of outpatients whereas most of the A. ursingii strains were isolated from blood cultures in seriously ill hospitalized patients. In our inpatient study, we found that 40% of A. ursingii isolates were recovered from blood cultures compared to 80% for A. schindleri isolates. In the case of catheter-related bacteremia, we can assume that Acinetobacter isolates were able to colonize the patients' skin and that the intravascular device served as a portal of entry for bloodstream infection (2). Major risk factors for nosocomial infections, such as cancer with recent chemotherapy, invasive devices and procedures, and long stays in ICUs, were found for all symptomatic patients. Note that all patients were either immunocompromised or catheterized (intravenous line or urinary catheter).

In our study, all of these isolates were initially phenotypically misidentified and retrospectively needed a molecular identification because the species names were absent in the databases of all commercial test systems. This limit of phenotypic methods is due partially to the improvement of the taxonomy with continuous descriptions of new species in the genus Acinetobacter, i.e., seven new species isolated from activated sludge plants described in 2003 (6). So, this implies that the manufacturers should rapidly update their databases according to the recent taxonomic changes. At present, no rapid and accurate commercial phenotypic identification exists for A. ursingii and A. schindleri. An array of 19 biochemical tests has been suggested by Bouvet and Grimont (5), but it is not easily applicable in medical practice and does not work to differentiate the species into the A. calcoaceticus-A. baumannii complex (11). However, we can suggest that the API 20 NE test represents a reasonably reliable phenotypic alternative for A. ursingii identification when the numerical code 0000071 is found. For other numerical codes (0000040, 0000050, 0000051, and 0000070), a definite confirmation by rrs gene sequencing should be performed because no reliable identification is possible. With the VITEK 2 system, A. ursingii and A. schindleri isolates, including each type strain, were misidentified as B. bronchiseptica (10/17) or Acinetobacter spp. (6/17) with acceptable criteria or not identified (1/17) (Table 3). This identification system yielded very major errors in the identification of A. ursingii and A. schindleri strains. Finally, note that this system does not use the result of oxidase reaction, which constitutes a simple test to easily differentiate the two genera Acinetobacter and Bordetella.

The rrs sequences of the clinical isolates A. ursingii and A. schindleri displayed more than 99.5% and 99.3% similarities, respectively, to the sequences previously deposited in databases and less than 97% similarity with those of other Acinetobacter spp. Besides, A. ursingii and A. schindleri isolates, respectively, shared at least 99.6% and 99.8% similarity to each other, which is lower than the intraspecies variability previously reported (23). Our results indicate that the rrs gene sequencing method is a specific and sensitive tool for identifying and differentiating A. ursingi and A. schindleri isolates.

Contrary to A. baumannii isolates, which may present high rates of antibiotic resistance, resulting in therapeutic problems for the treatment of patients with nosocomial infections (2, 28), clinical isolates of A. ursingii and A. schindleri seem to be more susceptible to antimicrobial agents. However, the prevalence and biochemical or genetic mechanisms of resistance have not been identified for these species. The main mechanism of resistance to β-lactam antibiotics in Acinetobacter spp. is due to the production of β-lactamases encoded either by the chromosome or by plasmids (28). The different studies demonstrate that overexpression of chromosomal cephalosporinase plays an important role in resistance to β-lactam antibiotics, especially in A. baumannii. Our results suggested the presence of a cephalosporinase with different susceptibility patterns according to the species (Table 4). Indeed, A. schindleri isolates appeared generally to be more susceptible to antibiotics than the A. ursingii isolates, especially for β-lactams. No strains of A. ursingii were susceptible to first-, second-, or third-generation cephalosporins, whereas A. schindleri isolates were susceptible to third-generation cephalosporins, with inconsistent resistance to cephalothin or cefoxitin. This might be due to differences in the nature and/or level of production of chromosomal β-lactamase, but further investigations should be performed with these two species. The resistance to aminoglycosides by the production of modifying enzymes is relatively common in clinical isolates of Acinetobacter baumannii (2); however, both A. ursingii and A. schindleri isolates are fully susceptible to the aminoglycosides tested in the present study. Quinolones have good activity against Acinetobacter strains compared to expanded-spectrum cephalosporins and aminoglycosides. In our study, the majority of A. ursingii isolates (9/10) and all A. schindleri isolates (5/5) were susceptible to nalidixic acid and fluoroquinolones. Finally, resistance to other antibiotics was variable and no specific studies on the prevalence and mechanisms of resistance in Acinetobacter spp. have been published.

In conclusion, the reliability of phenotypic tests for the identification of Acinetobacter species is inadequate, especially for non-A. calcoaceticus-A. baumannii complex species. Indeed, phenotypic analysis should be combined with molecular identification methods, such as rrs gene sequencing and amplified ribosomal DNA restriction analysis (8, 22). Newly described species, such as A. ursingii and A. schindleri, are frequently misidentified due to inadequacies of conventional biochemical testing. Even if their prevalence as well as their pathogenic potential and epidemiology remained unclear, this work highlights the importance of these two species among the opportunistic pathogens. Moreover, with regard to our results, we can recommend that the identification of A. ursingii be performed using the API 20 NE test system (with the numerical code 0000071) but not the VITEK 2 system. In doubtful cases and for the identification of A. schindleri, accurate identification needs a definite confirmation by molecular methods. Furthermore, A. ursingii strains appear to be more resistant to antibiotics than A. schindleri isolates, and that may imply therapeutic consequences. Finally, additional studies including more isolates will be needed to increase our knowledge of these two species.