Identification of Acinetobacter baumannii Strains with Monoclonal Antibodies against the O Antigens of Their Lipopolysaccharides

Abstract

Despite the emergence of Acinetobacter baumannii strains as nosocomial pathogens, simple methods for their phenotypic identification are still unavailable. Murine monoclonal antibodies specific for the O-polysaccharide moiety of the lipopolysaccharide (LPS) of two A. baumannii strains were obtained after immunization with heat-killed bacteria. The monoclonal antibodies were characterized by enzyme immunoassay and by Western and dot blot analyses and were investigated for their potential use for the identification of A. baumannii strains. The antibodies reacted with 46 of the 80 A. baumannii clinical isolates that were investigated, and reactivity was observed with 11 of 14 strains which were isolated during outbreaks in different northwestern European cities; no reactivity was observed with Acinetobacter strains of other genomic species, including the closely related genomic species 1 (Acinetobacter calcoaceticus), 3, and 13 sensu Tjernberg and Ursing, or with other gram-negative bacterial strains. The results show that O-antigen-specific monoclonal antibodies such as the ones described are convenient reagents which can be used to identify Acinetobacter strains in clinical and research laboratories.

The genus Acinetobacter belongs to the family Moraxellaceae (36), and its members are ubiquitous in the natural and clinical environments (1, 20, 38, 39, 44, 47). However, Acinetobacter strains have also been recognized as opportunistic nosocomial pathogens in recent years (2, 3). Urethritis, pneumonia, meningitis, and septicemia are the major diseases caused by these bacteria (3, 43), which, in addition, are usually highly resistant to a large spectrum of antibiotics (43). Acinetobacter baumannii (genomic species 2) is the most frequently occurring Acinetobacter sp. among clinical Acinetobacter isolates, and most hospital outbreaks are attributed to this species (3). Although other Acinetobacter species such as genomic species 3, A. johnsonii (genomic species 7), A. lwoffii (genomic species 8), and A. radioresistens (genomic species 12) may also be found among clinical Acinetobacter isolates, these species are usually considered to represent contamination or colonization rather than infection when they are isolated from clinical specimens (38, 40), particularly since these species are also present on the skin and mucous membranes of humans as part of their normal bacterial flora (39). Unfortunately, there is still lack of simple methods for the rapid identification of Acinetobacter strains in clinical laboratories (3, 6, 43). This is partly due to the confused taxonomic status associated with the genus (6, 11, 48) and also to the diversity of the strains, which is reflected in the different pheno- and genotypic groups that have been described (3, 6). To date, DNA-DNA hybridization studies have resulted in the delineation of 20 DNA homology groups, of which only 7 have received a formal species name (4, 5, 12, 42). Many strains described so far have remained unclassified (4, 5, 42).

Gram-negative bacteria express lipopolysaccharides (LPSs) at their outer surface (24, 25, 32). These LPSs consist of a polysaccharide covalently linked to a lipid component, termed lipid A, which anchors the LPS in the outer membrane (18, 31–33, 35). In enterobacteria, the polysaccharide is divided into the core oligosaccharide (linked to lipid A) and the O polysaccharide or O antigen (32, 33, 35). This type of LPS is referred to as the smooth- or S-form phenotype (33–35); the O antigens are characteristic for a given LPS and the parental bacterial strain, a fact on which serotyping schemes for various enteric and also nonenteric gram-negative bacteria are based (21, 27, 30, 32, 34, 35). Since all recently investigated LPSs from Acinetobacter strains have been shown to be of the smooth phenotype (13–17, 45, 46), a serotyping scheme for identification of members of this genus may also be possible. Recently, we reported on the specificity of hyperimmune rabbit sera against Acinetobacter LPS to examine the feasibility of such an identification scheme for Acinetobacter strains (29). Although they were shown to be useful (29), such antisera have certain disadvantages which make them unsuitable for routine applications, such as the presence of core-reactive antibodies as well as protein and possible capsular antibodies, which may lead to false-positive reactions when unabsorbed sera are used for O serotyping (23, 29, 37). Thus, to overcome this problem, we decided to generate monoclonal antibodies (MAbs) against the O antigens of various clinical and environmental Acinetobacter isolates.

In this report, we describe the generation and characterization of two MAbs specific for the O antigen of A. baumannii LPS and show that they can be used for the identification of A. baumannii strains, particularly for the tracing in hospital wards of strains originating from two epidemiologically important A. baumannii clonal groups.

MATERIALS AND METHODS

Bacteria.

The A. baumannii strains (n = 82) investigated in this study are listed in Table 1. They consisted mostly of clinical isolates from different parts of Scandinavia, The Netherlands, and the United Kingdom. Acinetobacter strains belonging to other genomic species were also examined (genomic species 1 [A. calcoaceticus; n = 8], genomic species 3 [n = 13], genomic species 4 [A. haemolyticus; n = 7], genomic species 5 [A. junii; n = 5], genomic species 6 [n = 1], genomic species 7 [A. johnsonii; n = 8], genomic species 8/9, which is considered a single entity [42] [A. lwoffii; n = 13], genomic species 10 [n = 3], genomic species 11 [n = 6], genomic species 12 [A. radioresistens; n = 7], genomic species 13 sensu Tjernberg and Ursing [42] [n = 11], and genomic species 14 [n = 4]). All Acinetobacter strains had previously been identified to the species level by DNA-DNA hybridization and other methods and were from the culture collection of L. Dijkshoorn (Leiden University Medical Center, Leiden, The Netherlands). The strains were originally obtained from A. Horrevorts (Canisius Wilhelmina Ziekenhuis, Nijmegen, The Netherlands), P. Gerner-Smidt (Statens Seruminstitut, Copenhagen, Denmark), T. L. Pitt (Central Public Health Laboratory, London, United Kingdom), I. Tjernberg and J. Ursing (Malmö University Hospital, Malmö, Sweden), and P. Janssen (University of Ghent, Ghent, Belgium). The non-Acinetobacter strains investigated in this study were obtained from R. Podschun (National Reference Center of Klebsiella species, Kiel, Germany) or were from our own culture collection (Salmonella spp. [n = 10], Escherichia coli [n = 4], Shigella sonnei [n = 8], Enterobacter spp. [n = 10], Pseudomonas sp. [n = 6], Stenotrophomonas maltophilia [n = 6], Serratia spp. [n = 10], Burkholderia cepacia [n = 2], Hafnia spp. [n = 10], and Proteus spp. [n = 20]).

TABLE 1.

Reactivities of MAbs S48-3-13 and S48-3-17 in dot and Western blots with LPSs from proteinase K-treated bacterial whole-cell lysates from A. baumannii clinical isolates investigated in this study and O-banding patterns obtained following acid hydrolysis of membrane-bound LPSs and immunostaining with MAb S1 of proteinase K-treated bacterial lysates from A. baumannii strains which did not react with MAb S48-3-13 or MAb S48-3-17

Straina	Sourceb	City (country)c	Positiond	Reactivity with MAbe		Pattern obtained after acid hydrolysisf
				S48-3-13	S48-3-17
ATCC 17904	Urine	Not known	E4	−	−	−
ATCC 19606T	Urine	Not known	E3	−	−	−
NCTC 7844	Not known	Not known	G8	−	−	C
9g	Sputum	Malmö (SE)	G7	−	−	−
10g	Wound	Rotterdam (NL)	G5	−	+	ND
11g	Wound	Rotterdam (NL)	G2	−	+	ND
12g	Toe web	Rotterdam (NL)	E6	−	+	ND
13g (I)	Bronchus	Nijmegen (NL)	F3	+	−	ND
14g	Not known	Not known	F1	−	−	−
16g	Urine	Rotterdam (NL)	E12	−	+	ND
17g	Wound	Rotterdam (NL)	G1	−	+	ND
18g	Urine	Rotterdam (NL)	G3	−	+	ND
19g	Pus	Leiden (NL)	G6	−	−	A
20g	Sputum	Rotterdam (NL)	F7	−	+	ND
21g	Wound	Malmö (SE)	E5	−	−	−
22g	Urine	Dordrecht (NL)	E9	+	−	ND
23g	Drain	Rotterdam (NL)	F6	−	+	ND
24g •	Urine	Rotterdam (NL)	NI	+	−	ND
26g	Wound	Nijmegen (NL)	E11	+		ND
27g	Sputum	Dordrecht (NL)	F8	+		ND
28g	Urine	Dordrecht (NL)	E10	+	−	ND
29g (I)	Sputum	Utrecht (NL)	F2	+	−	ND
30g	Urine	Rotterdam (NL)	F4	−	−	B
31g (II)	Urine	Rotterdam (NL)	F11	−	+	ND
32g	Blood	Dordrecht (NL)	F9	+	−	ND
33g (I)	Urine	Dordrecht (NL)	C12	+	−	ND
34g •	Urine	Rotterdam (NL)	NI	−	+	ND
36g	Sputum	Rotterdam (NL)	D12	−	−	D
37g	Blood	Rotterdam (NL)	E2	−	+	ND
38g	Sputum	Rotterdam (NL)	F10	−	+	ND
39g	Sputum	Rotterdam (NL)	F12	−	+	ND
40g	Sputum	Rotterdam (NL)	G4	−	+	ND
41g	Catheter tip	Utrecht (NL)	E1	+	−	ND
42g	Ear	Leiden (NL)	F5	−	−	A
RUH 508h	Bronchus	Nijmegen (NL)	D9	+	−	ND
RUH 513h	Bronchus	Nijmegen (NL)	D10	+	−	ND
RUH 733h	Bronchus	Nijmegen (NL)	D11	+	−	ND
RUH 735h	Bronchus	Nijmegen (NL)	D8	−	−	−
RUH 937h	Urine	Nijmegen (NL)	D4	+	−	ND
RUH 975h	Pus	Nijmegen (NL)	D5	+	−	ND
RUH 1027h	Urine	Nijmegen (NL)	D6	+	−	ND
RUH 1202h	Bronchus	Nijmegen (NL)	D7	+	−	ND
RUH 1205h	Bronchus	Nijmegen (NL)	D3	−	−	E
RUH 1277h	Bronchus	Nijmegen (NL)	D2	+	−	ND
RUH 3204i	Tube	Nijmegen (NL)	D1	−	−	F
RUH 1093j	Sputum	Rotterdam (NL)	C6	−	−	D
RUH 1486j	Navel	Rotterdam (NL)	C7	−	−	G
RUH 1752j (ug)	Bronchus	Enschede (NL)	C8	−	−	D
RUH 1907j	Bronchus	Rotterdam (NL)	C9	−	−	J
RUH 2180j	Sputum	Nijmegen (NL)	C10	−	−	H
RUH 2688j	Pharynx	Rotterdam (NL)	C11	−	−	ND
GNU 1078j (I)	Rectal mucosa	Leuven (BE)	B10	+	−	ND
GNU 1079j (I)	Tracheal site	Salford (UK)	B12	+	−	ND
GNU 1080j	Catheter urine	Salisbury (UK)	C1	−	+	ND
GNU 1081j (ug)	Tracheal aspirate	Cork (IE)	C2	−	−	−
GNU 1083j (I)	Urine	London (UK)	C3	+	−	ND
GNU 1084j (I)	Burn wound	Sheffield (UK)	C4	+	−	ND
GNU 1086j (II)	Respiratory tract	Newcastle (UK)	C5	−	−	B
MB 142j	Skin	London (UK)	B7	−	−	−
SJH 9/MB 264j	Skin	London (UK)	B8	−	−	−
SJH 26/MB 288j	Nail fold	London (UK)	B9	−	−	J
189k	Crural ulcer	Odense (DK)	B6	−	+	ND
9771k	Urine	Næstved (DK)	B11	−	−	G
10074k	Urine	Vejle (DK)	A1	−	−	−
10086k	Urine	Vejle (DK)	A2	−	−	I
2032l	Sputum	Venlo (NL)	A3	+	−	ND
2033l	Sputum	Venlo (NL)	A4	+	−	ND
2034l	Urine	Venlo (NL)	A5	+	−	ND
2036l	Pus	Venlo (NL)	A6	+	−	ND
2037l (I)	Sputum	Venlo (NL)	A7	+	−	ND
3242l (I)	Burn wound	Basildon (UK)	A8	+	−	ND
3344l	Burn wound	Basildon (UK)	A9	+	−	ND
3347l	Burn wound	Basildon (UK)	A10	+	−	ND
3348l	Burn wound	Basildon (UK)	A11	+	−	ND
3349l	Burn wound	Basildon (UK)	A12	+	−	ND
3370l	Respiratory tract	Newcastle (UK)	B1	−	−	B
3371l	Respiratory tract	Newcastle (UK)	B2	−	−	B
3372l	Respiratory tract	Newcastle (UK)	B3	−	−	B
3373l	Respiratory tract	Newcastle (UK)	B4	−	−	B
3374l	Respiratory tract	Newcastle (UK)	B5	−	−	B
133m	Wound	Malmö (SE)	E7	−	−	−
147m	Wound	Malmö (SE)	E8	−	−	−

^aATCC, American Type Culture Collection, Rockville, Md.; NCTC, National Collection of Type Cultures, London, United Kingdom; T, type strain; (I) and (II), outbreak strains belonging to clonal groups I and II, respectively (7); (ug), outbreak strains which could not be allocated to a particular clonal group (7); •, strains used to prepare monoclonal antibodies. 

^bSource or specimen from which the strain was originally isolated. 

^cCity and country where the strain was originally isolated. SE, Sweden; NL, The Netherlands; BE, Belgium; UK, United Kingdom; IE, Ireland; DK, Denmark. 

^dPosition of the bacterial strains in dot blots (see also Fig. 4); NI, not included (homologous antigen). 

^eDot blot assays were performed with proteinase K-digested whole-cell bacterial lysates (see Materials and Methods). +, positive reaction; −, no reaction. 

^fA to J, O-antigen patterns; −, no banding pattern observed; ND, not determined. 

^gDijkshoorn et al. (9). 

^hDijkshoorn et al. (10). 

ⁱHorrevorts et al. (19). 

^jDijkshoorn et al. (7). 

^kGerner-Smidt and Tjernberg (12). 

^lDijkshoorn et al. (8). 

^mTjernberg and Ursing (42). 

Bacterial LPSs, whole-cell lysates, and proteinase K digestion.

The Acinetobacter strains against which MAbs were prepared (see below) were grown in a fermentor (10 liters), and the cells were killed with phenol and centrifuged. LPS was extracted from the sedimented bacteria by the hot phenol-water method (49) and was lyophilized. Preparation of whole-cell lysates (undiluted or diluted 1:4 in sample buffer [45]) and proteinase K digestion were performed as described previously (29).

MAbs.

MAbs were prepared by conventional protocols after immunization of mice with heat-killed bacteria. A. baumannii 24 and 34, against which rabbit immune sera have been produced in a previous study (29), were selected as immunogens. BALB/c mice (four mice per antigen) were injected intravenously on days 0, 7, 14, and 21 with 20, 20, 60, and 120 μg of antigen, respectively. Animal sera were screened on day 28 for antibodies against the respective immunogen by a dot blot assay with purified LPS as the antigen (see below). The animal whose serum had exhibited the strongest reactivity was given a booster intravenous injection on day 125 and booster intraperitoneal injections on days 126 and 127, with 200 μg of antigen administered in each injection. Two days after the last injection, the animals were exsanguinated and the spleens were removed. Spleen cells were prepared and fused at a ratio of 1:1 with mouse myeloma X63Ag8 cells by using polyethylene glycol 1500 (Boehringer Mannheim) according to conventional protocols. Primary hybridomas were screened by dot blot and enzyme immunoassay (EIA) with isolated LPS as the antigen. Relevant hybridomas were cloned three times by limiting dilution, isotyped with a commercially available isotyping kit (Bio-Rad), and purified by affinity chromatography on Protein G (Pharmacia). The antibodies were checked for purity by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and subsequent Coomassie staining. The purified MAbs were stored at −20°C until further use.

Serological assays.

EIA and Western blotting were performed as described previously (29) with LPS and proteinase K-digested whole-cell lysates, respectively, as antigens. For Western blotting, 1:4-diluted bacterial lysates were treated with proteinase K, separated by SDS-PAGE with a 10% separating gel, electrotransferred overnight onto a polyvinylidene difluoride (PVDF) membrane, and immunostained as described previously (29, 45). For dot blotting, proteinase K-digested lysates were diluted 1:3 in distilled water, of which 1 μl was dotted onto nitrocellulose. For the screening of animal sera and primary hybridoma supernatants, 0.5 μg of purified LPS was applied to the nitrocellulose membrane. After drying (37°C, 2 h), the membranes were blocked in blot buffer (45) supplemented with 10% nonfat dry milk and were immunostained as described above for the Western blots (29, 45).

Acid hydrolysis of membrane-bound LPS.

Membrane-bound LPS was hydrolyzed under acidic conditions as described in a previous study (28). Briefly, lysates (undiluted) were digested with proteinase K, separated by SDS-PAGE with a 10% separating gel, and transferred onto PVDF membranes. The membranes were then incubated at 100°C for 1 h in heat-resistant glass containers containing 0.1 M HCl. After extensive washing in blot buffer (at least six times), they were blocked in blot buffer supplemented with 10% nonfat dry milk and were immunostained with lipid A-specific MAb S1 as described previously (28). Parallel gels were stained with alkaline silver nitrate as reported elsewhere (45).

RESULTS

Immunization of mice and preparation of MAbs.

BALB/c mice were successfully immunized with heat-killed bacteria from A. baumannii 24 or 34. The primary hybridomas (n = 864) were tested for antibody reactivity by dot blotting and EIA with purified LPS as the antigen. Eleven hybridomas reacted with strain 34, whereas only 1 was observed to react with strain 24. None of the hybridomas reacted with both strains. Among the 11 hybridomas which were found to react with strain 34, the 1 with the highest reactivity was selected for further studies. The antibodies were cloned by limiting dilution (three times), isotyped, and subsequently purified by chromatography on Protein G; purity was ascertained by Coomassie staining following SDS-PAGE (data not shown). MAb S48-3-13 against strain 24 was of the immunoglobulin G3 (IgG3) isotype, and MAb S48-3-17 against strain 34 was of the IgG1 isotype. The results described below were obtained with affinity-purified MAbs.

Specificities of MAbs.

The antibodies were tested by EIA with LPS (5 μg/ml; 50 μl/well) as the antigen. MAbs S48-3-13 and S48-3-17 reacted (optical density at 405 nm, >0.2) at concentrations of 5 and 40 ng/ml, respectively, with the homologous antigen (strains 24 and 34, respectively). No heterologous reactivity (concentration of antibody yielding an optical density at 405 nm of >0.2, >5,000 ng/ml) was observed. Next, checkerboard titrations were performed with antigen concentrations of between 32 and 4,000 ng/ml (1.6 to 200 ng of antigen per well) and antibody concentrations of between 0.5 and 1,000 ng/ml. The binding curves showed that both antibodies bind to the homologous LPS over a broad range of antigen concentrations (Fig. 1). Proteinase K-treated bacterial lysates or LPSs from A. baumannii 24 and 34 were separated by SDS-PAGE, blotted onto PVDF membranes, and immunostained with the homologous or heterologous antibody. A banding pattern characteristic of that of an O-polysaccharide chain could be observed for both strains (Fig. 2 and 3, lanes 1). Identical patterns were observed when LPS was used. No heterologous reactivity was observed, and no reaction with the core lipid A region was observed when the LPSs were separated on a 15% gel (data not shown). To show that both antibodies were indeed directed against the LPS and not another polysaccharide, the following experiment was performed (28); proteinase K-treated whole-cell lysates from both strains were separated by SDS-PAGE, transferred onto a PVDF membrane, and subjected to hydrolysis in 0.1 M HCl. The free 4′-monophosphoryl lipid A (41), which remained membrane bound, could be detected in situ by MAb S1, which recognizes this partial structure (22). For both strains, the pattern was indistinguishable from that observed when the LPS was stained with the homologous antibody, thus indicating that the antibodies were indeed directed against the O polysaccharide (data not shown).

FIG. 1.

Checkerboard titrations of MAbs S48-3-13 (A) and S48-3-17 (B) by EIA with LPSs from A. baumannii 24 and 34, respectively, as solid-phase antigens. Plates were coated with antigen at concentrations of 4,000 (●), 2,000 (■), 1,000 (▴), 500 (⧫), 250 (○), 125 (□), 63 (▵), and 32 (◊) ng/ml (50 μl per well) and were reacted with MAbs at the concentrations indicated on the abscissa. OD₄₀₅, optical density at 405 nm.

FIG. 2.

Representative Western blot of S48-3-13-positive A. baumannii clinical isolates after SDS-PAGE of the proteinase K-digested whole-cell lysates (10 μl each), transfer onto a PVDF membrane, and immunostaining with MAb S48-3-13. Bacteria are, from left to right, strain 24 (homologous strain; lane 1), strain 26 (lane 2), strain RUH 508 (lane 3), strain GNU 1084 (lane 4), strain GNU 1078 (lane 5), and strain 2032 (lane 6).

FIG. 3.

Representative Western blot of S48-3-17-positive A. baumannii clinical isolates after SDS-PAGE of the proteinase K-digested whole-cell lysates (10 μl each), transfer onto a PVDF membrane, and immunostaining with MAb S48-3-17. Bacteria are, from left to right, strain 34 (homologous strain; lane 1), strain 11 (lane 2), strain 18 (lane 3), strain 40 (lane 4), strain 10 (lane 5), strain 39 (lane 6).

The MAbs were subsequently tested by dot blotting with proteinase K-treated lysates from 80 A. baumannii clinical isolates. They were found to react with 46 strains (Fig. 4; Table 1); none of the strains reacted with both antibodies. The specificities of the reactions could be confirmed by Western blotting (Fig. 2 and 3, lanes 2 to 6). For both antibodies, the ladder patterns obtained were indistinguishable from those observed after immunostaining of the respective homologous LPS. No reactivity was observed when the two antibodies were tested with the Acinetobacter strains of other genomic species or with the non-Acinetobacter strains (data not shown).

FIG. 4.

Reactivities of MAbs with A. baumannii clinical isolates by dot blotting. Bacterial lysates were diluted 1:3 in distilled water, dotted onto nitrocellulose membranes (1 μl per dot), and developed with MAb S48-3-13 (A) or MAb S48-3-17 (B). The positions of the strains on the membrane are listed in Table 1.

Determination of LPS phenotypes.

To determine their LPS phenotypes, those strains (n = 34) which had not reacted with either antibody were also subjected to acid hydrolysis. A representative Western blot of the patterns obtained is shown in Fig. 5, and the patterns for the 34 strains are listed in Table 1. All patterns so obtained were distinguishable from those observed for strains 24 and 34 obtained by Western blotting with the homologous antibody and may thus represent the patterns for additional serotypes within the species A. baumannii. No banding patterns were observed following staining with alkaline silver nitrate (data not shown).

FIG. 5.

Representative Western blot of S48-3-13- and S48-3-17-negative A. baumannii clinical isolates after SDS-PAGE of the proteinase K-treated bacterial lysates (20 μl each), transfer onto a PVDF membrane, hydrolysis at 100°C for 1 h in 0.1 M HCl, and immunostaining with lipid A-specific MAb S1. Bacteria are, from left to right, strain 42 (lane A), strain GNU 1086 (lane B), strain NCTC 7844 (lane C), strain 36 (lane D), strain RUH 1205 (lane E), strain RUH 3204 (lane F), strain RUH 1486 (lane G), strain RUH 2180 (lane H), strain 10086 (lane I), and strain RUH 1907 (lane J).

DISCUSSION

LPSs are amphiphilic molecules imbedded in the outer membranes of gram-negative bacteria (24, 25, 32). Serological and chemical analyses of the LPSs from several Acinetobacter strains have shown they are of the smooth phenotype (13–17, 45, 46). Since S-LPS has been used as a basis for serotyping schemes for various bacterial species in the past (27, 32, 34, 35), such LPS antibodies could also be expected to be of value for the differentiation of Acinetobacter strains. This technique could then also be implementable in clinical microbiology laboratories, which lack simple methods for the phenotypic identification of Acinetobacter strains (3, 6, 43). By using rabbit antisera, this hypothesis was proven to be correct (29). However, although they were shown to be highly specific, the antisera have the disadvantage that they contain core-reactive and non-LPS antibodies, e.g., capsular and protein antibodies, which would lead to false-positive results when the sera are used for identification purposes (29). MAbs, however, which react only with the O antigen of the LPS can be generated, and this specific reactivity thus makes them more suitable for such a scheme. Moreover, the latter approach offers the advantage that virtually unlimited amounts of antibodies of homogeneous specificity can be made available.

Since numerous studies have now confirmed that A. baumannii is the most prevalent species associated with outbreaks (3, 43), MAbs were generated against the LPSs of two A. baumannii strains and were investigated for their potential use for the identification of strains belonging to this species. Two antibodies, S48-3-13 (IgG3) and S48-3-17 (IgG1), were selected and were shown to be highly specific for the homologous LPS by EIA and Western blotting. The MAbs were then tested by dot and Western blotting with 80 other clinical isolates previously identified by DNA-DNA hybridization to be A. baumannii. MAb S48-3-13 was found to react with 31 strains, and MAb S48-3-17 was found to react with 15 strains; none of the strains reacted with both antibodies. Among these 46 strains, reactivity was observed with 11 of 14 strains (Table 1) which we are certain were associated with nosocomial outbreaks in different northwestern European cities (7, 8), although one cannot exclude the possibility of potential “outbreak”-related strains among the other isolates included in the mentioned studies. In the study by Dijkshoorn et al. (7) performed in 1995, the distinction of four groups of A. baumannii strains (numbered 1 to 4) by at least one genomic and one other typing method was explained by the possibility that the respective strains originated from a common clonal origin (26). These four groups were therefore termed clones I to IV (7). Interestingly, the 11 outbreak strains which were noted to react with the two antibodies belonged to A. baumannii clonal group I or II (Table 1); MAb S48-3-13 reacted only with isolates of clone I (n = 9) and MAb S48-3-17 reacted only with those of clone II (n = 2). The only exception was strain GNU 1086 (clone II), which did not react with either of the two MAbs. Although initially surprising, this finding may be explained by observations made in the study by Dijkshoorn et al. (7), in which strain GNU 1086 was found to have a ribotype pattern different from those of the other strains within this clonal group, despite identical amplified fragment length polymorphism and protein profile patterns (biotypes and antibiograms are not considered since these were not unique for this group [7]). Thus, this strain could represent a variant within this particular clonal group with which MAb S48-3-17 is unable to react. Another possibility is that this strain has an entirely different clonal origin, despite having a high degree of similarity to the strains within clonal group II. This hypothesis is supported by the ladder pattern which was observed following acid hydrolysis of the LPS (see below), which differed from that observed for the LPS of strain 34 when the LPS was immunostained with MAb S48-3-17.

Strains of clones I and II have been proposed to have virulence factors related to invasiveness, transmissibility, or an enhanced ability to colonize immunocompromised patients and therefore should be a cause of concern in hospitals (7). Thus, since the two MAbs reacted with strains having characteristics of clones I and II, respectively, they provide an easy way of tracing such strains in the hospital environment (7), e.g., by means of a simple (latex) agglutination test. The other two isolates, strains GNU 1081 and RUH 1752, could not be grouped in the previously mentioned study (7) and, as postulated (7), may represent additional clones or groups within this species.

Since no staining with alkaline silver nitrate was observed for those strains which had failed to react with MAb S48-3-13 or MAb S48-3-17, we used a method in which lipid A, which is membrane bound following acid hydrolysis of LPS which has been blotted onto a PVDF membrane, is detected with specific MAbs, thus allowing LPS phenotype determination. Ten additional banding patterns were identified. None of the patterns were identical, and they also differed from the patterns observed for the LPSs from strains 24 and 34 when the LPSs were immunostained with the homologous antibody; thus, they may represent additional serotypes within this species. Some strains did not show a banding pattern, which may be due to a reduced level of O-antigen expression or the natural production of LPS which is of the rough phenotype. The lack of staining with alkaline silver nitrate has also been observed with other Acinetobacter strains, and possible reasons for this phenomenon have extensively been discussed elsewhere (13, 14, 17, 45, 46).

The possibility of the use of a scheme for the identification of Acinetobacter strains based on their O antigens is clearly demonstrated in this report. The two antibodies described herein are highly specific for the O antigen of the LPS from a large number of A. baumannii strains; no reactivity was observed with Acinetobacter strains of other genomic groups, including the pheno- and genotypically closely related genomic species 1 (A. calcoaceticus), 3, and 13 sensu Tjernberg and Ursing (42), or with strains of other gram-negative bacterial genera and species, such as Salmonella, E. coli, Serratia, Pseudomonas, S. maltophilia, or B. cepacia (data not shown). The LPSs of the A. baumannii strains which did not react with either of the two antibodies as well as the LPSs of strains belonging to genomic species which are not clinically relevant will be characterized in our laboratory, and MAbs against the O antigen will be generated in the future to fill the present gaps.