Prevalence of O1/K1- and O2/K3-Reactive Actinobacillus suis in Healthy and Diseased Swine

Abstract

A cell surface antigen-typing system was devised for the swine pathogen Actinobacillus suis and used to examine the prevalence of different lipopolysaccharide (O) types in healthy and diseased pigs. The strains examined in this study were isolated from a variety of locations in Canada and from Kansas. Lipopolysaccharide preparations of 151 isolates of A. suis were characterized by immunoblotting using polyclonal antisera generated to strains SO4 (O1/K1), H89-1173 (O2/K3), and VSB 3714, a rough strain. Approximately 54% (62 of 114) of A. suis isolates from diseased pigs, all (11 of 11) isolates from healthy pigs, and all (4 of 4) reference strains reacted with O1/K1 antiserum. More than 80% (18 of 22) of A. suis strains used for bacterin production and approximately 41% (47 of 114) of isolates from diseased pigs bound O2/K3 antiserum. One isolate appeared to be rough, and five were untypeable. O1/K1- and O2/K3-reactive strains were equally prevalent in Kansas, whereas O2/K3-reactive strains were more common in Québec and western Canada and O1/K1 strains were most common in Ontario. The fact that virtually all of the strains submitted for bacterin production were O2/K3-reactive strains is consistent with the notion that these strains may be more virulent than O1/K1 strains; alternatively, this may reflect geographic or other biases. In addition, we observed cross-reactivity between A. suis cell surface antigens and swine antisera to several other important pathogens. This finding may explain why previous attempts to develop a simple serodiagnostic test for A. suis have been unsuccessful.

The gram-negative bacterium Actinobacillus suis is an opportunistic pathogen of conventionally reared swine that can cause sporadic disease following stress (10–12, 16). In high-health-status herds, however, A. suis may be a considerable threat either when the organism is introduced or when high-health-status animals are mixed with conventionally reared swine. In very young pigs, A. suis infection is characterized by an acute septicemia with a high mortality rate (3, 15, 16, 18). Cyanosis, respiratory distress, neurological disturbances, and arthritis can also be seen (17). The course of the disease in mature animals in conventional herds may be less severe and can include erysipelas-like lesions, abortion, metritis, and meningitis (9, 12, 19). In grow/finish and adult animals in high-health-status herds, however, A. suis can cause septicemia with lung lesions that superficially resemble Actinobacillus pleuropneumoniae pleuropneumonia (25).

The pathogenicity of the A. suis disease is not well understood, although it is likely that two RTX toxins (ApxI_{var. suis} and ApxII_{var. suis} [22]), urease (2), capsular polysaccharide (CPS), and lipopolysaccharide (LPS) contribute to virulence. Although early studies suggested that A. suis isolates from Canada were homogeneous (1, 22), the present work has shown that there are at least two serologically distinct groups. Consistent with this finding, chemical characterization of the major surface polysaccharides of selected A. suis strains (13) revealed two different lipopolysaccharide O-antigen types; the O1 antigen, which is a homopolymer with the structure →6)-β-d-Glc-(1→6)-β-d-Glc-(1→, and the O2 antigen, which contains a [Glc, Gal₂, GlcNAc] branched tetrasaccharide (13). Therefore, the purpose of the present study was to determine the prevalences of different cell surface antigen types of A. suis and assess whether there was a correlation between antigen type and disease, date of isolation, or location of isolation.

MATERIALS AND METHODS

Bacterial strains and growth condition.

The A. suis strains used in this study were obtained from various locations throughout Canada and from a single laboratory in the United States (Table 1). Eleven isolates from healthy pigs and 66 isolates from diseased pigs have been described in detail previously (22). A further eight clinical isolates obtained from the Animal Health Laboratory, Guelph, Canada, in 1998 were also evaluated. The 19 strains from Kansas (isolated between 1995 and 1997) were a generous gift of B. Fenwick, Kansas State University. M. Gottschalk kindly donated 20 A. suis isolates obtained from outbreaks in Québec in the late 1990s. The clinical isolates studied were recovered in significant numbers from animals with a clinical picture consistent with A. suis infection (22). J. Gallant of Gallant Custom Laboratories Inc., Guelph, provided 22 A. suis strains from across Canada that were used for production of autogenous bacterins. These isolates, collected over a 2 1/2-year period (1997 to 1999), were all from animals in herds with serious A. suis disease outbreaks. Four reference strains from the American Type Culture Collection (ATCC) were included for comparison. These strains were obtained from clinically healthy pigs following exposure to atomic radiation (24). A. suis isolates were routinely grown overnight at 37°C in 5% CO₂ on blood agar plates containing 5% sheep blood or in brain heart infusion broth (Difco Laboratories, Detroit, Mich.).

TABLE 1.

Immunoblot reactions of Hitchcock-Brown preparations (6) of A. suis isolates from healthy and diseased animals with polyclonal hyperimmune rabbit O1/K1 and O2/K3 antisera

Source (n)	No. (%)
	Having reaction with:		Untypeable
	O1/K1	O2/K3
Clinical isolates
Total (114)	62 (54.4)	47 (41.2)	5 (4.4)
Ontario (75)	52 (69.3)	21 (28.0)	2 (2.7)
Kansas (19)	9 (47.4)	8 (42.1)	2 (10.5)
Québec (20)	1 (5.0)	18 (90.0)	1 (5.0)
Bacterina isolates
Total (22)	3 (13.6)	18 (81.8)	1 (4.5)
Alberta (6)	0 (0.0)	6 (100.0)	0 (0.0)
British Columbia (4)	0 (0.0)	4 (100.0)	0 (0.0)
Québec (8)	2 (25.0)	6 (75.0)	0 (0.0)
Saskatchewan (4)	1 (25.0)	2 (50.0)	1 (5.0)
Slaughterhouse isolates (11)	11 (100.0)	0 (0.0)	0 (0.0)
ATCC strains (4)	4 (100.0)	0 (0.0)	0 (0.0)

^aUsed as a source of autogenous vaccines. 

Cell surface polysaccharide preparation.

Whole-cell lysates were prepared using the proteinase K digestion method of Hitchcock and Brown (6) with a number of modifications. Five-milliliter A. suis cultures were grown in brain heart infusion broth overnight with vigorous aeration (200 rpm). The optical densities (at 600 nm) of the overnight cultures were adjusted to 1.00, and bacterial cells were harvested from 1.5 ml of culture by centrifugation (27,000 × g, 5 min, 20°C). After the supernatant was discarded, the cell pellets were solubilized in 100 μl of lysing buffer (2% sodium dodecyl sulfate, 4% 2-mercaptoethanol, 10% glycerol, 1 M Tris [pH 6.8], and 0.05% bromophenol blue) and heated at 100°C for 35 min. Proteinase K (25 mg/ml) (Gibco BRL, Life Technologies, Inc., Burlington, Ontario, Canada) was added, and the cell lysates were incubated at 65°C for 60 min. After protease digestion, an additional 250 μl of lysing buffer was added, and the lysates were incubated at 100°C for 5 min. The digested samples were stored at 4°C until analyzed.

Immunoblotting.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was carried out as described previously (7) using 12% resolving and 5% stacking gels. Immediately prior to loading, all samples were heat treated (5 min, 100°C), and 8 μl was loaded per lane. Electrophoretic transfer of gels to nitrocellulose membranes (0.45 μm pore size; Bio-Rad Laboratories) was performed overnight at 4°C with constant voltage (30 V) in a Bio-Rad transfer system with the transfer buffer described by Towbin et al. (21). Following transfer, the membranes were placed in 5% (wt/vol) skim milk (Difco Laboratories) in phosphate-buffered saline (PBS) and incubated at room temperature for 7 h to block remaining binding sites. The membranes were then incubated overnight with swine or rabbit antisera diluted 1:100 in 2% skim milk in PBS. After being washed three times with 0.05% Tween 20 in PBS for 5 min, the membranes were rinsed briefly three times with PBS and then incubated with alkaline phosphatase-conjugated protein A (Sigma Chemical Co.) diluted 1:500 in 2% skim milk. The washing steps were repeated, and the immunoblot reaction products were visualized in 10 ml of buffer (0.1 M Tris, 0.09 M NaCl, and 0.15 M MgCl₂ · 6H₂O [pH 9.5]) containing 35 μl of 5-bromo-4-chloro-3-indoylphosphate (BCIP) (Bio-Rad Laboratories) and 45 μl of nitroblue tetrazolium (Bio-Rad Laboratories). The reactions were terminated by washing the membranes two times for 5 min each with distilled water.

Antigen and antiserum preparation.

Whole-cell antigens from strains H89-1173 and VSB 3714 were prepared by washing confluent 18-h cultures from eight blood agar plates with approximately 4 ml of 0.25% formalin in sterile physiological saline. Bacteria were harvested by centrifugation (12,000 × g, 10 min, 4°C) and washed once with normal saline, and the optical density at 525 nm was adjusted to 1.00. Specific-pathogen-free (SPF) New Zealand rabbits were bled to obtain preimmune serum and then injected intravenously at 3- to 4-day intervals for 4 weeks with increasing doses of antigen (0.5 ml on day 1; 1.0 ml on day 5; 2.0 ml on day 9; and 3 ml on days 13, 17, 21, and 25). Antisera were collected 6 to 7 days after the final injection and stored at −20°C. For production of antiserum against strain VSB 3714, two additional injections were done (3 ml on day 48 and 3 ml on day 62). Antiserum against strain SO4 was generously provided by the late S. Rosendal, University of Guelph (22). Swine antisera to A. suis; A. pleuropneumoniae serotypes 1, 5, and 7; Pasteurella multocida; and Haemophilus parasuis were the kind gift of B. Fenwick, Kansas State University. These antisera were generated in SPF pigs known to be free of A. suis.

Serum absorption.

In a concurrent study, we found that antibodies to the A. suis O1/K1 antigen are ubiquitous in rabbits and other species (13). Accordingly, the A. suis H89-1173 (O2/K3) antisera were extensively preabsorbed with A. suis H93-0055 (O1/K1) antigen prior to use. Briefly, confluent 18-h cultures were washed off four blood agar plates with ∼4 ml of 0.4% formalin in sterile physiological saline per plate and concentrated by centrifugation (2,500 × g, 10 min, 4°C). The bacterial pellet was then suspended in 4.5 ml of PBS containing 0.02% (wt/vol) sodium Merthiolate as a preservative. After addition of 0.5 ml of the antiserum to be absorbed, the suspension was incubated at 37°C for 2 to 3 h on a rotary mixer and then overnight at 4°C. Bacterial cells were removed by centrifugation (27,000 × g, 1 h, 4°C) followed by filtration (0.2-μm-pore-size sterile Acrodisc; Gelman Sciences), and the absorbed antisera were stored at −20°C. The efficacy of absorption was verified by immunoblotting (Fig. 1B, lanes 1 to 4).

FIG. 1.

Immunoblots of Hitchcock-Brown preparations (6) of A. suis LPS probed with antisera to A. suis strains SO4 (O1/K1) (A), H89-1173 (O2/K3) (B), and VSB 3714 (rough/K?) (C). Lanes: 1, A. suis SO4; 2, H93-0055; 3, B49; 4, C84; 5, H91-0380; 6, H89-1173; 7, H92-2146; 8, VSB 3714.

RESULTS

Immunoblotting.

Representative immunoblots of A. suis Hitchcock-Brown preparations (6) probed with hyperimmune rabbit antiserum to strain SO4 (O1/K1), and absorbed H89-1173 (O2/K3) serum are shown in Fig. 1A and B. Molecules reacting with O1/K1 antiserum were seen as a broad high-molecular-weight smear (Fig. 1A, lanes 1 to 4). A narrower, high-molecular-weight band was seen in preparations that reacted with H89-1173 (O2/K3) antiserum (Fig. 1B, lanes 5 to 7). Material that reacted with the O1/K1 antiserum did not react with absorbed O2/K3 antiserum and vice versa (Fig. 1A, lanes 5 to 7, and B, lanes 1 to 4). A similar pattern of binding was seen when purified SO4 and H89-1173 LPSs were used (data not shown). Hitchcock-Brown preparations of A. suis VSB 3714 did not react with either the O1/K1 or the O2/K3 antiserum (Fig. 1A, lane 8, and B, lane 8, respectively). Conversely, when O1/K1- and O2/K3-reactive A. suis isolates were examined with antiserum to VSB 3714, a dense, low-molecular-weight band was observed (Fig. 1C).

Prevalences of serotypes.

The prevalences of O1/K1- and O2/K3-reactive strains of A. suis from both diseased pigs and healthy pigs are summarized in Table 1. Approximately 54% (62 of 114) of clinical isolates of A. suis, all (11 of 11) isolates from healthy pigs, and all (4 of 4) reference strains reacted with the O1/K1 antiserum. More than 80% (18 of 22) of A. suis strains used for bacterin production and approximately 41% (47 of 114) of clinical isolates bound O2/K3 antiserum. While the O1/K1-reactive strains were common in Ontario (approximately 52%), O2/K3-reactive isolates were more frequently isolated in Québec (approximately 90%). There was an almost equal distribution of the two serotypes in Kansas. Five strains (approximately 4%) had only a very weak reaction with O2/K3 antiserum and no reaction with O1/K1 antiserum. These five strains did, however, react with VSB 3714 antiserum.

Cross-reactivity.

All of the preimmune rabbit antisera used for this work reacted with purified O1/K1 antigen (data not shown) (13). Monteiro et al. have also shown that there is strong cross-reactivity with equine and bovine sera and weak cross-reactivity with preimmune SPF swine sera and pure O1 antigen (13). Similarly, cross-reactivity was observed when convalescent-phase sera from SPF swine infected with A. pleuropneumoniae serotypes 1, 5, and 7, P. multocida, or H. parasuis were used to examine the LPSs of O1/K1 strains. (Figs. 2A, lanes 1 to 4, and data not shown). In contrast, strong cross-reactivity was observed when LPS preparations from O2/K3-reactive strains were evaluated with swine antisera to A. pleuropneumoniae serotype 1 (Fig. 2A, lanes 5 to 8) and A. pleuropneumoniae serotype 7 (Fig. 2C, lanes 5 to 8). Some cross-reactivity could also be seen with the A. pleuropneumoniae serotype 5 antiserum (Fig. 2B, lanes 5 to 8) and these strains, while weak cross-reactivity with P. multocida antiserum (Fig. 2D, lanes 5 to 8) and very weak binding with H. parasuis antiserum (data not shown) were observed.

FIG. 2.

Hitchcock-Brown preparations (6) of selected A. suis strains probed with antisera to A. pleuropneumoniae serotype 1 (A), A. pleuropneumoniae serotype 5 (B), A. pleuropneumoniae serotype 7 (C), P. multocida (D). Lanes: 1, A. suis SO4; 2, H93-0055; 3, B49; 4, C84; 5, H91-0380; 6, H89-1173; 7, H92-2146; 8, H90-2521.

DISCUSSION

Based on a variety of genetic and phenotypic properties, it was initially thought that A. suis isolates were very homogeneous (1, 22). The present study has revealed, however, that there are at least two commonly occurring, serologically related groups of A. suis in Canada and Kansas: O1/K1 reactive and O2/K3 reactive. We designated the A. suis isolates studied O/K reactive rather than assigning them to a specific O serotype because crude preparations of cell surface antigens were used in these experiments. Although Hitchcock-Brown preparations (6) are considered to contain mostly LPS, CPS anchored to the lipid A core can also be present and could have cross-reacted with the polyclonal antiserum used. While there are some indications (discussed below) that this is not likely, we could not completely exclude this possibility.

In immunoblots using homologous sera, the O1/K1-reactive antigen was visible as a very broad high-molecular-weight band and the O2/K3-reactive antigen appeared as a somewhat less broad high-molecular-weight band. Although there was cross-reactivity between H91-0380 (O2/K2) Hitchcock-Brown preparations and O2/K3 (H89-1173) antisera, no reactivity was observed with C84 (O1/K2) Hitchcock-Brown preparations, indicating that band observed is likely due to the identical LPS rather than CPS. When VSB 3714 LPS was probed with homologous antiserum, a reaction with lower-molecular-weight bands was detected. An identical pattern was seen when whole-cell lysates of selected A. suis isolates were probed with VSB 3714 antiserum. These results and structural data (M. Monteiro, unpublished data) suggest that VSB 3714 is a rough strain and that there may be a common LPS core in A. suis. Since it is known that immunization against rough strains of Escherichia coli can protect swine against lethal challenge with A. pleuropneumoniae (4), it might be worthwhile to determine if rough LPS could be used in a vaccine to confer specific immunity to A. suis infection or to protect against other swine pathogens from the family Pasteurellaceae.

A few clinical isolates were analyzed that reacted strongly with VSB 3714 antiserum but did not react with antiserum raised against an O1/K1 isolate and reacted only very weakly with antiserum containing O2/K3 antibodies. It is not clear if these strains belong to a different serotype (or serotypes) that share some cross-reactive epitopes with the O2/K3-reactive antigen or if they were O2/K3-reactive strains which produced much less O/K-reactive antigen. At present, we consider these isolates untypeable.

Some associations between antigenic type and geographic location and virulence, but not date of isolation, could be made. All of the isolates from the healthy pigs, including the ATCC reference strains, were O1/K1 reactive, whereas only half of the clinical isolates examined were O1/K1 reactive. With a very few exceptions, the remainder of the clinical isolates were O2/K3 reactive. When strains used for bacterin production were probed with O2/K3 antiserum, more than two-thirds were found to express O2/K3-reactive antigens.

The association of O2/K3-reactive strains with disease may also be due in part to differences in CPS in these strains. Two different CPS types both rich in sialic acid, are associated with the O2-containing strain (M. Monteiro, unpublished data). In contrast, O1-containing isolates appear to have mostly →6)-β-d-Glc-(1→6)-β-d-Glc-(1→ capsule. Since antibodies to →6)-β-d-Glc-(1→6)-β-d-Glc-(1→ are frequently found in preimmune sera from swine and other species (13), it is likely that these antibodies confer some degree of protection to the O1/K1-reactive antigens of A. suis. In addition, the fact that →6)-β-d-Glc-(1→6)-β-d-Glc-(1→ has been shown to have an immunostimulatory effect in mice (23) further supports the notion that O1/K1-containing strains might be less virulent than O2/K3-reactive strains. Since an association between virulence and O/K types has been established for some gram-negative bacteria (5), such a relationship may also exist for some types of A. suis. To test the hypothesis that A. suis O2 strains are more virulent than O1 strains, the pathogenicities of A. suis serotype O1/K1, O1/K2, and O2/K2 strains were evaluated by intraperitoneal challenge. The O2/K2 strain caused the most severe peritonitis and disseminated most widely to other tissues. Moderate lesions were seen with the O1/K2 strain, while the O1/K1 strain caused mild lesions and remained largely localized to the peritoneum, again consistent with the notion that O2 strains may be more virulent than O1 strains but also pointing to a role for CPS (20).

Some differences in the geographic distribution of A. suis serotypes were observed. O2/K3-reactive strains were most prevalent in Québec and western Canada, while O1/K1-reactive strains were predominant in Ontario; both groups were isolated with equal frequency in Kansas (Table 1). Interestingly, no bacterin production strains originated from Ontario despite the presence of a very large pork industry. The reasons for this discrepancy are not presently clear. It may be that the distributions of either the O1/K1-reactive or O2/K3-reactive strains of A. suis are different in different geographic locations, but there are no data to support this hypothesis. Since it is difficult to ensure retrospectively that the strains studied were truly representative, these differences could also reflect a bias in sampling, but this is not likely. However, since this paper was initially submitted, a number of severe A. suis disease outbreaks have occurred in Ontario. From these cases of sudden death of grow/finish animals, the A. suis strains isolated were all O2/K3 reactive (Đ. Slavić, unpublished data).

Antibodies to the O1 antigen of A. suis were detected in all preimmune rabbit sera used in this study and in preimmune sera from several other species (13). Weak cross-reactivity was also detected with some swine antisera (13). In contrast, strong reactions to the O2/K3 antigens were observed with convalescent-phase swine antisera to A. pleuropneumoniae serotypes 1 and 7 (Fig. 2A and C). These results are in contrast to a report by Lebrun et al. (8), who observed no reactivity with a panel of monoclonal antibodies specific to the O side chain of A. pleuropneumoniae serotype 7 and LPS of a field isolate of A. suis. Since the O/K type of the A. suis isolate used in their study was not known, the observed discrepancy may be explained by the absence of O2/K3-reactive antigen. Some cross-reactivity was also observed when antisera to A. pleuropneumoniae serotype 5, P. multocida, and H. parasuis were used (Figs. 2B and D). The cross-reactivity between A. pleuropneumoniae convalescent-phase sera and the A. suis O2/K3-reactive isolates was unexpected, as neither the O side chains of LPS nor the CPSs of these organisms share a common structure (14). In the cases of P. multocida and H. parasuis, the structures of their cell surface carbohydrates are not known, so the basis of this cross-reactivity also remains to be determined.

In conclusion, this is the first report of the identification of two serologically distinct groups of A. suis. From these studies we have shown that polyclonal hyperimmune rabbit antisera to O1/K1 and O2/K3 serotypes may be used for routine serotyping of A. suis isolates, although O2/K3 antiserum must be absorbed to eliminate antibodies to (1→6)-β-d-glucan, which appear to be ubiquitous. Further, these experiments provide an explanation for why attempts to develop serodiagnostic tests for A. suis infection based on LPS or CPS have been plagued by false-positive results. In addition, these data provided a preliminary indication that O2-containing strains may be more virulent than their O1 counterparts.