Evidence for invasion of a human oral cell line by Actinobacillus actinomycetemcomitans

Actinobacillus actinomycetemcomitans, an oral bacterial species associated with periodontal disease, was found to invade human cell lines. Invasion was demonstrated by recovery of viable organisms from gentamicin-treated KB cell monolayers and by light and electron microscopy. Internalization occurred through a cytochalasin D-sensitive process. Invasion efficiencies of some A. actinomycetemcomitans strains were comparable to those of invasive members of the family Enterobacteriaceae. Differences in invasiveness were correlated with bacterial colonial morphology. Smooth variants invaded more proficiently than rough variants. A. actinomycetemcomitans can undergo a smooth-to-rough colonial morphology shift which results in the loss of invasiveness. Coordinated regulation of genes involved in the rough-to-smooth phenotypic transitions may play a role in the episodic nature of periodontal disease.

A longitudinal microbiological investigation of Actinobacillus actinomycetemcomitans and Eikenella corrodens in juvenile periodontitis

Longitudinal clinical and microbiological monitoring of subjects with localized juvenile periodontitis indicated that Actinobacillus actinomycetemcomitans and Eikenella corrodens were significantly associated (P less than 0.05) with active tissue destruction.

Ecology and significance of Pasteurellaceae in animals

The reservoir of eighty-one taxa/groups classified with the family Pasteurellaceae Pohl 1981 is reviewed based upon published data and own investigations. With the exception of certain strains of P. multocida, A. pleuropneumoniae and [H.] paragallinarum organisms belonging to this family are usually regarded as opportunistic, secondary invaders which under normal conditions coexist peacefully with the animal host on mucosal membranes of the upper respiratory- and lower genital tracts. Very little is known about factors that govern the ecological preferences that certain members of this family show for specific surfaces and hosts. Mechanisms of colonization, survival and multiplication, invasion and pathogenic action are incompletely understood. The significance of Pasteurellaceae in animals and man has recently been reviewed. Subsequent publications have underlined the significance of biovars 2 of P. canis and P. avium and ornithine negative P. multocida in pneumonia in cattle. In addition, differences in pathogenicity have been demonstrated for different serovars of [H.] parasuis. The disease potential of many taxa/groups is only incompletely known.

Actinobacillus species and their role in animal disease

Actinobacillus species are Gram-negative bacteria responsible for several quite distinct disease conditions of animals. The natural habitat of the organisms is primarily the upper respiratory tract and oral cavity. A. lignieresii is the cause of actinomycosis (wooden tongue) in cattle: a sporadic, insidiously-developing granulomatous infection. In sharp contrast is A. pleuropneumoniae which is responsible for a rapidly spreading often fatal pneumonia, common among intensively reared pigs. Detailed investigation of this organism has provided a much clearer picture of the bacterial factors involved in causing disease. A. equuli similarly causes a potent septicaemia in the neonatal foal; growing apparently unrestricted once infection occurs. Other members of the genus induce characteristic pathogenesis in their preferred host, with one, A. actinomycetemcomitans, being a cause of human periodontal disease. This article reviews recent understanding of the taxonomy and bacteriology of the organisms, and the aetiology, pathogenicity, diagnosis and control of animal disease caused by Actinobacillus species.

Oral pathogens: from dental plaque to cardiac disease

Oral bacteria exhibit highly specific adherence mechanisms and as a result they colonize and cause disease principally in the oral cavity. Oral pathogens, however, can produce systemic disease and are known causative agents of infective endocarditis. Recent studies have revealed that periodontal disease per se is also a statistically significant risk factor for cardiovascular disease. A link between the two diseases is the secretion and systemic appearance in periodontitis of pro-inflammatory cytokines capable of eliciting effects associated with atherosclerosis and coronary heart disease.

Receptors for transferrin in pathogenic bacteria are specific for the host's protein

Transferrin receptors detected by a solid-phase binding assay were shown to be specific for the host's transferrin in the representative bacterial pathogens Neisseria meningitidis (human), Pasteurella haemolytica (bovine), and Actinobacillus pleuropneumoniae (porcine). Consistent with the receptor specificity, iron-deficient bacteria were only capable of utilizing transferrin from the host as a source of iron for growth.

Regulation of leukotoxin in leukotoxic and nonleukotoxic strains of Actinobacillus actinomycetemcomitans

Actinobacillus actinomycetemcomitans is a gram-negative bacterium that has been implicated in the etiology of several forms of periodontitis, especially localized juvenile periodontitis. A potent leukotoxin (Lkt) is produced by most A. actinomycetemcomitans isolates from patients with periodontal disease, but some isolates are leukotoxin nonproducing (Lkt-). The molecular bases for the differences in leukotoxin expression are being explored to clarify the role of leukotoxin in pathogenesis. We have previously cloned the leukotoxin structural gene, lktA, from the leukotoxin-producing (Lkt+) strain JP2 and have shown that it is linked to three other genes, lktB, lktC, and lktD, whose gene products are thought to be required for activation and localization of the leukotoxin. These genes have now been used in Southern blot analysis to demonstrate that Lkt- strains, like Lkt+ strains, contain all four genes of the lkt gene cluster. While restriction fragment length polymorphisms were detected, they did not correlate with toxin phenotype. RNA blot analysis demonstrated that Lkt+ strains produced two transcripts, one 9.3 kb in length and the other 4.3 kb. They encode lktCABD and lktCA. respectively. Lkt- strains contained significantly lower levels of the 4.3-kb transcript with no discernible 9.3-kb message. The leukotoxic activity of the A. actinomycetemcomitans strains, measured by chromium release assays, correlated with the lkt RNA content. Therefore, a major component of leukotoxin regulation is at the level of RNA transcription or stability. Interestingly, the lkt RNAs in JP2 are regulated during growth phase, being greatly reduced in cells approaching stationary phase. Thus, the regulation of lkt RNA can be affected by both genotype and environment.

Immunoserological comparison of 104-kilodalton proteins associated with hemolysis and cytolysis in Actinobacillus pleuropneumoniae, Actinobacillus suis, Pasteurella haemolytica, and Escherichia coli

A homologous polyclonal antibody was produced in a rabbit to the 104-kilodalton (kDa) protein hemolysin of Actinobacillus pleuropneumoniae serotype 1 strain CM-5. In immunoblots, this antibody recognized a similar 104-kDa protein produced in culture supernatants by A. pleuropneumoniae serotypes 1 to 12 and taxon "Minor group" in addition to Pasteurella haemolytica, Actinobacillus suis, and alpha-hemolysin-producing Escherichia coli (but only weakly in the latter two organisms). These results were reproduced by using a mouse monoclonal antibody to the CM-5 104-kDa protein hemolysin, except that the monoclonal antibody bound more strongly to the alpha-hemolysin produced by E. coli, only weakly to the 104-kDa protein produced by "Minor group," and not at all to any extracellular antigens produced by A. suis. Pigs experimentally infected with A. pleuropneumoniae serotypes 1 to 10 and A. suis produced an antibody that recognized the 104-kDa hemolysin produced by CM-5. A pig challenged with a "Minor group" strain did not have such antibodies. Rabbit antiserum produced against the leukotoxin of P. haemolytica and alpha-hemolysin-producing E. coli also recognized the CM-5 hemolysin, but the latter only weakly. The hemolytic activity produced by CM-5 in culture supernatant was neutralized strongly by the pig serum to serotypes 1, 2, 5, 6, 9, and 10 and A. suis, only partially by serotype 8 antiserum and the rabbit antiserum to P. haemolytica leukotoxin, and not all by the antiserum to serotypes 3, 4, and 7 and "Minor group" and the E. coli alpha-hemolysin. These results indicate that a similar but not identical 104-kDa protein is produced in vitro and in vivo by all serotypes of A. pleuropneumoniae and may be related to cytolysins produced by other gram-negative bacteria.

Identification of the pangenome and its components in 14 distinct Aggregatibacter actinomycetemcomitans strains by comparative genomic analysis

Background

Aggregatibacter actinomycetemcomitans is genetically heterogeneous and comprises distinct clonal lineages that may have different virulence potentials. However, limited information of the strain-to-strain genomic variations is available.

Methodology/Principal Findings

The genome sequences of 11 A. actinomycetemcomitans strains (serotypes a-f) were generated de novo, annotated and combined with three previously sequenced genomes (serotypes a-c) for comparative genomic analysis. Two major groups were identified; serotypes a, d, e, and f, and serotypes b and c. A serotype e strain was found to be distinct from both groups. The size of the pangenome was 3,301 genes, which included 2,034 core genes and 1,267 flexible genes. The number of core genes is estimated to stabilize at 2,060, while the size of the pangenome is estimated to increase by 16 genes with every additional strain sequenced in the future. Within each strain 16.7–29.4% of the genome belonged to the flexible gene pool. Between any two strains 0.4–19.5% of the genomes were different. The genomic differences were occasionally greater for strains of the same serotypes than strains of different serotypes. Furthermore, 171 genomic islands were identified. Cumulatively, 777 strain-specific genes were found on these islands and represented 61% of the flexible gene pool.

Conclusions/Significance

Substantial genomic differences were detected among A. actinomycetemcomitans strains. Genomic islands account for more than half of the flexible genes. The phenotype and virulence of A. actinomycetemcomitans may not be defined by any single strain. Moreover, the genomic variation within each clonal lineage of A. actinomycetemcomitans (as defined by serotype grouping) may be greater than between clonal lineages. The large genomic data set in this study will be useful to further examine the molecular basis of variable virulence among A. actinomycetemcomitans strains.

Effects of cations and osmotic protectants on cytolytic activity of Actinobacillus actinomycetemcomitans leukotoxin

Actinobacillus actinomycetemcomitans leukotoxin permeabilized the plasma membrane of HL-60 promyelocytic leukemia cells, resulting in colloid osmotic lysis. These events were associated with efflux of 51chromium (from prelabeled cells), influx of propidium iodide, and ultrastructural evidence of cellular damage. Target cell lysis was inhibited by procedures which may interfere with the initial interaction of the toxin with the plasma membrane. For example, washing cultures (to dilute and remove toxin) or the addition of monoclonal antibodies (to neutralize toxin) or trypsin (to inactivate toxin) limited lysis when undertaken within the first 5 min of the reaction. The extent of injury was also diminished when radiolabeled HL-60 cells were exposed to toxin in the presence of unlabeled, toxin-sensitive cells (e.g., HL-60 cells or human neutrophils) or certain toxin-resistant target cells (e.g., human K562 erythroleukemia cells). This suggests that the association of the toxin with the cell membrane may not be sufficient to cause lysis without activation of additional effector mechanisms. The addition of specific trivalent (e.g., La3+) or divalent (e.g., Ca2+ and Zn2+) cations to toxin-treated cells appeared to enhance their capacity to repair or minimize the extent of toxin-mediated membrane damage. Depending on size, certain saccharides served as osmotic protectants: maltose almost completely inhibited radiolabel release, while smaller molecules provided correspondingly less protection. The results imply that the leukotoxin has membranolytic activity, producing pores in target cells with a functional diameter approximately the size of maltose (0.96 nm).

Role of lipopolysaccharides in adherence of Actinobacillus pleuropneumoniae to porcine tracheal rings

The ability of 17 Actinobacillus pleuropneumoniae isolates, representing serotypes 1, 2, 5, and 7, to adhere to tracheal rings maintained in culture was examined. Porcine tracheal rings were infected, and 8 h after inoculation, adherent bacterial cells were evaluated. A. pleuropneumoniae adhered to tracheal rings, and marked variations were observed between and even within serotypes, suggesting that adherence of this microorganism is not primarily related to the serotype of the isolate. No relationship was found between adherence to porcine tracheal rings and plasmid profiles, virulence in mice, hemagglutination, capsular material thickness, or whole-cell protein profiles. On the other hand, we observed that all isolates of serotypes 1 and 5 had a semirough-type lipopolysaccharide (LPS), whereas isolates of serotypes 2 and 7 had a smooth-type LPS (75%) or a semirough-type LPS (25%). Results showed that 83% of isolates with a smooth-type LPS adhered in large numbers to tracheal rings, whereas 80% of isolates with a semirough-type LPS adhered poorly (P less than 0.007). Our data indicated that the degree of adherence of A. pleuropneumoniae to porcine tracheal rings appeared to be related, at least in part, to LPS profiles. Furthermore, LPS seemed to be the adhesin of A. pleuropneumoniae, since purified LPS blocked adherence of this microorganism to porcine tracheal rings.

Genetic approach to the study of epidemiology and pathogenesis of Actinobacillus actinomycetemcomitans in localized juvenile periodontitis

Actinobacillus actinomycetemcomitans isolates from periodontal pockets were examined for restriction fragment-length polymorphism using a characterized 4.7-kb DNA probe. A total of 6 patterns of RFLP was found in 133 isolates originating from 12 subjects. No relatedness was found between RFLP types and serotypes. Different periodontal sites within the same subject and different individuals within the same family sometimes showed only one type of A. actinomycetemcomitans RFLP. When members among the same family showed 2 RFLP types, children were always infected with the A. actinomycetemcomitans strains found in at least one of the parents. These findings support the concept of familial spread of A. actinomycetemcomitans. A. actinomycetemcomitans RFLP type B, corresponding to reference strain JP2, seems to be particularly virulent, as indicated from the presence of RFLP type B in 3 subjects who converted from a healthy periodontal state to localized juvenile periodontitis. RFLP type B was not detected in any of the 21 A. actinomycetemcomitans-infected patients with adult periodontitis. The RFLP method seems to be useful in determining the epidemiology and possibly the potential virulence of periodontal strains of A. actinomycetemcomitans.

Hemolysin patterns of Actinobacillus pleuropneumoniae

The secreted hemolytic activities produced by the reference strains and field isolates of the 12 serotypes and 2 subtypes of Actinobacillus pleuropneumoniae were analyzed. Serotype 1 produced a Ca2(+)-inducible hemolysin, which was previously characterized as a 105-kilodalton protein and was named hemolysin I (HlyI). Serotypes 2, 4, 6, 7, and 8 produced a different hemolytic activity that was not inducible by Ca2+ but required this ion for its activity. The hemolytic activity produced by these serotypes was much weaker than that found in serotype 1 and was not neutralized by rabbit antibodies against HlyI. It was, however, neutralized by serum from pigs that were experimentally infected with a serotype 2 strain and was called hemolysin II (HlyII). Serotypes 5a, 5b, 9, 10, and 11 produced both HlyI and HlyII. In these strains, HlyI was the major contributor to the hemolytic activity. The remaining serotypes, 3 and 12, produced a very weak hemolytic activity, which was not further analyzed. Immunoblot analysis of the culture supernatants from all 12 serotypes with rabbit polyclonal antibodies directed against HlyI revealed reactions with a protein in the 105-kilodalton size range for all serotypes, indicating that HlyI and HlyII might be serologically related. Strains producing active HlyI seem to belong to serotypes that are generally considered to be virulent types and that are frequently isolated from pigs in severe pleuropneumonia outbreaks.

Humoral antibody response and protective immunity in swine following immunization with the 104-kilodalton hemolysin of Actinobacillus pleuropneumoniae

Five cesarean-derived, colostrum-deprived pigs were given three adjuvant-supplemented subcutaneous and one intravenous injection of the purified 104-kDa hemolysin from serotype 1 Actinobacillus pleuropneumoniae CM-5. Six control animals received phosphate-buffered saline only. Five of six control pigs died within 24 h after challenge. The sixth control pig was moribund and euthanized after 48 h. All six pigs had pleuropneumonia, and A. pleuropneumoniae was isolated from all six lungs. None of the vaccinated pigs died as a result of challenge. After being euthanized, two pigs in this group had no lung lesions but three had chronic pleuropneumonia involving 10, 20, and 40% of the lung tissue. A. pleuropneumoniae was isolated from lung lesions of these three animals but not from the two pigs without lesions. The prechallenge hemolysin-neutralizing antibody titers in the vaccinated pigs were 1:10,900, 1:10,600, 1:4,800, 1:3,900, and 1:3,000, in order of increasing lung involvement. None of the control pigs had neutralizing antibodies. Enzyme-linked immunosorbent assay (ELISA) antibodies to capsule, lipopolysaccharide, and hemolysin were not detected in serum samples collected from the control pigs. In the vaccinated group, prechallenge sera did not contain ELISA antibodies to capsule or lipopolysaccharide. ELISA antibodies to the hemolysin were detected only in the prechallenge and postchallenge serum samples. These results indicate that pigs immunized with the 104-kDa hemolysin of serotype 1 A. pleuropneumoniae are protected against challenge with virulent bacteria. The association between neutralizing antibodies and protection indicates indirectly that the hemolysin is an important virulence factor.

Actinobacillus spp. and related bacteria in infected wounds of humans bitten by horses and sheep

We describe the isolation of Actinobacillus lignieresii and an A. equuli-like bacterium from an infected horse-bite wound in a 22-year-old stable foreman and A. suis from a bite injury in a 35-year-old man who had been attacked by a horse. A. lignieresii was also isolated in pure culture from an infected sheep-bite wound in a rural worker. These species of the genus Actinobacillus are primarily associated with animals and animal diseases and are rarely isolated from humans. The purpose of this report is to raise awareness of the possible occurrence of Actinobacillus spp. in bite wounds inflicted by farm animals and to discuss the difficulties encountered in the identification of species of Actinobacillus and related bacteria.

Immune suppression induced by Actinobacillus actinomycetemcomitans: effects on immunoglobulin production by human B cells

Actinobacillus actinomycetemcomitans produces an immunosuppressive factor (ISF) which has been shown to suppress mitogen- and antigen-induced DNA, RNA, and protein synthesis in human T lymphocytes. In this study, we examined purified A. actinomycetemcomitans ISF for its ability to alter immunoglobulin production by human B cells. The ISF caused a dose-dependent inhibition of pokeweed mitogen (PWM)-induced immunoglobulin G (IgG) and IgM production. Preexposure to ISF was not required to achieve maximal inhibition of immunoglobulin synthesis, as previously observed for its effect on T-cell activation. Nevertheless, the ISF appeared to act by irreversibly affecting the early stages of cell activation. While PWM-induced immunoglobulin production is under the influence of T-regulatory circuits, it appears that the ISF interacts directly with B cells. First, ISF failed to alter either the synthesis of interleukin-2 (IL-2) or the expression of IL-2 receptors on T cells. Second, experiments in which individual purified populations of cells were exposed to ISF, washed, and placed back into tissue culture indicated that when all cells (i.e., T cells, B cells, and monocytes) were exposed to ISF, significant suppression was observed. However, when only one cell population was treated with ISF, suppression of both IgG and IgM synthesis was observed only when the B-cell-enriched population was exposed to ISF. These results in conjunction with our earlier findings suggest that the ISF functions via the activation of a regulatory subpopulation of B lymphocytes, which in turn either directly or indirectly (via suppressor T cells) downregulate both B- and T-cell responsiveness. Furthermore, it is hypothesized that patients who harbor A. actinomycetemcomitans could suffer from local or systemic immune suppression. This suppression may enhance the pathogenicity of A. actinomycetemcomitans itself or that of some other opportunistic organism.

Killing of human myelomonocytic leukemia and lymphocytic cell lines by Actinobacillus actinomycetemcomitans leukotoxin

The purified leukotoxin of Actinobacillus actinomycetemcomitans kills human leukemic cell lines (e.g., HL-60, U937, and KG-1) and human T- and B-cell lines (e.g., JURKAT, MOLT-4, Daudi, and Raji) in a dose- and time-dependent manner. The 50% effective doses for these cell lines are similar to those established for human polymorphonuclear leukocytes and monocytes. In contrast, other human and nonhuman tumor cell lines are not susceptible to the leukotoxin. These human leukemia and lymphoid cell lines will serve as useful model systems with which to study the molecular specificity and mechanism(s) of action of the actinobacillus leukotoxin.

Characterization of a non-hemolytic mutant of Actinobacillus pleuropneumoniae serotype 5: role of the 110 kilodalton hemolysin in virulence and immunoprotection

To determine the role of hemolysin(s) in virulence and immunoprotection, non-hemolytic mutants of Actinobacillus pleuropneumoniae serotype 5, strain J45, were isolated following chemical mutagenesis. One mutant was selected for extensive characterization. Differences in capsule content, or in lipopolysaccharide or membrane protein electrophoretic profiles of the parent and mutant were not detected. A predominant, calcium-inducible protein of 110 kDa was present in culture supernatant of the parent, but absent from the mutant. Two-dimensional (2-D) gel electrophoresis confirmed that the 110 kDa protein was absent in culture supernatant of the mutant, but few, if any, minor differences could be detected in whole-cell proteins between the parent and mutant. The mutant totally lacked extracellular hemolytic and cytotoxic activity. Lysates of whole cells of the mutant contained weak hemolytic activity, and the 110 kDa protein could be detected by immunoblotting. Neutralization titers were negative in pigs immunized with the mutant or purified, denatured hemolysin, although enzyme-immunoassay titers were detected. Four additional independently isolated non-hemolytic mutants were avirulent in pigs and mice at doses greater than 10 times the lethal dose of the parent. Neither pigs nor mice were protected against lethal infection following immunization with the non-hemolytic mutant. We conclude that the 110 kDa hemolysin plays an important role in bacterial virulence and the pathogenesis of pleuropneumonia, and that sufficiently high levels of neutralizing antibodies to the 110 kDa hemolysin may be required for protection of pigs against disease.

Production of Apx toxins by field strains of Actinobacillus pleuropneumoniae and Actinobacillus suis

The three Apx toxins of Actinobacillus pleuropneumoniae have potential value for use in vaccines and diagnostic tests which will be species specific instead of serotype specific, provided that the Apx toxins are species specific and all field strains produce these toxins. We examined 114 A. pleuropneumoniae field strains and found that they secreted either ApxI, ApxII, ApxI and ApxII, or ApxII and ApxIII and secreted no other cytolytic activities. However, proteins similar to ApxI and ApxII were also produced by Actinobacillus suis.

Association between bacteriophage-infected Actinobacillus actinomycetemcomitans and rapid periodontal destruction

Actinobacillus actinomycetemcomitans was isolated from periodontal pockets in a patient suffering from prepubertal periodontitis. Electron microscopy revealed 3 different groups of bacteriophages in filtrates of subgingival plaque from all the active periodontal lesions. Phage infected A. actinomycetemcomitans in this patient was restricted to periodontal pockets which, according to standardized roentgenograms, had shown bone destruction during the past 12 months. A follow-up study of 7 months revealed that a "burned out" site which harbored noninfected A. actinomycetemcomitans, turned into an active site at the same time as the A. actinomycetemcomitans of that site became infected with the phages. These findings indicate a relationship between rapid prepubertal periodontal destruction and phage-infected A. actinomycetemcomitans.

Bacterial heat shock protein-60 increases epithelial cell proliferation through the ERK1/2 MAP kinases

Heat shock proteins (hsp) have important roles in the regulation and protection of both prokaryotic and eukaryotic cells, especially during environmental stress. Hsps are also important bacterial virulence factors. We investigated whether bacterial hsp60 can alter epithelial cell mitogen-activated protein kinase (MAPK) signaling and cell proliferation. Human skin keratinocytes (HaCaT cell line) were cultured in the presence of hsp60 purified from Actinobacillus actinomycetemcomitans, an important oral pathogen. Protein kinases in the ERK1/2 and p38 MAPK signaling pathways were probed with kinase-specific and phosphorylation-site-specific antibodies on Western blots. In quiescent cultures, hsp60 increased ERK1/2 phosphorylation in a sustained manner and p38 phosphorylation transiently. Hsp60 also increased epithelial cell proliferation by about 30%. Inhibition of the ERK1/2 pathway by PD 98059 (a MEK1 inhibitor) reversed partially ERK1/2 phosphorylation and totally cell proliferation indicating that the ERK1/2 MAPK pathway is involved in the hsp60-induced cell growth. This was supported by findings that hsp60 stimulated phosphorylation of RSK1/2 and cyclic AMP response element-binding protein and increased expression of transcription factors c-Jun and c-Fos. Recombinant human hsp60 did not alter ERK1/2 or p38 phosphorylation and had no effect on epithelial cell proliferation. Inhibition of p38 MAPK pathway by SB 203580 increased both ERK1/2 phosphorylation and cell proliferation demonstrating that the inhibitor can either directly or indirectly activate the ERK1/2 MAPK pathway. The results show that exogenous bacterial hsp60 is able to activate ERK1/2 phosphorylation and thereby cause increased epithelial proliferation. In case of mucosal infection this effect may either lead to increased wound repair or participate in the pathological mechanism of some bacterial diseases that involve increased epithelial proliferation.

The cytotoxin of Actinobacillus pleuropneumoniae (pleurotoxin) is distinct from the haemolysin and is associated with a 120 kDa polypeptide

Mutants of Actinobacillus pleuropneumoniae strain HK 361 (serotype 2) were isolated which were deficient in type II (Ca2(+)-dependent) haemolysin activity (Hly-). Some of the Hly- mutants secreted a potent, heat-labile extracellular cytotoxic activity against porcine alveolar macrophages. Comparison of cell-free culture supernatant from the parent strain and some Hly- mutants by SDS-PAGE and immunoblotting revealed the loss of a major extracellular polypeptide of 109 kDa. Two Hly- mutants which in addition failed to secrete a 120 kDa polypeptide produced no extracellular cytotoxic activity, suggesting that the 120 kDa protein was the cytotoxin. Antiserum raised to the culture supernatant from a Hly- mutant lacking the 109 kDa polypeptide recognized the 120 kDa band, but not the 109 kDa band, in immunoblots and neutralized the cytotoxic activity, but not the haemolytic activity, of A. pleuropneumoniae. The 120 kDa polypeptide and extracellular cytotoxic activity were widespread among A. pleuropneumoniae strains, but absent from related bacterial pathogens of the pig: Actinobacillus suis, Haemophilus parasuis and Pasteurella multocida. A clear correlation was found between the presence of the 120 kDa polypeptide and cytotoxic activity in culture supernatants. The cytotoxic activity of all the strains tested was neutralized by antibody to the Hly- extracellular material and by convalescent pig serum. It is proposed that the 120 kDa polypeptide represents the cytotoxin of A. pleuropneumoniae, that it is distinct from the haemolysin, and that it be termed pleurotoxin.