It has been over 125 years since Louis Pasteur first identified that a bacterium was the causative agent of fowl cholera. In seminal experiments, he also showed that repeated passage of the bacteria produced an attenuated strain incapable of causing disease, but the inoculation of birds with this strain could elicit a protective immune response (Pasteur, 1880, 1881). In this review, we aim to outline the progress that has been made in understanding this highly versatile pathogen. Indeed, many of the molecular determinants for virulence are now identified and understood and we can begin to piece together many of the steps required for successful infection.
The species Pasteurella multocida is subdivided into four subspecies that include the type strain, multocida, and three others, gallicida, septica and the recently described tigris (Capitini et al., 2002). This review will consider only the type strain, subspecies multocida. Pasteurella multocida can cause disease in a wide range of animal species and is the causative agent of numerous, economically important diseases, including avian fowl cholera, bovine haemorrhagic septicaemia, enzoonotic pneumonia and swine atrophic rhinitis (De Alwis, 1992). Pasteurella multocida strains are classified into serogroups (A, B, D, E and F) based on capsule antigens and further classified into 16 serotypes (1–16) based primarily on lipopolysaccharide antigens using the Heddleston scheme (Carter, 1955; Heddleston et al., 1972).
Fowl cholera is a serious disease of poultry and can present in either acute or chronic forms. The majority of acute fowl cholera cases are caused by serogroup A strains of P. multocida. Obvious clinical signs of acute fowl cholera may not occur until very late in the infection and include depression, ruffled feathers, fever, anorexia, mucous discharge from the mouth, diarrhoea and an increased respiratory rate (Rhoades & Rimler, 1989).
Haemorrhagic septicaemia, predominantly caused by serotypes B and E strains of P. multocida, is an acute disease that affects cattle and buffalo and is characterized by oedematous swelling of the head and neck, and swollen haemorrhagic lymph nodes (Carter & De Alwis, 1989). Atrophic rhinitis in pigs is caused by strains of P. multocida that express P. multocida toxin (PMT), and typically belong to serogroup D. In pigs, the most obvious symptoms include the shortening and twisting of the snout, dark tear staining and pneumonia (Chanter & Rutter, 1989). Atrophic rhinitis rarely causes death, but it is economically important as it significantly reduces the growth rate of the infected pigs.
Pasteurella multocida can be a primary or secondary agent involved in pneumonia in cattle (predominantly caused by serotype A:1), pigs and, occasionally, sheep (Gilmour, 1978; Chanter & Rutter, 1989; Frank, 1989). Pasteurella multocida can also infect rabbits resulting in rhinitis (‘snuffles’) and pneumonia (Manning et al., 1989). Although relatively uncommon, human infections have been observed in a range of sites, commonly following cat or dog bites (Weber et al., 1984).
Of all the diseases caused by P. multocida, fowl cholera is probably best understood. In birds, it is widely believed that P. multocida enters the host through tissues of the respiratory tract. Adhesion of P. multocida to turkey air sac macrophages has been demonstrated, and virulent P. multocida inoculated into the upper respiratory tract or trachea of turkeys can be subsequently detected in internal organs between 6 and 12 h postinoculation (Rhoades & Rimler, 1990; Matsumoto et al., 1991; Pruimboom et al., 1996, 1999). The mechanisms by which P. multocida invades from the lungs into the bloodstream are poorly understood, although a study using heterophil-depleted chickens experimentally infected with P. multocida showed that heterophils may play a dual role. Initially, recruitment of heterophils into the lungs led to tissue damage and invasion, but as the infection progressed, heterophils limited the infection by promoting clearance of the bacteria from the lungs and spleen (Bojesen et al., 2004).
Although it has been generally accepted that fowl cholera is a septicaemic infection, bacteria can only be isolated in large numbers from the blood of birds very late in infection, and it has been proposed that this late re-emergence of blood-borne bacteria is due to the rupture of liver and spleen phagocytes (Pabs-Garnon & Soltys, 1971). In the terminal stages of fowl cholera, death is probably caused by massive bacteraemia and endotoxic shock (Carter, 1967; Rhoades & Rimler, 1984).
The first systemic host defence used against most invading bacteria involves the innate immune system and includes phagocytosis and bactericidal properties of serum components such as complement. However, P. multocida has mechanisms to evade this system, including the presence of a capsule (Boyce & Adler, 2000; Chung et al., 2001). Active immunity against P. multocida infection is mainly humoral; vaccination of birds with killed P. multocida stimulates protection against homologous challenge, and opsonization studies in mice have shown that bactericidal antibodies are produced (Wijewardana & Sutherland, 1990).
Pasteurella multocida can be classified by serological methods into five capsule groups designated A, B, D, E and F. The composition and structure of the capsular material found in P. multocida serotypes A, D and F are very similar to mammalian glycosaminoglycans and consist mainly of hyaluronan, heparosan and unsulphated chondroitin, respectively (Pandit & Smith, 1993; Rimler, 1994; DeAngelis, 1996; DeAngelis & Padgett-McCue, 2000). The genes required for the synthesis and transport of these capsular types are encoded within a single region on the genome (Townsend et al., 2001). However, recently, in type A, D and F strains, an additional gene encoding a heparosan synthase was identified outside of the known capsule biosynthesis region. The synthase encoded by this gene, renamed hssB (formerly pgla), is transcribed 10-fold less than the synthase within the capsule operon, uses a different acceptor and gives rise to smaller molecular weight polymer products. It was proposed that the expression of this gene may give rise to capsular variation (Deangelis & White, 2004).
In general, strains that possess a capsule are more virulent than their acapsular variants (Heddleston et al., 1964; Snipes et al., 1987; Tsuji & Matsumoto, 1989). The important role of capsule in the pathogenesis of P. multocida has been clearly demonstrated as genetically defined, acapsular mutants constructed from both serogroup A and B strains were strongly attenuated in mice (Boyce & Adler, 2000; Chung et al., 2001). The serogroup A acapsular mutant was also shown to be avirulent in chickens and was unable to establish growth in chicken muscle (Chung et al., 2001). Interestingly, despite its apparent lack of persistence, vaccination of chickens with high doses of this acapsular mutant stimulated protective immunity (Chung et al., 2005).
It is widely accepted that capsule plays a significant role in resistance to phagocytosis, and this has been demonstrated in vitro by Harmon et al. (1991) and others, who have correlated sensitivity to phagocytosis with the presence and thickness of the bacterial capsule (Truscott & Hirsh, 1988; Harmon et al., 1991; Pruimboom et al., 1996). Moreover, studies using murine macrophages clearly demonstrated that a genetically defined acapsular serotype B mutant was more susceptible to uptake than its wild-type parent (Boyce & Adler, 2000). Resistance to complement-mediated lysis is clearly important for virulence, and experiments on P. multocida type A strains have shown that serum resistance correlates with the possession of a capsule (Snipes & Hirsh, 1986; Hansen & Hirsh, 1989). A genetically defined acapsular mutant was no longer serum resistant in normal avian serum compared with the serotype A wild-type parent and complemented mutant (Chung et al., 2001). In contrast, a genetically defined, acapsular mutant of a serotype B strain showed no reduction in serum resistance in either bovine or murine serum compared with the parent strain (Boyce & Adler, 2000). The basis for this difference between capsular types remains unknown. However, despite this evidence suggesting no loss of serum resistance, the serotype B acapsular mutant was strongly attenuated in mice and had increased sensitivity to phagocytosis by murine macrophages (Boyce & Adler, 2000).
Pasteurella multocida lipopolysaccharide plays a critical role in the pathogenesis of disease. It stimulates humoral immunity and is considered to be a protective antigen. Monoclonal antibodies raised against the lipopolysaccharide from a serotype A strain were bactericidal and protected mice against homologous challenge (Wijewardana et al., 1990). In addition, an opsonic monoclonal antibody against lipopolysaccharide from a serotype B strain of P. multocida partially protected mice against P. multocida infection (Ramdani & Adler, 1991).
There are conflicting reports as to the endotoxic properties of lipopolysaccharide isolated from P. multocida. Intravenous inoculation of lipopolysaccharide from serotype B:2 stains could reproduce clinical signs of haemorrhagic septicaemia in buffalo (Horadagoda et al., 2002), but the endotoxic properties of lipopolysaccharide from serotype A strains is less clear. Studies have indicated that chicken embryos and mice are highly susceptible, but that turkey poults are relatively resistant (Ganfield et al., 1976; Rhoades & Rimler, 1987; Mendes et al., 1994).
It is clear that P. multocida requires a complete lipopolysaccharide structure in order to replicate in vivo and cause disease. In a recent study using signature-tagged mutagenesis, a mutant attenuated in mice and chickens had an insertional inactivation of the gene waaQPM (encoding a putative heptosyl transferase) required for the addition of heptose to lipopolysaccharide (Harper et al., 2004). Using mass spectrometry and nuclear magnetic resonance, it was demonstrated that the predominant lipopolysaccharide glycoforms extracted from this mutant were severely truncated. Complementation experiments demonstrated that providing a functional waaQPM gene in trans restored both the lipopolysaccharide structure and growth in mice to wild-type levels (Harper et al., 2004).
A galE mutant of P. multocida had significantly reduced viability in mice (Fernandez de Henestrosa et al., 1997). Included in the role of galE in bacteria is the epimerization of UDP-glucose to UDP-galactose before lipopolysaccharide assembly, and this mutant probably expressed an altered lipopolysaccharide, although the structural analysis of the lipopolysaccharide was not reported (Fernandez de Henestrosa et al., 1997).
The complete lipopolysaccharide structure of strains VP161 and X73 belonging to Heddleston serotype 1 and PM70, a Heddleston serotype 3 strain, have been determined and all possess a structure similar to the lipopolysaccharide or lipo-oligosaccharide (LOS) of Neisseria spp. and Haemophilus spp. The lipopolysaccharide from the P. multocida strains had highly conserved inner cores with a tri-heptose unit linked via a Kdo residue to lipid A, but there were significant variations in the outer core between the three different strains (St Michael et al., 2005a, b, c). Interestingly P. multocida lipopolysaccharide isolated from the two Heddleston type I strains contains terminal phosphocholine residues, which in other bacteria plays a key role in bacterial adhesion to, and invasion of, host cells by binding directly to the platelet-activating factor (PAF) receptor (Cundell et al., 1995; Schenkein et al., 2000; Swords et al., 2000; Serino & Virji, 2002). Although the role of phosphocholine residues on the lipopolysaccharide of P. multocida remains unknown, it has been shown in the bovine model that lipopolysaccharide from P. multocida assists in adhesion to neutrophils and transmigration through endothelial cells (Galdiero et al., 2000).
In addition to phosphocholine residues, all the P. multocida strains studied to date have phosphoethaolamine residues attached to various sites on their lipopolysaccharide. Notably, in addition to phosphocholine residues, the X73 strain has phosphoethaolamine residues attached to each of the terminal galactose sugars (St Michael et al., 2005a). In other Gram-negative bacteria, phosphoethaolamine is added to a number of sites within the inner core of lipopolysaccharide by specific transferases, and in Neisseria meningitidis the expression and position of these residues affects the ability of the bacteria to resist the innate immune response (Ram et al., 2003). A clear role for phosphoethaolamine in P. multocida has not been defined.
Fimbriae and adhesins
There are many genes, including ptfA, fimA, flp1, flp2, hsf_1 and hsf_2 on the P. multocida genome that encode proteins similar to fimbriae or fibrils in other bacteria.
It is likely that fimbriae play a role in the surface adhesion, as fimbriae have been observed on some P. multocida serotype A strains that were able to adhere to mucosal epithelium, but not on the surface of those strains unable to adhere (Glorioso et al., 1982; Rebers et al., 1988; Isaacson & Trigo, 1995; Ruffolo et al., 1997). Type IV fimbriae (pili) have been isolated and characterized from P. multocida serotypes A, B and D (Ruffolo et al., 1997) and are often associated with virulence in other bacteria because of their role in attachment to host cell surfaces. The subunit gene, ptfA, has been isolated and sequenced from a number of strains and the predicted protein sequences showed significant variation between strains (Doughty et al., 2000). However, the role of fimbrial structures in P. multocida virulence is still unproven.
The P. multocida Pm70 genome contains a region encoding proteins with significant similarity to the Flp pilin locus in Actinobacillus actinomycetemcomitans, encoding a recently described type IV fimbrial subfamily (Kachlany et al., 2001b; May et al., 2001). This locus encodes proteins predicted to be the Flp pilin subunits and the proteins required for the pilin assembly. Despite the lack of physical evidence for Flp pili in P. multocida, there is evidence through STM studies in mice that the products of these genes are required for virulence. Two genes, flp1, encoding a Flp pilin subunit and tadD, predicted to encode a component of the secretion apparatus required for Flp pilin assembly, have been inactivated in two independent STM studies using transposon mutagenesis, and both mutants were significantly attenuated in mice (Fuller et al., 2000; May et al., 2001; Kachlany et al., 2001a, b; Harper et al., 2003).
Two P. multocida genes, pfhaB1 and pfhaB2, share significant similarity with a class of genes that encode filamentous haemagglutinins, which in Bordetella pertussis play a major role in the colonization of the upper respiratory tract (Kimura et al., 1990; Mooi et al., 1992). Mutation of these genes in P. multocida resulted in significantly reduced virulence in mice (Fuller et al., 2000). More recently, a pfhaB2 mutant was constructed in a fowl cholera strain, P1059, and shown to be highly attenuated in turkeys when administered intranasally, but only moderately attenuated when given intravenously, suggesting that pfhaB2 has a significant role in initial colonization or invasion (Tatum et al., 2005).
In general, most P. multocida strains that cause fowl cholera, haemorrhagic septicaemia or pneumonia are not known to express any toxins. The dermonecrotic toxin, PMT, expressed mainly by serogroup D strains, is the only toxin identified to date and is responsible for the clinical and pathological signs of atrophic rhinitis (Foged et al., 1987; Rimler & Rhoades, 1989). Both purified native and recombinant PMT toxin can be used to experimentally induce clinical signs of disease (Foged et al., 1987; Lax & Chanter, 1990).
PMT acts intracellularly by modulating the Gαq subunit in the phospholipase C signal transduction pathway (Murphy & Rozengurt, 1992; Wilson et al., 1997). In pigs, this leads to atrophy of nasal turbinates where bone resorption occurs due to uncontrolled proliferation of osteoclasts, and regeneration of bone is prevented by the inhibition of osteoblasts (Sterner-Kock et al., 1995; Mullan & Lax, 1998). PMT is also a potent mitogen, inducing many cellular effects including rearrangements in the actin cytoskeleton (Zywietz et al., 2001). Like cholera and pertussis toxins, PMT activates dendritic cells to mature cells but, unlike the other toxins, it is a poor adjuvant and appears to suppress the antibody response (Bagley et al., 2005). It has been proposed that PMT blocks chemotaxis-induced migration of dendritic cells to regional lymph nodes and might, therefore, in a natural infection, limit the development of an adaptive immune response (Blocker et al., 2006).
The PMT toxin gene resides on a lysogenic bacteriophage belonging to the Siphoviridae family. As the PMT toxin has no signal sequence and no known mechanism of export, it has been suggested that the lytic phase of the bacteriophage may mediate the release of the toxin (Pullinger et al., 2004). The toxin is an effective immunogen; a toxoid developed by deletion mutagenesis of the cloned PMT toxin gene was found to protect mice and their offspring against challenge with purified PMT (Petersen et al., 1991). Similarly, a genetically modified PMT toxin, where there were two key amino acid substitutions, led to a nontoxigenic protein that protected pigs against experimental challenge with the wild-type strain (To et al., 2005).
Iron regulated and iron acquisition proteins
Iron is an essential element which must be acquired by bacteria in order to survive. Because of its inherent toxicity, the level of free iron available in vivo is very limited and P. multocida, like other bacterial species, has developed multiple mechanisms for iron uptake. Sequence analysis of P. multocida PM70 revealed that a relatively large proportion of the genome (over 2.5%) encodes 53 proteins with similarity to proteins involved in iron uptake or acquisition (May et al., 2001).
Comparisons of P. multocida grown in iron-rich, iron-depleted media or in vivo has demonstrated that many high molecular weight outer membrane proteins are regulated by iron levels and have therefore been called iron-regulated outer membrane proteins (IROMPs) (Snipes et al., 1988; Choi-Kim et al., 1991). Pasteurella multocida grown under iron-limited conditions also induces a stronger protective response in mice compared with the same strain grown under iron-replete conditions (Kennett et al., 1993), and it is thought that IROMPs may, therefore, play a significant role in cross-protective immunity (Glisson et al., 1993; Ruffolo et al., 1998). However, analysis of IROMPs using a proteomics approach identified only one protein, PM0805, that was up-regulated and only one, OmpW, that was down-regulated under low-iron conditions (Boyce et al., 2006).
Transferrin receptors utilized by bacterial species in the Pasteurellaceae and Neisseriaceae families usually consist of two iron binding receptors TbpA and TbpB (Gray-Owen & Schryvers, 1996), but recent evidence suggests that the transferrin receptor in bovine strains of P. multocida is composed of only a single protein TbpA (Ogunnariwo & Schryvers, 2001). However, this receptor may not be present in all P. multocida strains; a recent PCR and DNA hybridization study found that the tbpA gene is present only in some bovine and ovine clinical isolates (Ewers et al., 2006). Iron acquisition proteins are believed to play a role in the disease process; serogroup A mutants with inactivated ExbB, ExbD, TonB, or HgbA proteins are attenuated in mice (Fuller et al., 2000; Bosch et al., 2002a). ExbB, ExbD and TonB are part of the TonB transport complex, required to transport and provide energy for iron sequestration, and HgbA is predicted to be a haemoglobin-binding protein required for the acquisition of iron from host proteins (Bosch et al., 2002b).
Sialic acid metabolism
Sialidases are produced by some bacterial species and act to remove sialic acid from host glycosylated proteins and lipids for use as a carbon source. These enzymes may also enhance bacterial virulence by unmasking key host receptors and/or reducing the effectiveness of host defences such as mucin. Most P. multocida strains produce sialidase and both cell bound and extracellular sialidases have been reported in P. multocida (Scharmann et al., 1970; Drzeniek et al., 1972; White et al., 1995). Bacterial sialidase is produced in vivo in goats after transthoracic challenge with either P. multocida or M. haemolytica (Straus & Purdy, 1994; Straus et al., 1996). Passive protection of mice against P. multocida A:3 homologous challenge has been demonstrated using rabbit antisera raised against partially purified sialidase, although there was some difficulty in removing anti-LPS antibodies from the serum before administration into mice (Ifeanyi & Bailie, 1992), thus raising the question about the specificity of the protective antibodies.
Two sialidases, NanH and NanB, have been cloned and characterized from a fowl cholera isolate of P. multocida (Mizan et al., 2000). These sialidases differed in their specificity, with both able to utilize 2,3′ sialyl lactose, but only NanB able to fully utilize 2,6′ sialyl lactose. It was proposed that the presence of two sialidases with slightly different specificities would enhance the metabolic capacity of P. multocida in the host (Mizan et al., 2000).
In addition, there is increasing evidence to suggest that P. multocida is capable of scavenging sialic acid from the environment for both the sialylation of cell components and for nutrients via a catabolic breakdown pathway (Steenbergen et al., 2005). Furthermore, the uptake, but not catabolism, of sialic acid was shown to be essential for virulence in mice (Steenbergen et al., 2005). Two genes, pm0188 and pm0508, encoding sialic acid transferases, have been identified in the P. multocida genome. Interestingly, the product of pm0188 has been shown to be a multifunctional enzyme capable of four functions. It exhibits transferase activity linking sialic acid to galactose with either 2,3 or 2,6 linkages with varying activity. In lower, more acidic pH conditions, it can function both as a sialidase capable of cleaving 2,3 sialyl linkages, but not 2,6 linkages, and as a trans sialidase, transferring 2,3 linked sialyl groups to another galactose residue (Yu et al., 2005). An unequivocal role in pathogenesis has not yet been demonstrated.
Although the role of hyaluronidase in pathogenesis has not been determined, it is present in many of the serotype B strains of P. multocida that cause bovine haemorrhagic septicaemia. A study of 74 P. multocida strains representing all capsular serotypes found that only the type B strains, isolated from haemorrhagic septicaemia infections, produced hyaluronidase (Carter & Chengappa, 1980). Another study of 176 strains of P. multocida representing different serotypes also found hyaluronidase activity confined to serotype B, but more specifically B:2, and it was suggested that a test for hyaluronidase activity could be used to presumptively identify B:2 strains (Rimler & Rhoades, 1994).
Outer membrane proteins
Early studies on the P. multocida outer membrane showed that a 37 kDa protein was among five identified as possible protective immunogens based firstly on radioimmunoprecipitation results using protective immune rabbit sera, and secondly on their location in the outer membrane (Lu et al., 1988). Monoclonal antibodies raised against the 37 kDa protein were able to passively protect rabbits and mice against infection with P. multocida with strong protection afforded against homologous strains, and some limited protection against heterologous strains (Lu et al., 1991).
A protein of similar molecular mass (39 kDa) was identified in the P. multocida A:3 strain P1059; its expression was shown to correlate with the presence and amount of capsule present on the cell (Borrathybay et al., 2003b; Ali et al., 2004a). Pasteurella multocida can adhere to, and invade, chicken embryo fibroblasts, and this adherence was inhibited by both monoclonal and polyclonal antibodies raised against the 39 kDa protein (Borrathybay et al., 2003a; Al-haj Ali et al., 2004; Ali et al., 2004a, b). Passive immunization of mice with a monoclonal antibody against the 39 kDa protein or active immunization with affinity purified 39 kDa protein, demonstrated that antibodies raised against this protein were cross-protective against serovars A:1 and A:3 (Ali et al., 2004a, b). The actual identity of this 39 kDa protein was not reported, but recently a 39 kDa protein which can stimulate cross-serotype protection (Rimler, 2001) was isolated from outer membrane protein extracts of the same A:3 strain, P1059, used in the above studies. This protein was identified as PlpB (Pasteurella lipoprotein B), using peptide mass fingerprinting (Tabatabai & Zehr, 2004) and is predicted to be an ABC transport protein required for the uptake of methionine into the cell (Merlin et al., 2002).
One of the major outer membrane proteins of P. multocida is OmpH. Antibodies raised against this protein provide some protection against disease. Monoclonal antibodies specific for OmpH passively protect mice against P. multocida challenge (Marandi & Mittal, 1997) and vaccination with the native, but not recombinant, OmpH protein elicits protective immunity in birds against homologous challenge (Luo et al., 1997). In addition, antibodies raised to an OmpH synthetic peptide, Cyclic-L2, provided partial protection in chickens against homologous challenge (Luo et al., 1999).
Recent studies examining the ability of P. multocida to bind host extracellular matrix proteins have shown that the bacteria can adhere to fibronectin and collagen type IX. Proteins identified as possible adhesins include OmpA, Oma87, Pm1069 and the iron related proteins, Tbp (transferrin binding protein) and the putative TonB receptor HgbA (Dabo et al., 2005).
OmpA, a β-barrel, ion channel protein, has been specifically identified as having a direct role in adhesion. Homologs of this protein are important adhesins in Escherichia coli, Haemophilus influenzae and other bacteria. Notably, recombinant P. multocida OmpA binds to bovine kidney cells and interacts with host extracellular matrix molecules heparin and fibronectin (Dabo et al., 2003).
Pasteurella multocida is capable of causing disease in a wide range of animals and birds. There is a clear correlation between capsular type and the disease, with serotypes A and F typically associated with fowl cholera, serotypes B and E with haemorrhagic septicaemia in cattle, and serotype D strains expressing PMT toxin with atrophic rhinitis in swine. What is unclear is whether the expression of a particular type of lipopolysaccharide or the type and amount of proteins presented on the bacterial surface also contribute to the disease specificity. Other virulence factors such as neuraminidases, iron sequestering proteins and metabolic enzymes play key roles in acquiring and utilizing substrates for growth within the host, often a relatively nutrient poor and hostile environment.
There has been little progress in understanding exactly how P. multocida invades mucosal surfaces to gain access to the blood and how the host responds to the infection. However, significant advances have been made in identifying bacterial factors critical to P. multocida pathogenesis. It is known that capsule and lipopolysaccharide are essential for normal disease progression in fowl cholera and there is some progress in understanding the bacterial response to the in vivo environment (Boyce et al., 2002, 2004) and identifying bacterial factors required for disease progression (Fuller et al., 2000, Harper et al., 2003).