Introduction
Actinobacillus pleuropneumoniae is the major cause of porcine pleuropneumonia, a highly contagious respiratory disease responsible for major economic losses in the swine industry (Taylor, Reference Taylor, Straw, D'allaire, Megeling and Taylor1999). The disease is characterized by hemorrhagic, fibrinous and necrotic lung lesions and the clinical features range from acute to chronic. Pigs surviving the disease often suffer from reduced growth rates and frequently become asymptomatic carriers of the pathogen (Moller et al., Reference Moller, Andersen, Christensen and Kilian1993; Sidibe et al., Reference Sidibe, Messier, Lariviere, Gottschalk and Mittal1993) and are the main cause of bacterial dissemination (Taylor, Reference Taylor, Straw, D'allaire, Megeling and Taylor1999). To date, 15 serotypes of A. pleuropneumoniae based on capsular antigens have been described (Dubreuil et al., Reference Dubreuil, Jacques, Mittal and Gottschalk2000; Blackall et al., Reference Blackall, Klaasen, Van Den Bosch, Kuhnert and Frey2002); all serotypes are capable of causing disease, although differences in virulence have been described (Frey, Reference Frey1995b; Jacobsen et al., Reference Jacobsen, Nielsen and Nielsen1996). Several virulence factors are known for A. pleuropneumoniae, such as the Apx toxins (Frey, Reference Frey and Donachie1995a), the lipopolysaccharides (LPSs) (Jacques, Reference Jacques1996; Ramjeet et al., Reference Ramjeet, Deslandes, St Michael, Cox, Kobisch, Gottschalk and Jacques2005), the capsule and various outer membrane proteins (OMPs) (Haesebrouck et al., Reference Haesebrouck, Chiers, Van Overbeke and Ducatelle1997; Jacques, Reference Jacques2004).
The economic importance of this disease in the swine industry has stimulated intensive research in the past years in the A. pleuropneumoniae vaccination field. Many studies have been reported and several vaccines have been commercialized but complete satisfaction has not been obtained in the protection of pigs against A. pleuropneumoniae infection (Backstrom, Reference Backstrom1999; Haesebrouck et al., Reference Haesebrouck, Pasmans, Chiers, Maes, Ducatelle and Decostere2004). The purpose of this review is to summarize and present current knowledge on the achievements realized in vaccination against A. pleuropneumoniae. We will focus our interest on the evolution of A. pleuropneumoniae vaccines from the first commercialized whole-cell bacterins to more promising ones such as subunit vaccines and live attenuated vaccines. We will also discuss and provide more information on the importance of the administration route, vaccine formulation and adjuvants in the stimulation of specific immune responses in order to provide good protection. Finally, we will highlight new promising strategies and new developments in the A. pleuropneumoniae vaccination field.
The limits of inactivated whole-cell bacterial vaccines
The so-called ‘first-generation’ vaccines of whole-cell bacterins were the first commercialized vaccines against A. pleuropneumoniae infection and consisted of heat-killed bacteria or formalin-treated whole-cells. Inactivated whole-cell vaccines have the advantage of presenting a complex array of antigenic determinants to the immune system without any concern for reversion issues raised by live attenuated vaccines. In order to enhance the expression of immunogenic and protective antigens, bacteria can be grown in specific conditions mimicking the host environment, prior to bacterin preparation. Studies have shown that bacterins obtained from A. pleuropneumoniae serotype 10 grown in NAD-restricted conditions induced a better protection upon challenge (Van Overbeke et al., Reference Van Overbeke, Chiers, Donne, Ducatelle and Haesebrouck2003). However, the use of whole-cell bacterins as an A. pleuropneumoniae vaccine is limited as previous immunization and challenge experiments only showed partial protection with a slight reduction in mortality (Jolie et al., Reference Jolie, Mulks and Thacker1995; Furesz et al., Reference Furesz, Mallard, Bosse, Rosendal, Wilkie and Macinnes1997). The absence of secreted proteins such as the Apx toxins which are known to be highly immunogenic and essential for protection, might explain the limited protection observed with bacterins (Haga et al., Reference Haga, Ogino, Ohashi, Ajito, Hashimoto and Sawada1997; Seah et al., Reference Seah, Frey and Kwang2002). The alteration of antigenic characters of certain bacteria-associated virulence factors by heat, irradiation or chemical treatments during bacterin preparation might also affect the efficacy of the vaccine (Haesebrouck et al., Reference Haesebrouck, Chiers, Van Overbeke and Ducatelle1997). Moreover, bacterins offer limited cross-protection (Jolie et al., Reference Jolie, Mulks and Thacker1995) and do not prevent initial infection and colonization, which facilitates the emergence of healthy carriers. Indeed, one major problem encountered in using bacterins as vaccines is that they confer only partial protection against the homologous serotype and generally do not confer protection against challenge with heterologous serotypes (Higgins et al., Reference Higgins, Lariviere, Mittal, Martineau, Rousseau and Cameron1985; Thacker and Mulks, Reference Thacker and Mulks1988; Fenwick and Henry, Reference Fenwick and Henry1994). The low efficacy of bacterins might also be related to the spectrum of immune responses induced, usually limited to humoral response (Furesz et al., Reference Furesz, Mallard, Bosse, Rosendal, Wilkie and Macinnes1997), and the blood lymphocyte subset phenotypes displayed (Appleyard et al., Reference Appleyard, Furesz and Wilkie2002), which do not reflect natural infection. In fact, whole inactivated bacteria display no colonization of the respiratory tract, which is important for an effective immune stimulation.
New developments in inactivated whole-cell bacterial vaccines have shown a promising strategy in A. pleuropneumoniae vaccination in terms of antigen immunogenicity. Genetically-inactivated ghost vaccines are empty whole cell envelopes produced by controlled expression of bacteriophage PhiX174 lysis gene E (Witte et al., Reference Witte, Wanner, Blasi, Halfmann, Szostak and Lubitz1990, Reference Witte, Wanner, Sulzner and Lubitz1992). Expression of this gene from a plasmid in Gram-negative bacteria leads to the formation of a protein E specific tunnel which subsequently results in the outflow of cytoplasmic contents without any physical or chemical denaturation of the bacterial surface structures (Witte et al., Reference Witte, Wanner, Blasi, Halfmann, Szostak and Lubitz1990, Reference Witte, Wanner, Sulzner and Lubitz1992). Thus, bacterial ghosts have the advantage over bacterins of sharing functional and antigenic determinants with their living counterparts. Moreover, the activating potential of bacterial ghosts in the maturation and stimulation of immune cells has also been brought to light (Felnerova et al., Reference Felnerova, Kudela, Bizik, Haslberger, Hensel, Saalmuller and Lubitz2004). The use of this technology could offer some promising perspectives in vaccination, as recombinant ghost bacteria can be effectively used to enhance expression and delivery of antigens (Szostak et al., Reference Szostak, Hensel, Eko, Klein, Auer, Mader, Haslberger, Bunka, Wanner and Lubitz1996) in order to target a specific local immune response (Lubitz et al., Reference Lubitz, Witte, Eko, Kamal, Jechlinger, Brand, Marchart, Haidinger, Huter, Felnerova, Stralis-Alves, Lechleitner, Melzer, Szostak, Resch, Mader, Kuen, Mayr, Mayrhofer, Geretschlager, Haslberger and Hensel1999; Lubitz, Reference Lubitz2001; Jalava et al., Reference Jalava, Eko, Riedmann and Lubitz2003; Riedmann et al., Reference Riedmann, Kyd, Cripps and Lubitz2007). Studies have shown that immunization with A. pleuropneumoniae bacterial ghosts is more effective than bacterin vaccination in protecting pigs against lung colonization and infection, and could therefore prevent development of healthy carriers (Katinger et al., Reference Katinger, Lubitz, Szostak, Stadler, Klein, Indra, Huter and Hensel1999; Hensel et al., Reference Hensel, Huter, Katinger, Raza, Strnistschie, Roesler, Brand and Lubitz2000). Moreover, a cross-protective potential in those ghost vaccines has also been suggested (Huter et al., Reference Huter, Hensel, Brand and Lubitz2000). Despite the partial protection observed with bacterins and the encouraging preliminary trials with the bacterial ghost system, the use of inactivated whole-cell bacteria as vaccines is still compromised by the fact that one main concern in A. pleuropneumoniae vaccination is cross-protection. It has been shown that a pig that survives natural or experimental infections is immunized against all serotypes of A. pleuropneumoniae (Nielsen, Reference Nielsen1984). These observations suggest the presence of highly immunogenic bacterial antigens common to all serotypes which are expressed only within the host. In this context, neither bacterins nor bacterial ghosts seem to be suitable for effective protection as the main problem associated with the use of inactivated whole-cell bacteria is the in vivo environment expression which cannot be completely reproduced in vitro (Goethe et al., Reference Goethe, Gonzales, Lindner and Gerlach2000; Van Overbeke et al., Reference Van Overbeke, Chiers, Donne, Ducatelle and Haesebrouck2003). Consequently, recent research in the A. pleuropneumoniae vaccination field has mainly focused on finding antigens highly conserved among all serotypes which could be purified and used as potential subunit vaccines, and also in the development of live attenuated mutants in order to overcome the problem of failure of cross-protection.
Virulence factors of A. pleuropneumoniae and subunit vaccine candidates
Many virulence factors of A. pleuropneumoniae have been investigated for their protective potential (Table 1). In order to find candidates for the development of subunit vaccines, studies had first targeted the most accessible structures of the bacteria. Hence, components of the bacterial surface such as the capsule, LPS and several OMPs were first identified as potential vaccine candidates. An anionic fraction of a saline extract of A. pleuropneumoniae serotype 1 (ANEX) that contained polysaccharide, LPS and protein antigens showed protective immunity in pigs when combined with an effective adjuvant (Willson et al., Reference Willson, Rossi-Campos and Potter1995). However, the major difficulties encountered with the capsule and LPS are their high heterogeneity among the serotypes (Perry et al., Reference Perry, Altman, Brisson, Beynon and Richards1990; Dubreuil et al., Reference Dubreuil, Jacques, Mittal and Gottschalk2000). Thus, vaccination with these bacterial components failed to confer good protection against heterologous serotypes. Pigs and mice immunized with LPS were previously found to be partially protected upon homologous challenge with A. pleuropneumoniae serotype 1 (Rioux et al., Reference Rioux, Dubreuil, Begin, Laferriere, Martin and Jacques1997, Reference Rioux, Girard, Dubreuil and Jacques1998) while cross-serotype challenge experiments in mice vaccinated with LPS showed no protection (Rioux et al., Reference Rioux, Dubreuil, Begin, Laferriere, Martin and Jacques1997). Passive immunization of mice with monoclonal antibodies directed against LPS also failed to provide protection against the heterologous serotype of A. pleuropneumoniae (Saze et al., Reference Saze, Kinoshita, Shiba, Haga, Sudo and Hashimoto1994). Moreover, pigs immunized with purified LPS or capsule were not protected against challenge with the homologous A. pleuropneumoniae serotype 5 (Inzana et al., Reference Inzana, Ma, Workman, Gogolewski and Anderson1988), and other studies also showed that neither the capsule nor the LPS seemed to be directly correlated with protection of mice in A. pleuropneumoniae challenge experiments (Byrd and Hooke, Reference Byrd and Hooke1997). Consequently, research on subunit vaccines has mostly focused on finding conserved antigens such as OMPs and lipoproteins.
Table 1. Subunit vaccines against A. pleuropneumoniae infection

Endobr=endobronchial, ID=intradermal, IM=intramuscular, IN=intranasal, IP=intraperitoneal, SC=subcutaneous.
Although OMP profiles differ for most serotypes of A. pleuropneumoniae (Rapp et al., Reference Rapp, Munson and Ross1986), a few OMPs were characterized at a molecular level and found to be present in almost all A. pleuropneumoniae serotypes. These include the transferrin-binding protein TfbA (or TbpB) (Gonzalez et al., Reference Gonzalez, Caamano and Schryvers1990; Gerlach et al., Reference Gerlach, Klashinsky, Anderson, Potter and Willson1992b), a 42-kDa maltose-inducible protein (Deneer and Potter, Reference Deneer and Potter1989), the 14-kDa peptidoglycan-associated lipoprotein PalA (Frey et al., Reference Frey, Kuhnert, Villiger and Nicolet1996) and the 50-kDa lipoprotein OmlA (Gerlach et al., Reference Gerlach, Anderson, Klashinsky, Rossi-Campos, Potter and Willson1993). Many low molecular-mass OMPs of A. pleuropneumoniae were also detected using Surface Enhanced Laser Desorption Ionisation (SELDI) – ProteinChip™ technology. In fact, SELDI was shown to be a useful complementary approach to conventional proteomic analytical methods with A. pleuropneumoniae, particularly suitable for analysis of proteins in the <20-kDa mass range (Hodgetts et al., Reference Hodgetts, Bosse, Kroll and Langford2004). Among the bacterial surface components, lipoproteins are also known to be highly immunogenic and protective. A novel method using a mild detergent treatment was developed to enhance the release of immunogenic lipoproteins from the outer membrane in culture supernatant without bacterial lysis. The main advantage of this extraction method is that the resulting cell-free supernatant (CFS) can then be used as a non-recombinant subunit vaccine (Goethe et al., Reference Goethe, Gonzales, Lindner and Gerlach2000). Thus, a subunit vaccine based on detergent-prepared CFS from A. pleuropneumoniae serotypes 1, 2 and 5 grown under iron-restricted conditions showed good protective activity and cross-protection between serotypes 2 and 9 (Maas et al., Reference Maas, Meens, Baltes, Hennig-Pauka and Gerlach2006b). However, experiments to assess the potential capacity of OMPs and lipoproteins to induce protective immunity were mostly restricted to immunoblot analysis with convalescent sera, while many other immunogenic OMPs were only identified by their molecular mass without any further characterization (Cruz et al., Reference Cruz, Nedialkov, Thacker and Mulks1996). For example, the outer membrane lipoprotein PalA which was previously identified as a potential vaccine candidate based on its reactivity with pig immune sera (Frey et al., Reference Frey, Kuhnert, Villiger and Nicolet1996), was later found to have a negative effect on protective immunity against A. pleuropneumoniae in vaccinated pigs (Van Den Bosch and Frey, Reference Van Den Bosch and Frey2003).
The iron acquisition systems of A. pleuropneumoniae include several important uptake systems such as uptake of transferrin, hemoglobin, and ferrichrome, a hydroxamate siderophore (Jacques, Reference Jacques2004). Not only is iron essential for survival of the bacteria but iron-restriction is also an important signal controlling the expression of many genes including some coding for virulence factors (Deslandes et al., Reference Deslandes, Nash, Harel, Coulton and Jacques2007). Proteins involved in iron uptake are therefore potential candidates for the development of subunit vaccines and were investigated for their protective capacities. Three different transferrin-binding proteins B (TbpB) of 60, 62 and 65 kDa were identified among all A. pleuropneumoniae serotypes. Immunization of pigs with the 60 kDa Tbp conferred limited protection against challenge with the homologous strain (Gerlach et al., Reference Gerlach, Anderson, Potter, Klashinsky and Willson1992a; Rossi-Campos et al., Reference Rossi-Campos, Anderson, Gerlach, Klashinsky, Potter and Willson1992). An acellular pentavalent subunit vaccine (Pleurostar™ Novartis) was prepared with recombinant antigens from A. pleuropneumoniae and contains the transferrin-binding protein B of A. pleuropneumoniae serotype 7. This vaccine showed partial protection against severe challenge with A. pleuropneumoniae serotype 9 (Van Overbeke et al., Reference Van Overbeke, Chiers, Ducatelle and Haesebrouck2001). FhuA and HgbA, receptors for ferrichrome and hemoglobin, respectively, were also shown to be conserved among all serotypes and biotypes of A. pleuropneumoniae (Mikael et al., Reference Mikael, Pawelek, Labrie, Sirois, Coulton and Jacques2002, Reference Mikael, Srikumar, Coulton and Jacques2003; Srikumar et al., Reference Srikumar, Mikael, Pawelek, Khamessan, Gibbs, Jacques and Coulton2004; Shakarji et al., Reference Shakarji, Mikael, Srikumar, Kobisch, Coulton and Jacques2006). Pig infection experiments have highlighted the role of HgbA as an important virulence factor which could be of interest as a potential subunit vaccine (Shakarji et al., Reference Shakarji, Mikael, Srikumar, Kobisch, Coulton and Jacques2006).
Apx toxins are secreted toxins, members of the RTX toxins family. They represent major virulence factors of A. pleuropneumoniae and are known to be strongly immunogenic. The importance of Apx toxins in protective immunity against porcine pleuropneumonia was demonstrated in many studies (Inzana et al., Reference Inzana, Todd, Ma and Veit1991). It has been shown that neutralizing antibodies directed against Apx toxins protected neutrophils from being killed and consequently allowed them to efficiently clear the ingested bacteria (Cruijsen et al., Reference Cruijsen, Van Leengoed, Dekker-Nooren, Schoevers and Verheijden1992; Jansen, Reference Jansen1994). Protection of vaccinated pigs against an aerosol challenge with A. pleuropneumoniae serotype 1 has been shown to be correlated with the presence of IgG1 subclass anti-hemolysin (Furesz et al., Reference Furesz, Wilkie, Mallard, Rosendal and Macinnes1998). A hemolysin vaccine made of purified ApxI and ApxII showed good protective activity in pigs against A. pleuropneumoniae serotype 1 (Haga et al., Reference Haga, Ogino, Ohashi, Ajito, Hashimoto and Sawada1997) while the N-terminal fragment of ApxI was shown to elicit good protection in mice against various serotypes of A. pleuropneumoniae (Seah et al., Reference Seah, Frey and Kwang2002). N- and C-terminal domains as well as the activation domain of the RTX toxin ApxIII also displayed potential for further vaccination trials as pig antisera raised against those fragments expressed cytotoxin-neutralizing activities (Seah and Kwang, Reference Seah and Kwang2004). Immunization experiments with Apx toxins in combination with other bacterial compounds all showed that Apx toxins were essential vaccine components to confer protection against bacterial challenge (Byrd and Kadis, Reference Byrd and Kadis1992; Van Den Bosch et al., Reference Van Den Bosch, Jongenelen, Pubben, Van Vugt and Segers1992; Beaudet et al., Reference Beaudet, Mcsween, Boulay, Rousseau, Bisaillon, Descoteaux and Ruppanner1994; Jansen, Reference Jansen1994; Frey, Reference Frey and Donachie1995a; Madsen et al., Reference Madsen, Carnahan and Thwaits1995). Thus far, almost all commercially available A. pleuropneumoniae subunit vaccines known as ‘second-generation’ vaccines contain Apx toxins (Chiers et al., Reference Chiers, Van Overbeke, De Laender, Ducatelle, Carel and Haesebrouck1998; Van Overbeke et al., Reference Van Overbeke, Chiers, Ducatelle and Haesebrouck2001; Habrun et al., Reference Habrun, Bilic, Cvetnic, Humski and Benic2002; Van Den Bosch and Frey, Reference Van Den Bosch and Frey2003; Tumamao et al., Reference Tumamao, Bowles, Van Den Bosch, Klaasen, Fenwick, Storie and Blackall2004; Meeusen et al., Reference Meeusen, Walker, Peters, Pastoret and Jungersen2007).
Generally, traditional vaccine extracts are enriched for secreted or surface-exposed bacterial components, as shown above. However, internal proteins that are involved in cellular metabolism are also reported to induce a protective immunity in other systems despite their predicted periplasmic and cytoplasmic localizations (Mosier et al., Reference Mosier, Iandolo, Rogers, Uhlich and Crupper1998; Thomas et al., Reference Thomas, Dunkley, Moore, Reynolds, Bastin, Kyd and Cripps2000). The NADPH-sulfite reductase hemoprotein CysI of A. pleuropneumoniae was shown to be protective when tested as a subunit vaccine, as immunized pigs showed lower mortality and reduced clinical signs after challenge with virulent A. pleuropneumoniae (Willson et al., Reference Willson, Gerlach, Klashinsky and Potter2001).
Many virulence factors of A. pleuropneumoniae alone or more often as a cocktail, have been tested as subunit vaccines for their protective capacities. Despite all the advances made in the vaccination field, none of the subunit vaccines commercialized to date provide complete protection against A. pleuropneumoniae infection. The discovery of an effective subunit vaccine is also limited by the fact that many virulence factors (e.g. the toxin ApxIV) are only expressed in vivo (Schaller et al., Reference Schaller, Kuhn, Kuhnert, Nicolet, Anderson, Macinnes, Segers and Frey1999). Thus, studies are still progressing in the finding of new in vivo-expressed immunogenic antigens using powerful genetic tools.
Evolution of live vaccines towards the DIVA (Differentiating Infected from Vaccinated Animals) concept
The use of live attenuated bacteria in vaccination has always been associated with the possibility of reversion to a fully virulent phenotype and the risk of development of disease in immunocompromised vaccinated animals. Indeed, live attenuated A. pleuropneumoniae vaccines suffer from a number of drawbacks including the risk of inoculating animals with inadequately attenuated pathogens and the possibility that the attenuated bacteria may revert to a pathogenic state resulting in disease of the inoculated animals and the possible spread of the pathogens to other animals. Despite all the disadvantages mentioned above, live attenuated vaccines, along with subunit vaccines, represent the most promising research avenues in the A. pleuropneumoniae vaccination field. The major reason why the use of attenuated live vaccine is a good approach in vaccination against porcine pleuropneumonia is that pigs surviving natural infection were found to be completely protected against homologous infection and partially against heterologous serotypes of A. pleuropneumoniae (Nielsen, Reference Nielsen1984; Cruijsen et al., Reference Cruijsen, Van Leengoed, Ham-Hoffies and Verheijden1995; Haesebrouck et al., Reference Haesebrouck, Van De Kerkhof, Dom, Chiers and Ducatelle1996). This suggests that only live bacteria can confer cross-protection via in vivo-induced expression of protective antigens. A large number of mutants were generated and tested as live attenuated vaccines for their protective efficacy (Table 2). Intranasal immunization of mice with temperature-sensitive mutants of A. pleuropneumoniae serotype 1 induced protection against homologous challenge (Byrd and Hooke, Reference Byrd and Hooke1997). An experimental streptomycin-dependent strain of A. pleuropneumoniae was used as a live attenuated vaccine and showed protection upon homologous challenge with serotype 1 but not against serotype 15 (Tumamao et al., Reference Tumamao, Bowles, Van Den Bosch, Klaasen, Fenwick, Storie and Blackall2004). Several mutants in metabolic genes were generated and tested in a pig infection model. Creation of a riboflavin auxotroph mutant via the partial deletion of the riboflavin biosynthesis operon (ribGBAH) resulted in high attenuation in pigs (Fuller et al., Reference Fuller, Thacker and Mulks1996). Another metabolic mutant aroQ, affected in the aromatic/chorismate biosynthesis pathway was also found to be attenuated at a similar level as the riboflavin mutant (Ingham et al., Reference Ingham, Zhang and Prideaux2002). Mutation in the aroA gene, involved in the essential aromatic biosynthetic pathway, rendered the bacteria fully avirulent with no signs of respiratory disease or lung lesions in any of the animals infected with the mutant (Garside et al., Reference Garside, Collins, Langford and Rycroft2002). However, the use of those metabolic mutants as live vaccines could be a problem since there was no or poor persistence of the bacteria in the respiratory tract of pigs after infection. In fact, to be beneficial in generating a protective immune response, the bacteria have to persist sufficiently in the host to colonize the airways.
Table 2. Live vaccines candidates against A. pleuropneumoniae infection

IM=intramuscular, IN=intranasal, IP=intraperitoneal, SC=subcutaneous.
A dmsA mutant, affected in the putative catalytic subunit DmsA of anaerobic dimethyl sulfoxide reductase involved in oxidative metabolism under anaerobic conditions was found to be attenuated. Interestingly, the challenge mutant strain was reisolated on days 7 and 21 post-infection from the bronchoalveolar lavage fluid (BALF) from several pigs (Baltes et al., Reference Baltes, Hennig-Pauka, Jacobsen, Gruber and Gerlach2003), suggesting a live vaccine potential for this dmsA mutant. A superoxide dismutase sodC mutant that was sensitive to in vitro superoxide microbicidal action failed as an attenuated live vaccine as the mutant was still virulent and caused lung lesions (Sheehan et al., Reference Sheehan, Langford, Rycroft and Kroll2000). Although the bacteria are sensitive to superoxide-mediated killing by neutrophils and alveolar macrophages, they still secrete Apx toxins which rapidly kill host cells.
In order to find potential genes that could be targeted for preparation of live attenuated vaccines, isogenic mutants of A. pleuropneumoniae serotype 7 were generated for two virulence genes ureC and exbB, encoding respectively the urease and the ExbBD complex involved in iron uptake. Infection experiments showed that the ureC mutant but not the exbB mutant is able to survive in pigs and is slightly attenuated (Baltes et al., Reference Baltes, Tonpitak, Gerlach, Hennig-Pauka, Hoffmann-Moujahid, Ganter and Rothkotter2001). Urease can therefore be considered as a potential virulence factor that could be targeted in vaccination experiments.
Studies mentioned above showed that the use of attenuated live vaccines is often limited by the fact that the strain should be less virulent but must also be viable in the host and retain its colonization capabilities to induce a strong immune response. For example, an attenuated strain of A. pleuropneumoniae serotype 1 with a thinner capsule, strain CM5A, was able to persist in the tonsils and induce an effective protective immunity in pigs against challenge with the virulent strain CM5 (Bosse et al., Reference Bosse, Johnson, Nemec and Rosendal1992). Several important characteristics are thus essential for a good live attenuated vaccine: (i) the strain should remain highly immunogenic; (ii) the strain has to be less virulent and cause sufficient but minimum infection and lesions to avoid substantial infection. These suggest that in the case of specific gene inactivation, the targeted genes have to be important virulence factors without being essential for the viability of the bacteria. In this regard, an apxIA mutant of A. pleuropneumoniae serotype 10 producing a C-terminal truncated ApxI toxin was constructed by insertion of a chloramphenicol resistance gene cassette. This mutant offered partial cross-protection upon challenge of vaccinated pigs with serotypes 1 and 2 (Xu et al., Reference Xu, Chen, Shi, Yang, Wang, Li, Guo, Blackall and Yang2006). An apxII mutant of A. pleuropneumoniae serotype 7, lacking both apxIIA and apxIIC genes coding respectively for the structural toxin ApxIIA and the post-translational activating protein ApxIIC, was constructed using site-specific mutagenesis. The HS93Tox-mutant belongs to serotype 7 and as such, also lacks the apxIA and apxIC genes coding for the ApxI toxin. This mutant strain was transformed with a plasmid containing the apxIA gene so that it can express the ApxI structural protein but in a non-activated form. The mutant was shown to be attenuated in a mouse model and to be capable of inducing Apx-specific antibodies (Prideaux et al., Reference Prideaux, Pierce, Krywult and Hodgson1998). Vaccination of mice with the mutant offered protection against homologous wild-type serotype 7 challenge, as well as heterologous challenge with a serotype 1 strain (Prideaux et al., Reference Prideaux, Pierce, Krywult and Hodgson1998). The same group has also used site-specific mutagenesis to generate an apxIIC mutant that secretes an inactivated form of ApxII toxin. Vaccination experiments showed that pigs vaccinated with this serotype 7 live mutant strain via the intranasal route were protected against a cross-serotype challenge with a virulent serotype 1 strain of A. pleuropneumoniae (Prideaux et al., Reference Prideaux, Lenghaus, Krywult and Hodgson1999). The apx mutants mentioned above all displayed non-activated forms of Apx toxins that are still immunogenic. In fact, Apx toxins were shown to be essential for immunoprotection, as previous studies showed that immunization with a non-hemolytic mutant lacking the 110 kDa hemolysin was unable to protect pigs and mice against lethal infection (Inzana et al., Reference Inzana, Todd, Ma and Veit1991). However, the use of those attenuated mutants as live vaccine is again limited by the fact that they contain foreign DNA or antibiotic resistance genes. Indeed, licensing of mutants containing an antibiotic resistance marker for use in livestock might be difficult to obtain due to the risk of resistance transmission to other pathogens. Therefore, even if previous studies have confirmed the safety of mutants containing antibiotic resistance genes (Inzana et al., Reference Inzana, Glindemann, Fenwick, Longstreth and Ward2004), the introduction of mutations without antibiotic markers might prove valuable for future A. pleuropneumoniae vaccine development. Another apxIIC mutant of A. pleuropneumoniae serotype 7 containing no antibiotic resistance marker was generated and showed cross-protection in mice against A. pleuropneumoniae serotypes 1 and 3 as well as in pigs against serotype 1 (Bei et al., Reference Bei, He, Yan, Fang, Tan, Xiao, Zhou, Jin, Guo, Lv, Huang and Chen2005, Reference Bei, He, Zhou, Yan, Huang and Chen2007). Recently, a double ΔapxIC/ΔapxIIC mutant of A. pleuropneumoniae serotype 1 was constructed and investigated for its protective efficacy. This mutant secretes inactivated forms of both ApxI and ApxII which however retain their complete antigenicity. Upon homologous (serotype 1) and heterologous (serotype 9) challenges, intranasally vaccinated pigs were completely protected from clinical signs, showed no mortality and only few lung lesions. These results combined with the fact that the strain contains no foreign DNA suggest a significant live vaccine potential for this double mutant SLW03 (Lin et al., Reference Lin, Bei, Sha, Liu, Guo, Liu, Tu, He and Chen2007).
Another important concern in A. pleuropneumoniae vaccination is that bacterial vaccines currently in use do not allow the differentiation between a vaccinated animal and an infected one. Indeed, it is of major importance to discriminate between immunized and infected pigs for generating and maintaining specified pathogen-free herds which are the optimum choice with respect to long-term animal health and consumer protection. The problem is that live attenuated vaccines are not necessarily affected for surface-exposed and/or immunogenic virulence factors which are important for mounting an antibody-based immune response. Therefore, a serology-based discrimination is not always possible between the wild-type and the attenuated mutant strain. The DIVA concept can be used to allow that discrimination by introducing a negative marker in the live attenuated strain. In order to obtain a DIVA vaccine, a suitable marker has to be: (i) highly immunogenic, (ii) expressed in all serotypes and (iii) not essential for protective immunity.
In previous studies non-capsulated mutants of A. pleuropneumoniae serotypes 1 and 5, obtained following chemical mutagenesis, showed attenuation and good protection upon homologous and heterologous challenges. Interestingly, infected and immunized pigs could be discriminated since production of antibodies against the capsule was not induced in the latter (Inzana et al., Reference Inzana, Todd and Veit1993). Subsequently, Tonpitak et al. (Reference Tonpitak, Baltes, Hennig-Pauka and Gerlach2002) designed a DIVA-based vaccine against A. pleuropneumoniae. A double mutant ΔureCΔapxIIA of A. pleuropneumoniae serotype 2 was shown to be attenuated and protective against homologous challenge. In this mutant strain the toxin ApxII was used as a negative marker as it is highly immunogenic and is also present in 13 of the 15 serotypes of A. pleuropneumoniae. Thus, immunized pigs could be discriminated from infected ones by serological detection using an ApxIIA ELISA test. Starting from this double mutant prototype live negative marker vaccine, a six-fold ΔapxIIAΔureCΔdmsAΔhybBΔaspAΔfur mutant of A. pleuropneumoniae was further generated with additional mutations in three enzymes involved in anaerobic respiration and the Fur ferric uptake regulator (Maas et al., Reference Maas, Jacobsen, Meens and Gerlach2006a). Interestingly, this mutant did not cause clinical disease in contrast to the previously described double mutant which showed some lung lesions (Tonpitak et al., Reference Tonpitak, Baltes, Hennig-Pauka and Gerlach2002). Moreover, although highly attenuated, the six-fold mutant was still able to colonize and persist in intact lung tissue over a period of 6 weeks in small numbers, long enough to induce a humoral immune response. From a vaccination perspective, not only was this mutant in accordance with the DIVA concept, but it also showed significant protection upon heterologous infection with an antigenically distinct A. pleuropneumoniae serotype 9 challenge strain (Maas et al., Reference Maas, Jacobsen, Meens and Gerlach2006a). Despite these encouraging results, the protective efficacy of this six-fold mutant has to be further confirmed upon challenge with other serotypes before it can be used as a live attenuated vaccine. Moreover, the short rise in body temperature observed upon vaccination is not in accordance with current licensing rules for commercial vaccines. Research in the past few years has shown a great potential of live vaccines in A. pleuropneumoniae vaccination in terms of safety, efficacy, stability and also production costs. However, the use of live bacteria in vaccination is usually limited to experimental trials due to ethical issues and restrictive legislation. Vaccine strains should not persist in the host until slaughter age. Hence, further studies are required to increase the safe use of live vaccines and also to improve the efficacy of subunit vaccines which would be more attractive for commercialization.
Mucosal immunity and vaccination strategies
The initial step in the pathogenesis of porcine pleuropneumonia is the colonization of the porcine respiratory tract, followed by the induction of host clearance mechanisms and damage to lung tissues (Bosse et al., Reference Bosse, Janson, Sheehan, Beddek, Rycroft, Kroll and Langford2002). Thus, the epithelial lung surface constitutes the portal for entry of A. pleuropneumoniae via interaction with the pulmonary mucosal surface (Jacques et al., Reference Jacques, Belanger, Roy and Foiry1991; Dom et al., Reference Dom, Haesebrouck, Ducatelle and Charlier1994; Abul-Milh et al., Reference Abul-Milh, Paradis, Dubreuil and Jacques1999). Upon entry, the bacteria are captured by antigen presenting dendritic cells which subsequently migrate in organized mucosal lymphoid tissues such as the broncho-alveolar lymphoid tissue (BALT) to initiate the specific adaptative immune response. Activated lymphocytes are then directed back to the mucosa to mount a local immune response and produce antibodies at the site of infection (Kunkel and Butcher, Reference Kunkel and Butcher2003; Mora et al., Reference Mora, Bono, Manjunath, Weninger, Cavanagh, Rosemblatt and Von Andrian2003a). Thus, induction of mucosal immunity suggests the activation of both humoral and cell-mediated immune responses. One important characteristic of the mucosal immune response is the local production and secretion of dimeric immunoglobulin A (sIgA). This molecule is the major immunoglobulin found in the healthy respiratory tract and is believed to be the most important immunoglobulin for defense at this site (Pilette et al., Reference Pilette, Ouadrhiri, Godding, Vaerman and Sibille2001; Woof and Kerr, Reference Woof and Kerr2006). sIgA has the advantage over other antibody isotypes such as IgG to be resistant to degradation in the protease-rich external environment of mucosal surfaces (Kilian et al., Reference Kilian, Mestecky and Russell1988; Neutra and Kozlowski, Reference Neutra and Kozlowski2006). The majority of polymeric IgA produced in mucosal tissues is transported across the epithelium into the luminal environment where it promotes neutralization of antigens or microorganisms in the mucus by inhibiting the adherence to the mucosal surface. This mechanism is known as immune exclusion. Therefore, one main concern in A. pleuropneumoniae vaccination is to find the best vaccination strategies to stimulate an appropriate mucosal immune response which could provide an effective protection against A. pleuropneumoniae infection.
Several routes of administration of vaccines have been reported, such as systemic immunization via intradermal or intramuscular routes, and mucosal immunization via oral or intranasal routes (Hensel et al., Reference Hensel, Van Leengoed, Szostak, Windt, Weissenbock, Stockhofe-Zurwieden, Katinger, Stadler, Ganter, Bunka, Pabst and Lubitz1996). Most inactivated whole-cell vaccines were tested intramuscularly (Jolie et al., Reference Jolie, Mulks and Thacker1995; Furesz et al., Reference Furesz, Mallard, Bosse, Rosendal, Wilkie and Macinnes1997; Hensel et al., Reference Hensel, Huter, Katinger, Raza, Strnistschie, Roesler, Brand and Lubitz2000; Van Overbeke et al., Reference Van Overbeke, Chiers, Donne, Ducatelle and Haesebrouck2003). Indeed, bacterins were logically found to be less effective when used as mucosal vaccines (Hensel et al., Reference Hensel, Stockhofe-Zurwieden, Petzoldt and Lubitz1995) since killed bacteria cannot colonize the mucosal surface and therefore cannot induce an effective immune response. However, systemic immunization failed to be considered as a good vaccination method as it is associated with many disadvantages. First, the use of syringes is often associated with high risks of needle breakage and inflammatory responses at the site of injection, which could alter the quality of the product. Second, systemic immunization was generally found to be ineffective for the induction of mucosal IgA antibody response (McGhee et al., Reference Mcghee, Mestecky, Dertzbaugh, Eldridge, Hirasawa and Kiyono1992; Kaul and Ogra, Reference Kaul and Ogra1998; Liu et al., Reference Liu, Abdul-Jabbar, Qi, Frazer and Zhou1998; McCluskie et al., Reference Mccluskie, Weeratna, Payette and Davis2002; Goonetilleke et al., Reference Goonetilleke, Mcshane, Hannan, Anderson, Brookes and Hill2003) which is a key element in the protection against airway pathogens. The ideal vaccination strategy for respiratory pathogens should provide both humoral and cell-mediated protection, not only at the relevant mucosal surface, but also throughout the body. In this regard, the ability of mucosal immunization to prime the immune system for both systemic and mucosal responses (Kunkel and Butcher, Reference Kunkel and Butcher2003) suggests that mucosal vaccination might be a more suitable strategy to improve the efficacy of vaccines against A. pleuropneumoniae infection.
To date, three porcine mucosal vaccines are licensed in North America: two using the intranasal route of immunization against transmissible gastroenteritis virus and Bordetella bronchiseptica, and one using the oral route against rotavirus (Gerdts et al., Reference Gerdts, Mutwiri, Tikoo and Babiuk2006). The best way to obtain an effective mucosal immune response in the upper airway is thought to be through the nasal or tonsilar immunization route. Nonetheless, based on the concept of an integrated mucosal immune system which is supported by several oral immunization studies (Pabst and Binns, Reference Pabst and Binns1994; Ogra et al., Reference Ogra, Faden and Welliver2001; Cox et al., Reference Cox, Van Der Stede, Verdonck, Snoeck, Van Den Broeck and Goddeeris2002; Bouvet et al., Reference Bouvet, Decroix and Pamonsinlapatham2002), experimental oral vaccine prototypes against A. pleuropneumoniae infection have been developed. Oral vaccination offers many practical advantages over parenteral immunization. First, vaccine delivery is simple and does not require laborious and time-consuming procedures. Second, it eliminates the risk of inflammatory response at the injection site as well as stress to the animals. However, oral administration of antigens, especially non-replicating ones, presents several challenges that must be overcome in order to achieve an effective protection: the immunogen must maintain its native structure and antigenicity in the acidic pH of the stomach, and it must be stable to proteolytic enzyme digestion in the gastrointestinal tract. In this regard, a variety of oral delivery systems and mucosal adjuvants have been developed to enhance the oral immunogenicity of non-replicating antigens (Ryan et al., Reference Ryan, Daly and Mills2001; Liao et al., Reference Liao, Cheng, Yeh, Lin and Weng2001). Recombinant DNA technology has been used to generate a Saccharomyces cerevisiae strain (Shin et al., Reference Shin, Bae, Cho, Lee, Kim, Yang, Jang and Yoo2005) and a transgenic tobacco plant (Lee et al., Reference Lee, Kim, Kang, Kim, Chung, Yoo, Arntzen, Yang and Jang2006), both expressing the A. pleuropneumoniae ApxIIA toxin (Table 1). A killed whole-cell vaccine of A. pleuropneumoniae serotype 1 has also been incorporated into biodegradable microspheres (Liao et al., Reference Liao, Chiou, Yeh, Chen and Weng2003) in an attempt to protect antigens from the intralumenal environment and reduce the effective dose. However each system has met with little success: both ApxIIA-based oral vaccines induced only a weak antigen-specific immune response, causing a limited protection against A. pleuropneumoniae in a mouse model, while the oral-vaccine microspheres induced a mucosal IgA production but a low systemic immune response (Liao et al., Reference Liao, Chiou, Yeh, Chen and Weng2003). Furthermore, oral vaccination is limited by the fact that immunogens given orally can induce tolerance that reduces the efficacy of the vaccine. Indeed, immunogens fed daily in small doses or in a single high dose often induce oral tolerance that appears to be mediated by cellular or humoral suppressor factors (Mattingly and Waksman, Reference Mattingly and Waksman1980; Challacombe, Reference Challacombe1987; Sosroseno, Reference Sosroseno1995).
As noted above, an optimal immune response in the respiratory tract could be induced by intranasal immunization. As with oral immunization, the intranasal route offers many practical advantages, except that it requires a more complex immunization protocol with full co-operation from the producer. Many live attenuated A. pleuropneumoniae vaccines have been tested in intranasal immunization experiments (Bosse et al., Reference Bosse, Johnson, Nemec and Rosendal1992; Prideaux et al., Reference Prideaux, Lenghaus, Krywult and Hodgson1999; Tonpitak et al., Reference Tonpitak, Baltes, Hennig-Pauka and Gerlach2002; Maas et al., Reference Maas, Jacobsen, Meens and Gerlach2006a), and showed a more effective protection compared to the oral vaccines previously described. Interestingly, live attenuated A. pleuropneumoniae intranasal vaccines were also shown to induce a protective humoral immunity (Bosse et al., Reference Bosse, Johnson, Nemec and Rosendal1992).
One of the greatest challenges in vaccinology today is the development of novel mucosal vaccines and vaccine formulations that are safe, effective and yet cost effective. The delivery system is a critical factor in mucosal immunization (Ryan et al., Reference Ryan, Daly and Mills2001; Gerdts et al., Reference Gerdts, Mutwiri, Tikoo and Babiuk2006). In general, non-replicating antigens such as proteins and killed vaccines are poorly immunogenic when given mucosally. Hence, addition of adjuvants is particularly important in order to stimulate the mucosal immune system. However, the use of adjuvants is frequently associated with tissue damage, which is a main concern in food-producing animal. Thus, one has to choose the right combination of adjuvants in order to develop an effective vaccine that would protect against the disease, but not create unacceptable tissue reaction (Willson et al., Reference Willson, Rossi-Campos and Potter1995). Cholera toxin and heat labile enterotoxin have been shown to be effective mucosal adjuvants for nasal delivery of numerous antigens, but their use has been restricted due to their toxicity (Takahashi et al., Reference Takahashi, Marinaro, Kiyono, Jackson, Nakagawa, Fujihashi, Hamada, Clements, Bost and Mcghee1996; Rappuoli et al., Reference Rappuoli, Pizza, Douce and Dougan1999; Williams et al., Reference Williams, Hirst and Nashar1999). CpG oligonucleotides (ODN) are also known as potent adjuvants that significantly enhance cellular and humoral responses to co-administrated antigens when given parenterally or mucosally (McCluskie and Davis, Reference Mccluskie and Davis1999; Krieg, Reference Krieg2000). In pigs, CpG containing a GTCGTT motif has been shown to be important for optimal stimulation of porcine lymphocytes (Rankin et al., Reference Rankin, Pontarollo, Ioannou, Krieg, Hecker, Babiuk and Van Drunen2001). However, in vivo degradation of ODNs and antigens limits their uptake and their efficiency as immune stimulators. Hence, the formulation of the vaccine plays an important role in the efficiency of mucosal vaccines. Various vaccine-targeting adjuvants (VTA) formulations are suitable delivery systems for antigens and CpG ODNs by the intranasal route in pigs, notably incorporated into biphasic lipid vesicles (Alcon et al., Reference Alcon, Foldvari, Snider, Willson, Gomis, Hecker, Babiuk and Baca-Estrada2003, Reference Alcon, Baca-Estrada, Vega-Lopez, Willson, Babiuk, Kumar, Hecker and Foldvari2005). In fact, intranasal immunization of pigs with a combination of the lipoprotein OmlA and CpG ODNs in biphasic lipid vesicles induced a local immune response with significant amounts of IgG and IgA in nasal secretions (Alcon et al., Reference Alcon, Baca-Estrada, Vega-Lopez, Willson, Babiuk, Kumar, Hecker and Foldvari2005). A recent study showed that tracheal administration of the transferrin-binding protein TbpB of A. pleuropneumoniae in conjunction with an adjuvant formulation containing chitosan, a cationic polysaccharide, enhances both mucosal and systemic immune responses in pigs (Kim et al., Reference Kim, Kim and Lee2007). In light of all the studies performed in vaccination strategies against A. pleuropneumoniae, intranasal administration of antigens along with appropriate vaccine formulations seems to be an effective needle-free vaccine delivery route in pigs, inducing both systemic and local immune responses.
Screening for vaccine candidates using a genome-wide approach
In vivo expression technology (IVET)
Many new approaches have been used in the last decade to identify potential bacterial components to be included in subunit vaccines, or potential genes to be inactivated in live vaccine strains. Development of appropriate genetic tools has enabled the use of these new strategies in A. pleuropneumoniae. In many cases, researchers have tried to identify bacterial factors that are preferentially expressed in vivo, as these should have a role in virulence or persistence in the host. The IVET (Slauch et al., Reference Slauch, Mahan and Mekalanos1994) is a technique in which small genomic fragments potentially containing in vivo active promoters are linked to a gene essential for in vivo growth in an auxotrophic mutant. Using IVET, Fuller et al. (Reference Fuller, Shea, Thacker and Mulks1999) screened a library of 2400 clones, looking for promoters that were induced during an experimental infection in pigs. Ten unique genetic loci were identified and sequenced, and six of them had significant homology to known gene sequences. These genes were called ivi genes, for ‘in vivo induced’ genes. Although these genes seemed to be mostly involved in metabolic pathways, a few of them were found to be linked with virulence. One gene contained a sequence similar to the Haemophilus influenzae mrp gene involved in LPS biosynthesis, and another was later identified as an in vivo-induced organic hydroperoxide reductase that could protect A. pleuropneumoniae from oxidative stress encountered during the infection process (Shea and Mulks, Reference Shea and Mulks2002). One of the ivi genes, ilvI, encodes acetohydroxy acid synthase (AHAS) isoenzyme III, which catalyzes the reaction for the first step in the biosynthesis of the branched-chain amino acids (BCAA; isoleucine, leucine and valine). Enzymes involved in this pathway have been identified as in vivo induced in previous studies with other pathogens (Wang et al., Reference Wang, Mushegian, Lory and Jin1996; Mei et al., Reference Mei, Nourbakhsh, Ford and Holden1997; Sun et al., Reference Sun, Bakshi, Chalmers and Tang2000; Fuller et al., Reference Fuller, Kennedy and Lowery2000a), and it was hypothesized that BCAA biosynthesis is required for survival and virulence in lungs of mammalian hosts (Wagner and Mulks, Reference Wagner and Mulks2006). Using a chemically defined medium, Wagner and Mulks (Reference Wagner and Mulks2006) showed that eight out of ten ivi genes, iviG, iviI, iviP, iviS, iviU, iviX, iviY and ivi17g, had increased activity in BCAA deprived medium. In a subsequent study, a gene with similarity to the lrp gene of Escherichia coli, encoding the leucine-responsive regulatory protein (Lrp), was cloned, sequenced and expressed in vitro (Wagner and Mulks, Reference Wagner and Mulks2007). Electrophoretic gel mobility assays showed that the A. pleuropneumoniae Lrp binds to the iviG and iviI promoters, and might therefore regulate the expression of these genes. The riboflavin auxotroph mutant that was generated in order to conduct the IVET experiments was then used alone as a potential live attenuated vaccine (Fuller et al., Reference Fuller, Thacker, Duran and Mulks2000c). When supplied with limited amounts of riboflavin in order to permit a low level of in vivo replication, mortality was reduced in both homologous (serotype 1) and heterologous (serotype 5) challenges, even though there was no significant reduction in lung pathology.
Signature tagged mutagenesis (STM)
The same group also applied the STM system to A. pleuropneumoniae (Fuller et al., Reference Fuller, Martin, Teel, Alaniz, Kennedy and Lowery2000b). STM systems rely on the unique ‘tagging’ of each transposon mutant with small DNA sequences. Pools of mutants are then screened in vivo, and mutants that are not recovered in vivo but still show in vitro growth similar to that of the wild-type are further investigated. The selected mutants are thought to harbor a mutation in a gene that is essential for in vivo survival. Using over 800 A. pleuropneumoniae mini-Tn10 mutants, Fuller et al. (Reference Fuller, Martin, Teel, Alaniz, Kennedy and Lowery2000b) identified 110 potentially attenuated mutants representing 35 groups of unique loci. Competitive index (CI; [mutant cfu/wild-type cfu]input/[mutant cfu/wild-type cfu]output) determination for each mutant led to the identification of 20 mutants that were significantly attenuated in vivo. Seven mutants, including four mutants with relatively low in vivo CI (genes yaeE, fkpA, tig and HI0379) and three mutants for genes that had also been identified in a previous study in Pasteurella multocida (genes exbB, atpG and pnp), were selected for preliminary vaccine studies against homologous challenge. Although three out of the seven mutants caused some mortality when administered at very high doses (1010 CFU, with 50% mortality in one case), all surviving animals were well protected against homologous challenge, while animals that were vaccinated with a commercial bacterin showed 37.5% mortality (Fuller et al., Reference Fuller, Martin, Teel, Alaniz, Kennedy and Lowery2000b). This system was successful in identifying genes that are known to be involved in virulence processes such as exbB, which is involved in various iron acquisition systems in bacteria and was one of the mutated genes in the 20 significantly attenuated mutants. The exbB mutant, which showed a very low in vivo CI, caused no mortality when administered at very high dose and surviving animals showed complete protection and low lung lesion scores (Fuller et al., Reference Fuller, Martin, Teel, Alaniz, Kennedy and Lowery2000b).
Using a genetic system similar to that of Fuller et al. (Reference Fuller, Kennedy and Lowery2000a, Reference Fuller, Martin, Teel, Alaniz, Kennedy and Loweryb), Sheehan et al. (Reference Sheehan, Bosse, Beddek, Rycroft, Kroll and Langford2003) screened a total of 2064 mini-Tn10 mutants. Whereas bacteria were recovered by lung lavage following infection in the first STM study in A. pleuropneumoniae, Sheehan et al. (Reference Sheehan, Bosse, Beddek, Rycroft, Kroll and Langford2003) observed more consistent recovery of bacteria after homogenization of the entire porcine lung. Moreover, mutants were retained for further studies only if they could be identified as potentially attenuated after two consecutive in vivo screening experiments. Using this protocol, 105 potentially attenuated mutants were identified, with mutations in 55 individual genes. Some of these genes, such as those involved in capsular polysaccharide export, LPS biosynthesis and iron transport, were already known virulence genes in A. pleuropneumoniae, and only 3 genes (genes tig, pnp and apvD/macA) were common to those identified by Fuller et al. (Reference Fuller, Martin, Teel, Alaniz, Kennedy and Lowery2000b). Eleven of the 55 attenuated mutants also showed general growth defects in vitro. The in vivo CI was determined for 14 mutants, and eight of them showed high attenuation while the other six did not seem attenuated although there was consistent lack of recovery of these mutants after in vivo screening. This feature is common to some STM studies (Autret et al., Reference Autret, Dubail, Trieu-Cuot, Berche and Charbit2001; Maroncle et al., Reference Maroncle, Balestrino, Rich and Forestier2002), and the authors hypothesized that those mutants might have very subtle effects on virulence that are not seen at higher dose or in less diverse populations.
As in the IVET study and in the first STM study by Fuller et al. (Reference Fuller, Shea, Thacker and Mulks1999, Reference Fuller, Martin, Teel, Alaniz, Kennedy and Lowery2000b), Sheehan et al. (Reference Sheehan, Bosse, Beddek, Rycroft, Kroll and Langford2003) identified several new potential virulence-related genes in A. pleuropneumoniae. Furthermore, results from this study also helped to gain a better understanding of the diverse iron-acquisition systems of A. pleuropneumoniae, as a second TonB system was identified. Two mutants harboring disrupted tonB genes were identified as potentially attenuated, and DNA sequencing showed two distinct copies of the gene: tonB1, the original tonB gene in A. pleuropneumoniae, shares homology with the Neisseria meningitidis tonB (Beddek et al., Reference Beddek, Sheehan, Bosse, Rycroft, Kroll and Langford2004) and is located upstream of and is co-transcribed with genes exbB, exBD and tbp, coding for the transferring binding proteins (Tonpitak et al., Reference Tonpitak, Thiede, Oswald, Baltes and Gerlach2000). The tonB2 gene seems to form an operon with genes exbB2 and exbD2, and shares homology with tonB genes from P. multocida and Haemophilus sp. (Sheehan et al., Reference Sheehan, Bosse, Beddek, Rycroft, Kroll and Langford2003). The STM study also revealed, using in vivo CI experiments, that inactivation of tonB2, but not tonB1, leads to attenuation.
Selective capture of transcribed sequences (SCOTS)
SCOTS is another strategy that can lead to the identification of genes transcribed in vivo. During SCOTS, RNA mixes comprising pathogen and host molecules are reverse-transcribed to cDNA, and pathogen-specific sequences are captured with photobiotinylated gDNA previously blocked with rRNA-coding DNA sequences (Graham and Clark-Curtiss, Reference Graham and Clark-Curtiss1999). Enrichment of sequences specific for growth in the host is then performed by selective capture using, again, photobiotinylated gDNA, but this time previously blocked with cDNA recovered after growth of the pathogen in culture medium. The scope of SCOTS is therefore different from that of STM, which identifies only genes that are essential for in vivo survival, and similar to that of IVET which leads to the identification of in vivo-induced genes. The SCOTS approach was used with A. pleuropneumoniae: using necrotic porcine lung tissue, Baltes and Gerlach (Reference Baltes and Gerlach2004) identified 46 genes, 20 of which had previously been identified as induced in vivo or involved in virulence in other pathogens. Genes coding for the ApxIV toxin, the putative ABC transporter ApaA, the TbpB small subunit of the transferrin receptor and the dimethyl sulfoxide reductase, which had all previously been detected in vivo, were detected by SCOTS. Other known and putative virulence factors, such as the gene coding for the HgbA hemoglobin receptor and a gene coding for a putative Hsf autotransporter adhesin were also identified (Baltes and Gerlach, Reference Baltes and Gerlach2004). In H. influenzae, Hsf is thought to be the major non-pilus adhesin (St Geme et al., Reference St Geme, Cutter and Barenkamp1996; Cotter et al., Reference Cotter, Yeo, Juehne and St Geme2005).
The experiment was repeated using samples from chronically infected pigs (day 21 post-infection versus day 7 post-infection) (Baltes et al., Reference Baltes, Buettner and Gerlach2007). This time, 36 unique genes were identified, 21 of which code for proteins involved in metabolism. Three genes, coding for elongation factor EF-Tu, ubiquinone reductase, and RNA polymerase B had also been identified in the previous SCOTS study. Of particular interest were genes hlyX, coding for a global anaerobic regulator homologous to the E. coli fumarate and nitrate reduction (Fnr) protein, and gene aasP, coding for a putative autotransporter serine protease. The HlyX protein was shown to complement anaerobic respiratory deficiencies of fnr mutants of E. coli (Green et al., Reference Green, Sharrocks, Macinnes and Guest1992), and also to activate a cryptic hemolytic activity that was fnr-independent. It has been hypothesized that, inside necrotic lung lesions, A. pleuropneumoniae has to rely on anaerobic metabolism to survive, and therefore the overexpression of hlyX does not come as a surprise. Whether or not this regulator can also enhance in vivo transcription of virulence genes in A. pleuropneumoniae has yet to be shown. Properties of the aasP genes were further investigated, as numerous reports over the last few years have highlighted implication of autotransporters in virulence, often as Ig proteases (Mistry and Stockley, Reference Mistry and Stockley2006; Riesbeck and Nordstrom, Reference Riesbeck and Nordstrom2006). Both transcript and protein synthesis were shown to be increased during anaerobic growth, and a putative Fnr binding site was identified in the aasP promoter region. The gene sequence of aasP is identical to that of a putative autotransporter serine protease that was identified simultaneously in a microarray experiment under iron-restriction conducted by our group, and termed Ssa1 (Deslandes et al., Reference Deslandes, Nash, Harel, Coulton and Jacques2007).
DNA microarrays
Although the IVET, STM and SCOTS approaches do lead to the identification of genes putatively involved in virulence, none of these techniques can give an overall knowledge of gene expression in bacteria, or are as powerful as DNA microarrays. The use of microarrays can lead to better genome coverage than IVET and STM techniques, as each and every identified ORF of the bacterial genome is tested. As is the case for SCOTS, DNA microarrays enable the identification of genes that are overexpressed at different levels in a particular condition. Since these platforms have started to be overwhelmingly used at the end of the 1990s, scientists have taken advantage of this large genome coverage to gain an insight into genes that could potentially code for antigenic proteins. For years, scientists working with A. pleuropneumoniae were lacking a reliable and fully annotated genome sequence. Lately, serotype 5b strain L20 was sequenced by the team of John H. E. Nash (National Research Council, Ottawa, Canada). Using bioinformatic tools, 2170 ORFs were identified in the complete genomic sequence of this strain (Genbank, CP000569). This information was then used to generate a DNA microarray with 2033 ORFs, corresponding to 95% of the ORFs of length greater than 160 nt in the genome sequence (http://ibs-isb.nrc-cnrc.gc.ca/glycobiology/appchips_e.html). With the genome data and microarray technology in hand, we have undertaken, with collaborators, various genomic studies in order to identify new potential vaccine candidates. Our strategy enables us to handle efficiently one of the most challenging issues encountered when working with A. pleuropneumoniae, i.e. the existence of 15 distinct serotypes. Using bioinformatics and genome sequences, a list of genes that could putatively code for OMPs or lipoproteins was generated. This objective is the core of many reverse vaccinology projects (Mora et al., Reference Mora, Veggi, Santini, Pizza and Rappuoli2003b), a strategy which relies strictly on available genomic information in order to identify in silico potential vaccine candidates (Serruto and Rappuoli, Reference Serruto and Rappuoli2006). These candidates are then further investigated, and screened in order to satisfy existing criteria for the development of good vaccines for a particular pathogen. As an example, using this approach, approximately 600 novel vaccine candidates have been identified in the serogroup B N. meningitidis (MenB) by the first team to attempt experiments that would later be considered as the hallmark of reverse vaccinology (Pizza et al., Reference Pizza, Scarlato, Masignani, Giuliani, Arico, Comanducci, Jennings, Baldi, Bartolini, Capecchi, Galeotti, Luzzi, Manetti, Marchetti, Mora, Nuti, Ratti, Santini, Savino, Scarselli, Storni, Zuo, Broeker, Hundt, Knapp, Blair, Mason, Tettelin, Hood, Jeffries, Saunders, Granoff, Venter, Moxon, Grandi and Rappuoli2000). Of these novel candidates, 28 could elicit protective immunity and could eventually induce immunity against all meningococcal isolates.
Using multiple bioinformatic algorithms to scan the A. pleuropneumoniae 5b L20 genome, 93 genes were identified as putative OMPs or lipoproteins, and therefore encoding potential surface-exposed antigens (Chung et al., Reference Chung, Ng-Thow-Hing, Budman, Gibbs, Nash, Jacques and Coulton2007). OMPs were then enriched using various extraction protocols, which lead to the recovery of 50 of the 93 potential OMPs and lipoproteins identified in silico (53%) as identified with LC-MS/MS. To date, this study is the first to establish the OM proteome of A. pleuropneumoniae. In silico analyses, although powerful, have some limitations. While these analyses will enable us to generate a list of potential vaccine candidates, this list cannot be considered as entirely representative of the mechanisms that are used by bacteria in their natural host. In order to monitor interactions between bacteria and their environment, gene expression profiling with DNA microarrays can be conducted. By gathering information on the bacterial response to changes in its environment, it is likely that new genes expressed during infection conditions will be discovered. This strategy was used by researchers working on N. meningitidis, shortly after the first reverse-vaccinology experiments were conducted. In a study of gene expression following adhesion to human epithelial cells, approximately 350 genes showed differential expression, 189 of which were overexpressed (Grifantini et al., Reference Grifantini, Bartolini, Muzzi, Draghi, Frigimelica, Berger, Ratti, Petracca, Galli, Agnusdei, Giuliani, Santini, Brunelli, Tettelin, Rappuoli, Randazzo and Grandi2002). Twelve of those overexpressed genes, five of which could elicit production of bactericidal antibodies, were potentially involved in adhesion, and had not been previously identified in the in silico mining of MenB (Serruto and Rappuoli, Reference Serruto and Rappuoli2006). It is therefore clear that microarray technology can identify new potential vaccine candidates, and complement other genome mining methods.
Using DNA microarrays, we have identified genes that are expressed in conditions mimicking the in vivo environment (Deslandes et al., Reference Deslandes, Nash, Harel, Coulton and Jacques2007). Since iron-restriction has long been recognized as a condition encountered in the mammalian host, we first tested the effect of iron-restriction on A. pleuropneumoniae serotype 1. After supplementation of the culture medium with an iron chelator, we identified 210 genes that were differentially expressed, 92 of which were overexpressed. Logically, the major response of A. pleuropneumoniae to ironrestriction was the induction of genes involved in iron transport. While all previously known systems were shown to be upregulated, our experiments also lead to the identification of new potential iron-acquisition systems that could also potentially be induced in vivo. As an example, genes showing homology with the N. meningitidis HmbR receptor, specific for hemoglobin, and genes showing homology with the Yfe chelated-iron acquisition system were significantly upregulated. Of particular interest was also the identification of ORFs homologous to the Ssa1 protein of Mannheimia haemolytica. This protein belongs to the family of subtilisin-like serine proteases, and possesses an auto-transporter domain. The gene was termed aasP by Baltes et al. (Reference Baltes, Buettner and Gerlach2007) shortly after. Gene hlyX was also upregulated under iron-restriction.
Furthermore, we also investigated the transcriptional response of A. pleuropneumoniae after interaction with porcine lung epithelial cells. Transcriptional response of both planktonic bacteria and adherent bacteria was assessed, and major changes were observed. Most of the genes identified were metabolism-related, but some putative components that could be involved in adhesion were also identified (Deslandes V., Auger E. and Jaques M., unpublished observations). To date in A. pleuropneumoniae, only LPS has been shown to play a role in adhesion in vitro (Belanger et al., Reference Belanger, Dubreuil, Harel, Girard and Jacques1990; Paradis et al., Reference Paradis, Dubreuil, Rioux, Gottschalk and Jacques1994, Reference Paradis, Dubreuil, Gottschalk, Archambault and Jacques1999). It would be interesting to identify other genes that are expressed in vivo in the lungs of pigs. However, many technical limitations must be solved before a representative in vivo study can be conducted. Researchers who wish to perform these studies must find ways to isolate bacteria in sufficient amounts and then stabilize the transcriptome very rapidly. Furthermore, contamination with eukaryote mRNA is also a concern.
Finally, we are using DNA microarrays to perform comparative genomic hybridizations and to verify that genes of interest are highly conserved among the reference strains of the 15 serotypes of A. pleuropneumoniae, as well as in field strains of those serotypes most frequently isolated in North America. Those results, combined with the ones obtained in the proteomic and transcript profiling experiments, will enable us to identify new potential vaccine targets that are both expressed in vivo and conserved among all serotypes and biotypes.
Discussion and perspective
The wide spectrum of research in the vaccination field has allowed great developments in A. pleuropneumoniae vaccines. The use of inactivated whole-cell bacterial vaccines was clearly shown to be the least promising vaccination strategy in order to obtain efficient protection against A. pleuropneumoniae infection. In fact, killed bacteria display no colonization of the respiratory tract. Moderate persistence and colonization of the respiratory tract is important for the development of an effective immune response. The limited cross-protection and the absence of in vivo-expressed antigens in non-living vaccines also account for the inefficiency of bacterins. In contrast, this review shows the great potential of subunit and live attenuated vaccines. Despite, the numerous safety and ethical drawbacks associated with the use of live bacteria, live vaccination is probably the best approach against A. pleuropneumoniae as it reflects natural infection and allows the in vivo-expression of immunogenic antigens which are crucial for effective protection. Moreover, the DIVA concept which allows the differentiation between vaccinated and infected animal is an important feature that has to be considered in order to increase the reliability of live vaccines.
Subunit vaccines are another important research avenue in A. pleuropneumoniae vaccination and have the advantage over live vaccines of being less restricted by legislation issues. However the development of subunit vaccines is not an easy task as it suggests not only the discovery of highly immunogenic antigens with a broad protective activity, but also the use of adjuvants and formulations which are key elements for an appropriate stimulation of the host immune system. Indeed, one main concern in vaccination strategy is to find the best way to obtain an effective immune stimulation. These include not only the use of adjuvants and formulations but also the selection of the appropriate immunization route. Thus, among the different immunization methods tested we have highlighted in this review the high potential of intranasal inoculation in the stimulation of mucosal immunity.
Another feature that has to be considered in vaccination is that the ability of a vaccine to generate an effective protection-mediating immune response can differ depending on the genetic background of pigs in a population (Magnusson et al., Reference Magnusson, Bosse, Mallard, Rosendal and Wilkie1997). In this way, vaccination could be allied with commercial livestock breeding strategies in order to select for more responsive pigs. Passive immunization with antibodies is also worth further investigation as an alternative method for vaccination against A. pleuropneumoniae. This approach has become even more attractive in terms of cost and productivity with the large-scale production of IgY antibodies in egg yolks following immunization of hens with bacterial antigens (Shin et al., Reference Shin, Choi, Kim, Hur and Yoo2002). Despite the advances made especially for subunit and live attenuated vaccines, the incomplete knowledge on virulence factors and bacterial antigens expressed in vivo by A. pleuropneumoniae could be one of the reasons why a highly effective vaccine against A. pleuropneumoniae infection has not yet reached the market. Thus, the investigation for vaccine development cannot be dissociated from the new genetic tools available such as IVET, SCOTS and microarrays for the discovery of new in vivo-expressed antigens, and STM for the finding of essential genes for survival (Tables 3 and 4). We believe that those genetic tools in combination with trial experiments will definitely help explore new virulence pathways and subsequently allow the design of more effective vaccines against porcine pleuropneumonia.
Table 3. Proteins identified by 1D-gel and LC-MS/MS after enrichment for OMPsFootnote 1 for which the corresponding genes were also identified in gene expression experiments or by in silico prediction

1 As identified by Chung et al. (Chung et al., Reference Chung, Ng-Thow-Hing, Budman, Gibbs, Nash, Jacques and Coulton2007).
2 Transcriptional profiling under iron-restricted conditions (Deslandes et al., Reference Deslandes, Nash, Harel, Coulton and Jacques2007).
IVET=in vivo expression technology, STM=signature tagged mutagenesis, SCOTS=selective capture of transcribed sequences.
Table 4. Genes identified by more than one gene expression methodology

a Transcriptional profiling under iron-restricted conditions (Deslandes et al., Reference Deslandes, Nash, Harel, Coulton and Jacques2007).
Acknowledgments
Work in M.J. laboratory has been supported by grants from the Natural Sciences and Engineering Research Council of Canada (Discovery grant 003428, Strategic grants 224192 and 306730-04, and Research Networks grant 225155). V.D. is a recipient of a FQRNT scholarship and J.G. of a Michel-Saucier postdoctoral fellowship.