Bacillus subtilis as a tool for vaccine development: from antigen factories to delivery vectors

Bacillus subtilis and some of its close relatives have a long history of industrial and biotechnological applications. Search for antigen expression systems based on recombinant B. subtilis strains sounds attractive both by the extensive genetic knowledge and the lack of an outer membrane, which simplify the secretion and purification of heterologous proteins. More recently, genetically modified B. subtilis spores have been described as indestructible delivery vehicles for vaccine antigens. Nonetheless both production and delivery of antigens by B. subtilis strains face some inherent obstacles, as unstable gene expression and reduced immunogenicity that, otherwise, can be overcome by already available gene technology approaches. In the present review we present the status of B. subtilis-based vaccine research, either as protein factories or delivery vectors, and discuss some alternatives for a better use of genetically modified strains.


INTRODUCTION
Besides the well-known reputation in the industrial production of enzymes, such as proteases and amylases, some Bacillus species, with particular emphasis on B. subtilis, have been explored as a host for the expression of foreign proteins with pharmacological or immunological activities (Harwood 1992, Wong 1995).Such interest stems from a plethora of very solid reasons: (i) as gram-positive bacteria, B. subtilis strains do not have an outer membrane and, thus, all secreted proteins are released directly into the growth medium, which greatly simplifies and reduces the costs of downstream purification steps; (ii) the available knowledge on genetics and physiol-ogy of B. subtilis finds parallel only with Escherichia coli K12, making easier the development of controllable gene expression systems and adaptation to large-scale stream-line fermentation processes; (iii) B. subtilis strains have a well-established safety record and have deserved the GRAS (generally regarded as safe) status; (iv) production of spores, the most resistant life form found on earth, warrants easy preservation of strains even under harsh environmental conditions; and, finally, (v) the ability to grow in simple and non-expensive media at fast growth rates and a non-biased codon usage confer to this bacterial species a top candidate position for the expression of heterologous proteins, including those with potential use in vaccine development (Henner 1990, Wong 1995).
Modern vaccine development approaches have employed genetically modified bacteria in three distinct ways.First live vaccines based on attenuated pathogens can be obtained after introduction of mutations, usually deletion of genes required for the ability to grow or inflict damage into the mammal host (Hormaeche et al. 1999).Second, bacterial strains can be converted into convenient and low cost cellular factories for production of bioactive molecules such as proteins, polysaccharides and nucleic acids.Such compounds can be purified by genetically modified strains and incorporated as antigens into subunit-based vaccine formulations (Vilar et al. 2003).Finally, some bacterial species can be genetically modified in order to become live carriers delivered either by parenteral or, preferentially, by mucosal routes, of passenger antigens derived from one or several pathogens, either as intact proteins or peptides genetically fused to host bacterial proteins (Medina andGuzman 2001, Curtiss et al. 1989).
So far attempts to use B. subtilis strains in vaccine development have focused mainly on the production of recombinant antigens, both as intracellularly expressed proteins or soluble proteins secreted into the extra-cellular medium.More recently, particular attention was given to the use of genetically modified B. subtilis spores expressing surface-exposed antigens genetically fused to spore coat proteins.In this review, both approaches will be considered and some positive and negative aspects will be discussed taking into account the possible impact of this technology on the field of vaccine development.
Modern subunit vaccines require production of highly purified antigens, which can be more easily achieved by genetically modified bacterial or yeast strains, as successfully illustrated by the recombinant hepatitis B vaccine (McAleer et al. 1984).The use of a recombinant microbial host avoids direct contact with the pathogen and usually allows reproducible and high recovery yields at much re-duced costs.So far production of antigens for vaccine use have been mainly based on yeast or E. coli K12 strains as cell factories (Cereghino andCregg 2000, Cornelis 2000).Nonetheless, some Bacillus species, particularly B. subtilis strains, have been already employed in the production of heterologous antigens due to both the availability of several well-established expression systems and the ability to secrete recombinant proteins into the culture medium.In spite of these attractive features only a small number of research groups have consistently worked with recombinant B. subtilis strains as hosts for expression of antigens with potential vaccine application in humans (Table I).
The protective antigen (PA) of B. anthracis is a major protective antigen and is included in the presently available acellular anthrax vaccine (Belton andStrange 1954, Puziss et al. 1963).As a common receptor-binding domain of two B. anthracis toxins, lethal factor and edema factor, PA represents the main target of specific neutralizing antibodies that confer the protective status to the presently available vaccines both for human and veterinary use (Baillie 2001, Friedlander et al. 1999, 2002).As a protein produced by a highly pathogenic Bacillus species, the expression and purification of the PA antigen by recombinant B. subtilis strains represented a safer choice for the production of anti-anthrax acellular vaccines.The PA-encoding sequence was cloned in expression plasmids and successfully expressed and secreted by recombinant strains (Baillie et al. 1994, Ivins and Welkos 1986, McBride et al. 1998).Recombinant PA purified from culture supernatants of B. subtilis cultures proved to be fully immunogenic and conferred protection both to mice and Guinea pigs challenged with lethal doses of B. anthracis spores (Ivins andWelkos 1986, McBride et al. 1998).An expression system based on an inducible B. subtilis lysogenic phage has also been employed to express a secreted and immunogenic form of the B. anthracis PA (Baillie et al. 1994).
Plasmid-based B. subtilis expression systems have also been successfully applied in the production of pertussis toxin (PT) subunits and the Neisseria meningitides class I outer membrane protein (P1), either as soluble secreted or intracellular insoluble proteins (Idänpään-Heikkila et al. 1995, 1996, Himanen et al. 1990, Nurminen et al. 1992, Saris et al. 1990).In both cases protective antibody responses were achieved in animals immunized with the purified proteins but additional steps had to be included in order to enhance or activate the immunogenicity of the encoded antigens.In the case of the P1 protein, incorporation into liposomes or detergent micelles were required for the stabilization of the rather hydrophobic outer membrane protein, while a refolding step was required after denatur-ing of pertussis toxin subunits from inclusion bodies (Idänpään-Heikkila et al. 1995, 1996, Himanen et al. 1990, Nurminen et al. 1992).In both cases the reduced immunogenicity of the antigens expressed by the B. subtilis strains was attributed to the lack of conformational epitopes, lost during expression or purification steps, required for induction of protective antibody responses.Three Chlamydia pneumoniae proteins, two outer membrane proteins and one cytoplasmic heatshock inducible protein, were cloned in expression plasmid vectors and produced in B. subtilis strains either as insoluble (membrane proteins) or soluble cytoplasmic proteins with histidine tags (Airaksinen et al. 2003).Purified proteins elicited specific serum antibodies, after parenteral administration to mice, which reacted with the native proteins expressed by C. pneumonia strains and induced cell proliferation after exposure of splenocytes to elementary bodies (Airaksinen et al. 2003).Other reports have also described the use of recombinant B. subtilis strains for the production of pneumolysin, a pneumococcal toxin, and an outer membrane protein of Haemophilus influenzae, either as secreted or cytoplasmic proteins, encoded by multi-copy plasmidbased expression systems (Srikumar et al. 1993, Taira et al. 1989).Both proteins were suitable for immunological assays but, in the case of the H. influenzae porin, the altered conformation of the recombinant protein preclude the generation of neutralizing antibodies (Taira et al. 1989).
The rather restricted list of proteins expressed by B. subtilis strains aiming the development of subunit vaccines contrasts with the potential advantages of the Bacillus model as a host for protein expression.At least in part, the reduced interest on the use of B. subtilis as a host for protein expression can be ascribed to commercially available expression systems based in E. coli K12 of yeast systems (Cereghino andCregg 2000, Studier et al. 1990).Moreover, the reduced recovery yields obtained with some expression systems have been frequently cited as an argument for the low interest for B. subtilis as a protein factory (Saris et al. 1990, Himanen et al. 1990, Taira et al. 1989).Reduced expression levels of heterologous proteins by recombinant B. subtilis strains may be ascribed to three main factors.First, the unstable gene expression, associated either with the loss of the recombinant plasmid (segregational instability) or rearrangement of the cloned gene (structural instability), may reduce gene expression by loss of the genetic information (Bron et al. 1988, Ehrlich et al. 1986).Most B. subtilis strains produce large amounts of extracellular proteases, which can quickly degrade most the foreign proteins produced and secreted by the recombinant strains (Wang et al. 1988).Finally, the limited number of promoters available for construction of plasmid expression vectors restricted the development of strains with a better performance as cell factories for recombinant proteins.In all three situations, technical solutions for each problem have already been obtained and are available for use in heterologous protein expression.
Integration of the heterologous gene into specific sites of the B. subtilis chromosome has been frequently employed to enhance gene expression stability.The presence of sequences sharing sequence homologies with target chromosomal genes and selective antibiotic markers allows the integration of recombinant genes under the control of different promoters after allelic exchange by homologous recombination (Härtl et al. 2001, Mogk et al. 1996).Stable gene expression in B. subtilis can also be achieved with plasmids replicating via doublestrand intermediates, similarly to plasmids derived from gram-negative bacteria (Jannière et al. 1990).In contrast to chromosome integration methods, enhanced structural and segregational gene expression could be achieved with such vectors without reduction of gene copy number (Jannière et al. 1990).The identification of secreted and cell-bound proteases of B. subtilis permitted the generation of strains defective in 6, 7 or 8 different proteases, which greatly improved the recovery of recombinant proteins from the growth medium of genetically modified strains (Wu et al. 1991(Wu et al. , 2002)).The complete sequencing of the B. subtilis genome and the availability of promoter trap systems greatly simplified the task of identification and evaluation of promoters with potential applications for heterologous gene expression (Gat et al. 2003).Based on the evidences cited above, development of efficient antigen expression systems is clearly feasible with the available knowledge of B. subtilis.

B. subtilis SPORES AS LIVE CARRIERS OF ANTIGENS
Both gram-negative and -positive bacteria have been extensively used as live carriers of vaccine antigens.Such strains can be delivered via the oral route avoiding the use of needles and eliminating the risk of iatrogenic transmission of infectious diseases.In addition, there is no need to purify the expressed antigens, which are produced by the biosynthetic apparatus of the carrier strain.Given by the oral route, such live vaccines can induce both mucosal and systemic immune responses to endogenous and heterologous antigens.They can have a bi-or multivalent character according to the number of expressed antigens and the genetic background of the bacterial host.In contrast to the gram-negative hosts, which are usually represented by attenuated strains of enteric pathogens such as Salmonella and Shigella species (Curtiss 2002, Curtiss et al. 1989, Medina and Guzman 2001), most gram-positive species used as live carrier of vaccine antigens are harmless posing no potential risk to humans or animals (Pozzi and Wells 1997).Indeed most of the gram-positive bacterial species employed as oral vaccine antigen carriers belong to our natural microbiota and/or take part of our daily meals, as lactic bacteria including Streptococcus gordoni, Lactococcus lactis, and several Lactobacillus species (Pozzi andWells 1997, Seegers 2002).As an additional appealing feature, some gram-positive bacterial species used as live carrier of vaccine antigens also display a probiotic action and, therefore, may play a dual role on the treatment and prevention of enteric infections (Holzapfel and Schillinger 2002).
The use of genetically modified B. subtilis strains as live carriers of vaccine antigens was recently reported and, so far, has relied on the unique feature of a few gram-positive bacteria, the ability to form endospores. Bacterial endospores represent the most resistant life form found on our planet and potentially would have an endless shelflive (Nicholson et al. 2000).Moreover, B. subtilis spores have also a record of therapeutic application as probiotics both for humans and animals (Mazza 1994).The initial reports that recombinant B. subtilis spores could carry surface-exposed antigens and induce both systemic and secreted antigen-specific antibody responses after oral deliv-ery to mice were received as new and promising alternative for development of oral vaccines (reviewed by Duc and Cutting 2003, Oggioni et al. 2003, Ricca and Cutting 2003).
Two spore coat proteins, CotB and CotC, have been used to display tetanus toxin fragment C (TTC) of Clostridium tetani and the B subunit of the heat labile toxin (LTB) produced by some enterotoxigenic E. coli strains (ETEC) as fused proteins expressed on the surface of B. subtilis spores (Isticato et al. 2001, Duc et al. 2003a, Mauriello et al. 2004) (Table I).Both antigens were expressed as C-terminal fusions with the Cot proteins, which are by themselves anchored at the outer layer of the spore coat at a number of approximately 1,000 molecules/spore (Isticato et al. 2001, Duc et al. 2003a).In contrast to the attempts to employ B. subtilis strains as expression systems of heterologous proteins for subunit vaccines, the genes encoding hybrid spore coat proteins were integrated into specific sites at the bacterial chromosome, which conferred stability to the expression of the recombinant genes (Duc et al. 2003a, Isticato et al. 2001, Mauriello et al. 2004).Mice orally treated with three consecutive daily doses of recombinant spores, given three times at intervals of two weeks, developed statistically significant secreted fecal (IgA) and serum (IgG) antibody levels which, at least in the case of TTC, could confer protection to mice challenged with lethal doses of the toxin (Duc et al. 2003a, Mauriello et al. 2004).
In spite of the clear interest in the development of a new mucosal vaccine delivery method that could combine both prophylatic and therapeutic effects in an indestructible carrier particle, there are some points which clearly demand further improvements before B. subtilis spores could be efficiently employed as live carriers of vaccine antigens.At present, the main concern is the low immunogenicity of B. subtilis spores in mammal hosts.In contrast to B. anthracis, B. subtilis spores do not efficiently activate strong local or systemic antibody responses, against itself and carrier proteins, particularly when delivered via the oral route (Duc et al. 2003a, Mauriello et al. 2004).Such features could be attributed to the ubiquitous presence of spores in the environment and its chemical composition that is devoid of compounds known to induce inflammatory responses in mammal hosts.Thus, oral immunization regimens with recombinant B. subtilis spores require high antigen loads (above 10 10 spores/dose) and repeated immunizations (Duc et al. 2003a, Mauriello et al. 2004).Indeed some experimental evidences suggested that most of the local and systemic immunogenicity of recombinant B. subtilis spores is triggered during the intracellular germination and transient persistence of the nascent vegetative cell in the phagosome of phagocytic cells, as macrophages (Duc et al. 2003b(Duc et al. , 2004)).

OTHER Bacillus SPECIES EMPLOYED IN VACCINE DEVELOPMENT
B. brevis strains have been successfully used as a host for the production of recombinant proteins with biotechnological or pharmaceutical interest such as cytokines, hormones and single chain antibodies, based on multicopy plasmid expression systems able to promote secretion of encoded proteins into the growth media (Kozuka et al. 1996, Udaka et al. 1989, Udaka and Yamagata 1993, Shiroya et al. 2001, Takimura et al. 1997).Of particular interest for the vaccine field was the production of recombinant toxin subunits of B. pertussis and the LTB subunit of LT produced by ETEC strains, to be used either as antigens for subunit vaccines or as a mucosal adjuvant in the case of LTB, by recombinant B. brevis strains (Byun et al. 2001, Kozuka et al. 2000).Nonetheless, in contrast to the B. subtilis expression systems, recombinant proteins expressed and secreted in B. brevis strains usually require rather long cultivation periods (from 5 to 8 days) in order that maximal expression levels could be achieved (Byun et al. 2001, Kozuka et al. 2000).
A considerable increase in the interest to develop new and more efficient anthrax vaccines has been observed as a direct response to the growing attention to the terrorist threat (Baillie 2001, Friedlander et al. 1999, 2002).B. anthracis vaccines based on the spores of an attenuated strain or as a cell-free vaccine are already available both for humans and animals since many years (Barnard andFriedlander 1999, Cohen et al. 2000).Nonetheless, some research groups have evaluated the behavior of attenuated B. anthracis strains as live carriers of antigens derived from other gram-positive bacterial antigens (Mesnage et al. 1999, Sirard et al. 1997a, b).The Ib component of iota toxin from Clostridium perfringens, the listeriolysin from Listeria monocytogenes and the fragment C of tetanus toxin were successfully expressed by vegetative cells of a B. anthracis vaccine and parenteral administration of 10 8 spores at a single or two dose-based immunization regimens elicited high and protective antigenspecific systemic antibody responses (Mesnage et al. 1999, Sirard et al. 1997a, b).In spite of the excellent immunogenicity in laboratory animal models, clinical use of B. anthracis bivalent vaccine strains seems to be remote but the vaccine system have a clear interest for veterinary use (Brossier et al. 1999).

IMPROVING THE USE OF B. subtilis IN VACCINE DEVELOPMENT
The available evidences indicate that genetically modified B. subtilis strains, as well as other Bacillus species, can represent useful tools for the vaccine research field either as cell factories or live carriers of heterologous antigens.However, both applications demand technical improvements to enhance gene expression and/or immunogenicity of the produced antigens (Table II).At present, most of the technical problems faced by those who had employed B. subtilis strains for the production of vaccine antigens have already been surpassed, as discussed above.
Expression systems located either on stable plasmids (Titok et al. 2003) or allowing ectopic integration at different sites of the chromosome (Guérout-Fleury et al. 1996, Härtl et al. 2001, Middleton and Hofmeister 2004, Shimotsu and Henner 1986)  Improving the use of B. subtilis as a tool for vaccine development.
As a cell factory of antigens for subunit vaccines • Gene expression stabilization by chromosome integration of the recombinant genes or use of structurally stable multi-copy plasmids • Identification of new promoters or/and optimization of existing promoters allowing high level expression and optimal secretion of encoded antigens • Protease-deficient B. subtilis strains as hosts for antigen production As live carriers of orally delivered antigens • Strains able to germinate earlier and persist longer in phagocytic cells • Strains with enhanced resistance to the gastrointestinal environment • Development of expression systems able to drive gene expression under stress conditions • Co-administration of mucosal adjuvants either for biotechnological or biochemical applications (Simonen and Palva 1993, Udaka et al. 1989, Udaka and Yamagata 1993, Wang et al. 1988, Wong 1995, Wu et al. 2002, Wu and Wong 1999).Strong promoters with tightly regulated activity, as those induced by xylose or IPTG, are available for B. subtilis and some of them have already been tested for expression of heterologous or endogenous proteins either as intracellular or secreted proteins (Conrad et al. 1996, Kim et al. 1996).Similarly, B. subtilis strains unable to secrete several proteases have been shown to reduce degradation of secreted heterologous proteins to trace levels, as compared to the previously used laboratory strains, and can be used as more appropriate hosts for the production of vaccine antigens (Wu et al. 1991(Wu et al. , 2002)).
Although promising as a vaccine tool, application of B. subtilis as an antigen delivery vehicle, either as spores or vegetative cells, will demand a better understanding of the fate of cells and spores after oral ingestion or parenteral administration to mammal hosts.Definition of specific events relevant in the triggering of humoral and cellular immune responses can contribute to the development of more rational and efficient vaccine strategies with enhanced immunogenicity.Recent evidences indicate that germination of B. subtilis spores located into phagosomes is a key step on the activation of immune responses since it permit antigen presenting cells to sample, process and display peptides to lymphocytes (Hoa et al. 2001).Similar events seem to occur with the spores of the more immunogenic B. anthracis vaccine strains, which germinate and transiently multiply inside macrophages (Cohen et al. 2000).Thus, development of B. subtilis strains with enhanced ability to elicit immune responses to passenger antigens shall explore mimic some B. anthracis features as early and more efficient intracellular spore germination, ability to transiently survive and multiply inside phagocytic cells or in vivo express antigens more efficiently after phagocytosis.
Expression of listeriolysin has allowed phagosome escape and transient intracellular multiplication of recombinant B. subtilis (Bielecki et al. 1990, Hoa et al. 2001).Mutations allowing early germination of B. subtilis spores are known but the role during the transit into the mammal host is unknown (Setlow 1994).Finally, hundreds of genes activated during exposure to stressful conditions, as low pH or anaerobiosis, are controlled by alternative sigma factor and their promoters could be use to drive the in vivo expression of antigens (Schumann 2003, Wiegert et al. 2001).Similarly strains able to constitutively express σ B -dependent stress regulon may better resist the transit through the gastrointestinal tract as vegetative cells (Völker et al. 1999).Finally, mucosal and systemic immunogenicity of both spores and vegetative cells may be increased by co-administration of mucosal adjuvants as CT or LT (Bowman andClements 2001, Guillobel et al. 2000).Such experimental alternatives can be easily tested and might contribute to the more efficient use of B. subtilis strains as vaccine carries.

THE FUTURE OF B. subtilis AS A TOOL FOR VACCINE DEVELOPMENT
Although some of the most important diseases that scourged mankind during centuries have been eradicated or kept under control by the use of vaccines, several diseases causing millions of casualties are still among us without an efficient vaccine alternative.The search of new vaccines has experienced an extraordinary increase during the last two decades and, at present, hundreds of distinct vaccine formulations are under laboratory testing or clinical evaluation.With this notion in mind, the use of B. subtilis strains, either as a host for production of purified antigens or as orally-delivered carrier of vaccine antigens, may find a more broad interest especially for diseases affecting the gastrointestinal tract, which can also obtain benefits from the probiotic nature of spores.The foundations of B. subtilis-based vaccines have been settled and the instruments to achieve a more efficient performance of the system are known.It is now a question of persistence and dedication of those interested in the field to obtain new B. subtilis strains that could fulfill more efficiently the task.

115 TABLE I Antigens expressed by recombinant B. subtilis strains aiming the production of vaccines.
are available for B. subtilis and, some of its close relatives such as B. brevis.Several antigens have been successfully expressed and secreted into the growth medium in their biological active form amenable