OUP user menu

Towards a Brucella vaccine for humans

Stuart D. Perkins, Sophie J. Smither, Helen S. Atkins
DOI: http://dx.doi.org/10.1111/j.1574-6976.2010.00211.x 379-394 First published online: 1 May 2010


There is currently no licensed vaccine for brucellosis in humans. Available animal vaccines may cause disease and are considered unsuitable for use in humans. However, the causative pathogen, Brucella, is among the most common causes of laboratory-acquired infections and is a Center for Disease Control category B select agent. Thus, human vaccines for brucellosis are required. This review highlights the considerations that are needed in the journey to develop a human vaccine, including animal models, and includes an assessment of the current status of novel vaccine candidates.

  • brucellosis
  • Brucella
  • vaccine
  • human


Brucellosis, and the causative bacterial species, Brucella, have been present for thousands of years (Capasso et al., 2002). David Bruce first isolated Brucella melitensis from the spleen of a British solider who had died of a febrile illness common among military personnel stationed on the island of Malta (known as Malta Fever) in 1886 (Bruce, 1887). The bacterium was termed Micrococcus melitensis, with ‘melitensis’ derived from the Roman name for Malta, ‘Melita’. Contagious abortion in cattle was ascribed to Bacillus abortus by Bernhard Bang in 1897 (Bang, 1897). Later, in 1917, the causes of the two diseases were found to be identical, and renamed Brucella in honour of Bruce.

Brucella are small (0.5–0.7 μm diameter, 0.6–1.5 μm length), nonmotile, non-spore-forming aerobic Gram-negative coccobacilli, found in high prevalence in many areas of the world. Six species of Brucella are classically recognized and classified based mainly on their preferred hosts and pathogenicity: B. melitensis (usually found in sheep and goats), Brucella abortus (cattle and buffalo), Brucella suis (pigs), Brucella ovis (rams), Brucella canis (dogs) and Brucella neotomae (desert wood rat). The main pathogenic species worldwide are B. melitensis, B. abortus and B. suis. Recently, two new Brucella spp., Brucella pinnipedialis (seals and walruses) and Brucella ceti (whales, dolphins and porpoises), have been isolated from marine mammal hosts (Ewalt et al., 1994; Ross et al., 1996), and Brucella microti has been isolated from voles (Scholz et al., 2008). New species have also been isolated from a breast implant (Brucella inopinata) (Scholz et al., 2009) and a captive baboon colony (Schlabritz-Loutsevitch et al., 2009). Overall, there is a high interspecies homology, and the phenotypic and host preference differences are likely attributed to the variation in their proteomes.

Brucellosis is a zoonotic infection of domestic and wild animals. In domestic animals and wild animals, Brucella infection predominantly causes late gestation abortion in pregnant females and orchitis and epididymitis in males (Enright, 1990).

The need for a human vaccine

Four species of Brucella are known to be pathogenic for humans (primarily B. melitensis, B. abortus and B. suis, but also B. canis occasionally). Of these, B. melitensis is thought to cause the most severe and acute form of the disease in humans, with B. suis, B. abortus and B. canis causing milder diseases, in descending order. Human cases of brucellosis are found worldwide every year, although the number of cases reported is considered to be largely underestimated (Pappas et al., 2005). The vast majority of human cases of brucellosis worldwide are attributed to B. melitensis.

Although brucellosis in humans is rarely fatal, it can be severely debilitating and disabling. Acute brucellosis occurs within 8 weeks postexposure and is associated with nonspecific flu-type symptoms such as intermittent fever (rising to 38–41 °C during the evening) and associated sweats, chills, malaise and nausea, as well as anorexia, headache, myalgias and back pain (Sarinas & Chitkara, 2003). Brucellosis also causes enlargement of the liver, spleen and superficial lymph nodes. In addition, neurobrucellosis may occur in up to 5% of patients due to invasion of the central nervous system by the bacteria (Shakir et al., 1986). Chronic brucellosis may occur a year after infection, presenting as chronic fatigue, depression, weight loss and reactive arthritis. Disease may persist for months and relapses are common within 3–6 months of stopping treatment (Solera et al., 1998).

Brucella remains a significant cause of laboratory-acquired infection (Noviello et al., 2004). The characteristics that facilitate easy transmission of Brucella easily transmitted include the low infectious dose for humans and the variety of ways in which the bacteria can enter the body, for example via the respiratory mucosa, oral mucosa (with ensuing colonization of lymph nodes of the head), conjunctivae or abraded skin. Additionally, the organism is able to survive under various environmental conditions including in soil, manure, water and animal tissues (Corbel et al., 1989; Crawford et al., 1990).

Airborne transmission of brucellosis occurs primarily in abattoir workers (Kaufmann et al., 1980) and presents the same symptoms as for disease resulting from other routes of infection. Transmission by the aerosol route suggests utility as a biological weapon. Indeed, the United States began developing B. suis as such in 1942. Although the offensive programme was terminated in 1969, and munitions were never used in combat, Brucella was formulated into bombs to maintain long-term viability and field tested (Purcell et al., 2007). In recent years, interest in airborne Brucella has increased in light of the emerging threat from terrorist attacks. Brucella is a Center for Disease Control (CDC) category B select agent, because it is considered moderately easy to disseminate, can cause moderate morbidity and low mortality, requires specific enhancements to the CDC's diagnostic capacity and enhanced disease surveillance (Greenfield et al., 2002; Sarinas & Chitkara, 2003).

The intracellular lifestyle of Brucella means that only a limited number of antibiotics are effective against these organisms once they have entered their intracellular niche. Commonly used antibiotics for the treatment of human brucellosis include tetracycline, trimethoprim-sulphamethoxazole, aminoglycosides, rifampicin, quinolones and chloramphenicol (reviewed in Al-Tawfiq et al., 2008). These antibiotics are often used in combination due to the high incidence of disease relapse of single-agent therapy, relapse rates often reported to be between 5% and 40% (Colmenero Castillo et al., 1989; Ariza et al., 1992; Montejo et al., 1993).

As a result, and coupled with the current lack of a licensed vaccine, there is a significant need for effective vaccines or treatments for human brucellosis.

Available vaccines for humans

An ideal vaccine for use in humans would be safe and effective. It should not produce disease or more than minimal local or systemic reactions and should provide long-lived protection. Although conceptually a single-administration vaccine may be preferable, this may depend on the efficacy of the vaccine.

Bacillus abortus S19 was widely used as a human vaccine in the former Soviet Union (Vershilova et al., 1961). However, S19 and B. melitensis Rev.1 (Spink et al., 1962) can cause brucellosis in humans and are therefore now considered unsuitable for human vaccination. A variety of live attenuated strains, such as B. abortus strain 19BA or B. melitensis 104M, have been used at some time in the former Soviet Union or China, respectively, but have tended to be reactogenic and of limited efficacy.

In addition, various Brucella fractions have been studied as potential human vaccines, but their efficacy remains uncertain. They include the phenol-insoluble sodium dodecyl sulphate fraction of B. abortus or B. melitensis (also known as fraction PI), developed by the French (Merieux PI), and composed mainly of peptidoglycan, proteins and smooth lipopolysaccharides (Lopez-Merino et al., 1976). This vaccine preparation appeared to be effective. It was immunogenic and protective against Brucella infection in the mouse model. In humans, it was immunogenic (Rasooly et al., 1968; Renoux et al., 1973), but it was also highly reactogenic and caused severe local pain at the site of injection and postvaccination fever (Hadjichristodoulou et al., 1994). The vaccine is no longer produced.

A polysaccharide fraction produced by mild acid hydrolysis developed in the former Soviet Union seems to be protective with minimal reactogenicity in clinical trials, but its availability is uncertain (Dranovskaya et al., 1991).

Requirements of a human vaccine

In order to develop an effective vaccine for controlling human brucellosis, an understanding of the immune response to Brucella is required.

The relative significance of CD4+ and CD8+ T cells in controlling Brucella infection is unclear. Adoptive transfer of CD4+, CD8+ or whole T cell populations from immunized mice to naïve mice resulted in lower splenic colonization after infection than that observed in untreated mice (Araya et al., 1989), suggesting that both T cell populations play a role in Brucella immunity. However, other studies have suggested that CD8+ T cells play a more critical role than CD4+ T cells in controlling brucellosis (Pavolv et al., 1982; Oliveira & Splitter, 1995), particularly in the production of Th1-type cytokines such as IFN-γ and IL-2 (Oliveira & Splitter, 1995). The importance of IFN-γ is supported by studies in which IFN-γ−/− mice infected with Brucella organisms died within 6 weeks (whereas wild-type mice do not succumb to infection) (Murphy et al., 2001). Thus, overall, CD4+ T cells are considered to play a more limited role in controlling Brucella infection than CD8+ T cells, although there are a higher number of CD4+ T cells, and the population does secrete IFN-γ, and so they should not be dismissed as unimportant. In humans infected with B. melitensis, the Vγ9Vδ2 T cell population, possessing the Vγ9Vδ2 T cell receptor, is drastically increased in peripheral blood during the acute phase of infection (Bertotto et al., 1993). These cells are stimulated to produce TNF-α, IFN-γ and a number of different cytokines and are able to reduce the intracellular growth of Brucella either by contact-dependent mechanisms, such as the delivery of lytic granules, or from a distance by the release of soluble factors (Oliaro et al., 2005).

Proinflammatory cytokines such as TNF-α, IFN-γ and IL-12, produced at the onset of infection, have been shown to play a central role in the clearance of the bacterial organism in mice (Dornand et al., 2002). IFN-γ activates the bactericidal function of macrophages, a function that can be maximized by the production of TNF-α by macrophages and natural killer cells (Zhan et al., 1996; Zhan & Cheers, 1998). It is hypothesized that should TNF-α play a similar role in humans, this could be a crucial mechanism for bacterial elimination (Dornand et al., 2002).

Experiments involving the passive transfer of serum in mice suggest a role for humoral immunity in murine brucellosis. For example, the passive transfer of sera containing anti-lipopolysaccharides (Montaraz et al., 1986; Araya et al., 1989; Winter et al., 1989) or an oligopolysaccharide-specific monoclonal antibody (Montaraz et al., 1986; Phillips et al., 1989) protects against B. abortus infection. The dominant antibody isotypes detected in B. abortus-infected mice are IgG2a and IgG3, similar to the natural bovine host, suggesting that a Th1-type immune response occurs in brucellosis (Elzner et al., 1994). Generally, opsonization is considered the most important protective role of antibodies in Brucella infection. Furthermore, a high concentration of IgG during infection of cattle appears to prevent complement-mediated extracellular bacterial lysis and promotes phagocytosis of Brucella, enhancing the extension of the disease (Hoffmann & Houle, 1995). Thus, the role of antibody in protecting against brucellosis is not clearly understood.

Overall, there are three main mechanisms of the adaptive immune response in brucellosis that appear to be important. First, IFN-γ produced by CD4+, CD8+ and γδ T cells activates the bactericidal action of macrophages to hamper the intracellular survival of Brucella. In addition, the cytotoxic action of CD8+ and γδ T cells kills infected macrophages and, finally, Th1-type antibody isotypes opsonize the bacteria to facilitate phagocytosis. To maximize the likelihood of developing effective vaccines, these immunological findings highlight that a predominantly Th1-type immune response should be induced and that a cytotoxic response may be important in controlling human infection (Fugier et al., 2007).

Models for vaccine efficacy

Because brucellosis is a zoonotic disease found naturally in a range of animal species, there are numerous citations of different Brucella species infecting different animal hosts, and a range of clinical signs and pathological observations have been recorded.

Mouse model

Although naturally a disease of larger mammals, a small rodent model of Brucella would be beneficial for the determination of vaccine efficacy (as well as other types of studies) because rodents are routinely bred, housed and handled for such studies, and ethically and financially, larger numbers of smaller animals may be used. However, in most mice strains, there are no outward signs of Brucella infection, and so protection is measured as the ‘reduction at a specified time after challenge of the number of colony forming units (CFU) of Brucella recoverable from the spleen, or liver, or both’ (World Health Organization). A similar definition of protection was used in original studies to determine the protective efficacy of the Rev.1 live vaccine strain, where protection was assessed by the ‘ability of animals to sterilize spleen tissue of the organisms taking up the challenge’ (Herzberg et al., 1953). Most studies with mice use bacterial loads in the spleen as an indication of infection. For example, Jiménez de Bagüés et al. (1993) showed that the peak bacterial spleen load with B. ovis was between 2 and 4 weeks postinfection, but bacteria could still be detected after 26 weeks. Infection by the intraperitoneal or the intravenous routes yielded bacterial counts between 104 and 1010 CFU, whereas the subcutaneous route of infection resulted in less bacteria per spleen. Not surprisingly, Jimenez de Bagues and colleagues observed that higher infection doses resulted in higher spleen counts.

Immunocompromized mice have been used as mouse models to determine the virulence of different Brucella strains (Ko et al., 2002), particularly interferon regulatory factor (IRF)−/− mice. These are IRF-1-deficient mice defective in the production of IL-12, IFN-γ and cytotoxic CD8+ T-cells necessary for the clearance of bacteria. IRF−/− mice appear to have a low level of immunity and can control some levels of infection as a critical level of Brucella is needed to cause death (Ko et al., 2002). Infection with 5 × 105 CFU of B. abortus S2308 intraperitoneally resulted in 10/10 deaths by 4 weeks postchallenge. Recent work has also shown that IRF-1−/− mice succumb to infection with 1 × 107 CFU of the B. melitensis strain 16M after 9 days (Rajashekara et al., 2005). In this study, Rajashekara and colleagues used bioluminescence to monitor the spread of disease in these knockout mice. IRF-1−/− mice appeared sick by day 5 and died 8 days postinfection and >109 CFU bacteria could be isolated from the liver and spleen. Chronic infection was also associated with bacteria in the joints and bacteria were localized in the salivary glands, which may be significant for the oral route of infection. In comparison, Rag1−/− mice lacking the recombination-activating gene 1 (a defect resulting in a lack of mature T or B cells) did not succumb to disease, but failed to clear infection, with 104B. melitensis 16M organisms delivered via the intranasal route (Izadjoo et al., 2000). Such mice, mutated in components of the immune response, are useful for assessing the virulence or the attenuation of Brucella, as well as for determining the role of different immune components in a protective response. As such, they have a role to play in developing vaccine candidates, although wild-type animal models are also important as they reflect the contribution of the ‘normal’ unaltered innate and adaptive immune response to infection and its role in protection.

A mouse infection method has been standardized for the evaluation of live smooth Brucella vaccines that is currently accepted by the European Union, allowing the safety and immunogenicity of B. melitensis Rev.1 and B. abortus S19 to be determined on two axes of (1) the residual virulence of a live vaccine, i.e. the persistence in the spleen determined as the recovery time 50 of the vaccine and (2) the immunogenicity of a vaccine, i.e. the ability to protect against a challenge with a virulent strain determined as the number of virulent bacteria retained in the spleen. Defined experimental conditions are followed, including the number and breed of animals, route of inoculation, doses, challenge strain and time intervals at which the spleen counts are determined (Grillü et al., 2000). Ideally, evaluation of new vaccine candidates in a mouse model would involve a comparison of vaccinated vs. naïve animals and include an established live attenuated animal vaccine for comparison (Fig. 1).

Figure 1

Schematic representation of the method for evaluating novel vaccine candidates for brucellosis in mice.

Recently, there have been three reports on the exposure of mice to Brucella by the aerosol route, which represents a common natural route of infection, and the route most likely to be encountered in a bioterrorism scenario. In one study, the Madison Chamber was used to expose BALB/c mice to three different aerosol doses (107–109 CFU) of B. abortus 2308 or B. melitensis 16M (Kahl-McDonagh et al., 2007). This work determined the bacterial loads in the liver, lungs and spleen at time points between 1 and 8 weeks postinfection and showed that B. abortus caused chronic infection in lung tissue, whereas B. melitensis numbers decreased more rapidly. This study also showed an increase in spleen weight with increased dose and increased time postinfection and demonstrated protection, through reduced bacterial loads, after intraperitoneal immunization with unmarked deletion strains of Brucella (Kahl-McDonagh et al., 2007). The same B. melitensis and B. abortus strains were used for aerosol delivery via a jet nebulizer, a study that also showed consistent colonization of liver, lung and spleen tissues, but failed to show vaccine-induced protection with 1 × 107 CFU B. abortus RB51 against an aerosol challenge (Olsen et al., 2007). Most recently, our laboratory has both confirmed and extended the data generated with B. melitensis and, additionally, developed a mouse model of B. suis infection following inhalational exposure (Smither et al., 2009). In our study, B. suis strain 1330 caused a more acute infection than B. melitensis 16M in BALB/c mice, with higher numbers of bacteria colonizing the spleen and lungs and bacterial loads peaking earlier in infection. Together, these aerosol models supplement a previously described intranasal route of infection model in mice in which mice were given a range of doses of B. melitensis 16M (Mense et al., 2001). The mass of the spleens of infected mice increased between 3 and 7 weeks postinfection and bacteria were detected in the spleen up to 140 days postinfection and were also recovered from the lungs up to 4 weeks postinfection. An increase in the area of the white pulp was observed in the spleen, but no histological changes were identified in the lungs (Mense et al., 2001).

A range of other small rodents were evaluated for their susceptibility to Brucella infection by Thorpe et al. (1967). The ID50 dose in a range of wild mice species varied from 20 organisms to >107 organisms and there were marked differences in susceptibility with the same organism among different species. Generally, rats, lagomorphs and squirrels were more resistant than smaller rodents. Another study demonstrated that rats infected with B. abortus appear lethargic, febrile and anorexic within 24 h, but the symptoms are resolved within a week postinfection (Baek et al., 2005). In this model, bacteria may be cultured from the blood, spleen and testes, and vertical transmission of the bacteria can occur.

Other animal models

Clearly, an ideal animal model for brucellosis in humans would show different susceptibilities to the different Brucella species, reflecting the virulence of the bacteria in humans. Braude (1951) showed that, in guinea-pigs, B. melitensis causes a severe illness with clinical signs including visible abscesses, a ruffled coat and hair loss as well as bacterial colonization of the lungs, liver and testes, and resulting in three of 18 male animals succumbing to the infection. In comparison, B. suis infection resulted in a number of visible abscesses, mostly in the spleen, but no animals succumbed to infection during the course of the experiment and B. abortus infection caused no signs other than an enlarged spleen upon necropsy (Braude et al., 1951). Elberg & Henderson (1948) reported that guinea-pigs are also susceptible to B. melitensis and B. suis infection by the inhalational route. In this work, the bacteria were not localized in the pulmonary tract, suggesting that Brucella organisms were disseminated from the respiratory route of infection.

Because brucellosis has natural animal hosts, the pathogenesis of brucellosis and the evaluation of vaccine candidate efficacy may be studied in these animals. In Beagles, an oral or an intravenous infection with B. abortus resulted in a variety of clinical signs, most commonly a vaginal discharge in females (Palmer & Cheville, 1997). Brucella-specific antibodies were detected for up to 7 weeks postinfection and bacteria were isolated from a range of tissues including the liver, spleen and lymph nodes, but not from the blood, urine or faeces. At necropsy, only pregnant females had any lesions, mostly in the uterus. In pregnant goats, B. abortus is distributed through a range of tissues including the liver, lung, spleen, uterus, udder, blood and a range of lymph nodes following infection (Meador & Deyoe, 1986). The infection of goats with B. melitensis and B. abortus produces reproducible colonization and abortion profiles enabling the caprine model to be used to test vaccine candidates (Elzer et al., 2002). Such a model may prove useful in the evaluation of test vaccine candidates following initial efficacy evaluation in mice, although these studies take a long time to complete and may also have limited relevance to vaccination against human brucellosis.

Nonhuman primates

For the development of a human brucellosis vaccine, the most useful model of the disease may be a nonhuman primate. In an early study in 1929, rhesus macaques were infected with B. melitensis, B. suis or B. abortus by the oral route. Some macaques fed B. abortus became emaciated, appeared disinterested and had decreased appetite, and yet their Brucella-specific antibody titres remained low and bacteria were only recovered from the spleen (Huddleston & Hallman, 1929). Brucella suis appeared to be the most infectious in this species, with very high antibody titres against Brucella and bacteria cultured from the blood, kidneys, liver, lung and spleen. In another study, stump-tail macaques were infected with B. canis orally, intravenously and through the conjunctival sac (Percy et al., 1972). No clinical signs were observed and haematological values remained normal throughout infection, but Brucella organisms were isolated from the blood from 2 weeks postinfection in all animals. Brucella canis was also isolated from one animal's liver and kidney, and from the uterus of a different female. Histopathological changes were observed in the liver, spleen, lymph nodes, heart and reproductive tract of some of the infected monkeys and antibodies to B. canis were detected from 2 to 5 weeks postinfection.

More recently, rhesus macaques were exposed to aerosolized B. melitensis. In this study, the spleen was the organ most colonized, followed by the liver, lung and blood (Mense et al., 2004). Splenomegaly was very common, liver abscesses were observed with a high frequency at higher Brucella inoculum doses and lesions were apparent in the kidney and lymph nodes. One monkey showed signs of epididymitis, orchitis and splenitis, observations believed to be similar to observations of brucellosis in humans. Necropsy of infected animals showed that the white pulp area of the spleen increased with higher inoculum doses, as did the total area of hepatitis. In the blood, there was a correlation between inoculum size and anti-Brucella titre.

From these studies in animals, it is clear that, irrespective of the route of infection, brucellosis is a multiorgan disease and the Brucella organisms quickly disseminate from the site of infection via the mononuclear phagocyte system to all areas of the body, most commonly recovered in the highest amounts from the spleen. In practice, most vaccine candidates for brucellosis are initially tested in the established mouse model of Brucella infection via an injection. Promising live Brucella vaccines for animal use could be evaluated in larger animal models such as the caprine model, while human vaccines should ideally be tested in nonhuman primates. Considering the possible use of Brucella as a biological weapon, it may be necessary to evaluate human vaccine candidates in aerosolized Brucella infection models.

Human brucellosis vaccine candidates

Because there is currently no licensed vaccine for human use, there is a desire to develop a safe and efficacious vaccine for the protection of at-risk workers, general populations living in endemic areas or in the event of a bioterrorism scenario, or for military personnel for protection against Brucella used in biowarfare. The administration route of a human vaccine needs to be able to generate the appropriate immune response, but also needs to take into account practical considerations of administration.

Live attenuated vaccines

Live attenuated vaccines are generally considered to be the vaccine type of choice for intracellular bacteria such as Brucella spp., providing in vivo-expressed antigens to which an appropriate immune response may be generated. Using modern recombinant DNA techniques, new candidate live-attenuated vaccine strains of Brucella have been engineered for evaluation. Various attenuated mutants have been identified in Brucella via random mutagenesis approaches to identify genes encoding virulence factors (Allen et al., 1998; Hong et al., 2000; Ficht et al., 2002). This range of attenuated mutants provides strains that are able to persist in host tissues following infection for varying amounts of time and may be evaluated as new live attenuated vaccine candidates. In order to evaluate the protective efficacy of attenuated Brucella strains as a function of persistence of the organism, Kahl-McDonagh et al. (2006) constructed unmarked deletion mutants in B. melitensis of three candidate genes that persisted for different periods. The Δasp24 mutant, which persisted for the longest period of time in vivo, was found to be the most effective strain at providing protection against subsequent B. melitensis infection in the mouse model. This mutant persisted in the mouse at 16 weeks postinoculation, at the end of the described study, highlighting that its protection was afforded by the long period of survival of the strain. Overall, a correlation was observed between persistence of the candidate live attenuated vaccine strain and protection against infection. The Δasp24 mutant has also been evaluated in the goat model, providing protection against infectious abortion (Kahl-McDonagh et al., 2006). A potential problem with the use of live attenuated Brucella vaccines for use in humans involves the potential for incomplete clearance of persistent strains. One approach to addressing this issue that is currently being evaluated is to use highly attenuated live vaccine candidates and enhance their immunogenicity by encapsulating the bacteria in microspheres for slow release (Arenas-Gamboa et al., 2008, 2009). It should be noted, however, that S-lipopolysaccharide-containing candidates are almost certain to interfere with serological tests used for diagnosing human brucellosis. This would require the development of appropriate assays to differentiate the responses induced by vaccines from that induced from field pathogenic strains. A subunit vaccine approach, for example, would not require this.

Subunit vaccines

Subunit vaccines are attractive for vaccination purposes because of their safety profile. However, such vaccines often fail to elicit the magnitude of immune response that live attenuated vaccination does. Therefore, much effort is expended in increasing immune responses to specific antigens to achieve this maximal effect. Because the mouse model of Brucella infection is nonlethal, a positive control is required to understand the log-fold decrease in bacterial colonization that vaccination strategies should achieve. Therefore, animal vaccines, although unusable in humans, may be used to set a ‘standard’ of protection that subunit vaccines aspire to achieve.

Recombinant proteins

Recombinant protein-based vaccines have been licensed for use in humans. For example, vaccines have been developed for hepatitis B and human papillomavirus (Plotkin & Plotkin, 2004). A number of subunits of Brucella have been examined as recombinant protein vaccines in mouse models of disease, and some of these have shown protective efficacy (Table 1). However, recombinant proteins generally require an adjuvant to improve immune responses. Thus, the key to developing a protein-based vaccine for brucellosis is the identification of the correct antigenic formulation coupled with an adjuvant that drives an appropriate immune response for protection against Brucella species.

View this table:
Table 1

Protective proteins in mouse models of brucellosis

Antigenic proteinPropertiesFormulationStrains protected againstEvidence of similar or improved protective responses compared with live animal vaccines
22.9 kDaUnknownProtein (Cespedes et al., 2000)B. abortusProtein formulated in IFA was able to confer protective efficacy equivalent to RB51 (Cespedes et al., 2000)
32.2 kDaUnknownProtein (Cespedes et al., 2000)B. abortus
BLSLumazine synthaseProtein (Velikovsky et al., 2003) DNA vaccine (Velikovsky et al., 2002, 2003)B. abortus
Cu–Zn SODCu–Zn superoxide dismutaseDNA vaccine (Onate et al., 2003; Munoz-Montesino et al., 2004; Gonzalez-Smith et al., 2006; Singha et al., 2008) SFV particles (Onate et al., 2005) E. coli (Onate et al., 1999; Andrews et al., 2006)B. abortusSFV particle-delivery of SOD elicited similar protection as RB51 (Onate et al., 2005) DNA vaccine able to stimulate similar protection as RB51 in mice (Onate et al., 2003) and able to elicit immune responses in cattle (Saez et al., 2008)
DnaKMolecular chaperone of hsp70 familyProtein (Delpino et al., 2007)B. abortus
GAPDHGlutaraldehyde-3-phosphate dehydrogenase enzymeDNA vaccine (Rosinha et al., 2002)B. abortus
IalBInvasin B proteinDNA vaccine (Commander et al., 2007)B. melitensis
L7/L12Ribosomal proteinProtein (Oliveira & Splitter, 1996a) DNA vaccine (Kurar & Splitter, 1997) L. lactis-vectored (Shi Da et al., 2006) Escheriosome-delivered (Mallick et al., 2007)B. abortusL7/L12-containing escheriosomes are able to elicit protection in mice equivalent to S-19 (Mallick et al., 2007)
Omp16Outer membrane proteinProtein (Pasquevich et al., 2008)B. abortusAble to induce protective immunity equivalent to S19 whether delivered by the intraperitoneal or oral routes (Pasquevich et al., 2008)
Omp19Outer membrane proteinProtein (Pasquevich et al., 2008)B. abortusAble to induce protective immunity equivalent to S19 whether delivered by the intraperitoneal or oral routes (Pasquevich et al., 2008)
Omp31Outer membrane proteinProtein and Peptide (Cassataro et al., 2005a) DNA vaccine (Cassataro et al., 2005b) DNA vaccine–protein (Cassataro et al., 2007c)B. ovis B. melitensisrOmp31, peptide and DNA vaccine are able to elicit similar levels of protection as H38 against B. ovis, but not B. melitensis (Cassataro et al., 2005a, b)
P39Periplasmic binding proteinProtein (Al-Mariri et al., 2001)B. abortusProtection against challenge equivalent to B19 at 4 weeks, but protection was not sustained (Al-Mariri et al., 2001)
SurAPeriplasmic peptidyl cistrans isomeraseProtein (Delpino et al., 2007)B. abortus
BCSP31, SOD and L7/L12Outer membrane protein, Cu–Zn superoxide dismutase, ribosomal proteinCoimmunization of DNA vaccine plasmids (Yu et al., 2007)B. abortusAble to elicit 10–20-fold higher levels of antibody and greater protection than the S19 vaccine (Yu et al., 2007)
BLS and Omp31 peptideLumazine synthase, outer membrane proteinChimeric protein (Cassataro et al., 2007a) Multivalent DNA vaccine (Cassataro et al., 2007b)B. ovis B. melitensisInsertion of Omp31 peptide into BLS conferred equivalent protection vs. B. ovis as Rev.1 as both protein (Cassataro et al., 2007a) and DNA vaccine (Cassataro et al., 2007b)
bp26 and TFDiagnostic antigen, trigger factor (immunophilin)Protein (Yang et al., 2007) DNA vaccine (Yang et al., 2005)B. melitensisCoimmunization of recombinant bp26 and TF in CT was able to reduce bacterial load by a factor of three (Yang et al., 2007)
L7/L12 and Omp16Ribosomal protein, outer membrane proteinFusion protein and DNA vaccine (Luo et al., 2006a)B. abortusDNA vaccine able to elicit protective effect equivalent to RB51 (Luo et al., 2006a)
L7/L12 and P39Ribosomal protein, periplasmic binding proteinDNA vaccine (Luo et al., 2006b)B. abortus

The type of immune response elicited by recombinant Brucella protein vaccines may be influenced more by the antigen rather than by the choice of adjuvant. For example, the 18-kDa cytoplasmic protein Brucella lumazine synthase (BLS) was administered to mice with different adjuvants (Velikovsky et al., 2003). BLS was formulated with an aluminium hydroxide gel (AL) or incomplete Freund's adjuvant (IFA), both of which tend to drive a Th2-dominated immune response, or with monophosphoryl lipid A, which tends to drive a Th1-type immune response. Each of these adjuvants increased antibody responses, but did not alter the predominance of IgG1 over IgG2a. Additionally, all three adjuvants were able to upregulate IL-2, IFN-γ and IL-10 production. The level of protection against B. abortus challenge observed in the spleen following BLS immunization was also independent of adjuvant, but AL and IFA were more effective at providing protection in the liver. Although there was an improvement over BLS without an adjuvant, none of these formulations offered the same level of protection as the B. melitensis H38 vaccine strain. This data therefore suggest that an adjuvant is necessary to augment both the IgG response and to elicit cell-mediated responses when immunizing mice with BLS, but that the choice of adjuvant is less critical. This also appears to be the case with outer membrane protein (Omp)31 (Cassataro et al., 2005a), Omp16 and Omp19 (Pasquevich et al., 2008), and may be an important observation in view of the limited numbers of adjuvants licensed for human use.

Data generated with a putative periplasmic-binding protein of Brucella (P39) support the need for an adjuvant in recombinant protein vaccines for brucellosis (Al-Mariri et al., 2001). The addition of CpG oligodeoxynucleotide to P39 allowed mice to elicit a Th1-orientated immune response in BALB/c mice and an IFN-γ proliferative response. Protection against B. abortus 544 challenge was only provided if P39 was formulated with the CpG and the level of protection afforded by P39-CpG to the mice at 4 weeks postchallenge was equivalent to the live smooth B19 strain. Although this level of protection was not sustained and was less than that afforded by B19 at 8 weeks postvaccination, these data provide encouragement that a recombinant protein-based vaccine can be effective in providing protection against brucellosis.

Additionally, a number of other recombinant Brucella proteins have been shown to afford a level of protection to BALB/c mice equivalent to that of a live vaccine strain. The 31-kDa Omp31 is a haemin-binding protein that has been used to immunize BALB/c mice when formulated in IFA, eliciting high IgG titres and specific Th1 responses as characterized by the induction of IFN-γ and IL-2 (Delpino et al., 2006). Although a CD8+ CTL response was evident, these cells were not protective in vivo. Instead, the protection elicited by Omp31 was mediated by CD4+ cells. Omp31 elicits a similar level of protection as the H38 vaccine against challenge with B. ovis, but is not as effective against a challenge with the more virulent B. melitensis strain (Cassataro et al., 2005a). The 22.9-kDa protein of B. abortus formulated in IFA was as effective in protecting against B. abortus challenge as the live B. abortus RB51 strain vaccine, but was not effective without an adjuvant (Cespedes et al., 2000). Similarly, coimmunization of recombinant bp26 (a Brucella periplasmic protein) and trigger factor (TF; suggested to be a chaperone based on homology with an Escherichia coli gene), administered by the intranasal route to mice, was able to protect against B. melitensis challenge, reducing the bacterial load by a factor of three (Yang et al., 2007). Other recombinant Brucella proteins shown to have some protective efficacy against B. abortus include the ribosomal L7/L12 protein (Oliveira & Splitter, 1996b), the stress proteins DnaK and SurA (Delpino et al., 2007) and the 32.2-kDa protein of B. abortus (Al-Mariri et al., 2001). However, despite the formulation of these proteins with Freund's adjuvants, they afforded lower levels of protection than either RB51 or S19.

A number of Brucella proteins have proved nonprotective in animal models. These include bacterioferritin, a T dominant antigen that, although able to produce an appropriate immune response in mice when formulated with CpG ODN, was unable to protect against B. abortus 544 challenge (Al-Mariri et al., 2001). Brucella heat-shock proteins also appear to be ineffective as recombinant protein vaccines. GroEL formulated in RIBI (Bae et al., 2002) or IFA (Leclerq et al., 2002), and GroES, HtrA, GroEL+GroES and GroEL+GroES+HtrA formulated in RIBI all failed to provide protection against B. abortus 2308 despite evidence that these proteins are able to elicit Brucella protein-specific immune responses (Oliveira et al., 1996; Bae et al., 2002). Some proteins shown to induce immune responses such as YajC (Vemulapalli et al., 1998) and UvrA (Oliveira et al., 1996) of B. abortus remain unproven against challenge. Other proteins, such as SecD and ssb, are unable to elicit an immune response (Oliveira et al., 1996; Vemulapalli et al., 1998).

The combining of antigens has met with mixed success. Coadministration of proteins does not necessarily induce either an additive or a synergistic effect (Cassataro et al., 2007a; Delpino et al., 2007). However, attempts to further define the protective regions of Brucella antigens have allowed the identification of an Omp31 peptide (amino acids 48–74) able to elicit protection against B. melitensis challenge similar to that of the full-length protein (Cassataro et al., 2005a). Crucially, this exposed loop of Omp31 can be inserted into the N-termini of the BLS protein to produce a chimeric molecule that confers enhanced protection to mice such that they are able to resist B. ovis challenge infection similar to that of the B. melitensis Rev.1 live attenuated vaccine (Cassataro et al., 2007a).

Vectored vaccines

Vector systems for vaccination can elicit potent and wide-ranging immune responses in vivo because the antigen is synthesized and processed in the host. One feature of vaccine vectors is the potential for delivery by less invasive routes. For example, Semliki Forest virus (SFV) particles (Jerusalmi et al., 2003), Yersinia enterocolitica (Al-Mariri et al., 2002) and Lactococcus lactis (Xin et al., 2003; Bermudez-Humaran et al., 2005) may all be administered by mucosal routes of administration. Some of the antigens produced as recombinant proteins have also been successfully delivered to mice using vector-based systems. For example, RNA encoding the Brucella superoxide dismutase (SOD) protein and packaged into replication-deficient SFV particles was able to protect mice against B. abortus 2308 challenge at a level similar to that of the live RB51 vaccine (Onate et al., 2005). However, it is important that careful consideration is given to the vector system used. For example, the use of the vaccinia virus to deliver GroEL (Baloglu et al., 2000; Bae et al., 2002) or L7/L12 (Baloglu et al., 2005) was unable to induce a protective response. Although not surprising for GroEL, L7/L12 is protective in mice when administered as a protein or encoded in a DNA vaccine. Because a critical property of vaccinia virus is the ability to downregulate the host's IFN-γ response, the very response that significantly contributes to protection against Brucella, the vaccinia virus may be an inappropriate delivery vehicle for Brucella vaccines.

Mice vaccinated with live E. coli expressing Cu–Zn SOD were able to elicit cytotoxic T cells and a Th1-type immune response, evidenced by IFN-γ production and an IL-4 response (Andrews et al., 2006), and were protected against challenge with B. abortus 2308 (Onate et al., 1999). A separate approach using E. coli has demonstrated that E. coli lipid-based liposomes (escheriosomes) are able to fuse with the membrane of antigen-presenting cells and deliver entrapped solutes to their cytosol. The L7/L12 protein entrapped in this system was able to confer superior protective immunity to protein formulated in IFA against B. abortus 544 (Mallick et al., 2007).

Similarly, the food-grade lactic acid bacterium L. lactis is an attractive delivery vector from a safety point of view. Protective immune responses elicited by the delivery of antigens from HIV (Xin et al., 2003) and human papillomavirus (Bermudez-Humaran et al., 2005) indicate that this is a viable strategy for vaccination. Although L. lactis secreting the Brucella GroEL protein remains untested in mice (Miyoshi et al., 2006), L. lactis producing the Brucella L7/L12 protein is able to confer partial protection against B. abortus challenge (Shi et al., 2006). Overall, these data suggest that this delivery vehicle warrants further investigation.

DNA vaccines

DNA vaccines are able to elicit both humoral and cellular immune responses. However, it is generally perceived that they are able to induce less potent immune responses than protein vaccines. However, this may not be the case. For example, a DNA vaccine expressing Omp31 appears to elicit similar levels of protection as the recombinant Omp31 protein formulated in IFA against both B. melitensis and B. ovis (Cassataro et al., 2005a, b). Furthermore, a BLS-expressing DNA vaccine is a more effective vaccine than the BLS protein against B. abortus (Velikovsky et al., 2002), but similar if the protein is formulated in an adjuvant (Velikovsky et al., 2003). Additionally, differences in the immune response may be generated by these different vaccine strategies, which implies a difference in the mechanism of protection. For example, BLS expressed by a DNA vaccine does not elicit IL-10-producing cells, whereas BLS protein does. This may be relevant because IL-10 can downregulate protective immunity during primary B. abortus infection (Fernandes & Baldwin, 1995). Furthermore, following Omp31 delivery by a DNA vaccine, a weak humoral response and strong CD8+ CTL responses are stimulated. This is in contrast to recombinant protein delivery in which CD4+ Th1 cells mediate protection (Cassataro et al., 2005a).

DNA vaccines may also be used to successfully identify novel Brucella proteins for assessment as vaccine candidates. For example, the invasion protein B (IalB) was highlighted from a panel of four Brucella proteins with unknown immunogenicity as having a protective effect against B. melitensis (Commander et al., 2007). It is the ease of production of DNA vaccines that confers them with additional utility as, not just potential vaccines, but also as tools for the assessment of candidate proteins and the evaluation of immune responses.

Improving DNA vaccination

Strategies to enhance the efficacy of DNA vaccines are constantly emerging in order to maximize immune responses. This is particularly important to aid the transition of these vaccines into larger animal models and, of course, humans. A number of these strategies have been utilized for Brucella antigens.

The expression and subsequent targeting of the expressed antigen from a DNA vaccine can be altered by manipulation of the DNA construct. Changing the promoter from the commonly used CMV promoter to a bovine MHC I promoter (p6) substantially reduced protective efficacy in a DNA vaccine expressing L7/L12 (Kurar & Splitter, 1997). The cellular location of the expressed protein can also be altered by the inclusion of simple targeting sequences or secretion signals. For example, the human tissue plasminogen activator signal sequence was used to secrete GroEL from cells transfected with the GroEL-expressing DNA vaccine (Leclerq et al., 2002). However, this strategy induced lower levels of antibody in immunized mice than nonsecreted GroEL. In a separate study, the Igκ chain signal sequence was used to secrete the SOD protein expressed from a DNA vaccine and assessed against B. abortus challenge (Gonzalez-Smith et al., 2006). This construct was not directly compared with a DNA vaccine in which the SOD protein was not secreted, but it was able to elicit a protective effect, albeit a log lower than the B. abortus 2308 control. Thus, it is not clear whether secretion of a Brucella antigen is the most appropriate strategy for improving the efficacy of a Brucella DNA vaccine. Secretion of a protein from the transfected cell in vivo predominantly improves humoral responses. An alternative strategy in which an intracellular targeting signal, such as ubiquitin, is fused to the Brucella antigen may prove to be more effective. This has been successful for other antigens (Brandsma et al., 2007; Dobano et al., 2007; Ilyinskii et al., 2008), but has not been attempted so far with a Brucella antigen.

Another strategy for improving brucellosis DNA vaccines is to modulate the immune response by the coexpression of cytokines. This can be achieved using different approaches. The cytokine may be fused to the antigen within the DNA vaccine. Both IL-2 (Gonzalez-Smith et al., 2006) and IL-18 (Singha et al., 2008) gene fusions to SOD were expressed from a single DNA vaccine in order to retain the cytokine effect in the local environment to the antigen. In both these cases, inclusion of the cytokine failed to increase protective efficacy over the SOD-expressing DNA vaccines. Alternatively, the gene may be coadministered on a separate plasmid. For example, an IL-12-expressing DNA vaccine was coadministered with a glyceraldehyde-3-phosphate-dehydrogenase enzyme (GAPDH)-expressing DNA vaccine (Rosinha et al., 2002). The GAPDH has been associated with pathogenesis and shown to be a putative vaccine candidate in other infectious disease models such as Schistosoma mansoni (Argiro et al., 2000). Brucella GAPDH expressed from a DNA vaccine was protective in mice if coadministered with the IL-12 DNA vaccine, although the level of protection was lower than the S19 vaccine (Rosinha et al., 2002). Clearly, the use of cytokines for immune modulation of vaccines needs to be considered carefully for safety in humans.

Many of the brucellosis DNA vaccine studies described above rely on intramuscular immunization. However, this approach requires large amounts of DNA and may pose scale-up problems for studies in larger animal models such as humans. Other routes of injection have been tested including the direct injection of a SOD-expressing DNA into the spleen of BALB/c mice (Argiro et al., 2000). However, particle-mediated epidermal delivery (PMED) of DNA vaccines is showing promise in humans (Drape et al., 2006) and requires significantly less plasmid DNA than intramuscular immunization. Although convention suggests that PMED-based DNA vaccination generally elicits Th2-type immunity, the responses vary with the antigen and the species (Payne et al., 2002) and strong Th1 responses can also be produced (Haynes et al., 1996; Nakano et al., 1997; Fuller et al., 2002). Whether PMED-based DNA vaccination will be appropriate for brucellosis remains to be seen. One study that demonstrated no protection afforded by a PMED-based DNA vaccine for brucellosis encoded the GroEL antigen, which has similarly failed to afford protection when delivered as a protein or as an intramuscular DNA vaccine (Leclerq et al., 2002).

Improvement of the delivery of plasmid DNA using live attenuated Y. enterocolitica strains has been evaluated for Brucella antigens. For example, delivery of the O:9 strain encoding P39 or bacterioferritin was able to confer protection against B. abortus 544 challenge (Al-Mariri et al., 2002). However, it was noted that the O:9 control strain of Y. enterocolitica (not expressing Brucella antigens) was able to confer a similar level of protection. Because B. abortus and Y. enterocolitica serotype O:9 share common epitopes on their lipopolysaccharide chains, it is possible that the lipopolysaccharides from the Yersinia makes a significant contribution to this protection. Indeed, the delivery of these antigens from Y. enterocolitica serotype O:3, which has an antigenically unrelated lipopolysaccharides, is unable to induce protection.

Prime-boost vaccination regimens utilize DNA vaccines to prime an immune response to an antigen before recombinant proteins or viral vectors are used as a heterologous boost. This approach has commonly improved responses to DNA vaccines encoding vaccine antigens from other infectious diseases (Leung et al., 2004; Perkins et al., 2006; McConnell et al., 2007), although boosting with the recombinant Omp31 protein only moderately improved the response to an Omp31-expressing DNA vaccine (Cassataro et al., 2007c). This approach may be useful for consideration in future studies.

One of the advantages of DNA vaccination is that multiple antigens can be expressed in vivo with relative ease. A divalent fusion DNA vaccine of the Brucella antigens L7/L12 and Omp16 induced a log-fold greater protection in BALB/c mice than DNA vaccines expressing these proteins alone. Impressively, this protection was equivalent to the live B. abortus RB51 (Luo et al., 2006a). This level of protection was also much greater than immunization with an L7/L12–Omp16 fusion protein, although it should be noted that this was without formulation in an adjuvant. Other studies also show that fusion of L7/L12 and P39 within a DNA vaccine improves protective efficacy when compared with individual DNA vaccines (Luo et al., 2006b) and that coadministration of individual DNA vaccines can also improve protective efficacy. Similarly, expression of the Brucella antigens bp26 and TF from DNA vaccines elicited greater protective efficacy against B. melitensis when administered together rather than alone (Yang et al., 2005). Coadministration of multiple plasmids expressing BCSP31, SOD or L7/L12 strikingly elicited 10–20-fold higher levels of antibody and greater protection against Brucella than the S19 vaccine (Yu et al., 2007). Furthermore, addition of the exposed loop epitope of Omp31 to the N-termini of BLS and expression from a DNA vaccine was able to elicit better and similar protection against B. ovis and B. melitensis, respectively, than live vaccine strain immunization (Cassataro et al., 2007b).

Achieving sufficient immune responses in the mouse model to warrant a transition to a larger species is the challenge facing the development of subunit vaccines for brucellosis. It is encouraging to note therefore that a brucellosis DNA vaccine expressing Cu/Zn SOD, able to elicit a good protective immune response in mice (Onate et al., 2003), can also induce both antibody and specific cell-mediated immune responses in cattle, albeit at a low level (Saez et al., 2008). This encourages the potential of successfully transferring vaccine candidates that show promise in animal models (Table 1) into the clinic for evaluation in humans.


A number of considerations will need to be addressed in developing a brucellosis vaccine for use in humans, including the steps needed to license a product, as well as the concept of use. Vaccine efficacy will need to be clearly demonstrated in at least two animal models. This will likely include one small animal (such as a mouse) and one larger animal (preferably a nonhuman primate). Licensing will also entail a demonstration of its safety, immunogenicity and efficacy in humans. It may be possible to evaluate a human brucellosis vaccine in field trials. However, where it is not possible to clearly demonstrate protective efficacy in humans, a prediction of its efficacy is required, probably the demonstration of stimulation of a surrogate marker of protective efficacy.

While live attenuated vaccines for human brucellosis clearly offer potential advantages in terms of immunogenicity and protective efficacy, the licensing of such a vaccine could prove difficult. Some vaccines consisting of purified recombinant proteins, including those for plague and anthrax (Williamson et al., 2005; Marano et al., 2008), are currently performing well in human clinical trials and thus offer an alternative approach to vaccine development. Subunit vaccines are considered safe, but it is likely that multiple subunits may be needed to confer a high level of protection against brucellosis. Vaccines based on live Brucella may be costly to produce because the production would require high-level containment facilities. A vaccine based on purified proteins would likely require refrigeration and medical professionals to administer. This may not be ideal for some areas of the world where brucellosis is prevalent. In this situation, the use of vector-based vaccine delivery systems such as DNA vaccines or live attenuated vaccines may enable the codelivery of multiple protective subunits, as well as potentially offering the advantages of noninjectable delivery and rapid and cost-effective vaccine production.

Concluding remarks

In summary, subunit vaccines are able to elicit protection against Brucella. Naturally, a combination of the correct antigen and appropriate delivery to the immune system are required. Encouragingly, some antigens are able to protect BALB/c mice (mostly against B. abortus) as effectively as a live control vaccine challenge. These include the 22.9-kDa protein, SOD (delivered as a DNA vaccine, an SFV replicon or a peptide), L7/L12+–Omp16 (DNA vaccine) or Omp31 (protein or DNA vaccine against B. ovis). Other antigens are nonprotective regardless of the delivery system such as the heat shock proteins, which do not induce a protective immune response against Brucella whether delivered as a protein, from a DNA vaccine or from the vaccinia virus, for example. The challenge for these subunit proteins is to transition to larger animal models and to demonstrate efficacy against different species of Brucella. For this, it is likely that further optimization will be required regardless of the delivery system.


  • Editor: Simon Cutting


View Abstract