Animal brucellosis is a disease affecting various do-mestic and wild life species and is caused by an infection with bacteria belonging to the genus Brucella. The genus has several species and each species has a preference for specific animal hosts, for example, B. abortus mainly infects cattle, B. melitensis infects mainly goats and sheep and B. suis infects mainly pigs. These small Gram negative bacteria form a genetically coherent taxon which is related most closely to Ochrobactrum and more distantly to Agrobacterium and Rhizobium within the alpha-2 sub-group of the Proteobacteriaceae.

Most of the Brucella nomen species are transmissible to humans where it can cause serious acute and chronic disease with some cases resulting in death, making brucellosis an important zoonosis. Brucella is also a facultative intracellular parasite; the pathogenesis of brucellosis and the nature of the protective immune response are closely related to this property (^{Cheers 1984}).

Because of the serious economic and medical conse-quences of brucellosis for both cattle farmers and humans in general, efforts have been made to prevent the infection through the use of vaccines (^{Nicoletti 1990}). These were initially developed on an empirical basis, but with our current ability to manipulate the genome of the bacteria, more rational designs are being used for the development of better vaccines against the disease.

The lipopolysaccharide (LPS) molecule is closely associated with the phenotype of Brucella colonies in culture. Brucella can present itself upon culture with either a "smooth" or a "rough" colony morphology, with some strains presenting a "mucoid" phenotype (^{White and Wilson 1951}). It is possible for smooth colonies to become rough spontaneously and some rough Brucella strains may revert to the smooth morphology making the degree of stability an important consideration in vaccine development. Virulence of B. abortus is associated with smooth morphology and therefore, B. abortus field strains producing acute and chronic brucellosis in cattle (and humans) are smooth. Smooth organisms have an LPS molecules containing a polysaccharide O-chain made from a homopolymer of perosamine (N-formyl-4-amino,4,6-dideoxy mannose), while rough organisms lack this O-chain on their LPS molecule; some strains possess only a greatly truncated portion of it (^{Caroff et al 1984}, ^{Moreno 1984}). The O-chain of smooth Brucella is located on the surface of the organ-ism. Essentially all animals infected with smooth strains respond immunologically to this O-chain by making an-tibodies against it. This is a very important point because, all serological tests used in the diagnosis of infected cattle are based on the detection of O-chain antibodies (^{Diaz R et al 1968}, ^{Nielsen et al 1988}, ^{Vemulapalli et al 2000}). Also, it is very important to be aware that antibodies against the O-chain do not protect cattle against infection with B. abortus and in general, a clear role of any other Brucella antibodies in protection against cattle brucellosis has not been demonstrated. An effective immune response against Brucella infection in cattle requires a strong Cell Mediated Immune (CMI) response based on both, active T-helper1 and T-cytotoxic cells (^{Schurig et al 2002}). The current understanding is that protection in cattle is highly dependent on the induction of these specific T lymphocytes and the concurrent activation of macrophages by T-helper1 cells secreting interferon-gamma (Schurig et al 2002,Vemulapalli et al 2000^c).

The vaccine that has been most widely used to prevent bovine brucellosis is B. abortus strain 19 (^{Nicoletti 1990}).

This strain was first described in 1930 and was originally isolated from the milk of a Jersey cow as a virulent strain in 1923, but after being kept in the laboratory at room temperature for over a year, was found to have become attenuated (^{Buck 1930}). This strain was able to induce a certain level of protective immunity in cattle. Effectiveness fluctuated depending on a variety of variables including age of vaccination, dose, route and prevalence of brucellosis in vaccinated herds (^{Nicoletti 1990}). Strain 19 is an attenuated organism with a smooth morphology normally unable to grow in the presence of erythritol (^{Jones et al 1965}). Although strain 19 is of low virulence for cattle, vaccination of pregnant cows can result in abortions. This event is rather rare, ranging from less than 1 up to 2.5% under field conditions and between 10-12% under experimentally controlled conditions (^{Mingle et al 1941}, ^{Moore 1950}, ^{Manthei 1952}, ^{Beckett and Mc Diarmid 1985}, ^{Tabynov et al 2016}). Intravenous injections of pregnant cattle with strain 19 induces 100% abortions, while RB51 did induce up to 25% abortion (^{Palmer et al1996}) .

Strain 19 is pathogenic for humans and can lead to chronic infections if cases are not treated with appropriate antibiotics (^{Revich et al 1961}, ^{Wallach et al 2008}). The most common route of human infections with strain 19 is accidental inoculation during vaccination of cattle. Strain 19 can also be found in the milk of cattle if vaccination is carried out during adulthood (^{Samartino et al 2000}).

Since strain 19 is a smooth species of Brucella, the presence of LPS with O-chain on its surface is the reason for the appearance and persistence of O-chain antibodies in serum following administration of this vaccine (^{Diaz et al 1968}, ^{Nielsen et al 1988}). Since these antibodies are the basis of all diagnostic tests, strain 19 vaccination prevents easy differentiation of vaccinated from infected animals and therefore, delays eradication efforts and leads to over condemnation of cattle. In order to decrease this problem to some extent, calves are vaccinated rather early in life since their antibody response to Brucella is weaker at that time. The problem becomes worse if calves are vaccinated for the first time later in life or are re-vaccinated for booster purposes since antibody levels will increase even more and be very persistent (^{Manthei 1952}, ^{Samartino et al 2000}).

Several undesirable characteristics of strain 19, particularly its confusing effect on serodiagnosis, led to the development of B. abortus strain RB51 vaccine (^{Schurig et al 1991}). Strain RB51 is a highly attenuated mutant of B. abortus strain 2308. It is a very stable strain meaning that it does not change colony morphology upon multiple passages in culture or animals (Schurig et al 1991, ^{Colby 1997}) . Strain RB51 has a rough colony morphology and is essentially devoid of O-chain (Schurig et al 1991). The wboA gene of strain RB51, a gene involved in the biosynthesis of the O-chain, is interrupted by an IS711 element impeding the production of O-chain (^{Vemulapalli et al 1996}, 2000^a). Since strain RB51 lacks O-chain, vaccinated cattle do not respond to the O-chain and do not develop antibodies to this antigen. Therefore, vaccinated animals remain serologically negative in all serological diagnostic tests. Furthermore, re-vaccination at any age does not induce O-chain antibodies in the animals allowing the application of "booster" vaccinations without affecting serological diagnostic tests necessary for disease eradi-cation (^{Samartino et al 2000}, ^{Dorneles et al 2015}). When used in cattle, one vaccination with strain RB51 induces protection levels similar to those induced by strain 19 and protection of the vaccinated animals can vary from 65% to 100% depending on the conditions prevailing in a specific herd or experiment (Schurig et al 2002). Revaccination with strain RB51 for a second time appears to increase immunity and represents an additional advantage of RB51 over strain 19. Since 1996, millions of cattle have been immunised with strain RB51 and reversion to a virulent form has never been observed testifying to its extreme stability.

As mentioned before, CMI responses play a critical role in resistance against intracellular bacterial infections (^{Schurig et al 2002}, ^{Vemulapalli et al 2000c}). It is therefore critical that vaccines are presented to the immune system in a way that they induce a T cell response that can protect against the infection. Live bacterial vaccines are considered essential to induce an appropriate T cell mediated CMI protective response. Although the exact reasons for needing live organisms to induce the right response are debatable and are probably multiple, the synthesis of antigens by the live organisms during the immune response induction phase appears critical. Replication of the vaccine strain may be less critical than synthesis of new antigens since irradiated vaccines, where the bacteria do not replicate but are able to synthesize antigens after the irradiated vaccine application, are protective. Irradiated Brucella are able to induce protection through CMI responses while killed organisms are not (Montaraz and Winter 1968, ^{Magnani et al 2009}, ^{Moustafa et al 2011}). Since live bacteria are able to induce protective CMI responses, attenuated strains of bacteria are often used as live vaccines to protect against intracellular bacterial infections. Both B. abortus strain RB51 and strain 19 are live, attenuated vaccines able to induce protection while the same vaccines rendered metabolically inactive are not unless potent adjuvants are used (Montaraz and Winter 1968). This observation has important practical implications since live, attenuated vaccines have to be handled carefully (kept cold, use only shortly after reconstituting in the field, etc) to maintain their effectiveness.

In many cases live, attenuated vaccines do not provide high levels of protection, particularly if animals are confronted with high numbers of infectious bacteria (^{Manthei CA 1968}). For example, a Brucella vaccinated cow will have a very high likelihood of being well protected against infection if she is exposed to a low number of field B. abortus bacteria (Manthei CA 1968). Not only does level of exposure affects the protective ability of the vaccine, vaccine dose will also affect the protective outcome (Manthei CA1968, ^{Confer et al 1985}). In contrast, a cow may not be protected when exposed to the placenta and fetus of a B. abortus abortion where Brucella number in the hundreds of billions. Therefore, eradication of brucellosis not only depends on vaccination, it depends on a sustained vacci-nation program and good management where vaccinated animals are separated as much as possible from high level infection sources and where seropositive animals are consistently removed from the herd.

The fact that full protection in cattle is probably not achieved under most field conditions, made it relevant to work on approaches that could improve effectiveness of the current strain RB51 vaccine without changing its positive characteristics of being highly attenuated and leaving animals seronegative after one or multiple vaccinations. We hypothesized, some time ago, that over-expression (production of more than normal quantities) of a Brucella protective antigen by vaccine strain RB51 would result in enhancement of the vaccine's efficacy (^{Vemulapalli et al 2000b}, Vemulapalli et al 2002). Our studies demonstrated that we could over-express B. abortus Cu/Zn superoxide dismutase (SOD) protein in vaccine strain RB51(Vemulapalli et al 2002). SOD is considered to be one (^{He et al 2002}) of probably many protective Brucella antigens and significantly increases the vaccine's protective capabilities in the murine model of brucellosis without altering the attenuation, stability or serological characteristics of the vaccine. This stim-ulated our interest in pursuing this approach to produce a more effective RB51 vaccine. Interestingly, even though homologous over-expression of Cu/Zn superoxide dismutase SOD enhanced protection significantly in mice, it did not enhanced protection against B. abortus infection in bison and elk (^{Olsen et al 2009}, ^{Nol et al 2016}). In bison, the overexpressing RB51-SOD vaccine was actually less efficacious and was cleared faster from the vaccinated animals than the parenteral RB51 strain (Olson et al 2009). Faster clearance may have resulted in a weaker immune response and may explain the de-crease in efficacy. This indicates that the murine model is not always an indicator of what may happen in cattle or other animal species making protection experiments in cattle crucial before a new vaccine can be called more effective in the target species.

Since homologous over-expression of a protective antigen can lead to a more effective vaccine, it was logical to think that expression by strain RB51 of a protective antigen derived from an unrelated infectious agent, for which protection is mediated by a CMI response, would protect against the unrelated disease as well as against infection with Brucella. Consistent with this hypothesis, protection against infection with Neospora caninum was demonstrated in the mouse model using a strain RB51 strain expressing parasite protective antigens (^{Rajasekaran et al 2008}). Therefore, it may be possible to create strain RB51vaccines able to protect cattle against Brucella infection as well as protect against unrelated diseases si-multaneously. Our approach to create a strain RB51 vaccine able to protect against Brucella and Mycobacterium bovis infections simultaneously is described below.

In order to use strain RB51 as a platform vaccine able to induce specific immune responses against a variety of homologous and/or heterologous antigens, a plasmid containing the gene encoding the foreign antigen along with an antibiotic resistance gene (as a plasmid marker) had been employed during our developmental work carried out in the past (^{McQuiston et al 1995}, ^{Vemulapalli et al 2000c}, Vemulapalli et al 2002). Because Brucella spp. have no naturally occurring plasmids, we adapted a broad host range plasmid that had been shown to replicate in Brucella (^{Kovach et al 1994}). We eliminated extraneous plasmid DNA sequences and introduced a variety of promoters to allow for expression of genes derived from Brucella (homologous overexpression) as well as derived from other bacterial species (heterologous overexpression) and the resulting plasmid was designated as pNS or pLeuB (^{Seleem et al 2004}, see figure 1).

The approach of using an antibiotic resistance gene as a selection marker has been criticized as not being environmentally safe because in the vaccinated animals, it has the potential to introduce the antibiotic resistance gene into normal and pathogenic flora. Therefore, expressing the protective antigens from a plasmid that is not dependent on an antibiotic resistance gene for selection of the desired strain, would be an advantage, as it would have a minimum environmental risk.

The leuB gene, encoding isopropyl malate dehydroge-nase, is one of the four genes essential for the biosynthesis of leucine in B. abortus (^{Essenberg et al 1993}). B. abortus is known to survive and replicate in nutrient-limited environments inside a host. Therefore, a B. abortus mutant lacking a gene for the biosynthesis of an essential amino acid like leucine is unlikely to survive in that environment including being able to replicate in minimal media (^{Essenberg et al 1993}, ^{Bacon et al 1951}, ^{Bange et al 1996}). The complementation of such an auxotroph with a plasmid carrying the wild-type leuB gene (encoding the enzyme necessary for the amino acid synthesis) would provide a means of selection and maintenance of this plasmid in B. abortus under minimum media conditions. Thus, we produced a leuB mutant in strain RB51 using allelic exchange (^{Rajasekaran et al 2008}). The resultant leuB auxotroph (RB51leuB) cannot grow in leucine-deficient conditions but when complemented with a plasmid carrying the wild-type B. abortus leuB gene, the leucine deficiency of RB51leuB is eliminated (Rajasekaran et al 2008). Thus, the complemented RB51leuB strain can be used to over-express homologous and/or heterologous antigens and eliminate the environmental concerns related to antibiotic resistance. Importantly, the complemented RB51leuB strain retained the basic characteristics of the original strain RB51 which are attenuation, no induction of O-chain antibodies and protection against B. abortus infection.

Figure 1 Schematic of pLeuB plasmid containing the leuB gene of B. abortus; a multiple cloning site (MCS) and an origin of replication (Ori). The trcD promoter is a very strong promoter as it is a hybrid between the lacZ and tryptophan promoter. The signal sequence (SigSeq) is from the SOD gene and followed by an ATG start codon and an in-frame histidine tag (6XHis) followed by a protective antigen from BP26 of B. abortus. 

Using strain RB51 leuB as a platform and pNSLeuB, an antibiotic-resistance marker free plasmid, we constructed several strains over-expressing homologous antigens. For example, strains RB51leuB/SOD (superoxide dismutase), RB51leuB/SOD/L7/L12 (ribosomal proteins) and RB51leuB/ SOD/WboA (glycosyl transferase) were constructed to over-express the selected Brucella antigens: SOD alone, SOD and ribosomal protein L7/L12 or SOD and glycosyl transferase, respectively. The ability of these vaccine candidates to protect against a virulent B. suis challenge were evaluated in a mouse model. All vaccine groups protected mice significantly (P<0.05) when compared to the control group. Within the vaccine groups, the mice vaccinated with strain RB51leuB/SOD/WboA were significantly better protected than those that were vaccinated with either strain RB51leuB/SOD or RB51leuB/SOD/L7/L12. These results suggest that Brucella antigens can be over-expressed in strain RB51leuB and can elicit enhanced protective immune responses against brucellosis at least in the murine model of the disease (^{Rajasekaran et al 2011}). The murine model for Brucellosis has been used by numerous authors to test protective abilities of vaccines (Montaraz and Winter 1968). Nevertheless, it is not clear if protective abilities observed in mice translate into protective abilities in cattle or other species. For example, work carried out with strain RB51 in mice indicating protection by strain RB51 was later demonstrated in cattle (^{Schurig et al 1991}, ^{Cheville et al 1993}). On the other hand, RB51 overexpressing Cu/Zn superoxide dismutase showed higher protection than RB51 in mice but did not protect bison against Brucella challenge (^{Vemulapalli et al 2002}, Olson et al 2009). It remains to be seen if similar results can be obtained in cattle.

Control of bovine tuberculosis is an important ele-ment in the control or eradication of human tuberculosis. Unfortunately, there is no practical vaccine available for the control of tuberculosis in cattle even so, the problem is very significant. To illustrate the dimension of the problem, long term data analysis of bovine tuberculosis in India suggests that approximately 7.3% of cattle are potentially infected. This means that there are 21.8 million TB infected cattle, which amounts to more than all cattle in United States. This systematic analysis is even more alarming as approximately 10% of human tuberculosis is due to M. bovis infection. Therefore, control of bovine tuberculosis is an important component related to efforts of human tuberculosis eradication, especially in the de-veloping world, as test and slaughter may not be an option due to economic and religious considerations (^{Srinivasan et al 2018}).

Bacille Calmette Guerin (BCG) vaccine is the current TB vaccine used in humans. It is an attenuated strain of M. bovis (^{Behr et al 1997}, ^{Murray 2004}). BCG was attenuated by serially passing a virulent M. bovis strain on oxbile medium for 230 times in the laboratory. This process led to attenuation of the virulent M. bovis strain as indicated by self-limiting infection as well as partial resistance to reinfection with M. bovis (^{Takamura et al 2005}). One of the serious disadvantages of BCG vaccine is that vaccinated humans and animals turn skin test pos-itive. Therefore, BCG is rarely used in cattle to protect against bovine tuberculosis (^{Waters et al 2012}). Recent, largescale field trials with cattle using very low doses of BCG demonstrated high levels of protection (vaccine efficacy between 85.7%-86.7%) without affecting their ante mortem skin or blood tests (^{Nugent et al 2018}). This low dose approach may make vaccination of cattle with BCG more acceptable than in the past.

Following our previously defined strategy of over-expressing heterologous antigens in strain RB51, we speculated that expression of protective antigens from M. bovis in B. abortus strain RB51, a vaccine strain that is USDA approved and extensively used throughout the world, could make an effective dual vaccine able to protect against Brucellosis and Tuberculosis in cattle simultaneously. To prove this point, we selected the following protective Mycobacterium antigens: 85B, ESAT6 and Rv2660c for this purpose (Al Qublan, Dissertation, Virginia Tech, 2014). These genes were cloned and expressed in our auxotrophic leuB deletion mutant of strain RB51 (RB51leuB). As explained before, the advantage of using the RB51leuB deletion mutant is that one can have a plasmid encoding the protective antigens along with the leuB gene instead of a drug resistance gene for selection purposes

The reasons for selecting these antigens are outlined by characteristic that define them as protective antigens.