Recombinant BCG as a vaccine vehicle to protect against tuberculosis

Abstract

Mycobacterium bovis Bacille Calmette Guérin (BCG) was first administered to humans in 1921 and has subsequently been delivered to an estimated 3 billion individuals, with a low incidence of serious complications. The vaccine is immunogenic and is stable and cheap to produce. Additionally, the vaccine can be engineered to express foreign molecules in a functional form, and this has driven the development of BCG as a recombinant vector to protect against infectious diseases and malignancies such as cancer. However, it is now clear that the existing BCG vaccine has proved insufficient to control the spread of tuberculosis, and a major focus of tuberculosis vaccine development programs is the construction and testing of modified forms of BCG. This review summarizes the strategies employed to develop recombinant forms of BCG and describes the potential of these vaccines to stimulate protective immunity and protect against Mycobacterium tuberculosis infection.

Bacille Calmette Guérin (BCG) was developed at the Institut Pasteur by Albert Calmette and Camille Guérin and first administered to humans in 1921.¹ The strain was developed by continual passaging of the bovine tuberculosis strain, Mycobacterium bovis, in media containing potato bile, which resulted in an attenuated vaccine strain with reduced virulence in animals.² The vaccine was initially administered to newborns as an oral vaccine of three doses, however current practice is to deliver a single dose intradermally. The early work of Calmette demonstrated that the vaccine was safe for use in humans and while its use was initially restricted to France, international distribution and re-appropriation of the vaccine started in 1927.³ This redistribution and local preparation of vaccine led to the generation of daughter strains of BCG, which can be distinguished based on genomic comparisons.⁴ Analysis of randomized controlled trials and case-control studies have shown good efficacy of BCG vaccination against severe childhood forms of tuberculosis, however its efficacy against adult pulmonary disease is variable.^5,6 It is yet to be established if the genetic and phenotypic variation in BCG strains impacts on their capacity to protect against tuberculosis. The variable efficacy of BCG, coupled with the inability of current control strategies to reduce tuberculosis incidence,⁷ have driven efforts to develop more effective vaccines for tuberculosis. One strategy is to maintain use of BCG yet modify the vaccine to improve its effectiveness, as the vaccine does possess qualities that make it suitable as a vaccine vehicle; it can invoke some level of protective immunity (particularly in children), the vaccine is well tolerated and does possess potent immunostimulatory capacity which has been applied to the control of other conditions such as bladder cancer.⁸ This has maintained interest in the development of recombinant BCG strains for the control of tuberculosis, which is the focus of this review.

The Birth of Recombinant BCG

BCG possesses a number of characteristics that, compared to commonly used bacterial hosts such as Escherichia coli, render it relatively difficult to manipulate genetically. BCG is naturally resistant to a variety of commonly used antibiotics, replicates slowly (doubling time of approximately 20 hours), is not readily transformed with DNA, and early attempts at allelic exchange in slow-growing mycobacteria resulted in a high degree of illegitimate recombination.⁹ In addition, differences in promoter recognition, replication machinery and codon usage has required the use of specific mycobacterial vectors to effectively express recombinant proteins in BCG. Many of these issues have been systemically addressed over the past two decades. Shuttle vectors have been developed that permit the exchange of plasmids between E. coli and mycobacteria^10,11 and these vectors have been improved to allow high-level secretion of recombinant proteins,¹² or inducible systems used to permit regulated gene expression.^13–15 Expression of foreign antigens in BCG was first demonstrated by using the secreted Ag85B protein of mycobacteria as a carrier for a B cell epitope from the HIV p17^gag protein, with the recombinant protein being detected by immunoblotting in the supernatant of recombinant (r) BCG cultures.¹⁶ In 1991, three groups reported the development of rBCG strains that were capable of inducing antigen-specific immune responses against the encoded antigens.^17–19 In the report of Stover et al. both multicopy and integrative plasmids were developed and used to express a number of antigens under the control of promoter elements from BCG heat shock proteins, and these vectors form the basis of many of the expression systems used today.¹⁷ While heat shock protein promoters have been widely used, other systems have proven useful, such as the M. paratuberculosis pAN promoter used in the early studies of Winter et al.¹⁹ and the M. fortuitum pBlaF* promoter, which permits stable, high-level expression of recombinant genes in mycobacteria.²⁰

The development of systems for foreign antigen expression in BCG has permitted the construction of rBCG strains modified to induce protective immunity in models of experimental infection. Most of the early studies were focused on expressing in BCG well-defined protective antigens from pathogens other than mycobacteria. These early studies are reviewed in detail by Ohara and Yamada,²¹ and will not be further described here. For mycobacteria the hierarchy of protective antigens is less well defined, particularly as BCG is known to express major protective antigens that are shared with M. tuberculosis, such as the Ag85 complex.²² However, the pressing need for development of new vaccines to contain the spread of tuberculosis has resulted in a number of strategies being employed to modify BCG to improve it's effectiveness, which are described below.

Expanding the Antigenic Repertoire of BCG

The Ag85 complex, comprising of Ag85A, Ag85B and Ag85C, are major protective antigens shared by both BCG and M. tuberculosis.²² These antigens are protective in animal models of tuberculosis and are strongly recognised by M. tuberculosis-infected individuals.^23,24 The first reports examining the efficacy of BCG overexpressing mycobacterial antigens exploited the antigenicity of members of the Ag85 complex. Ohara et al. demonstrated that BCG overexpressing Ag85A was able to limit the replication of Mycobacterium leprae in a mouse footpad model of infection.²⁵ Horwtiz et al. reported that Ag85B overexpression by BCG afforded significant reduction in bacterial loads²⁶ and improved survival of M. tuberculosis-challenged guinea pigs compared to BCG alone.²⁷ This vaccine has undergone phase I testing in humans and in initial reports resulted in enhanced anti-BCG immunity in vaccinated individuals.²⁸ Other rBCG strains overexpressing members of the Ag85 complex have been developed and assessed for their protective efficacy. An Ag85A-overexpressing BCG strain resulted in a reduction in lung pathology and bacterial load after M. tuberculosis challenge in both rhesus and cynomolgus monkeys compared to BCG alone.^29,30 Similarly, lung pathology and bacterial load in M. tuberculosis-challenged guinea pigs was reduced after vaccination with BCG overexpressing Ag85C, with protection associated with reduced levels of pro-inflammatory cytokines in the lung.³¹ Despite the success of expression of individual components of the Ag85 complex in BCG, a vaccine over expressing all three components of the complex has yet to be developed; such a vaccine may serve to expand the breadth of BCG-reactive T cells generated after vaccination. Combination of other antigens have been tried however. BCG co-expressing Ag85B, Mpt64 and Mtb8.4 induced a similar reduction in bacterial load after M. tuberculosis infection of mice compared to BCG alone, however mice were significantly protected against infection-induced weight loss.³²

The attenuation of BCG is due predominately to the deletion of DNA fragments from M. bovis during development of the vaccine, termed regions of difference (RD), which has resulted in the current version of the vaccine strain.^33–35 There has been much effort in determining which deletions are responsible for the attenuated phenotype of BCG, and if such information can be exploited to develop recombinant vaccines with improved efficacy. The M. tuberculosis ESAT-6 and CFP10 antigens are located in RD1 and are therefore absent from BCG; the role of these two proteins in virulence has been investigated in detail.³⁶ Re-introduction of RD1 into BCG results in a persistent BCG strain that improves protection against disseminated M. tuberculosis infection in mice and guinea pigs.³⁷ It is of interest to note that BCG secreting ESAT-6 alone did not show increased virulence or improved protective efficacy in mice compared to conventional BCG.³⁸ While this may be due to differences in ESAT-6 export efficiency between the strains, it does suggests that factors other than ESAT-6 secretion may be responsible for the improved protective effect of BCG:RD1, such as the persistence of the strain³⁹ and/or additional antigens expressed by the recombinant vaccine. Similarly, BCG, which overexpressed ESAT-6 and Ag85B only showed an improved protective effect in mice when the two antigens were expressed as a fusion protein, as BCG expressing ESAT-6 or Ag85B alone did not result in enhanced protective efficacy.⁴⁰

The genealogy of BCG strains is complex and subtle genetic differences, such as single nucleotide polymorphisms, exist between BCG strains and parental M. bovis, which may be responsible for differences in virulence and antigenicity.⁴¹ For example, BCG is unable to utilize L-alanine and this deficiency is thought to inhibit the growth of the vaccine in vivo and limit vaccine efficacy.⁴² Loss of this function is due to a point mutation in the BCG gene encoding L-alanine dehydrogenase (Ald) and L-alanine catabolism can be conferred on BCG by introduction of the Ald gene from M. tuberculosis.⁴³ Restoration of Ald activity however did not alter the in vivo persistence of BCG in mice, and protection against aerosol M. tuberculosis infection was not altered by addition of ald to the BCG vaccine.⁴³ None-the-less, it is clear that small genetic changes alter mycobacterial phenotype,^42,44,45 and such changes may be exploited for the future development of recombinant BCG vaccines.

An additional feature of BCG vaccination is that certain antigens are not recognized by immunized-individuals even though the vaccine can express the proteins. One example is the HspX latency antigen, which is recognized by T cells from M. tuberculosis-infected individuals, particularly those classified to have latent M. tuberculosis infection.⁴⁶ Although both M. tuberculosis and BCG express a functional form of HspX, BCG-vacinees do not recognize the protein.^47,48 Engineering of BCG to constitutively express HspX restores immune recognition of the protein in mice, however the protective effect against M. tuberculosis was not improved.⁴⁹ As other ‘dormancy antigens’ are recognized during latent M. tuberculosis infection,⁵⁰ a possible vaccine strategy is the coordinate overexpression of multiple dormancy antigens by BCG to improve protective immunity, especially against latent disease.

Turning on the Immune System

An alternative approach to improve BCG efficacy is to add immuno stimulatory components to the vaccine in order to manipulate the response at the time of vaccination, thus prolonging the quality and longevity of memory T cell responses (Fig. 1). One of the most intriguing aspects of BCG as a recombinant vaccine vehicle is that the bacterium can produce mammalian molecules in a functional form, which may relate to the capacity of BCG to permit the glycosylation and methylation of recombinant products.^51,52 Most reports have focused on the development of BCG secreting mammalian cytokines and related molecules, and a number of different functional categories have been expressed in BCG. These include pro-inflammatory cytokines such as TNF,⁵³ members of the interferon class such as IFNγ and IFNα2B,^53,54 lymphoid stimulatory cytokines including IL-2 and IL-18,^53,55,56 and myeloid growth factors such as GM-CSF and Flt3 ligand.^53,57 In essentially all cases, BCG secreting these molecules have resulted in increased generation of BCG-reactive T cell responses after delivery to mice or improved release of cytokines from human cells.^{53,54,56–58} These strains have also proven useful to investigate immune processes, such as the role of IFNγ and TNF in the control of mycobacterial growth^59,60 and the influence on TNF levels on mycobacterial infection.⁶¹ BCG secreting the chemokine MCP3 has also been constructed, which was able to improve lymphocyte migration and augment antigen-specific T cell responses in mice.⁶²

Figure 1.

An overview of the immune processes targeted by rBCG strain to improve anti-mycobacterial immunity. Professional antigen presenting cells (APC), such as dendritic cells, can present peptide fragments to CD8⁺ and CD4⁺ T cells, and induction of autophagy can improve this process (BCG:Ag85B). T cells can be expanded in the presence of IL-2 (BCG:IL-2), and numbers of antigen-specific cells can be increased by overexpression of protective antigens in BCG, such as Ag85B (BCG:Ag85B). APC-derived cytokines such as IL-12 and IL-18 (BCG:IL-18) can enhance the expansion of ‘Th1-like’ T cells secreting IFNγ and TNF, which can then act on infected host cells to aid elimination of ingested mycobacteria (BCG:IFNγ, BCG:TNF). The generation and maintenance of antigen-specific memory T cells can be improved by cytokines such as IL-15 (BCG-Ag85B-IL-15). Growth factors including GM-CSF and Flt3L can induce differentiation and activation of APCs and host cells such as macrophages to improve antigen presentation and aid destruction of internalized bacteria (BCG:GM-CSF, BCGFlt3L), while increased apoptosis of host cells can facilitate antigen transfer to APC and improve T cell presentation (BCG:LLO). Many of the cytokines and cell types described have additional roles/functions which are not depicted here. This schematic is only a representation of the vaccine strategies employed, a more comprehensive outline is provided in the text.

Although many of these ‘immunostimulatory’ BCG vaccines improve anti-BCG immune responses in animal models, for the most part these strains have not improved the protective effect of the vaccine. For example, mice vaccinated with BCG secreting either IL-2, IL-18, TNF or IFNγ induce strong ‘Th1-like’ immunity yet don't improve the protective effect of BCG against M. tuberculosis infection^63,64 (Ryan and Triccas, unpublished data). Moreover, when used as immunotherapeutic vaccines, a subset of these rBCG strains can actually exacerbate disease in M. tuberculosis infected animals.⁶⁴ This may suggest that the direct stimulation of T cells (IL-2 and IL-18) or targeting macrophage activation (TNF and IFNγ) may not be ideal strategies to improve anti-mycobacterial protective immunity, at least in the mouse models used for these studies. However, improving the activation and expansion of dendritic cells by use of BCG secreting the GM-CSF resulted in improved protection against disseminated M. tuberculosis infection compared to parental BCG.⁵⁸ BCG secreting a fusion of Ag85B and IL-15, a cytokine important for the maintenance of memory CD8⁺ T cells,⁶⁵ resulted in increased generation of memory T cells and improved protection in the lungs of vaccinated mice.⁶⁶ BCG secreting either MCP-3 or Flt3L afforded protective efficacy similar to BCG alone, however both strains are significantly less virulent in immunodeficient mice compared to BCG alone.^57,62 The reduced virulence is most likely related to increased clearance of these strains within the host, and such strains could be useful in areas where there is a high prevalence of immunodeficiency.

Targeting other immune processes through the use of recombinant BCG has also provided protective vaccines. BCG engineered to secrete listeriolysin (LLO), a pore-forming cytolysin that facilitates the entry of Listeria monocytogenes into the cytoplasm of cells, increased host cell apoptosis and improved protection against aerosol M. tuberculosis infection.⁶⁷ Importantly, this strain (BCG:LLO) can protect mice against a hyper-virulent M. tuberculosis strain, which is not controlled by vaccination with BCG alone. It has also recently been shown that vaccination with BCG overexpressing Ag85B could augment protective immunity in mice by stimulating autophagy of antigen presenting cells,⁶⁸ a process used by the cell to facilitate antigen presentation and suppress growth of intracellular bacteria.⁶⁹

Taking all the Best Bits: Combination Vaccines

Efforts have been made to improve the efficacy of recombinant BCG vaccines by using the strains to target multiple immune processes, or combine existing vaccines with other leading vaccine candidates. A BCG strain expressing three immunodominant antigens (Ag85A, Ag85B, TB10.4) together with the cytolysin perfringolysin O from Clostridium perfringens improved immune responses in mice and resulted in improved survival of mice challenged with the hyper-virulent HN878 strain of M. tuberculosis.⁷⁰ Vaccination of rhesus macaques with the same rBCG vaccine and subsequently boosted with an adenovirus vector expressing a fusion of Ag85A, Ag85B and TB10.4 resulted in increased IFNγ and antigen-specific T cell proliferation compared to BCG alone, however the protective effect of this vaccination regimen has yet to be assessed.⁷¹ BCG:LLO has also been used in a prime-boost approach, in combination with a modified vaccinia virus expressing M. tuberculosis Ag85A (MVA-85A), which is the most well developed TB vaccine currently in human trials.⁷² Despite the fact that MVA-85A boosting led to significant increases in Ag85A-specific T cell responses, this approach did not result in improved efficacy compared to BCG:LLO alone.⁷³ This highlights the need for better correlates of protective immunity induced by candidate tuberculosis vaccine strains.

Concluding Remarks

The favourable safety profile of BCG and its efficacy in protecting children against tuberculosis supports retention of BCG in current immunization schedules. To date, BCG overexpressing Ag85B is the only rBCG strain for which clinical data from assessment in humans has been reported.²⁸ If a recombinant form of BCG is to replace the current vaccine will require entry of a sufficient number of vaccines into controlled field trials to select the best candidate, as it is possible that for some candidates the promising efficacy and safety seen in animal models may not be transferred to human populations. It is also evident that BCG-induced immunity is not life-long, a finding highlighted in field trials and even identified by Calmette himself in early animal studies,^3,74 and it is unknown if modified BCG vaccines will be capable of significantly extending the duration of anti-BCG immunity. This suggests that any new BCG vaccine candidate may need to be delivered in combination with boosting agents to stimulate prolonged protective immunity. Of those tuberculosis vaccines in clinical trials, the majority are designed as boosting vaccines, and include fusion proteins in selected adjuvants (e.g., Ag85B-ESAT-6, Mtb72f) or viral vectors expressing dominant antigens (e.g., MVA-Ag85A, Aeras-402-Ad35).⁷⁵ Therefore, both rBCG strains and suitable boosting vaccines should be developed and assessed in parallel to identity the most effective vaccine regimen to protect against tuberculosis.