key: cord-342847-sjcanw19
authors: Leung, Andrea S; Tran, Vanessa; Wu, Zuowei; Yu, Xuping; Alexander, David C; Gao, George Fu; Zhu, Baoli; Liu, Jun
title: Novel genome polymorphisms in BCG vaccine strains and impact on efficacy
date: 2008-09-15
journal: BMC Genomics
DOI: 10.1186/1471-2164-9-413
sha: 
doc_id: 342847
cord_uid: sjcanw19

Bacille Calmette-Guérin (BCG) is an attenuated strain of Mycobacterium bovis currently used as a vaccine against tuberculosis. Global distribution and propagation of BCG has contributed to the in vitro evolution of the vaccine strain and is thought to partially account for the different outcomes of BCG vaccine trials. Previous efforts by several molecular techniques effectively identified large sequence polymorphisms among BCG daughter strains, but lacked the resolution to identify smaller changes. In this study, we have used a NimbleGen tiling array for whole genome comparison of 13 BCG strains. Using this approach, in tandem with DNA resequencing, we have identified six novel large sequence polymorphisms including four deletions and two duplications in specific BCG strains. Moreover, we have uncovered various polymorphisms in the phoP-phoR locus. Importantly, these polymorphisms affect genes encoding established virulence factors including cell wall complex lipids, ESX secretion systems, and the PhoP-PhoR two-component system. Our study demonstrates that major virulence factors are different among BCG strains, which provide molecular mechanisms for important vaccine phenotypes including adverse effect profile, tuberculin reactivity and protective efficacy. These findings have important implications for the development of a new generation of vaccines.

Bacille Calmette-Guérin (BCG) is an attenuated strain of Mycobacterium bovis and is the only available vaccine against tuberculosis (TB). Since 1974, BCG vaccination has been included in the World Health Organization (WHO) Expanded Program on Immunization. It is estimated that more than 3 billion individuals have been immunized with BCG and over 100 million doses of BCG are administered annually. Multiple studies have confirmed that BCG is generally safe and can protect children against disseminated disease, including tuberculosis meningitis [1, 2] . BCG also provides cross-protection against leprosy [3] . However, the success of BCG against pulmonary TB in adults is still debated, since randomized clini-cal trials have reported protection efficacy ranging from 0-80% [4, 5] . Several hypotheses for the variation in observed efficacy have been proposed [6] [7] [8] [9] .

One explanation concerns the heterogeneity of the BCG strains [6] . The original BCG was derived from a virulent strain of M. bovis isolated from a cow. From 1908 through 1921, this isolate was subjected to 230 passages on glycerinated potato bile medium, which generated an attenuated strain termed BCG [10] . Distribution and widespread use of BCG started around 1924 and was accompanied by changes in the manufacturing process in production facilities. For instance, while BCG in Sweden was transferred without interruption from bile potato to bile potato medium in accordance with Calmette's original practice [11] , BCG production in Denmark involved alternating rounds of growth on potato bile medium and Sauton broth until 1949 when it was grown exclusively in Sauton medium [12] . Prior to the establishment of seed stocks in the 1960s, BCG was passaged continuously, and the changes in media and transfer schedules contributed to the "in vitro evolution" of BCG [6] . It is estimated that as many as 49 production substrains have been used at one time or another in various parts of the world [13] , including the four major BCG vaccines in current use (BCG-Pasteur, -Danish, -Glaxo, and -Japan) [14] . The relative protective efficacy of BCG substrains is currently unknown [6, 15] .

Anecdotal reports have long indicated that BCG substrains exhibit phenotypic differences in growth characteristics, biochemical activities, ability to protect against challenge with Mycobacterium tuberculosis (M. tb), and residual virulence [16] . Over the past decade, numerous groups have sought to identify the genomic changes responsible for these phenotypes. The earliest whole genome comparisons confirmed that BCG was indeed related to, but distinct from M. tb and M. bovis [17] [18] [19] . Subsequent analyses of multiple vaccine strains have uncovered extensive genome diversity including both deletions and duplications in BCG substrains [18, [20] [21] [22] . The phylogeny established by these molecular methods is consistent with the historical records of BCG dissemination [20, 23, 24] . For example, BCG strains acquired after 1927 exhibit the RD2 deletion, while nRD18 is only deleted in strains obtained after 1933. Other genomic changes are exclusive to individual daughter strains, and are associated with vaccine production at specific locations [22, 24] .

A number of molecular techniques have been used to investigate genomic polymorphisms in BCG strains. Early efforts using subtractive hybridization [18] and spotted oligonucleotide arrays [20, 22, 25] effectively identified large sequence polymorphisms, but lacked the resolution to identify smaller changes. More recently, complete genome sequencing has enabled high-resolution analysis of BCG-Pasteur 1173P2 [24] , but sequences for other BCG lineages have yet to be determined. To identify potential genomic polymorphisms in other BCG substrains, we have employed a tiling array platform developed by Nim-bleGen Systems. This DNA microarray-based comparative genome sequencing technique allows high resolution detections of sequence polymorphisms [26] [27] [28] . Using this technique, in tandem with DNA resequencing, we have identified a number of novel genomic polymorphisms in BCG strains. Importantly, these polymorphisms affect genes that are known virulence factors and are expected to have a major impact on the immunogenicity and efficacy of individual vaccine strains.

We have used NimbleGen tiling arrays to analyze the genomic variability of 13 BCG strains, including BCG-Russia, -Japan, -Moreau, -Sweden, -Birkhaug, -China, -Prague, -Glaxo, -Danish, -Tice, -Phipps, -Frappier and -Pasteur. All of these strains, except BCG-China, have previously been subjected to genomic analysis by other methods [18, 20, 22, 24, 25] . The complete genome sequence of BCG-Pasteur 1173P2 is available [24] . The same BCG-Pasteur strain was included in the analysis to serve as an internal control for our experiments in addition to validating the NimbleGen technique. In each experiment, genomic DNA from M. tb H37Rv [29] acted as the common referent.

A total of 42 deletions were identified. Twenty-five of these have been described previously [18, 20, 22, 24, 25] . Thirteen more represent transposons (e.g., IS6110) present in the referent strain (M. tb H37Rv), but absent from the M. bovis and BCG lineages [24, 29, 30] . Six duplications were identified, four (DU1, DU2-I, -II, -III) of which have been described previously [21, 24] . These results confirm the validity of our approach, and the utility of tiling arrays for comparative genomics. A total of 4 novel deletions and 2 duplications were identified in our analysis. These novel deletions and duplications are described below.

Two deletions specific to BCG-Moreau were identified. The first is a 975 bp deletion ( Table 1 ) that eliminates the distal end of fadD26 (Rv2930/BCG2952) and the start of ppsA (Rv2931/BCG2953). These genes are part of the genetic locus required for the biosynthesis of phthiocerol dimycocerosates (PDIMs) and phenolic glycolipids (PGLs) [31] , two cell wall lipids known to be important for the virulence of M. tb and M. bovis [32] [33] [34] . In previous work, we demonstrated that BCG-Moreau does not pro-duce PDIMs or PGLs [35] , which is now explained by the fadD26-ppsA deletion identified in the current study.

The second novel polymorphism in BCG-Moreau is an 1128 bp deletion within Rv3887c/BCG3942c (Table 1) . Although intact in other BCG substrains, this region overlaps with a 2.4-kb deletion (termed RDpan) found in some M. bovis strains [36] , including the sequenced strain, AF2122/97 [30] . The Rv3887c/BCG3942c gene encodes a membrane transport protein and is part of the ESX-2 type VII secretion system [37] . The role of the ESX-2 system in virulence is unknown considering its variable presence among clinical M. bovis isolates from both France and England [36] . However, loss of the Rv3887c membrane transporter likely eliminates the secretion of ESAT-6-and CFP-10-like antigens [37] and influences the immunogenicity of the vaccine strain.

Two novel deletions were identified in BCG-Sweden and BCG-Birkhaug. These polymorphisms are identical between the two BCG strains, which is consistent with their genealogy [24] . The first deletion comprises 110 bp and disrupts the promoter and translational start site of whiB3 (Rv3416/BCG3486) [see Additional file 1]. The other is a 245 bp deletion within trcR (Rv1033c/ BCG1091c) ( Table 1 ). Both genes encode transcriptional regulators known to impact virulence.

WhiB3 belongs to a family of seven M. tb transcriptional regulatory proteins that contain iron-sulfur clusters and are predicted to regulate gene expression in response to environmental stimuli [38] . WhiB3 responds to oxygen and nitric oxide, and is important for regulation of carbon metabolism [39] . The deletion of whiB3 in M. bovis attenuates in vivo growth in guinea pigs [40] . TrcR is the response regulator of the TrcR-TrcS two-component system. Deletion of trcS from M. tb generates a hypervirulent phenotype such that the strain exhibits increased lethality in SCID mice [41] .

Although the genomic profiles of BCG-Birkhaug and BCG-Sweden are similar, we have also found that BCG-Birkhaug is distinguished by a strain-specific duplication, named DU-Birkhaug. This spans the origin of replication and is analogous to the DU1 duplication in BCG-Pasteur [21, 24] (Fig. 1A) . However, the borders of the DU-Birkhaug are different. Whereas DU1 encompasses 29.6 kb from Rv3910 to pknB/Rv0014, DU-Birkhaug spans a slightly different region, from trxB/Rv3913 to rodA/ Rv0017c. Most of the genes in these regions are involved with DNA replication and cell division. Unlike DU1, DU-Birkhaug also appears to be in a genomic location distant to its original copy. Initial PCR-based attempts to characterize the boundaries of this duplication assumed that the second copy was nearby failed to detect a product (data not shown). As such, the genome location of DU-Birkhaug remains unknown.

Our analysis also revealed a novel duplication in the genome of BCG-Tice termed DU-Tice. It comprises a 22kb duplication that encompasses Rv1782-Rv1800 (Fig.  1B ). The precise boundaries and location of this duplication were determined using primers at the junction (Fig.  1C) . Interestingly, DU-Tice encodes the ESX-5 secretion system [37, 42] . This includes several conserved membrane transporters (Rv1782, Rv1783, Rv1795, and Rv1797), a membrane associated ATPase (Rv1784), a set of PE/PPE genes (Rv1787-Rv1792) and the ESAT-6 and CFP-10 family proteins (esxM and esxN) [37] . ESX-5 is absent from the genome of the fast-growing, non-pathogenic M. smegmatis, but present in both the M. avium complex and M. marinum. The role of ESX-5 in virulence has been demonstrated in M. marinum [37, 42] . It has been suggested that the ESX clusters evolved via gene duplication [43] and DU-Tice offers the first snapshot of such an event.

To our knowledge, we have conducted the first genomic analysis of BCG-China, which is a descendant of BCG-Danish obtained from the Statens Serum Institut around 1947. Consistently, BCG-China exhibits the DU2-III duplication and deletion of RD2 (data not shown), which is similar to other BCG-Danish derivatives, including BCG-Prague (obtained in 1946 from passage 725) [44] , 

The coordinates correspond to the genome of M. tb H37Rv. The two deletions and DU-Tice were confirmed by PCR application and DNA sequencing.

Novel duplications identified in BCG-Birkhaug and BCG-Tice by NimbleGen tiling array Figure 1 Novel duplications identified in BCG-Birkhaug and BCG-Tice by NimbleGen tiling array. Sections of the ratio plot are shown. The ratio of the reference (M. tb H37Rv) probe intensity (Cy5) was divided by the test (BCG strain) probe intensity (Cy3). Reference probes and test probes that do not span a mutation should represent full-length perfect match hybridization, and thus should have similar intensities, with a reference/test ratio near 1. If the test genome contains an amplification event (increased copy number when compared to the reference), then the reference/test ratio will shift below 1. (A) Novel duplication (DU-Birkhaug) identified in BCG-Birkhaug, which is analogous to the DU-Pasteur (DU1) but has different borders. The same genomic region of BCG-Sweden, which is closely related to BCG-Birkhaug, is shown for comparison. (B) Novel duplication (DU-Tice) identified in BCG-Tice. Three other BCG strains belonging to the same group (DU2-IV) are shown for comparison. (C) The precise border of DU-Tice is mapped by PCR amplification using primers specific to the junction. The two copies are immediately adjacent to each other and overlap by 1 bp.

BCG-Glaxo (obtained in 1954, from passage 1077) and BCG-Danish (lyophilized in 1961, from passage 1331) [45] . However, BCG-China and -Prague do not contain the previously described deletion of Rv1810, which is characteristic of BCG-Glaxo and -Danish. As such, the Rv1810 deletion must have occurred between 1947 and 1954. Coincidentally, this period corresponds to the replacement of potato bile medium by Sauton medium for BCG production in Denmark [12] .

The PhoP-PhoR system is one of the 11 two-component systems found in the M. tb genome [29] . The PhoR protein is a transmembrane histidine kinase that transmits signals from the environment by autophosphorylation. The phosphoryl group is then transferred to PhoP, a response regulator that regulates the expression of multiple genes [46] . Recently, several studies have demonstrated that the PhoP-PhoR system, particularly PhoP, plays an essential role in M. tb virulence [26, [46] [47] [48] . A single point mutation (S219L) in the DNA binding region of PhoP partially accounts for the attenuation of the H37Ra strain of M. tb [26] . Furthermore, a phoP mutant of M. tb was found to be more attenuated than BCG-Pasteur in SCID mice infections [47] . Our NimbleGen analysis revealed some weak signals in the phoP-phoR region (not shown), which prompted us to resequence these genes. The DNA fragment containing the promoter region of phoP, the ORFs of phoP and phoR, and the intergenic region was PCR amplified from each BCG strain and determined by DNA sequencing.

Our sequence analysis revealed a number of polymorphisms in the phoP-phoR locus in various BCG strains compared to the genome sequence of M. bovis. The three early BCG substrains, BCG-Russia, -Japan, and -Moreau, contain an identical IS6110 (1,356 bp) insertion at nucleotide 851593 of the M. tb genome, which is 18 bp upstream of the start codon of phoP (Fig. 2) . This IS6110 element is identical to many other copies of IS6110 found in various locations in the M. tb genome. It is flanked by a 3-bp direct repeat (GAA) on both sides and is in an inverse orientation of phoP-phoR (Fig. 2) . The presence of an IS6110 element in the promoter region of phoP in BCG-Russia, -Japan, and -Moreau has been described previously, but its insertion site and orientation were not determined until now [49] . Although not present in M. tb H37Rv or M. bovis AF2122/97, an IS6110 insertion in the phoP promoter was found in a clinical strain of M. bovis termed B strain, which was responsible for a severe nosocomial outbreak of multidrug resistant TB in humans in Spain [50, 51] . However, unlike the three BCG strains, the IS6110 insertion in the M. bovis B strain is located at 75 bp upstream of the start codon of phoP and is in the same orientation as phoP-phoR [50] . The potential effect of IS6110

IS6110 insertion in the phoP promoter in BCG-Russia, -Moreau, and -Japan on phoP expression is described in the 'Discussion' section.

Three other novel phoP-phoR polymorphisms that likely impact their functions were also uncovered by our sequencing analysis. An identical, 11-bp deletion within the ORF of phoR was uncovered in BCG-Sweden and BCG-Birkhaug (ACCGGACTGGG, nucleotides from 853689 to 853699, M. tb genome coordinates). This deletion changes the amino acid sequence of 54 residues (residues 432 to 485) in the C-terminal of PhoR. This polymorphism is different than the previously described 10 bp deletion within phoR present in BCG-Danish and BCG-Glaxo, which affects residues 91-485 [24] . BCG-Frappier also contains a single nucleotide deletion (A at 852701, M. tb genome coordinates), causing a frame-shift mutation that affect residues 103-485 of PhoR. Together, these results indicate that besides BCG-Danish and BCG-Glaxo, BCG-Sweden, -Birkhaug, and -Frappier also contain a defective phoR gene.

A single nucleotide insertion within the ORF of phoP was uncovered in BCG-Prague (G, between nucleotides 852067 and 852068, M. tb genome coordinates) [see Additional file 2]. This frame shift mutation changes the C-terminal sequence (residues 154-247) of PhoP, which is the DNA binding domain (residues 144-247) [52] [53] [54] .

As such, BCG-Prague is a natural phoP mutant.

Single point mutations in PhoP or PhoR are also found in various BCG strains and are summarized in Table 2 . In contrast, sequences of the phoP-phoR locus of BCG-Phipps, -Tice, and -Pasteur are identical to the published sequence of BCG-Pasteur and M. bovis [24, 30] .

The loss of the RD1-encoded ESX-1 protein secretion system during 1908-1921 contributes to the attenuation of BCG ( [55] , see also Fig. 3) . However, because reintroduction of ESX-1 into BCG does not restore full virulence, other genetic lesions are also involved [56] . As such, some strains are more virulent than others in animal models of infection [57] and also exhibit differential ability to induce adverse reactions (reactogenicity) following vaccination in neonates [58] . Our current work begins to provide some explanation for these observed differences (Fig. 3) .

Consistent with a previous study [58] , we find that the earliest distributed BCG strains, BCG-Russia, -Japan, and -Moreau, all contain a second copy of IS6110 that is inserted in the promoter region of phoP. A similar, albeit distinct, insertion of IS6110 in the phoP promoter was also found in a virulent strain called M. bovis B strain [50] . The presence of IS6110, which is in the same orientation as phoP, increases the expression of phoP and (the resulting [30] . BCG-Danish and -Glaxo contains a 10 bp deletion within phoR, which was described previously [24] . ND: polymorphisms not detected. NA: no affect.

increase in virulence) was thought to be responsible for the outbreak of M. bovis B strain in humans [50] . Similarly, the expression level of phoP was found to be higher in BCG-Japan than in BCG-Pasteur [24] . However, in BCG-Japan, -Russia, and -Moreau, the IS6110 is in the inverse orientation of phoP (Fig. 2 ). As such, how IS6110 upregulates phoP expression in BCG is not immediately apparent and likely involves a different mechanism. One possibility is the elimination of phoP autoregulation. It was shown that PhoP protein, albeit from H37Ra, binds to three 9-bp direct repeats within the phoP promoter sequence and represses its own expression [52] . In BCG, IS6110 is inserted between the PhoP binding sites and the start codon of phoP, which could impair the repression by PhoP and subsequently increase phoP expression. Alternatively, an unidentified promoter sequence within IS6110 in the same orientation of phoP could drive the expression of phoP. The presence of the second copy of IS6110 in these early BCG strains also suggests that the original BCG isolated in 1921 might have been derived from a highly virulent M. bovis strain containing the same IS6110 element. This IS6110 was subsequently lost in other BCG strains (Fig. 3) and is not present in most clinical strains of M. bovis and M. tb isolated in modern times [50, 59] .

Given the important role of PhoP in M. tb virulence, higher expression of phoP could explain why BCG-Russia is generally considered more virulent than other BCG strains [6] . However, in the other early strains, BCG-Moreau and BCG-Japan, the loss of lipid virulence factors PDIMs and PGLs appears to have a more pronounced effect on virulence. Consequently, these two strains, together with BCG-Glaxo, which also lacks PDIMs and PGLs, and as we have described previously, are more attenuated and less reactogenic than other BCG strains [35] . The deletion of fadD26-ppsA described here provides a genetic mechanism for the defective PDIM/PGL biosynthesis in BCG-Moreau. However, this region is intact in BCG-Japan and BCG-Glaxo, indicating that other mechanisms may also lead to the PDIM/PGL defect.

BCG-Sweden was obtained from the Institut Pasteur in 1926 while Konrad Birkhaug acquired the strain that bears his name around 1927 [6] . Previous studies indicated that these strains differ from other early BCG strains (i.e., BCG-Russia, -Japan, -Moreau) only by the loss of the IS6110 element described above. Our current work reveals three novel deletions shared by BCG-Sweden and BCG-Birkhaug (Fig. 3) , which distinguish them from other early strains. Two deletions affect the regulatory proteins WhiB3 and TrcR, and have different impacts on virulence.

The whiB3 gene appears to be important for virulence. The M. bovis whiB3 mutant is attenuated for growth in guinea pigs but not in mice [40] . Conversely, the trcRS two-component system has a negative impact on virulence. Deletion of trcS from M. tb generates a hypervirulent phenotype in SCID mice [41] . BCG-Sweden was used in Sweden from 1926 until 1978 and was then replaced by BCG-Danish because of the high frequency of osteitis associated with the former strain [60] . The deletion of trcR may contribute to the reactogenicity of BCG-Sweden. Figure 3 Refined genealogy of BCG vaccines. The genealogy is modified from a previous model [24] . Genetic markers identified in this work are highlighted.

The other deletion found in BCG-Sweden and BCG-Birkhaug affects the phoR gene of the phoP-phoR two-component system. Remarkably, three other late BCG strains, BCG-Danish, -Glaxo, and -Frappier also contain a defective phoR gene. Together a total of five BCG strains are natural phoR mutants. However, three distinct mutations are found among these five strains, which correspond to their genealogy (Fig. 3) . The role of phoR in virulence is less understood than for phoP. Among its many functions, phoP is required for the biosynthesis of trehalose-containing cell wall lipids [48, 61, 62] . Contrastingly, phoR does not seem to be required for this function [62] . Nevertheless, the fact that the phoR mutation has been acquired by different groups of BCG strains by three independent events and genetic mechanisms suggests that there was a common selective pressure and an important role for this gene during the in vitro evolution of BCG.

Another BCG strain that contains a major mutation in the phoP-phoR system is BCG-Prague. A single nucleotide insertion in the ORF of phoP changes the C-terminal sequence, which contains the DNA binding domain of PhoP [52] [53] [54] . As such, BCG-Prague is a natural phoP mutant and likely to be more attenuated than other BCG strains. This is consistent with the study by Lagranderie et al., which showed in mice models of infection that BCG-Prague exhibited more attenuated phenotypes compared to three other BCG strains (BCG-Russia, -Pasteur, and -Glaxo) [57] . Compared to 11 other BCG strains, including BCG-Russia, -Moreau, -Japan, -Sweden, -Danish, -Glaxo, and -Pasteur that have been analyzed in the current study, BCG-Prague consistently exhibited the weakest ability to induce delayed type hypersensitivity to tuberculin in children [63] or in guinea pig models [64] . Because of the traditional presumption that tuberculin reactivity is associated with vaccine potency, BCG-Prague, which was used in Czechoslovakia between 1951-1980 and appeared to be effective, was replaced by BCG-Russia in 1981 [58] . An immediate increase of BCG-induced osteitis cases was observed in Czechoslovakia following the switch of BCG-Prague to BCG-Russia [65] . The phoP mutation detected in the current study may explain the weak tuberculin sensitivity induced by BCG-Prague. It was recently shown that a phoP mutant of M. tb was more attenuated than BCG-Pasteur and confers an equivalent protection in mice against M. tb challenge. In the guinea pig model, the M. tb phoP mutant showed superior protection to BCG-Pasteur against a high dose challenge with M. tb [47] . Consequently, the M. tb phoP mutant is now being evaluated as a vaccine candidate to replace BCG [66] . Since BCG-Pasteur contains an intact phoP gene, and in light of our finding, it would be worthy to compare the M. tb phoP mutant with BCG-Prague in terms of safety and protective efficacy.

The novel duplication uncovered in BCG-Tice (DU-Tice) may have an impact on its residual virulence and immunogenicity. DU-Tice contains the entire ESX-5 secretion system, which is one of the five type VII secretion systems found in the M. tb complex [37] . Importantly, besides the RD1-encoded ESX-1, ESX-5 is the only other ESX system that has been shown to be involved in virulence thus far [37] . ESX-5 is conserved in other pathogenic mycobacteria and reported to facilitate the cell-to-cell spread of M. marinum in infected macrophages, a function shared by ESX-1 [42] . However, ESX-5 does not complement the loss of virulence caused by ESX-1 deletion, suggesting that they play distinct roles in virulence [37] . Horwitz and co-workers have used BCG-Tice as the host strain to overexpress antigen 85B. This resulted in a recombinant strain termed rBCG30 that exhibits superior protective efficacy over BCG-Tice and is currently being evaluated as a vaccine candidate in human clinical trials [67] [68] [69] [70] [71] . The rBCG30 Tice strain also showed significantly stronger immune response and better protection against M. tb challenge than the rBCG30 strain based on BCG-Connaught [69] .

The duplication of ESX-5 in BCG-Tice, which could increase the residual virulence and immunogenicity, may partially account for the benefit associated with rBCG30 Tice.

Our current work has uncovered six large sequence polymorphisms not described previously, including two deletions exclusive to BCG-Moreau, two deletions shared by BCG-Sweden and BCG-Birkhaug, as well as the DU-Birkhaug and DU-Tice duplications. Moreover, we have uncovered a number of polymorphisms in the phoP-phoR locus in various BCG strains. Remarkably, these polymorphisms affect genes that are well known to have major impact on the virulence of M. tb or M. bovis. These include genes involved in the biosynthesis of lipid virulence factors PDIMs/PGLs, genes that encode the ESX family type VII secretion system, and the phoP-phoR two-component regulatory system. [72] . In addition, a mutant of M. bovis deficient in PDIMs/PGLs is being considered as a vaccine to protect wildlife against bovine tuberculosis [73] . Our previous study [35] and current work provide direct evidence that BCG vaccine strains are different in major virulence factors, and likely have different vaccination properties including safety, immunogenicity, and efficacy. Since new vaccine candidates are evaluated for their vaccination properties relative to BCG, the appropriate choice of BCG strain for these studies is critical. Furthermore, because it is likely that BCG will continue to play a role in tuberculosis control by being included in forthcoming clinical trials, as either a primer to be boosted by new components (e.g. subunit or DNA vaccine) or as an integral component (e.g. recombinant BCG) of new vaccines, greater attention must be given to the benefits that a particular strain may -or may notoffer.

The mycobacterial strains used in this study were: 

Mutation mapping microarrays were designed with Nim-bleGen algorithms that select a 29-mer oligonucleotide every 7 bases on each strand of the reference genome sequence (Genbank Accession AL123456) [29] . All probes were synthesized in parallel on a four-array set using a Digital Light Processor™ (Texas Instruments, Plano Texas) and photoprotected by phosphoramidite chemis-try (Maskless Array Synthesis) (NimbleGen Systems, Madison WI) in a random probe layout [74, 75] .

Labeled genomic DNA was hybridized to arrays in the NimbleGen Hybridization Buffer at 42°C for 16 hr using a MAUI hybridization system (BioMicro Systems, Inc. Salt Lake City, Utah). Labeled genomic DNA (5 μg) from the reference strain M. tb H37Rv and from each BCG strain were co-hybridized to each array. Arrays were washed with NimbleGen wash buffer, and were then spun dry in a microarray high-speed centrifuge (TeleChem International, Inc., Sunnyvale, CA) and stored until scanned.

Microarrays were scanned at 5 μm resolution using the Genepix ® 4000B scanner (Axon Instruments, Union City CA), and pixel intensities were extracted using NimbleS-can™ v2.4 software (NimbleGen). Probes that spanned potential mutations were identified by NimbleGen software. Probe sequences corresponding to all possible candidate mutation sites were selected for resequencing. The strategy that was used to automatically generate the sequencing array is similar to that described previously [28] . Briefly, 8 probes per base position were analyzed, 4 per genome strand. These probes contain all possible alleles at a centrally located position. The length, melting temperature and mismatch position of each probe were optimized. When target DNA is hybridized to these arrays the perfectly matched probe will hybridize more strongly than the three corresponding mismatched probes for each strand. The differential signal intensity between the perfectly matched probe and mismatched probes allows the base to be determined precisely. These resequencing arrays were synthesized, hybridized with labeled genomic DNA from each BCG strain and scanned as above. Sequence base assignments were made using a machinelearning algorithm [76] . Putative mutation-containing DNA segments were PCR amplified and verified by capillary sequencing [see Additional file 3]. The microarray data has been deposited in the Center for Information Biology Gene Expression Database (CIBEX; http:// cibex.nig.ac.jp), with the accession number of CBX70.

ASL, VT, ZW, and XY performed the experiments and participated in data analysis. DCA participated in data analysis and co-authored the manuscript. GFG oversaw the experiments. BZ oversaw the experiments and participated in data analysis. JL oversaw the experiments, analyzed the data, and wrote the manuscript.

The efficacy of bacillus Calmette-Guerin vaccination of newborns and infants in the prevention of tuberculosis: meta-analyses of the published literature

Effect of BCG vaccination on childhood tuberculous meningitis and miliary tuberculosis worldwide: a meta-analysis and assessment of cost-effectiveness

Protective effect of bacillus Calmette Guerin (BCG) vaccine in the prevention of leprosy: a meta-analysis

Preventing tuberculosis with bacillus Calmette-Guerin vaccine: a meta-analysis of the literature

Efficacy of BCG vaccine in the prevention of tuberculosis. Meta-analysis of the published literature

BCG-different strains, different vaccines? Lancet Infect Dis

Failure of the Mycobacterium bovis BCG vaccine: some species of environmental mycobacteria block multiplication of BCG and induction of protective immunity to tuberculosis

Field trials of tuberculosis vaccines: how could we have done them better? Control Clin Trials

Differential effects of prior exposure to environmental mycobacteria on vaccination with Mycobacterium bovis BCG or a recombinant BCG strain expressing RD1 antigens

The 'experimental stable' of the BCG vaccine: safety, efficacy, proof, and standards

The Swedish strain of BCG

Changes in BCG strains

Report on a WHO consultation on the characterisation of BCG strains

Relationship between bacille Calmette-Guerin (BCG) strains and the efficacy of BCG vaccine in the prevention of tuberculosis

Effect of freeze-drying on the colony morphology of various daughter strains of BCG

Use of a Mycobacterium tuberculosis H37Rv bacterial artificial chromosome library for genome mapping, sequencing, and comparative genomics

Molecular analysis of genetic differences between Mycobacterium bovis BCG and virulent M. bovis

Physical mapping of Mycobacterium bovis BCG pasteur reveals differences from the genome map of Mycobacterium tuberculosis H37Rv and from M. bovis. Microbiology

Comparative genomics of BCG vaccines by wholegenome DNA microarray

Comparative genomics uncovers large tandem chromosomal duplications in Mycobacterium bovis BCG Pasteur

The in vitro evolution of BCG vaccines

A historical and molecular phylogeny of BCG strains

Genome plasticity of BCG and impact on vaccine efficacy

Detection of deleted genomic DNA using a semiautomated computational analysis of GeneChip data

Control of M. tuberculosis ESAT-6 secretion and specific T cell recognition by PhoP

Identification of a nitroimidazo-oxazine-specific protein involved in PA-824 resistance in Mycobacterium tuberculosis

Tracking the evolution of the SARS coronavirus using high-throughput, high-density resequencing arrays

Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence

The complete genome sequence of Mycobacterium bovis

The dimycocerosate ester polyketide virulence factors of mycobacteria

Complex lipid determines tissue-specific replication of Mycobacterium tuberculosis in mice

Transposon mutagenesis of Mb0100 at the ppe1-nrp locus in Mycobacterium bovis disrupts phthiocerol dimycocerosate (PDIM) and glycosylphenol-PDIM biosynthesis, producing an avirulent strain with vaccine properties at least equal to those of M. bovis BCG

A glycolipid of hypervirulent tuberculosis strains that inhibits the innate immune response

Differential productions of lipid virulence factors among BCG vaccine strains and implications on BCG safety

A novel polymorphic genetic locus in members of the Mycobacterium tuberculosis complex

Type VII secretionmycobacteria show the way

Differential gene expression in response to exposure to antimycobacterial agents and other stress conditions among seven Mycobacterium tuberculosis whiB-like genes

Mycobacterium tuberculosis WhiB3 responds to O2 and nitric oxide via its [4Fe-4S] cluster and is essential for nutrient starvation survival

Mycobacterium tuberculosis WhiB3 interacts with RpoV to affect host survival but is dispensable for in vivo growth

Deletion of two-component regulatory systems increases the virulence of Mycobacterium tuberculosis

Bitter W: A specific secretion system mediates PPE41 transport in pathogenic mycobacteria

Evolution and expansion of the Mycobacterium tuberculosis PE and PPE multigene families and their association with the duplication of the ESAT-6 (esx) gene cluster regions

Some biological, biochemical and morphological changes and their correlations in selected BCG vaccine strains

Development of the Mycobacterium bovis BCG vaccine: review of the historical and biochemical evidence for a genealogical tree

Mutation in the transcriptional regulator PhoP contributes to avirulence of Mycobacterium tuberculosis H37Ra strain

The live Mycobacterium tuberculosis phoP mutant strain is more attenuated than BCG and confers protective immunity against tuberculosis in mice and guinea pigs

The Mycobacterium tuberculosis PhoPR two-component system regulates genes essential for virulence and complex lipid biosynthesis

Use of gene probes based on the insertion sequence IS986 to differentiate between BCG vaccine strains

IS6110 mediates increased transcription of the phoP virulence gene in a multidrug-resistant clinical isolate responsible for tuberculosis outbreaks

High rate of tuberculosis reinfection during a nosocomial outbreak of multidrug-resistant tuberculosis caused by Mycobacterium bovis strain B

Transcriptional autoregulation by Mycobacterium tuberculosis PhoP involves recognition of novel direct repeat sequences in the regulatory region of the promoter

PhoP-PhoP interaction at adjacent PhoP binding sites is influenced by protein phosphorylation

Structure of the DNA-binding domain of the response regulator PhoP from Mycobacterium tuberculosis

Mycobacterial virulence and specialized secretion: same story, different ending

Recombinant BCG exporting ESAT-6 confers enhanced protection against tuberculosis

Comparison of immune responses of mice immunized with five different Mycobacterium bovis BCG vaccine strains

Quality control of BCG vaccine by WHO: a review of factors that may influence vaccine effectiveness and safety

Mapping of IS6110 insertion sites in Mycobacterium bovis isolates in relation to adaptation from the animal to human host

BCG complications. Estimates of the risks among vaccinated subjects and statistical analysis of their main characteristics

A point mutation in the two-component regulator PhoP-PhoR accounts for the absence of polyketide-derived acyltrehaloses but not that of phthiocerol dimycocerosates in Mycobacterium tuberculosis H37Ra

The virulence-associated two-component PhoP-PhoR system controls the biosynthesis of polyketide-derived lipids in Mycobacterium tuberculosis

Tuberculin sensitivity and skin lesions in children after vaccination with 11 different BCG strains

Tuberculin sensitivity in guinea-pigs after vaccination with varying doses of BCG of 12 different strains

Second IUATLD study on complications induced by intradermal BCG-vaccination

Live tuberculosis vaccines based on phoP mutants: a step towards clinical trials

Protective immunity against tuberculosis induced by vaccination with major extracellular proteins of Mycobacterium tuberculosis

Recombinant bacillus calmette-guerin (BCG) vaccines expressing the Mycobacterium tuberculosis 30-kDa major secretory protein induce greater protective immunity against tuberculosis than conventional BCG vaccines in a highly susceptible animal model

A new vaccine against tuberculosis affords greater survival after challenge than the current vaccine in the guinea pig model of pulmonary tuberculosis

Enhancing the protective efficacy of Mycobacterium bovis BCG vaccination against tuberculosis by boosting with the Mycobacterium tuberculosis major secretory protein

Maslesa-Galic S: A novel live recombinant mycobacterial vaccine against bovine tuberculosis more potent than BCG

Advances in tuberculosis vaccine strategies

A new attenuated Mycobacterium bovis vaccine protects brushtail possums (Trichosurus vulpecula) against experimental tuberculosis infection

Gene expression analysis using oligonucleotide arrays produced by maskless photolithography

Light-directed 5'-->3' synthesis of complex oligonucleotide microarrays

A self-tuning method for one-chip SNP identification

This work was supported by an award from the National Natural Science Foundation of China (NSFC) (to JL), and research grants from Canadian Institutes of Health Research (CIHR) (MOP-15107 and MOP-82772 to JL), and a grant (Z0005190043521) from Beijing Municipal Science And Technology Commission to (BZ).