The first T-cell epitope mapping of Ag85A was performed 20 years ago in seven different mouse strains vaccinated with live M. bovis BCG (12). Twenty-eight overlapping 20-mer peptides covering the complete mature 295-amino-acid (AA) protein were synthesized. Significant interleukin-2 (IL-2) and gamma interferon (IFN-γ) secretion was measured following in vitro stimulation of spleen cells with these peptides. H-2^d haplotype mice (BALB/c and DBA/2) reacted preferentially against the amino-terminal half of the protein, i.e., against peptide 5 (aa 41–60) and especially against peptide 11 (aa 101–120), which contains a predicted I-E^d binding motif. H-2^b haplotype mice, on the other hand, reacted against peptides from both amino- and carboxy-terminal halves of the protein, peptide 25 (aa 241–260) and peptide 27 (aa 261–280) being the most potent stimulators of IL-2 and IFN-γ production. Finally, CBA/J (H-2^k) and major histocompatibility complex class II mutant B6.C.bml2 mice, carrying a mutant I-A^bml2 allele on an H-2^b background, reacted only very weakly to the 85A peptides. (12).

BALB/c and C57BL/6 mice were infected intravenously with Mycobacterium tuberculosis H37Rv and Th1-type spleen cell cytokine secretion was analyzed in response to purified Ag85A, Ag85B, and Ag85C components and synthetic overlapping peptides covering the three mature sequences (13). Tuberculosis-infected C57BL/6 mice reacted strongly to some peptides from Ag85A and Ag85B but not from Ag85C and more specifically strong responses were detected against peptide 25 (aa 241–260) of Ag85A and against the same sequence of Ag85B (13). This latter peptide region was also identified by Yanagisawa et al. (14) and TCR-transgenic mice with MHC class II A^b-restricted CD4⁺ T-cells expressing TCR α and β chains for the mycobacterial Ag85B_240–254 have been generated (15). Tuberculosis-infected BALB/c mice reacted only to peptides from Ag85A: p11 (aa 101–120), p16 (aa 151–160), and p20 (aa 191–210) (13).

Plasmid DNA vaccination is a powerful tool to identify protective antigens of tuberculosis and to identify immunodominant CD4⁺ and particularly CD8⁺ T-cell epitopes (16). BALB/c and C57BL/6 mice were vaccinated intramuscularly with plasmid DNA encoding the three components of the Ag85 complex. Ag85A and Ag85B encoding plasmids induced a robust Th1 like response to native Ag85 purified from BCG CF, characterized by elevated levels of IL-2, IFN-γ, and TNF-α. Levels of IL-4, IL-6, and IL-10 were low or undetectable. Plasmid encoding Ag85C was only weakly immunogenic. Whereas BALB/c mice reacted preferentially to the Ag85A component, C57BL/6 mice reacted to both Ag85A and Ag85B with more or less the same magnitude (17). Furthermore, vaccination with plasmid DNA encoding Ag85A or Ag85B but not Ag85C conferred significant protection against mycobacterial replication in lungs from C57BL/6 mice (17). T-cell epitopes could be identified in BALB/c and C57BL/6 mice vaccinated with plasmid DNA encoding Ag85A, Ag85B, and Ag85C DNA using synthetic peptides spanning the three Ag85 proteins, and the epitope repertoire was found to be broader than in infected mice (13). Despite pronounced sequence homology, a number of immunodominant regions contained component specific epitopes. Thus, BALB/c mice vaccinated against all three Ag85 antigens reacted against the same amino-acid region, 101–120 (already identified in BCG vaccinated and TB infected mice) but responses were completely component specific. In C57BL/6 mice, a cross-reactive T-cell response was detected against two carboxy-terminal peptides spanning amino acids 241–260 and 261–280 of Ag85A and Ag85B. These regions were not recognized at all in C57BL/6 mice vaccinated with Ag85C DNA.

T-cell repertoire of BALB/c mice vaccinated with plasmid DNA encoding Ag85A was broader that of Mtb infected mice (13, 18). Besides peptide regions spanning aa 11–30 and 191–210 inducing both IL-2 and IFN-γ responses, three peptides induced strong IFN-γ but weak to no IL-2 responses. More detailed analysis of the Ag85A sequence for predicted MHC class I binding motifs using “human leukocyte antigen (HLA) peptide motif” (http://www-bimas.cit.nih.gov/molbio/hla_bind/) showed that these three peptides spanned four predicted CD8⁺ T-cell epitopes (18). The following half-time dissociation scores (reflecting affinity for the respective MHC class I molecules) were found: aa 61–68 YDQSGLSV: half-time dissociation score 600, predicted K^d; aa 71–78 PVGGQSSF: half-time dissociation score 390, predicted L^d; aa 145–152 YAGAMSGL: half-time dissociation score 2000, predicted K^d; aa 161–168 PTLIGLAM: half-time dissociation score 150, predicted L^d. CTL activity against these peptides was demonstrated using a ⁵¹Cr release assay (Figure 2) and showed cross-reactive responses against Ag85B for both K^d restricted peptides (18). A particularly interesting region was identified in peptide 15 spanning aa 141–160, which besides the K^d restricted epitope 145–152 also contains a CD4⁺ epitope with a predicted Rothbard and Taylor motif spanning aa 147–154 and an amphipathic stretch spanning aa 149–157 (according to T sites program) (19).

Immunization with DNA followed by modified vaccinia virus Ankara strain, both expressing the antigen 85A, induced both CD4⁺- and CD8⁺-T-cell responses in BALB/c mice, directed against the K^d restricted epitope WYDQSGLSV (aa 60–67) and the I-E^d restricted epitope TFLTSELPGWLQANRHVKPT (aa 99–119), respectively (20). DNA priming, followed by a MVA85A boost induced both CD4⁺ and CD8⁺ responses, whereas priming with MVA followed by a DNA boost only induced CD4 responses. Following immunization with dendritic cells pulsed with the antigen 85A CD4⁺- or CD8⁺-restricted epitope, alone or in combination, copresentation of both epitopes on the same dendritic cell was required for protection, demonstrating that induced CD8⁺ T-cells can play a protective role against tuberculosis (20).

A single intranasal, but not i.m., immunization with a recombinant replication-deficient adenoviral-based vaccine expressing Ag85A (AdAg85A) provided potent protection against airway M. tuberculosis challenge at an improved level over that by cutaneous BCG vaccination. Systemic priming with an Ag85A DNA vaccine and mucosal boosting with AdAg85A conferred a further enhanced immune protection, which was remarkably better than BCG vaccination. Such superior protection triggered by AdAg85 mucosal immunization was correlated with much greater retention of Ag-specific T-cells, particularly CD4 T-cells, in the lung and was shown to be mediated by both CD4+ (LTSELPGWLQANRHVKPTGS, aa 101–120) and CD8⁺ (MPVGGQSSF, aa 70–78) T-cells (21).

For H-2^b haplotype mice, no MHC class I restricted epitopes have been identified so far on Ag85A or Ag85B to our knowledge, neither in BCG or plasmid DNA vaccinated nor in TB infected mice.

In our first paper on Ag85A (called P32 at that time), we reported that healthy Mantoux positive volunteers showed a much stronger lymphoproliferative and IFN-γ response to this antigen than tuberculosis patients (22). This was the initial indication that T-cell responses against this protein could confer protection against Mtb. Subsequently, we reported on T-cell epitope mapping of Ag85A from Mtb using peripheral blood mononuclear cell (PBMC) cultures from healthy tuberculin-positive volunteers and from patients with tuberculosis, using the same synthetic 20-mer peptides of the murine study. Peptide recognition was largely promiscuous, with a variety of HLA haplotypes reacting to the same peptides. PBMC from all tuberculin-positive subjects reacted to Ag85A, and the majority proliferated in response to peptide 6 (amino acids 51–70), peptides 13, 14, and 15 (amino acids 121–160), or peptides 20 and 21 (amino acids 191–220). PBMC from tuberculosis patients demonstrated a variable reactivity to Ag85 and its peptides, and the strongest proliferation was observed against peptide 7 (amino acids 61–80) (23). Nine out of ten of the tuberculin-positive volunteers in this study reacted to aa 141–160, precisely the peptide characterized by the presence of both a CD4⁺ and a CD8⁺ epitope in BALB/c mice. In contrast, the most immunogenic CD4⁺ peptide of Ag85A for BALB/c mice, i.e., p11 was not recognized by PBMCs from healthy PPD-positive humans. However, two reports published in 2000 and 2001 showed that aa 100–117 of Ag85B is recognized in a similar promiscuous manner by T-cells from a majority of PPD-positive human volunteers (24, 25). Ag85A differs from Ag85B in three aa in this region Gly107Gln, Gln110Ser, and His114Ala. Ag85A sequences from M. ulcerans, M. leprae, Map, and Maa also show strong differences as compared to MtbAg85A sequence in this region (see Figure 1). These aa shifts probably explain the difference in the human responses to Ag85A and Ag85B, as also in DNA vaccinated BALB/c mice, IL-2 and IFN-γ responses to region 100–120 are specific for both Ag85 components (13).

In 1995, Silver et al. had already assessed the T-cell epitopes of Mtb Ag85B using blastogenic responses of PBMC from 12 healthy purified protein derivative-positive subjects to a set of synthetic 15-mer peptides based on the full 325-amino-acid sequence (leader sequence included) (26). Seven immunodominant regions were identified and each subject responded to at least one of the two most dominant epitopes, which corresponded to aa 91–115 and aa 193–217. Peptides of these two epitopes induced production of IFN-γ by sorted CD4⁺ T-cells.

Human leukocyte antigen-transgenic mice can be a powerful tool to identify human T-cell epitopes in an experimental mouse model. In 1998, A. Geluk reported on the identification of an HLA-class II restricted epitope of Ag85. HLA-DRA/B1*0302 (DR3) transgenic mice were vaccinated with Ag85 protein, purified from Mtb CF in Incomplete Freund’s adjuvant (27). Using 20-mer peptides, covering the entire Mtb 85B sequence, they identified one single peptide epitope in the NH₂-terminal region, spanning aa 11–30 (aa 51–70 in the numbering including the signal peptide): LQVPSPSMGRDIKVQFSGG. This sequence is identical in Ag85A and Ag85B, but differs in two positions in Ag85C: Pro16Ala and Ser28Gly. Also in M. ulcerans, Map, and Maa, position 16 has the shift to Alanine. In M. leprae there is also one aa shift: Ser28Asn.

Whereas most of these studies on human T-cell epitope mapping were performed during the mid-nineties, one more recent paper of Lindestam Arlehamn et al. reported on the memory phenotype of Mtb-specific CD4⁺ T-cells, using HLA-class II tetramers for a peptide shared between Ag85A and Ag85B, i.e., aa 90–104 AGCQTYKWETFLTSE in healthy PPD-positive donors, latently infected with Mtb (28). Tetramer positive T-cells predominantly consisted of CD45RA⁻CCR7⁺ central memory T-cells in all donors tested, followed by effector memory (CD45⁻CCR7⁻) T-cells. Only a minor fraction appeared to be naïve or effector T-cells. Interesting to note that this sequence is shared with Ag85A of M. leprae that there is only Q93T shift in M. ulcerans, Map, and Maa but that the sequence of Mtb Ag85C differs in five positions.

Less is known on human MHC class I restricted epitopes of Ag85. Klein et al. reported on a HLA-B*35 restricted CD8⁺ T-cell epitope of Ag85C (29). Using reverse immunogenetics, they tested 23 motif-bearing peptides of the Ag85 complex for binding to HLA-B*35, one of the most common HLA-B types in West Africa. Three 9-mer peptides bound with high affinity to HLA-B*3501. Peptide MPVGGQSSFY (spanning aa 70–79 of the mature protein), a highly conserved region shared by all three members of the Ag85 complex of Mtb and also identical in M. ulcerans, M. avium, and M. leprae. This peptide encompasses an L^d predicted epitope, recognized by BALB/c mice vaccinated with pAg85A DNA (see Plasmid DNA vaccines/viral vectors encoding Ag85A, Ag85B, and Ag85C of Mtb) and also an IL-2/IFN-γ inducing region for C57BL/6 mice vaccinated with pAg85C (13). Peptide WPTLIGLAM of Ag85C (spanning aa 160–168) with a W160G change as compared to all other sequences, and a T162S change in Ag85B, and the three other non-tuberculous mycobacteria. Finally peptide IPAEFLENF of Ag85B (spanning aa 224–232), with an isoleucine in position 224 shared with Ag85C, and a leucine in Ag85A of Mtb and the four non-tuberculous mycobacteria. WPTLIGLAM stimulated effector cells were able to kill Mtb or BCG infected macrophages and produced IFN-γ and TNF-α (30). Interestingly, an L^d restricted epitope spanning the same aa 161–168 (PTLIGLAM) was identified in Ag85A DNA vaccinated BALB/c mice, which did not cross-react with the corresponding Ag85B peptide (because of the Thr162Ser shift).

A comprehensive epitope mapping to HLA-A*0101, A*0201, A*1101, A*2402, B*0702, B*0801, and B*1501 of Ag85B was published in 2007 (31). Affinity and half-life (t_1/2 off-rate) analysis for individual peptide species on HLA-A and HLA-B molecules revealed binding ranges between 10⁻³ and 10⁻⁷ M. After selection of the best matches, major histocompatibility complex class I/peptide tetramer complexes were constructed to measure the CD8⁺ T-cell responses directly ex vivo in PBMC derived from 57 patients with acute pulmonary tuberculosis. Three patterns of (allele-) specific CD8⁺ recognition were identified: (a) Focus on one dominant epitope, (b) Co-dominant recognition of two distinct groups of peptides, and (c) Diverse and broad recognition of peptides (presented by HLA-A*0201). Peptides that bound with slow off-rates to class I alleles, that is HLA-A*0201, were associated with low frequency of CD8⁺ T-cells in PBMCs from patients with tuberculosis. HLA-B alleles showed fast off-rates in peptide binding and restricted high numbers (up to 6%) of antigen-specific CD8⁺ T-cells in patients with pulmonary tuberculosis (31). Functional analysis (in vitro IL-2 and IFN-γ production) revealed that tetramer-binding T-cells in PBMCs from these patients were little or not responsive to the nominal peptide epitope, confirming the notion of a deficient Ag85 specific T-cell response in TB patients. The study focused on TB patients and not on latently infected subjects, which could have been more relevant in the context of TB vaccine development.

In 1994, Roche et al. reported on the T-cell determinants of Ag85B of M. bovis (MPB 59) in BCG vaccinees and TB patients. The mature 85B protein of M. bovis (MPB59) has a high degree of amino-acid identity with the M. bovis 85A protein (76%) and the Mtb 85B (99%) and Mtb85A (76%) proteins. Proliferative assays with recombinant MPB59 demonstrated that PBMC from 95% of BCG vaccinees and 52% of tuberculosis patients responded to the whole mature protein. Using a set of synthetic 20-mer peptides, five peptides were found to be recognized in more than half of the MPB59 responders. The T-cell-reactive regions were essentially identical in the M. bovis and Mtb 85B proteins. Subjects with a variety of HLA-DR phenotypes responded to a number of these peptides and there was no difference in the pattern of responses between BCG vaccinees and TB patients. A promiscuous recognition pattern was observed in response to peptides spanning aa 51–70 (recognized by 87% of the responsive BCG vaccinees and 93% of the responding TB patients) and aa 11–30 (recognized by 73% of the responders). Peptides in the C-terminal region (aa 131–150 and aa 191–210) were more frequently recognized by patients than by BCG vaccinees (32). More recently, Finan et al. reported on 236 healthy Gambian babies vaccinated at birth with M. bovis BCG (33). Using a whole blood assay 2 months after vaccination, cytokine analysis showed that 89% of the babies produced Ag85 complex specific IFN-γ responses, albeit that response varied up to 10 log-fold within this population and 25 and 31% of the babies also produced detectable levels of the Th2 cytokines IL-5 and IL-13, respectively. Unfortunately, T-cell epitopes were not mapped in this study.

A. Geluk et al. reported on the identification of two HLA-A*0201 restricted CD8⁺ T-cell epitopes of Ag85B using pDNA vaccination encoding Ag85B of HLA-A2/K^b transgenic mice. (34). HLA-A*0201 is one of the most prevalent class I alleles, with a frequency of over 30% in most populations. The two peptides spanned aa 145–152 FIYAGSLS and aa 199–207 KLVANNTRL. As already mentioned, the first region is also recognized by K^d restricted CD8⁺ T-cells of Ag85A/B DNA vaccinated mice. The second peptide differs from that of Ag85A only in position 201, with a leucine in the Ag85A and a valine in the Ag85B sequence (both with a non-polar side chain), change that does not affect the binding affinity for HLA-A*0201 (34). As precursor frequencies of these cells were low in the periphery of human BCG vaccinees, restimulation with M. bovis BCG was necessary to visualize the cells by tetramer staining. Stable human CD8⁺ T-cell lines were generated against the two peptides, using CD4 depletion and peptide-pulsed autologous Dcs derived, from HLA-A*0201+ BCG-responsive donors. These T-cell lines were able to lyse HLA-A*0201+ peptide-pulsed targets and produced the pro-inflammatory cytokines IFN-γ and TNF-α. The group of H. Dockrell also demonstrated Ag85A specific CTL responses using BCG-specific cell lines generated from PBMC of BCG-vaccinated donors stimulated for 2 weeks with live M. bovis BCG in the presence of IL-2 and IL-7 (35). In this study, two HLA-A*0201 restricted epitopes were identified, one spanning aa 5–13 GLPVEYLQV and the other spanning aa 199–207 KLIANNTRV. This second peptide spans exactly the same region of Ag85B identified by Geluk et al., using HLA-A2/K^b transgenic mice, suggesting the existence of a cross-reactive CTL epitope in this region for Ag85A and Ag85B.

In 1994, we also reported on T-cell epitope mapping of MtbAg85A using PBMCs from healthy lepromin-positive volunteers and from patients with leprosy. As for tuberculosis, peptide recognition was largely promiscuous, with a variety of HLA haplotypes reacting to the same peptides. However, despite a 90% homology between the 85A proteins of M. leprae and Mtb, the peptides recognized were different. PBMC from lepromin-positive healthy contacts reacted against peptide 2 (aa 11–30), peptide 5 (aa 41–60), and peptides 25 and 26 (aa 241–270). PBMC from paucibacillary patients reacted preferentially against peptide 1 (amino acids 1–20) and peptide 5. Multibacillary patients were not reactive to Ag85 or the Ag85A peptides (23). It is interesting to note that responses to aa 11–30 were also identified in BCG vaccinees (27, 32). It is well known that BCG vaccination exerts some degree of protection against leprosy and that a second BCG immunization can increase this protection (36).