Six- to eight-week old CB6F1 mice (male C57BL/6 × female BALB/c F1) were purchased from Charles River. All animal studies were ethically approved by the State Office for Health and Social Services, Berlin, Germany. We designed two protocols, one for single vaccination and one for prime-boost vaccination, each consisting of 10 mice per timepoint divided into two groups of five mice for manageability and performed in two separate experiments.

The full-length H56 and H56_E cDNA (33) was codon optimized, and inserted at the 3′end of TTFC cDNA (44, 45) into the pVAX1 vector (Invitrogen) as described previously (43). DNA tattoo immunization was performed with 15 µl cDNA (2 µg/µl) in Tris-EDTA buffer with a 9-needle bar mounted on a tattoo rotary device (Cheyenne) adjusted to 100 Hz at 1 mm depth for 1 min (46), and the mice were under isoflurane anesthesia.

The Mtb H37Rv (American Type Culture Collection; catalog no. 27294) and BCG Danish 1331 (BCG SSI) (American Type Culture Collection; catalog no. 35733) were grown in Middlebrook 7H9 broth (BD) supplemented with albumin-dextrose-catalase enrichment (BD), 0.2% glycerol, and 0.05% Tween 80 or on Middlebrook 7H11 agar (BD) containing 10% (vol/vol) oleic acid-albumin-dextrose-catalase enrichment (BD) and 0.2% glycerol. BCG was grown to mid-log phase, washed with phosphate-buffered saline (PBS) and stored at −80°C in PBS/10% glycerol. BCG was washed in PBS before vaccination and administered at a dose of 10⁶ CFU in 100 µl for s.c. immunization and 10⁶ CFU in 25 µl for the i.d. tattoo. Aerosol challenge with Mtb was performed using an inoculum of 20–50 CFU.

The s.c. injection of BCG is known to lead to an approximately 1 log reduction in bacterial burden in mouse models (47, 48) and was included as the standard against which the efficacy of i.d. tattoo immunization with BCG or DNA were compared. The DNA vaccines included H56, a DNA vector containing the full-length codon-optimized H56 gene fused to TTFC cDNA, and H56_E, in which six CD8⁺ T cell epitope flanking sequences have been optimized to enhance proteasome-mediated processing (43), and were administered at day 0, 3, and 6. BCG was administered s.c. at day 0. One group of mice was also immunized by i.d. tattoo with BCG (day 0 only), in order to compare s.c. and i.d. administration of BCG. Immune responses were measured at day 21 and the efficacy of the methods was tested by subjecting the mice to an aerosol challenge with Mtb at day 60.

Two groups of mice were s.c. immunized with BCG at day 0, and one of these was boosted by i.d. H56_E DNA tattoo at day 40 for a heterologous immunization. Another group was primed and boosted i.d. with H56_E cDNA tattoo. Unvaccinated mice were included as controls. Immune responses were measured 2 weeks after the booster vaccination. Mice were infected with Mtb at day 100 by aerosol challenge and bacterial burdens and lung pathology were assessed at day 190.

For measurements of the bacterial load, lungs and spleens were harvested and serial dilutions were performed in PBS-0.05% Tween 80 and plated on Middlebrook 7H11 agar. The percentage of inflammation per lung area was measured in a blinded manner on formaldehyde fixed tissue sections stained in Giemsa. Whole lung sections were scanned with a ZEISS Axioscan Z1 driven by ZEN and lung pathology analysis was performed using Volocity (Perkin Elmer). Lung images shown were chosen based on the image analysis, with the lung having the value closest to the average result per group shown here.

All epitopes, as determined previously (43), were synthesized using Fmoc solid phase chemistry. The sequence enumeration of the synthetic peptides referred to the vaccine H56 (33). CD4⁺ T cell epitopes included H56_242–262 (Ag85B derived) and H56_288–307 (ESAT-6 derived). The CD8⁺ T cell epitopes included H56_354–363 (ESAT-6 derived) and five epitopes of Ag85B.

The eighty-five 15mer peptides used as pool to measure the total amount of H56-specific T cells (provided as a kind gift from Dr. Donatella Negri) spanned the entire amino acid sequence of H56, overlapping by 10 amino acid residues, and were synthesized by PRIMM Srl.

For intracellular cytokine production, splenocytes were plated overnight with or without 1 µg/ml peptide or peptide pool containing 1 µg/ml of each peptide at 37°C. During the last 4 h, brefeldin (5 µg/ml, Sigma) was added to wells incubated with peptide or phorbol myristate acetate (50 ng/ml, Sigma) and ionomycin (250 ng/ml, Sigma) or to cells incubated with medium alone. Cells were then stained for the cell surface markers named below, fixed with 2% paraformaldehyde, permeabilized with saponin buffer (saponin, 1 g/l; CaCl₂, 0.11 g/l; MgSO₄, 0.125 g/l; NaN₃, 0.5 g/l; bovine serum albumin (BSA) 1 g/l; 10 mM HEPES in PBS, pH 7.4), and stained for intracellular cytokines as described (43). The cell surface- and intracellular cytokine panel consisted of anti-TCRβ-A700 (Biolegend; clone H57-597), anti-CD4 Pacific Blue (Biolegend; clone GK1.5), anti-CD8 PerCP (Biolegend; clone 53-6.7), anti-IL-2 APC (Biolegend; clone JES6-5H4), anti-IFNγ PE-Cy7 (Biolegend; clone XMG1.2), anti-IL-17 PE (Biologend; clone TC11-18H10.1), and anti-TNF-α FITC (clone MP6-XT-22, grown in-house). Samples were acquired on an LSR II cytometer (BD Biosciences) with BD FACS Diva software and analyzed using FlowJo v10 (TreeStar).

Multiscreen ELISPOT plates (Millipore) were coated with 2 µg/ml anti-mouse IFN-γ (clone AN18, grown in-house) in PBS overnight at 4°C. Wells were washed and blocked with 5% BSA/PBS. 5 × 10⁵ erythrocyte depleted lymph node cells were plated with or without 1 µg/ml synthetic peptide overnight in Iscove’s modified Dulbecco’s medium (IMDM) with 10% fetal calf serum (FCS) Pen-Strep at 37°C. Plates were washed with PBS plus 0.01% Tween 20 (PBS-T), and IFN-γ was detected with biotinylated anti-IFN-γ (clone XMG1.2), followed by alkaline–phosphatase (AP)-conjugated streptavidin (homemade) in PBS-T supplemented with 2% BSA. The assay was developed with NBT/BCIP substrate (Thermo Fisher Scientific) and analyzed using an Immunopost S6 Ultra-V Analyzer (Cellular Technology Limited).

Spleens and inguinal lymph nodes were collected and single-cell suspensions were generated in IMDM 10% FCS Pen-Strep. Cells were surface stained to quantify cell populations. T cell panel: anti-CD3 Alexa 700 (BioLegend; clone 17A2), anti-CD4 PE-Cy7 (BioLegend; clone RM4-5), anti-CD8 V500 (BD Horizon, clone 53-6.7), anti-CD62L APC (BD Pharmingen; clone MEL-14), anti-CD44 Pacific Blue (clone IM7, grown in-house), anti-CXCR5 PE (BioLegend; clone J252D4), and anti-CCR7 PerCP (BioLegend; clone 4B12). B cell panel: anti-B220 PE (BioLegend; clone RA3-6B2), anti-CD38 APC (BioLegend; clone 90), anti-Fas PE-Cy7 (BioLegend; clone Jo2), anti-GL7 FITC (BioLegend; clone GL7), anti-MHCII Pacific Blue (BioLegend; clone M5/114.15.2), anti-CD19 Alexa 700 (BioLegend; clone 6D5), and anti-CD138 Percp (BioLegend; clone 281-2). CD4⁺ T_CM were CD3⁺CD4⁺CD44^highCD62^high, CD4⁺ T_EM were CD3⁺CD4⁺CD44^highCD62L^low, T follicular helper cells (T_FH) were CD3⁺CD4⁺CXCR5⁺, germinal center (GC) B cells were B220⁺CD19⁺GL7⁺Fas⁺, and plasma cells B220⁺CD19⁺CD138⁺CD38⁺.

Mycobacteria-, Ag85B- and ESAT-6-specific antibodies in serum were measured by indirect enzyme-linked immunosorbent assay using Mtb H37Rv lysate (BEI Resources), recombinant Ag85B (BEI Resources), or recombinant ESAT-6 (BEI Resources) to coat and AP-labeled anti-mouse IgG, IgG1, and IgG2c for detection (SouthernBiotech).

The scripts for Figures 1 and 4 are available upon request. FACS data visualization was designed using R-package ggplot2 (v2.1.0; Springer-Verlag New York, NY, USA, 2009.) in R programming language [v3.2.3; R Development Core Team (2008)]. The mean values of percentages of CD4⁺ or CD8⁺ T cells producing cytokines after peptide stimulation were normalized to control samples stimulated with medium and visualized using heat maps. ELISpot data visualization was designed using R-package ggplot2 (v2.1.0; Springer-Verlag New York, NY, USA, 2009) in R programming language [v3.2.3; R Development Core Team (2008)]. The mean numbers of IFN-γ-producing cells per million cells after peptide stimulation were normalized to control samples stimulated with medium. For visualization purposes, the mean value for each experimental condition was transformed to square root of mean response signal and displayed as a heat map.

Data were tested for normality with Levene tests. Kruskal–Wallis with Dunn’s multiple comparison test was used to analyze differences in bacterial burdens between groups. Differences in immunological parameters between groups were analyzed using a one-way ANOVA with Tukey’s multiple comparisons test. p < 0.05 was considered significant. The differences in cytokine frequencies were assessed by creating linear regression models with treatment as the predictor and measured cytokine frequency as the dependent variable. The models were created for CD4⁺ cells, CD8⁺ cells, all peptide stimulations, and measured cytokines separately. Benjamini–Hochberg method was used to correct for multiple testing. p-values <0.05 were considered significant.

Mice were immunized with BCG (s.c. or i.d.), with H56, a DNA vector containing the full-length codon-optimized H56 gene fused to TTFC cDNA, or with H56_E, in which six CD8⁺ T cell epitope flanking residues in the H56 sequence were optimized to enhance proteasome-mediated processing (43) (Figure 1A). Previously, we identified seven novel CD8⁺ T cell epitopes in the H56 fusion protein (43). In order to compare Ag85B- and ESAT-6-specific T cell responses induced by the cDNA vaccines and BCG, we measured peptide-specific immune responses after the different immunization strategies, including responses to an H56 peptide pool. Splenocytes or lymph node cells from vaccinated mice were re-stimulated ex vivo with Ag85B- and ESAT-6-derived CD8⁺ (43) and CD4⁺ (49, 50) T cell epitopes and responses in the spleen were measured using intracellular cytokine staining and flow cytometry (Figures 1B–D) or in the draining lymph nodes by IFN-γ ELISpot (Figure 1E). The mean percentages of cytokine-producing CD4⁺ or CD8⁺ T cells after peptide stimulation were normalized to control samples stimulated with medium and visualized using heat maps (Figures 1C,D). Results of statistical analysis are shown in Table S1 in Supplementary Material. Incubation of splenocytes with a pool of overlapping 15-mers covering the entire H56 sequence induced increased IFN-γ production by CD4⁺ T cells from H56 and H56_E i.d. vaccinated mice compared to unvaccinated controls (Figure 1C). H56 cDNA immunized mice had increased IFN-γ production in response to both the Ag85B-derived peptide H56_242–262 and the ESAT-6-derived peptide H56_288–307. No differences in TNF-α, IL-17, or IL-2 secretion by spleen CD4⁺ T cells were detected between the groups (Figure 1C). Differences in CD8⁺ T cell IFN-γ production were not significant, although there was a trend toward increased recognition of the H56 protein (as measured by responses to the peptide pool) in the H56 or H56_E cDNA i.d. immunized mice (Figure 1D). IFN-γ ELISpot analysis of total lymph node cells also demonstrated increased numbers of cells producing IFN-γ in the H56 cDNA i.d. vaccinated mice after stimulation with the H56 peptide pool compared to both unvaccinated and BCG s.c. vaccinated mice (Figure 1E; Table S2 in Supplementary Material). Overall, the H56 cDNA tattoo immunization increased H56-specific IFN-γ-producing T cell responses.

T cell memory populations and B cell responses after vaccination were measured by flow cytometry at day 21 in spleens and draining lymph nodes (Figure 2A). Both s.c. and i.d. BCG immunization resulted in increased percentages of CD4⁺ T_EM among spleen CD4⁺ T cells compared to unvaccinated mice (Figure 2B), with CD4⁺ T_FH showing similar trends (Figure 2C). Frequencies of CD4⁺ T_CM were higher in the spleens of DNA i.d. vaccinated mice than in BCG s.c. vaccinated mice (Figure 2D). T_EM responses differed in draining lymph nodes in that only H56_E immunized mice had increased T_EM (Figure 2E). A similar pattern was observed for T_FH in lymph nodes compared to spleen, with all responses increased in all vaccinated groups compared to unvaccinated controls (Figure 2F). This pattern was also seen for T_CM in draining lymph nodes (Figure 2G). There were no significant differences in frequencies of CD8⁺ T_EM or CD8⁺ T_CM between the groups (data not shown).

Humoral responses may contribute to protection against TB (51–53). We measured the frequencies of plasma cells and GC cells among B cells by flow cytometry (Figure 2H) in spleens (Figures 2I,J) and draining lymph nodes (Figures 2K,L). Frequencies of GC B cells were increased in spleens of both groups vaccinated with BCG compared to unvaccinated mice (Figure 2I), with a similar trend in plasma cells (Figure 2J). In the draining lymph nodes, GC B cells appeared increased in all immunized groups (Figure 2K) while plasma cells were similar between groups (Figure 2L). Antibody responses were very variable between animals, possibly because of the mixed background of CB6F1 mice, which are first-generation offspring of female BALB/c and male C57/BL6 mice. Although antibody responses varied between mice, the trend was directed toward increased levels of mycobacteria-specific IgG, including IgG1 and IgG2c, in mice vaccinated with BCG, which expresses multiple mycobacterial proteins (Figure 3A) and increased levels of Ag85B-specific IgG, IgG1, and IgG2c in the H56 cDNA vaccinated group (Figure 3B). ESAT-6-specific antibodies were not detected (data not shown). Curiously, antibody levels were lower in mice immunized with H56_E DNA than with H56 DNA, suggesting that B cell epitopes may have been influenced by the sequence modification.

To evaluate the protective efficacy of the different vaccines, bacterial loads were measured in lungs and spleens at day 90 after infection with Mtb and lung pathology was quantified as cell infiltration per lung area in a blinded manner. Mice vaccinated with BCG s.c. benefited from an approximately 1 log reduction in bacterial load in the lung compared to unvaccinated mice, whereas mice vaccinated with BCG i.d. did not (Figure 3C). Bacterial loads were not decreased after i.d. immunization with H56 or H56_E cDNA. A similar trend was seen in the spleen (Figure 3D). Furthermore, mice vaccinated with BCG strains had strongly decreased lung inflammation compared to unvaccinated mice (Figures 3E,F). H56 cDNA immunization did not decrease lung pathology compared to unvaccinated mice; however, there was reduced lung pathology in H56_E cDNA immunized mice (Figures 3E,F). Overall, cDNA tattoo immunization i.d. with H56 constructs was not as protective as BCG s.c. against TB in our model, but H56_E immunization ameliorated pathology.

Although i.d. H56_E DNA tattoo vaccination did not reduce Mtb burdens, it ameliorated lung pathology, and we hypothesized that it may increase the efficacy of the BCG vaccine in a prime-boost system by boosting responses to Ag85B and inducing responses to the Mtb-specific antigens ESAT-6 and Rv2660c found in H56. Therefore, we tested a heterologous prime-boost vaccination regimen (Figure 4A). In addition, we tested whether repeating i.d. H56_E DNA tattoo immunization (homologous prime/boost) could improve its protective efficacy. Spleen H56-specific CD4⁺ IFN-γ responses were increased at day 14 after boosting of BCG with i.d. H56 and H56_E DNA tattoo immunization as well as by homologous H56_E cDNA immunization, compared to BCG-only immunized groups (Figure 4B; Table S3 in Supplementary Material). In addition, spleen H56-specific CD4⁺ TNF-α responses were increased in the homologous H56_E cDNA immunized group compared to the BCG s.c. immunized group, and in both groups boosted with H56_E DNA compared to the unvaccinated controls. The BCG/H56_E immunized group had increased frequencies of Ag85B-specific CD4⁺TNF-α⁺ cells. Homologous cDNA prime-boost induced increased percentages of H56-specific and ESAT-6-specific IFN-γ CD8⁺ T cells compared to both BCG s.c. immunized and unvaccinated groups (Figure 4C; Table S3 in Supplementary Material). BCG s.c. immunized mice had increased Ag85B-specific CD8⁺IFN-γ⁺ responses compared to unvaccinated mice, which were further boosted by H56_E cDNA compared to BCG alone. ELIspots on lymph node cells showed increased IFN-γ responses to Ag85B and ESAT-6 CD4 epitopes as well as the H56 peptide pool in mice that had received homologous H56_E cDNA prime-boost (Figure 3D; Table S4 in Supplementary Material). Analysis of T cell subsets demonstrated no differences in frequencies of spleen T_EM, T_CM, or T_FH after DNA boosting (Figures 4E–G), with a similar pattern in the draining lymph nodes (Figures 4H–J). Frequencies of CD8⁺ T_CM and CD8⁺ T_EM were not significantly different after different immunization strategies, similar to the single vaccination experiments (data not shown).

Frequencies of H56-specific multi-cytokine-producing CD4⁺ T cells after different vaccination regimes were analyzed (Figure 5), since cytokine-producing capacity has been correlated to differentiation state (40, 54, 55). BCG induced much lower frequencies of H56-specific cytokine-producing cells overall. Strikingly, there were differences in cytokine production profiles between groups receiving BCG alone, heterologous prime-boost with BCG and H56_E cDNA, or homologous H56_E cDNA prime-boost. Specifically, in the heterologous prime-boosted group, CD4⁺ T cells responding to Ag85B-derived H56_242–262 were primarily IFN-γ⁺TNF-α⁺, while in the H56_E/H56_E cDNA vaccinated groups they expressed a more mixed cytokine-producing profile. By contrast, CD4⁺ T cells responding to ESAT-6-derived H56_288–307 were TNF-α⁺IL-2⁺ in the BCG/H56_E group and had mixed profiles in the H56_E/H56_E and the BCG-only vaccinated groups. Responses to the peptide pool were different again, with BCG prime/DNA boost mice showing mostly IFN-γ⁺ cells, while the H56_E/H56_E vaccinated group had a mixed profile.

No significant differences were found in lymph node GC and plasma B cells between groups (data not shown), and antibody levels were also similar (Figures 6A,B) although the anti-mycobacteria IgG1 titer was increased in mice receiving a booster immunization with H56_E cDNA compared to the group vaccinated with BCG alone (Figure 6A). Titers of anti-mycobacteria antibodies in H56_E cDNA/H56_E cDNA prime-boost mice remained as low as in unvaccinated controls (Figure 6A), but Ag85B-specific IgG2c titers were raised (Figure 6B).

To determine the protective efficacy of the different prime-boost strategies, bacterial loads, and lung pathology were measured 90 days after Mtb challenge. BCG and BCG/H56_E cDNA immunized groups had reduced lung bacterial burdens compared to the unvaccinated mice; however, there was no significant improvement in bacterial loads after boosting with H56_E cDNA (Figures 6C,D). Boosting H56_E cDNA prime with a homologous immunization did not increase efficacy, and mice were not protected over unvaccinated controls. Bacterial loads in spleens followed a similar pattern (Figure 6D), although vaccination was not as protective as in the lung. Quantification of infiltrating cells in the lungs by image analysis also revealed H56_E homologous prime-boost mice to be the least protected (Figures 6E,F). Mice primed with BCG and boosted with H56_E cDNA were the only group to show significant improvement in lung pathology compared to the unvaccinated group and tended to have less cell infiltration than the BCG-vaccinated mice too, suggesting that the H56-specific T cell responses associated with boosting with H56_E cDNA did have a beneficial effect in ameliorating disease. Overall, protective efficacy relied on the presence of BCG in the vaccine regimen, suggesting that broader anti-mycobacterial responses may be required for optimal protection in addition to the generation of highly specific responses against particular antigens.