The M. bovis BCG Tokyo 172 strain was cultured on Sauton medium (26) as a pellicle according to the approved production procedure.

A lot of BCG culture filtrate (80 L), which was a by-product from the validated BCG vaccine production process, was collected at the early stationary phase using centrifugal ultrafiltration (Pelicon cassette; 5 kDa; Merck Millipore Co. Ltd., Burlington, MA, USA). Saturated ammonium sulfate solution was added to obtain a 50% saturation. After storage at 9°C for 16 h, precipitated proteins were recovered via centrifugation at 10,440 × g and 4°C for 30 min. The residue was washed three times with the 50% saturated ammonium sulfate solution and dissolved in phosphate-buffered saline (PBS). After desalting through dialysis, the protein amount was determined using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) to obtain 1.04 g CFP. These procedures are repeated for preparation of another lot of CFP. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and two-dimensional PAGE were performed as previously described (27). A flask of BCG culture on the Sauton medium was heated at 90°C for 1 hour. After heat treatment, residual BCG was filtered out and then proteins were purified as described for CFP. This preparation of CFP with heat-treatment was named CFPH.

Totally 277 C57BL/6j and 30 BALB/c female six-week-old mice were purchased from Japan SLC, Inc., Shizuoka, Japan. OT-I mice were provided by the National Institute of Infectious Diseases, Japan and obtained through home breeding in the animal experimentation facility of Japan BCG Laboratory. Animals were kept under specific pathogen-free conditions. All animal experiments were approved by the Institutional Animal Care and Use Committee of Japan BCG Laboratory.

C57BL/6j mice were subcutaneously inoculated with 1 × 10⁶ colony forming units of BCG in 0.1 mL PBS. Six weeks after BCG priming, the mice were boosted with 50 µg CFP, PPD, or CFPH three times at two-week intervals. The mice were euthanized two weeks after the last boost. Next, the lungs were removed and used to prepare cell suspensions at 5 × 10⁶ cells/mL in the culture medium. Lung cells were suspended to 2 × 10⁵ cells/well in RPMI-1640 medium (Fujifilm Wako, Tokyo, Japan) with 10% fetal bovine serum (FBS) (Thermo Fisher Scientific), 1% penicillin/streptomycin (Thermo Fisher Scientific), and 50 µM 2-mercaptoethanol (2ME) (Thermo Fisher Scientific). They were subsequently stimulated with 10 µg/mL GM10 peptide [CD8+ T cell epitope of mycobacterial Mtb32a protein (PepA), GAPINSATAM (28)], or 10 µg/mL peptide25 [CD4+ T cell epitope of Ag85B, FQDAYNAAGGHNAVF (29)] on ELISPOT plates (Mabtech, Nacka Strand, Sweden). The cells were incubated at 37°C for 24 h. Next, plates were washed with PBS and incubated with 1 µg/mL biotinylated anti-mouse IFN-γ mAb R4-6A2 (Mabtech) in PBS containing 0.5% FBS for 2 h at 25°C. Then, 100 µL streptavidin-alkaline phosphatase (Mabtech) was added to each well at a 1/500 dilution. After 1 h, spots were visualized using a peroxidase stain diaminobenzidine kit (Nacalai Tesque, Kyoto, Japan) and counted using the KS ELISPOT reader (ZEISS, Oberkochen, Germany).

OVA was purchased from Merck Millipore. MHC class I-binding OVA-peptide (SIINFEKL) and B16 melanoma peptide TRP-2 (VYDFFVWL) were purchased from Medical & Biological Laboratories Co., Ltd. (Tokyo, Japan). Monoclonal antibodies, including anti-CD11c (N418), anti-MHC class I (AF6.88.5.5.3), anti-MHC class II (M5/114.15.2), and CD86 (B7-2), were purchased from Thermo Fisher Scientific to analyze BMDC surface antigens.

EL-4 (Public Health England, Salisbury, UK) and Colon-26 (Tohoku University Cell Resource Center for Biomedical Research/Cell Bank, Miyagi, Japan) cells were grown in RPMI-1640 medium with 10% FBS, 1% penicillin/streptomycin, and 50 µM 2ME. EG7 cells (ATCC CRL-2113), which are the OVA gene transfectants of EL4 cells, were grown in RPMI-1640 medium with 40 µg/mL G-418 sulfate (Fujifilm Wako). Furthermore, B16F10 cells (ATCC CRL-6475) were grown in D-MEM (Fujifilm Wako) with 10% FBS, 1% penicillin/streptomycin, and 2ME. BMDCs were prepared from C57BL/6j mice as previously described (30). Brifely, the mouse bone marrow cells were grown in an RPMI-1640 medium supplemented with 10 ng/mL recombinant mouse granulocyte/macrophage colony-stimulating factor (rGM-CSF) (Thermo Fisher Scientific) and recombinant mouse IL-4 (rmIL-4) (R&D Systems, Minneapolis, MN, USA) at 37°C under 5% CO₂ in a humidified incubator. On Day 3, nonadherent cells were collected from the supernatant, gently washed, and added to fresh RPMI-1640 medium supplemented with 10 ng/mL rGM-CSF and 4 ng/mL rmIL-4. On Day 6, nonadherent cells were collected via centrifugation, resuspended in fresh RPMI-1640 medium with rGM-CSF and rmIL-4, and cultured for an additional 24 h in cell culture dishes. CD11c expression was confirmed in the cells, and CD11c positivity was >92%.

Immature BMDCs (1 × 10⁶ cells/mL) were plated in 96-well round-bottom tissue culture plates with 10 µg/mL OVA mixed with CFP (200 µg/mL) or without. The cells were subsequently cultured at 37°C under 5% CO₂ for 24 h. OVA-treated BMDCs were cultured with 10 µg/mL mitomycin C (MMC) (Merck Millipore) for 45 min. The cells were washed, suspended in fresh medium with 2.5 × 10⁶ cells/mL CD8+ T cells from OT-I mice, and cultured for 24 h under the same conditions. CD8+ T cells were isolated from the spleens of OT-I mice using an EasySep Mouse CD8+ T cell Isolation Kit (Veritas, Tokyo, Japan). Purities of the CD8+ T cells were >95%, as determined using flow cytometry (FACSCalibur, BD Bioscience, San Jose, CA, USA). The concentration of IL-2 in the culture supernatants was measured according to the manufacturer’s instructions using an enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems).

On Days 0 and 10, C57BL/6j mice were immunized through subcutaneous injection of OVA (10 µg/mouse) or B16F10 cell lysate (equivalent to 1 × 10⁶ cells/mouse) mixed with or without CFP (200 µg/mouse) on the abdomen. BALB/c mice were immunized with Colon-26 cell lysate (equivalent to 1 × 10⁶ cells/mouse) using the same procedure as the OVA and B16F10 lysate. To prepare a tumor cell lysate, B16F10 or Colon-26 cells were divided into 1.5-mL tubes (5 × 10⁷ cells per 500 µL PBS/tube) and disrupted using a Bioruptor (Cosmo Bio, Tokyo, Japan). Disrupted cells were centrifuged at 14,000 × g for 15 min at 4°C, and the supernatant was used as a cell lysate (100 µL is equivalent to 1 × 10⁶ cells).

Immune mouse spleens were removed seven days after the second immunization with OVA. Next, spleen cells (1 × 10⁷) were cultured in vitro with OVA-peptide (10 µg/mL) at 37°C for two days. IFN-γ production of the spleen cells in the culture supernatant was quantified using an ELISA kit (R&D Systems) according to the manufacturer’s instructions.

Seven days after the second immunization with OVA or B16F10 cell lysate, immune mouse spleen cells (1 × 10⁷) were cultured in vitro with MMC-treated EG7 cells (6 × 10⁵) or TRP-2 peptide (10 µg/mL), respectively. The MMC-treated EG7 cells were obtained by culturing EG7 cells in 50 µg/mL MMC for 45 min and washing them. Floating cells were collected after five days of culture, and lymphocytes were prepared using Lympholyte M Cell Separation Media (TCP Analytical group, Isle of Palms, SC, USA) for effector cells. EG7 and B16F10 or TRP2 peptide-pulsed EL-4 cells were used as target cells for OVA- and TRP-2-specific cytotoxicity, respectively. The TRP-2 peptide-pulsed EL-4 cells were obtained by culturing EL-4 cells in 5 µg/mL TRP-2 for 90 min and washing them. The antigen-specific cytotoxicity evaluation was performed using a 7-AAD/CFSE Cell-Mediated Cytotoxicity Assay Kit (Cayman Chemical Company, Ann Arbor, MI, USA). Target cells were stained with 5- (6)-carboxyfluorescein diacetate succinimidyl ester (CFSE) and incubated at 37°C for 30 min. CFSE-stained target cells were incubated with various effector cells to target cell ratios at 37°C for 3 h. They were subsequently stained with 7-amino-actinomycin D (7AAD). Finally, CFSE+7AAD+ cells were measured using FACSCalibur (BD Bioscience). Data were analyzed using the FlowJo software (BD Biosience).

The prevention model was constructed by subcutaneously administering EG7 (1 × 10⁶ cells/mouse: n = 10), B16F10 (1 × 10⁵ cells/mouse: n = 20) and Colon-26 cells (2.5 × 10⁵ cells/mouse: n = 10) in OVA-, B16F10 cell lysate-, and Colon-26 cell lysate-immunized mice, respectively. The cells were administered on the ventral side seven days after the second immunization. Tumor size was measured twice weekly using a microcaliper and calculated as follows: tumor size (mm³) = length × width². To ensure a humane endpoint, the animal was considered dead and euthanized by hyperanesthesia if body weight decreased by > 10% in one week.

To construct the treatment model, EG7 (1 × 10⁶ cells/mouse: n = 10) or B16F10 (1 × 10⁵ cells/mouse: n = 10) were subcutaneously transplanted on the ventral side of C57BL/6j mice. On Days 7, 10, 14, 17, 21, and 24 after tumor transplantation, OVA (10 µg/mouse) or B16F10 cell lysate (equivalent to 1 × 10⁶ cells/mouse) mixed with or without CFP (200 µg/mouse) were intradermally injected around the tumor for the EG7- or B16F10-challenged mice, respectively. Assessment of the tumor size and humane endpoint were as described. For the improved treatment model, BCG Tokyo (0.1 mg/mouse: n = 6) (Japan BCG Laboratory, Tokyo, Japan) was subcutaneously administered on the back of the neck in C57BL/6j mice. Six weeks after BCG injection, EG7 cells (1 × 10⁶ cells/mouse) were subcutaneously transplanted into the left ventral side on Day 0. On Days 7, 10, 14, 17, 21, and 24 after EG7 cell transplantation, OVA (10 µg/mouse) mixed with or without CFP (200 µg/mouse) was intradermally injected around the tumor to treat the mice. Assessment of the tumor size and humane endpoint were as described.

Data are presented as means ± SD. Statistical analyses were performed through the Mann-Whitney U test using GraphPad Prism 9 (GraphPad, Software La Jolla, CA, USA). Significant differences compared with the control are indicated by asterisks. P < 0.05 was considered statistically significant. *P < 0.05; **P < 0.01; n.s., not significant.

A flow chart describing the preparation of the BCG CFP is shown in Supplementary Figure 1 . In this study, two-dimensional gel electrophoresis revealed >100 protein spots ( Supplementary Figure 1 ). CFP induced PepA (GM10 CTL epitope)-specific CD8+ T cells in mice primed with the BCG Tokyo vaccine strain and intranasally boosted with CFP three times. Although purified protein derivatives (PPD)-specific CD4+ T cell induction was observed ( Supplementary Figure 2 ), this result was not observed with PPD boosted using gamma-interferon (IFN-γ) enzyme-linked immunospot (ELISPOT) assays. Granzyme B expression indicated that the induced CD8+ T cells would be functional CTLs. PPD and CFP possess similar components. Therefore, we suspected that the difference could be due to heat treatment during PPD preparation. To clarify this point, we tested the effect of CFP heat treatment ( Supplementary Figure 1 ) in the same immunization protocol. CFPH did not induce the PepA-specific CTLs ( Supplementary Figure 2 ), suggesting that the presence of a higher-order protein structure is involved in CFP activity. Based on these observations, we hypothesized that the PepA-specific CTLs might be induced by an unknown protein.

To understand CD8+ T cell activation by CFP, we first examined its action in vitro by culturing bone marrow-derived dendritic cells (BMDCs) from C57BL/6j mice with CFP. The expression of major histocompatibility complex (MHC) class I was not significantly different, whereas that of MHC class II was significantly reduced compared with the BMDCs cultured without CFP. In contrast, CFP activated CD86, a ligand of CTLA4 on T cells (31) ( Figure 1A ). In addition, BMDCs showed significant production of proinflammatory cytokines such as tumor necrosis factor (TNF)-α, interleukin (IL)-6, and IL-12p40 ( Figure 1B ). Moreover, an inhibitory cytokine IL-10 was produced (data not shown). The lower expression of MHC class II on BMDCs may be due to the increased IL-10 production by these cells. Furthermore, ovalbumin (OVA)-specific CD8+ T cell activation was examined by coculture with BMDC incorporated with CFP and OVA, which mimicked an in vivo CTL reaction. In this system, CFP clearly enhanced IL-2 production in the BMDC with a dose-dependency ( Figure 1C ), suggesting that CFP-induced CD8+ T cell activation may be involved in the activation of antigen presentation cells (APCs) such as dendritic cells.

Here, we also examined CFP action in vivo. To directly prove our hypothesis, we mixed OVA, which is a protein antigen unrelated to BCG, with CFP to examine OVA-specific CTL induction. Although EL4 cells were not lysed as a target cell control, EG7 cells were killed in an effector vs target ratio-dependent manner ( Figure 2A ), indicating that the OVA-specific CTLs were highly induced. IFN-γ+ T cell induction and EG7 tumor growth suppression were reproducible even when different lots of CFP were used in the independent experiments ( Supplementary Figure 3 ). In addition, splenocytes from OVA/CFP-immunized mice showed enhanced IFN-γ release into the culture supernatant ( Figure 2B ). These data supported our hypothesis and clearly demonstrated that CFP acts as a CTL adjuvant. However, CFP did not show an adjuvant effect in the induction of OVA-specific antibody production (data not shown). To clarify the CTL adjuvant activity of CFP, we induced CTLs against another antigen. B16F10 melanoma cell lysate was used with CFP in the immunization of mice, and the CTL response to tyrosinase-related protein-2 (TRP-2) tumor antigen (32) was analyzed. CTL induction was demonstrated in killing assays using both B16F10 ( Figure 2C ) and peptide-pulsed EL4 cells as target cells ( Figure 2D ). This suggests that CFP can induce CTLs against various protein antigens.

To examine whether the CTL response enhanced with CFP exhibited a protective effect, we analyzed the preventive effect of CFP against tumor cell growth in mice and EG7 tumor growth was markedly suppressed by immunization ( Figure 3A ). We adopted B16F10 cells to confirm the applicability of the CFP adjuvant not for an artificial EG7 tumor but for natural cancer cell growth suppression. CFP mixed with B16F10 cell lysate was administered to mice, and the mice were subsequently challenged with B16F10 cells. The immunized mice considerably suppressed B16F10 cell growth ( Figure 3B ). Moreover, the tumor cell lysate/CFP vaccine regimen effectively suppressed the growth of the Colon-26 murine colon carcinoma cell line. BALB/c mice immunized using Colon-26 cell lysate with CFP showed reduced tumor cell growth ( Figure 3C ). Thus, the CTL adjuvant activity of CFP was unaffected by MHC class I restriction because tumor growth suppression was demonstrated in both C57BL/6 and BALB/c mice. These data suggest that antigen-specific CTL induction by the CFP adjuvant could effectively prevent the growth of various tumor cells.

Mice were inoculated with EG7 cells and then treated with the OVA/CFP mixture several times from Day 7 post inoculation. The EG7 tumor size was considerably reduced in mice treated with the OVA/CFP mixture from 11 days post inoculation ( Figure 4A ). In addition, this treatment effect was enhanced by BCG priming ( Figure 4C ). Three out of six mice treated with the OVA/CFP mixture showed tumor regression 24 days after EG7 cell transplantation ( Figure 4D ). We applied the regimen in the B16F10 treatment models similar to the EG7 experiment to elucidate the tumor suppressive effect of CTLs. The B16F10 cell lysate and CFP mixture-immunized mice showed suppressed melanoma cell growth ( Figure 4B ). These data indicate that CFP could have potential as a CTL adjuvant for cancer immunotherapy.