Introduction

Personalized neoantigen mRNA vaccines showed high potency in early-phase clinical trials to treat advanced melanoma and pancreatic cancer1,2. In these trials, patients underwent surgical tumor removal followed by next-generation sequencing (NGS) to identify tumor-specific mutations. The neoepitopes surrounding tumor-specific mutations were predicted for their ability to bind MHC molecules. The top ranked neoepitopes were chosen to develop neoantigen mRNA vaccines for personalized cancer immunotherapy. Neoantigen mRNA vaccine (mRNA-4157) plus pembrolizumab were found to significantly prolong recurrence-free survival as compared to pembrolizumab alone in resected stage IIIB-IV melanoma2. Neoantigen mRNA vaccine (autogene cevumeran) plus atezolizumab and chemotherapy (mFOLFIRINOX) significantly extended median recurrence-free survival in vaccine responders (not reached, n = 8) than vaccine non-responders (13.4 months, n = 8) in resected pancreatic ductal adenocarcinoma (PDAC) patients1. Inspired by these promising results, phase 2 or 3 clinical trials are underway to further evaluate efficacy and safety of personalized neoantigen mRNA vaccines to treat advanced melanoma and pancreatic cancer. Neoantigen mRNA vaccines are also under active exploration to treat other cancer types, such as glioblastoma3.

Various strategies are under exploration to enhance neoantigen mRNA vaccine efficacy in tumor therapy. Better neoantigen screening algorisms are needed to increase prediction accuracy and mRNA vaccine response rates. A LinearDesign algorism was developed to optimize mRNA design and codon usage and substantially improve mRNA stability and protein expression4. This strategy was found to increase antibody titer by up to 128 times compared with codon-optimization benchmark for mRNA vaccines4. A modified computational workflow was developed to account for allele-specific anchor position preferences to enhance neoantigen prediction5. Besides prediction algorism improvement, the core-shell structure of lipopolyplex was used to formulate mRNA to prevent rapid degradation of mRNA with good safety profiles in preclinical and clinical studies6. Besides the above approaches, a constitutively active mutation (V155M) of the stimulator of interferon (IFN) genes (STING) was co-formulated with antigenic mRNA to significantly enhance antigen-specific CD8+ T cell responses and anti-tumor immunity in TC-1 tumor models7.

Vaccine adjuvants have been essential to develop protein/subunit vaccines by potentiation of antigen uptake and stimulation of dendritic cell (DC) maturation8,9. mRNA vaccines have no protein-based antigens available at the time of administration. Thus, traditional adjuvants that act solely by enhancing antigen uptake may not work for mRNA vaccines. Additionally, adjuvant-induced signaling pathways may interfere mRNA translation 10. Flagellin is a protein-based toll-like receptor (TLR) 5 agonist and has been actively explored as both vaccine adjuvants and carriers in vaccine development against bacterial, viral, and parasitic diseases11. Flagellin can also be prepared as mRNA platform for co-delivery with antigenic mRNA to minimize the potential negative effects on mRNA translation. Flagellin has been also explored as an immunomodulatory agent to modify the immunosuppressive tumor microenvironment or as an adjuvant to enhance vaccine-induced anti-tumor immunity11. Attenuated Salmonella typhimurium bacteria secreting Vibrio vulnificus flagellin B (FlaB) in tumor tissues were found to suppress tumor growth and metastasis in mouse models via TLR4 and TLR5-mediated two-step enhancement approach12. FlaB co-administration was found to potentiate human papillomavirus 16 E6 and E7 (E6/E7) peptide-based cancer vaccine efficacy by promotion of IFNγ-secreting CD8+ T cells responses13. Our previous studies found Salmonella typhimurium flagellin (FljB) could serve as an adjuvant and a carrier for cytotoxic T lymphocyte (CTL) epitope of ovalbumin (OVA) to elicit potent anti-tumor immunity in OVA-expressing B16F10 melanoma and E.G7 lymphoma models, while CTL epitope of OVA alone showed no significant anti-tumor effects14.

This study explored potential adjuvant effects of FljB mRNA to potentiate OVA mRNA-induced CTL responses and anti-tumor immunity in OVA-expressing B16F10 (B16F10-OVA) models. The relative potency of FljB co-expression to separate expression in enhancing OVA mRNA-induced CTL responses and anti-tumor immunity was also compared.

Materials and methods

Reagents

Lipid Nanoparticle (LNP-102) Exploration Kit was purchased from Cayman Chemical (Ann Arbor, MI, USA). The RiboGreen RNA Quantification Kit was purchased from Thermo Fisher Scientific (R32700, Waltham, MA, USA). Fluorescence-conjugated antibodies, anti-CD28 antibody, and brefeldin A were obtained from BioLegend (San Diego, CA, USA). ACK lysing buffer was purchased from Quality Biological (118-156-101, Gaithersburg, MD, USA). 3,3′,5,5′-Tetramethylbenzidine (TMB) substrate was purchased from Thermo Fisher Scientific (34028, Waltham, MA, USA). Horseradish peroxidase (HRP)-conjugated sheep anti-mouse IgG secondary antibody was purchased from Cytiva (NA931, Marlborough, MA, USA). Enzyme-linked immunosorbent assay (ELISA) Max™ Deluxe Set Mouse IL-6 was purchased from BioLegend (431304, San Diego, CA, USA). Dulbecco’s Modified Eagle Medium (DMEM) was purchased from ATCC (30-2002, Manassas, VA, USA). RPMI-1640 media was purchased from Thermo Fisher Scientific (21870-076, Waltham, MA, USA). Collagenase D was purchased from Millipore Sigma (11088866001, Burlington, MA, USA) and Dispase was purchased from Life technologies (17105-041, Carlsbad, CA, USA). Ficoll-Paque PREMIUM sterile solution was purchased from Cytiva (17544202, Marlborough, MA, USA).

Mice

Male C57BL/6 mice (6–8 weeks old) were purchased from the Jackson Laboratory (Bar Harbor, ME) and housed in the animal facilities of the University of Rhode Island. Mice were euthanized in their home cages by delivering compressed CO2 in gas cylinders Via a Euthanex® lid in the procedure room. All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Rhode Island (AN1415-009) and conducted in accordance with National and Institutional Guidelines and Regulations. Animal experiments were reported in accordance with the ARRIVE guidelines.

mRNA Preparation

OVA-encoding DNA was obtained from NCBI GenBank (MH360742.1). DNA encoding FljB of S. Typhimurium strain LT2 lacking D2 and D3 domains (FljB∆D2D3 or FljBΔ176−414) was obtained from our previous studies15. To obtain DNA encoding FljB-OVA fusion protein, OVA-encoding DNA was inserted to 3’-end of FljB∆D2D3 (short as FljB) after flexible linker (G4S)×4-encoding sequence. DNA encoding signal peptide (SP, MFVFLVLLPLVSSQCV) was added to 5’-end of OVA, FljB and FljB-OVA. The assembled sequences were codon-optimized for human expression (Table. S1). The various mRNA sequences were custom synthesized by Trilink Biotechnologies (San Diego, CA).

LNP encapsulation and characterization

Lipid Nanoparticle (LNP-102) Exploration Kit was used to encapsulate mRNA into LNPs according to manufacturer’s recommendations. In brief, lipids were dissolved in pure ethanol at molar ratios of 50:10:38.5:1.5 (SM-102:1,2-DSPC: cholesterol: DMG-PEG 2000). mRNA was dissolved in 50 mM sodium acetate (pH 5.0). The lipid mixture was mixed with the aqueous mRNA solution at a weight ratio of 10:1 (lipid: mRNA) and a volume ratio of 1:3 (ethanol: aqueous) using the NanoAssemblr® Ignite™ (Precision Nanosystems) at a total flow rate of 9 ml/min, followed by dialysis against PBS and then 8.7% sucrose in PBS. Following dialysis, LNPs were passed through a 0.22 μm filter and concentrated with 3 kDa molecular-weight-cutoff centrifugal filters. LNP sizes were measured by dynamic light scattering (DLS) analysis in Malvern Zetasizer and also by Cryo-EM. To measure the encapsulation efficiency, LNPs were treated with 2% Triton X-100 to disrupt the lipid structure. The total and unencapsulated mRNA concentrations were measured by RiboGreen RNA Quantification Kit. The encapsulation efficiency (EE%) was calculated as (total mRNA – unencapsulated mRNA)/total mRNA × 100%.

Immunization

Mice were intramuscularly immunized with LNPs that contained an equimolar amount of OVA mRNA (5 µg, 1483 NT), FljB-OVA mRNA (7.8 µg, 2325 NT), OVA mRNA (5 µg) admixed with FljB mRNA (3.7 µg, 1110 NT), or empty LNPs. Boost immunization was similarly conducted 3 weeks later.

Tumor challenge

B16F10-OVA cells were cultured in DMEM medium supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, and 100 U/mL penicillin and 100 µg/mL streptomycin. Cells were harvested at around 80% confluency. Cells were thoroughly washed in PBS and adjusted to 107 cells/mL. B16F10-OVA cells (106 in 100 µl volume) were subcutaneously injected into the right flank of mice.

Tumor volume measurement

Tumor length and width were measured with a digital caliper as in our previous report14. Tumor volume was calculated using the formula: v=\(\:\frac{1}{2}a{b}^{2}\), where a and b were long and short diameter of the tumor, respectively. Tumor-bearing mice were euthanized when reaching humane endpoints.

Antibody responses

Serum anti-OVA and anti-FljB antibody responses were measured using ELISA as previously described15. Briefly, 96-well ELISA plates were coated with 10 µg/ml OVA or 1 µg/ml FljB (full-length) and incubated at 4 °C overnight. After blocking with 5% non-fat milk, 4-fold serial dilutions of serum samples were added and the plates were incubated at room temperature for 90 min. After washing, HRP-conjugated anti-mouse IgG (NA931, Cytiva), subtype IgG1 (PA1-74421, Invitrogen), and IgG2c antibodies (A90-136P, Bethyl Laboratories) were added and the plates were incubated at room temperature for 1 h. After washing, TMB substrates were added and the plates were incubated in the dark for 15–30 min. After adding 2M H2SO4, optical absorbances were read at 450 nm and 570 nm in a microplate reader (Molecule Devices).

Cytokine measurement

Serum IL-6 were measured using Mouse IL-6 ELISA MAX Deluxe Set according to manufacturer’s recommendations. Briefly, 96-well plates were coated with capture antibody and incubated at 4 °C overnight. Following blocking with assay diluent A, standard and diluted serum samples were added and plates were incubated at room temperature for 2 h. After washing, detection antibody was added and plates were incubated at room temperature for 1 h. After washing, Avidin-HRP solution was added and plates were incubated at room temperature for 30 min. After washing, 3,3′,5,5′-tetramethylbenzidine (TMB) substrate was added and plates were incubated in the dark for 20 min. After adding stop solution, optical absorbances were read at 450 nm and 570 nm in a microplate reader (Molecule Devices).

Single-cell suspension Preparation

Splenocytes were prepared by passing spleens through 70 μm cell strainers followed by red blood cell (RBC) lysis with ACK buffer. Tumor infiltrating lymphocytes (TILs) were prepared as reported with slight modifications16. Briefly, tumor tissues were cut into small pieces followed by digestion with 0.2% Collagenase D and 0.6U Dispase in 37 °C water bath with constant shaking at 100 rpm for 1 h. Tumor tissues were then passed through 70 μm cell strainers for 3 times, followed by resuspension in 10 mL RPMI-1640 medium. Next, 4 mL Ficoll-Paque media was added to the bottom of a 15-mL conical tube. 10 mL tumor cell suspensions were gently layered on top of the Ficoll-Paque media. Tubes were centrifuged at 1,025 × g for 20 min at 20 °C (acceleration 1, deceleration 0). The mononuclear cell layers at the interface were transferred to new tubes followed by washing with RPMI-1640 medium three times.

Flow cytometry

Splenocytes or TILs were stimulated with 1 µg/mL endo-free OVA in the presence of 4 µg/mL anti-CD28 antibodies (37.51) overnight. Next day, brefeldin A was added 5 h before cell harvest. Splenocytes were stained with fluorescence-conjugated anti-CD4 (GK1.5) and anti-CD8 (53 − 6.7) antibodies, fixed and permeabilized, and then stained with fluorescence-conjugated antibodies against IFNγ (XMG1.2), granzyme B (NGZB), and perforin (S16009A). TILs were stained with fluorescence-conjugated anti-CD3 (17A2), anti-CD4 (GK1.5) and anti-CD8a (53 − 6.7) antibodies, fixed and permeabilized, and then stained with fluorescence-conjugated anti-IFNγ (XMG1.2), anti-granzyme B (NGZB) and anti-perforin (S16009A) antibodies. Cells were then subject to flow cytometry analysis in BD FACSVerse. Flow cytometry data were analyzed using FlowJo™ software (version 10, https://www.bdbiosciences.com/en-us/products/software/flowjo-v10-software). Unstained and single-color stained samples were used for compensation and gating adjustment.

Statistical analysis

Values were expressed as mean ± SEM (standard error of the mean). One-way analysis of variance (ANOVA) with Fisher’s LSD test was used to compare differences for more than two groups, except otherwise specified. P-value was calculated by GraphPad PRISM software (version 10.2.1, https://www.graphpad.com/) and considered significant if it was less than 0.05.

Results

mRNA vaccine Preparation and characterization

OVA mRNA, FljBΔD2D3 mRNA (short as FljB), FljBΔD2D3-OVA mRNA (short as FljB-OVA) were encapsulated in LNPs (Fig. 1A). FljB-OVA mRNA-encapsulated LNPs were 118 nm in diameter (Fig. 1B) and mainly spherical shaped (Fig. 1C). FljB and OVA mRNA-encapsulated LNPs showed a similar size and shape to FljB-OVA mRNA-encapsulated ones (data not shown).

Fig. 1
Fig. 1
Full size image

mRNA design and LNP characterization. (A). Illustration of OVA, FljBΔD2D3, FljBΔD2D3-OVA mRNA design (SP omitted). B-C. DLS (B) and Cryo-EM (C) analysis of FljB-OVA mRNA-encapsulated LNPs. Scale in C: 100 nm.

FljB co-expression potentiates Th2-biased IgG1 antibody responses

Mice were intramuscularly immunized with LNP(OVA mRNA), LNP(OVA mRNA)/LNP(FljB mRNA), LNP(FljB-OVA mRNA), or empty LNP. Serum anti-OVA IgG and subtype IgG1 and IgG2c antibody titers were measured 3 weeks later as shown in Fig. 2A. Incorporation of FljB mRNA in OVA mRNA immunization significantly reduced anti-OVA IgG antibody production regardless of co-expression or separate expression (left, Fig. 2B). Interestingly, FljB co-expression or separate expression significantly increased anti-OVA IgG1 antibody production with co-expression showing better enhancement (middle, Fig. 2B). FljB co-expression or separate expression profoundly reduced anti-OVA IgG2c antibody production (right, Fig. 2B). These results hinted FljB co-expression potentiated OVA mRNA-induced Th2-biased IgG1 antibody responses. We further measured anti-FljB antibody levels. FljB mRNA immunization with D2/D3 domains removed elicited baseline levels of anti-FljB antibody responses (Fig. 2C). Interestingly, FljB-OVA mRNA elicited potent anti-FljB antibody responses (Fig. 2C). FljB-based vaccines have been known to induce systemic IL-6 release14,17. Here, we found that FljB mRNA/OVA mRNA immunization significantly increased serum IL-6 levels 3 h after prime or boost (Fig. 2D-E). FljB-OVA mRNA immunization also increased serum IL-6 levels but to a far lesser extent than that induced by FljB mRNA/OVA mRNA immunization (Fig. 2D-E). Serum IL-6 levels significantly reduced 24 h after prime or boost and were still significantly higher in FljB mRNA/OVA mRNA group than that in OVA mRNA group after prime (Fig. 2D). Serum IL-6 levels showed no significant difference among groups 48 h after prime or boost (Fig. 2D-E).

Fig. 2
Fig. 2
Full size image

FljB co-expression potentiates Th2-biased IgG1 antibody responses. (A). Illustration of experimental schedules. Note: mice in empty LNP group were euthanized on day 59 and one mouse in OVA mRNA group was euthanized on day 62 due to reaching humane endpoint. All other mice were euthanized on day 64. (B). OVA-specific IgG (left) and subtype IgG1 (middle) and IgG2c antibody responses (right) after prime. (C). FljB-specific IgG antibody responses after prime. D-E. Serum IL-6 levels 3, 24, and 48 h after prime (D) or boost (E). Two-way ANOVA with Dunnett’s multiple comparison test was used to compare differences of other groups with OVA mRNA group in B-E. n = 5. *, p < 0.05; **, p < 0.01; ***, p < 0.001. NS, not significant.

FljB co-expression potentiates splenocyte CD8+ and CD4+ T cell responses

Mice were challenged with B16F10-OVA melanoma on day 39 (Fig. 2A). The majority of mice were euthanized on day 64 (Fig. 2A). Spleens were collected for analysis of cytokine-secreting CD8+ and CD4+ T cells after in vitro stimulation. As shown in Fig. 3A-B, FljB co-expression but not separate expression significantly increased granzyme B (GrB)+ CD8+ T cell levels. FljB co-expression increased GrB+ CD8+ T cells by 2.5 folds compared with OVA mRNA alone and twofolds compared with FljB separate expression (Fig. 3A-B). Similar results were observed for IFNγ+ CD8+ T cells and perforin+ CD8+ T cells. FljB co-expression increased IFNγ+ CD8+ T cell levels by 2.6 folds compared with OVA mRNA alone and 1.9 folds compared with FljB separate expression (Fig. 3C-D). FljB co-expression increased perforin+ CD8+ T cell levels by 2.1 folds compared with OVA mRNA alone and 1.6 folds compared with FljB separate expression (Fig. 3E-F). MFI of cytokine levels showed the same trend as cytokine+ CD8+ T cell levels (Fig.S2). FljB co-expression but not separate expression also significantly increased dual and triple cytokine-secreting CD8+ T cell levels. FljB co-expression increased GrB+ perforin+, GrB+ IFNγ+, and IFNγ+ perforin+ CD8+ T cell levels by 2.9, 4.0, and 2.9 folds, respectively, as compared to OVA mRNA alone, and by 2.1, 2.6, and 1.8 folds, respectively, as compared to FljB separate expression (Fig. 3G-I). FljB co-expression increased GrB+ perforin+ IFNγ+ CD8+ T cell levels by 3.5 and 2.3 folds as compared to OVA mRNA alone and FljB separate expression, respectively (Fig. 3J).

Fig. 3
Fig. 3
Full size image

Splenocyte CD8+ T cell responses. Splenocytes of tumor-bearing mice were stimulated with OVA overnight followed by immunostaining and flow cytometry analysis of GrB+, IFNγ+, and perforin+ CD8+ T cell responses. Gating strategies were shown in Fig.S1. (A). Representative dot plots about GrB+ cells in CD8+ T cells. (B). Comparison of GrB+ CD8+ T cell levels among groups. (C). Representative dot plots about IFNγ+ cells in CD8+ T cells. (D). Comparison of IFNγ+ CD8+ T cell levels among groups. (E). Representative dot plots about perforin+ cells in CD8+ T cells. (F). Comparison of perforin+ CD8+ T cell levels among groups. G-I. Comparison of GrB+ perforin+ (G), GrB+ IFNγ+ (H), and IFNγ+ perforin+ (I) CD8+ T cell levels among groups. (J). Comparison of GrB+ perforin+ IFNγ+ CD8+ T cell levels among groups. One-way ANOVA with Fisher’s LSD test was used to compare differences between groups. n = 4–5. *, p < 0.05; **, p < 0.01; ***, p < 0.001.

FljB co-expression but not separate expression also significantly increased GrB+, IFNγ+, and perforin+ CD4+ T cell levels. FljB co-expression increased GrB+ CD4+ T cell levels by 2.3 folds compared with OVA mRNA alone and 2.2 folds compared with FljB separate expression (Fig. 4A-B). FljB co-expression increased IFNγ+ CD4+ T cell levels by 2.1 folds compared with OVA mRNA alone and 1.9 folds compared with FljB separate expression (Fig. 4C-D).

FljB co-expression increased perforin+ CD4+ T cell levels by 2.8 folds compared with OVA mRNA alone and 2.0 folds compared with FljB separate expression (Fig. 4E-F). MFI of cytokine levels showed the same trend as cytokine+ CD4+ T cell levels (Fig.S2). FljB co-expression but not separate expression also significantly increased dual cytokine-secreting CD4+ T cell levels. FljB co-expression increased GrB+ perforin+, GrB+ IFNγ+, and IFNγ+ perforin+ CD4+ T cell levels by 2.6, 4.1, and 2.6 folds, respectively, as compared to OVA mRNA alone, and by 1.8, 3.1, and 1.5 folds, respectively, as compared to FljB separate expression (Fig. 4G-I). Interestingly, no triple cytokine-secreting CD4+ T cells were induced in all groups.

Fig. 4
Fig. 4
Full size image

Splenic CD4+ T cell responses. Splenocytes of tumor-bearing mice were stimulated with OVA overnight followed by immunostaining and flow cytometry analysis of GrB+, IFNγ+, and perforin+ CD8+ T cell responses. Garting strategies were shown in Fig.S1. (A). Representative dot plots about GrB+ cells in CD4+ T cells. (B). Comparison of GrB+ CD4+ T cell levels among groups. (C). Representative dot plots about IFNγ+ cells in CD4+ T cells. (D). Comparison of IFNγ+ CD4+ T cell levels among groups. (E). Representative dot plots about perforin+ cells in CD4+ T cells. (F). Comparison of perforin+ CD4+ T cell levels among groups. G-I. Comparison of GrB+ perforin+ (G), GrB+ IFNγ+ (H), and IFNγ+ perforin+ (I) CD4+ T cell levels among groups. One-way ANOVA with Fisher’s LSD test was used to compare differences between groups. n = 4–5. *, p < 0.05; **, p < 0.01; ***, p < 0.001.

FljB reduces tumor-infiltrating CD8 + T lymphocytes

TILs were isolated for analysis of CD4+ and CD8+ T cell levels. As shown in Fig. 5A-B, OVA mRNA significantly increases tumor-infiltrating CD8+ T cell levels as compared to empty LNP, while FljB co-expression or separate expression reduced its levels as compared to OVA mRNA alone. FljB co-expression showed significantly reduced tumor-infiltrating CD4+ T cell levels as compared to FljB separate expression (Fig. 5C). Interestingly, the ratio of tumor-infiltrating CD8+ to CD4+ T cells was significantly increased in OVA mRNA alone and FljB co-expression groups as compared to empty LNP (Fig. 5D). FljB separate expression showed significantly reduced ratio of tumor-infiltrating CD8+ to CD4+ T cells as compared to OVA mRNA alone or FljB co-expression (Fig. 5D).

Fig. 5
Fig. 5
Full size image

Tumor-infiltrating lymphocyte analysis. TILs were isolated and stimulated with OVA followed by immunostaining and flow cytometry analysis. Gating strategies were shown in Fig.S3. (A). Representative dot plots about percentages of tumor-infiltrating CD4+ and CD8+ T cells. (B). Percentages of tumor-infiltrating CD8+ cells in CD3+ T cells. (C). Percentages of tumor-infiltrating CD4+ cells in CD3+ T cells. D. Ratio of CD8+ to CD4+ TILs. One-way ANOVA with Fisher’s LSD test was used to compare differences between groups. n = 4–5. *, p < 0.05; **, p < 0.01; ***, p < 0.001. Note: One mouse each in empty LNP and OVA mRNA groups gave rise to very few TILs and were excluded from the analysis.

FljB co-expression increases perforin-secreting TILs

The ability of TILs to secret GrB, IFNγ, and perforin was also analyzed. FljB co-expression but not separate expression significantly increased tumor-infiltrating perforin+ CD8+ T cells but not GrB+ or IFNγ+ CD8+ T cells (Fig. 6A-D). FljB co-expression increased tumor-infiltrating perforin+ CD8+ T cells by 45% as compared to OVA mRNA alone and 33% as compared to FljB separate expression (Fig. 6B). FljB co-expression but not separate expression also increased tumor-infiltrating perforin+ and GrB+ CD4+ T cells (Fig. 6E-F). FljB co-expression increased tumor-infiltrating perforin+ CD4+ T cells by 3.5 and 3.0 folds as compared to OVA mRNA alone and FljB separate expression, respectively (Fig. 6E). FljB co-expression increased tumor-infiltrating GrB+ CD4+ T cells by 4.3 and 2.3 folds as compared to OVA mRNA alone and FljB separate expression, respectively (Fig. 6F). No significant differences were found in tumor-infiltrating IFNγ+ CD4+ T cell levels among group (Fig. 6G). MFI of perforin, GrB, and IFNγ showed the same trend as percentages of perforin+, GrB+, and IFNγ+ cells in tumor-infiltrating CD4+ and CD8+ T cells (Fig.S4). FljB co-expression also significantly increased tumor-infiltrating GrB+ perforin+ CD4+ T cells as compared to OVA mRNA alone (Fig.S5).

Fig. 6
Fig. 6
Full size image

Cytokine-secreting TIL analysis. TILs in Fig. 5 were further analyzed about cytokine-secreting CD4+ and CD8+ T cells. (A). Representative dot plots about perforin+ cells in CD8+ TILs. B-D. Percentages of perforin+ (B), GrB+ (C), and IFNγ+ (D) in CD8+ TILs. E-G. Percentages of perforin+ (E), GrB+ (F), and IFNγ+ (G) in CD4+ TILs. One-way ANOVA with Fisher’s LSD test was used to compare differences between groups. n = 5. *, p < 0.05; **, p < 0.01. Empty LNP group was excluded from the analysis due to the small number of CD4+ and CD8+ T cells from this group.

FljB co-expression more significantly reduces tumor growth rate than FljB separate expression

Tumor growth rates were also monitored. We found OVA mRNA vaccination significantly reduced tumor growth rate as compared to empty LNP vaccination (Fig. 7A-B). OVA mRNA vaccination with FljB separate expression didn’t further reduce tumor growth rate (Fig. 7C), while OVA mRNA vaccination with FljB co-expression further reduced tumor growth rate (Fig. 7D). On day 23 after tumor inoculation, one in 5 mice in OVA mRNA and FljB mRNA/OVA mRNA groups had tumor volumes below 25% the average of tumor volumes of empty LNP group on day 20, while 3 in 5 mice in FljB-OVA mRNA group had tumor volume below this threshold (Fig. 7A-D). FljB/OVA separate expression significantly increased tumor volume on day 23 as compared to OVA mRNA alone (Fig. 7E), while FljB-OVA co-expression slightly reduced tumor volume despite lacking a statistically significant difference. FljB-OVA co-expression showed significantly reduced tumor volumes on day 23 as compared to FljB/OVA separate expression (Fig. 7E).

Fig. 7
Fig. 7
Full size image

Tumor growth rate. Tumor volumes of the differentially immunized mice were monitored. Tumor volumes of individual mice in empty LNP, OVA mRNA, FljB mRNA/OVA mRNA, and FljB-OVA mRNA groups were shown in A-D, respectively. Overall tumor growth was summarized in E. Dashed line: 25% of tumor volume of empty LNP group on day 20. Two-way ANOVA with Dunnett’s multiple comparisons test was used to compare differences between groups in E. n = 5. *, p < 0.05; ***, p < 0.001.

Discussion

This study found FljB co-expression with OVA (FljB-OVA mRNA) could significantly increase cytokine-secreting CD8+ T cell responses, while FljB separate expression (FljB mRNA/OVA mRNA) showed minimal effects. GrB, IFNγ, and perforin are cytokines secreted by CTLs with anti-tumor effects. GrB activates target cell’s intrinsic cell death pathways, while perforin facilitates GrB delivery into target cells18. IFNγ exerts its anti-tumor effects by inducing apoptosis of cancer cells, reducing the number of endothelial cells and inducing blood vessel destruction, and sustaining M1 status of tumor-associated macrophages19. FljB-OVA mRNA immunization induced at least 2-fold higher levels of GrB, IFNγ, and perforin-secreting CD8+ T cells in spleen of tumor-bearing mice as compared to OVA mRNA immunization, while FljB mRNA/OVA mRNA immunization failed to significantly increase these cells. These results hint the importance FljB co-expression with OVA to form FljB-OVA fusion proteins rather than separate expression in elicitation of CTL responses. Similar results were observed in CD4+ T cells. FljB-OVA mRNA induced more than 2-fold higher levels of GrB, IFNγ, and perforin-secreting CD4+ T cells as compared to OVA mRNA immunization, while FljB mRNA/OVA mRNA immunization failed to significantly increase these cells. FljB co-expression induced 10-fold lower systemic IL-6 levels as compared to FljB separate expression (Fig. 2D-E), hinting a better safety of the former to enhance mRNA vaccine-induced cellular immune responses considering flagellin-based vaccines often induce systemic IL-6 release with a close link to its systemic adverse reactions17,20.

The ability of FljB-OVA mRNA to induce significant cytokine-secreting CD8+ and CD4+ T cell responses was also observed in tumor tissues. FljB-OVA mRNA induced more than 2.5-fold higher GrB and perforin-secreting CD8+ TIL levels than OVA mRNA alone. FljB-OVA mRNA also induced more than 4.5-fold higher GrB-secreting CD4+ TIL levels than OVA mRNA alone. In contrast, FljB mRNA/OVA mRNA failed to significantly increase GrB, IFNγ, and perforin-secreting CD8+ or CD4+ TIL levels. To our surprise, significantly increased GrB, IFNγ, and perforin-secreting CD8+ or CD4+ T cell levels in FljB-OVA mRNA group didn’t translate into more potent anti-tumor immunity. FljB-OVA mRNA only slightly reduced tumor growth rate as compared to OVA mRNA alone. This might be explained by significantly reduced tumor infiltration of CD8+ T cells in FljB-OVA mRNA group as compared to that in OVA mRNA group. Limited tumor infiltration has been a common challenge for CAR-T cell therapy of solid tumors21. Limited tumor infiltration of activated CTLs was also reported in various scenarios. In one study, polyethyleneimine (PEI)-based personalized vaccine platform carrying neoantigen peptides and CpG adjuvants in a compact nanoparticle was found to elicit high levels of neoantigen-specific CD8+ T cells in the circulation but with poor tumor infiltration and anti-tumor efficacy22. In a different study, therapeutic CD25 depletion significantly increased systemic CD4+ and CD8+ T cell responses23. Yet, these cells failed to efficiently infiltrate the tumor and increase the intratumor ratio of effector T cells to Tregs23. According to reports22,24,25, modification of tumor microenvironment by intratumor delivery of immunostimulatory agents and cytokines was found to significantly increase tumor infiltration of activated CTLs and is worth of investigation to enhance tumor infiltration of FljB-OVA mRNA-induced CTLs.

Besides cellular immune responses and anti-tumor immunity, we also measured anti-OVA and anti-FljB antibody responses. We found FljB-OVA mRNA and FljB mRNA/OVA mRNA induced more potent anti-OVA IgG1 but weaker IgG2c antibody responses as compared to OVA mRNA alone. These results hinted FljB co-expression or separate expression could potentiate Th2-biased IgG1 antibody responses. Induction of Th2-biased IgG isotype switching by FljB mRNA was in line with prior reports that soluble flagellin potentiated Th2-biased antibody responses26. Furthermore, FljB co-expression or separate expression reduced overall anti-OVA IgG production, indicative of lack of adjuvant effects of FljB mRNA in potentiation of OVA mRNA-induced antibody responses. Th2-biased antibody responses seemed contradictory with the induction of potent Th1 responses (Fig. 4). A prior study found soluble FliC elicited Th2-biased antibody responses and yet promoted Tbet-regulated Th1 clearance after Salmonella typhimurium infection27. Our study was in line with this finding in that FljB-OVA mRNA elicited Th2-biased antibody responses and yet induced strong Th1 and CTL responses after B16F10-OVA challenge. We further found FljB mRNA/OVA mRNA failed to induce significant anti-FljB antibodies. This is less likely due to weak expression of FljB protein considering this group showed significantly increased serum IL-6 levels. Weak induction of anti-FljB antibodies was most likely due to the lack of D2/D3 domains, which were found to be the major immunodominant domains to elicit anti-flagellin antibody responses28. In further support, deletion of D3 domain reduced anti-flagellin antibody production, while deletion of D2/D3 domains further reduced anti-flagellin antibody production29. Interestingly, FljB-OVA mRNA induced significant anti-FljB antibody responses. These results hinted D0/D1 domains of FljB possessed B cell epitopes to elicit potent antibody responses. The weak anti-FljB antibody responses induced by FljB mRNA/OVA mRNA remained to be explored but might be due to a lack of helper T cell epitopes in the D0/D1 domains of FljB to provoke strong helper T cell responses, while the OVA portion of the fusion FljB-OVA protein provides CD4+ T cell epitopes to elicit potent helper T cell responses, critical for elicitation of potent anti-FljB antibody responses. Similar to our finding, replacing D2/D3 domains of flagellin with HIV-1 p24 protein induced significant anti-flagellin antibodies following intranasal immunization30.

Personalized neoantigen mRNA vaccines are under active development in tumor immunotherapy. Strategies to further enhance their therapeutic efficacy are highly demanded. The use of traditional adjuvants to potentiate mRNA vaccine-induced innate immune responses without a significant inhibition of mRNA translation holds a promise to elicit more potent CTL responses and anti-tumor immunity. This study supports preparation of flagellin-fused tumor antigen mRNA vaccines to elicit potent CTL responses. Yet, strategies are needed to promote tumor infiltration of elicited CTLs to achieve better therapeutic efficacy.