Multi-antigen DNA vaccine delivered by polyethylenimine and Salmonella enterica in neuroblastoma mouse model

Preparation of DNA vaccine

Plasmid vector pING (courtesy of Jedd D. Wolchok, USA) was used a backbone of the DNA construct. The fragment of TH gene (encoding 124 amino acids from 278 to 401 residues of native tyrosine hydroxylase) which contains the highest density of MHC class I epitopes and protein mouse/human homology was used as an antigen. Full-length cDNA was used for Survivin gene (sequence of 140 amino acids, Ref. # CCDS25694.1). The fragment of PHOX2B gene encoding the N-terminal 161 amino acids with full mouse/human homology was used. The gene of potato virus X coat protein (PVXCP) was added to the C-end of vaccine construct antigen-encoding insert as a fusion via AAAGPGP peptide linker. Immunoglobulin heavy chain leader peptide was added to the N-end of protein to ensure secretion of the protein out of the cell. DNA–PEI conjugate was prepared by mixing 10 μg pDNA with a solution of 15 μg PEI 20 kDa, pH 7.0 (Sigma-Aldrich). An oral form of vaccine was obtained by electroporation of newly designed (aroA, guaAB) attenuated strain of Salmonella enterica serovar typhimurium—SS2017 (data not published) with plasmid DNA (DNA–SL). The doses of bacteria (10⁸ per mouse) were measured in CFU, established by serial dilutions of a suspension of bacteria and seeding on plates for counting colonies. Bacterial culture was grown on the LB agar with 50 μg/mL kanamycin.

RT-PCR antigen expression

Total RNA was extracted from the 10⁶ murine neuroblastoma NB41A3 cells with TRIzol reagent (Invitrogen). 1 μg of total RNA was used for cDNA synthesis. Real-time PCR was performed on the CFX96 One-Touch (Bio-Rad). Total volume of reaction was 25 μL and included forward and reverse primers (6 pmol), TaqMan probe (4 pmol), 2 × TaqMan MasterMix (Applied Biosystems) and MQ water. HPRT gene was used as an internal control. Sequences of the primers are shown in Supplementary Table S1. Reactions were performed in duplicate. Antigen expression was calculated according to the M.V. Pfaffl using calibration curves for each antigen [21].

Cells and mice

NB41A3 murine NB cells were grown in F10 medium (Gibco) supplemented with 10% FBS and 1% penicillin/streptomycin. C57Bl/6 healthy mice (n = 60) 8–10 weeks of age were used for immunogenicity testing. Three consecutive experiments were conducted with each antigen (TH, Survivin, PHOX2B). A half of set included two control groups: empty vector pING (plasmid control) and bacterial cells without incorporated plasmid DNA (bacterial control). One group of animals was vaccinated thrice intramuscularly (i.m.) with DNA–PEI conjugate and another group with DNA–SL. Bacterial suspension was delivered by intragastric administration. After 2 weeks of monitoring, the mice were killed and all required biological material was extracted. Syngeneic A/J mice (Harlan Lab) (n = 64) of 8–10 weeks of age were used for tumor modeling. Primary tumors were induced by subcutaneous injection of 10⁶ NB41A3 cells and tumor growth was measured by a microcaliper. The tumor volume was calculated according to the formula: width × length × height. Experimental groups were the same as in immunogenicity testing. Mice received three therapeutic vaccinations: 7 days before, 5 and 10 days after tumor engraftment. When primary tumors reached a volume of 4000 mm³ and/or showed signs of necrosis, the mice were killed.

MHC epitope prediction

The MHC-I binding predictions for antigen immunogenicity modeling were made on 12/6/2019 using the IEDB analysis resource Consensus tool [22], which combines predictions from ANN aka NetMHC (4.0) [23–25], SMM [26] and CombLib [27].

Cytotoxicity assay

At the end of the experiment, splenocytes were harvested and incubated in the presence of irradiated NB41A3 NB cells in RPMI 1640 [10% FBS, 1% AA, 50 μmol/L β-mercaptoethanol and 100 units/mL IL-2] at a target:effector (T:E) ratio of 1:10 and 1:25 overnight. Target cells were plated out at a density of 10,000 per well in a 96-well plate. Effector cells were added to duplicate wells at varying T:E ratios. Cytotoxic activity of splenocytes was determined by using a flow cytometry assay (FC500, Backman Coulter, USA). Target cells were prestained with CFSE (carboxyfluorescein succinimidyl ester) to separate them from effectors; propidium iodide was used for vitality staining. Cytotoxicity was calculated according to lysis (%) = [experimental count of dead cells (cdc)-spontaneous cdc]/[100 − spontaneous cdc] × 100%.

ELISA

The samples of the whole blood were collected from the heart in aseptical conditions. Then, they were centrifuged at 5000 × g to separate serum from cellular fraction. Experimental serum was tested in duplicate in the 1:100 and 1:200 ratios in 96-well plate coated with PVXCP protein. Immobilized anti-PVXCP antibodies were detected with goat anti-mouse antibodies IgG + IgM labeled with horse radish peroxidase (Thermo Fisher Scientific, USA). The plates were read using a Sunrise™ Absorbance reader (TECAN). Twofold dilution of mouse serum collected from the animals immunized with PVX virus was used as calibrators. Each dilution had a certain amount of related units per mL (1000-125).

ELISPOT

IFN-gamma secretion was determined using the BD™ ELISPOT mouse IFN-γ ELISPOT Set (BD Biosciences, USA). Splenocytes were harvested from individual mice after immunization with a DNA vaccine. Effector cells were plated in duplicate at 2.5 × 10⁵/well in a 200-μL final volume with medium alone, 10 μg of the TH/Survivin/PHOX2B peptide library per ml, and 20 μg of the PVXCP protein per mL or 4 μg of the phytohemagglutinin. Plates were analyzed with an enzyme-linked immunospot assay (ELISPOT) reader (CTL ImmunoSpot^® S5 UV Analyzer, C.T.L.,). The mean number of spots from duplicate wells was calculated for each responder animal and adjusted to represent the mean number of spots per 10⁶ splenic cells.

Statistics

The statistical significance of difference between experimental groups was determined by non-parametrical Mann–Whitney U test using GraphPad Prism 6.0 software. Findings were regarded as significant if p values were < 0.05.

Antigen expression in NB41A3

Murine NB41A3 brain origin NB cells were analyzed for antigen expression to establish suitable mouse tumor model. mRNA expression of TH, Survivin and PHOX2B genes was measured by RT-PCR. All investigated antigens were expressed in NB41A3. As shown in Fig. 1, PHOX2B was the most highly expressed antigen. The expression of Survivin was three times less abundant than that of PHOX2B (7.27 ± 2.95 vs 21.07 ± 4.67), and TH demonstrated the lowest expression level (3.57 ± 2.76). RT-PCR data suggest that the chosen antigens could be used as targets for DNA vaccination. Moreover, the absence of genetically engineered enhanced expression allows to analyze the influence of antigen expression level on immunogenicity.

Antigen immunogenicity

Antigen immunogenicity (IG) depends on several parameters such as the size of a molecule, amino acids diversity, presence of aromatic amino acids, and solubility. The efficiency of MHC-ligand processing and binding affinity are additional factors determining the induction of immune response [29]. Selected antigens were analyzed for the most common immunogenicity parameters to predict the best option for DNA vaccination. Antigens were scored points for each parameter from 1 (the worth result) to 3 (the best result). Summing up the points resulted in the final scores represented in Table 1. PHOX2B protein contains more amino acids than other contenders, the majority of which are aromatic. MHC-I epitope binding prediction was performed using immune epitope database-analysis resource (IEBB-AR) for both mouse strains used in the experiments. Only T cell epitopes with percentile affinity rank less than 10 were considered as highly immunogenic. TH and PHOX2B antigens have more suitable epitopes for H2K^b haplotypes corresponding to C57Bl/6 mice and Survivin includes epitopes predominantly for H2K^k from A/J mice. Median affinity binding level of selected epitopes also have different values for mouse strains. According to the total IG score, theoretical immunogenicity of antigens increased in the following order: TH → Survivin → PHOX2B for both mouse strains.

Preliminary screening of the antigens for immunogenicity revealed that PHOX2B had the greater potential for inducing a strong immune response than other investigated proteins according to in silico testing. There are many factors that can be taken into account in mathematical models, and their biological testing on animals is a prerequisite in the process of finding the ideal candidate for vaccination.

To assess immunogenicity of selected antigens in vivo, three independent experiments were conducted. C57Bl/6 mice were vaccinated three times with each of DNA vaccine. As was shown by a cytotoxicity test, DNA vaccination led to increase of target cell killing in almost all treated groups of mice compared to the vector control. As shown in Fig. 2a–c, the DNA–PEI conjugate elicited significant T-cell cellular immune response for each antigen and their combination except TH (Fig. 2a), where the percentage of lysed cells was comparable with that in the control group (3.4% and 4.9% in control vs 2.9 and 5% in the TH–PEI, target:effector ratio 1:10 and 1:25, respectively). Oral gavage with S. enterica induced a strong cellular immune response against target cells, which was at least two times higher than that after DNA–PEI for all antigens (p < 0.05). The median percentage of killing NB cells significantly increased in mice vaccinated with DNA–SL vs DNA–PEI: 10.2% vs 4.9% for TH, 23.8% vs 11.5% for Survivin, 21.23% vs 7.9% for PHOX2B (data shown for T:E ratio 1:25).

Cytotoxic activity (CTA) of splenocytes after DNA vacination with different antigens was evaluated to reveal the most immunogeneic tumor-associated protein. To overcome the variations in cytotoxic response in different experiments induced by the dynamic settings, data were normalized to the corresponding control group and are depicted in Fig. 2d. There was a significant CTA increase in the following order of antigens: TH → Survivin → PHOX2B when they were conjugated with PEI. The immunogenecity of TH was lower than that of Survivin and PHOX2B (p < 0.05). Oral vaccination with SL produced comparable results for all three antigens (nearly 2 times higher than in control), but TH produced the best result. CTA after vaccination with DNA–PEI and DNA–SL had close values with a tendency of higher response induced by DNA–SL.

ELISPOT assay was performed to assess IFNγ secretion. Data are shown in Fig. 3a–c. Splenocytes were stimulated with peptide library of distinct antigens and with PVXCP protein. IFNγ production in response to PVXCP stimulation was demonstrated in 62.5% of vaccinated mice. The highest antigen-specific IFNγ secretion was detected in TH–PEI and PHOX2B–PEI groups after incubation with peptide libraries, 7.2 and 8.4 times higher (respectively) than in the control group. Mice vaccinated with bacterial form developed a stronger response to antigen stimulation. Thus, the level of IFNγ in TH-SL was 35 times higher and PHOX2B-SL was 11.6 times higher than in the placebo group. Vaccination with Survivin had no effect on the IFNγ production. To compare different antigens with each other, the same normalization to the control like in CTA analysis was performed (Fig. 3d). IFNγ response was considered positive if it exceeded double increase in comparison with the placebo group. There were no differences between antigens and delivery methods in response to PVXCP. DNA vaccination with TH was accompanied by higher IFNγ secretion (against peptide library) than other antigens with both carriers. When SL was used, TH was significantly more effective than Survivin (Chi square, p < 0.05). A quantity of IFNγ was higher for all antigens when they were used with PEI compared to SL (p = 0.03 for PHOX2B).

Titer of anti-PVXCP antibodies was not different from control group in all experiments.

Tumor progression in mice

Vaccination was performed according to the scheme shown in Fig. 4a. The dynamics of tumor progression was monitored during 30 days after the engraftment of NB cells. Tumor volume [mm³] was measured by digital caliper every 5 days. Histological analysis showed that engrafted NB was represented by solid layers of a malignant mesenchymal tumor tissue composed of large cells with round, light-colored nuclei with several hyperchromic nucleoli, and with very high mitotic activity (up to 30 mitoses/10HPF). In the tumor, foci of necrosis and foci of ingrowth into the skin were determined. In separate segments, the signs of the beginning of ganglion differentiation were observed (Fig. 4b).

It was established that all types of the DNA vaccine used were able to slow down tumor growth (Fig. 5a–c). Among plasmid DNA carriers, PEI showed better results, when compared with Salmonella in all groups of mice. Median tumor volume after vaccination with TH–PEI on the 25th day was 918.2 vs 2328 mm³ with TH–SL (2664 mm³ in the control group). Comparable results were obtained from the mice vaccinated with Survivin. Final tumor volume after vaccination with Survivin–PEI was 2.7 times less than in control (512 vs 1417 mm³) and after vaccination with Survivin–SL two times less (1302 vs 2664 mm³). Immunization with PHOX2B antigen was the most successful in preventing tumor progression. PHOX2B–PEI vaccination led to median final NB tumor volume of 420 vs 2664 mm³ in the control group. PHOX2B–SL effect was reflected as 4.7 times decrease of tumor size (634 mm³).

Antitumor effect of vaccination with different antigens was compared by the ratio of mean tumor volume of experimental mice to control mice at the last time point (Fig. 5d). It was shown that vaccination with PHOX2B elicited the strongest antitumor effect when compared with TH or Survivin antigens. TH was the second most effective antigen in tumor containment and Survivin was the least effective.

Conflict of interest

The authors declare that they have no conflict of interest.

Ethics approval

All animal experiments were conducted according to the Belarussian guide for the Care and Use of Laboratory Animals and approved by the ethics committee of the Institute of bioorganic chemistry of the National Academy of Sciences of Belarus.