Article

ISSN 0006-2979, Biochemistry (Moscow), 2025, Vol. 90, No. 6, pp. 754-772 © Pleiades Publishing, Ltd., 2025.

Published in Russian in Biokhimiya, 2025, Vol. 90, No. 6, pp. 812-832.

754

Multiepitope mRNA Vaccine mRNA-mEp21-FL-IDT

Provides Efficient Protection against M.tuberculosis

Alisa A. Kazakova

, Galina S. Shepelkova

, Ivan S. Kukushkin

Vladimir V. Yeremeev

, Roman A. Ivanov

, and Vasiliy V. Reshetnikov

1,a

Sirius University of Science and Technology, 354340 Sirius Federal Territory, Russia

Central Tuberculosis Research Institute, 107564 Moscow, Russia

e-mail: reshetnikov.vv@talantiuspeh.ru

Received January 18, 2025

Revised June 9, 2025

Accepted June 9, 2025

Abstract— Tuberculosis is a leading cause of death from a bacterial infection agent. The development of

new tuberculosis vaccines can reduce the number of new cases and tuberculosis-related deaths. One of the

most promising areas in vaccination is development of mRNA vaccines, which have already proven their

high effectiveness against COVID-19 and other viral infections. Using modern immunoinformatic methods,

we developed four new antituberculosis multiepitope mRNA vaccines differing in the encoded adjuvants

and codon composition and tested their immunogenicity and protectivity in mice. Most of the developed

mRNA vaccines induced the formation of both cellular and humoral immunity. The adaptive response was

stronger for the vaccines with the RpfE adjuvant; however, the best protective response was elicited by the

mRNA-mEp21-FL-IDT vaccine with the FL adjuvant. This vaccine reduced the mycobacterial load in the lungs

of mice infected with Mycobacterium tuberculosis and increased their survival rate. Altogether, our results

indicate that the mRNA-mEp21-FL-IDT vaccine ensures effective protection against tuberculosis comparable

to that provided by the BCG vaccine.

DOI: 10.1134/S0006297925600073

Keywords: mRNA vaccine, tuberculosis, multiepitope vaccines, molecular adjuvants, adaptive immunity,

protective immunity

* To whom correspondence should be addressed.

INTRODUCTION

Tuberculosis is an infectious disease caused by

the Mycobacterium tuberculosis bacterium. In 2023

alone, 10.7  million new cases of tuberculosis have

been registered, and 1.25 million people died of this

disease [1]. Currently, Bacillus Calmette–Guérin (BCG)

is only one certified vaccine against tuberculosis; how-

ever, it is ineffective against pulmonary tuberculosis,

which is the main form of tuberculosis in adults [2].

In this regard, there is an urgent need for new highly

efficient antituberculosis vaccines.

mRNA vaccines have gained widespread recogni-

tion due to the success of the RNA-1273 (Moderna) and

BNT162b2 (Pfizer) vaccines against SARS-CoV-2. Cur-

rently, new mRNA vaccines are being designed against

HIV-1, Zika virus, influenza virus, rabies virus, and

other pathogens [3-6]. mRNA vaccines present several

advantages over other types of vaccines, including low

production costs, rapid development, high efficacy,

noninfectivity, and the absence of integration into the

genome. Although mRNA vaccines have proven their

effectiveness and safety [7, 8], researchers have not

yet succeeded in creating an antituberculosis mRNA

vaccine that would ensure strong protection compara-

ble to that provided by the BCG vaccine [9, 10].

The key role in the effectiveness of mRNA vac-

cines against bacterial pathogens is played by the

target antigen sequences and molecular adjuvants.

Currently, the most promising strategy for choosing

an antigen for vaccines is the multiepitope design, i.e.,

inclusion of many individual epitopes from different

pathogen’s antigens [11, 12]. The design of a multie-

pitope mRNA vaccine allows the incorporation of the

MULTIEPITOPE mRNA VACCINE PROTECTS AGAINST TUBERCULOSIS 755

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

most immunogenic and safest epitopes [13-16], which

will activate cytotoxic T  lymphocytes (CTL), helper T

lymphocytes (HTL), and B  cells when presented by the

major histocompatibility complex (MHC). As a rule, the

antigenic proteins used in the vaccine creation are sur-

face and secreted proteins of a pathogen, which are the

first to come into contact with host’s immune system.

The immunogenicity and safety of epitopes are

determined by their affinity to the MHC, stability (re-

sistance to proteolytic cleavage in a cell), and lack

of homology with the organism’s own proteins. The

three-dimensional structure is especially important

for the B  cell epitopes, since B  cell receptors recog-

nize conformational epitopes on the antigen surface.

The binding to the epitope leads to the activation of

B cells and their differentiation into plasma cells that

produce the antibodies specific to this epitope.

T  cell epitopes (CTL and HTL) are formed by the

antigen cleavage in the proteasome and binding to

the MHC class I (for CD8

  T  cells) or class II (for CD4

T  cells). The MHC–epitope complex is transported to the

cell surface, where it is recognized by T cells through

the T  cell receptors, leading to the activation of T cells

and triggering of the immune response.

Beside the epitopes, the effectiveness of mRNA

vaccines is determined to a great extent by the en-

coded molecular adjuvants. For example, incorpora-

tion of the FMS-like tyrosine kinase-3 ligand (FL) as

an encoded adjuvant increased the level of Th1 cy-

tokine (IFN-γ and IL-12) production, the number of

T  cells secreting IFN-γ, the activity of cytotoxic T  lym-

phocytes, and the level of IgG antibodies, which to-

gether allowed to develop an antituberculosis DNA

vaccine with a better efficacy than BCG [17]. Other

molecules have also been used as adjuvants in an-

tituberculosis vaccines, including resuscitation-pro-

moting factor E (RpfE), which helps restore the via-

bility of M.  tuberculosis and ensures the activation

of dendritic cells (DCs), as well as heparin-binding

hemagglutinin (HBHA) and 50S ribosomal protein

(Rv0652), which act as bacterial agonists of Toll-like

receptors (TLRs) and promote the activation of TLR4.

RpfE, HBHA, and Rv0652 induce the maturation of

DCs by enhancing the surface expression of matura-

tion markers (CD40, CD80/CD86, and class I/II MHC)

and production of IL-6, IL-1β, IL-23p19, IL-12p70, and

TNF-α in a TLR4-dependent manner [18-20]. RpfE also

promotes the differentiation of CD4

T cells into Th1

and Th17 subpopulations through the modulation of

DC function.

Beside the coding sequence of antigens, molecular

adjuvants, and signal peptides [21,  22], the secondary

structure of the RNA molecule itself plays an import-

ant role in the effectiveness of mRNA therapeutics,

as it determines the translational efficiency of mRNA

and its stability in the cells [23, 24].

In our study, we focused on designing an effec-

tive multiepitope mRNA antituberculosis vaccine by

in silico immunoinformatic methods, which substan-

tially reduced the duration and cost of vaccine de-

velopment due to the rapid and effective prediction

of suitable vaccine antigens, epitopes, and adjuvants

[25]. We employed two sequences of molecular adju-

vants – RpfE and FL – and used two contrasting strat-

egies for the RNA secondary structure optimization.

To test the functional efficacy of the four newly de-

veloped mRNA vaccines and to compare them to BCG,

we evaluated their immunogenicity and protectivity

invivo in mice with the opposite sensitivity to tuber-

culosis.

MATERIALS AND METHODS

The choice of target proteins. All amino acid se-

quences were obtained from the NCBI Protein (https://

www.ncbi.nlm.nih.gov/) and UniProt (http://www.

Uniprot.org) databases. Linear  B  lymphocyte (LBL),

CTL, and HTL epitopes with the optimal characteris-

tics were chosen within the amino acid sequences of

selected M. tuberculosis proteins.

Prediction and evaluation of LBL, CTL, and HTL

epitopes. To predict the LBL epitopes, we selected 10

M.  tuberculosis antigens that induce a pronounced

IgG response and/or are membrane proteins [26,

27]: ESAT6, CFP10, Ag85A, Ag85B, PE, PPE55, PPE68,

MPT83, HRP1, and HspX. The LBL epitopes were se-

lected with the ABCpred online server (https://webs.

iiitd.edu.in/raghava/abcpred/ABC_submission.html)

[28], which predicts epitopes using a recurrent neural

network trained on epitopes from the Bcipep database

(https://webs.iiitd.edu.in/raghava/bcipep/info.html)

[29]. The parameters applied for the epitope search

were as follows: LBL epitope length, 16 amino acid(a.a.)

residues; specificity threshold, 0.51; overlapping filter:

ON. All LBL epitopes predicted by the server were

subjected to subsequent analysis.

The CTL epitopes were chosen using the Next

Generation T Cell Prediction Class I online resource

(https://nextgen-tools.iedb.org/)  [30]. The epitope lengths

were set to 9-10  a.a. The full reference set of human

leukocyte antigens (HLA) and the NetMHCpan 4.1 EL

mathematical model were used to predict the epitopes.

The obtained CTL epitopes were sorted by IC50, and

peptides with IC50 <  500 were selected for subsequent

analysis.

The HTL epitopes were predicted with the MHC-

II Binding Predictions server (http://tools.iedb.org/

mhcii/). The length of the epitopes was set to 15  a.a.

The epitopes were predicted using the full reference

set of HLA and the NetMHCIIpan 4.1 EL mathemati-

cal model. The obtained peptides were sorted by their

KAZAKOVA et al.756

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

percentile rank, and epitopes with a rank >90 were

selected for further analysis.

To predict the potential autoimmunogenicity, all

selected epitopes were checked for homology with

the human and mouse proteomes using the protein–

protein BLAST service (https://blast.ncbi.nlm.nih.gov/

Blast.cgi?PROGRAM=blastp&PAGE_TYPE=BlastSearch

&LINK_LOC=blasthome) and Homo sapiens and Mus

musculus protein databases (Taxid:9606 and Taxid:

10090, respectively). To prevent the autoimmune re-

sponse, all chosen peptides lacked homologs among

human and mouse proteins with an E-value  <  0.1  [31].

To predict the antigenicity, the selected epitopes

were analyzed with the VaxiJen software v3.0 (https://

www.ddg-pharmfac.net/vaxijen3/)  [32]. The epitopes

that were identified as potentially antigenic with a

probability of at least 66% according to the analy-

sis results were subjected to further investigation.

To predict the allergenicity of the epitopes, the

AllerTOP server v2.0 (https://www.ddg-pharmfac.net/

AllerTOP/) was used with the default parameters; only

nonallergenic epitopes were selected [33]. To predict

the toxicity of the epitopes, we used the next-gener-

ation ToxiPred2 web server (https://webs.iiitd.edu.in/

raghava/toxinpred2/ige.html) under the MERCI predic-

tion model [34]. Only the epitopes that were found to

be antigenic, nonallergenic, and nontoxic were select-

ed for further analysis.

Next, the LBL epitopes were assessed for their

putative ability to induce in a complex with MHC the

synthesis of IgG and IgE using the IgPred Web serv-

er (https://webs.iiitd.edu.in/raghava/igpred/prot-vari-

pred.html) with a threshold of 0.9 [35].

The HTL epitopes were tested for their tenta-

tive ability to induce in a complex with MHC the

synthesis of IFN-γ, IL-4, and IL-10. To evaluate the

secretion of IFN-γ, we employed the IFNepitope Web

service (https://webs.iiitd.edu.in/raghava/ifnepitope/

predict.php) using the hybrid prediction approach

“SVM + motif” with the “IFN-γ versus non IFN-γ” pre-

diction module [36]. The secretion of IL-4 was assessed

with IL4pred (https://webs.iiitd.edu.in/raghava/il4pred/

predict.php) [37] using the “SVM + motif” hybrid pre-

diction model with a threshold of 0.2. To predict the

secretion of IL-10, IL-10Pred (https://webs.iiitd.edu.in/

raghava/il10pred/predict3.php) was utilized with the

SVM predictive model with a threshold of −0.3 [38].

Only the epitopes that induced the secretion of IFN-γ

but not of IL-4 or IL-10 were selected for further studies.

The final vaccine sequence included 7 LBL epi-

topes, 7 HTL epitopes, and 7 CTL epitopes.

Analysis of MHC allele frequency in the popu-

lation. The estimated coverage of the population was

predicted for all the epitopes selected at the previous

stages using the corresponding MHCI and MHCII al-

leles with the Population Coverage tool (http://tools.

iedb.org/population/) in the IEDB 59 database [39]. The

prediction was made using the full set of MHC alleles

similar to the set utilized at the epitope prediction

stage, as well as the full “Russia” dataset. Based on the

analysis of population coverage, we selected epitopes

with >90% prevalence of respective MHC alleles in the

population.

Molecular docking between T-lymphocyte epi-

topes and MHC alleles. Molecular docking of the

HTL and CTL epitopes was performed on the CABS-

dock Web server (https://biocomp.chem.uw.edu.pl/

CABSdock/) [40]. After the docking, we used the Hawk-

Dock server (http://cadd.zju.edu.cn/hawkdock/) to esti-

mate the binding free energy in order to find the best

conformation identified by CABS-dock [41].

The selected MHC I alleles were HLA-A*11:01

(PDB  ID:  6ID4), HLA-A*02:01 (PDB  ID:  7RTD), and

HLA-A*24:02 (PDB  ID:  5WWI); the selected MHC  II

alleles were HLA-DRB1*01:01 (PDB  ID:  1AQD),

HLA-DRB1*15:01 (PDB  ID:  8TBP), HLA-DRB1*07:01

(PDB  ID:  7Z0Q), and HLA-DPA1*01:03/HLA-DPB1*02:01

(PDB  ID:  3LQZ). These alleles were chosen as some of

the most common alleles in the Russian population

(https://allelefrequencies.net/default.asp) [42].

The resulting docking models were assessed using

the root-mean-square distance (RMSD), a parameter

indicating the quality of docking and directly related

to the quality of ligand–receptor binding. According

to the developers’ assessment, the docking poses with

RMSD  >  3.00  Å have only few key interactions char-

acteristic of true docking positions, while RMSD  <  3  Å

indicates a large number of key interactions, suggest-

ing strong ligand binding. RMSD from 3 to 5.5  Å cor-

responds to a moderate strength of ligand binding,

and at RMSD  >  5.5  Å, the ligand presumably does not

bind at all.

Design of multiepitope vaccine proteins and

vaccine mRNAs. We used four types of cleavable

linkers (EAAAK, GPGPG, KK, and AAY) in the design

of the vaccine protein to ensure the hydrolysis and

necessary flexibility/rigidity of the synthesized protein

molecules for the efficient epitope presentation. The

EAAAK linker was placed between the tissue plas-

minogen activator (tPA) and the adjuvant sequences

to improve the stability of the fusion protein [43].

The GPGPG and AAY linkers were placed between the

HTL and CTL epitopes, respectively, to improve the

presentation [44]. The HTL epitopes were connected

to each other via the GPGPG linkers, the LBL epi-

topes – via the KK linkers, and the CTL epitopes –

via the AAY linkers. The optimized Kozak sequence

(GCCACAAUGgg) was included in the mRNA sequence

[45]. A special stop codon sequence (UGAUGAUGA)

was chosen to achieve the maximal translation ter-

mination. Two signal sequences were added to

the mRNA to strengthen the antigen presentation:

MULTIEPITOPE mRNA VACCINE PROTECTS AGAINST TUBERCULOSIS 757

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

the tPA secretory signal (UniProt ID: P00750) was

placed at the 5′ end, and the MHC I-targeting domain

(MITD; UniProt ID: Q8WV92) was placed at the 3′ end.

The full-length RpfE (CCP45243.1) and FL

(NP_001450.2) were used as encoded adjuvants. RpfE

(Rv2450c) was chosen because it binds to TLR4, in-

duces proliferation of T cells and differentiation into

Th1 and Th17 lymphocytes with a pronounced anti-

mycobacterial activity, stimulates the production of

proinflammatory cytokines, promotes the cellular im-

mune response, and enhances the adaptive immune

response [18]. FL, which is a growth factor, stimulates

the growth of T  cells, B  cells, and DCs, increases the

immunogenicity of DNA vaccines; and enhances the re-

cruitment and expansion of DCs, thereby contributing

to the formation of effective immune response [17].

The TPL sequence was used as the 5′ untranslat-

ed region (UTR). The TPL has a high median ribo-

some loading and improves the mRNA translational

efficiency (https://optimus5.cs.washington.edu/MRL)

[46,  47]. The 3′ UTR sequence from the RNA-1273

vaccine against SARS-CoV-2 (Moderna) served as the

3′  UTR. Furthermore, to increase the mRNA stability, a

standard 110-nt polyA tail and the AG sequence were

added downstream of the T7 promoter in the expres-

sion cassette. The AG sequence ensures the cotrans-

criptional incorporation of the synthetic cap analog

7,3

′-OGpppAmG (CapAG) during transcription invitro.

Therefore, the overall structure of the vaccine

mRNA was as follows: 5′-cap  →  5′  UTR  →  Kozak se-

quence  →  tPA signal peptide  →  EAAAK linker  →  RpfE/

FL (adjuvant)  →  GPGPG linker → HTL epitopes → KK

linker → LBL epitopes → AAY linker  →  CTL epitopes  →

AAY linker → MITD → stop codons → 3′ UTR → polyA tail

(Fig.1). Four mRNA sequences were generated: mRNA-

mEp21-RpfE-LD, mRNA-mEp21-FL-LD, mRNA-mEp21-

RpfE-IDT, and mRNA-mEp21-FL-IDT that differed in

the adjuvant sequence (RpfE or FL) and optimization

of the mRNA secondary structure (LD or IDT).

Prediction of antigenicity, allergenicity, toxici-

ty, and physicochemical properties of vaccine pro-

teins. For the analysis of antigenicity, allergenicity, and

toxicity of vaccine proteins, we used the amino acid

sequences that included the adjuvants, epitopes, and

linkers, but lacked the tPA and MITD sequences. The

antigenicity of the potential vaccines was predicted

with the VaxiJen server v3.0. We also used the ANTI-

GENpro server (https://scratch.proteomics.ics.uci.edu/)

based on the microarray data and machine learning

algorithms [48]. The allergenicity, toxicity, and homol-

ogy of the vaccine proteins were assessed with the

AllerTOP v.2.0, ToxiPred2, and BLAST, respectively, as

described above.

The physicochemical parameters of protein se-

quences lacking the tPA and MITD fragments were

evaluated with the Expasy server (https://web.expasy.

org/protparam/) [49]. The examined characteristics

included the number of amino acid residues, molecu-

lar weight, theoretical isoelectric point (pI), aliphatic

index (AI), instability index (II), and global average

hydropathicity (GRAVY).

Determination and confirmation of the tertia-

ry structure of the vaccine proteins. The hypothet-

ical tertiary structures were obtained using Robetta

(https://robetta.bakerlab.org/submit.php) [50] and then

evaluated with the ERRAT, Verify 3D, and PROCHECK

web services (https://saves.mbi.ucla.edu/).

Prediction of conformational B-cell epitopes.

The tertiary structure of a protein can give rise to new

conformational B cell epitopes [51]. To predict both

linear and conformational B cell epitopes, we used

the ElliPro tool (http://tools.iedb.org/ellipro/), which

analyzes geometric characteristics of 3D models and

offers the best area under the curve (AUC) (0.732) for

any protein model [52, 53].

Molecular docking of vaccine proteins with TLR4.

The docking was performed between each putative

vaccine protein and TLR4 (PDB  ID:  3FXI) using the

ClusPro 2.0 server (https://cluspro.bu.edu/home.php)

and the PIPER algorithm [54]. RpfE (TLR4 agonist)

served as a positive control [18]. The binding free

energy (∆G) and the dissociation constant (K

) were

calculated at 37°C with the help of the PRODIGY tool

from the HADDOCK server (https://rascar.science.uu.nl/

prodigy/) [55]. The PDBsum web server (https://www.

ebi.ac.uk/thornton-srv/databases/pdbsum/Generate.

html) was used to analyze and visualize the receptor–

ligand interactions [56].

Fig. 1. Structure of developed mRNA vaccines. Four mRNA sequences were generated: mRNA-mEp21-RpfE-LD, mRNA-mEp21-

FL-LD, mRNA-mEp21-RpfE-IDT and mRNA-mEp21-FL-IDT, differing in the adjuvant sequence (RpfE or FL) and optimization

of the mRNA secondary structure (LD or IDT). 7 HTL epitopes, 7 LBL epitopes, and 7 CTL epitopes were included in each

vaccine variant.

KAZAKOVA et al.758

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

In silico immunization. The time course of the

immune response was simulated for each vaccine

with the C-ImmSim server (https://kraken.iac.rm.cnr.it/

C-IMMSIM/index.php?page=1) employed [57]. Three

in silico immunizations (injections) of 1000 units of

vaccine per dose were performed on days 1, 28, and

56 of the simulation. The remaining parameters were

defaults.

Optimization of codon composition and remov-

al of miRNA-binding sites. Optimization algorithms

are commonly used to increase the mRNA stabili-

ty and translation efficiency. Optimization was per-

formed with the following algorithms: Linear Design

(https://github.com/LinearDesignSoftware/Linear

Design), RiboTree (https://github.com/philarevalo/

RiboTree), IDT (https://eu.idtdna.com/pages), genewiz

(https://www.genewiz.com/en-GB/), and iCodon (https://

bazzinilab.shinyapps.io/icodon/) [24, 58, 59]. The ob-

tained mRNA sequences were analyzed in the Super -

folder (https://github.com/eternagame/superfolder-covid-

mrna-vaccines).

The Superfolder, which includes special software

packages for determining the degradation time, sec-

ondary structure, ∆G, and other critical characteristics

of mRNA, was used to evaluate the following param-

eters: mRNA half-life, average unpaired probability

(AUP; the total proportion of unpaired nucleotides;

lower AUP increases the life span of mRNA), AUP init

14 (the total proportion of unpaired nucleotides with-

in the first 14 nucleotides; higher AUP init 14 increas-

es the translatability of mRNA), and codon adaptation

index (CAI, codon optimization parameter; higher CAI

increases the translatability of mRNA). The in  silico

parameters predicted by the Superfolder programs

have not been experimentally verified.

For further work, we chose the Linear Design and

IDT algorithms because according to the insilico mod-

eling results, Linear Design increased the half-life of

RNA better than the other algorithms, whereas IDT

helped to achieve the optimal mRNA translatability.

To prevent the cleavage and translation inhibition

of mRNA, we searched the designed mRNA sequences

for the binding sites for H.  sapiens and M.  musculus

miRNAs using the miRDB website (https://mirdb.org/

custom.html). If the probability of the binding site

presence was ≥65 (conventional units of this server;

scale, from 0 to 100), the corresponding codons were

replaced with synonymous ones to decrease the prob-

ability below 65, after which the resulting sequences

were rechecked for the presence of miRNA-binding

sites and translated insilico to correct possible errors

caused by the codon replacement (http://molbiol.ru/

scripts/01_13.html).

Animal experiment design. Female mice of the

C57BL/6JCit (B6) (genetically resistant to tuberculosis)

and I/StSnEgYCit (I/St) (genetically sensitive to tuber-

culosis [60]) strains aged 3 to 4 months and weighing

at least 21  g were used in the experiments. The an-

imals were kept under standard conditions with the

ad libitum access to food and water at the Animal Fa-

cility of the Tuberculosis Research Institute (Moscow,

Russia).

To assess the immunogenicity of the candidate

vaccines, B6 mice were divided into 7 groups (5 mice

per group). The animals of the experimental groups

were immunized intramuscularly twice (5 μg/mouse) at

an interval of 4 weeks with one of the four mRNA vac-

cines (mRNA-mEp21-RpfE-IDT, mRNA-mEp21-FL-IDT,

mRNA-mEp21-RpfE-LD, or mRNA-mEp21-FL-LD). Mice

receiving lipid nanoparticles (LNPs) without mRNA or

phosphate-buffered saline (PBS (Sigma-Aldrich, USA),

pH 7.5) served as negative controls. Positive controls

were mice that were inoculated subcutaneously with

100,000 colony-forming units (CFU) of BCG one time,

five weeks before the experiment.

Four weeks after the second immunization, the

antigen-specific production of IFN-γ was assessed by

the ELISpot assay ex  vivo in isolated splenocytes. The

titer of IgGs (the maximum serum dilution at which

specific IgGs are detected) to the Mycobacterium tu-

berculosis sonicate, which was obtained by ultrasoni-

cally disintegrating bacterial cells, was determined in

animal blood serum. The serum was diluted with PBS

at a 1  :  50 to 1  :  400 ratio. The intensity of the humoral

response was assessed by absorption at 450 nm ac-

cording to the previously described method [61].

The protective properties of mRNA vaccines were

evaluated in I/St mice genetically sensitive to tuber-

culosis. The animals were divided into seven groups

each containing 15 mice. Mice from 4 groups were in-

jected with the mRNA-mEp21-RpfE-IDT, mRNA-mEp21-

FL-IDT, mRNA-mEp21-RpfE-LD or mRNA-mEp21-FL-LD

vaccines; one group was immunized with the BCG,

and two groups served as negative controls (mice in-

jected with mRNA-free LNPs and PBS (pH 7.5)). Im-

munization with the mRNA vaccines was performed

twice with an interval of 4  weeks intramuscularly at

a dose of 5  μg/mouse. Four weeks after the second

immunization, the mice were infected intravenously

with the virulent M.  tuberculosis H37Rv strain at a

dose of 500,000  CFU/mouse. The BCG vaccination was

performed once subcutaneously at a dose of 100,000

CFU/mouse 5 weeks before the infection. The weight

loss in the infected mice was assessed by weekly

weighing the animals. Fifty days after the infection,

the mycobacterial load of the spleen and lungs (CFU)

(5  mice/group) and the dynamics of mouse death

(10  mice/group) were evaluated.

Cloning and in vitro transcription. The con-

structs for subsequent in  vitro RNA transcription

were cloned into the commercial pSmart vec-

tor (Lucigen, USA) with the introduced polyA tail.

MULTIEPITOPE mRNA VACCINE PROTECTS AGAINST TUBERCULOSIS 759

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

The 5′ and 3′ UTRs were fused to the coding sequence

by the overlap PCR. The resulting insert consisting of

the 5′ UTR, coding sequence, and 3′ UTR was cloned

into the vector at the EcoRI and BglII restriction sites.

The obtained vectors were used for the transforma-

tion of NEB-stable cells (New England Biolabs, UK)

grown at 30°C on a shaker (180rpm). The nucleotide

sequence of the construct was confirmed by Sanger

sequencing. Plasmid DNA was isolated with a Qiagen

Plasmid Maxi Kit (Qiagen, USA) and linearized at the

unique AhlI restriction site located after the polyA tail.

In vitro transcription was performed in the reac-

tion mixture containing 5 μg of the linearized plas-

mid, reagents from the RNA-20 Kit (Biolabmix, Russia),

1.2  mM synthetic analog m

7,3

′-OGpppAmG (CapAG;

Biolabmix), 40  U of RiboCare ribonuclease inhibitor

(Evrogen, Russia), and 0.002  U of inorganic pyrophos-

phatase (New England Biolabs) according to the man-

ufacturer’s protocol. A standard set of nucleotides

(ATP, GTP, CTP, UTP) was used for in  vitro transcrip-

tion. The reaction was carried out for 2  h at 37°C, af-

ter which 3  mM each ribonucleoside triphosphate was

added, and the reaction mixture was incubated for

another 2  h. DNA was hydrolyzed with RQ1 nuclease

(Promega, USA), and RNA was precipitated with LiCl

(final concentration, 0.32 M) and EDTA (final concen-

tration, 20mM; pH8.0), followed by incubation on ice

for 1  h. The length and purity of the synthesized RNA

molecules were assessed by capillary electrophoresis

in a Qsep100 Bio-Fragment Analyzer (BiOptic, Taiwan).

mRNA encapsulation into LNPs. RNA was dis-

solved at 0.2  mg/ml in 10 mM citrate buffer (pH  3.0)

and mixed with lipids dissolved in absolute ethanol in

a microfluidic cartridge in a NanoAssemblr™ nanopar-

ticle formulation system (Precision Nanosystems,

Canada). The lipid mixture contained 42.7% choles-

terol (Sigma-Aldrich), 23.15% synthetic lipid ALC-0315

(BroadPharm, USA), 9.4% 1,2-distearoyl-sn-glycero-3-

phosphocholine (DSPC; Avanti Polar Lipids, USA); 1.6%

polyethylene glycol 2000 (PEG 2000; BroadPharm);

and 23.15% cationic lipid SM-102 (Cayman Chemical,

USA). The content of mRNA in the LNPs was 0.04% (by

weight). To obtain the LNPs, the aqueous and alcohol

phases were mixed at a 3  :  1 (v/v) ratio; the total mix-

ing rate was 10 ml/min. The obtained particles were

dialyzed against PBS (pH 7.4) for 18  h. The quality of

the LNPs was assessed using four parameters: particle

size, polydispersity index (Zetasizer Nano ZSP, USA),

mRNA loading, and RNA integrity. Particle measure-

ments were performed with a Nanobrook Series in-

strument (Brookhaven Instruments, USA) in a DTS0012

cuvette. The original LNP suspension was diluted

100-fold with PBS (pH  7.4). The concentration of

mRNA packed into the LNPs was calculated from the

difference in fluorescence before and after disruption

of the nanoparticles stained with RiboGreen (Thermo

Fischer Scientific, USA). The particles were disrupted

with Triton X-100 (Sigma-Aldrich). The fraction of RNA

encapsulated into the LNPs in all samples was more

than 90% of total (encapsulated  +  free) RNA. The par-

ticle size was 97-118  nm, and the polydispersity index

was 0.149-0.186. The RNA quality and integrity in the

particles was assessed by capillary electrophoresis

with a Qsep100 Bio-Fragment device after LNPs dis-

ruption (Online Resource 1).

Quantitative assessment of the T cell response

to immunization. The number of splenocytes secret-

ing IFN-γ in response to the stimulation with the

M.  tuberculosis sonicate (soluble fraction of the ultra-

sonic lysate of M.  tuberculosis H37RV cells) was de-

termined by the ELISpot assay using a Mouse IFN-γ

ELISpot Kit (BD, USA) and an AEC substrate kit (BD),

which allowed to quantify the level of the T  cell re-

sponse in immunized mice.

Splenocytes were isolated from the spleen of im-

munized and control mice under sterile conditions as

described previously [62]. The splenocytes were seed-

ed at 200,000  cells/well in ELISpot multi-well plates

containing a PVDF membrane (BD) and simultaneous-

ly stimulated by adding the M.  tuberculosis sonicate

(10  μg/ml). The total volume of the liquid in each well

was brought up to 200  μl with RPMI 1640 culture me-

dium (PanEco, Russia) supplemented with 10% fetal

bovine serum (Biowest, France) and the cells were

incubated for 20  h in a CO

incubator (5%  CO

, 37°C).

The cells that had not been stimulated and cells stim-

ulated with concanavalin  A (Sigma-Aldrich) at a dose

of 5  μg/ml served as the negative and positive con-

trols, respectively. Spots corresponding to splenocytes

secreting IFN-γ were counted with an S6 Ultra ELISpot

Reader (ImmunoSpot, USA).

Mycobacterial antigens. Mycobacterial antigens

were kindly provided by V.  G.  Avdienko. M.tuberculo-

sis cells (strain H37Rv) disrupted by sonication were

used as antigens. To obtain the sonicate, mycobacteria

were grown in Sotona synthetic broth (HiMedia, India)

for 28 days at 37°C. The washed bacterial mass was

disintegrated by sonication with a Soniprep 150 Ultra-

sonic Disintegrator (MSE, United Kingdom) according

to the procedure described in [63]. Protein concentra-

tion was determined by the Bradford method.

Determination of the titers of IgG against M.tu-

berculosis antigens. The titers of IgG against myco-

bacterial antigens were determined in mouse blood

serum by the enzyme immunoassay using the previ-

ously described technique [61]. Briefly, 96-well plates

(Helicon, Russia) were coated with the mycobacterium

sonicate (10  μg/ml in 50  mM carbonate-bicarbonate

buffer, pH  6.1) and incubated for 18-hour at 4°C. Next,

mouse blood serum from the immunized and control

animals was added to the wells at the dilutions from

1  :  50 to 1  :  400. The reaction was visualized with

KAZAKOVA et al.760

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

horseradish peroxidase-labeled polyclonal antibodies

against mouse IgG (Jackson ImmunoResearch Labo-

ratories, USA) using tetramethylbenzidine (TMB; R&D

Systems, USA) as a peroxidase substrate.

Evaluation of the protective response after im-

munization with the mRNA vaccine. The mycobac-

terial load in the internal organs of infected mice was

analyzed on the 50th day after the infection. Spleen and

lung tissue homogenates were prepared under sterile

conditions. Serial 10-fold dilutions of the homogenates

were plated onto Petri dishes (50μl per dish) with Mid-

dlebrook agar 7H10 (Liofilchem, Italy) and incubated

at 37°C. After 21 days, the macrocolonies of M. tuber-

culosis H37Rv were counted in the dishes, and these

data were converted to the colony number per organ.

Statistical analysis. The data were processed by

the one-way ANOVA with the Dunnett’s correction for

multiple comparisons. The overall survival was esti-

mated by the Kaplan–Meier method. The significance

of differences in the overall survival was calculated

by the Mantel–Cox log-rank test. The differences be-

tween the groups were considered statistically signif-

icant at p<0.05. Data analysis and visualization were

performed with the GraphPad Prism 9.5.1 software.

RESULTS

Formation of adaptive immunity and effective

protection against M.  tuberculosis via activation of

B and T lymphocytes can achieved by selection of

optimal HTL, CTL, and LBL epitopes for the antitu-

berculosis vaccine. In this work, we used various im-

munoinformatic tools to predict and evaluate B- and

T-cell epitopes. The efficacy and safety of all obtained

epitopes were assessed by analyzing their coverage in

the population of the Russian Federation, antigenicity,

nonallergenicity, nontoxicity, and autoimmunogenici-

ty (Fig. 2). In total, seven epitopes of each type were

chosen for the prospective antituberculosis vaccine.

Also, the tPA and MITD sequences were inserted into

the coding sequence to direct the encoded protein to

the endoplasmic reticulum and Golgi apparatus for

its secretion and presentation by the MHC molecules.

Furthermore, it is known that the tPA sequence in-

creases the immunogenicity of a vaccine without

changing the balance between the Th1 and Th2 im-

mune responses [64,  65]. We assumes that the inclu-

sion of two signal sequences would ensure a balance

between the secretion of the target peptide and its

Fig. 2. Stages in the development of the multiepitope mRNA vaccine against tuberculosis.

MULTIEPITOPE mRNA VACCINE PROTECTS AGAINST TUBERCULOSIS 761

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

Fig.  3. Design of the in vivo experiments. a) Evaluation of the vaccine immunogenicity in B6 mice; b) Evaluation of the

vaccine protective properties in I/St mice.

proteasomal degradation and presentation in the con-

tent of MHC I. We also used two adjuvant sequenc-

es (RpfE from M.  tuberculosis and human FL). After

designing the amino acid sequences of the vaccines,

we applied two cardinally different optimization algo-

rithms, IDT and LD, to obtain four mRNA sequences

(mRNA-mEp21-RpfE-IDT, mRNA-mEp21-FL-IDT, mRNA-

mEp21-RpfE-LD, and mRNA-mEp21-FL-LD). All mRNAs

contained a set of essential mRNA elements, such as

the cap, 5′ and 3′ UTRs, and the polyA tail of 110 ad-

enine residues.

The designed mRNAs were synthesized, encap-

sulated into LNPs, and injected into mice for testing

their immunogenicity and protective efficacy against

M. tuberculosis infection. The full workflow of the

study is presented in Fig. 3.

The target M. tuberculosis proteins that play a

key role in the infection, associated pathological pro-

cess, and induction of a strong immune response,

were selected based on our recent systematic review

of vaccines against M. tuberculosis [26]. We select-

ed the most commonly used M.  tuberculosis anti-

gens: ESAT6 (WGG95196.1), CFP10 (WGG95197.1),

Ag85B (WP_262239103.1), Ag85A (P9WQP3.1), HspX

(CCE37510.1), MPT83 (ANZ83596.1), PE/PPE family

proteins PPE55 (VCU51654.1) and PPE68 (CCE39296.1,

SGL17895.1, CNN07764.1, and AAC32213.1), Mtb8.4

(CAA15851), Rv1733c (P9WLS9.1), MPT64 (CAA53143.1),

and Hrp1 (CCP45424.1). ESAT6 and CFP10 are the

low-molecular-weight secretory proteins that are im-

portant for the mycobacterial virulence and pathoge-

nicity and encoded by genes of the RD1 locus, which

is absent in the BCG [66, 67]. The secreted proteins

Ag85A and Ag85B, which bind fibronectin and have

the mycolyltransferase activity, are essential for main-

taining the integrity of the mycobacterial cell wall.

The HspX chaperone is a dominant antigen produced

by M. tuberculosis at the latent infection stage; it is

thought to enhance the long-term stability of proteins

and cellular structures, thereby helping the tuberculo-

sis pathogen to survive for a long time [68, 69]. MPT83

is a glycosylated lipoprotein found on the surface of

mycobacterial cells that evokes a strong T-cell re-

sponse and antibody production [70]. Proteins of the

PE/PPE family are characterized by the presence of

conserved Pro-Glu (PE) and Pro-Pro-Glu (PPE) motifs

in their N-terminal regions, respectively; they induce

activation of B and T cells and modulate the host’s

immune response [71]. Mtb8.4 is an immunoreactive

antigen for T cells and is found in humans with la-

tent mycobacterial infections; it induces high levels

of IFN-γ (Th1 cytokine) secretion [72]. Rv1733c is the

KAZAKOVA et al.762

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

main latency antigen of M.  tuberculosis; it is actively

expressed by latent M.  tuberculosis cells and is well

recognized by T cells of infected individuals  [73].

MPT64 promotes the antituberculosis cellular immuni-

ty and activates macrophages, thus inducing secretion

of IL-1β, IL-6, IL-10, and TNF-α by these cells. MPT64

can also partially inhibit the apoptosis in macrophages

[74]. HRP1 is a secreted protein encoded by the DosR

locus and has a strong antigenic activity [75,  76].

Prediction and evaluation of LBL, CTL, and

HTL epitopes. Using the ABCpred online server, we

obtained 448 LBL epitopes. Their analysis in VaxiJen,

ToxinPred, AllerTOP 2.0, and BLAST yielded 185 an-

tigenic, nontoxic, and nonallergenic epitopes with

no homology to human and mouse sequences. From

these, we selected seven most antigenic epitopes.

According to the IgPred server, none of these epitopes

stimulated IgE production; at the same time, all epi-

topes exhibited a high similarity to the epitopes in-

ducing IgG production.

Using the T Cell Prediction Class I service, 18,547

CTL epitopes were obtained for all the studied anti-

gens; 176 epitopes demonstrating the best binding to

MHC I were chosen for further analysis. Their assess-

ment with VaxiJen produced 126 epitopes proven to

be antigenic with an over 66% probability. According

to ToxinPred, 125 of them were nontoxic, and among

these, 72 epitopes were identified as nonallergenic

by AllerTOP 2.0. Analysis for the homology with hu-

man and mouse sequences limited this number to 70

epitopes, from which seven most antigenic epitopes

were selected (the probability of the selected epitopes

being antigens was 100%, based on the verification

by the VaxiJen v3.0 service) and used for construc-

tion of the vaccine proteins and molecular docking

with MHC I.

Using the MHC-II Binding Predictions service,

150,000 HTL epitopes were found in all antigen se-

quences. After all necessary checks, 192 HTL epitopes

were selected, which were assessed for their ability to

induce the production of IFN-γ, IL-4, and IL-10. Given

that IFN-γ is a key cytokine in the antituberculosis

response, we mostly selected the epitopes capable of

inducing the synthesis of this protein. On the other

hand, induction of the anti-inflammatory cytokines

IL-4 and IL-10 positively correlates with the increase

in the M.tuberculosis load and, therefore, is undesir-

able. Accordingly, only HTL epitopes that presumably

did not induce the expression of these two cytokines

were used in our research. Seven most antigenic epi-

topes (100% probability of being antigens based on the

verification by the VaxiJen v3.0 service) were utilized

for the construction of the vaccine protein and mo-

lecular docking with MHC II. More detailed informa-

tion on the selected epitopes is presented in Table  S1

in Online Resource 2.

Molecular docking between T-lymphocyte epi-

topes and MHC alleles. The results of the HLA pro-

tein docking with CTL and HTL epitopes are presented

in Fig.  S1 in Online Resource  2. The best RMSD and

binding free energy values for MHCI were found for

the interaction between the AGNFERISGD epitope and

HLA-A*24:02. For MHC II, the best parameters were

found for the GLLDPSQAMGPTLIG epitope interaction

with HLA-DRB1*15:01 (Fig. S2 in Online Resource 2).

For 4 out of 7 CTL epitopes, the RMSD was less than

3  Å, implying a high probability of interaction with

MHC I. Among the HTL epitopes, six peptides out of

seven had the RMSD values < 3  Å. Nevertheless, we

decided to include all the studied epitopes in the vac-

cine, since these epitopes had been predicted to in-

teract with other MHC alleles other than alleles used

for the docking procedure, although this interaction

could not be confirmed by docking because of the ab-

sence of corresponding 3D structures in the databases.

It is likely that the docking results obtained for such

alleles and their epitopes could be even more accept-

able than those presented in this article. Moreover,

the docking with MHC  I and MHC  II was only one of

the ways to assess the quality of the epitopes used

in our study. Moreover, the other parameters demon-

strated by the epitopes with less acceptable RMSD

values were comparable to the values exhibited by

the epitopes with “good” RMSD, therefore, we did not

exclude them from the study.

Prediction of antigenicity, allergenicity, toxici-

ty, and physicochemical properties of the vaccine

proteins. The amino acid sequences of the vaccine

proteins were tested for antigenicity, allergenicity, and

toxicity using the VaxiJen, ANTIGENpro, AllerTOP, and

ToxinPred servers. Both mRNA-mEp21-FL and mRNA-

mEp21-RpfE vaccines types were found to be antigen-

ic, nonallergenic, nontoxic, and stable, with II  <  40

(conventional units of this server). The mRNA-mEp21-

FL vaccine was also thermally stable; the GRAVY pa-

rameter was < 0 for both vaccine types, indicating the

hydrophilicity of the resulting constructs (Table  S2 in

Online Resource 2).

Determination and confirmation of the tertia-

ry structure of vaccine proteins. The tertiary struc-

ture was obtained with the Robetta server. Because

Robetta offered five models for each construct, we

used PROCHECK to select a model with the best ste-

reochemical accuracy for further work (Table  S3 in

Online Resource  2). The best models are presented in

Fig.  S3 in Online Resource  2.

Prediction of B-cell conformational epitopes.

The formation of the tertiary structure gives rise to

B-cell conformational epitopes, which were predict-

ed using the ElliPro server. The evaluation of dis-

continuous conformational epitopes is presented in

Table  S4 in Online Resource  2. In mRNA-mEp21-FL,

MULTIEPITOPE mRNA VACCINE PROTECTS AGAINST TUBERCULOSIS 763

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

294  a.a. were found in the conformational B-epitopes

with the predicted values from 0.915 to 0.56 (conven-

tional units ranging from 0 to 1 showing how much

an isolated epitope “protrudes” beyond the average

boundaries of the protein represented as an ellipsoid;

larger values indicate that the protein region is more

likely to protrude outward, i.e., likely to act as a B  cell

epitope). In mRNA-mEp21-RpfE, 269 a.a. were found

with the predicted values from 0.861 to 0.527. In the

RpfE-containing vaccine protein, the bulk of the adju-

vant and most of the LBL and CTL epitopes turned out

to be B-cell conformational epitopes. By contrast, in

the FL-containing vaccine protein, the conformational

epitopes were mostly HTL and LBL epitopes. The 2D

and 3D models of the B-cell conformational epitopes

in the vaccine proteins are shown in Figs.  S4-S6 in

Online Resource 2.

Molecular docking of the vaccine proteins with

TLR4. The affinity of the mEp21-RpfE and mEp21-

FL proteins toward TLR4 was comparable to that

of RpfE, which is expected to stimulate an immune

response against M.  tuberculosis (Table S5 in Online

Resource  2). The hypothetical interactions of the TLR4

complex with the mEp21-RpfE and mEp21-FL proteins

(Fig.  S7 in Online Resource 2) were examined on the

PDBsum server (Fig.  S8 in Online Resource 2). The re-

sults of molecular docking indicated that the vaccine

proteins presumably have an affinity for TLR4 that

is similar to the affinity of RpfE, as well as possess

comparable adjuvant characteristics.

In  silico immunization. The used model includ-

ed three injections of each investigated vaccine to

simulate the immune response, and produced similar

predicted immune responses for all the tested vaccines

(Fig.  S9 in Online Resource2). The modeling results im-

plied a higher extent of IgM induction as compared to

IgG. Furthermore, after a decrease in the vaccine anti-

gen level, the content the immunoglobulins remained

high, and the number of memory B  cells increased,

possibly indicating formation of the immune memo-

ry after exposure to the antigen. The emergence of

memory cells and active Tc (T  cytotoxic) lymphocytes

among the Th (T  helper) and Tc  cell populations, re-

spectively, was possible. The levels of IFN-γ and IL-2

(important cytokines involved in the progression of

immune response against tuberculosis) were also

predicted to increase after the immunization. The

Simpson index  (D) remained low during the entire

period of insilico immunization, implying a potential

production of interleukins and cytokines. Therefore,

the results of in silico modeling demonstrated that all

variants of the developed vaccines were presumably

effective and safe.

Optimization of codon composition and remov-

al of miRNA-binding sites. The employed optimiza-

tion algorithms (Linear Design, RiboTree, IDT, genewiz,

and iCodon) were compared with each other via the

Superfolder algorithm (Fig.S10 in Online Resource 2).

Comparison of the key parameters of mRNA stability

and translatability allowed us to select two optimi-

zation algorithms: Linear Design and IDT. According

to the results of in silico analysis, optimization with

Linear Design considerably extended the half-life of

RNA, suggesting mRNA stabilization and a hypotheti-

cal increase in the number of synthesized protein mol-

ecules. However, Linear Design showed a low AUP init

14 value, which could potentially diminish the transla-

tional efficiency, as low AUP values could slow down

the movement of ribosomes along the highly struc-

tured RNA molecule. The IDT algorithm yielded high-

er AUP and AUP init 14 values, presumably affecting

the availability of mRNA for protein synthesis. Various

optimization algorithms were applied to each vaccine

protein. As a result, four mRNA sequences were de-

veloped: mRNA-mEp21-RpfE-LD, mRNA-mEp21-FL-LD,

mRNA-mEp21-RpfE-IDT, and mRNA-mEp21-FL-IDT

(Fig.4). The stability characteristics of the four mRNA

vaccine sequences are presented in Fig.  S11 in Online

Resource 2.

We also removed the miRNA-binding sites from

all four constructs, because of the ability of such sites

causing RNA interference and translation inhibition.

The final secondary structures of all four full-length

mRNAs obtained using RNAfold (http://rna.tbi.univie.

ac.at/cgi-bin/RNAWebSuite/RNAfold.cgi) are presented

in Fig.4. As one can see, the sequences obtained with

the Linear Design algorithm are much more struc-

tured than the sequences obtained with IDT.

The efficacy of the developed mRNA vaccines in

in vivo experiments. To evaluate the immunogenic-

ity of the antituberculosis multiepitope mRNA-mEp21

vaccines in B6 mice, we quantified the cellular and

humoral immune responses to the mycobacterial

antigens. The cellular response was assessed by the

ELISpot assay, and the humoral immunity was deter-

mined as the titers of IgG antibodies to mycobacterial

antigens from the M. tuberculosis sonicate.

After immunization, there were differences in the

number of IFN-γ-producing cells between the mouse

groups [ANOVA: F(6,  28)  =  11.58, p  <  0.001]. Thus, the

number of IFN-γ-producing cell increased only in ani-

mals immunized with mRNA-mEp21-RpfE-LD and BCG

compared to the control group (p  =  0.042 and 0.012,

respectively; Fig.  5a). Hence, only one of the devel-

oped mRNA vaccines ensured the induction of cellular

immunity against M.  tuberculosis antigens, and this

response was less pronounced compared to the im-

munization with BCG.

Immunization induced a significant increase in

the titers of IgG antibodies against mycobacterial an-

tigens at four serum dilutions: 1  :  50, 1  :  100, 1  :  200,

and 1  :  400 [ANOVA, 1  :  50: F(6,  28)  =  13.23, p  <  0.0001;

KAZAKOVA et al.764

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

Fig.  4. Secondary MFE (minimum free energy) structures of vaccine mRNAs. a) mRNA-mEp21-RpfE (Linear Design);

b)  mRNA-mEp21-RpfE (IDT); c)  mRNA-mEp21-FL (Linear Design); d)  mRNA-mEp21-FL (IDT). The color scale indicates the

probability of base pairing, where 0 is the minimal probability and 1 is the maximal probability.

1  :  100: F(6,  28)  =  16.31, p  <  0.001; 1  :  200: F(6,  28)  =

10.91, p  <  0.001; 1  :  400: F(6,  28)  =  10,62, p  <  0.001].

The maximum serum titer at which significant differ-

ences were observed between the mice receiving the

vaccine and the negative control group was 1  :  50 for

mRNA-mEp21-FL-IDT, 1  :  100 for mRNA-mEp21-FL-LD,

1  :  200 for mRNA-mEp21-RpfE-IDT, and 1  :  400 for

mRNA-mEp21-RpfE-LD (Fig.  5d). Moreover, no signif-

icant differences were found among different mRNA

vaccines. On the other hand, immunization with BCG

produced higher IgG titers. Therefore, immunization

with experimental mRNA vaccines led to the forma-

tion of the humoral IgG immune response to mycobac-

terial antigens at all four serum dilutions. However,

similar to the cellular immune response, the induced

immunity was manifested less compared to the vac-

cination with BCG.

To assess the protective properties of the mRNA

vaccines, prophylactically immunized I/St mice were

infected with M.  tuberculosis, and the mycobacte-

rial load in their lungs and spleen was determined

50  days after the infection. We found that immuni-

zation affected the bacterial loads in the lungs and

spleen [F(6,  28)  =  4.807, p  =  0.01, and F(6,  28)  =  10.28,

p  <  0.001]. In the lungs, both BCG and mRNA-mEp21-

FL-IDT decreased the bacterial load (p  <  0.001 and

p  =  0.002, respectively, Fig.  5b), whereas in the spleen,

the reduction in the bacterial load was observed only

after vaccination with BCG (p  =  0.011). Immunization

with any mRNA vaccine did not significantly lower

the bacterial load in the spleen, even in the case of

mRNA-mEp21-FL-IDT (p  =  0.105, Fig.  5c). Administra-

tion of mRNA-mEp21-FL-IDT diminished the num-

ber of CFU by 79.37% (0.686  log

  units) in the lungs

and by 86.2% (0.86  log

  units) in the spleen, where-

as BCG reduced the number of CFU in the lungs by

90.6%  (1.025  log

  units) and by 80.4% in the spleen

(0,708  log

  units).

Mouse survival after infection with M.  tubercu-

losis was evaluated on day 111 after the infection

(Fig.  5e). In the control group (mice that received PBS),

eight out of 10 animals had died by the end of the

experiment, whereas in the BCG and mRNA-mEp21-

FL-IDT groups, only two mice out of 10 had died.

The Kaplan–Meier survival analysis revealed a sig-

nificant extension of the lifespan of mice vaccinated

MULTIEPITOPE mRNA VACCINE PROTECTS AGAINST TUBERCULOSIS 765

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

Fig.  5. In  vivo experiments. The number of splenocytes secreting IFN-γ, IgG titers, CFU in the lungs and spleen (n  =  5 mice

per group), and survival after infection (n  =  10 mice per group) were evaluated. a) Estimation of the number of cells se-

creting IFN-γ in response to the stimulation with M.  tuberculosis antigens by the ELISpot assay; b and c)evaluation of the

bacterial load in the (b) lungs and (c) spleen (the number of M.tuberculosis CPUs in the tissue homogenate 50 days after

infection); d)IgG titers in response to the stimulation with M.  tuberculosis antigens; e)survival of mice after infection with

M. tuberculosis; *  p  <  0.05; **  p  <  0.01; ***  p  <  0.001; ****  p  <  0.0001.

with BCG or mRNA-mEp21-FL-IDT as compared to an-

imals of the control group (Mantel–Cox log-rank test,

p  =  0.0011). It should be noted that the weight loss

dynamics in mice of the mRNA-mEp21-FL-IDT group

was similar to that in the BCG group (data not shown).

Overall, our results showed that despite induc-

ing a less pronounced adaptive response, the mRNA-

mEp21-FL-IDT mRNA vaccine provided efficient pro-

tection (comparable with that of BCG) after a chal-

lenge with M. tuberculosis.

DISCUSSION

Here, we used immunoinformatic methods to de-

velop a new multiepitope antituberculosis mRNA vac-

cine encoding 21 epitopes from 11 proteins of M.  tu-

berculosis. We also used the sequences of molecular

adjuvants RpfE and FL and employed two strategies

for optimizing the secondary structure of the RNA

molecule (Linear Design and IDT), which allowed

us to obtain four mRNA vaccines, which efficacy

KAZAKOVA et al.766

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

was tested in mice. All the developed vaccines induced

humoral immunity against M. tuberculosis antigens,

which corresponded to the results of in silico immu-

nization. On the other hand, only the mRNA-mEp21-

RpfE-IDT vaccine induced the cellular immunity,

which is at odds with the data obtained by in  silico

immunization. According to these data, vaccines with

the RpfE adjuvant should cause weaker cellular immu-

nity compared to vaccines with the FL adjuvant, since

the former is less likely to stimulate the maturation

of cytotoxic T-lymphocytes. In our experiments, the

most efficient protection against tuberculosis was pro-

vided by the mRNA-mEp21-FL-IDT vaccine containing

the FL molecular adjuvant. Our results have shown

that the two-dose (5μg each dose) immunization with

RNA-mEp21-FL-IDT reduced the mycobacterial load in

the lungs and increased the animal survival rate to

80%, which was comparable to the efficacy of the BCG

vaccine.

The observed differences in the effects of the in-

corporated molecular adjuvants may be explained by

their functions. The FL sequence has been widely used

in the vaccines against the pseudorabies virus, simi-

an immunodeficiency virus, etc., as well as in cancer

immunotherapy [77-79]. FL mobilizes and stimulates

DCs, natural killer cells, and B cells. Its inclusion in the

DNA vaccines enhances the immune response [80-82].

In our work, the FL-based vaccine induced the hu-

moral response and the protective immunity, but the

cellular response was modest (less than 12 IFN-γ spots

per well). In other studies on the development of DNA

vaccines against tuberculosis, incorporation of the FL

sequence has caused more pronounced formation of

both humoral response (as judged by the IgG titer)

and cellular response (as evidenced by the elevated

production of IFN-γ by splenocytes after their specific

activation), as well as caused a more significant decline

in the mycobacterial load in the lungs and spleen (al-

though this effect was still less pronounced compared

with the BCG vaccine) [17, 83-85]. In most of research

articles, however, the efficacy of antituberculosis DNA

vaccines has been inferior to that of BCG [26].

RpfE is a M. tuberculosis protein that is an ag-

onist of TLR4. TLR4 plays an important role in the

protection against tuberculosis: the knockout of the

TLR4 gene leads to the increase in the bacterial load

in infected mice [86]. In our study, incorporation of

the RpfE sequence ensured a more pronounced for-

mation of cellular immunity but did not ensure the

protection of the animals after the infection. Simi-

larly to our findings, Xin et  al.  [87] showed that the

inclusion of RpfE strengthened the cellular immunity

against M.  tuberculosis but did not reduce the bac-

terial load in the infected animals. We believe that

the lack of the protective effect may be due to the

excessive immunogenicity of the adjuvant and to the

development of systemic inflammation after its ad-

ministration. Thus, we observed spleen enlargement,

inflammation at the injection site, and intestinal in-

flammation in some mice of the mRNA-mEp21-RpfE

group (data not shown). However, no negative effects

of RpfE have been observed in other studies of this

adjuvant [88, 89].

A distinctive feature of our work is application of

two cardinally different algorithms (LD and IDT) for

optimizing the RNA secondary structure. Presumably,

LD promotes an increase in the RNA lifetime and de-

creases the number of unpaired bases, whereas IDT

ensures the balanced codon use and allows to avoid

rare codons. We did not find significant differences

in the efficacy of mRNA vaccines developed with dif-

ferent optimization algorithms, although it has been

reported earlier that the LD algorithm notably en-

hanced the humoral immune response in mice after

administration of mRNA vaccines against COVID-19

and chickenpox virus [58].

Overall, we used immunoinformatic methods to

create a multiepitope antituberculosis vaccine, mRNA-

mEp21-FL-IDT, which induced the humoral immunity

and reduced bacterial load in the lungs, while improv-

ing animal survival after infection with M.tuberculo-

sis with the efficiency comparable to that of the BCG

vaccine. However, the results of our work do not allow

to conclude about the efficacy of the adjuvants or op-

timization algorithms. In our previous studies, the de-

veloped multiepitope mRNA antituberculosis vaccines

encoding five epitopes of the immunodominant pro-

tein ESAT6 (MTB-mEp-5-1 and mEpitope-ESAT6) were

highly immunogenic, but their protective properties

were worse than those of BCG [90, 91]. Other articles

on the design of antituberculosis DNA and RNA vac-

cines also indicate that in most cases, the vaccines

induced the adaptive immunity but failed to pro-

tect from M.  tuberculosis infection  [10,  26]. Moreover,

even when the protective response was successfully

induced, only in rare cases immunization with nu-

cleic acid-based antituberculosis vaccines provided a

stronger protection compared to BCG [26]. In this con-

text, the most efficient DNA vaccines have been those

encoding the ESAT6, CFP10, CFP21, MTP64, Ag85B, or

MPT63 antigens [17, 92, 93]. Among them, ESAT6 pro-

vided the best protection, while the addition of the

adjuvant (FL, IL-12, or IL-23) has further reduced the

number of CFU.

In conclusion, here we combined for the first time

the multiepitope design of an mRNA vaccine, inclusion

of molecular adjuvant sequence, and RNA sequence

optimization, which allowed us to obtain an effective

antituberculosis mRNA vaccine (mRNA-mEp21-FL-IDT)

with a protective effect against M. tuberculosis infec-

tion. The efficacy and safety of mRNA-mEp21-FL-IDT

should be confirmed in expanded preclinical studies.

MULTIEPITOPE mRNA VACCINE PROTECTS AGAINST TUBERCULOSIS 767

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

Abbreviations. AUP, average unpaired probabil-

ity; BCG, Bacillus Calmette–Guérin vaccine; CFU, colo-

ny-forming unit; CTL, cytotoxic T lymphocyte; DC, den-

dritic cell; FL, FMS-like tyrosine kinase-3 ligand; HLA,

human leukocyte antigens; HTL, helper T lymphocyte;

LBL, linear B lymphocyte; LNP, lipid nanoparticle;

MHC, major histocompatibility complex; MITD, MHC

I-targeting domain; RMSD, root-mean-square distance;

RpfE, resuscitation-promoting factor; TLR, Toll-like

receptor; tPA, tissue plasminogen activator; UTR, un-

translated region.

Supplementary information. The online version

contains supplementary material available at https://

doi.org/10.1134/S0006297925600073.

Acknowledgments. The authors are grateful to

O. V. Zaborova (Lomonosov Moscow State University)

for her assistance with the mRNA encapsulation into

LNPs. Equipment from the Resource Centers for ge-

netic engineering and biotechnological products at the

Sirius University of Science and Technology was used

in the study.

Contributions. A.A.K., I.S.K., and V.V.R. developed

the genetic constructs, produced the plasmids, pre-

pared mRNAs, and wrote the manuscript; R.A.I. and

V.V.R. developed the study concept, supervised the

research, discussed the results, and edited the manu-

script; G.S.Sh. and V.V.E. conducted invivo experiments

and edited the manuscript.

Funding. The project was supported by the Min-

istry of Science and Higher Education of the Russian

Federation (Agreement no.075-10-2025-017).

Ethics approval and consent to participate. The

animal study protocol was approved by the Ethics

Committee of the Central Tuberculosis Research In-

stitute (protocol no. 3/2; May 11, 2023) in accordance

with the EU Directive 2010/63/EU on animal experi-

ments and was implemented in accordance with the

European Convention ETS no. 123 for the Protection of

Vertebrate Animals used for Experimental or Scientif-

ic Purposes (Strasbourg) (1986, with the 2006 amend-

ment), the International Agreement on Humane Treat-

ment of Animals (1986), and the Guide for the Care

and Use of Laboratory Animals, 8th ed. (2010).

Conflict of interest. The authors of this work de-

clare that they have no conflicts of interest.

REFERENCES

1. Global tuberculosis report 2024 (2024) URL:

https://www.who.int/teams/global-programme-on-

tuberculosis-and-lung-health/tb-reports/global-

tuberculosis-report-2024.

2. Colditz, G. A., Brewer, T. F., Berkey, C. S., Wilson,

M. E., Burdick, E., Fineberg, H. V., and Mosteller, F.

(1994) Efficacy of BCG vaccine in the prevention of

tuberculosis. Meta-analysis of the published liter-

ature, JAMA, 271, 698-702, https://doi.org/10.1001/

jama.1994.03510330076038.

3. Gandhi, R. T., Kwon, D. S., Macklin, E. A., Shopis,

J. R., McLean, A. P., McBrine, N., Flynn, T., Peter, L.,

Sbrolla, A., Kaufmann, D. E., Porichis, F., Walker,

B. D., Bhardwaj, N., Barouch, D. H., and Kavanagh,

D. G. (2016) Immunization of HIV-1-infected per-

sons with autologous dendritic cells transfected with

mRNA encoding HIV-1 Gag and Nef: results of a ran-

domized, placebo-controlled clinical trial, J. Acquir

Immune Defic., 71, 246-253, https://doi.org/10.1097/

QAI.0000000000000852.

4. Richner, J.M., Himansu,S., Dowd, K. A., Butler, S.L.,

Salazar, V., Fox, J. M., Julander, J. G., Tang, W. W.,

Shresta, S., Pierson, T. C., Ciaramella, G., and Dia-

mond, M. S. (2017) Modified mRNA vaccines protect

against Zika virus infection, Cell, 6, 1114-1125, https://

doi.org/10.1016/j.cell.2017.02.017.

5. Alberer, M., Gnad-Vogt, U., Hong, H. S., Mehr, K. T.,

Backert, L., Finak, G., Gottardo, R., Bica, M. A.,

Garofano,A., Koch, S.D., Fotin-Mleczek,M., Hoerr,I.,

Clemens,R., and von Sonnenburg,F. (2017) Safety and

immunogenicity of a mRNA rabies vaccine in healthy

adults: an open-label, non-randomised, prospective,

first-in-human phase 1 clinical trial, Lancet, 390, 1511-

1520, https://doi.org/10.1016/S0140-6736(17)31665-3.

6. Bahl, K., Senn, J. J., Yuzhakov, O., Bulychev, A.,

Brito, L. A., Hassett, K. J., Laska, M. E., Smith, M.,

Almarsson,Ö., Thompson,J., Ribeiro, A.M., Watson,M.,

Zaks, T., and Ciaramella, G. (2017) Preclinical and

clinical demonstration of immunogenicity by mRNA

vaccines against H10N8 and H7N9 influenza virus-

es, Mol. Ther., 6, 1316-1327, https://doi.org/10.1016/

j.ymthe.2017.03.035.

7. Pateev,I., Seregina,K., Ivanov,R., and Reshetnikov,V.

(2023) Biodistribution of RNA vaccines and of their

products: evidence from human and animal stud-

ies, Biomedicines, 12, 59, https://doi.org/10.3390/

biomedicines12010059.

8. Hu, C., Bai, Y., Liu, J., Wang, Y., He, Q., Zhang, X.,

Cheng, F., Xu, M., Mao, Q., and Liang, Z. (2024) Re-

search progress on the quality control of mRNA

vaccines, Expert Rev. Vaccines, 23, 570-583, https://

doi.org/10.1080/14760584.2024.2354251.

9. Fan, X.Y., and Lowrie, D.B. (2021) Where are the RNA

vaccines for TB? Emerg. Microbes Infect., 10, 1217-

1218, https://doi.org/10.1080/22221751.2021.1935328.

10. Khlebnikova,A., Kirshina,A., Zakharova,N., Ivanov,R.,

and Reshetnikov, V. (2024) Current progress in the

development of mRNA Vaccines against bacterial

infections, Int. J. Mol. Sci., 25, 13139, https://doi.org/

10.3390/ijms252313139.

11. Cianci, R., and Franza, L. (2022) Recent advances in

vaccine technology and design, Vaccines (Basel), 10,

624, https://doi.org/10.3390/vaccines10040624.

KAZAKOVA et al.768

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

12. Zhang, L. (2018) Multi-epitope vaccines: a promising

strategy against tumors and viral infections, Cell.

Mol. Immunol., 15, 182-184, https://doi.org/10.1038/

cmi.2017.92.

13. Mahapatra, S.R., Dey,J., Kaur,T., Sarangi,R., Bajoria,

A.A., Kushwaha, G.S., Misra,N., and Suar,M. (2021)

Immunoinformatics and molecular docking studies

reveal a novel multi-epitope peptide vaccine against

pneumonia infection, Vaccine, 39, 6221-6237, https://

doi.org/10.1016/j.vaccine.2021.09.025.

14. Dey,J., Mahapatra, S.R., Lata, S., Patro, S., Misra, N.,

and Suar, M. (2022) Exploring Klebsiella pneumoni-

ae capsule polysaccharide proteins to design multie-

pitope subunit vaccine to fight against pneumonia,

Expert Rev. Vaccines, 21, 569-587, https://doi.org/

10.1080/14760584.2022.2021882.

15. Dey, J., Mahapatra, S. R., Patnaik, S., Lata, S.,

Kushwaha, G.S., Panda, R.K., Misra,N., and Suar,M.

(2022) Molecular characterization and designing of

a novel multiepitope vaccine construct against Pseu-

domonas aeruginosa, Int. J. Pept. Res. Ther., 28, 49,

https://doi.org/10.1007/s10989-021-10356-z.

16. Alharbi, M., Alshammari, A., Alasmari, A. F., Alharbi,

S. M., Tahir Ul Qamar, M., Ullah, A., Ahmad, S.,

Irfan, M., and Khalil, A. A. K. (2022) Designing of a

recombinant multi-epitopes based vaccine against

Enterococcus mundtii using bioinformatics and im-

munoinformatics approaches, Int. J. Environ. Res.

Public Health, 19, 3729, https://doi.org/10.3390/

ijerph19063729.

17. Jiang, Q., Zhang, J., Chen, X., Xia, M., Lu, Y., Qiu, W.,

Feng,G., Zhao,D., Li,Y., He,F., Peng,G., and Wang,Y.

(2013) A novel recombinant DNA vaccine encoding

Mycobacterium tuberculosis ESAT-6 and FL pro-

tects against Mycobacterium tuberculosis challenge

in mice, J. Biomed. Res., 5, 406-420, https://doi.org/

10.7555/JBR.27.20120114.

18. Choi, H. G., Kim, W. S., Back, Y. W., Kim, H., Kwon,

K.W., Kim, J.S., Shin, S.J., and Kim, H.J. (2015) Myco-

bacterium tuberculosis RpfE promotes simultaneous

Th1- and Th17-type T-cell immunity via TLR4-depen-

dent maturation of dendritic cells, Eur. J. Immunol.,

45, 1957-1971, https://doi.org/10.1002/eji.201445329.

19. Jung, I. D., Jeong, S. K., Lee, C. M., Noh, K. T., Heo,

D. R., Shin, Y. K., Yun, C. H., Koh, W. J., Akira, S.,

Whang,J., Kim, H.J., Park, W.S., Shin, S.J., and Park,

Y.M. (2011) Enhanced efficacy of therapeutic cancer

vaccines produced by co-treatment with Mycobac-

terium tuberculosis heparin-binding hemagglutinin,

a novel TLR4 agonist, Cancer Res., 71, 2858-2870,

https://doi.org/10.1158/0008-5472.CAN-10-3487.

20. Lee, S.J., Shin, S.J., Lee, M.H., Lee, M.G., Kang, T.H.,

Park, W. S., Soh, B. Y., Park, J. H., Shin, Y. K., Kim,

H. W., Yun, C. H., Jung, I. D., and Park, Y. M. (2014)

A potential protein adjuvant derived from Mycobac-

terium tuberculosis Rv0652 enhances dendritic cells-

based tumor immunotherapy, PLoS One, 9, e104351,

https://doi.org/10.1371/journal.pone.0104351.

21. Gote, V., Bolla, P. K., Kommineni, N., Butreddy, A.,

Nukala, P.K., Palakurthi, S.S., and Khan,W. (2023) A

comprehensive review of mRNA vaccines, Int. J. Mol.

Sci., 24, 2700, https://doi.org/10.3390/ijms24032700.

22. Muslimov, A., Tereshchenko, V., Shevyrev, D.,

Rogova, A., Lepik, K., Reshetnikov, V., and Ivanov,R.

(2023) The dual role of the innate immune system in

the effectiveness of mRNA therapeutics, Int. J. Mol.

Sci., 24, 14820, https://doi.org/10.3390/ijms241914820.

23. Mauger, D.M., Cabral, B.J., Presnyak,V., Su, S.V., Reid,

D.W., Goodman,B., Link,K., Khatwani,N., Reynders,J.,

Moore, M.J., and McFadyen, I.J. (2019) mRNA struc-

ture regulates protein expression through changes in

functional half-life, Proc. Natl. Acad. Sci. USA, 116,

24075-24083, https://doi.org/10.1073/pnas.1908052116.

24. Wayment-Steele, H. K., Kim, D. S., Choe, C. A., Nicol,

J. J., Wellington-Oguri, R., Watkins, A. M., Parra

Sperberg, R. A., Huang, P. S., Participants, E., and

Das,R. (2021) Theoretical basis for stabilizing messen-

ger RNA through secondary structure design, Nucleic

Acids Res., 49, 10604-10617, https://doi.org/10.1093/

nar/gkab764.

25. María, R. R., Arturo, C. J., Alicia, J. A., Paulina, M. G.,

and Gerardo, A. O. (2017) The impact of bioinfor-

matics on vaccine design and development, InTech,

https://doi.org/10.5772/intechopen.69273.

26. Kazakova, A., Zhelnov, P., Sidorov, R., Rogova, A.,

Vasileva, O., Ivanov, R., Reshetnikov, V., and

Muslimov, A. (2024) DNA and RNA vaccines against

tuberculosis: a scoping review of human and ani-

mal studies, Front. Immunol., 15, 1457327, https://

doi.org/10.3389/fimmu.2024.1457327.

27. Yang, Y., Chen, Y. Z., and Xia, T. (2024) Optimizing

antigen selection for the development of tuberculo-

sis vaccines, Cell Insight, 3, 100163, https://doi.org/

10.1016/j.cellin.2024.100163.

28. Saha, S., and Raghava, G. P. (2006) Prediction of

continuous B-cell epitopes in an antigen using re-

current neural network, Proteins, 1, 40-48, https://

doi.org/10.1002/prot.21078.

29. Saha, S., and Raghava, G. P. S. (2004) BcePred: pre-

diction of continuous B-Cell epitopes in antigenic

sequences using physico-chemical properties, Lec-

ture Notes Comput. Sci., 3239, 197-204, https://

doi.org/10.1007/978-3-540-30220-9_16.

30. Fleri, W., Paul, S., Dhanda, S. K., Mahajan, S., Xu, X.,

Peters, B., and Sette, A. (2017) The immune epitope

database and analysis resource in epitope discov-

ery and synthetic vaccine design, Front. Immunol.,

8, 278, https://doi.org/10.3389/fimmu.2017.00278.

31. Johnson,M., Zaretskaya,I., Raytselis,Y., Merezhuk,Y.,

McGinnis,S., and Madden, T.L. (2008) NCBI BLAST: a

better web interface, Nucleic Acids Res., 36, W5-W9,

https://doi.org/10.1093/nar/gkn201.

MULTIEPITOPE mRNA VACCINE PROTECTS AGAINST TUBERCULOSIS 769

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

32. Doytchinova, I.A., and Flower, D.R. (2007) VaxiJen: a

server for prediction of protective antigens, tumour

antigens and subunit vaccines, BMC Bioinform., 8, 4,

https://doi.org/10.1186/1471-2105-8-4.

33. Dimitrov, I., Bangov, I., Flower, D. R., and Doytchi-

nova, I. (2014) AllerTOP v. 2 – a server for in silico

prediction of allergens, J. Mol. Model, 20, 2278, https://

doi.org/10.1007/s00894-014-2278-5.

34. Sharma, N., Naorem, L. D., Jain, S., and Raghava,

G. P. S. (2022) ToxinPred2: an improved method for

predicting toxicity of proteins, Brief. Bioinform., 23,

bbac174, https://doi.org/10.1093/bib/bbac174.

35. Gupta,S., Ansari, H.R., Gautam,A., Open Source Drug

Discovery Consortium, and Raghava, G.P. (2013) Iden-

tification of B-cell epitopes in an antigen for induc-

ing specific class of antibodies, Biol. Direct., 8, 27,

https://doi.org/10.1186/1745-6150-8-27.

36. Dhanda, S. K., Vir, P., and Raghava, G. P. (2013) De-

signing of interferon-gamma inducing MHC class-II

binders, Biol. Direct., 8, 30, https://doi.org/10.1186/

1745-6150-8-30.

37. Dhanda, S.K., Gupta,S., Vir,P., and Raghava, G.P. (2013)

Prediction of IL4 inducing peptides, Clin. Dev. Immu-

nol., 2013, 263952, https://doi.org/10.1155/2013/263952.

38. Nagpal, G., Usmani, S. S., Dhanda, S. K., Kaur, H.,

Singh,S., Sharma,M., and Raghava, G.P. (2017) Com-

puter-aided designing of immunosuppressive pep-

tides based on IL-10 inducing potential, Sci. Rep., 7,

42851, https://doi.org/10.1038/srep42851.

39. Bui, H. H., Sidney, J., Dinh, K., Southwood, S.,

Newman, M. J., and Sette, A. (2006) Predicting pop-

ulation coverage of T-cell epitope-based diagnostics

and vaccines, BMC Bioinformatics, 7, 153, https://

doi.org/10.1186/1471-2105-7-153.

40. Kurcinski,M., Jamroz,M., Blaszczyk,M., Kolinski,A.,

and Kmiecik,S. (2015) CABS-dock web server for the

flexible docking of peptides to proteins without prior

knowledge of the binding site, Nucleic Acids Res., 43,

419-424, https://doi.org/10.1093/nar/gkv456.

41. Weng, G., Wang, E., Wang, Z., Liu, H., Zhu, F., Li, D.,

and Hou, T. (2019) HawkDock: a web server to pre-

dict and analyze the protein-protein complex based

on computational docking and MM/GBSA, Nucle-

ic Acids Res., 47, 322-330, https://doi.org/10.1093/

nar/gkz397.

42. Gonzalez-Galarza, F.F., McCabe,A., Santos, E.J.M.D.,

Jones,J., Takeshita,L., Ortega-Rivera, N.D., Cid-Pavon,

G.M.D., Ramsbottom,K., Ghattaoraya,G., Alfirevic,A.,

Middleton,D., and Jones, A.R. (2020) Allele frequency

net database (AFND) 2020 update: gold-standard data

classification, open access genotype data and new

query tools, Nucleic Acids Res., 48, 783-788, https://

doi.org/10.1093/nar/gkz1029.

43. Arai, R., Ueda, H., Kitayama, A., Kamiya, N., and

Nagamune, T. (2001) Design of the linkers which

effectively separate domains of a bifunctional fu-

sion protein, Protein Eng., 8, 529-532, https://doi.org/

10.1093/protein/14.8.529.

44. Meza, B., Ascencio, F., Sierra-Beltrán, A. P., Torres, J.,

and Angulo, C. (2017) A novel design of a multi-an-

tigenic, multistage and multi-epitope vaccine against

Helicobacter pylori: an in silico approach, Infect.

Genet. Evol., 49, 309-317, https://doi.org/10.1016/

j.meegid.2017.02.007.

45. Noderer, W. L., Flockhart, R. J., Bhaduri, A., Diaz

deArce, A.J., Zhang,J., Khavari, P.A., and Wang, C.L.

(2014) Quantitative analysis of mammalian transla-

tion initiation sites by FACS-seq, Mol. Syst. Biol., 10,

748, https://doi.org/10.15252/msb.20145136.

46. Sample, P. J., Wang, B., Reid, D. W., Presnyak, V.,

McFadyen, I. J., Morris, D. R., and Seelig, G. (2019)

Human 5′ UTR design and variant effect predic-

tion from a massively parallel translation assay,

Nat. Biotechnol., 7, 803-809, https://doi.org/10.1038/

s41587-019-0164-5.

47. Reshetnikov, V., Terenin, I., Shepelkova, G., Yere-

meev, V., Kolmykov, S., Nagornykh, M., Kolosova, E.,

Sokolova,T., Zaborova,O., Kukushkin,I., Kazakova,A.,

Kunyk, D., Kirshina, A., Vasileva, O., Seregina, K.,

Pateev, I., Kolpakov, F., and Ivanov, R. (2024) Un-

translated region sequences and the efficacy of mRNA

vaccines against tuberculosis, Int. J. Mol. Sci., 25, 888,

https://doi.org/10.3390/ijms25020888.

48. Magnan, C. N., Zeller, M., Kayala, M. A., Vigil, A.,

Randall, A., Felgner, P. L., and Baldi, P. (2010) High-

throughput prediction of protein antigenicity using

protein microarray data, Bioinformatics, 23, 2936-

2943, https://doi.org/10.1093/bioinformatics/btq551.

49. Gasteiger, E., Gattiker, A., Hoogland, C., Ivanyi, I.,

Appel, R.D., and Bairoch,A. (2003) ExPASy: The pro-

teomics server for in-depth protein knowledge and

analysis, Nucleic Acids Res., 13, 3784-3788, https://

doi.org/10.1093/nar/gkg563.

50. Yang, J., Anishchenko, I., Park, H., Peng, Z.,

Ovchinnikov, S., and Baker, D. (2020) Improved pro-

tein structure prediction using predicted interresidue

orientations, Proc. Natl. Acad. Sci. USA, 3, 1496-1503,

https://doi.org/10.1073/pnas.1914677117.

51. Kringelum, J. V., Nielsen, M., Padkjær, S. B., and

Lund, O. (2013) Structural analysis of B-cell epi-

topes in antibody:protein complexes, Mol. Immunol.,

53, 24-34, https://doi.org/10.1016/j.molimm.2012.06.001.

52. Ponomarenko, J., Bui, H. H., Li, W., Fusseder, N.,

Bourne, P. E., Sette, A., and Peters, B. (2008) ElliPro:

a new structure-based tool for the prediction of

antibody epitopes, BMC Bioinform., 9, 514, https://

doi.org/10.1186/1471-2105-9-514.

53. Solanki, V., and Tiwari, V. (2018) Subtractive pro-

teomics to identify novel drug targets and reverse

vaccinology for the development of chimeric vaccine

against Acinetobacter baumannii, Sci. Rep., 8, 9044,

https://doi.org/10.1038/s41598-018-26689-7.

KAZAKOVA et al.770

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

54. Kozakov, D., Hall, D. R., Xia, B., Porter, K. A.,

Padhorny,D., Yueh,C., Beglov,D., and Vajda,S. (2017)

The ClusPro web server for protein-protein dock-

ing, Nat. Protoc., 2, 255-278, https://doi.org/10.1038/

nprot.2016.169.

55. Xue, L. C., Rodrigues, J. P., Kastritis, P. L., Bonvin,

A. M., and Vangone, A. (2016) PRODIGY: a web serv-

er for predicting the binding affinity of protein-pro-

tein complexes, Bioinformatics, 23, 3676-3678, https://

doi.org/10.1093/bioinformatics/btw514.

56. Laskowski, R. A., Jabłońska, J., Pravda, L., Vařeková,

R. S., and Thornton, J. M. (2018) PDBsum: Structur-

al summaries of PDB entries, Protein Sci., 1, 129-134,

https://doi.org/10.1002/pro.3289.

57. Rapin,N., Lund,O., Bernaschi,M., and Castiglione,F.

(2010) Computational immunology meets bioinformat-

ics: the use of prediction tools for molecular binding

in the simulation of the immune system, PLoS One,

5, e9862, https://doi.org/10.1371/journal.pone.0009862.

58. Zhang, H., Zhang, L., Lin, A., Xu, C., Li, Z., Liu, K.,

Liu, B., Ma, X., Zhao, F., Jiang, H., Chen, C., Shen, H.,

Li,H., Mathews, D.H., Zhang,Y., and Huang,L. (2023)

Algorithm for optimized mRNA design improves sta-

bility and immunogenicity, Nature, 621, 396-403,

https://doi.org/10.1038/s41586-023-06127-z.

59. Diez, M., Medina-Muñoz, S. G., Castellano, L. A.,

da Silva Pescador, G., Wu, Q., and Bazzini, A. A.

(2022) iCodon customizes gene expression based on

the codon composition, Sci. Rep., 12, 12126, https://

doi.org/10.1038/s41598-022-15526-7.

60. Meade, R. K., and Smith, C. M. (2024) Immuno-

logical roads diverged: mapping tuberculosis out-

comes in mice, Trends Microbiol., 33, 15-33, https://

doi.org/10.1016/j.tim.2024.06.007.

61. Eremeev, V. V., Dukhovlinov, I. V., Orlov, A. I.,

Shepelkova, G. S., Fedorova, E. A., Balazovsky,

M. B., and Gergert, V. Ya. (2018) Study of the dura-

tion of the immune response induced by a vaccine

based on recombinant proteins Ag85, TB10 and FliC

[in Russian], Med. Immunol., 20, 271-276, https://

doi.org/10.15789/1563-0625-2018-2-271-276.

62. Abou-Zeid, C., Gares, M. P., Inwald, J., Janssen, R.,

Zhang,Y., Young, D.B., Hetzel,C., Lamb, J.R., Baldwin,

S.L., Orme, I.M., Yeremeev,V., Nikonenko, B.V., and

Apt, A. S. (1997) Induction of a type 1 immune re-

sponse to a recombinant antigen from Mycobacteri-

um tuberculosis expressed in Mycobacterium vaccae,

Infect. Immun., 65, 1856-1862, https://doi.org/10.1128/

iai.65.5.1856-1862.1997.

63. Avdienko, V. G., Babayan, S. S., and Bocharova, I. V.

(2024) Characteristics and application of mono-

clonal antibodies against Mycobacterium kansa-

sii [in Russian], Vestnik CNIIT, 8, 51-63, https://

doi.org/10.57014/2587-6678-2024-8-4-51-63.

64. Kou, Y., Xu, Y., Zhao, Z., Liu, J., Wu, Y., You, Q.,

Wang, L., Gao, F., Cai, L., and Jiang, C. (2017) Tissue

plasminogen activator (tPA) signal sequence enhanc-

es immunogenicity of MVA-based vaccine against tu-

berculosis, Immunol. Lett., 190, 51-57, https://doi.org/

10.1016/j.imlet.2017.07.007.

65. Kreiter, S., Selmi, A., Diken, M., Sebastian, M.,

Osterloh, P., Schild, H., Huber, C., Türeci, O., and

Sahin, U. (2008) Increased antigen presentation effi-

ciency by coupling antigens to MHC class I traffick-

ing signals, J. Immunol., 1, 309-318, https://doi.org/

10.4049/jimmunol.180.1.309.

66. Yang, H., Chen, H., Liu, Z., Ma, H., Qin, L., Jin, R.,

Zheng,R., Feng,Y., Cui,Z., Wang,J., Liu,J., and Hu,Z.

(2013) A novel B-cell epitope identified within My-

cobacterium tuberculosis CFP10/ESAT-6 protein,

PLoS One, 8, e52848, https://doi.org/10.1371/journal.

pone.0052848.

67. Ganguly, N., Siddiqui, I., and Sharma, P. (2008) Role

of M. tuberculosis RD-1 region encoded secreto-

ry proteins in protective response and virulence,

Tuberculosis, 88, 510-517, https://doi.org/10.1016/

j.tube.2008.05.002.

68. Yuan,Y., Crane, D.D., and Barry, C.E. (1996) Station-

ary phase-associated protein expression in Mycobac-

terium tuberculosis: function of the mycobacterial al-

pha-crystallin homolog, J. Bacteriol., 178, 4484-4492,

https://doi.org/10.1128/jb.178.15.4484-4492.1996.

69. Cunningham, A. F., and Spreadbury, C. L. (1998)

Mycobacterial stationary phase induced by low

oxygen tension: cell wall thickening and localiza-

tion of the 16-kilodalton alpha-crystallin homolog,

J. Bacteriol., 180, 801-808, https://doi.org/10.1128/

JB.180.4.801-808.1998.

70. Becker, K., and Sander, P. (2016) Mycobacterium tu-

berculosis lipoproteins in virulence and immunity –

fighting with a double-edged sword, FEBS Lett., 21,

3800-3819, https://doi.org/10.1002/1873-3468.12273.

71. Ahmed,A., Das,A., and Mukhopadhyay,S. (2015) Im-

munoregulatory functions and expression patterns of

PE/PPE family members: roles in pathogenicity and

impact on anti-tuberculosis vaccine and drug de-

sign, IUBMB Life, 67, 414-427, https://doi.org/10.1002/

iub.1387.

72. Coler, R.N., Campos-Neto,A., Ovendale,P., Day, F.H.,

Fling, S.P., Zhu,L., Serbina,N., Flynn, J.L., Reed, S.G.,

and Alderson, M.R. (2001) Vaccination with the T cell

antigen Mtb 8.4 protects against challenge with Myco-

bacterium tuberculosis, J. Immunol., 166, 6227-6235,

https://doi.org/10.4049/jimmunol.166.10.6227.

73. Coppola,M., van den Eeden, S.J., Wilson,L., Franken,

K. L., Ottenhoff, T. H., and Geluk, A. (2015) Synthetic

long peptide derived from Mycobacterium tuberculo-

sis latency antigen Rv1733c protects against tubercu-

losis, Clin. Vaccine Immunol., 22, 1060-1069, https://

doi.org/10.1128/CVI.00271-15.

74. Fan,Q., Lu,M., Xia, Z.Y., and Bao,L. (2013) Mycobac-

terium tuberculosis MPT64 stimulates the activation

MULTIEPITOPE mRNA VACCINE PROTECTS AGAINST TUBERCULOSIS 771

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

of murine macrophage modulated by IFN-γ, Eur. Rev.

Med. Pharmacol. Sci., 24, 3296-3305.

75. Lin, M. Y., Geluk, A., Smith, S. G., Stewart, A. L.,

Friggen, A. H., Franken, K. L., Verduyn, M. J., van

Meijgaarden, K. E., Voskuil, M. I., Dockrell, H. M.,

Huygen, K., Ottenhoff, T. H., and Klein, M. R. (2007)

Lack of immune responses to Mycobacterium tuber-

culosis DosR regulon proteins following Mycobacteri-

um bovis BCG vaccination, Infect. Immun., 75, 3523-

3530, https://doi.org/10.1128/IAI.01999-06.

76. Sharpe, M. L., Gao, C., Kendall, S. L., Baker, E. N.,

and Lott, J.S. (2008) The structure and unusual pro-

tein chemistry of hypoxic response protein 1, a la-

tency antigen and highly expressed member of the

DosR regulon in Mycobacterium tuberculosis, J. Mol.

Biol., 383, 822-836, https://doi.org/10.1016/j.jmb.

2008.07.001.

77. Yao, L., Hu, Q., Chen, S., Zhou, T., Yu, X., Ma, H.,

Ghonaim,A., Wu,H., Sun,Q., Fan,S., and He,Q. (2021)

Recombinant pseudorabies virus with TK/gE Gene

deletion and Flt3L co-expression enhances the innate

and adaptive immune response via activating den-

dritic cells, Viruses, 13, 691, https://doi.org/10.3390/

v13040691.

78. Boopathy, A.V., Nekkalapudi,A., Sung,J., Schulha,S.,

Jin, D., Sharma, B., Ng, S., Lu, S., Wimmer, R.,

Suthram, S., Ahmadi-Erber, S., Lauterbach, H.,

Orlinger, K. K., Hung, M., Carr, B., Callebaut, C.,

Geleziunas,R., Kuhne,M., Schmidt,S., and Falkard,B.

(2024) Flt3 agonist enhances immunogenicity of are-

navirus vector-based simian immunodeficiency virus

vaccine in macaques, J. Virol., 7, e0029424, https://

doi.org/10.1128/jvi.00294-24.

79. Zhu,J., Ke,Y., Liu,Q., Yang,J., Liu,F., Xu,R., Zhou,H.,

Chen, A., Xiao, J., Meng, F., Yu, L., Li, R., Wei, J., and

Liu, B. (2022) Engineered Lactococcus lactis secret-

ing Flt3L and OX40 ligand for in situ vaccination-

based cancer immunotherapy, Nat. Commun., 13,

7466, https://doi.org/10.1038/s41467-022-35130-7.

80. Maraskovsky, E., Brasel, K., Teepe, M., Roux, E. R.,

Lyman, S.D., Shortman,K., and McKenna, H.J. (1996)

Dramatic increase in the numbers of functionally

mature dendritic cells in Flt3 ligand-treated mice:

multiple dendritic cell subpopulations identified,

J. Exp. Med., 184, 1953-1962, https://doi.org/10.1084/

jem.184.5.1953.

81. Brasel,K., De Smedt,T., Smith, J.L., and Maliszewski,

C. R. (2000) Generation of murine dendritic cells

from flt3-ligand-supplemented bone marrow cul-

tures, Blood, 96, 3029-3039, https://doi.org/10.1182/

blood.V96.9.3029.h8003029_3029_3039.

82. Pulendran,B., Banchereau,J., Burkeholder,S., Kraus,E.,

Guinet, E., Chalouni, C., Caron, D., Maliszewski, C.,

Davoust, J., Fay, J., and Palucka, K. (2000) Flt3-li-

gand and granulocyte colony-stimulating factor mo-

bilize distinct human dendritic cell subsets in vivo,

J. Immunol., 165, 566-572, https://doi.org/10.4049/

jimmunol.165.1.566.

83. Triccas, J. A., Shklovskaya, E., Spratt, J., Ryan, A. A.,

Palendira,U., Fazekas de StGroth,B., and Britton, W.J.

(2007) Effects of DNA- and Mycobacterium bovis BCG-

based delivery of the Flt3 ligand on protective immu-

nity to Mycobacterium tuberculosis, Infect. Immun.,

75, 5368-5375, https://doi.org/10.1128/IAI.00322-07.

84. Ahn, S.S., Jeon, B.Y., Kim, K.S., Kwack, J.Y., Lee, E.G.,

Park, K.S., Sung, Y.C., and Cho, S.N. (2012) Mtb32 is

a promising tuberculosis antigen for DNA vaccination

in pre- and post-exposure mouse models, Gene Ther.,

19, 570-575, https://doi.org/10.1038/gt.2011.140.

85. Xu,J., Xu,W., Chen,X., Zhao,D., and Wang,Y. (2008)

Recombinant DNA vaccine of the early secreted an-

tigen ESAT-6 by Mycobacterium tuberculosis and

Flt3 ligand enhanced the cell-mediated immunity in

mice, Vaccine, 26, 4519-4525, https://doi.org/10.1016/

j.vaccine.2008.06.044.

86. Branger, J., Leemans, J. C., Florquin, S., Weijer, S.,

Speelman, P., and Van Der Poll, T. (2004) Toll-like

receptor 4 plays a protective role in pulmonary tu-

berculosis in mice, Int. Immunol., 3, 509-516, https://

doi.org/10.1093/intimm/dxh052.

87. Xin, Q., Niu, H., Li, Z., Zhang, G., Hu, L., Wang, B.,

Li,J., Yu, H., Liu,W., Wang, Y., Da,Z., Li,R., Xian,Q.,

Wang,Y., Zhang,Y., Jing,T., Ma,X., and Zhu,B. (2013)

Subunit vaccine consisting of multi-stage antigens

has high protective efficacy against Mycobacterium

tuberculosis infection in mice, PLoS One, 8, e72745,

https://doi.org/10.1371/journal.pone.0072745.

88. Oksanen, K. E., Myllymäki, H., Ahava, M. J., Mäkin-

en, L., Parikka, M., and Rämet, M. (2016) DNA vac-

cination boosts Bacillus Calmette-Guérin protection

against mycobacterial infection in zebrafish, Dev.

Comp. Immunol., 54, 89-96, https://doi.org/10.1016/j.

dci.2015.09.001.

89. Vakili, B., Nezafat,N., Zare,B., Erfani, N., Akbari,M.,

Ghasemi, Y., Rahbar, M. R., and Hatam, G. R. (2020)

A new multi-epitope peptide vaccine induces im-

mune responses and protection against Leishmania

infantum in BALB/c mice, Med. Microbiol. Immunol.,

209, 69-79, https://doi.org/10.1007/s00430-019-00640-7.

90. Kozlova, A., Pateev, I., Shepelkova, G., Vasileva, O.,

Zakharova, N., Yeremeev, V., Ivanov, R., and

Reshetnikov,V. (2024) A cap-optimized mRNA encod-

ing multiepitope antigen ESAT6 induces robust cellu-

lar and humoral immune responses against Mycobac-

terium tuberculosis, Vaccines (Basel), 12, 1267, https://

doi.org/10.3390/vaccines12111267.

91. Shepelkova, G. S., Reshetnikov, V. V., Avdienko, V. G.,

Sheverev, D.V., Yeremeev, V.V., and Ivanov, R.A. (2023)

Impact of untranslated mRNA sequences on immuno-

genicity of mRNA vaccines against M. tuberculosis in

mice, Bull. RSMU, 6, 27-33, https://doi.org/10.24075/

brsmu.2023.054.

KAZAKOVA et al.772

BIOCHEMISTRY (Moscow) Vol. 90 No. 6 2025

92. Wu, M., Li, M., Yue, Y., and Xu, W. (2016) DNA vac-

cine with discontinuous T-cell epitope insertions

into HSP65 scaffold as a potential means to improve

immunogenicity of multi-epitope Mycobacterium tu-

berculosis vaccine, Microbiol. Immunol., 60, 634-645,

https://doi.org/10.1111/1348-0421.12410.

93. Cai,H., Tian,X., Hu,X., Pan,Y., Li,G., Zhuang,Y., and

Zhu, Y. (2003) Immunogenicity and protective effica-

cy study using combination of four tuberculosis DNA

vaccines, Sci. China C. Life Sci., 46, 495-502, https://

doi.org/10.1360/02yc0121.

Publisher’s Note. Pleiades Publishing remains

neutral with regard to jurisdictional claims in published

maps and institutional affiliations. AI tools may have

been used in the translation or editing of this article.