*3.2. Design of Artificial Poly-CTL-Epitope Antigen of Ebola Virus*

For the purposes of designing poly-CTL-epitope antigen (EV.CTL), we used Immune Epitope Database (http://iedb.org) [24] to select known T-cell epitopes and peptide fragments of antigens of different Ebola virus strains with an experimentally verified capacity to bind to different allomorphs of MHC molecules. In total, at the time of antigen designing (2016) the database contained information on 1134 unique peptides from 65 antigens of 16 Ebola virus strains verified for their capacity to bind to 60 allomorphs of MHC class I molecules (56 Human Leukocyte Antigen (HLA) allelic variants). To analyze conservation of peptides, we used 14,556 amino acid sequences from NCBI ProteinBank (ncbi.nlm.nih.gov/protein) belonging to different Ebola viruses (Zaire ebolavirus, Sudan ebolavirus, Bundibugyo ebolavirus, Tai Forrest ebolavirus, and Reston ebolavirus). We considered peptides with experimentally verified cytotoxic activity. Furthermore, when designing target immunogens, we selected those with sufficiently high binding affinity to different HLA class I molecule variants (pIC50 > 6.3).

After that, we selected peptides identified at least in 1000 known viral sequences and interacting with at least two allelic HLA molecule variants. In total, we selected 44 peptides which cumulatively were restricted by 34 allelic HLA class I molecule variants including the most globally widespread ones (Table 1). It is known that optimally selected epitopes restricted by ten different HLA class I alleles cover virtually the entire population of any geographic region [46,47].

Based on the selected T-cell epitopes, we designed EV.CTL poly-CTL-epitope antigen using TEpredict/PolyCTLDesigner software we developed earlier [25,26] that we regard as a universal platform for rational design of polyepitope immunogens—candidate DNA vaccines to induce T-cell immunity to infectious as well as oncological diseases. PolyCTLDesigner enables us to select a minimal set of epitopes with known or predicted specificity to different allelic variants of MHC class I molecules covering a selected repertoire of HLA alleles with a preset degree of redundancy. After that, PolyCTLDesigner predicts binding affinity to TAP for the selected set of known or predicted epitopes using a model developed by Peters et al. [48] and when required adds TAP-specific amino acid residues (no more than three) to epitope N-terminus to optimize binding.


**Table 1.** Predicted CD8+ cytotoxic T-lymphocytes (CTL)-epitopes in the sequences of Ebola virus proteins (antigens).

At the next step, PolyCTLDesigner analyzes all possible matchings of the selected peptides and detects the optimal spacer sequence for each pair providing an appropriate cleavage of epitopes with a release of proximal peptide C-terminus. To predict proteasomal and/or immunoproteasomal cleavage, PolyCTLDesigner uses models developed by Toes, et al. [49].

When analyzing matchings of epitopes, PolyCTLDesigner forms a direct graph where nodes denote epitopes and ribs correspond to acceptable matchings. Each rib has a relevant weight vector characterized by effective proteasomal cleavage, spacer length, and a number of predicted non-target epitopes at the joint. At the final stage, the software designs the optimal resultant of polyepitope immunogen sequence determined as a full simple way in the formed graph with the least length (weight).

In this study, we used PolyCTLDesigner to predict binding affinity of the selected peptides (Table 1) to TAP; when required software added alanine residue to peptides N-terminus to enhance interaction efficiency. Poly-CTL-epitope fragment of EV.CTL was designed using a degenerated spacer motif [ARSP][DLIT][LGA][VKA] with optimization of proteasomal cleavage and 10% exactness of proteasomal filter.

To test the immunogenicity of the designed vaccine construct in mice using ELISpot and ICS, we selected seven additional peptides with proven ability to induce cytotoxic response of T-lymphocytes in BALB/c mice: *KFINKLDALH, NYNGLLSSI, PGPAKFSLL, YFTFDLTALK, EYLFEVDNL, LFLRATTEL, and LYDRLASTV*. Based on the selected peptides, mouse polyepitope fragment included at C-terminus of the polyepitope construct was designed with PolyCTLDesigner. To verify synthesis of the designed antigen in transfected cells, we included C-terminal marker epitope EPFRDYVDRFYKTLR of p24 HIV-1 protein recognized by monoclonal antibodies 29F2 in the final construct.

Designed amino acid sequence appears as follows (mouse epitopes are italicized):

**MMVIFRLMR**—**ADLS**—GHMMVIFRL—**KK**—VQLPQYFTF—**ADLS**—KQIPIWLPL—**RK**—EYAPFA RLL—RVPTVFHKK—FIYFGKKQY—**R**—VLYHRYNLV—**ADL**—YQGDYKLFL—AFPRCRYVHK—ATP VMSRFAA—AFAEGVVAFL—KVYWAGIEF—**R**—TVAPPAPVY—TLASIGTAF—**R**—TTIGEWAFW— **RK**—LANETTQAL—FLLQLNETI—**R**—FVHSGFIYF—**K**—IISDLSIFI—**R**—NFFHASLAY—**RR**—LAN PTADDF—**K**—ILMNFHQKK—**ADLS**—FTPQFLLQL—YSGNIVHRY—**ADLA**—RTSFFLWVI—RTF SILNRK—**RK**—LSDLCNFLV—**ADLV**—HMMVIFRLM—**ADLK**—IMYDHLPGF—ALPQYFTFDL—YL EGHGFRF—**R**—FLSFASLFL—**R**—TRSFTTHFL—RLMRTNFLI—**ADG**—FRLMRTNFL—**R**—GQFLSFA SL—**R**—SFASLFLPK—RLASTVIYR—ARLSSPIVL—AHPLARTAKV—QFLSFASLF—**R**—GYLEGTRTL —**R**—FRYEFTAPF—**KK**—*YFTFDLTALK—EYLFEVDNL—R—PGPAKFSLL—RK—LFLRATTEL—RK— NYNGLLSSI—R—LYDRLASTV—R—KFINKLDALH*—SGSG—**EPFRDYVDRFYKTLR**

The length of the designed polyepitope EV.CTL is 547 amino acids; a share of spacer sequences is 12.76%. To target polyepitope immunogen into proteasome, we added ubiquitin sequence to N-terminus of the final poly-CTL-epitope construct.
