Identification of Regulators for Antigen-Specific CD8⁺ T Cells in African Swine Fever Virus-Restored Pigs

1. Introduction

African swine fever virus (ASFV) was first identified in Kenya, Africa, in 1921, and it is currently prevalent in several African countries or regions [1]. The disease spread to China in 2018 [2]. By 2024, the Ministry of Agriculture of China reported over 190 outbreaks of ASFV, leading to the culling of more than 1.2 million pigs. This has resulted in a significant decline in the pig inventory and a notable increase in pork prices. ASFV poses serious challenges to the development of China's pig industry and stability of pork supply.

At present, the development of ASFV vaccines remains a major challenge. The immunogenicity of inactivated vaccines, subunit and gene deletion vaccines is not fully understood. These vaccines may fail to provide comprehensive protective after vaccination [3,4,5]. Research has demonstrated that key structural proteins of ASFV, such as the p72 protein on the capsid surface, and the CD2v protein on the envelope, are important for eliciting a humoral immune response [6]. Both CD2v and p72 are involved in the adsorption of ASFV to erythrocytes and macrophages. These proteins play a role in inducing neutralizing antibodies, thereby preventing ASFV from binding to target cells. These proteins are commonly used as key markers for ASFV serological diagnosis and vaccine development [7,8,9]. However, substantial individual variations in immune response to ASFV inactivated, subunit, and gene deletion vaccines have been observed. Sandra Blome et al. found that approximately 60% of the pigs vaccinated with the inactivated ASFV vaccine exhibited acute and fatal symptoms after ASFV challenge [10,11,12]. Similarly, Jankovic et al. demonstrated that the subunit vaccine provided protection in up to 60% of the pigs [13,14,15,16]. The protective effect of the gene deletion vaccine is uncertain at present [5], and the immunogenicity of ASFV is reduced to 66% due to the deletion of some virulence genes [17]. Huaji Qiu, Chungen Pan, and others reviewed of the latest research progress and concluded that a significant gap remains between the development of the ASFV vaccine and its practical application. Ogweng Bisimwa et al. further highlighted that differences in immune responses to ASFV among individuals might be related to genetic variations within the pigs [18,19].

A pivotal aspect of developing novel control strategies lies in understanding the intricate virus-host interplay. ASFV, with its relatively compact genome, is heavily reliant on hijacking the host's cellular machinery to facilitate its own replication. Therefore, identifying key host transcriptional factors, such as ELK4 and MECP2, that are cooped by the virus is particularly valuable. These regulators represent critical cellular dependencies that ASFV must exploit to successfully establish infection. Targeting these host factors therapeutically could potentially restrict viral replication without directly targeting the virus itself, thereby circumventing the challenge of viral mutation and offering a promising avenue for the development of broad-spectrum antiviral interventions.

Identifying the TFs and molecular markers associated with the ASFV cellular immune response can be achieved by screening individuals with strong anti-ASFV capabilities, collecting samples with varying levels of cellular immune responses, and integrating pig genome sequencing with bioinformatics analysis. This approach is critical for advancing ASFV research, vaccine development, and the creation of ASFV-resistant pig breeds. In animals, cellular and humoral immunity play vital roles in resistance to viral infection [20]. Antigen presentation bridges cellular and humoral immune responses within adaptive immunity [21]. The Activation of T and B cells is dependent on antigen presentation in ASFV-infected pigs [2,22]. Antigen-presenting cells (APCs), which include dendritic cells, macrophages, and B cells, capture foreign antigens through phagocytosis, endocytosis, or receptor-mediated internalization. These cells then process the antigens into short peptide fragments using lysosomes or proteasomes. Subsequently, these antigenic peptide fragments bind to major histocompatibility complex (MHC) class I or II molecules, forming antigenic peptide-MHC complexes. These complexes are then presented on the cell surface for recognition [23]. APC migrates to the secondary lymphoid organs where they activate CD8⁺ T cells or CD4⁺ Th cells through the specific recognition of MHC-antigenic peptide-T cell receptor (TCR) ternary complex. This, in turn, results in the further activation of both cellular and humoral immunity [21]. In cellular immunity activation, the efficiency and accuracy of antigen presentation directly determine the intensity and quality of subsequent CD8⁺ T and CD4⁺ Th cell responses [24].

In this study, we first identified the potential antigenic epitopes of the ASFV p72 protein by MHC tetramers staining and antigen-specific CD8⁺ T cell flow cytometry. We then compared the transcriptomes of submandibular lymph nodes with different antigen-specific CD8⁺ T cells (ACD8⁺) levels by RNA-seq. A weighted Kendall's Tau rank correlation test of extent of differential expression and potential of TF binding as employed to identify TFs that regulate the ACD8⁺. ELK4 was identified as a key candidate gene in this analysis. Finally, we analyzed the role of ELK4 in regulating CD8⁺ T cell abundance using SNP-scanning, MHC-I tetramer-specific T cell flow cytometry, and association analysis.

2. Materials and Methods

2.1. Experimental Materials

The peripheral blood samples of the Landrace × Large White × Duroc pigs (4 weeks of age) in this study were obtained from the ASFV-infected pig farm in Jingzhou, Hubei Province, where ASFV infection was first detected in August 2019. Following the implementation of Real-time PCR detection and precision depopulation measures, the pig farm successfully curtailed the ASFV outbreak and achieved re-population in September 2022. In this experiment, the 112 animals were negative for ASFV based on Real-time PCR assay and exhibited no symptoms of infection. However, 57 (51%) animals were positive for ASFV p72 antibody. Peripheral blood samples were collected from the animals, diluted with an equal volume of RPMI, and then, PBMCs were isolated using Ficoll-Hypaque density gradient centrifugation and cryopreserved.

The collection of samples in this study was conducted in accordance with the Regulations for the Administration of Experimental Animals issued by the Science and Technology Commission of China (NO. 2006-398). All procedures involving animals were approved by the Animal Ethics Commission of Yangtze University (Jingzhou, Hubei, China). The piglets were first sedated via intramuscular injection of a combination of tiletamine-zolazepam (5 mg/kg) and xylazine (2 mg/kg). Following the induction of loss of consciousness, complete euthanasia was achieved by the administration of an intravenous overdose of sodium pentobarbital (100 mg/kg). Death was confirmed by the absence of a corneal reflex and the cessation of both respiration and heartbeat. Immediately, submandibular lymph node samples were meticulously collected using sterile surgical instruments. Each sample was divided into aliquots, snap-frozen in liquid nitrogen, and stored at -80°C for subsequent molecular analysis.

The MHC tetramers were generated following the method outlined by Pedersen [25,26,27]. First, transfer 30 µL of fixed peptide monomer into a 1.5 mL Eppendorf tube. Subsequently, 3.3 µL of FITC-conjugated streptavidin (Cell Signaling, 34524) was added, and the mixed vigorously. Incubate the mixture on ice in the dark for 30 minutes. Prepare the blocking solution by adding 1.6 µL of 50 mM D-Biotin and 6 µL of 10% (w/v) NaN3 to 192.4 µL of PBS and mixing by vertexing. After the incubation, add 2.4 µL of blocking solution to the mixture to stop the reaction. Incubate the tubes at 4°C overnight.

2.2. Testing the abundance of CD8⁺ T cells specific for ASFV p72 protein peptide

The flow cytometry assay was performed in accordance with the following protocol. First, the assembled tetramers were centrifugated in tubes at 2500 x g for 5 minutes at 4°C. Subsequently, the solution was kept on ice in the dark. Add 2 x 106 cells to a 96-well U-bottom plate, then adjust the volume of the solution to 200 µL with Cell Staining Buffer. Add 2 µL of the previously prepared tetramers. Mix and incubate on ice in the dark for 30 minutes. Prepare the PE-conjugated CD8⁺ surface marker antibody (ThermoFisher, 76-2-11) and incubate for 30 minutes on ice in the dark. Wash the cells with staining buffer twice. Resuspend cells with staining buffer. The acquisition of samples on a flow cytometer, with the appropriate settings within two hours. Live cells were gated based on FSC-A and SSC-A characteristics to exclude debris. Live cells were subsequently gated using PE-conjugated CD8⁺ versus FITC-conjugated streptavidin signal to identify ACD8⁺ cells.

2.3. High Throughput Sequencing

From the 112 animals assayed for the ACD8⁺, three animals for each group were selected for antigen-specific CD8⁺ T cell abundance at 5% (low), 50% (medium), and 95% (high) of all the samples (six females and three castrated males). Tissue samples were mixed with 1 mL of Trizol reagent and snap frozen in liquid nitrogen. Subsequently, the samples were transferred to the designated DNA facility with dry ice for RNA-seq analysis. The samples for RIN scores > 0.9 were selected for the following analysis. The sequencing library preparation was carried out following the protocol stipulated by the Illumina TruSeq RNA sample preparation kit (Illumina Inc., USA). Approximately 10 μg of total RNA from each sample was utilized for library construction and RNA sequencing. The Sequencing was performed on an Illumina HiSeq 2500 sequencer (Illumina Inc., USA) using a single-read sequencing method (50bp). Subsequently, a data filtration process was implemented to obtain high-quality, clean reads, and to remove low-quality reads present in the raw reads, and the quality of the raw sequencing reads was assessed as recommended by the manufacturer. The Hisat2 [28] software was employed to map clean reads to the reference genome (Sus Scrofa 11.1), which was extracted from the NCBI genome database [29]. The calculated reads count per gene was estimated by Htseq-count [30] and used to compare the difference in gene expression among samples. The library was normalized using median of ratios method and the DEseq2 R package [31] was utilized for the identification of differentially expressed genes with an FDR (false discovery rate) of ≤ 0.05 and an FC (fold change) of ≥ 1.5 or ≤ 0.67. The differentially expressed genes were annotated using DAVID annotation tool. All data relevant to this study have been deposited in the NCBI GEO database under accession number GSEXXXXX (Data temporarily available at https://t3.znas.cn/g6ujwMnbv63, will be deposited in GEO after NCBI services are recovered.).

2.4. Identification of key transcription factors

The identification of the key TFs that regulate the differentially expressed genes was achieved through a multifaceted approach. Initially, the gene sets, comprising genes targeted by a specific TF, and the gene list, containing the differentially expressed genes, were constructed. Subsequently, correlation analysis was implemented on these lists to elucidate the regulatory mechanisms. The TFBS data utilized in this study were obtained from the prediction results of GRIT-2.0. GRIT-2.0 utilized a mixed Student's t-test approach to predict TFBS, incorporating both the binding score of binding sites, denoted as Jindex (Equation 1), and the conservation characteristics of binding sites across species [32]

$$Jindex={Max}_s\left\{\mathit{ln}\sqrt[w]{{\prod }_{k=1}^w{\displaystyle \frac{q\left(k, L_k\right)}{p\left(L_k\right)}}}\right\}; 1\le s\le l-w+1$$

(1)

As demonstrated by Huang et al. [33] the Jindex quantifies the maximum of repeated averaging of log likelihood ratios (LLRs), which are an indicative factor of the potential presence of a motif at a specific location in a sequence. The correlation analysis was executed using the FLAVER software package. The present study employed the strategy developed by Yao [34], which was achieved by testing the significance of the correlation between the order of genes in the gene set and the corresponding order in the gene list. The analysis accentuates genes with larger weights while diminishing the emphasis on genes with smaller weights. As Shieh's research indicates, the weighted Kendall's τ assumes the form of Equation (2). The limiting distribution (LD) can be derived from Equation (3). As the value of n approaches infinity, the LD value approaches N(0, 1), and thus, the value of the p-value can be estimated.

$${\tau }_w=2/\left[{\left({\sum }_i^n{v}_i\right)}^2-{\sum }_i^n{v}_i^2\right]·{\sum }_{i>j}^n{v}_i{v}_{j}sgn\left(i-j\right)sgn\left(R_i-R_j\right)$$

(2)

If X <, =, or > 0, then sgn (X) = -1, 0, or 1. The vi denotes the weighting function, which is bounded by [1, n] and ranges from 0 to 1.

$$LD=\sqrt{n}{\tau }_w{\displaystyle \frac{3\underset{n\to \infty }{\mathit{lim}}{n}^{-1}\sum _x^n{v}_x}{2\sqrt{\underset{n\to \infty }{\mathit{lim}}{n}^{-1}\sum _x^n{v}_x^2}}}$$

(3)

The weighting function vx employed in the test is delineated in equation (4), where x ranges from 1 to n. The vxs and vxl are the gene weights of genes in the gene set and the gene list, respectively.

$$v_x={\left[(1-{\displaystyle \frac{v_{xs}}{\mathit{max}\left\{v_s\right\}}})·(1-{\displaystyle \frac{v_{xl}}{\mathit{max}\left\{v_l\right\}}})\right]}^{0.5}$$

(4)

Figure 1. A schematic summary of the experimental workflow.

Figure 1. A schematic summary of the experimental workflow.

2.5. Genotyping of single-nucleotide polymorphisms (SNPs) based on Sanger sequencing

The primers were designed according to the flanking sequence of the SNP sites, covering approximately 900 base pairs (bp) of the upstream region of the transcription start site (FW: 5'-AGTGCTTTTCAGATATTTCGTGT-3', RV: 5'-AGACATTCAGTCGTAGCTCCA-3'). Peripheral blood genomic DNA was extracted and utilized as a template, and the target DNA fragment was amplified by PCR (40 cycles of 95°C for 15s, 56°C for 20s, and 72°C for 15s). The amplified product is then purified to remove any residual primer and nucleotide and sent to the DNA facility. The Sanger sequencing was performed using the BigDye Terminator v3.1 kit on an ABI 3730 sequencer. The sequencing results were analyzed using DNASTAR Lasergene software. The SNP association analysis was performed using the ANOVA method in R software package. Since the 112 pigs were from a field population and the population stratification information is unknown thus confounding is not considered. We controlled the false discovery rate (FDR) using the Benjamini-Hochberg procedure for all statistical analysis in this study. A schematic summary of the experimental workflow was provided as Fig. 1. Animal information, SNP genotypes, and ACD8⁺ for each animal were provided in Supplemental Data 1.

3. Results

3.1. Specificity of the SLA-Peptide Tetramer Complex

The FITC-labeled tetrameric complexes of SLA-1 (HET-1) folded with the SQIEETHLV peptide from the ASFV p72 protein, and FITC-labeled tetrameric complexes of SLA-1 (HET-2) folded with the FVTPEIHNL peptide from the ASFV p72 protein, were prepared. The flow cytometry analysis obtained over 10⁶ events for each sample. The results demonstrated that HET-1-stained specific T cells from ASFV p72 antibody-positive individuals (Figure 2B) but not from other ASFV p72 antibody-negative individuals (Figure 2A). This finding thus confirmed the specificity of tetramer staining. Similarly testing the specificity of HET-2 found that the tetramer could stain CD8⁺ T cells from p72 antibody-positive individuals, but not CD8⁺ T cells from other p72 antibody-negative individuals (Figure 1C and D). Given that HET-1 exhibits a marginally higher degree of sensitivity compared to HET-2, the HET-1 was utilized in the subsequent experimental phase to investigate the ACD8⁺.

3.2. Lymph node transcriptome analysis of individuals with varying levels of antigen specific CD8⁺ T cells

A total of 112 animals were assayed for the ACD8⁺, the results of which are shown in Fig. 3. From this data set, three animals for each group were selected for antigen-specific CD8⁺ T cell abundance at 5% (low), 50% (medium), and 95% (high) of all the samples. No criteria were set for including and excluding animals, and confounders were not controlled. Subsequently, RNA-seq sequencing of submandibular lymph node tissue was performed in three animals per group. The RNA-seq sequencing results demonstrated an average of 32 million sequences per sample. A total of 22,301 transcripts has been identified, with an average 21 million reads per sample were obtained. The analysis revealed that 2,049 transcripts exhibited significant differential expression (FDR < 1E-8) between the groups with high and low ACD8⁺ levels. The top 30 differentially expressed genes are listed in Table 1. A total of 1,017 transcripts were found to be significantly differentially expressed (FDR < 1E-8) between the medium and low ACD8⁺ level groups. The top 30 differentially expressed genes are listed in Table 2. A total of 196 transcripts were significantly differentially expressed (FDR < 0.05) between the high and medium levels of antigen-specific CD8⁺ T cells. For a comprehensive list of the differential genes compared in each group, refer to Supplemental Data 2. Four genes, ELK4, ETS1, MECP2, and ZBTB33, were selected for Real-time PCR validation, and the Real-time PCR results are consistent with the RNA-seq results (Fig. 4, primer sequences shown in Supplemental Document 1).

ASFV has been shown to cause host transcriptome remodeling by regulating inflammatory response, interferon response, apoptosis, autophagy, antigen presentation, and adaptive immunity [35,36]. Comparing the animals exhibited high and low levels of ACD8⁺ revealed the presence of 61 inflammation-related genes, including IL33, FASN, TGFB1, DAGLA, and PSMA1, which showed significant upregulation. This upregulation was associated with the characteristic symptoms of ASF, including high fever and systemic inflammation, which are the primary causes of mortality. A significant number of proteins encoded by ASFV, including I329L, A528R, and EP402R (CD2v), can effectively inhibit the production and signal transduction of interferon within the host organism, thus enabling evasion of the innate immune response [37,38,39]. In this study, 14 interferon-production-related genes, such as ZFPM1, BCL3, SCRIB, RARA, and RBX1, exhibited significant differential expressions between high and low ACD8⁺ cells. In certain instances, the virus has been observed to inhibit early apoptosis, thereby promoting its replication. Conversely, in other instances, the virus has been found to trigger late apoptosis, thus facilitating viral release [40,41,42]. In this study, we observed four apoptosis-related genes (BOK, TAOK1, CASP8, CASP3) and 51 autophagy-related genes (SEC16A, USP20, RAB1A, PLEKHF1, ATG2A) that exhibited significant differential expressions between high and low levels of ACD8⁺ cells. Most importantly, ASFV inhibits the expression of major histocompatibility complex (MHC) class Ⅰ and Ⅱ molecules. This results in a direct impairment of T-cell recognition and activation, consequently leading to a failure of the adaptive immune response. This phenomenon, among others, has led to the observation that some pigs that have recovered from infection often exhibit weak immunity, which complicates the development of vaccines [43,44,45]. In the present study, seven MHC I-associated genes and six MHC II genes (i.e., MARCHF1, CIITA, IDE, CTSS, TAF7) were observed to be differentially expressed. In addition, ten adaptive immune response genes (i.e., CTSS, CREG1, JCHAIN, AKIRIN2, and CSK) were identified as differentially expressed. ASFV has been observed to regulate the antigen presentation process in the host by altering the expression pattern of these genes.

Figure 3. The 112 pigs’ population exhibited different abundances of antigen specific CD8⁺ T cells. Bar height is a quantitative metric representing the proportion of antigen-specific CD8⁺ T-positive cells as determined by flow cytometry. The yellow bars represent the samples selected for RNA-seq sequencing, with the selection criteria delineated as follows: animals at 5% (low), 50% (medium), and 95% (high).

Figure 3. The 112 pigs’ population exhibited different abundances of antigen specific CD8⁺ T cells. Bar height is a quantitative metric representing the proportion of antigen-specific CD8⁺ T-positive cells as determined by flow cytometry. The yellow bars represent the samples selected for RNA-seq sequencing, with the selection criteria delineated as follows: animals at 5% (low), 50% (medium), and 95% (high).

Figure 4. Real-time PCR validation of representative differentially expressed genes. Plot A to D shows results for ELK4, ETS1, MECP2, ZBTB33, respectively. The height of the bars shows relative expression levels which is normalized with the house keeping gene GAPDH.

Figure 4. Real-time PCR validation of representative differentially expressed genes. Plot A to D shows results for ELK4, ETS1, MECP2, ZBTB33, respectively. The height of the bars shows relative expression levels which is normalized with the house keeping gene GAPDH.

3.3. Identification of key transcription factors for antigen-specific CD8⁺ T cell's abundance

The FLAVER analysis identified 95 TFs as significant regulators in the correlation test between the differentially expressed gene list identified in the high and low ACD8⁺ cell groups (FDR < 1E-7, top 30 listed in Table 3). A total of 77 TFs were identified as significant regulators for the genes in the differentially expressed gene list between the medium and low ACD8⁺ cell groups (FDR < 1E-7, top 30 listed in Table 4). Among the groups, the most significant TFs included G4204_MECP2, MA0098.3_ETS1, G56805_ZBTB33, G2002_ELK1, MA0076.2_ELK4, MA1564.1_SP9, and G104394_E2f4. A correlation graph for a representative TF, ELK4, was shown on Fig. 5. MECP2 itself was down-regulated 5.19-fold (q-value < 1E-11) between the high and low groups and 3.14-fold (q-value < 1E-5) between the medium and low groups. The transcription levels of ETS1 and E2f4 did not significantly differ between the high/low and medium/low groups. ZBTB33 demonstrated a 9.79-fold increase (q-value < 1E-9) between the high and low groups and a 7.48-fold increase (q-value < 1E-6) between the medium and low groups. The transcription levels of ELK1 and SP9 in submandibular lymph nodes were below 10 copies. ELK4 expression was found to be significantly elevated in both the high- and medium-low groups, with a 10.05-fold increase (P-value < 1E-24) and a 9.25-fold increase (P-value < 1E-18), respectively. These results suggest that ASFV may achieve immune evasion by altering the transcription levels of MECP2, ZBTB33, and ELK4 in the host. Given the absence of a substantial difference in the transcription levels of ETS1 and E2F4 among the high/low and t medium/low groups, it was hypothesized that ASFV might exert its function by modulating their product at post-transcriptional or protein levels. Functional annotation found that 18 of the 44 genes comprising the class Ⅰ MHC mediated antigen processing presentation signaling pathway have ELK4 binding sites in their promoters as predicted by GRIT software package (Table 5). ELK4 may be an important target molecule responsible for the differences in the ACD8⁺ between groups.

Figure 5. The correlation graph for target genes for ELK4 transcription factor. Plot A and B illustrated the genes have or have not ELK4 binding site. The x axis represented the rank values of the extent of deferential expression. Bar height is a quantitative metric representing the potential of binding with ELK4. Bars highlighted with green represented the antigen processing and representation genes.

Figure 5. The correlation graph for target genes for ELK4 transcription factor. Plot A and B illustrated the genes have or have not ELK4 binding site. The x axis represented the rank values of the extent of deferential expression. Bar height is a quantitative metric representing the potential of binding with ELK4. Bars highlighted with green represented the antigen processing and representation genes.

3.4. SNP genotyping and association analysis with the abundance of antigen-specific T cells

DNA sequencing revealed six DNA mutations in the promoter region of ELK4. Although ELK4 is a TF, it does not have enzymatic activity itself but plays a pivotal role in the regulation of genes function in selective protein degradation within cells. This process is critical for various biological functions, including the cell proliferation cycle, stress response, and immune response. The proteasome degradation pathway facilitates this regulatory function of ELK4. Polypeptides formed by the degradation of foreign proteins by ELK4’s targets are utilized for MHC class I antigen presentation, thereby inducing specific cellular immune responses. Furthermore, ELK4’s targets genes have been identified as being closely associated with antigen presentation. The S.-404A>G mutation of the ELK4 gene formulated three distinct genotypes among the 112 pigs examined. The observed genotype frequencies were AA: 0.05, AG: 0.21, and GG: 0.74, which is significantly deviated from the Hardy-Weinberg equilibrium. The frequency of allele A and G was determined to be 0.15 and 0.85 respectively. The ACD8⁺ in the samples of 112 pigs with GG, AG, and AA genotypes was 0.21 ± 0.102, 0.28 ± 0.077, and 0.39 ± 0.075, respectively. This indicated a trend of GG < AG < AA. A statistically significant discrepancy was observed between AA and AG, GG genotype (q-value < 0.05). Three genotypes of the ELK4 gene were formulated in the study of 112 pigs through the S.-668C>T mutation. The observed genotype frequencies were CC: 0.02, CT: 0.3, and TT: 0.68, which is significantly deviated from the Hardy-Weinberg equilibrium. The frequency of allele C and T was 0.17 and 0.83, respectively. The ACD8⁺ in samples of 112 pigs with the CC genotype was 0.41 ± 0.067, the CT genotype was 0.27 ± 0.061, and the TT genotype was 0.19 ± 0.079. The ACD8⁺ exhibited a descending trend, with TT ranking at the highest, CT and CC ranking at the lowest. The CC genotype exhibited a significantly higher ACD8⁺ than that CT and TT genotypes (q-value < 0.05). ACD8⁺ was not significant between genotypes of other mutation sites. Further analysis indicated that the sex factor does not associate with ACD8⁺.

Table 6. Antigen-specific CD8⁺ T Cell Abundance and the Genetic Variation of Single-Nucleotide Polymorphisms (SNP) in Individuals.

SNP ID	Antigen-specific CD8+ Cell Abundance for Genotype			Overlap with Transcription Factor Binding Site
	XX	Xx	xx
S.-291C>T	0.36±0.1a	0.34±0.1a	0.31±0.11a	G7566_ZNF18
S.-404A>G	0.39±0.08a	0.28±0.08b	0.21±0.1b	G15376_Foxa2, MA0036.3_GATA2, MA0037.4_Gata3, MA0482.2_GATA4, MA1104.2_GATA6, MA1970.1_TRPS1
S.-463A>C	0.36±0.09a	0.35±0.1a	0.34±0.09a	G20852_Stat6
S.-604C>T	0.36±0.1a	0.35±0.08a	0.36±0.11a	G4772_NFATC1
S.-668C>T	0.41±0.07a	0.27±0.06b	0.19±0.08b	G10062_NR1H3, G12355_Nr1i3, G22337_Vdr, G5914_RARA, G7421_VDR, G9970_NR1I3
S.-808G>A	0.36±0.09a	0.35±0.11a	0.42±0.1a	MA0102.4_CEBPA

Table 6. Antigen-specific CD8⁺ T Cell Abundance and the Genetic Variation of Single-Nucleotide Polymorphisms (SNP) in Individuals.

SNP ID	Antigen-specific CD8+ Cell Abundance for Genotype			Overlap with Transcription Factor Binding Site
	XX	Xx	xx
S.-291C>T	0.36±0.1a	0.34±0.1a	0.31±0.11a	G7566_ZNF18
S.-404A>G	0.39±0.08a	0.28±0.08b	0.21±0.1b	G15376_Foxa2, MA0036.3_GATA2, MA0037.4_Gata3, MA0482.2_GATA4, MA1104.2_GATA6, MA1970.1_TRPS1
S.-463A>C	0.36±0.09a	0.35±0.1a	0.34±0.09a	G20852_Stat6
S.-604C>T	0.36±0.1a	0.35±0.08a	0.36±0.11a	G4772_NFATC1
S.-668C>T	0.41±0.07a	0.27±0.06b	0.19±0.08b	G10062_NR1H3, G12355_Nr1i3, G22337_Vdr, G5914_RARA, G7421_VDR, G9970_NR1I3
S.-808G>A	0.36±0.09a	0.35±0.11a	0.42±0.1a	MA0102.4_CEBPA

Note: S.-291C > T indicates a C to T mutation at base 291, located upstream of the transcription start site. This nomenclature is universally applied to analogous sites. XX represents the dominant homozygote genotype, Xx represents the heterozygote genotype, and xx represents the recessive homozygote genotype.

4. Discussion

Antigen-specific T cells are a critical component of the adaptive immune response. These cells specifically recognize antigenic peptides presented by major histocompatibility complex (MHC) molecules through their T cell receptors (TCRs), and they play a pivotal role in anti-infection, anti-tumor, and immune regulation [46]. The Activation of these cells requires a dual signal: a primary signal derived from the MHC-peptide-TCR trimolecular complex, and a secondary signal provided by a costimulatory molecule such as CD28/B7. This process ultimately results in the differentiation of effector T cells, including CD8⁺ cytotoxic T cells, CD4⁺ helper T cells, and memory T cells [47]. Research has demonstrated that the clonal expansion, functional polarization, and memory formation of antigen-specific T cells directly impact the strength and durability of the immune response [48].

This study examined the ACD8⁺ abundance variations (ASFV p70 protein) in a small population of ASFV restored pig farms. We combined this with transcriptome sequencing to identify the key TFs regulating ASFV ACD8⁺ abundance. In this study, we sought to quantify and characterize ASFV antigen-specific CD8⁺ T cells directly. To this end, we prepared tetrameric complexes of SLA-1 folded with the SQIEETHLV epitope of the ASFV p72 protein and the FVTPEIHNL epitope, respectively. Preliminary studies have demonstrated that the tetrameric complex stains ACD8⁺ from individuals who are positive for p72 antibody, but not from individuals who are negative for p72 antibody. Subsequently, the tetramer was employed to investigate the ACD8⁺ abundance within the experimental population. The method provides a reliable quantitative measure for the antigen presentation process in ASFV infection. The present study also employed correlation analysis, which was implemented in the FLAVER software package developed by Yao et al. [32,33,34] to test the significance of the correlation between the order of genes in the gene set and the corresponding order in the gene list. The FLAVER is a software package that has been developed to identify the key TFs from data derived from transcriptomes.

Among the key TFs identified by transcriptome sequencing, MECP2 has been shown to recognize and bind methylated cytosine (5mC) on DNA, and to recruit other protein complexes (such as histone deacetylase HDACs, Sin3A, etc.) to modify chromatin structure [49]. MECP2 has been reported to suppress the expression of antigen presentation–related molecules. In macrophages, MECP2 has been observed to bind to the promoter region of the CIITA (an MHC-Ⅱ trans-activator) gene, thereby negatively regulating the expression level of MHC-Ⅱ molecules [50]. Antigen-presenting cells (APCs) lacking functional MECP2 may therefore display elevated levels of antigen presentation. ETS1 has been demonstrated to play an active and positive regulatory role in antigen presentation. This agent enhances the function of APCs, such as dendritic cells, by directly binding to and activating genes encoding key antigen-presenting elements. This, in turn, promotes the activation of T cells and the initiation of adaptive immune responses [51]. The ETS1 protein has been observed to bind directly to the promoter region of the CIITA gene. ETS1 is also play roles in regulating genes involved in antigen processing, such as CD74 [52], a molecule that plays a key role in the assembly and peptide loading of MHC-Ⅱ molecules. E2F4 can bind directly to the promoter region of the CIITA gene. This binding may result in the regulation of MHC-Ⅱ related genes. In contrast, the roles of ZBTB33, ELK1, and SP9 in antigen presentation have received less attention from the research community.

Previous research has shown that the knockout of ELK4 impairs cell proliferation and disrupts the cell cycle in bone marrow-derived mast cells (BMMCs), which is associated by reduced transcription of cell cycle related genes [53]. Furthermore, the study observed a decrease in the transcriptional activation of cytokines and chemokines, accompanied by an increase in mast cell degranulation, in ELK4 knockout BMMCs [53]. Studies showed that the repression of ELK4 results in the augmentation of macrophage markers, including CD86 and iNOS, as well as p38/JNK phosphorylation. Concurrently, this process fosters the expression of mesenchymal markers, such as N-cadherin and Vimentin, while concomitantly inhibiting E-cadherin [54]. The reduced expression of ELK4 activates the p38 and ERK signaling pathways in the MAPK signaling pathway, promoting the polarization of macrophages toward the M1 phenotype [54]. In mice lacking ELK4 and ELK1, higher level of innate-like αβ CD8⁺ T cells develop, which populate the periphery [55]. Further studies have indicated that SRF utilizes MKL1/2 to fulfill steady-state cellular functions, including cytoskeletal organization, and utilizes ELK4 to facilitate acute responses to external infection [56].

In this study, ELK4 regulated several important genes in CLASS I MHC mediated antigen processing presentation signaling pathways. Furthermore, ELK4 expression was significantly overexpressed in high antigen-specific CD8⁺ T cells. It is hypothesized that the increase in ELK4 expression level may promote the expression of antigen-presenting genes in MHC-I, which could result in a shift in the ACD8⁺ among groups. The reduced expression level of ELK4 in the mutant group relative to the wild-type group may be attributable to the S.-404A>G and S.-668C>T mutations in the promoter region. These mutations appear to disrupt the binding of TFs FOXA2, GATAs, TRPS1, NR1H3, RARA, VDR, and NR1I3, thereby reducing the transcription level of ELK4. A key limitation of this study is the relatively small sample size (n=9). The statistical power to detect differentially expressed genes (DEGs) is constrained. This increases the likelihood of both Type II errors (false negatives), where true differential expression is missed, and complicates the reliable estimation of gene expression variance. Consequently, our findings should be interpreted as highlighting the most robust transcriptional changes. While our transcriptome data provides strong correlation evidence for the transcription factors and the target genes, this study does not include functional validation to establish a direct causal relationship. To address this, future in vivo assays and animal models research is recommended to elucidate the underlying mechanism and strengthen our conclusions.