Functional Analyses of the Histone-like A104R Protein of African Swine Fever Virus and of a Homologous Pseudogene Product Found in Soft Tick Genomes

TL;DR

This study investigates the role of the A104R protein in African swine fever virus and its homolog in ticks, finding it not essential for replication but affecting virus growth.

Contribution

The study reveals the functional role of A104R and its tick homolog in ASFV replication and protein interactions.

Findings

A104R is not essential for in vitro replication of ASFV.

Loss of A104R leads to reduced virus titers and plaque sizes.

pA104R interacts with viral and cellular proteins independently of DNA.

Abstract

African swine fever virus (ASFV) causes a fatal disease in domestic pigs and wild boars (Sus scrofa), leading to nearly 100% mortality during acute infection and significant economic losses in swine production. Unlike other eukaryotic viruses, ASFV encodes a histone-like nucleic acid-binding protein, pA104R, which is highly conserved and present in all described ASFV isolates of different genotypes. Moreover, A104R-like sequences have been identified in the genomes of soft ticks, which can replicate and transmit ASFV. Using a virulent genotype IX field isolate from Kenya, we analyzed the importance of A104R for viral replication in a permissive wild boar cell line (WSL). In this study, we confirmed that A104R is not essential for in vitro replication of ASFV. Loss of A104R did not detectably affect viral DNA replication or RNA transcription but led to a moderate reduction in virus titers and plaque sizes. Substitution of A104R with a similar ASFV-like element derived from the genome of an Ornithodoros moubata soft tick was not capable of rescuing the deletion mutant phenotype. In contrast, reintroduction of the authentic A104R open reading frame (ORF) into the deletion mutant fully restored wild-type virus growth properties. In accompanying studies, we verified the DNA-binding activities of the ASFV- and tick-derived A104R proteins and performed mass spectrometric analyses of the pA104R interactome. These experiments revealed, besides DNA-dependent co-precipitated proteins, specific DNA-independent protein–protein interactions of pA104R with other viral and cellular proteins.

Figures

Full text

1. Introduction

African swine fever virus (ASFV, Asfivirus haemorrhagiae) is the causative agent of African swine fever (ASF), a lethal viral disease of domestic swine and its direct ancestor, the Eurasian wild boar (Sus scrofa). ASF was first described in 1921 [1], and is endemic in Sub-Saharan Africa where 23 different genotypes forming six groups were described based on the gene sequence of the major structural protein p72, but only two genotypes (I, II) have hitherto spread to other continents [2,3,4,5,6]. Currently, a highly virulent genotype II virus causing mortalities of up to 100% has spread over major parts of Europe and Asia, as well as some Caribbean countries [7,8]. Given the current lack of generally approved vaccines or antiviral pharmacological interventions, ASFV is regarded as a global threat to pig husbandry. While lethal to domestic swine and wild boar, ASFV infections are mostly subclinical in African wild pig species such as warthogs and bush pigs. Between these wild pigs, ASFV is transmitted in a sylvatic cycle by permissive soft ticks, such as the Ornithodoros moubata (O. moubata) complex [9,10].

Outside Africa, ASFV outbreaks have occurred from the late 1950s in Portugal and other countries of Southern and Western Europe, as well as in South America (the Caribbean and Brazil). These outbreaks were contained and eliminated [10]. In 2007, ASFV was introduced into Georgia and spread through the Transcaucasian countries and the Russian Federation, which was followed by dissemination to the European Union [11] and East Asia, including China [12]. The introduction of ASFV elicits trade bans of pork products, resulting in significant socioeconomic loses [13].

ASFV is a double-stranded DNA (dsDNA) virus and is currently the only characterized member of the family Asfarviridae in the order Asfuvirales, as well as the only known DNA arbovirus infecting mammals [14]. Together with the Poxviridae, it has been classified into the phylum Nucleocytoviricota and the class Pokkesviricetes (https://talk.ictvonline.org/taxonomy (accessed on 1 February 2026)). ASFV is a complex virus with an icosahedral shape, consisting of concentric layers of protein shells and two lipid envelopes. Depending on the isolate, the dsDNA genome ranges from 170 to 193 kbp and contains between 150 and more than 190 potentially protein-coding open reading frames (ORFs) [15]. ASFV predominantly replicates in virus factories located in the cytoplasm [15] but has also been reported to undergo an intranuclear viral DNA replication phase [16].

The ASFV ORF A104R encodes a histone-like protein (pA104R), which is recruited into viral factories [17] and is present in virus particles [18,19] as part of the nucleoid, a structure containing the viral genome, nucleoproteins, and the transcriptional machinery [19,20]. Furthermore, it has been proposed to participate in the heterochromatization of the host cell genome [17,21]. pA104R binds single-stranded (ss) and dsDNA, with a higher affinity for dsDNA [17,22]. Collaboratively with the topoisomerase II identified in ASFV (pP1192R), pA104R has been reported to induce DNA supercoiling and to be involved in viral replication and transcription [17]. The amino acid sequence of pA104R is highly conserved across multiple ASFV genotypes [17]. pA104R shares detectable sequence homology (25–30%) with two families of bacterial histone-like proteins (IHF and HU) [22,23], which have been termed nucleoid-associated proteins (NAPs). NAPs are reported to contribute to both the control of gene expression, through wrapping, bending, and bridging of DNA, and the organization of the nucleoid, conferring structure on the bacterial genome [24]. Whether pA104R fulfills similar functions remains to be elucidated, and recent analyses of the crystal structure of pA104R in complex with DNA might contribute to this [25].

Next-generation sequencing (NGS) has revealed the presence of ASFV-like integrated (ASFLI) elements in the genome of O. moubata complex soft ticks, which included ASFLI-A104R [26]. Although it was proposed that the integration of these elements occurred more than 1.47 million years ago, ASFLI-A104R was highly conserved in nucleotide sequence (83–85%) and in deduced amino acid sequence identities (94–95%). However, the insertion of one nucleotide in a homopolymer region led to a frameshift and a premature stop codon [26].

In the past, A104R has been considered to be essential for ASFV replication and has been suggested as a candidate gene for the development of ASFV disabled infectious single-cycle (DISC) vaccines [17,27]. In a recent study with the Georgia 2010 strain (genotype II), an A104R-deleted mutant showed drastically reduced virulence in experimentally infected swine; however, the deletion mutant was not able to induce protection against lethal challenge with the parental virus in these animals [28].

In this study, we analyzed whether or not pA104R is also dispensable for in vitro replication of a virulent genotype IX field isolate from Kenya (ASFV-IX-Kenya-1033, AK1033) [29]. Unlike current European genotype II isolates, AK1033 replicated efficiently in a permanent wild boar lung cell line (WSL) without the need for adaptation [30]. Based on this ASFV isolate, we generated A104R-deleted mutants with and without reporter gene insertions, a virus revertant, and a substitution mutant, containing the ASFLI element-based A104R derived from soft ticks, in which the frameshift had been repaired. Presence or absence of pA104R was confirmed through Western blot and indirect immunofluorescence assays (IFA). In vitro growth kinetic studies, as well as plaque assays on WSL cells, were used for comparative analysis of in vitro replication properties. Furthermore, the kinetics of viral DNA replication in cells infected with wild-type, mutant, and revertant virus (qPCR) and the kinetics of viral early (CP204L/p30) and late gene (B646L/p72) transcription (RT-qPCR) were compared. Moreover, the DNA-binding properties of in vitro-expressed native and ASFLI element-based pA104R, as well as the protein interactomes of the GFP-tagged pA104R in plasmid-transfected WSL cells in the presence and absence of DNA were investigated.

2. Materials and Methods

2.1. Cells and Viruses

The ASFV-permissive wild boar lung cell line WSL [31,32], as well as a rabbit kidney cell line (RK13), were cultivated in Iscove′s modified Dulbecco’s medium with Ham’s F-12 Nutrient Mix and 10% fetal bovine serum (FBS) at 37 °C and maintained in the same medium with only 5% FBS after ASFV infection or plasmid transfection. For plaque assays, the medium was supplemented with 6.7 g/L methyl cellulose.

The virulent field isolate AK1033 (GenBank# OZ005801) [29] was kindly provided by R. Bishop and L. Steinaa from the International Livestock Research Institute (ILRI) in Nairobi, Kenya. The virus was propagated in WSL cells and used for the construction of the described virus recombinants. The cell culture-adapted ASFV isolate Armenia 2008 has also been described previously [30].

Porcine peripheral blood mononuclear cells (PBMCs) were prepared as described [33]. One day after isolation, fresh medium supplemented with 5 ng/mL granulocyte-macrophage colony-stimulating factor (GM-CSF) (Kingfisher Biotech, Saint Paul, MN, USA) was added to enhance differentiation, viability, and susceptibility to productive ASFV infection [34]. The differentiated and activated cells were then used for comparative replication studies of the generated ASFV mutants.

2.2. Construction of Plasmids and Virus Recombinants

For cloning of the A104R gene region of AK1033 (Figure 1A), a 2794 bp fragment was amplified through PCR from genomic virus DNA using primers AKA104RR-F and AKA104RR-R (Table S1), as well as KOD Xtreme Hot Start DNA Polymerase (Merck, Darmstadt, Germany). In these primers, the ASFV-specific sequences (bold) were preceded by vector-specific sequences (in Italics), which permitted insertion into the SmaI-digested plasmid pUC19 through ligase-free hot fusion cloning [35], resulting in pUC-AKA104RR.

To obtain an A104R deletion plasmid, pUC-AKA104RR was subjected to PCR using primers AKDA104RHF-F and AKDA104RHF-R (Table S1), which amplified a 5280 bp fragment, including the vector, but lacking the A104R ORF from codon 3 to the end. The 5′ parts of the primers (in Italics) matched a LoxP site-flanked expression cassette containing the eGFP and human CD4 ORFs under control of the ASFV p72 and p30 promoters, respectively [33], which had been isolated as a 2459 bp HindIII/SmaI fragment from pUC-LoxPPp72GFPPp30huCD4. Hot fusion of the two fragments resulted in plasmid pUC-AKΔA104R-GFP (Figure 1B).

In a final cloning step, the modified human CD4 ORF was replaced by a synthetic version of the frameshift-corrected A104R-like gene form O. moubata soft ticks [26]. To this end, a 6355 bp fragment of pUC-AKΔA104R-GFP lacking the CD4 ORF was amplified through PCR with primers ASFVPolyA-F and ASFVPp30-R (Table S1) and used for hot fusion with the synthetic A104Rtick gene, which had been amplified as a 371 bp fragment using primers PA104RZ-F and PA104RZ-R (Table S1), resulting in plasmid pUC-AKA104Rtick-GFP (Figure 1B, Figure S1).

For the generation of GFP-expressing A104R deletion or substitution mutants of ASFV (AKΔA104R-GFP and AKA104Rtick-GFP), WSL cells were grown overnight to subconfluent monolayers and transfected with the corresponding transfer plasmids using the K2 Transfection System (Biontex, München, Germany), as recommended by the manufacturer, but including a centrifugation of the plate for 1 h at 25 °C and 690× g after DNA addition. After 6 h, the transfection solution was removed, and cells were infected with AK1033 at a multiplicity of infection (MOI) of approx. 1 through centrifugation for 1 h at 37 °C and 690× g. Then, the inoculum was replaced by fresh medium and cells were incubated at 37 °C until green fluorescence and cytopathic effect (CPE) became visible (approx. 5 d). After freeze-thawing of the cultures, serial dilutions of the virus progenies were analyzed through plaque assays on WSL cells. Single foci of green fluorescent cells were identified by microscopy, labeled, and aspirated, and plaque purification was repeated until the virus populations of ASFV AKΔA104R-GFP and AKA104Rtick-GFP (Figure 1B) appeared homogeneous (3 to 5 times).

The clean deletion mutant AKΔA104R-loxP (Figure 1C) was obtained through Cre-Lox recombination after infection of WSL-CreCneo cells [33] with the plaque-purified mutant AKΔA104R-GFP and selection of non-fluorescent progeny virus plaques.

To facilitate generation and isolation of the A104R revertant AKA104RR (Figure 1C) and rescuants of other GFP-expressing ASFV mutants, an eGFP gene-specific guide RNA gene sequence [36] was inserted into the modified CRISPR/Cas9 vector pX330-ΔNLS1/2neoR, and the plasmid was used for stable transfection of WSL cells, as described [30]. The obtained WSL-Cas9EGFPgRneoR cells were used for transfection with plasmid pUC-AKA104RR and subsequent infection with ASFV AKΔA104R-GFP, and non-fluorescent progeny virus was plaque-purified to homogeneity and characterized by DNA and protein analyses.

For transient expression studies, the synthetic, frameshift-corrected A104Rtick ORF [26] and a similar synthetic codon-adapted version of the authentic A104R ORF of AK1033 (GeneArt, Thermo Fisher Scientific, Regensburg Germany) were inserted into the EcoRI-digested mammalian expression vector pCAGGS (GenBank# LT727518), resulting in pCAGGS-A104Rtick and pCAGGS-AK104R.

For interactome studies, another codon-adapted synthetic ASFV A104R ORF was generated (Twist Bioscience, South San Francisco, CA, USA) and shuttled via Gateway technology (Invitrogen, Thermo Fisher Scientific, Darmstadt, Germany) into the pcDNA6.2/N-EmGFP-DEST plasmid backbone (Invitrogen) to permit the expression of a GFP fusion protein (pA104R-GFP) (Figure S2). The empty vector pcDNA6.2/N-emGFP-DEST was used for the expression of GFP as a negative control.

2.3. Virus Propagation and DNA and RNA Preparation from Infected WSL Cells and ASFV Virions

WSL cells were grown in tissue culture flasks (25 to 175 cm^2^) to subconfluent monolayers and infected at a low MOI with the investigated ASFV mutants. After the development of pronounced CPE, the cell cultures were lysed by freeze-thawing, and centrifuged for 10 min at 2000× g and 4 °C. The supernatants were stored as virus stocks at −80 °C, whereas the cell pellets were used for preparation of total DNA, as described [37].

For the preparation of high-quality ASFV DNA for next-generation sequencing (NGS), 35 mL of virus supernatants were centrifuged in a Beckman SW32 rotor for 1 h at 20,000 rpm and 4 °C. The pellet was resuspended in 1 mL TE (pH 7.4), and, after addition of 10 µL 1 M MgCl_2_, 20 µL DNase I (5 U/µL), and 10 µL RNase A (10 µg/µL), incubated for 1 h at 37 °C. In a SW32 centrifugation tube 30 mL of phosphate-buffered saline (PBS) was under-layered with 6 mL of 30% sucrose in PBS, and the nuclease-treated particles were added on top. After centrifugation for 2 h at 20,000 rpm and 4 °C, the supernatant was completely aspirated, and the pellet was used for DNA preparation as above [37].

For small-scale nucleic acid extraction for qPCR, infected WSL cells in 24well plates were harvested using TRIzol™ LS Reagent (Thermo Fisher Scientific) and processed according to the manufacturer’s protocol for separate isolation of DNA and RNA. RNA was further purified using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and incubated with RNase-free DNase I (Qiagen) on the spin columns before elution.

2.4. Sequence Analyses

The A104R gene region of all obtained ASFV Kenya mutants was amplified through PCR using KOD Xtreme Hot Start DNA Polymerase (Merck) and primers AKA104RRHF-F and AKA104RRHF-R (Table S1). After separation on agarose gels, the PCR products were isolated (Zymoclean Gel DNA Recovery Kit, Zymo Research, Freiburg, Germany) and investigated through Sanger sequencing using the BigDye™ Terminator v1.1 Cycle Sequencing Kit (Thermo Fisher Scientific) and the above-mentioned primers, as well as primers A104RgR1-F, A104RgR1-R, A224LR2-R, EGFP-SNF, EGFP-SNR, EGFP-SCF, or huCD4-OUT (Table S1) for complete coverage of the amplification products. DNA sequences were determined in an Applied Biosystems 3500 Genetic Analyzer (Thermo Fisher Scientific) and evaluated using the Geneious Prime 2021.0.1 software package (available from https://www.geneious.com).

The complete genome sequence of ASFV KΔA104RGFP was determined from purified virion DNA by custom Illumina NovaSeq NGS (Eurofins Genomics, Ebersberg, Germany), yielding 3,771,507 pairs of 150 nt reads, of which approximately 1.15 million were identified as ASFV-specific, resulting in a mean coverage of > 900 reads per genome position. The reads were assembled using the Geneious Prime program “Map to Reference” and the in silico mutagenized genome sequence of AK1033 [29] (GenBank# OZ005801), as well as the sequence of the related genotype IX isolate ASFV R7 from Uganda [38] (GenBank# MH025917) as references. The assembled and annotated 189,148 bp sequence has been deposited in GenBank (# MZ566623).

2.5. Western Blot Analyses

ASFV-infected (MOI 2, 48 h p.i.) and uninfected WSL cells were lysed, and proteins were separated through discontinuous sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose membranes, and blots were subsequently incubated and blocked with TBS-T with 5% w/v skimmed milk powder, as described [39]. Membranes were incubated with polyclonal rabbit antisera against either pA104Rtick [26], p30 [36], or p72 [33] and the α-tubulin specific mouse monoclonal antibody B-5-1-2 (Sigma-Aldrich, Merck, Schnelldorf, Germany) at dilutions of 1: 10,000 in TBS-T with 0.5% w/v skimmed milk powder overnight. Membranes were subsequently washed with TBS-T and co-incubated for 1 h with IRDye 800 CW Donkey anti-Rabbit and IRDye 680 RD Donkey anti-Mouse antibodies (LI-COR, Lincoln, NE, USA) at dilutions of 1:20,000 in TBS-T, each. After repeated washing with TBS-T, the membranes were imaged using the LI-COR Odyssey CLx machine.

2.6. Immunofluorescence Analyses and Plaque Size Determination

WSL cell monolayers grown on 24well plates were infected with serial dilutions of wild-type (WT) or mutant ASFV and centrifuged for 1 h at 690× g at 37 °C. Then the inoculum was replaced by semi-solid methyl cellulose medium, and the cells were incubated for 4 d at 37 °C. The cells were fixed with acetone/methanol (1:1) for 30 min at −20 °C. In appropriate wells, the cells were rehydrated and washed with PBS, prior to being blocked with 10% FBS in PBS for 1 h. Then, the cells were incubated with either rabbit α-p72 (1:100) or α-pA104Rtick (1:200) sera in the same buffer for an additional hour at RT. After repeated washing with PBS, the cells were incubated with AlexaFluor 488-conjugated goat α-rabbit IgG (Invitrogen, Thermo Fisher Scientific), diluted 1:1000 in PBS for 1 h at RT. The plates were again washed with PBS and analyzed using a Leica DMi8 motorized fluorescence microscope and Leica Application Suite X software in version 3.6.0.20104 (© 2018). Whole wells were imaged using the automated well coverage function, and the resulting mosaic images were merged. For each virus, the areas of ≥50 plaques stained for p72 were determined using ImageJ (https://imagej.nih.gov) in 4 independent experiments. Mean relative sizes compared to WT ASFV plaques, which were set at 100%, as well as standard deviations, were calculated. The statistical significance of differences was calculated using GraphPad Prism 10.4.1 (released 2024) software (one-way ANOVA and Tukey’s multiple comparisons test). Furthermore, the effect sizes (Cohen’s d) were calculated.

2.7. Virus Replication Kinetics

Virus replication kinetics were performed on WSL cells or differentiated PBMCs grown overnight to confluent monolayers in 24well plates. The cells were infected at an MOI of approx. 0.03 with all investigated ASFV variants in four (WSL) or two (PBMC) replicas for each incubation time. The infection of WSL cells was synchronized through centrifugation of the plates for 1 h at 690 × g and 37 °C. The infected PBMCs were incubated for 1 h at 37 °C without centrifugation. Then, the inoculum was replaced by fresh medium containing penicillin and streptomycin, and single plates were frozen at −80 °C immediately thereafter and after 24, 48, 72, 96, 120, and 144 h at 37 °C. After thawing, the lysates were serially diluted, and virus was titrated through plaque assays on WSL cells grown in 96well plates, as described above. After 120 h at 37 °C, the cells were fixed, and appropriate wells were analyzed through indirect IFAs with the α-p72 rabbit serum as above to facilitate plaque counting. Mean virus titers of the four replicas of each mutant at any time and standard deviations were calculated and plotted. The statistical significance of differences was calculated using GraphPad Prism 10.4.1 software (one-way ANOVA and Dunnett’s multiple comparisons test). Furthermore, the effect sizes (Cohen’s d) were calculated.

2.8. DNA Replication and RNA Expression Kinetics

For investigation of viral DNA replication and viral early CP204L (p30) and late B646L (p72) gene transcription, WSL cells grown in 24well plates were infected as described above with wild-type, mutant, and revertant ASFVs at an MOI of 3 in two replicas for each time and virus. After 0, 2, 4, 8, 16, and 32 h at 37 °C, the medium was aspirated, and the cells were washed with PBS and lysed with TRIzol™ LS Reagent (Thermo Fisher Scientific) for separate preparation of total DNA and RNA (see above). DNA and RNA concentrations were determined using a Nano-Photometer (Implen, München, Germany), and 10 ng/µL dilutions were prepared in nuclease-free water.

Normal qPCR for DNA detection was performed using the QuantiTect Multiplex PCR NoROX kit (Qiagen) in 20 µL reactions containing 20 ng of infected or uninfected cell DNA according to the manufacturer’s instructions. The ASFV B646L-specific primer pairs, AKB646L-408F and AKB646L-507R (Table S1), as well as the β-actin gene-specific primer pair, ACT-CP-F and ACT-CP-R (Table S1), were included at 800 nM, and the TaqMan probes AKB646L-460P and ACT-CP-P (Table S1, purchased from Eurogentec, Seraing, Belgium) were included at 160 nM final concentrations. The samples were incubated for 15 min at 95 °C, followed by 45 cycles of 30 s 95 °C, 30 s 55 °C, and 30 s 68 °C in a Bio-Rad C1000/CFX96 real-time PCR machine, and results were analyzed using CFX Maestro software (Bio-Rad, Hercules, CA, USA). To rule out errors caused by slightly different input DNA amounts or inhibitory effects in certain samples, the results of the B646L-specific PCR component were normalized to those of the actin gene-specific one in the same samples. ASFV genome copies were determined based on standard curves generated by reactions containing 10^8^, 10^6^, 10^4^, 10^2^, or 10^0^ copies of a p72-expression plasmid (pCAGGS-p72-Georgia, kindly provided by G.M. Keil).

RT-qPCR reactions were performed using the One-Step RT-qPCR ToughMix (Quantabio, Beverly, MA, USA) in 20 µL reactions containing 20 ng of total RNA from infected or uninfected cells according to the manufacturer’s instructions. The ASFV-specific primer pairs, AKB646L-408F and AKB646L-507R, or AKCP204L-346F and AKCP204L-426R (Table S1), as well as ACT-CP-F and ACT-CP-R, were included at 600 nM, and the TaqMan probes, AKB646L-460P or AKCP204L-370P (Table S1), and ACT-CP-P at 120 nM final concentrations. The samples were incubated for 10 min at 50 °C and 2 min at 95 °C, followed by 45 cycles of 15 s 95 °C, 30 s 55 °C, and 45 s 60 °C in a Bio-Rad C1000/CFX96 cycler. The results of the B646L- or CP204L-specific PCR components were again normalized to those of the actin gene-specific ones in the same samples. The mean qPCR and RT-qPCR results of the two replicas of each virus and time were plotted, including standard deviations.

2.9. DNA-Binding Studies

The DNA-binding properties of pA104R and pA104Rtick were compared using the EpiQuik General Protein-DNA Binding Assay Kit (Epigentek, Farmingdale, NY, USA), as recommended by the manufacturer, with previously described 30 bp random dsDNA oligonucleotides [40]. A total of 400 pmol of the 5′-biotin-TEG labeled primers each, DBtest-1F or DBtest-2F, and of the corresponding complementary primers, DBtest-1R or DBtest-2R (Table S1), were mixed in 200 µL H_2_O, heated to 95 °C for 10 min, slowly cooled to room temperature (RT), and stored at −20 °C.

RK13 cells were grown to sub-confluent monolayers in 6well plates and transfected with pCAGGS-AKA104R or pCAGGS-A104Rtick using XtremeGENE HP DNA transfection reagent (Merck). After 24 h, transfected and non-transfected control cells were washed with PBS, scraped into 1 mL/well cold cell lysis buffer containing 0.75% N-dodecyl β-D-maltoside, 150 mM NaCl, 100 mM Tris-HCl (pH 8.0), and complete protease inhibitor cocktail (Roche #04693159001, Merck), and incubated for 1 h in a rotator at 4 °C. Finally, cell debris was removed through centrifugation for 20 min at 12,000× g and 4 °C, and the supernatants were stored at -80 °C.

A total of 2 µL of either of the hybridized oligonucleotide pairs, and 5 µL of the supernatants of transfected or non-transfected cells diluted in assay binding buffer were added to the EpiQuik test wells and incubated for 1 h at RT. As controls, wells without DNA were also included. Then, the wells were incubated with the 1:500 diluted pA104Rtick-specific rabbit antiserum [26] for 1 h and with 1:10,000 diluted peroxidase-conjugated goat anti-rabbit IgG (Jackson Immunoresearch, West Grove, PA, USA) for 30 min. The plates were repeatedly washed after each step and incubated with developing solution for 15 min. After the addition of stop solution, the adsorption at 450 nm was measured in an Infinite 200 PRO ELISA reader (Tecan, Männedorf, Switzerland). The results of three independent experiments with all cell supernatants and oligonucleotide hybrids were normalized to those obtained in the respective negative controls (without DNA), which were set to one. Standard deviations and statistical significance of differences (two-sided t-tests) were determined.

2.10. A104R Interactome Studies

2.10.1. Transfection

For affinity purification, WSL cells were plated in 6well plates and transfected with 2.4 μg of GFP-tagged expression constructs or GFP control according to the manufacturer’s recommendations (K2 Transfection System, Biontex). For each bait, n = 3, independent biological replicates were prepared for affinity purification.

2.10.2. Affinity Purification

Six hours after plasmid transfection, a total of approximately 5 × 10^6^ WSL cells (expressing GFP or pA104R-GFP) were mock-infected or infected with cell culture-adapted ASFV Armenia 2008 at an MOI of 2 PFU/cell. After 18 h (24 h after transfection), the cells were washed thrice with PBS and lysed on ice in 1 mL cold immunoprecipitation (IP) buffer [(50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM MgCl_2_) supplemented with 0.5% Nonidet P40 substitute (NP-40; Sigma-Aldrich #I8896) and complete mini EDTA-free protease inhibitor cocktail (Roche #04693159001)]. To distinguish between DNA/RNA-dependent and DNA/RNA-independent protein–protein interactions, we lysed cells under two conditions: (1) lysis buffer supplemented with benzonase (25 U/mL, Sigma-Aldrich #E8263) and (2) lysis buffer without benzonase. The cells were lysed through sonication 3 × 30 sec, 80% amplitude (Branson Digital Sonifier 450, Emerson St. Louis, MO, USA) and incubated on a tube rotator for 30 min at 4 °C and another 30 min on a thermomixer at 37 °C with constant shaking at 900 rpm. The lysates were cleared through centrifugation at 13,000× g and 4 °C for 15 min. From each lysate, 50 µL were removed for immunoblotting (whole cell lysate fraction). GFP-trap agarose beads (50 µL, ChromoTek, Planegg, Germany) were washed twice in the IP buffer and incubated with the remaining lysate for 1 h with constant rotation at 4 °C. The beads were washed twice with 1 mL of IP buffer containing 0.05% NP-40, followed by two washes in a detergent-free IP buffer. A total of 10 µL bead slurry was removed for immunoblotting (IP fraction). The remaining 40 µL beads prepared in at least three independent biological replicates for each bait were kept for mass spectrometric analysis.

2.10.3. Sample Preparation for Mass Spectrometry

After IP, beads were suspended in 300 µL freshly made UA buffer [8 M urea, 100 mM Tris-HCl pH 8], loaded onto 10 kDa filter units (Sartorius, Göttingen, Germany) and centrifuged at 12,000× g at 20 °C for 30 min. Filter-aided sample preparation (FASP) trypsin digestion was performed as described previously [41]. The beads were trypsinized to digest the baits and the interacting proteins in 100 μL of digest buffer [1 M urea, 50 mM Tris-HCl pH 7.5 and 5 μg/mL Trypsin (Promega, Madison, WI, USA)]. The digestion was performed overnight at 37 °C with shaking. The next day, the flow-through containing peptides was collected, and the samples were inactivated for 10 min at 95 °C. The peptides were acidified with formic acid (1% final concentration) and desalted using C18-100 µL tips (Thermo Fisher Scientific) according to the manufacturer’s instructions, as well as dried by vacuum centrifugation and reconstituted in 20 μL of 0.1% formic acid prior to mass spectrometry (LC-MS/MS).

2.10.4. Protein Identification Through LC-MS/MS

Digested peptide mixtures were analyzed through LC-MS/MS on a timsTOF Pro (Bruker Daltonics, Bremen, Germany), which was coupled online to a nanoElute nanoflow liquid chromatography system (Bruker Daltonics) via a CaptiveSpray nano-electrospray ion source. Peptides (corresponding to 100 ng) were separated on a reversed-phase C18 column (25 cm × 75 µm i.d., 1.6 µm, IonOpticks, Collingwood, Australia) with a binary buffer system of buffer A (water with 0.1% formic acid, v/v) and buffer B (acetonitrile with 0.1% formic acid, v/v). Peptides were separated running a gradient of 2% to 95% mobile phase B over 115 min (2% to 16% solvent B (0–60 min), 15% to 24% solvent B (60–90 min), 24–34% solvent B (90–105 min), 35% to 95% solvent B (105–107 min), and 95% solvent B (107–115 min)) at a constant flow rate of 400 nL/min. The column temperature was controlled at 40 °C. MS analysis of eluting peptides was performed in the dda-PASEF mode (1.1 s cycle time), as recommended by the manufacturer.

2.10.5. Data Analysis

Mass spectrometry raw files were processed with MaxQuant (v.2.0.2.0) [42]. Peptide search was performed against an ENSEMBL [43] Sus scrofa proteome database (v.11.1.2021-11-10) and an NCBI ASFV Georgia (v.FR682468.2) proteome database. The following modifications were included in the search parameters: trypsin digestion with a maximum of 2 missed cleavages, carbamidomethylation of cysteine as fixed modification, protein N-terminal acetylation, and oxidation of methionine as variable modifications. Mass error tolerance was set to 20 ppm for the full scan (MS1) and 40 ppm for MS/MS (MS2) spectra. False discovery rate (FDR) on peptide and protein level was set to 0.01, the minimum peptide length to seven amino acids, and the match between runs option was used with a 0.7 min match window and 20 min alignment time. The output files were analyzed using the Perseus software (v.2.0.3.0) [44]. Proteins identified only by modified peptides, reverse hits, and contaminants were filtered out. Protein was considered identified if at least two unique peptides of this protein were found in 50% of the replicates.

Proteins specifically binding to pA104R baits were filtered out by removing the GFP background. The background list consists of proteins identified in our GFP-negative controls and proteins identified as common protein contaminants for AP-MS experiments in HEK293 cells and deposited in the Contaminant Repository for Affinity Purification (CRAPome) [45]. Potential interactors were considered specific if they were identified only in GFP-bait pulldowns, or if the log2 fold change (between GFP control and GFP-bait) was greater than 2, and the p-value of a two-sided t-test was < 0.01. Protein interaction data were imported into Cytoscape (version 3.9.1) [46] to construct a protein–protein interaction network.

2.11. Biosafety

The Friedrich-Loeffler-Institut is licensed by the competent German authority to work with African swine fever virus, and all experiments with infectious ASFV were performed in improved biosafety level (BSL) 3 laboratories, which fulfill BSL 4 standards with respect to air filtration and effluent treatment.

3. Results

3.1. The A104R Gene Is Highly Conserved in Different ASFV Isolates

An amino acid sequence alignment of pA104R from our isolate AK1033 [29] and its homologs in various published isolates from different genotypes revealed amino acid sequence identities of 98 to 100% and also 95% identity to the frameshift-corrected ASFLI element-based pA104R in soft ticks. Few viral ASFV genome sequences contain an additional upstream in-frame ATG leading to N-terminally elongated deduced A104R proteins of 127 amino acids (labeled with asterisks in Figure 2). However, since this ATG is not conserved in other ASFV strains, it is presumably not functional as an initiation codon and might be located upstream of the transcription start point. Bioinformatics analysis showed that arginine-69, which is part of the DNA-binding domain [17,47,48], is conserved in all ASFV isolates, including the tick derived ASFLI-pA104R.

3.2. Generation and Characterization of A104R Deletion and Substitution Mutants of AK1033

A previous study utilized a small interfering RNA (siRNA) approach to knock down pA104R, reporting a A104R mRNA decrease of approximately 27% that resulted in reduced viral infection [17], and, recently, a A104R-deleted mutant of the genotype II ASFV isolate Georgia 2010 was shown to exhibit delayed growth in porcine macrophages and drastically reduced virulence in pigs [28]. To assess the functional relevance in AK1033, we generated A104R deletion mutants, a substitution mutant harboring the ASFLI element-based A104R and a WT A104R revertant virus (Figure 1).

Deletion and substitution mutants were obtained through homologous recombination between parental ASFVs and transfer plasmids in WSL cells, following previously described methodologies [49,50,51,52]. The used plasmids (pUC-AKΔA104RLoxPGFPhuCD4 and pUC-AKΔA104RLoxPGFPA104Rtick) contained LoxP-flanked expression cassettes for GFP under the control of the ASFV p72 promoter and either human CD4 [33] or the tick-derived and frameshift-repaired A104R under control of the ASFV p30 promoter (Figure S1) at the deleted A104R locus. Flanking sequences included the intact A224L and A240L genes (Figure 1B). The first two 5′-terminal codons of A104R were left in place to exclude the impairment of the transcription of the A224L ORF that runs in the opposite direction. After plasmid transfection, the cells were infected with ASFV WT, and plaque assays of diluted virus progenies permitted identification and purification of the desired GFP-expressing recombinants. To preclude possible adverse effects of transgene expression on replication of the deletion mutant AKΔA104R-GFP (Figure 1B), it was passaged on Cre recombinase-expressing WSL-CreCneo cells [33], which led to precise excision of the reporter cassette in the resulting virus, AKΔA104R-loxP (Figure 1C). To obtain the virus revertant ASFV-KenyaA104RR (Figure 1C), WSL-Cas9EGFPgRneoR cells expressing an eGFP gene-specific guide RNA were transfected with a recombination plasmid containing the authentic A104R gene region (pUC-AKA104RR), followed by infection with AKΔA104R-GFP and plaque purification of non-fluorescent progeny virus.

Following the generation of this panel of mutant viruses, their genomic DNA was prepared from infected WSL cells and analyzed through PCR amplification and sequencing of the A104R gene region using specific primers (Table S1). Compared to WT, the PCR products of all virus mutants exhibited the expected size shifts, and the revertant ASFV-KenyaA104RR again showed a wild-type-sized product (Figure 3A). No additional products indicating wild-type contaminations of any of the mutant virus stocks were detectable, and sequence analyses confirmed the presence of the desired deletions and insertions. Furthermore, virus particles of AKΔA104R-GFP were purified, and the prepared virion DNA was analyzed through NGS (GenBank# MZ566623). A comparison of the obtained 189,148 bp sequence to the available genome sequence of parental WT virus (GenBank# OZ005801) [29] confirmed the desired mutation of A104R and revealed the additional presence of only five single-base substitutions or indels (Table S2). Three of them were located in non-coding regions, and the others led to single amino acid substitutions or insertions in predicted proteins of unknown functions and were unlikely to be responsible for the observed phenotypic effects. No sequence reads of the A104R ORF were found in AKΔA104R-GFP, confirming the homogeneity of the deletion mutant. The phenotypic irrelevance of the five additional mutations was confirmed through the restoration of wild-type replication properties in the virus revertant AKA104RR (see below), which was obtained after homologous recombination of the deletion mutant with a plasmid-cloned 2.8-kbp genome fragment containing only the WT A104R gene region (Figure 1C).

We further investigated the virus mutants through Western blot analysis of WSL cell lysates harvested two days after infection at an MOI of 2 (Figure 3B). Lysates were probed for the viral proteins pA104R, the phosphoprotein p30 and the major capsid protein p72, and the host cell protein α-tubulin. The analysis with a monospecific rabbit antiserum raised against the tick-derived, and frameshift-corrected pA104R [26] showed the presence of similar amounts of pA104R in lysates infected with WT, the revertant AKA104RR, and the substitution mutant AKA104Rtick-GFP (Figure 3B, upper panel). pA104R could not be detected in cells infected with either of the deletion mutants. However, p30 (an early expressed protein) and p72 (a late expressed protein) were detected in all infected cell lysates at similar levels, indicating successful infection and productive virus replication in these cells (Figure 3B). Taken together, these data show that the A104R ORF could be successfully deleted without detectably affecting the expression of other virus proteins.

3.3. Deletion or Substitution of A104R Affects Productive Virus Replication in WSL Cells and Porcine Peripheral Blood Mononuclear Cells

Based on its DNA-binding properties and its presence in the viral nucleoid [17,18,19], important functions of pA104R for virus replication were suggested. Therefore, we proceeded to determine whether the absence of pA104R resulted in phenotypic changes with respect to virus growth kinetics in cell culture. WSL cells were infected with the virus mutants and wild-type ASFV at an MOI of 0.03 and harvested immediately thereafter (0 h) and every 24 h up to 144 h post-infection (p.i.) by freezing individual plates at −80 °C. Total progeny virus titers were determined through plaque assays on WSL cells. During the first round of virus replication, between 0 and 24 h p.i, the titers of the A104R deletion and substitution mutants increased only slightly less than those of the WT or revertant viruses (Figure 4A). At later times, the differences became more pronounced, resulting in an approximately 10fold reduced titer of the A104R mutants when compared to WT and the revertant viruses, which remained almost constant until the end of the experiment after 144 h (Figure 4A). WT titers at 72 h and 120 h p.i. were significantly higher (p < 0.0001) than those of the three A104R mutants, and the calculated Cohen’s d values indicated large effect sizes (AKΔA104R-GFP: 4.04 after 72 h, 13.51 after120 h; AKΔA104R-loxP: 4.24 after 72 h, 14.28 after 120 h; AKΔA104Rtick-GFP: 4.24 after 72 h, 14.30 after 120 h).

When GM-CSF-differentiated and activated PBMCs were infected with the same mutant and WT viruses, we observed an even more salient effect. After 24 h, the titers of the A104R deletion and substitution mutants showed an approximate 2 log_10_ reduction compared to both the WT and revertant viruses (Figure 4B). This difference also almost remained constant until the end of the experiment at 120 h p.i., with the GFP harboring deletion mutant displaying a slightly lower titer compared to the substitution and GFP-lacking deletion mutant between 24 h and 72 h p.i. (Figure 4B). Statistical analyses confirmed that WT titers at 72 h and 120 h were significantly higher (p < 0.001) than those of AKΔA104R-GFP (d = 6.06 after 72 h, d = 5.04 after 120 h), AKΔA104R-loxP (d = 6.04 after 72 h, d = 5.01 after 120 h), or AKΔA104Rtick-GFP (d = 6.05 after 72 h, d = 5.02 after 120 h). The somewhat lower significances and effect sizes observed in PBMCs compared to WSL cells were due to the smaller numbers of repetition (two versus four experiments). Thus, the deletion of A104R affected the replication of ASFV, most prominently in the GM-CSF-differentiated and activated porcine PBMCs, resulting in a moderate to severe reduction in the production of infectious virus particles. Remarkably, the expression of the frameshift-corrected tick-derived A104R homolog did not rescue this defect in either WSL cells nor in the PBMCs, whereas the restoration of the authentic A104R gene did.

3.4. Deletion or Substitution of A104R Significantly Reduces Plaque Sizes

Cell-to-cell spread of the ASFV mutants and wild-type viruses was compared through plaque assays on WSL cells, which were incubated under semi-solid methylcellulose medium. The cells were fixed 96 h after infection and analyzed through indirect IFAs with monospecific sera against the major capsid protein p72 and pA104R (Figure 5A). As expected, cells infected with these viruses showed p72 staining, whereas pA104R could only be observed in cells infected with wild-type virus, the revertant, and the substitution mutant AKA104Rtick-GFP (Figure 5A). To determine plaque sizes, the fluorescent areas detected with the p72-specific antiserum were measured. In four independent experiments, 50 plaques of each virus were analyzed, and the mean areas were normalized to WT virus plaque sizes from the same experiment (Figure 5B). Compared to wild-type ASFV, plaque sizes of both A104R-deleted mutants were significantly decreased by approximately 36% (AKΔA104R-GFP, p = 0.0025, d = 4.44) and 42% (AKΔA104R-loxP, p < 0.0001, d = 4.05), and the d values indicated large effect sizes. The A104R mutant virus harboring the repaired ASFV-like element-based pA104R (AKΔA104Rtick-GFP) also produced significantly (p = 0.0153, d = 3.06), approximately 29% smaller plaques than wild-type virus. The minor increase in plaque sizes compared to the deletion mutants might indicate a partial functional complementation by pA104Rtick but was statistically not significant (Figure 5B). In contrast, the revertant virus AKA104RR exhibited a WT-like plaque phenotype (Figure 5B). These results indicate that a functional A104R protein is relevant for efficient cell-to-cell spread of ASFV.

3.5. Deletion or Substitution of A104R Does Not Impair Viral DNA Replication nor Viral Transcription

To elucidate possible reasons for the reduced virus titers (Figure 4) and plaque sizes (Figure 5) of the A104R mutants, we analyzed the kinetics of viral DNA replication and transcription of viral early (CP204L/p30) [53,54,55] and late (B646L/p72) [53,56,57] genes. To this end, WSL cells were infected with the investigated viruses at an MOI of 3 and harvested at 0, 2, 4, 8, 16, and 32 h p.i. After TRIzol extraction, total DNA and RNA was prepared from the same samples. Total DNA was used for determining ASFV genome copy numbers, using duplex TaqMan qPCR reactions for the detection of the viral B646L gene and the host cell β-actin gene as an internal control (Figure 6). B646L-specific probes showed moderately increased DNA amounts of all investigated virus variants after 4 h and an exponential replication phase until 8 h p.i. At later times, the detected amount of ASFV DNA remained almost constant at calculated levels of approximately 1000 genome copies per cell. The replication profiles of the A104R deletion and substitution mutants were similar to those of WT or revertant viruses, demonstrating that pA104R is not relevant for viral DNA replication.

Similarly, the corresponding RNA samples were used to quantify mRNA levels of the early CP204L (p30) and late B646L (p72) ASFV genes, as well as of the host cell β-actin gene for normalization. Viral and cellular mRNA levels were quantified using duplex RT-qPCR. At 2 h p.i., the CP204L-specific quantification cycle (Cq) values of all infected samples decreased by approximately 6 compared to input levels, indicating an > 50-fold increase in mRNA amounts (Figure 7A). After 8 h, the CP204L mRNA amounts dropped slightly, followed by a moderate re-increase until the end of the experiment, which was in line with previous observations (Figure 7A) [33]. In contrast, transcription of the capsid protein gene B646L was not observed before 4 h p.i. (Figure 7B), i.e., the onset of viral DNA replication (Figure 6). B646L-specific mRNA amounts of all tested virus variants increased gradually until 16 h p.i. to > 50,000-fold higher than input levels (Figure 7B). While the relative amounts of B646L transcripts of A104R-expressing and A104R-deleted viruses were similar at all investigated times, early transcription of CP204L appeared even slightly enhanced in cells infected with the GFP-expressing A104R mutants of ASFV (Figure 7A). Thus, pA104R is obviously not required for efficient early or late transcription of ASFV genes nor for the replication of viral DNA. Therefore, the moderate growth defects of A104R-deleted ASFV (Figure 4 and Figure 5) must be due to other reasons, e.g., due to less efficient packaging of viral DNA into virus particles.

3.6. Authentic and Tick-Derived pA104R Exhibit DNA-Binding Activity

To test whether the described DNA-binding properties of ASFV pA104R [17,22] were also preserved in its frameshift-corrected tick-derived homolog [26], rabbit kidney (RK13) cells were transfected with expression plasmids for either protein (pCAGGS-AKA104R or pCAGGS-A104Rtick). Clarified cell lysates were incubated in streptavidin-coated test strips (EpiQuik General Protein-DNA Binding Assay Kit, Epigentek) together with biotinylated, previously described 30 bp random dsDNA oligonucleotides [40]. After washing, the pA104Rtick-specific rabbit antiserum [26] and peroxidase-conjugated anti-rabbit IgG were used for colorimetric detection of the bound proteins. The results of three independent experiments revealed a significantly (p < 0.001) increased binding of both pA104R WT and pA104Rtick in wells containing either of the oligonucleotides (DBtest-1F/R or -2F/R) compared to wells without DNA (Figure 8). However, whereas the absorption was increased approximately 2.8-fold in the pA104R WT samples, the pA104Rtick samples showed only a 1.75-fold increase. At present it is not clear whether this difference is due to a lower DNA-binding affinity of pA104Rtick or due to a reproducibly lower expression rate of the only frameshift-corrected tick-derived gene, compared to the codon-optimized WT gene. Semiquantitative Western blot analyses of the samples indicated approximately 3-fold higher amounts of pA104R WT (Figure S3). As expected, no pA104R-specific reactions were observed in wells containing mock-transfected cell lysates (Figure 8).

3.7. DNA-Dependent and -Independent Protein Interactions of pA104R

To investigate the pA104R interactome in the context of infection and DNA binding, we expressed the A104R-GFP fusion protein in ASFV-infected WSL cells. As a negative control, we expressed GFP alone. Since in the presence of nucleic acids, completely unrelated other DNA-binding proteins might be coprecipitated indirectly with pA104R, portions of the prepared cell lysates were treated with the endonuclease Benzonase^®^ (Sigma-Aldrich # E8263), which degrades all forms of nucleic acids to short (3 to 5 nt) oligonucleotides to determine DNA-independent direct protein–protein interactions of pA104R. The interacting proteins were purified with a GFP-trap system and subjected to mass spectrometric (MS) analysis in biological triplicates. To filter out the false-positive identifications, proteins bound to GFP in the absence of pA104R were excluded from further analysis. In this way, we selected 119 true protein interactions for pA104R (Table S3) and constructed a pA104R protein interaction network (Figure 9). Next, we performed functional enrichment analysis of the 119 proteins co-purifying with pA104R using gProfiler [58]. This analysis revealed significant enrichment of Gene Ontology (GO) terms [59] related to nucleic acid binding (GO:0003676), chromatin remodeling (GO:0006338), and mitochondrial gene expression (GO:0140053) (Table S4). To further characterize the functional properties of pA104R interactors, the proteins were grouped based on common Gene Ontology Biological Process (GO:BP) annotations. Additionally, we used the STRING database [60] to identify interactions between host proteins that have been experimentally validated and the EBI Complex Portal [61] database to assign the proteins into specific protein complexes. We identified 83 DNA-dependent and 45 DNA-independent protein interactions (PPIs) unique to pA104R, with only 9 PPIs common to both conditions. This indicates that DNA binding might block domains of pA104R required for direct interactions with other proteins or induce conformational changes in pA104R. Among the DNA-dependent, possibly indirect pA104R interaction partners, host proteins engaged in nucleosome assembly (such as H1-3, H1-0, H3C1, H2AZ2, and HIST1H1E) were identified as high-confidence interactors, as indicated by their abundance (log_10_ iBAQ [intensity-based absolute quantitation]). Conversely, proteins involved in histone modification (e.g., ANP32D, ANP32E, and SET) and the sialic acid synthase (NANS) were identified as high-confidence interactors only after benzonase treatment. Interestingly, several of the few host proteins interacting with pA104R under both conditions are involved in oxidative stress response (NFE2L2/NRF2 pathway) and apoptosis, which are both modulated by ASFV infection (Figure 9). Furthermore, several viral proteins were found to interact with pA104R under either of the two experimental conditions. Remarkably, only for two of the six ASFV proteins interacting with pA104R in a DNA-dependent manner (Table S3) direct DNA interactions have been described previously: pK78R and the DNA ligase encoded by NP419L [15]. On the other hand, the three DNA-independent viral binding partners of pA104R, i.e., the chaperone pB602L, the E2-like ubiquitin-conjugating enzyme pI215L, and the abundant structural protein pA137R, are considered to play important roles in virion morphogenesis [15,62,63,64].

4. Discussion

ASFV is a highly complex virus encoding more than 160 predicted proteins [15], and many, but not all, of them could be detected through proteome analyses of infected cells [65] and/or virus particles [19]. However, up to now, the functions of most of these proteins are either anticipated from sequence homologies to characterized viral or cellular gene products or completely unknown. As a major obstacle, the establishment of permissive trans-complementing cell lines, which would allow the deletion of essential or highly relevant ASFV genes, was hitherto not successful. The histone-like A104R protein belongs to the ASFV gene products, which were initially characterized by homology-based in silico predictions and functional in vitro studies [17,22]. Since siRNA knockdown of the A104R mRNA considerably affected virus titers in cell culture, this gene was previously considered to be essential for replication [17], and, based on this fallacy, it was speculated that successful deletion of A104R in trans-complementing cells to be established might lead to DISC vaccine candidates [27]. A crucial role of pA104R was also indicated by the described inhibition of its DNA interaction with certain stilbene derivatives, as well as a concomitant reduction in ASFV DNA and virus replication in macrophage cultures [25]. However, recently, A104R could be successfully deleted from genotype II ASFV, resulting in moderately affected in vitro replication, but in a drastic reduction in virulence in pigs [28]. Regrettably, immunization with the obtained A104R-deleted mutant did not confer protection against infection with virulent ASFV [28].

For more precise investigation of A104R functions, we took advantage of a wild boar lung cell line (WSL), which has been shown to permit efficient replication of certain virulent ASFV field isolates, including the genotype IX virus AK1033 [30,32]. We were able to generate A104R deletion mutants of AK1033 after transfection of WSL cells with recombination plasmids and subsequent WT virus infection. In vitro characterization of the isolated deletion mutants, AKΔA104R-GFP and AKΔA104R-LoxP, in WSL cells revealed moderate but statistically significant replication defects, leading to plaque size reductions of approximately 40%, and titer reductions by more than one log_10_ compared to WT and A104R rescued viruses. Similar titer reductions were observed in previous studies after siRNA knockdown of the A104R mRNA in ASFV-infected Vero cells [17] or after treatment of infected macrophages with stilbene derivatives [25]. However, unlike in these studies, we could not observe an effect on viral DNA replication or viral RNA transcription of early or late genes between 0 and 32 h after synchronized high MOI of WSL cells. This might indicate that these previously observed effects of siRNA knockdown of pA104R or of inhibition of pA104R-DNA interaction by stilbene derivatives [17,25] were indirect consequences of an impaired productive virus replication after low MOI infection, or that they resulted from off-target inhibitory effects on other viral or cellular gene products required for efficient DNA replication of ASFV.

The in vitro growth curves of the WT and A104R revertant viruses after low multiplicity infection of WSL cells, as well as of GM-CSF differentiated and activated PBMCs, were almost identical. In WSL cells, the time kinetics of titer increases in the A104R deletion and substitution mutants were also wild-type like, but all mutants exhibited reduced titers from 48 h p.i. until the end of the experiment (Figure 4A). This was in line with the observation that DNA replication and early (CP204L) and late (B646L) gene transcription kinetics of wild-type and mutant ASFVs were very similar (Figure 6 and Figure 7), indicating that the histone-like pA104R might play its accessory role afterwards, e.g., during packaging of virus DNA as a prerequisite for the efficient formation of infectious particles, as proposed earlier [17]. This could be mediated by a cooperation of pA104R with pP1192R, a virally encoded topoisomerase II [66], which modulates DNA supercoiling [17], and would explain the presence of pA104R in the nucleoid of virus particles [22]. Future electron microscopic studies might contribute to the elucidation of the role of pA104R in virion morphogenesis. Another possible function of pA104R might be an interaction with the host genome, leading to heterochromatinization and gene silencing, as suggested by the detection of pA104R not only in the cytoplasmic viral factories but also in the nucleus [17,21]. Such effects might contribute to immune evasion in vivo but could also directly enhance the expression of viral proteins due to increased amounts of available amino acids, tRNAs, and ribosome components, resulting in the production of more infectious virus particles per cell. Comparative proteome analyses of cells infected with wild-type and pA104R-deleted ASFVs might elucidate such functions.

An indication for a more important role of pA104R in vivo was the pronounced reduction in virus titers of the substitution and deletion mutants observed after infection of primary porcine PBMCs. Like described for a genotype II ASFV Georgia 2010-based mutant in primary swine macrophages [28], the titers of our genotype, IX AK1033-based A104R mutants, were also reduced by approximately three log_10_ after 48 h (Figure 4B). However, unlike in the published study, our A104R deletion and substitution mutants could not close this gap at later times after infection.

A further interesting aspect of our data shows that replacing the A104R ORF with a frameshift-corrected form of a described ASFLI-A104R element from tick genomes [26] did not restore a WT phenotype, although pA104Rtick was abundantly expressed and could be detected by Western blot and indirect IFA staining (Figure 3 and Figure 5). It has been proposed that such elements have been acquired from ancient ASFV infections of ticks and may have become part of the tick immune system, which serve as templates for small interfering RNAs (siRNAs and piRNAs). Although not yet identified, such RNAs could act to protect ticks from ASFV infection or lethality [26]. Due to the lack of eukaryotic promotors and transcript processing signals, as well as to the presence of a frameshift, protein expression from the A104R-like pseudogene was also not detectable in O. moubata ticks [26].

Since there was obviously no selective pressure to maintain an intact, functional A104R ORF in ticks during evolution, it was not surprising that substitutive insertion of the frameshift-repaired ASFLI-A104R element into the ASFV genome failed to restore a WT phenotype. The 5% amino acid sequence difference between pA104R of ASFV Kenya and of the ASFLI-A104R element is not much higher than the differences between several other viral A104R proteins, but the five amino acid substitutions in pA104Rtick affect fully conserved positions that are not mutated in any of the compared viral homologues (Figure 2). Thus, although arginine-69 (R^69^) as a key amino acid of the DNA-binding domain of pA104R [17] is also conserved in pA104Rtick, and, in line with this, pA104Rtick still binds to DNA (Figure 8), the observed mutations might abrogate its function in virus replication. Furthermore, A104R has been reported to be transcribed mainly as a viral late gene [17], whereas pA104Rtick was expressed under the control of the strong early promotor of the p30-coding gene CP204L, which is also expressed at late times after infection [53,57] in the ASFV recombinant generated in the present study. Moreover, the used p30 promoter corresponds to the functionally characterized promP30_2 sequence [53] and contains an upstream in-frame ATG (see Figure S1), which might be used instead of the desired start codon of A104Rtick. Thus, different expression kinetics and protein levels of pA104R or an N-terminal 8 amino acid “tag” might also contribute to the observed phenotype of ASFV AKA104Rtick-GFP, which was similar to that of A104R-deleted mutants. Generation and characterization of ASFV recombinants in which pA104R or pAK104Rtick are expressed under control of the A104R promoter and in which the differences between the native and the tick-derived protein are consecutively corrected might further elucidate the functional relevance of individual amino acids and sequence motifs within pA104R. Furthermore, it would be interesting to investigate the effects of ASFV A104R mutations or deletion not only in porcine cells but also in susceptible tick cells.

Given the prominence of pA104R in the central nucleoid structure of virions [18,19,22], possible influence on the host cell [17,21], and the considerable effect of its previously reported knockdown [17], it was speculated that its complete removal would elicit a lethal phenotype. However, our current results confirm that pA104R is not required for productive replication of ASFV in cell culture, including the wild boar lung cell line WSL. Possibly, other cellular or viral DNA-binding proteins like pK78R of ASFV, which are also components of the viral nucleoid [19], may compensate for the absence of pA104R. Nevertheless, it is conceivable that the deletion of A104R affects productive replication of ASFVs in the natural host more severely, as evinced by our PBMC-based growth kinetics and an independent in vivo study using a genotype II-based A104R-deleted mutant of ASFV [28]. Animal experiments might elucidate whether A104R-deleted AK1033, like the corresponding ASFV Georgia 2010 mutant, is also sufficiently attenuated to prevent lethal, as well as chronic infections in pigs, but, on the other hand, still capable to induce a protective immune response. Regrettably, the previously described A104R-deletion mutant did not protect swine from a lethal challenge with the parental genotype II ASF virus [28], and, therefore, several described ASFV deletion mutants lacking other genes like I177L [67] seem to be more promising vaccine candidates, which should also be adapted to other genotypes.

To complement the phenotypical studies, we investigated the interactions of pA104R with host and viral proteins, addressing a gap in current knowledge. Our pA104R interactome analysis identified proteins localized in the cytosol and nucleus, consistent with pA104R subcellular distribution in infected cells [17,21]. Among all interactions, we have found 83 DNA-dependent direct or indirect and 45 DNA-independent direct protein–protein interactions. Remarkably, only nine interacting proteins were detected under both conditions, which justified the chosen division of the pA104R interactome (Figure 9). These striking differences might be explained by (a) an enhanced affinity of pA104R to specific proteins upon DNA-binding, e.g., by binding-induced conformational changes; (b) independent binding of pA104R and other proteins to the same DNA molecules; or (c) the influence of DNA-binding proteins associated with pA104R, which may modify the interaction landscape in its vicinity.

Interestingly, in the absence of DNA, the pA104R coimmunoprecipitate was highly enriched with the SET protein and the ANP32D and ANP32E members of the acidic leucine-rich nuclear phosphoprotein 32 family, which play a role in histone modification processes. Functionally, they are involved in histone chaperoning by inhibiting acetylation of histones and nucleosomes [68,69]. Thus, hypothetically, by recruiting these proteins, pA104R could facilitate deacetylation of histones and nucleosomes, leading to the spreading of heterochromatization [70] of the host cell genome, as previously reported [21].

Presumably, several of the observed “interactions” of pA104R in the presence of DNA are independent of each other and not relevant for function. However, although it is not surprising that host histone proteins involved in nucleosome assembly are highly abundant in the DNA-dependent pA104R interactome, several studies have linked viral protein–histone interactions to genome packaging of different viruses (reviewed in [71]). Therefore, it is plausible to hypothesize that pA104R could also regulate ASFV genome organization into nucleosome-like structures via histone–chaperone interactions. Although pA104R’s involvement in viral genome stabilization and packaging has been proposed [72], experimental validation is lacking. Our studies highlight potential functional interactions between pA104R and host and also other important virus proteins, like the chaperone pB602L, the ubiquitin-conjugating enzyme pI215L, or the immunogenic phosphoprotein p30 (pCP204L). The nonstructural ASFV protein pB602L has been shown to be required for nucleocapsid formation and is involved in maturation of the major capsid protein p72 and correct intracellular localization of the capsid protein E120R, as well as the processing of the viral polyproteins pp220 and pp62 [63]. It is conceivable that the DNA-independent interaction with the chaperone pB602L is also required for proper folding of pA104R. Furthermore, the interaction with pI215L might indicate that pA104R is a target of or plays a role in ubiquitination of other virus or host proteins, which is considered to be relevant for evasion of antiviral host responses, as well as for virion maturation of ASFV [64]. However, further molecular investigations like reciprocal co-immunoprecipitations, and, as far as possible, double knock-outs or knock-downs, are necessary to verify the relevance of these interactions.

In summary, our data show that in the absence of the histone-like DNA-binding protein pA104R, ASFV remains infectious and fully capable of genome replication and transcription in cell culture. However, the decreased plaque size and lower titers of A104R-deleted ASFV suggest that the production of infectious virions occurs with decreased efficiency. This supports the hypothesis that pA104R facilitates efficient genome packaging [17,66,73,74,75,76]. Furthermore, our data show that a restored version of the ASFVLI-A104R sequence found in the genome of O. moubata ticks [26] is still DNA-binding but not functional during ASFV replication. Moreover, we provide the first pA104R interactome as a basis for future mechanistical studies to further elucidate interactions between ASFV proteins, as well as ASFV host cell interactions.