Deletion of CD2-Like (CD2v) and C-Type Lectin-Like (EP153R) Genes from African Swine Fever Virus Georgia-∆9GL Abrogates Its Effectiveness as an Experimental Vaccine

African swine fever virus (ASFV) is currently the most dreaded infectious disease affecting the global swine production industry. There is no commercial vaccine available, making the culling of infected animals the current solution to control outbreaks. Effective experimental vaccines have been developed by deleting virus genes associated with virulence. Deletion of the ASFV 9GL gene (∆9GL) has resulted in the attenuation of different ASFV strains, although the degree of attenuation varies across isolates. Here, we investigated the possibility of the increased safety of the experimental vaccine strain ASFV-G-Δ9GL by deleting two additional virus genes involved in pathogenesis, CD2v, a CD2 like viral encoded gene from the EP402R open reading frame (ORF), and C-type lectin-like viral gene, encoded from the EP153R ORF. Two new recombinant viruses were developed, ASFV-G-Δ9GL/ΔCD2v and ASFV-G-Δ9GL/ΔCD2v/ΔEP153R, harboring two and three gene deletions, respectively. ASFV-G-Δ9GL/ΔCD2v/ΔEP153R, but not ASFV-G-Δ9GL/ΔCD2v, had a decreased ability to replicate in vitro in swine macrophage cultures when compared with parental ASFV-G-Δ9GL. Importantly, ASFV-G-Δ9GL/ΔCD2v and ASFV-G-Δ9GL/ΔCD2v/ΔEP153R induced almost undetectable viremia levels when inoculated into domestic pigs and failed to protect them against challenge with parental virulent ASFV-Georgia, while ASFV-G-Δ9GL offered robust protection during challenge. Therefore, the deletion of CD2-like and C-type lectin-like genes significantly decreased the protective potential of ASFV-G-Δ9GL as a vaccine candidate. This study constitutes an example of the unpredictability of genetic manipulation involving the simultaneous deletion of multiple genes from the ASFV genome.


Introduction
African swine fever virus (ASFV), the only member of the Asfarviridae family, has a 180-190 kilobase double-stranded DNA genome that encodes for more than 150 open reading frames (ORFs) and is the etiological agent of African swine fever (ASF), a highly dreaded disease of RPMI 1640 Medium (Life Technologies, Grand Island, NY, USA) with 30% L929 supernatant and 20% fetal bovine serum (HI-FBS, Thermo Scientific, Waltham, MA, USA) for 48 h at 37 • C under 5% CO 2 . Adherent cells were detached from the plastic by using 10 mM EDTA in phosphate buffered saline (PBS) and were then reseeded into Primaria T25, 6-or 96-well dishes at a density of 5 × 10 6 cells per ml for use in assays 24 h later. ASFV Georgia (ASFV-G) was a field isolate kindly provided by Dr. Nino Vepkhvadze, from the Laboratory of the Ministry of Agriculture (LMA) in Tbilisi, Republic of Georgia [4].
Comparative growth curves between parental ASFV-G, ASFV-G-∆9GL, ASFV-G-∆9GL/∆CD2v and ASFV-G-∆9GL/∆CD2v/∆EP153R were performed in primary swine macrophage cell cultures. Preformed monolayers were prepared in 24-well plates and infected at a MOI (Multiplicity of Infection) of 0.1 (based on tissue culture infectious doses (TCID 50 ) previously determined in primary swine macrophage cell cultures). After 1 h of adsorption at 37 • C under 5% CO 2 the inoculum was removed, and the cells were rinsed two times with PBS. The monolayers were then rinsed with macrophage media and incubated for 2, 24, 48, 72 and 96 h at 37 • C under 5% CO 2 . At appropriate times post-infection, the cells were frozen at ≤−70 • C and the thawed lysates were used to determine the titers by TCID 50 /mL in primary swine macrophage cell cultures. All samples were run simultaneously to avoid inter-assay variability. Virus titration was performed on primary swine macrophage cell cultures in 96-well plates. Virus dilutions and cultures were performed using a macrophage medium. Presence of the virus was assessed by the presence of the cytopathogenic effect (confirmed by hemadsorption in the case of ASFV-G, ASFV-G-∆9GL and fluorescence in the case of ASFV-G-∆9GL/∆CD2v and ASFV-G-∆9GL/∆CD2v/∆EP153R. Virus titers were calculated as previously described [14].

Construction of the Recombinant Viruses
Recombinant ASFV-G-∆9GL/∆CD2v and ASFV-G-∆9GL/∆CD2v/∆EP153R were generated by homologous recombination using parental virus ASFV-G-∆9GL and recombination transfer vectors p72GFP∆CD2v and p72GFP∆CD2v/∆EP153R following previously described procedures [6]. The left flanking genomic region between ASFV genome positions 72377-73377 for p72GFP∆CD2v and 71822 and 72822 for p72GFP∆CD2v/∆EP153R, while the right arms for both vectors are the same between genome positions 74452 and 75452. Both transfer vectors harbor a reporter gene cassette containing the green fluorescent protein (GFP) gene under the control of the ASFV p72 late gene promoter as previously described [15]. Recombinant transfer vectors were obtained by DNA synthesis (Epoch Life Sciences, Sugar Land, TX, USA). This construction created a 1073-nucleotide deletion between nucleotide positions 73378-74451 in the case of ASFV-G-∆9GL/∆CD2v and a 1629-nucleotide deletion between nucleotide positions 72823-74451 in the case of ASFV-G-∆9GL/∆CD2v/∆EP153R ( Figure 1). These constructs produce the deletions in CD2v (leaving only the first 9 nucleotides in CD2) for ASFV-G-∆9GL/∆CD2v and completely deleting CD2 and EP152R genes in ASFV-G-∆9GL/∆CD2v/∆EP153R. Macrophage cell cultures were infected with ASFV-G-∆9GL and transfected with either p72GFP∆CD2v or p72GFP∆CD2v/∆EP153R. CD2v recombinant viruses were purified to homogeneity by successive rounds of limiting dilution purification. ASFV DNA was extracted from infected cells and full-length sequence using next generation sequencing (NGS) was performed as described previously [4] using an Ion PGM sequencer and standard sequencing protocols. The analysis of the sequence was done using CLC Genomics Workbench software version 20 (QIAGEN, Hilden, Germany).

Assessment of the Ability of Recombinant ASFV-G-Δ9GL/ΔCD2v and ASFV-G-Δ9GL/ΔCD2v/ΔEP153R to Replicate in Swine Macrophages
To evaluate the potential effect of the deletion of CD2v and EP153R genes in the process of virus replication, the in vitro growth kinetics of recombinants ASFV-G-Δ9GL/ΔCD2v and ASFV-G-Δ9GL/ΔCD2v/ΔEP153R were assessed in swine macrophage cultures, and compared with those of the parental ASFV-G-Δ9GL in a multistep growth curve. Primary cultures of swine macrophages were infected with either recombinant ASFV-G-Δ9GL/ΔCD2v, ASFV-G-Δ9GL/ΔCD2v/ΔEP153R, parental ASFV-G-Δ9GL or field isolate ASFV-G. Infections were conducted at an MOI of 0.01 and the samples were collected at 2, 24, 48, 72, and 96 hpi.
Results demonstrated that, as previously reported, ASFV-G-Δ9GL has a decreased replication when compared with its parental virus, the field isolate ASFV-G ( Figure 2). Significant differences in virus titers (between 2.5 and 3 log10) were observed starting at 48 hpi until the end of the experimental period. In addition, ASFV-G-Δ9GL/ΔCD2v presented growth kinetics almost undistinguishable from that of its parental ASFV-G-Δ9GL. Conversely, ASFV-G-Δ9GL/ΔCD2v/ΔEP153R presented a clear decreased replication when compared to either its parental virus ASFV-G-Δ9GL or ASFV-G-Δ9GL/ΔCD2v. Significant titer differences (about 1-1.5 log10) were observed starting at 24 hpi until the end of the experiment.

Detection of Anti-ASFV Antibodies
Anti-ASFV antibodies in the sera of infected animals were quantified using an in-house developed ELISA as previously described [9]. Briefly, ASFV was amplified in Vero cells [16], infected cells were then treated with Tween 80 (G-Biosciences, St. Louis, MO, USA) and sodium deoxycholate (Sigma, Saint Louis, MO, USA) to a final concentration of 1% (v/v) and cell antigens were stored at ≤−70 • C. Maxisorb ELISA plates (Nunc, St. Louis, MO, USA) were coated with 1 µg per well of either infected cell or uninfected cell antigen and blocked with phosphate buffered saline containing 10% skim milk (Merck, Kenilworth, NJ, USA) and 5% normal goat serum (Sigma). Swine sera were tested at multiple dilutions against both infected and uninfected cell antigen, with reactivity detected by an anti-swine IgG-horseradish peroxidase conjugate (KPL, Gaithersburg, MD, USA) and SureBlue Reserve peroxidase substrate (KPL). Plates were read at OD630 in an ELx808 plate reader (BioTek, Shoreline, WA, USA). Swine sera were considered positive for ASFV-specific antibodies if the OD630 ratio of the reaction against infected cell antigen to uninfected cell antigen was higher than 2.2.

Animal Experiments
ASFV-G-∆9GL, ASFV-G-∆9GL/∆CD2v and ASFV-G-∆9GL/∆CD2v/∆EP153R were assessed for their protective effect using 80-90 pound commercial breed swine. Groups of four pigs were inoculated intramuscularly (IM) at day 0 and 21 with 10 3 TCID 50 of the prescribed virus and at day 35 all animals (including a mock vaccinated control group) were IM challenged with 10 3 TCID 50 of highly virulent parental ASFV-G. Clinical signs (anorexia, depression, fever, purple skin discoloration, staggering gait, diarrhea and cough) and changes in body temperature were recorded daily throughout the experiment.

Ethics Statement
Animal experiments to collect blood for swine macrophages were performed under biosafety level 3AG conditions in the animal facilities at Plum Island Animal Disease Center (PIADC). All experimental
Next generation sequencing (NGS) was used to assess both the accuracy of the genetic modifications introduced during recombination and the genomic integrity of the rest of the virus genome. Full-length genomic comparison of the recombinant ASFV-G-∆9GL/∆CD2v and ASFV-G-∆9GL/∆CD2v/∆EP153R and parental ASFV-G-∆9GL demonstrated a deletion of 1082 and 1629 nucleotides in ASFV-G-∆9GL/∆CD2v and ASFV-G-∆9GL/∆CD2v/∆EP153R genomes, respectively, and of the insertion of a 1229-nucleotide construct corresponding to the p72-GFP cassette sequence. No additional significant differences were observed between these two recombinant viruses and the parental virus genomes. This confirmed that ASFV-G-∆9GL/∆CD2v and ASFV-G-∆9GL/∆CD2v/∆EP153R did not acquire additional mutations during the process of homologous recombination or purification steps. Additionally, NGS data indicated the absence of any residual parental ASFV-G-∆9GL genome as contaminant of the ASFV-G-∆9GL/∆CD2v and ASFV-G-∆9GL/∆CD2v/∆EP153R stocks.
Results demonstrated that, as previously reported, ASFV-G-∆9GL has a decreased replication when compared with its parental virus, the field isolate ASFV-G ( Figure 2). Significant differences in virus titers (between 2.5 and 3 log 10 ) were observed starting at 48 hpi until the end of the experimental period. In addition, ASFV-G-∆9GL/∆CD2v presented growth kinetics almost undistinguishable from that of its parental ASFV-G-∆9GL. Conversely, ASFV-G-∆9GL/∆CD2v/∆EP153R presented a clear decreased replication when compared to either its parental virus ASFV-G-∆9GL or ASFV-G-∆9GL/∆CD2v. Significant titer differences (about 1-1.5 log 10 ) were observed starting at 24 hpi until the end of the experiment. While deletion of the CD2v gene does not significantly affect the ability of ASFV-G-Δ9GL to replicate in swine macrophages, the additional deletion of the EP153R gene significantly affected replication. As previously reported, the deletion of the CD2v gene does not significantly affect the replication of ASFV field isolates Malawi [18] or Georgia2010 [11]. Therefore, it is not surprising that the deletion of CD2v from ASFV-G-Δ9GL does not significantly affect virus growth in swine macrophage cultures. Regarding the EP153R gene, one wild-type (Malawi isolate) and one Vero celladapted (Ba71V) ASFVs were evaluated after deletion of EP153R. In both cases, the deletion of EP153R did not affect the replication of either isolate in vitro [19,20]. Interestingly, the deletion of the EP153R gene in the context of ASFV-G-Δ9GL/ΔCD2v, producing the triple deletion gene mutant ASFV-G-Δ9GL/ΔCD2v/ΔEP153R, significantly decreased the replicative abilities in swine macrophage cultures. It is not clear why the deletion of the EP153R gene affects the replication of ASFV-G-Δ9GL/ΔCD2v while it does not affect that of the Malawi or BA71V isolates [19,20]. Although swine red blood cell hemadsorption in infected cells has been shown to be solely dependent on the expression of the CD2v gene [18], it has been reported that the deletion of the EP153R gene from the BA71V genome affected CD2v expression and hemadsorption [20]. It is possible that CD2v and EP153R have differential and overlapping functions, explaining why simultaneous deletion severely affects virus replication.

Assessment of Recombinant ASFV-G-Δ9GL/ΔCD2v and ASFV-G-Δ9GL/ΔCD2v/ΔEP153R to Induce Protection in Immunized Pigs
ASFV-G-Δ9GL has been shown to be attenuated when intramuscularly (IM) inoculated in swine at doses of 10 3 TCID50 and to induce a strong protective immunity when challenged with parental virulent ASFV-G isolate. The effect of the deletion of CD2v and EP153R genes in the protective While deletion of the CD2v gene does not significantly affect the ability of ASFV-G-∆9GL to replicate in swine macrophages, the additional deletion of the EP153R gene significantly affected replication. As previously reported, the deletion of the CD2v gene does not significantly affect the replication of ASFV field isolates Malawi [18] or Georgia2010 [11]. Therefore, it is not surprising that the deletion of CD2v from ASFV-G-∆9GL does not significantly affect virus growth in swine macrophage cultures. Regarding the EP153R gene, one wild-type (Malawi isolate) and one Vero cell-adapted (Ba71V) ASFVs were evaluated after deletion of EP153R. In both cases, the deletion of EP153R did not affect the replication of either isolate in vitro [19,20]. Interestingly, the deletion of the EP153R gene in the context of ASFV-G-∆9GL/∆CD2v, producing the triple deletion gene mutant ASFV-G-∆9GL/∆CD2v/∆EP153R, significantly decreased the replicative abilities in swine macrophage cultures. It is not clear why the deletion of the EP153R gene affects the replication of ASFV-G-∆9GL/∆CD2v while it does not affect that of the Malawi or BA71V isolates [19,20]. Although swine red blood cell hemadsorption in infected cells has been shown to be solely dependent on the expression of the CD2v gene [18], it has been reported that the deletion of the EP153R gene from the BA71V genome affected CD2v expression and hemadsorption [20]. It is possible that CD2v and EP153R have differential and overlapping functions, explaining why simultaneous deletion severely affects virus replication.  The presence of viremia in the experimentally inoculated animals was quantified at different days post-inoculation in swine macrophage cell cultures. As expected, animals inoculated with 10 3 TCID50 of virulent parental ASFV-G-Δ9GL had relatively high virus titers in blood (as high as 10 6 TCID50/mL) until the day of challenge (Figure 4). Conversely, animals inoculated with mutants ASFV-G-Δ9GL/ΔCD2v or ASFV-G-Δ9GL/ΔCD2v/ΔEP153R had either very low or undetectable (test sensitivity is ≥10 1.8 TCID50/mL) virus titers in blood. Therefore, it appears that the in vivo replication of both ASFV-G-Δ9GL/ΔCD2v and ASFV-G-Δ9GL/ΔCD2v/ΔEP153R was severely affected by the gene deletions. This is interesting since the deletion of CD2v from the genome of virulent isolates Georgia and Malawi [11,18] only produced a delay in the appearance of viremia and a moderate reduction in virus titers. Furthermore, the deletion of EP153R in the context of the field isolate Malawi does not have any effect on the viremia kinetics in infected animals [19]. It is interesting to note that, at least in the case of the deletion of the CD2v gene, although virus replication is not affected when evaluated in vitro using swine macrophage cultures, it is severely altered during the infection in swine, even using a route of inoculation (IM) that evades several host innate immune barriers. It is possible that these observed differences could be due to unidentified host factors that affect virus replication. The presence of viremia in the experimentally inoculated animals was quantified at different days post-inoculation in swine macrophage cell cultures. As expected, animals inoculated with 10 3 TCID 50 of virulent parental ASFV-G-∆9GL had relatively high virus titers in blood (as high as 10 6 TCID 50 /mL) until the day of challenge (Figure 4). Conversely, animals inoculated with mutants ASFV-G-∆9GL/∆CD2v or ASFV-G-∆9GL/∆CD2v/∆EP153R had either very low or undetectable (test sensitivity is ≥10 1.8 TCID 50 /mL) virus titers in blood. Therefore, it appears that the in vivo replication of both ASFV-G-∆9GL/∆CD2v and ASFV-G-∆9GL/∆CD2v/∆EP153R was severely affected by the gene deletions. This is interesting since the deletion of CD2v from the genome of virulent isolates Georgia and Malawi [11,18] only produced a delay in the appearance of viremia and a moderate reduction in virus titers. Furthermore, the deletion of EP153R in the context of the field isolate Malawi does not have any effect on the viremia kinetics in infected animals [19]. It is interesting to note that, at least in the case of the deletion of the CD2v gene, although virus replication is not affected when evaluated in vitro using swine macrophage cultures, it is severely altered during the infection in swine, even using a route of inoculation (IM) that evades several host innate immune barriers. It is possible that these observed differences could be due to unidentified host factors that affect virus replication. Presence of anti-ASFV antibodies in the serum of animals inoculated with ASFV-G-Δ9GL, ASFV-G-Δ9GL/ΔCD2v, or ASFV-G-Δ9GL/ΔCD2v/ΔEP153R was detected at the time of challenge using an in-house developed direct ELISA (see Materials and Methods) ( Figure 5). While animals inoculated with ASFV-G-Δ9GL presented, at the time of challenge, an average antibody titer of log10 4.00 (SD ± 0.00) antibody titers in those inoculated with ASFV-G-Δ9GL/ΔCD2v or ASFV-G-Δ9GL/ΔCD2v/ΔEP153R were much lower (log10 2, SD ± 0.82 and log10 1.5, SD ± 0.58, respectively). As we previously reported, in our experience, the induction of virus-specific antibodies was directly related to the presence of the replication of all the attenuated viruses tested [3,4,6,7,9,10,21].  Presence of anti-ASFV antibodies in the serum of animals inoculated with ASFV-G-∆9GL, ASFV-G-∆9GL/∆CD2v, or ASFV-G-∆9GL/∆CD2v/∆EP153R was detected at the time of challenge using an in-house developed direct ELISA (see Materials and Methods) ( Figure 5). While animals inoculated with ASFV-G-∆9GL presented, at the time of challenge, an average antibody titer of log 10 4.00 (SD ± 0.00) antibody titers in those inoculated with ASFV-G-∆9GL/∆CD2v or ASFV-G-∆9GL/∆CD2v/∆EP153R were much lower (log 10 2, SD ± 0.82 and log 10 1.5, SD ± 0.58, respectively). As we previously reported, in our experience, the induction of virus-specific antibodies was directly related to the presence of the replication of all the attenuated viruses tested [3,4,6,7,9,10,21]. Presence of anti-ASFV antibodies in the serum of animals inoculated with ASFV-G-Δ9GL, ASFV-G-Δ9GL/ΔCD2v, or ASFV-G-Δ9GL/ΔCD2v/ΔEP153R was detected at the time of challenge using an in-house developed direct ELISA (see Materials and Methods) ( Figure 5). While animals inoculated with ASFV-G-Δ9GL presented, at the time of challenge, an average antibody titer of log10 4.00 (SD ± 0.00) antibody titers in those inoculated with ASFV-G-Δ9GL/ΔCD2v or ASFV-G-Δ9GL/ΔCD2v/ΔEP153R were much lower (log10 2, SD ± 0.82 and log10 1.5, SD ± 0.58, respectively). As we previously reported, in our experience, the induction of virus-specific antibodies was directly related to the presence of the replication of all the attenuated viruses tested [3,4,6,7,9,10,21].  In order to assess the effect of inoculation with ASFV-G-∆9GL/∆CD2v or ASFV-G-∆9GL/ ∆CD2v/∆EP153R on the induction of protection against disease, ASFV-G-∆9GL/∆CD2v-or ASFV-G-∆9GL/∆CD2v/∆EP153R-exposed animals were challenged with parental virulent ASFV-G, with their responses compared to those infected with ASFV-G-∆9GL. Animals previously inoculated IM at day 0 and 21 pi with 10 3 TCID 50 of either ASFV-G-∆9GL, ASFV-G-∆9GL/∆CD2v or ASFV-G-∆9GL/∆CD2v/∆EP153R were IM challenged at 35 dpi with 10 3 TCID 50 of virulent parental ASFV-G. Animals were monitored daily for clinical signs and changes in body temperature. In addition, four naïve animals that were challenged with parental ASFV-G using the same route and dose served as a control group. Animals inoculated with ASFV-G displayed ASF-related signs by 4-5 days post-challenge (dpc), evolving into a more severe disease in the following days and all animals being euthanized around 7 dpc (Table 1 as well as Figures 3 and 6). As previously described [5], all animals receiving ASFV-G-∆9GL survived the challenge without presenting any clinical signs related to ASF, not even a transitory rise of body temperature that is sometimes observed in protected animals by day 4-5 pc. Conversely, all animals exposed to ASFV-G-∆9GL/∆CD2v or ASFV-G-∆9GL/∆CD2v/∆EP153R succumbed to the challenge with the parental virulent virus. Clear statistical differences were shown between the animals exposed to ASFV-G-∆9GL and ASFV-G (p < 0.0002), and between ASFV-G-∆9GL and ASFV-G-∆9GL/∆CD2v or ASFV-G-∆9GL/∆CD2v/∆EP153R (p < 0.0001). The kinetics of the appearance of clinical signs, including a rise of the body temperature of ASFV-G-∆9GL/∆CD2v and ASFV-G-∆9GL/∆CD2v/∆EP153R challenged animals, were similar to those of the control animals challenged with ASFV-G although, interestingly, significant differences arose as a result of the accelerated mortality of the pigs infected with ASFV-G-∆9GL/∆CD2v and ASFV-G-∆9GL/∆CD2v/∆EP153R compared with the mock vaccinated group (p-0.0403). Table 1. Swine survival and fever response following infection with either ASFV-G-∆9GL, ASFV-G-∆9GL/∆CD2v or ASFV-G-∆9GL/∆CD2v/∆EP153R and challenge with ASFV-G 10 3 TCID 50 . Viremia kinetics after the challenge followed the presentation of clinical signs associated with disease. As expected, mock vaccinated animals presented very high virus titers in blood (as high as 10 8 -10 9 TCID50/mL) until the day they were euthanized ( Figure 4). Conversely, the animals exposed to ASFV-G-Δ9GL experienced a continuous decrease in their viremias until they were undetectable (test sensitivity is ≥10 1.8 TCID50/mL) by the end of the experimental period. Animals exposed to ASFV-G-Δ9GL/ΔCD2v or ASFV-G-Δ9GL/ΔCD2v/ΔEP153R had a rapid increase in viremia at the only time point determined (4 dpc) since the animals were euthanized around day 5-6 pc due to the severity of the clinical signs of the disease. Therefore, it appears that the presence of pre-challenge viremia in Figure 6. Evolution of mortality in animals (4 animals/group) IM infected with 10 3 TCID 50 of either ASFV-G-∆9GL, ASFV-G-∆9GL/∆CD2v, or ASFV-G-∆9GL/∆CD2v/∆EP153R and challenged IM with 10 3 TCID 50 of either ASFV-G. Comparison between the survival curves produced by groups of pigs infected with different viruses were assessed by the Log-rank (Mantel-Cox). Analysis was conducted on Graphpad Prism software version 8.2.1.
Viremia kinetics after the challenge followed the presentation of clinical signs associated with disease. As expected, mock vaccinated animals presented very high virus titers in blood (as high as 10 8 -10 9 TCID 50 /mL) until the day they were euthanized ( Figure 4). Conversely, the animals exposed to ASFV-G-∆9GL experienced a continuous decrease in their viremias until they were undetectable (test sensitivity is ≥10 1.8 TCID 50 /mL) by the end of the experimental period. Animals exposed to ASFV-G-∆9GL/∆CD2v or ASFV-G-∆9GL/∆CD2v/∆EP153R had a rapid increase in viremia at the only time point determined (4 dpc) since the animals were euthanized around day 5-6 pc due to the severity of the clinical signs of the disease. Therefore, it appears that the presence of pre-challenge viremia in animals inoculated with ASFV-G-∆9GL associates with protection and an apparent limited or null replication of the challenge virus.
In this regard, it should be mentioned that all attenuated vaccine candidates produced in our laboratory (ASFV-G-∆9GL, ASFV-G-∆MGF, ASFV-G-∆9GL/∆UK, and ASFV-G-∆I177L strains) [3][4][5][6] which efficiently protect animals against disease after the challenge with parental ASFV-G presented, at some point before the challenge, significant viremia values. Interestingly, attenuated strains which were unsuccessful in inducing protection (as ASFV-G-∆9GL/∆MGF, or ASFV-G-∆TK) [7,22] failed to induce significant viremias after their inoculation. Although the association between the induction of viremia after the administration of attenuated virus strains and protection at the challenge is clear, we cannot establish a clear cause-effect relationship between these phenomena. Host mechanisms mediating protection against ASFV infection or disease remain still largely unknown and the possible relationship between the replication of attenuated strains and the protection achieved by them should be the object of further study.
In light of the previous consideration, one possible explanation of the failure of ASFV-G-∆9GL/∆CD2v and ASFV-G-∆9GL/∆CD2v/∆EP153R to induce protection could be the low level of replication of these viruses after administration with the possible lack of adequate antigenic exposure, resulting in a failed immune response unable to limit the replication of the challenge virus. Both CD2 and EP153R antigens have been mentioned as important in the induction of a protective immune response [23]. The host immune mechanisms responsible for protection against ASF are still under discussion. In our experience, the only immune mechanism induced by attenuated virus strains that is clearly associated with protection against challenge is the presence of virus-specific circulating antibodies [9]. In all cases, including this report, there is a strong association between the presence of ASFV-specific circulating antibodies and protection against challenge [3][4][5]9].
Genomic manipulation of ASFV seeking to improve/modify the phenotypic characteristics of a certain strain is a process of difficult predictability. It has been reported that the deletion of a specific virus gene, even those highly conserved across isolates, may produce a different phenotypic change as described for genes 9GL, TK, NL, UK, and DP148R [5,8,10,22,[24][25][26][27][28]. These results suggest that the different phenotypic effects observed after deleting specific genes from two different ASFV genomes is strongly dependent on the genomic background of the viral isolate.

1.
There are previous reports providing comparable data as those reported here. The deletion of a group of genes (MGF360/505) from ASFV-G-∆9GL reduced its protective effect [7]. Similarly, Abrams et al., [29] showed that the deletion of the NL (DP71L) and UK (DP96R) genes from the genome of the naturally attenuated OUR T88/3 strain strongly reduced its ability to protect pigs against challenge with virulent virus.
It is not known what is the exact function of the AFSV CD2v and EP153R genes. CD2v has been described as responsible for mediating adhesion to swine red blood cells in the infected animals, probably facilitating the transport of the virus during infection [11,18]. In addition, the presence of CD2v produced decreased lymphoproliferative responses in swine peripheral blood mononuclear cells to specific white cell mitogens [18], although the immunomodulatory role of the protein remains to be shown. The interaction of CD2v with host protein AP-1 has been described [30] and believed to modulate vesicle transport, although the importance of that interaction during the processes of virus replication or virulence are currently unknown. Regarding EP153R, it has an inhibitory effect on the caspase-3 activation and the apoptosis induced both in ASFV-infected cells and in cell lines transfected, either stably or transiently, with the EP153R gene and is treated with different pro-apoptotic stimuli [31]. At this point, it is not clear how any of the putative functions ascribed to either CD2v or EP153R may influence the inhibitory effect induced by their deletion as described in this report.
The results presented here clearly indicate that the rational development of novel ASFV vaccines requires caution, avoiding the direct extrapolation of the results obtained by specific gene deletions obtained in a particular virus strain to novel field virus isolates. Moreover, it confirms that, although pursuing variations for specific qualities by genetically modifying ASFV is methodologically feasible, the outcome of those modifications is quite unpredictable and at each instance it will require a thorough experimental validation.