1. IntroductionAfrican Swine Fever (ASF) is a highly contagious, often lethal hemorrhagic disease of both domestic and wild suids. ASF virus (ASFV), a large double-stranded DNA arbovirus and the only member of the family Asfarviridae, is the causative agent of this disease, infecting mononuclear phagocytic cells. ASF was first characterized in Western scientific literature in Kenya in 1921 [1], it is currently endemic in almost two dozen African countries, and it has been reported in 32 countries on the African continent since 2005 [2]. ASFV infection follows a sylvatic cycle, transmitting between agricultural populations of hogs and wild reservoirs of warthogs and ticks of the genus Ornithodoros. Over the 20th Century, several outbreaks of ASF were reported in Europe, East Asia and the Dominican Republic with devastating economic impacts. In 2007, an accidental introduction of ASFV into the Republic of Georgia resulted in a wide-scale epidemic throughout eastern Europe and Russia of a highly virulent strain known as the Georgia strain (ASFV-G) [3]. This epidemic has since touched off a fast-moving outbreak in Southeast Asia in early 2019 [4,5] and an outbreak in the Dominican Republic in the summer of 2021, followed by Haiti reporting presence of the disease to the World Organization for Animal Health (OIE) a few months later [6]. This latest outbreak on the island of Hispaniola is the first time in 40 years that ASF has been documented in the Western hemisphere, elevating the disease to the level of a global pandemic.Due to the complex series of reservoirs for ASFV, eradication is not currently a feasible means of disease control. Instead, vaccines would be the preferred method of control, though a successful commercial vaccine has been, until recently, elusive. Recently, a live attenuated vaccine developed by introducing a single gene deletion of I177L in the currently circulating ASFV-G strain was found to be highly effective at preventing clinical disease in animals later challenged with wild type orthologous virus [7,8] as well as a currently circulating field strain in Southeast Asia [9]. However, the immunological correlates of protection are not perfectly understood for this or other naturally attenuated ASFV strains. Protection induced by passive transfer of hyperimmune serum from convalescent to naive animals [10] as well as in vitro naïve PBMCs [11] suggests a role for neutralizing antibodies in protection. Recently, studies of the immune response induced by naturally attenuated ASFV strains correlated protection with increased numbers of interferon (IFN)γ producing cells [12], although this correlation was not always observed [13,14]. Furthermore, depletion of CD8+ cells using specific antibodies diminished protection, demonstrating the importance of the cellular response in protection against ASF [15]. Besides, the innate immune response has also been described to take a part in protection against ASF. Leitão et al. showed how protection induced by attenuated NH/P68 isolate correlated with an early increase in Natural Killer (NK) cells activation [16]. On the other hand, although preliminary data indicated that ASFV developed mechanisms to escape dendritic cells (DC) defenses [17] little has been described about correlation between DCs and protection against ASF.

In the present study, ASFV-G-ΔI177L-vaccinated swine were challenged with the orthologous wild type virus at 28 dpi and samples were collected over the course of the experiment to assess humoral and cellular immune responses. Parameters were compared with baseline data and also with time-matched placebo control animals. Results indicate that not only does the ASFV-G-ΔI177L vaccine elicit a strong antibody response, but it also stimulates an antiviral (IFNγ) response among CD4+, CD8+, and CD4+CD8+ memory T cells along with NK cells and γδT cells.

3. DiscussionThe current study provides a detailed analysis of both the humoral and cellular immune response landscape following vaccination with a highly effective live attenuated ASF vaccine, along with the immune landscape following challenge with its wild-type homologue. Development of a highly effective vaccine against ASFV has long been an elusive goal, due in no small part to: a lack of consensus on what defines a protective immune response; the difficulties involved with studying a virus that requires a high level of biocontainment; and the challenges of working with a virus without any closely-related viral relatives except for a recently discovered putative relative, which infects abalone and is far less well-understood even than ASFV [29]. Our study confirms past studies out of our lab showing that this novel vaccine candidate elicits a strong humoral immune response within the first 7–11 days post vaccination [7,8]. In fact, serum anti-ASFV IgG titer among ASFV-G-ΔI177L-vaccinated swine in the current study plateaued by 11 dpi. In a related study with this vaccine candidate, Tran et al. observed 50% mortality among Vietnamese swine and ~30% mortality in European swine challenged with a locally circulating Vietnamese strain of ASFV-G at 14 dpi, while survival in both pig strains rose to 100% when challenged just 7 days later at 21 dpi, suggesting that high antibody titers may not be sufficient on their own to protect against wild-type challenge [9]. A 2015 study assessing the protective efficacy of an ASFV-G-Δ9GL vaccine candidate found that the numbers of IFNγ-producing cells, which rose between 8–28 dpi, were well-correlated with protection from homologous challenge in their vaccinated group [30]. Conversely, a 2017 study by the same group of a double-gene knockout vaccine candidate ASFV-G-Δ9GL/ΔUK found that survival of vaccinated swine was 100% when challenged at 14 dpi and did not seem to be correlated with T cell immunity [14]. However, the method used to assess this endpoint in both studies was an IFNγ ELISpot, which does not take cell type or memory polarization into account. In a 2016 study by Carlson et al. using an ASFV-Pretoriuskop/96/4-Δ9GL single gene deletion vaccine candidate, the authors found that survival increased in a stepwise fashion from 40% to 80% when vaccinated swine were later challenged at 7, 10, 14, 21 or 28 dpi with the parental strain [13]. The authors found no clear correlation at any challenge time point between survival and IFNγ-producing cells, as measured by IFNγ ELISpot, nor was there any strong correlation between survival and any of the measured parameters at the 7, 10 and 14 dpi challenge groups, while at the later challenge time points, 21 and 28 dpi, ASFV-specific antibodies became a more reliable correlate of protection. In the present study, while the proportions of both CD8+, CD4+ and CD4+CD8+ Tmem expressing IFNγ begin to rise to the level of significance by 14 dpi, in terms of Tmem orientation to either central or effector phenotype upon ex vivo restimulation, this response does not become significant in vaccinated swine until 28 dpi. Interestingly, we see a sharp spike in IFNγ MFI at 14 dpi among CD8+ Tcm as well as CD8+, CD4+ and CD4+CD8+ Tem. By 28 dpi, IFNγ MFI has returned to a non-significant level in all Tem populations, while remaining elevated, in CD4+ and CD8+ Tcm. Meanwhile, an anomalously high IFNγ-PE MFI among CD4+ Tcm from a control pig at 14 dpi along with another high IFNγ-PE MFI among CD4+CD8+ Tcm from two control swine at 28 dpi reflect noise in the data—the number of events falling within the IFNγ+ gate used to determine these 3 MFI was either 1 or 2 in each case, an indication of just how unresponsive T cells from control pigs were to ex vivo stimulation with ASFV-G. Taken together, this may suggest that it is not enough to have IFNγ-producing, ASFV-G-reactive T cells by 14 dpi, but that proper orientation of those T cells as either effector or central memory may play a role in protective immunity. It is important to mention here that even though vaccinated animals showed detectable levels of viremia peaking at between 7 and 14 days post-inoculation, viremia continued to decrease even after challenge, indicating that challenge virus was effectively neutralized at the inoculation site by the primed immune cells and/or circulating neutralizing antibodies. A study by Iyer et al. in 2009 found that a highly effective live attenuated recombinant vaccine against West Nile Virus (a virus that infects monocytes in humans) was also correlated with reduced CD62L expression in IFNγ+ CD8+ T cells prior to challenge [31]. A 2005 study of a live attenuated human immunodeficiency virus (HIV) vaccine candidate constructed by incorporating HIV gene p160 into the vaccinia virus sought to establish a more thorough phenotype of memory T cells associated with long-term proliferative capacity [32]. The authors found that while HIV-tetramer+ CD8+ T cells underwent an abrupt and transient downregulation of CD62L expression in the first several days following inoculation, expression levels of this marker continued to rise out to 250 dpi and were not correlated with the ability of these cells to proliferate or produce IFNγ or TNFα in response to in vitro stimulation, similar to the current study’s finding that IFNγ production capacity does not correlate with surface CD62L expression. Other viruses such as Ebola virus [33], HIV [34,35], and others, encode proteins that inhibit and modulate CD62L expression and ectodomain shedding as part of their host immune evasion strategy. It has been observed that the shedding of CD62L is necessary for CD8+ T cells to gain lytic activity, as assessed by CD107a expression [36]. Future studies of ASF vaccine candidates ought to measure the full polarization of memory type T cells in terms of both CD62L and CCR7 expression, and should consider not only IFNγ expression, but also CD107a, perforin, granzyme B and granulysin expression, as well as the direct cytolytic capacity of these memory cells over time.In the present study, a strong Type II IFN response was noted in NK cells by 14 dpi among ASFV-G-ΔI177L-vaccinated swine, and this response remained strong after challenge, suggesting that perhaps these cells play a role in developing antiviral immunity and in fighting infection. A positive correlation between NK cells and protection from ASFV has also been documented in a previous study of the attenuated strain ASFV/NH/P68 [16]. Conversely, a study using the virulent Malta 78 strain of ASFV demonstrated suppression of NK cell activity between 3 and 6 dpi [37], which is in agreement with the lack of NK cell IFNγ activity noted post-challenge in our mock-vaccinated swine. These results suggest that the ability of NK cells to participate in ASFV immunity is dependent upon the virulence of the strain used. In a study of NK cell activity in vitro using both a non-virulent and virulent strain, Martins and Leitão found that while the non-virulent NK/P68 ASFV strain stimulated NK cell activity, exposure of NK cells to the virulent Lisbon 60 strain suppressed their activity [38]. Taken together, all of these findings concord well with the results of the present study. On the other hand, it is surprising to observe such a strong NK cell response upon re-stimulation from 14 dpi onwards. It is possible, that NK cells from ASFV-G-ΔI177L-vaccinated swine are undergoing an increased non-specific response to a secondary stimulation known as “trained immunity”, as previously described for cytomegalovirus (another double stranded DNA virus) [39,40]. Future research should further explore this understudied innate immune cell and its role in vaccine-induced ASFV immunity.While several other studies have measured changes in percentages of γδT cells and Treg, to the best of our knowledge, the current study is the first to assess antigen-specific responsiveness of these cell subsets via IFNγ production by flow cytometric analysis. The proportion of γδT cells that were IFNγ+ increased significantly by 7 dpi in vaccinated swine, producing prolific amounts of the cytokine at what is a relatively early time point compared to αβT cells and NK cells. Hühr et al. also observed a significant decrease in the percentage of γδT cells in the blood by 7 dpc in response to the highly virulent strain ASFV-Armenia08 in commercial swine [27], while a study by the same group observed a significant, yet transient, increase in this population at 5 dpi in commercial swine in response to infection with the milder ASFV-Estonia2014 [28]. However, neither study measured Type II IFN response in these cells. On 7 dpi, there was a simultaneous spike in the relative abundance of Treg among vaccinated swine following ex vivo stimulation with ASFV-G. A similar spike in Treg was also observed in the 2020 paper by Hühr et al. in the blood and spleen of both domestic swine and wild boars, as well as the gastrohepatic lymph node of commercial swine at 7 dpc with the virulent ASFV-Armenia08 strain. A more recent paper by the same group revealed a similar increase in Treg by 10 dpc in the lungs of commercial swine and the spleen and lungs of wild boars following challenge with a strain less virulent to commercial swine ASFV-Estonia2014 [28]. Interestingly, we see no commensurate spike in Treg following challenge in our mock-vaccinated swine, again pointing to the potential importance of strain virulence in the magnitude and temporality of the cellular immune response to infection. Possibly this early spike in Treg cells functions to rein in and control the αβT cell, γδT cell and NK cell pro-inflammatory response against the virus. While there is not much in the virology literature regarding IFNγ+ Treg, the spike in induction of this cytokine at 14 dpi may be related to control and orientation of the virus response. In cancer literature, IFNγ induction and IL-10 suppression in Treg appears to be an early requirement for Treg to properly mediate an anti-tumor environment, clear more tolerant Treg from their surroundings and boost anti-tumor immunity [41,42]. In ASFV literature, detectable circulating levels of IL-10 are associated with a derailed immune response to ultimately fatal ASFV infection [24]. Given this information, measuring intracellular and circulating IFNγ and IL-10 may be useful in future studies of ASF vaccine candidates to assess the appropriateness of the immune response.Even though few changes were seen in either DC-like cells or monocyte numbers as a percentage of the parent population following ex vivo stimulation with ASFV-G, only few changes were seen with relative consistency in terms of MFI of the surface markers SLA-II and CD14. SLA-II, the swine MHC-II surface protein, was upregulated at every timepoint following ASFV-G-ΔI177L-inoculation among vaccinated swine (4–28 dpi), as well as at 4, 7 and 14 dpi in control swine (Supplemental Table S4). While the cause of the upregulation in control swine is unclear, the result of an upregulation of MHC-II in ASFV-G-ΔI177L-inoculated swine is increased antigen presentation to and activation of CD4+ T cells. At 14 and 28 dpi, we also observed a significant downregulation in the expression of CD14, an LPS receptor, compared to baseline among ASFV-G-ΔI177L-vaccinated swine, along with a significantly lower expression at 14 dpi compared to mock-vaccinated swine. A recent in vitro study found that infection of macrophages and monocytes in culture with both a tissue culture adapted avirulent strain ASFV-BA71V as well as a virulent Sardinian strain ASFV-22653/14 resulted in significantly reduced CD16 expression, consistent with our findings, and a lack of change in MHC-II expression on these cells, inconsistent with our findings [43]. In another in vitro study of adherent porcine bone marrow cells, infection with virulent ASFV-Benin97/1 was again correlated with downregulated surface expression of CD16 and unaltered MHC-II expression [44]. In a follow-up study, Franzoni et al. found that monocyte-derived macrophages (moMφ) infected in vitro with virulent ASFV-22653/14 had a reduced capacity to release IL-6, IL-12 and TNFα upon classical stimulation or stimulation with a TLR2 agonist [21]. While this effect was partially abrogated during infection with the attenuated strain ASFV-NH/P68, production of these cytokines was still impaired compared to mock-infected cells. Taken together, these results suggest that ASFV is replicating surreptitiously inside its target cells and evading immune surveillance early in infection. While the upregulation of SLA-II observed in the current study is encouraging, the downregulation of CD16 may contribute to a slower cell-mediated immune response to the vaccine strain and thus delay the onset of full, protective immunity. Future studies of genetically attenuated vaccine candidate strains should incorporate in vitro assessment of monocyte, moMφ, and Mφ disfunction into their suite of immunogenicity testing to ensure optimal and timely effector cell priming and activation downstream of these target APCs.Interestingly, while it has been observed that strong cytokine responses are necessary for eliciting an immune response against ASFV [24], this study found very little modulation of a variety of proinflammatory cytokines following immunization, possibly as a downstream result of the downregulation of CD16 observed in monocytes. While nonsignificant trends appear in the data such as a peak in serum TNFα at 7 dpi and a small elevation in IFNα among ASFV-G-ΔI177L at 4 dpi, the only significant increases in serum cytokines among vaccinated swine were at 4 and 7 dpi for IL-1Ra and closer to challenge, at 14 and 28 dpi for IL-12p40. While Type I IFNs and IL-12 are crucial co-stimulatory signals for promoting IFNγ production and cytotoxic capacity in NK and T cells [24], the timing and serum levels observed post-inoculation in the current study do not correlate well with the IFNγ responses we clearly see in our NK and T cells. For example, significantly elevated levels of IL-12p40 are observed coincident with elevations in intracellular IFNγ expression, rather than prior to. Curiously, while we observed strong Type II IFN responses upon ex vivo stimulation of NK and T cells, serum cytokine levels of IFNγ never rose above the limit of detection, despite using two separate commercially available detection kits. This finding, while curious, is similar to findings in a recent study by Wang et al. [20]. The Multigene Family (MGF) genes MGF360 and MGF505 play a role in downregulating Type I IFN release from infected monocytes and macrophages early in infection [23,45]. While studies of vaccine candidates involving deletions of the MGF genes demonstrate attenuation of the virus and protection against subsequent homologous challenge, IFN responses were either not measured [46] or were inconclusive [23]. While Reis and colleagues found that knocking out MGF360 and MGF505 from highly virulent ASFV-Benin resulted in increased mRNA levels of IFNβ in macrophages co-cultured with the virus, along with high serum concentrations of IFNγ in swine vaccinated with this candidate strain between 5 and 7 dpi, IFNγ ELISpot revealed low numbers of T and/or NK cells producing this cytokine at 46 dpi and only modest increases following challenge. In another recent study by Ran et al., the authors identified GC-enriched regions of the ASFV-NH/P68 genome as having a strong inverse correlation with IFNβ, IL-6 and TNFα expression, particularly the gene I267L, which they found inhibits the RIG-I pathway via interaction with Riplet [47]. Deletion of this gene yielded a vaccine candidate with 80% efficacy and significantly elevated serum IFNβ concentration at 5 dpc. In our study, we observed an increase in some of the analyzed cytokines directly correlated with disease in control swine following challenge. Conversely, vaccinated animals—none of which displayed clinical disease—generally showed no change or a slight decrease in serum concentrations of most cytokines following challenge.The current study represents the first time such an exhaustive and cell-type-specific analysis of IFNγ production following ex vivo stimulation has been reported for either a wild-type or an attenuated strain of ASFV. While we observe that high antibody titers correlate with protection, we also observe strong correlation with: skewing of memory T cells away from a central memory phenotype and towards an effector memory phenotype; an upregulation of IFNγ production in CD8+ memory T cells, NK cells and γδT cells; and a downregulation of IFNγ production in CD4+ Tem cells. These correlates of protection were largely maintained in the post-challenge period for ASFV-G-ΔI177L-vaccinated swine, while control swine showed no signs of an adaptive immune response following challenge. It is worth mentioning that although it has been previously demonstrated that the levels of specific antibody titers induced by vaccination are the same regardless of vaccine dose used [8], studying the effect of vaccine dosage on the stimulation of cellular immune response deserves further research. While this single gene deletion LAV vaccine provided perfect protection and strong correlates of protection measurements at the 28 dpi challenge over this and several other studies [7,8,9], future studies should assess how durable this immunity is many months after inoculation. Future work should also explore the mechanistic underpinnings of the findings observed here. For example, future studies may consider focusing on: the involvement of monocytes and DCs on initial reaction to attenuated vaccine strains; how early antigen presenting cells responses affect downstream memory T cells stimulation; or what epigenetic and metabolic changes occur in NK cells following inoculation that result in the exuberant IFNγ response following restimulation, among others.