**Fetal Lymphoid Organ Immune Responses to Transient and Persistent Infection with Bovine Viral Diarrhea Virus**

**Katie J. Knapek 1,2,**†**,**‡**, Hanah M. Georges 1,**†**, Hana Van Campen 1,2, Jeanette V. Bishop 1, Helle Bielefeldt-Ohmann 3, Natalia P. Smirnova 1,§ and Thomas R. Hansen 1,\***


Received: 5 June 2020; Accepted: 24 July 2020; Published: 28 July 2020

**Abstract:** Bovine Viral Diarrhea Virus (BVDV) fetal infections occur in two forms; persistent infection (PI) or transient infection (TI), depending on what stage of gestation the fetus is infected. Examination of lymphoid organs from both PI and TI fetuses reveals drastically different fetal responses, dependent upon the developmental stage of the fetal immune system. Total RNA was extracted from the thymuses and spleens of uninfected control, PI, and TI fetuses collected on day 190 of gestation to test the hypothesis that BVDV infection impairs the innate and adaptive immune response in the fetal thymus and spleen of both infection types. Transcripts of genes representing the innate immune response and adaptive immune response genes were assayed by Reverse Transcription quatitative PCR (RT-qPCR) (2−ΔΔCq; fold change). Genes of the innate immune response, interferon (IFN) inducible genes, antigen presentation to lymphocytes, and activation of B cells were downregulated in day 190 fetal PI thymuses compared to controls. In contrast, innate immune response genes were upregulated in TI fetal thymuses compared to controls and tended to be upregulated in TI fetal spleens. Genes associated with the innate immune system were not different in PI fetal spleens; however, adaptive immune system genes were downregulated, indicating that PI fetal BVDV infection has profound inhibitory effects on the expression of genes involved in the innate and adaptive immune response. The downregulation of these genes in lymphocytes and antigen-presenting cells in the developing thymus and spleen may explain the incomplete clearance of BVDV and the persistence of the virus in PI animals while the upregulation of the TI innate immune response indicates a more mature immune system, able to clear the virus.

**Keywords:** bovine viral diarrhea virus; fetus; thymus; immune response

#### **1. Introduction**

Bovine viral diarrhea viruses (BVDV) cause significant economic losses in all sectors of cattle production worldwide [1–4]. BVDVs are small single-stranded RNA viruses belonging to the Pestivirus genus in the family Flaviviridae [5]. Isolates of BVDV are classified into two genotypes, type 1 and type 2, with several subtypes and two biotypes: cytopathic (cp) and noncytopathic (ncp) [6]. Acute BVDV

infection of immunocompetent cattle results in diverse clinical presentations, including subclinical infection, fever, nasal and/or ocular discharge, pneumonia, severe systemic disease, hemorrhage, and peracute death [7]. Importantly, BVDV infection of pregnant cattle results in fetal infection and reproductive losses, including early embryonic death, abortion, and stillbirth [8–11]. If maternal infection with ncp BVDV occurs prior to 125 days of gestation, the fetus becomes persistently infected (PI) with the virus and is born without BVDV-specific antibodies [12,13]. PI animals will shed BVDV throughout life, serving as the main source of infection for other cattle. BVDV infection of pregnant cows after 150 days of gestation results in a transient infection (TI) of the fetus. These TI calves are born with BVDV-specific antibodies indicative of a functional adaptive immune response and clearance of the virus [9,14,15]. Virus-specific immune responses in the bovine fetus develops between days 125 and 150; therefore, BVDV infection during this time may result in either a PI or TI, depending on the individual fetuses [16]. The differences in the outcomes of PI and TI fetal infections have been attributed to the maturation and function of the fetal immune system at the time of infection [17].

At least two explanations have been tendered to explain the persistence of BVDV in PI cattle. First, ncp BVDV has been shown to inhibit type I interferon (IFN) induction in vitro through the actions of virally encoded RNase (Erns) and the N-terminal protease (Npro) (reviewed in [18–21]). The latter Npro degrades *interferon regulatory factor 3* (IRF3), thus inhibiting the transcription of IFNB and its antiviral activity in adjacent cells [22–24]. Erns binds to and degrades double-stranded RNA (dsRNA), preventing its binding to cells and induction of IFNs [25]. Inhibition of the IFN response by Npro and Erns allows BVDV to persist and replicate in cells. Conversely, in an in vivo model, mutations of these two viral genes affecting their functions negate persistent infection [26]. In another in vivo model using BVDV strain Pe515nc, fetuses were inoculated in utero (amniotic fluid) and collected on days 3, 5, and 7 post-inoculation [27]. BVDV was confirmed in fetal tissues; however, an innate immune response was not seen, suggesting an inhibition of the fetal innate immune response to BVDV [27]. A second explanation of BVDV persistence in fetal infections might be that BVDV does not inhibit the IFN response, but instead, another unknown mechanism is responsible for persistence. Previously, we observed that bovine fetuses whose dams were inoculated with ncp BVDV 96B2222 on day 75 of gestation had a peak in BVDV RNA in fetal blood on day 97 of gestation, 21 days post-maternal inoculation (dpmi), followed by a 10-fold decrease in viral RNA on days 192 and 245 of gestation, suggesting partial viral clearance by an active immune response in the PI fetus, or possibly a decrease in infected cell types/numbers [28–30]. Interferon stimulated genes (ISG), such as *ISG15*, *protein kinase RNA-activated* (*PKR*), as well as the RNA helicases: *DEAD Box Protein 58* (*DDX58*), also known as *retinoic acid-inducible gene* (*RIGI), melanoma di*ff*erentiation-associated protein 5* (*MDA5*); and *DExH-box helicase 58* (*DHX58*) were also shown to be chronically upregulated in the PI animal postnatally [31]. These results indicate that PI fetuses and placenta respond to BVDV with an innate immune response, albeit somewhat reduced compared to TI fetuses [28,29,31]. These findings indicate that the fetus responds to ncp BVDV infection with an innate immune response, much like the innate response to ncp BVDV infection in postnatal calves shown by Palomares et al. (2013); hence, inhibition of the innate immune response by viral proteins does not entirely explain viral persistence in vivo [32]. Another in vivo postnatal model used BVDV strain 11,249, which was previously shown to inhibit the IFN response in vitro [33]. The study revealed that in vivo, this BVDV strain does induce IFN responses [33]. The stimulation of an IFN response to BVDV has been exhibited in in vitro studies, in vivo fetal infection studies, and in vivo postnatal infections. Differences in these theories could be explained by differences in BVDV strains used and/or differences in experimental models. It is important to examine these differences and theories when considering the results of the present study.

An additional explanation for viral persistence is that the presence of BVDV during the development of the T cell repertoire permits its antigens to be accepted as "self" antigens, resulting in a state referred to as immunotolerance in which elements of the adaptive immune system do not respond to viral antigens and do not clear the virus from fetal tissues [34]. We hypothesize that ncp BVDV fetal infection early in gestation (<125 days) interferes with T cell development in the bovine fetal thymus during a critical period in which T cells are selected based on their recognition of self-antigens, affecting T cell response to BVDV in the spleen, resulting in immunotolerance to the virus and persistent infection in the calf. BVDV fetal infection later in gestation (>150 days) results in an active innate immune response, which clears the virus infection and causes a delay in antigen-specific T and B cell responses. To examine the effect of ncp BVDV infection on the bovine fetal thymus and spleen, transcripts of genes representative of the innate and adaptive immune response pathways were compared with control fetuses collected on day 190 of gestation.

#### **2. Materials and Methods**

#### *2.1. Animals*

Animal experiments were performed as previously described [28,34]. All animal experiments were approved by the Institutional Animal Care and Use Committees at the University of Wyoming, approval 05-265A-02 (19/10/2005) and 08-16A-01 (11/01/09). BVDV antigen negative and seronegative yearling Hereford heifers were synchronized for estrus (ovulation) and artificially inseminated with BVDV-free semen. Pregnancy was confirmed by ultrasound examination on days 35 to 40 and on day 70 of gestation.

#### *2.2. Experimental Design: BVDV Inoculation and Fetal Collections*

A power analysis was performed, and a group of 6 animals per treatment group (control, TI, and PI) was determined to be appropriate for a power of 1. Therefore, 18 unvaccinated pregnant heifers were randomly placed into treatment groups and inoculated intranasally with 2 mL culture media to generate sham-treated controls (*n* = 6) or with media containing 4.4 log10 TCID50/mL of ncp BVDV2 strain 96B2222 on day 75 of gestation to generate PI (*n* = 6) fetuses and on day 175 to generate TI (*n* = 6) fetuses [34]. Treatment groups were kept in widely separated pens and fed at the end of feeding rounds to minimize viral transmission. To capture the fetal immune response during maternal seroconversion (in the PI group), day 190 of gestation was chosen for fetal collections. Eighteen fetuses were collected by Cesarean section and necropsied on day 190 of gestation. Samples of fetal thymuses and spleens were frozen in liquid nitrogen and stored at −80 ◦C.

#### *2.3. RNA Extraction and RT-qPCR*

Total RNA from 70 mg of frozen thymus and 50 mg of frozen spleen was isolated using TRIzol reagent according to the manufacturer's instructions (Invitrogen ThermoFisher, Rockford, IL, USA). The isolated RNA was treated with DNase I (Qiagen, Germantown, MD, USA) and purified using the RNeasy MiniElute Cleanup Kit (Qiagen, Germantown, MD, USA). RNA concentration and 260/280 and 260/230 ratios were measured using the NanoDrop 1000 Spectrophotometer (ThermoScientific, Rockford, IL, USA). The primer sequences and gene accession numbers are listed in Table 1. One μg of RNA was reverse transcribed to synthesize cDNA using iScript™ Reverse Transcription Supermix (Bio-Rad, Hercules, CA, USA). Reverse transcription quantitative polymerase chain reaction (RT-qPCR) was performed with iQ™ SYBR® Green Supermix (Bio-Rad, Hercules, CA, USA). Each cDNA reaction was diluted (1:5 for thymus and 1:10 for spleen) with RNase-free water. Primers were used at 3 μM concentration. Each sample was assayed in duplicate wells on a 384-well plate. Four biological replicates of RT-qPCR plates were performed at one cycle of 95 ◦C for 3 min, 40 cycles of 95 ◦C for 30 s, 58 ◦C for 30 s, and 72 ◦C for 15 s with a final 5 min elongation in a LightCycler-480 Instrument (Roche, Basel, Switzerland). Upon completion of RT-qPCR, melting curve analysis was performed to assess the quality of amplification.


**Table 1.** Primers utilized for RT-qPCR.

*DDX58*: DExD/H-Box Helicase 58; *IRF7*: Interferon Regulatory Factor 7; *NFKB*: Nuclear Factor Kappa-Light-Chain -Enhancer of Activated B Cells; *IFNB*: Interferon Beta; *ISG15*: Interferon Stimulated Gene 15; *STAT4*: Signal Transducer and Activator of Transcription 4; *IFI30*: IFI30 Lysosomal Thiol Reductase; *PSMB9*: Proteasome 20S Subunit Beta 9; *PSMB8*: Proteasome 20S Subunit Beta 8; *B2M*: Beta-2-Microglobulin; *CXCL10*: C-X-C Motif Chemokine Ligand 10; *CXCL*: C-X-C Motif Chemokine Ligand 16. FW: Forward primer; REV: Reverse Primer.

#### *2.4. RT-qPCR Targets and Validation*

RT-qPCR targets for each tissue were chosen based on current knowledge and available tissue. Targets shared between the thymus and spleen were chosen to understand basic changes in both the innate and adaptive branches of the immune system. For the innate immune branch, targets chosen were *DDX58*, *NFKB*, *IRF7*, *IFNB*, and *ISG15*. For the adaptive branch, targets chosen were *PSMB9*, *IFI30*, *CD4*, *CD8A*, *CD8B*, and *CD79B*. Thymic tissue samples were more abundant than the spleen, allowing for additional targets to be studied in the thymus; *STAT1*, *IFI6*, *CXCL10*, *CXCL16*, *CXCR6*, *TAP1*, *B2M*, *CIITA*, and *CD46*. These results can be found in Table A1. A previous study which used a microarray to study BVDV-infected spleens indicated *STAT4* and *PSMB8* as targets of interest specific for splenic samples [35]. Therefore, both *STAT4* and *PSMB8* were additional targets for splenic samples in this study. BVDV viral RNA was also targeted to confirm tissue infection in TI and PI fetuses. These results can also be found in Table A1. Standard RT-qPCR primers were validated and evaluated based on MIQE guidelines [36].

#### *2.5. Morphogenesis of Thymus during Bovine Fetal Development and E*ff*ect of In Utero BVDV Infection*

Fetal thymus and spleen samples were fixed for 48 h in 10% neutral-buffered formalin, transferred into 70% ethanol, and paraffin-embedded. Four to 5 μm sections were cut and stained with eosin and hematoxylin (H&E). Microscopic assessment was performed with special attention to the components of the reticulo-endothelial network, myeloid, and lymphoid cells, including where the latter two cell types appear in the organ primordium. The relative cortical-to-medulla ratio was assessed, including full encirclement of the medulla by cortical thymocytes. Additionally, attention was paid to myoid cells and Hassall's corpuscles in terms of appearance, frequency, and size.

#### *2.6. Statistical Analysis*

RT-qPCR data are presented according to MIQE guidelines and statistically analyzed using the 2−ΔΔCq method [37]. Briefly, the Δ quantification cycle (Cq; also known as Ct) was calculated by subtracting the mean Cq of the reference gene (18S rRNA), within the treatment group, from the Cq of the target gene (same treatment group as the reference). The average of the Δ Cq for controls for the target gene were then subtracted from each infection treatment Δ Cq, for the same target gene as controls, to calculate the fold change (2−ΔΔCq). Statistical analysis of data obtained by RT-qPCR (2−ΔΔCq) was performed in GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA). The data were checked for normality using Shapiro-Wilks test; normally distributed data were analyzed with a one-way ANOVA and Dunnett's multiple comparisons test, while non-normally distributed data were analyzed with a Kruskal-Wallis test and Dunn's multiple comparisons test. Significant differences were at *p* < 0.05 and tendencies/trends were at *p* < 0.10. Data are presented as the mean ± SEM. Graphical data represents 2−ΔΔCq of which statistical analysis was run. In the text, fold change is presented for ease of understanding up- and down-regulation. For 2−ΔΔCq values > 1, fold changes were reported as averaged 2−ΔΔCq values. For 2−ΔΔCq values between 0 and 1 (down regulation), the negative inverse (−1/(2−ΔΔCq)) was calculated and reported as fold change.

#### **3. Results**

#### *3.1. Detection of BVDV RNA Expression in Thymus and Spleen and BVDV Receptor CD46 in Fetal Thymuses*

BVDV RNA was not detected in any of the control fetal thymus RNA samples and was present in 6 of 6 PI and 5 of 6 TI fetal thymic RNA samples with PI thymuses having 9.7- to 68.8-fold higher BVDV RNA concentrations compared to TI thymuses (Figure 1A). PI fetal spleens had a significantly increased amount of BVDV RNA (*p* < 0.001, 2−ΔΔCq = 3037-fold greater) compared to controls (2−ΔΔCq = 1.0) and very low amounts of BVDV RNA in TI fetal spleens (Figure 1B). The concentration of BVDV receptor *CD46* mRNA in PI fetal thymuses was decreased 6.2-fold (*p* < 0.001) and increased in TI fetal thymuses (*p* < 0.05) compared to controls on day 190 (Figure 1C).

**Figure 1.** Bovine Viral Diarrhea Virus (BVDV) expression in (**A**) thymus and (**B**) spleen samples (Control, TI, and PI). Expression of BVDV was significantly upregulated in PIs compared to controls in both thymic and splenic samples. Expression of BVDV in TIs was higher than controls, but not significant due to individual variation. (**C**) BVDV receptor CD46 expression in thymus (Control, TI, and PI). *CD46* expression was significantly down-regulated in PI thymuses. Control: Uninfected, blue; TI: Transiently Infected, yellow; PI: Persistently Infected, red. \* *p* < 0.05, \*\*\* *p* < 0.001.

#### *3.2. Thymic Responses*

#### 3.2.1. Innate Immune Responses in TI and PI Fetal Thymuses

In TI fetal thymuses, mRNA concentrations were significantly increased (*p* < 0.01) for the innate immune response genes *NFKB* (1.5-fold) and *IRF7* (4.5-fold) compared to controls on day 190 (15 dpmi). Neither *DDX58*, *IFNB*, nor *ISG15* were significantly different than the controls (Figure 2).

**Figure 2.** Control, TI, and PI thymus transcripts associated with the innate immune response. Interferon transcriptional regulators *NFKB* and *IRF7* were significantly upregulated in TI fetuses compared to controls. In PI fetal thymuses, only IFN transcriptional regulator *NFKB* and *IFNB* were down-regulated compared to control fetuses. TI: Transiently Infected, PI: Persistently Infected, \*\* *p* < 0.01, \*\*\*\* *p* < 0.0001.

In the PI fetal thymuses, the mRNAs of the innate response genes *NFKB* (−2.8-fold) and *IFNB* (6.0-fold) were significantly downregulated in PI thymuses compared to controls on day 190 (*p* < 0.01) (Figure 2); whereas *DDX58*, *IRF7*, and *ISG15* concentrations remained at control levels and were not significantly different between control and PI thymuses. A trending decrease in the type I IFN stimulated transcripts *STAT1* (−22.7-fold; Table A1) was found in PI fetal thymuses compared to controls (*p* < 0.10), which was consistent with the observed downregulation of *IFNB*. Type II IFN stimulated transcripts, transitional responses from the innate to adaptive immune response, *IFI16* (−11.3-fold, trend *p* < 0.10) and *CXCL16* (−22.4-fold, *p* < 0.05) were similarly down-regulated (Table A1). *CXCR6* mRNA concentration was not different between control and PI fetal thymuses (Table A1).

#### 3.2.2. Adaptive Immune Responses in TI and PI Fetal Thymuses

In TI fetal thymuses, the expression of *MHC I* mRNA *PSMB9* (Figure 3), *TAP1* (Table A1), and *B2M* (Table A1) were not significantly different from controls. T cell markers *CD8A*, *CD8B*, and *CD4* and activated B cell marker *CD79B* mRNAs were not significantly different in TI fetal thymuses compared to controls.

The relative expression of MHC I antigen presentation pathway genes *PSMB9* (−5.2-fold), *TAP1* (−5.5-fold), and *B2M* (−27.9-fold) transcripts were significantly decreased in PI fetal thymuses relative to controls on day 190 of gestation (*p* < 0.05) (Figure 3, Table A1 respectively). CTL co-receptors *CD8A* (−16.2-fold) and *CD8B* (−9.6-fold) mRNA concentrations were significantly decreased in PI fetal thymuses compared to the controls on day 190 of gestation (*p* < 0.05) (Figure 3). *CIITA* (−3.4-fold, trend *p* < 0.10) and *CD4* (−8.1-fold, *p* < 0.0001) mRNA associated with MHC II antigen presentation were significantly decreased in PI fetal thymuses on day 190 (Table A1 and Figure 3, respectively). Activated B cell marker, *CD79B*, mRNA concentration was significantly decreased (−3.8-fold) in PI fetal thymuses compared to control (*p* < 0.001) thymuses at day 190 of gestation (Figure 3).

**Figure 3.** Thymic (Control, TI, and PI) transcript expression associated with the adaptive immune response. No significant differences were found in TI fetal thymuses compared to controls. However, all adaptive genes, with the exception of *IFI30*, were significantly down-regulated in PI fetal thymuses compared to controls. TI: Transiently Infected, PI: Persistently Infected. \* *p* < 0.05, \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001.

3.2.3. Morphogenesis and Histology of the Thymus during Bovine Fetal Development and the Effect of in Utero BVDV Infection

There were no discernible differences between the control or PI fetuses with regard to the thymus morphology at day 190. The morphology of the fetal thymus had attained almost a neonatal appearance, with the cortex having expanded to completely surround the medulla, and the cortico-medullary ratio being ≥2. The medulla contained large, often complex Hassall's corpuscles, large numbers of myoid cells, and eosinophilic granulocytes. The latter could be seen throughout the medulla, but appeared in highest frequency in the cortico-medullary border zone. The myoid cells were large, round cells with a deeply eosinophilic cytoplasm, either homogenous or with finely fibrillar patterning. The almost euchromatic nucleus was round and centrally located, or more oval and acentric. A small amount of chromatin clumping along the nuclear membrane and occasionally a small nucleolus was present. These cells may, in some cases, be considered "single-cell" Hassall's corpuscles.

#### *3.3. Splenic Responses*

#### 3.3.1. Innate Immune Responses in TI and PI Fetal Spleens

*DDX58* (25-fold; *p* < 0.05), IRF7 (22-fold; *p* < 0.05), and ISG15 (160-fold; *p* < 0.05) mRNA were upregulated in TI fetal spleens compared to controls, whereas *NFKB* and *IFNB* mRNA concentrations remained at basal levels and were not significantly different (Figure 4). The innate immune response in PI fetal spleens was similar to controls with only *NFKB* being significantly downregulated in PI fetal spleens compared to controls (0.1532-fold, *p* < 0.05, Figure 4).

**Figure 4.** Spleen (Control, TI, and PI) expression of transcripts associated with the innate immune response. TI fetal spleens exhibited an upregulation of *DDX58*, *IRF7*, and *ISG15* compared to controls, indicating an active type I IFN response. In PI fetal spleens, only *NFKB* was downregulated compared to controls. TI: Transiently Infected, PI: Persistently Infected. \* *p* < 0.05.

3.3.2. Adaptive Immune Responses in TI and PI Fetal Spleens

Adaptive immune response genes, such as *STAT4* (Figure 5), *PSMB9* (MHC class I, Table A1), and *IFI30* (MHC class II; Figure 5) were significantly upregulated in TI spleens compared to controls (5.54-fold, *p* < 0.01; 12.9-fold, *p* < 0.05 respectively); however, *PSMB8* (MHC class I) was not significantly different between the two groups (Figure 5). T and B lymphocyte markers, *CD4, CD8A, CD8B*, and *CD79B*, were not different in TI compared to the control spleen (Figure 5).

**Figure 5.** Splenic (Control, TI, and PI) expression of transcripts associated with the adaptive immune response. In TI fetal spleens, only *PSMB9* and *IFI30*, associated with antigen presentation, were significantly different to the controls. In PI fetal spleens, T cell markers *CD4* and *CD8A* were significantly downregulated compared to controls. TI: Transiently Infected, PI: Persistently Infected. \* *p* < 0.05, \*\* *p* < 0.01.

In PI fetal spleens, *STAT4* (Table A1), *PSMB8*/*9* (Table A1 and Figure 5, respectively), and *IFI30* (Figure 5) were not differentially expressed compared to control fetal spleens. PI fetal spleens exhibited a downregulation of T lymphocyte markers mRNA (*CD4*, and *CD8B*) compared to controls (0.1465-fold, *p* < 0.5; 0.4005-fold, *p* < 0.05, respectively) (Figure 5) but T lymphocyte marker *CD8A* and B lymphocyte marker *CD79B* were not significantly different between the two groups.

#### **4. Discussion**

#### *4.1. Transient Infection on Day 175 of Gestation Elicits an Immunocompetent Response in the Bovine Fetal Thymus and Spleen on Day 190*

BVDV takes approximately 7–14 days to cross the placenta and infect the fetus following intranasal inoculation of the dam [38]. The TI fetal spleens and thymuses collected at day 190 of gestation, 15 days post-maternal infection were positive for BVDV RNA, indicating that the fetuses were infected for 1–7 days, which is sufficient time to have elicited an innate immune response in all TI fetuses and initial adaptive responses in some TI fetuses.

The rudimentary thymus is first detected around day 27 of gestation in ruminants with colonization and lymphoid development occurring at day 42 of gestation. Day 42 progenitor cells come from the fetal liver until the bone marrow develops and matures enough (around day 70) to take over as the primary source of thymocytes [39]. The bovine fetal thymus contains a discernable cortico-medullary delineation by day 42, Hassall's corpuscles by day 65, and IgM positive cells by day 70 to 100 [40]. In addition to T lymphocytes and prothymocytes of T cell lineage, the postnatal thymus contains thymic epithelial cells, myoid cells, fibroblasts, dendritic cells, macrophages and miscellaneous neutrophils, eosinophils and B cells [41,42]. These cell types would be present in the thymus at the time of maternal inoculation with BVDV (75 dpmi and 175 dpmi) and subsequent fetal infection in this experimental model. In utero BVDV infections have been shown to result in thymic hypoplasia in bovine fetuses characterized by marked lymphocyte depletion in the cortex, accompanied by hypertrophy of the epithelial stroma and infiltration by monocytes and mature macrophages with BVDV antigens primarily localized to macrophages [43–45].

The innate immune gene transcripts, *NFKB*, and *IRF7* were significantly upregulated in TI fetal thymuses indicative of a robust activation of the innate immune response to BVDV, which agrees with previous findings [34,35,38]. Significant changes were not observed in adaptive response genes, including the T cell marker transcripts, *CD8A*, *CD8B*, *CD4*, or B cell activation marker *CD79B* compared to controls, as expected due to the short course between fetal infection and tissue collection. In addition, the thymus is a primary lymphoid organ responsible for the selection and maturation of T cells but with limited antigen-specific immune responses. The decreased BVDV RNA concentrations in TI thymuses compared to PI thymuses, however, is supportive evidence that TI fetuses were able to mount an effective immune reaction against the virus. Despite the ability of TI fetuses to clear the virus, BVDV fetal infection during day 175–190 of gestation occurs during a critical stage of thymus and spleen development, as well as T cell selection and maturation in the thymus. BVDV fetal infection during this stage of fetal development may alter the animal's ability to fight other infections postnatally.

The spleen is a secondary lymphoid organ responsible for rapid immune responses to bloodborne pathogens like BVDV. During embryonic development, mesenchymal progenitor cells differentiate into marginal reticular cells, fibroblastic reticular cells, and follicular dendritic cells (reviewed in [46]). These cells will contribute to the spleen stroma and begin to form the white pulp, a lymphocyte-rich area which is the site of antigen-specific immune responses to blood-borne pathogens [46]. White pulp establishment occurs at approximately 15 weeks of gestation in humans, assumed to be a similar time point for cattle [40,46]. At this early time point, few B cells cluster around splenic arterioles, and 3 weeks later, are joined by T cells [46]. The T cells form the periarteriolar sheath (PALS), while the B cells form a follicle around the PALS [46]. Lymphocyte infiltration is believed to occur well after persistent infection with BVDV. However, TI fetuses were infected following the establishment of the

white pulp and lymphocyte infiltration of the spleen. This may play a large role in the transient and persistent infection of BVDV in fetuses, as discussed later.

In TI fetal spleens, the cytoplasmic sensor mRNA *DDX58*, transcriptional factor *IRF7*, and interferon-stimulated gene *ISG15* were upregulated indicating the detection of BVDV RNA by the immune system of TI fetuses and activation of the innate immune response. Transcripts for genes associated with antigen presentation, *PSMB9* and *IFI30*, were upregulated in the TI fetal spleen, suggesting that the antigen processing mechanisms of cells had been stimulated. However, transcripts for T and B cell markers were not increased as viral peptide presentation in the context of MHC I and MHC II would not have stimulated the proliferation of T and B cells in the short time period between fetal infection and tissue collection. In addition, T and B cell activation may also have been delayed in the TI fetuses. BVDV is known to inhibit lymphocyte activation, resulting in delayed T and B cell responses during acute infections of postnatal animals [47,48]. The low amounts of BVDV RNA detected in the TI fetal spleen compared to PI fetal spleens suggest the innate immune mechanisms effectively inhibited viral replication prior to the elaboration of an adaptive immune response. Additionally, the fetal spleen is assumed to have both T and B cells on day 105 of gestation. It can be assumed these lymphocytes in the spleen, once activated, contribute to the clearance of the virus, seen in TI postnatal animals [9,15].

#### *4.2. PI Thymus and Spleen Exhibit Diminished or Inhibited Immune Responses at Day 190 of Gestation*

The mechanism of BVDV PI is controversial. Some of the first in vitro studies on BVDV infection concluded that BVDV infection blocked the host's interferon response contributing to the persistence of BVDV in fetuses infected before day 120 of gestation [21,23,24]. In a paper published in 2001, the authors infected fetuses in utero (through amniotic fluid) on day 60 of gestation and collected fetuses 3, 5, and 7 days post-maternal inoculation [27]. It was concluded that ncp BVDV inhibits the fetal IFN response [27]. Subsequently, using a model of maternal infection, the fetal IFN response was stimulated by BVDV 22 days post-maternal infection [28,31,34,35,38]. However, it is important to note the differences in BVDV strains used and the differences in experimental models. This current study follows several others with similar experimental models and the same BVDV strain that have demonstrated an active innate immune response in PI fetuses soon after the establishment of fetal infection. The discussion below highlights these previous findings [28,31,34,35,38].

Previously, it was established that the early gestation bovine fetus (day 75) could respond to ncp BVDV infection with an innate immune response indicated by the upregulation of cytosolic dsRNA sensors in fetal blood and spleen, type I IFNs in cotyledons, and ISGs in fetal tissues, including the liver and spleen [28,31,38]. Moreover, the IFNG protein, an important bridge between the innate and the adaptive immune response, was detected in the amniotic fluid and fetal blood, and *IFNG* mRNA was upregulated in liver, spleen, and thymus of PI fetuses 14–22 dpmi (89–97 days of gestation) [34]. The activation of these genes corresponded with a reduction in BVDV titer in PI fetal blood at day 190 and 245, suggesting partial clearance of the virus by an active innate immune response; however, BVDV continued to be present in fetal tissues, albeit at a reduced level (reviewed in [14,28,34]). We hypothesized that the persistence of ncp BVDV was due to exhaustion of the initial innate immune response, and impaired steps in the adaptive immune response leading to incomplete viral clearance in the PI fetus. In support of this hypothesis, two innate immune response mRNAs, *NFKB* and *IFNB*, the bridging mRNA *CXCL16*, and the adaptive immune response mRNAs, *PSMB9, TAP1, B2M, CD8A, CD8B, CD4*, and *CD79B*, were found to be significantly decreased on day 190 in PI fetal thymuses compared to controls, and T cell marker mRNAs (*CD4* and *CD8B*) were significantly downregulated in the PI spleen. However, as discussed below, it is believed attenuation of the immune system may be caused by Treg development, instead of our original hypothesis of immune system exhaustion [35,49–51].

*DDX58* binds to and recognizes viral RNA in the innate immune system pathway. A reduction in *DDX58* concentrations could negatively impact the ability of fetal cells to respond to BVDV RNA and subsequently impair the induction of IFN antiviral activities. Persistent inhibition of NFKB has been associated with inappropriate immune cell development or delayed cell growth. For example, mice null for NFKB signaling molecules have impaired thymic medullary epithelial cell formation and autoimmune disease [52]. The transcription factor NFKB is both induced by type 1 IFNs and is a transcriptional regulator of cytokine expression, including IFNB and IFNG [53]. The decreased *IFNB* mRNA observed in the PI fetal thymus on day 190 coincided with the decreased expression of *NFKB* and the IFNI-inducible gene mRNAs, *ISG15* and *CXCL16*. Reduced *IFNB* mRNA may be explained by a paucity of dendritic cells in the PI fetal thymus at this stage of development; however, differences in the dendritic cell population between PI and control thymuses collected on day 190 were not observed histologically. IFI16 is a cytosolic sensor for viral RNA/DNA, which transcriptionally regulates the expression of type I IFNs in a positive manner [54]. The trended downregulation of *IFI16* observed in the PI thymus could explain the decreased expression of *IFNB*.

The PI fetal thymus exhibited differences in the innate immune response genes at day 190 compared to those previously observed in the fetal spleen and blood. *DDX58, MDA5*, and *ISG15* were previously observed to be increased in PI fetal blood from day 97 to day 192, and *ISG15* was increased in the PI fetal spleen, bone marrow, caruncle, and cotyledon, reflecting an active innate response in these tissues [28,31]. In contrast, *NFKB* and *IFNB* were significantly downregulated in PI fetal thymuses on day 190, but only *NFKB* was downregulated in spleens. A previous study by our group indicated a significant down-regulation of *IFNB* in day 190 PI fetal spleens, but no significant change in *NFKB* [35]. These studies were independent, and although results differ slightly, they both indicate an inhibition of the innate immune response 115 days post-maternal infection. Regulatory T cells (Tregs) originate in the thymus during fetal development and create tolerance of self in the developing fetus to prevent autoimmune disorders [49,50]. Due to the fetal BVDV infection occurring early in fetal development, prior to lymphocyte infiltration in the spleen, it is possible the Treg cells are creating tolerance to the BVDV viral antigen, allowing the virus to continue replication without an immune response from the host [49,50]. The development of Tregs in the thymus could be contributing to the difference in immune gene expression between the thymus and spleen [49,50]. On day 190 of gestation in PI animals, Tregs may be inhibiting an immune response in the thymus first before the Treg cells migrate to the periphery and reach the spleen.

Thymic concentrations of adaptive immune response transcripts, such as MHC I antigen presentation (*PSMB9, TAP1*, and *B2M*), CTL co-receptors *CD8A* and *CD8B*, and T cell marker CD4 were also decreased in PI fetal thymuses compared to controls consistent with an impaired adaptive immune response. The reduction in antigen presentation gene transcripts in PI fetal thymuses were associated with reduced *CD79B*, a component of antigen recognition and activation in B cells. In this study, *PSMB9* expression in PI thymuses was decreased at day 190 of gestation, suggesting a reduction in the normal function of the immunoproteasome [55]. Following proteolysis of viral proteins by the immunoproteasome, TAP1 transports peptides to the endoplasmic reticulum (ER) where peptides bind to MHC class I molecules for presentation on the cell membrane. Decreased *TAP1* concentrations could negatively impact antigen presentation to immune cells. In addition, *B2M* mRNA in PI thymuses was significantly decreased at day 190 compared to controls. The B2M protein assembles with the MHC I molecule in the ER to form a stable MHC I complex necessary for antigen binding. When B2M is not present, MHC I remains in the ER, and the MHC I complex is not expressed on the cell surface [56]. Decreased expression of *B2M* would result in additional dysfunction of the endogenous viral antigen presentation mechanism. In PI fetal spleens, *PSMB9* and *IFI30* were not significantly different compared to controls. This contradicts a previous study in which PSMB9 and IFI30 were significantly downregulated in day 190 PI spleens compared to controls [35]. This difference between the studies reflect increased individual variability between controls in the present study or possibly due to slight variation in infection/fetal collection times. The difference in the expression of antigen presentation mRNAs between the spleen and thymus may be indicative of APC and lymphocyte migration from the primary lymphoid organ, that is, the thymus, to secondary lymphoid organs, such as the spleen. In secondary lymphoid organs, such as the spleen, APCs reside and actively survey the blood and lymph for antigens; thus, the secondary lymphoid organs will have higher gene expression of mature APCs and lymphocyte markers [57].

Cytotoxic T cells recognize peptides presented in the context of the MHC I and B2M complex on target cells by the T cell receptor (TCR) and co-receptor, CD8. Antigen recognition by the CD8-TCR complex triggers the release of perforin, granzymes and cytokines from the lytic granules of CTL. *CD8A* and *CD8B* mRNA concentrations in PI fetal thymuses and *CD8B* in spleens were significantly decreased at day 190, suggesting either a reduced number of CTLs or decreased CD8A/B expressed per cell. Similarly, *CD4*, the co-receptor of the T cell receptor on T helper cells, mRNA was decreased in PI fetal thymuses and spleens at day 190 compared to controls and may reflect a decrease in T helper cells or the number of co-receptors per T helper cell. Supporting the results in a previous study, a reduction in BVDV-specific CTLs, T helper cells, or in antigen recognition would negatively impact the clearance of BVDV-infected cells in the thymus, spleen, and other fetal tissues, thus contributing to viral persistence [35].

The B-cell antigen receptor complex-associated protein Beta chain, *CD79B*, is part of the B lymphocyte antigen receptor complex and is required for T cell-dependent activation of B cells. In cooperation with CD79A, CD79B activates B cells through the interaction with CD4 T cells. After this interaction, B cells then have the capability to proliferate, secrete antibodies, cytokines, and chemokines, and initiate memory cell formation [58]. CD79B gene expression was decreased in PI thymuses on day 190 compared to controls, but were not downregulated in PI fetal spleens. However, a previous study by our group did find a decrease in CD79B expression in PI fetal spleens [35]. This difference may be due to individual variation of animals in the current study or slight differences in infection/collection timing. A sustained decrease in CD79B expression could result in the depletion of B cells later in gestation or postnatally, contributing to the overall deficit in adaptive immunity, failed viral clearance, and BVDV persistence. The outcome would be apparent immunotolerance to BVDV and the inability of PI animals to combat secondary viral and bacterial infections, such as bovine respiratory disease in postnatal life.

CD46 has been shown to be a receptor for BVDV entry [59,60]. CD46 is a regulator of complement activation through its interaction with plasma serine protease factor I, which cleaves complement factors C3b and C4b to prevent cell injury from complement attack. The concentration of *CD46* was significantly decreased in PI fetal thymuses compared to controls at day 190. The reduced concentration of *CD46* in the PI thymuses may reflect a general diminishment of constitutively expressed cellular mRNAs due to viral mechanisms that usurp cellular transcriptional machinery in favor of the production of viral proteins [61]. Alternatively, expression of the BVDV receptor might be inhibited in response to the binding of BVDV to its receptor and the presence of high viral titers during persistent infection. The downregulation of *CD46* in PI thymuses might also reduce the regulatory activity of CD46 on the complement system and increase the vulnerability of host cells to complement-mediated injury. A soluble form of CD46 which neutralizes BVDV in vitro is present in adult bovine plasma and may play a role in viral clearance in postnatal acute infections. However, soluble CD46 is not found in fetal or neonatal serum, and is therefore unlikely to impact PI fetal BVDV infections [62].

The comparison of TI and PI fetal immune responses in developing lymphoid organs has not been previously examined in experimental settings. Unfortunately, bovine-specific antibodies for the genes of interest are unavailable, and limited this study to RNA expression only. The inability to obtain protein data does limit interpretation. However, the results of this transcriptome study exhibit a drastic difference in immune response competence and maturity between PI and TI fetuses. The upregulation of the innate immune response of TI lymphoid organs, along with the inability of BVDV levels to be significantly different than controls indicate that the TI fetus is effectively fighting the virus, unlike the PI fetus. The timepoint of TI fetal collections may have been too early to capture the activation of the adaptive immune response, or the virus is possibly able to temporarily inhibit the adaptive immune response. It is further hypothesized that although the TI fetuses are able to respond appropriately to the virus, this viral insult during fetal development epigenetically alters the TI animals' DNA and causes further immune-related issues postnatally. This will require further studies postnatally.

#### **5. Conclusions**

Although animals infected with BVDV in utero may be born morphologically normal, ncp BVDV fetal infection has significant effects on the expression of innate and adaptive immune responses. In TI animals infected late in gestation, the immune system is developed enough to fight and clear the virus prior to or following parturition. We interpret these RT-qPCR data to reflect an active and upregulated innate immune response in TI thymuses and spleens, seen as an increase in type I IFN transcriptional regulators (Figure 6). In the case of PI fetuses, the virus infects the fetus prior to complete immune system development, ultimately causing drastic attenuation of the immune system, possibly due to the fetus identifying the virus as "self" (Figure 6). Specifically, the downregulation of genes, including type I IFN transcriptional regulators, antigen presentation, and T cell markers in PI fetal tissues was found. These trends in TI and PI fetuses were seen in both the thymic and splenic tissues, suggesting that these changes occur simultaneously in tissues critical for immune system education and development (Figure 6). These types of changes are believed to affect the infected animals postnatally, possibly through epigenetic changes.

**Figure 6.** *Cont.*

**Figure 6.** Graphic summarizing significant changes in fetal immune responses based on type of infection and tissue, as interpreted by RT-qPCR analysis. BVDV transient infection represents a fetal infection after gestational day 160 in which the fetus is able to clear the virus. Trends include an upregulation in genes associated with the innate immune response. Day 190 of collections may have been too early to capture an adaptive immune response in TI fetuses. However, we hypothesize that these pathways contribute to BVDV clearance in TI animals. BVDV-persistent infection represents a fetal infection before gestational day 120 in which the fetus is unable to clear the virus. Trends include a drastic down-regulation of the innate and adaptive branches of the immune system, and may lead to immunotolerance of the fetus to BVDV. In future studies, TI BVDV clearance and PI immunotolerance will be studied further with protein targets and flow cytometry. Green text, up arrow left of text, and listed genes represent pathways and genes which were significantly upregulated, as determined by RT-qPCR. Red text, down arrow left of text, and listed genes represent pathways and genes which were significantly downregulated as determined by RT-qPCR.

**Author Contributions:** Conceptualization, K.J.K. and T.R.H.; methodology, N.P.S., K.J.K.; validation, K.J.K., H.M.G., N.P.S.; formal analysis K.J.K., H.M.G., H.V.C.; investigation, K.J.K., H.M.G., H.V.C., H.B.-O.; resources, J.V.B.; writing—original draft preparation, K.J.K., H.M.G.; writing—review and editing, H.V.C., H.B.-O., T.R.H.; visualization, K.J.K., H.M.G.; supervision, J.V.B., H.B.-O.; project administration, T.R.H.; funding acquisition, T.R.H., H.M.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by USDA AFRI NIFA Predoctoral Fellowship 2019-67011-29539/1019321 to H.M.G.; AFRI NIFA National Needs Fellowships Grant Program 2016-38420-25289, USDA-AFRI Grants 2004-35204-17005 and 2008-35204-04652, and W3112 Reproduction in Domestic Ruminants HATCH project #1011648 from the USDA National Institute of Food and Agriculture to T.R.H.

**Acknowledgments:** The authors would like to thank Ann Hess for her assistance with statistical analysis.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **Appendix A**


**Table A1.** Table of genes and their expression in respective tissues. These genes were not measured in both tissues, only one tissue.

SEM: Standard Error Mean; TI: Transient Infection; PI: Persistent Infection; C: Control; STAT1: Signal Transducer and Activator of Transcription 1; IFI6: Interferon Alpha Inducible Protein 6; CXCL10: C-X-C Motif Chemokine Ligand 10, CXCL16: C-X-C Motif Chemokine Ligand 16; TAP1: Transporter 1 ATP Binding Cassette Subfamily B Member; B2M: Beta-2-Microglobulin; CIITA: Class II Major Histocompatibility Complex Transactivator; STAT4: Signal Transducer and Activator of Transcription 4; PSMB8: Proteasome 20S Subunit Beta 8.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **The E**ff**ect of Bovine Viral Diarrhea Virus (BVDV) Strains and the Corresponding Infected-Macrophages' Supernatant on Macrophage Inflammatory Function and Lymphocyte Apoptosis**

**Karim Abdelsalam 1,2,\*, Mrigendra Rajput 1, Gamal Elmowalid 2, Jacob Sobraske 1, Neelu Thakur 1, Hossam Abdallah 2, Ahmed A. H. Ali <sup>2</sup> and Christopher C. L. Chase 1,\***


Received: 4 May 2020; Accepted: 23 June 2020; Published: 29 June 2020

**Abstract:** Bovine viral diarrhea virus (BVDV) is an important viral disease of cattle that causes immune dysfunction. Macrophages are the key cells for the initiation of the innate immunity and play an important role in viral pathogenesis. In this in vitro study, we studied the effect of the supernatant of BVDV-infected macrophage on immune dysfunction. We infected bovine monocyte-derived macrophages (MDM) with high or low virulence strains of BVDV. The supernatant recovered from BVDV-infected MDM was used to examine the functional activity and surface marker expression of normal macrophages as well as lymphocyte apoptosis. Supernatants from the highly virulent 1373-infected MDM reduced phagocytosis, bactericidal activity and downregulated MHC II and CD14 expression of macrophages. Supernatants from 1373-infected MDM induced apoptosis in MDBK cells, lymphocytes or BL-3 cells. By protein electrophoresis, several protein bands were unique for high-virulence, 1373-infected MDM supernatant. There was no significant difference in the apoptosis-related cytokine mRNA (IL-1beta, IL-6 and TNF-a) of infected MDM. These data suggest that BVDV has an indirect negative effect on macrophage functions that is strain-specific. Further studies are required to determine the identity and mechanism of action of these virulence factors present in the supernatant of the infected macrophages.

**Keywords:** bovine viral diarrhea virus; cytopathic BVDV; immunosuppression; lymphocyte apoptosis; monocyte-derived macrophages; non-cytopathic BVDV

#### **1. Introduction**

Bovine viral diarrhea virus (BVDV) causes immune dysfunction in cattle and other ruminants, resulting in severe economic losses. The immune dysfunction associated with BVDV is believed to be, in part, a consequence of lymphoid depletion that could be mild or severe based on the virulence of the BVDV strains [1–4]. Furthermore, BVDV induces multiple abnormalities to immune cell function and immune mediators that could potentiate viral infection and pathogenesis [5,6]. BVDV infects neutrophils and decreases their microbicidal activity [7]. It also infects alveolar macrophage and decreases the expression of complement Fc receptors and chemokine production, reducing their ability to engulf opsonized pathogens [8].

Although only cytopathogenic (Cp) strains can induce direct apoptosis to peripheral blood mononuclear cells (PBMCs) [9,10], other studies have suggested that Ncp BVDV induces apoptosis in vitro in PBMCs isolated from BVDV-infected animals [11]. Ncp BVDV causes the most severe

BVDV immune dysfunction and is associated with severe lymphoid depletion [2,3,12–15]. Some Ncp BVDV strains have been categorized under a new biotype, lymphocytopathogenic, as they can induce apoptosis in lymphocytes while not effecting epithelial cells [3,16]. In addition, Ncp BVDV has been reported to induce apoptosis in vitro in PBMCs isolated from BVDV-infected animals [11].

Factors associated with Cp BVDV-induced apoptosis include: (i) oxidative stress and (ii) soluble factor(s) produced by monocytes that induced apoptosis in uninfected cells after lipopolysaccharide (LPS) treatment, [9,10,17,18]. However, the exact mechanism by which Ncp/lymphocytopathogenic BVDV induced the apoptotic effect is not well understood. Several studies suggest the role of macrophages in lymphoid depletion and immune dysfunction associated with BVDV. These studies demonstrated the importance of monocytes/macrophage in lymphocyte apoptosis in a mixed population of PBMCs [9,13]. However, the role of apoptotic-associated cytokines, which are important for classic swine fever virus (CSFV)-induced lymphocyte apoptosis, do not appear to be important for BVDV [13]. BVDV and CSFV are classified under the same genus: *Pestivirus*, Family: *Flavivirdae.* It has also been hypothesized that the *Pestivirus*' specific secreted Erns glycoprotein could play a role in the apoptosis mechanism associated with CSFV [19,20].

Immune mediators potentiate viral infection and pathogenesis. CSFV [21], and Cp BVDV [17], prime uninfected cells for activation-induced apoptosis. The mechanism for activation-induced apoptosis is virus-specific, where direct contact between infected and uninfected cells may also be involved in CSFV [21]. CSFV infection of bone marrow hematopoietic cells does not induce apoptosis in the majority of infected cells but does cause apoptosis in the uninfected cells, suggesting an apoptotic role for soluble factors released from infected cells. Cp BVDV induces apoptosis, while Ncp BVDV inhibits apoptosis [9,10,17,18,22,23].

Cp BVDV induces apoptosis in epithlial cellls in cell culture and in PBMCs, making it different from Lymphocytopathogenic BVDV, which can only induce apoptosis in lymphocytes. The soluble factor(s) involved in apoptosis has not been well characterized in any of these studies. Several studies have identified cytokines that prime for apoptosis, including IL-4 [24], TNF-α, IFN-γ [25–27] and IL1beta [28]. However, these cytokines are produced by lymphocytes rather than macrophages [23].

To investigate the indirect effect of different strains of BVDV on macrophage activity and lymphocytes apoptosis, both peripheral blood lymphocytes and a BVDV-free BL-3 cell line were used [3]. The direct apoptotic effect of BVDV was examined by the infection of lymphocytes with different BVDV strains. The indirect effect was investigated by infection of the MDM with one of two different strains of BVDV, followed by treatment of lymphocytes with the irradiated virus-free supernatant from BVDV-infected macrophages [29].

#### **2. Materials and Methods**

#### *2.1. Viruses and Cells*

Two Ncp strains of BVDV 2 were used in the current study: a high virulent 1373 and, a low virulent 28508-5, in addition to Cp 296C strain as control for apoptosis [30]. All strains were titrated using TCID50 titration assay [31].

BVDV-free Madin Darby bovine kidney MDBK cells (passage 113–135) were grown in minimal essential medium (MEM, Gibco BRL, Grand Island, NY, USA) (pH 7–7.4) supplemented with 10% BVDV-free fetal bovine serum (FBS) (Hyclone Laboratories, Logan, UT, USA), penicillin (100 U/mL) and streptomycin (100 μg/mL). MDBK cells were maintained at 37 ◦C with 5% CO2 in humidified incubator and were used for viral titration and propagation as well as apoptosis experiments.

Bovine lymphosarcoma cell line (Bl-3) is a non-adherent, BVDV-free B cell line clone derived from BL-3.1 cell line that was obtained from ATCC, Manassas, VA, USA. BL-3 cell line was cultured as previously described [3] and they were used to examine the direct and indirect apoptotic effect of different BVDV strains.

#### *2.2. Isolation of Monocytes and Di*ff*erentiation to Monocyte Derived Macrophage (MDM)*

PBMCs were isolated as previously described [32,33] with modifications. Briefly, the heparinized blood was diluted 1:1 with PBS and overlaid over histopaque 1083 gradient using SepMateTM 50 mL tubes (Stemcell Technologies, Cambridge, MA, USA) and centrifuged at 1200× *g* for 20 min at room temperature (RT). The buffy-coat was then transferred to a clean 50 mL conical tube and washed five times with PBS, followed by centrifugation at 120× *g* for 10 min at RT. The viability of PBMCs was determined by trypan blue exclusion assay according to Strobber [34]. The PBMCs were suspended in RPMI 1640 medium (GE Healthcare, Hyclone Laboratories, Logan, UT, USA) supplemented with 10% FBS, penicillin (100 U/mL) and streptomycin (100 μg/mL) to achieve a final concentration of 1 × 10<sup>6</sup> cells/mL. The cells were incubated in T175 flasks for 3 h at 37 ◦C. Then, the adherent monocyte was washed with PBS five times and detached by incubation for 10 min with AccutaseTM (eBioscience, San Diego, CA, USA). Detached monocytes were PBS-washed twice to get rid of Accutase. The isolated monocyte cultured in complete RPMI 1640, as described by Elmowalid [29], at a concentration of 10<sup>5</sup> cells/well in 48-well plate, followed by incubation for 5 days at 37 ◦C. The incubated cells were fed every other day by replacing half of the conditioned media with fresh complete RPMI. At day 5, the MDM were characterized phenotypically as MHCI-, MHCII-, CD11b- and CD14-positive cells.

#### *2.3. Production and Inactivation of Infected MDM Supernatant*

The MDM were infected with 100 μL of 105 TCID50 BVDV strains at MOI of 1 in triplicate as described by Elmowalid [29], with modifications. The infected cells were incubated for 1 h, then washed to remove the excess unbound virus, and 500 μL of complete RPMI 1640 medium was added to each well in a 24-well plate. At least one column of the plate was mock-infected with complete RPMI 1640 medium as a negative control. The infected MDM were incubated for 12, 24 or 48 h at 37 ◦C in CO2 incubator. The BVDV-infected macrophage supernatants were collected at 12, 24 and 48 hpi (hours post-infection) and centrifuged at 1000× *g* for 10 min at RT to remove cellular debris. The supernatant was UV-inactivated for 20 min on ice to exclude the direct virus effect [35]. The absence of any infectious viral particles in the treated supernatants was confirmed by inoculation on MDBK cells followed by a 5-day incubation and BVDV specific immune-staining of the inoculated MDBK cells using both immune-peroxidase and immunofluorescence. The positive control was 1373 infected MDBK cells.

#### *2.4. Phagocytosis*

One mL of virus-free (UV-inactivated) supernatants collected at 24 or 48 hpi from Ncp1373 or 28508-5 BVDV strains or mock-infected MDM were used to treat MDM, cultured on 24-well plates (approximately 5 <sup>×</sup> <sup>10</sup><sup>5</sup> cells/well whose viability was <sup>&</sup>gt;92%), for 24 h. This was followed by washing the treated cells twice with PBS and then exposed to 250 <sup>μ</sup>L containing approximately 2 <sup>×</sup> 107 of TRITC-labeled *C. albicans* (50 yeast/macrophage) and incubated for 30 min at 37 ◦C. Finally, the cells were washed twice in cold PBS and re-suspended in 200 μL/well freshly prepared paraformaldehyde (PFA) to be examined under UV-microscopy. The number of yeast/cell counted and the cells were classified into two groups: cells that contained >20 TRITC labeled yeast/cell that indicated normal phagocytic activity and cells that contained <20 TRITC labeled yeast/cell that indicated insufficient phagocytic activity. A total number of 100 MDM containing yeast were counted and the percentage of phagocytic activity was calculated according to the following formula: Phagocytic% = (number of MDM containing >20 yeasts/total number of MDM) × 100.

#### *2.5. Bactericidal Activity*

MDM were treated with 1 mL/well of UV-inactivated supernatant from MDM infected with 1373, 28508-5 or mock-infected 24 and 48 h supernatant, and incubated for 24 h at 37 ◦C. The cells were then washed twice in PBS and incubated for 30 min or more (up to 120 min) with 1 <sup>×</sup> 107 *E. coli* (20 bacteria/macrophage) suspended in RPMI 1640 supplemented with 5% FBS and no antibiotics

to allow bacterial phagocytosis. One group of the cells were washed three times with cold PBS 30 post-incubation to stop phagocytosis and to remove excess bacteria (time 0) and were lysed and the lysates were centrifuged at 800× *g* for 10 min and the supernatants were aspirated, while the pellets were suspended in LB broth and stored at 4 ◦C. The other group of cells were incubated in 5% FBS, and no antibiotics for extra 90 min to allow killing (total 120 min) followed by washing and lysis, as described above. Cell lysates of both 0 and 90 min timepoints were placed in 96-well plates and incubated for 18 h at 37 ◦C, followed by adding 30 μL/well of of 50 mg/mL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT, Sigma, St. Louis, MO, USA) and the plates were incubated at 37 ◦C for another 6–8 h. Optical density was measured at 540 nm wavelengths in ELISA reader (Biotek, ELx808, Winooski, VT, USA). The percentage of killing was calculated by the following formula: Killing% = (absorbance after incubation for 90 min/Absorbance at time 0) × 100.

#### *2.6. Nitric Oxide Production*

Nitric oxide (NO) assay was done in 24-well plates. MDM were treated by adding 1 mL/well of UV-treated supernatant from 1373 or by direct infection with 1373 strain of BVDV (MOI of 1) kept with the cells. Mock-infected supernatant or RPMI 1640 media containing 5% FBS were used as a control. The cells were incubated for 24 h at 37 ◦C. The cells were then stimulated with lipopolysaccharide (25 μg/mL) (LPS, Sigma, St. Louis, MO, USA) for another 24 h. Then, supernatants were collected and nitric oxide (NO) concentration was determined using Griess reagent (Sigma, St. Louis, MO, USA) using Nitrite (NO2) measurement as an indicator of NO production according to the protocol of Blond et al., 2000: briefly, 100 μL of culture supernatant was added to 100 μL of Griess reagent, made of a 1/1 mixture of 1% (wt/vol) sulfanilamide and 0.5% (wt/vol) *N*-(1-naphthyl)ethylenediamine dihydrochloride (Sigma, St. Louis, MO, USA) in 30% acetic acid, in each well of a 96-well plate. Reactions were performed in triplicate at RT for 10 min. Chromophore absorbance was then measured at 550 nm in a microplate reader (Bio-TEK Instruments model EL 311; OSI, Paris, France). Nitrite concentration was evaluated by comparison with the nitrate standard curve (Sigma, St. Louis, MO, USA). The lower limit of detection for nitrite is 250 nM.

#### *2.7. Immunostaining and Flow Cytometry of CD14 and MHC II*

To examine the effect of BVDV direct infection or its supernatant on macrophage surface marker expression, the MDM were either infected or treated with different BVDV-48 h-supernatants: including 1373 and 28508-5 strains. This was in addition to the supernatant from mock-infected MDM. Surface marker expression was investigated 48 hpi, starting with using AccutaseTM (eBioscience, San Diego, CA, USA) to detach the MDMs that were washed twice in PBS and plated in a U-bottom-96-well plate by adding100 <sup>μ</sup>L of cells (2 <sup>×</sup> 105 cell) per well. Cells were then suspended in blocking buffer (50 μL/well) (PBS containing 2% FCS and 0.1% sodium azide) for 20 min at RT. 50 μL/well of mouse anti-bovine CD14, or MHC II (VMRD, Pullman, WA, USA) diluted at 1:200 in PBS was added and the cells incubated for 20 min at 4 ◦C, followed by washing three times in PBS. Subsequently, the cells were incubated with FITC-conjugated goat anti-mouse Ig G (Ig G whole molecule, ICN/Cappel laboratories, Santa Mesa, CA, USA) diluted at 1:1000 in staining buffer for 20 min in dark at 4 ◦C, followed by washing three times in PBS. Fifty (50) μL/well of propidium iodide (1 μg/mL) were added directly before analysis to stain the dead cells. Both cell control (without staining) and the FITC control (just FITC staining without the primary anti-bovine antibodies) were included to eliminate nonspecific background. For each sample, mean fluorescent intensity of immune-stained cells was estimated and analyzed on FACScan flow cytometry (Becton Dickinson, Mountain View, CA, USA) using CELL Quest software (BD Biosciences).

#### *2.8. Examining the Indirect or Direct E*ff*ect of BVDV Strains on Peripheral Blood Lymphocyte Population and MDBK*

To examine the indirect effect of BVDV strains, peripheral blood lymphocytes, BL-3 or MDBK were treated with various time-point irradiated supernatants from BVDV-infected macrophages at a dilution of 1:1 and incubated for 12, 24 or 48 h post treatment, as previously described [36]. To examine the direct effect of BVDV strains, peripheral blood lymphocytes or BL3 were infected with BVDV strain at MOI of 1, as described by Ridpath et al. [37]. The effect of the direct infection of lymphocyte with Cp 296C strain was included as a positive apoptosis control.

#### *2.9. Chromatin Condensation*

The apoptotic effect of the infected MDM supernatant on epithelial cells was investigated by incubating MDBK cells for 12, 24, or 48 h at 37 ◦C with 1 mL/well of UV-treated supernatant from 1373, or 28508-5 as well as the non-infected supernatant. The cells were then washed twice in PBS, and then stained using 4,6-diamidino-2-phenylindole (DAPI, Sigma, St Louis, MO, USA) according to the protocol of Mi Hyeon et al., 2002. Simply, one volume of pre-staining solution [citric acid (2.1 g), Tween 20 (0.5 mL), and distilled H2O (100 mL)] was added and the cells were incubated for 5–7 min at RT in the dark. After the incubation, another six volumes of staining solution [(citric acid (2.1 g), DAPI (0.2 mg/mL), 1 mg DNAse, and distilled H2O (100 mL)] were added for another 5–7 min. Cells were washed three times in PBS and examined under fluorescence microscopy. Two hundred cells (200) were counted and the mean value was calculated. The apoptotic index was the mean of three independent experiments.

#### *2.10. Annexin V Staining*

Both direct BVDV infection or the indirect effect of its supernatant on bovine lymphocytes and MDBK cells was investigated using the AnnexinV staining kit (eBioscience, San Diego, CA, USA). Briefly the infected or supernatant-treated lymphocytes were washed once with PBS and the cells were re-suspended in 100 μL of 1x binding buffer supplied by the kit. Five μL of FITC conjugated annexin v antibody added to the lymphocytes suspended in 100 μL of binding buffer and incubated for 10 min in the dark. This was followed by washing the cells twice by suspending the cells in 1x binding buffer followed by adding another five μL of propidium iodide to exclude the dead cells. Finally, the cells were re suspended in 200 μL of 1x BD FACSTM lysing solution followed by analysis with BD AccuriTM C6 Plus Flow Cytometer (BD Biosciences, San Jose CA, USA). The result was expressed as % of apoptosis.

#### *2.11. Protein Electrophoresis*

The BVDV-infected MDM supernatant was fractionated on 10% sodium dodecyl sulphate polyacrylamide gel (SDS-PAGE) according to the protocol of Schagger and von Jagow, 1987. Briefly, 40 μL of the supernatant (around 50 μg) were added to 10 μL of 5x sample buffer (65 mM Tris-HCL, pH 7.0, 2% SDS, 10% glycerol, 5% β mercaptoethanol, and 0.001% bromophenol blue) with heating at 100 ◦C for 3–5 min. Protein electrophoresis was then done by loading the treated sample (50 μL) on the SDS-PAGE gel at 25 mA current for 45–60 min followed by staining with Coomassie blue overnight (in some experiments for 4 h) with shaking at RT. Finally, the gel was immersed in de-staining buffer (40% methanol and 10% glacial acetic acid) with shaking for 15 min at RT until the protein bands become clear [38].

#### *2.12. Quantification of Apoptosis-Related Cytokines by qRT-PCR of BVDV-Infected MDMs*

Both the infected and mock-infected MDM were pelleted at 500× *g* for 8 min, then washed twice in PBS. The infected MDM were lysed and nucleic acid (NA) extraction done using RNeasy extraction kit (Qiagen, Valencia, CA, USA). The extracted NA of different samples was normalized to 5 ng/μL using Nanodrop ND-1000 Spectrophotometer (Fisher Scientific, Portsmouth, NH, USA). Nucleic acid extracted from mock-infected MDM with complete RPMI 1640 medium was used as a negative control while NA extracted from MDM treated with Concanavalin A (ConA) or lipopolysaccharide (LPS) (Sigma-Aldrich, St. Louis, MO, USA) was used as a positive control. Around 5 μL of the normalized NA was used for relative quantification of apoptosis-related cytokines; TNF-α, IL-1α, IL-1β and IL-6 for each sample in duplicates using quantitative reverse transcriptase PCR (qRT-PCR) and Power SYBR® Green RNA-to-CtTM 1-Step Kit (Thermo Fisher Scientific, Millersburg PA, USA). Additional cytokines were also measured that include: IFN–α, β and γ, IL-4, 8, 10, and 12. The cytokine primers used were described in Table 1 [39]. The relative expression of mRNA was standardized using beta actin and GAPDH as housekeeping genes. Each qPCR experiment was followed by heat dissociation curve step to exclude nonspecific amplification. qRT-PCR results were analyzed using relative expression software tool (REST©2009 software) [40].


**Table 1.** The set of primers used for quantifications of apoptosis-related cytokine mRNA.

#### *2.13. Supernatant Neutralization*

To exclude the apoptotic effect induced due to possible viral secreted proteins, 24 or 48 h supernatants from the BVDV-infected macrophages with 1373 strain were incubated at 37 ◦C for 1 h with Erns specific 15c5 monoclonal antibody (IDEXX Laboratories, Westbrook, ME, USA) or with BVDV-polyclonal antibody (kindly supplied by Dr. Robert Fulton, Food Animal Research, Oklahoma State University). The mixture of the infected supernatant and antibody was prepared as two parts of infected supernatant incubated with one part of concentrated antibody. Three (3) parts of the supernatant/antibody mixture were added to one part of BL-3 cell line seeded at 1 <sup>×</sup> 105 cells/well, followed by incubation at 37 ◦C for 36 h post-treatment.

#### *2.14. Statistical Analysis*

Data were analyzed using a Student's *t*-test (Microsoft EXCEL, MAC 2011) to assess the significance of the differences between mean values of treated and control samples at each timepoint. Differences were considered significant at *p* < 0.05, however some treatments showed very significant difference with *p* value of <0.01. Every experiment was achieved using at least three different animals and each experiment was done in triplicates. The variations in results were calculated by standard deviations at each timepoint. For cytokine analysis, REST© 2009 [40], that is based on analysis of variance (AVOVA).

#### **3. Results**

#### *3.1. Inactivation of Infected Supernatant*

The irradiated supernant-treated MDBK cells had no immune-peroxidase stain for the UV-irradiated supernant (Figure 1A) compared to the positive intracytoplasmic staining of 1373-infected-MDBK non-irradiated control (Figure 1B). There was no fluorescence signal for the irradiated strain (Figure 1C) compared to the non-irradiated infection control (Figure 1D).

**ϭϯϳϯ/ƌƌĂĚŝĂƚĞĚƐƵƉĞƌŶĂƚĂŶƚ**

**ϭϯϳϯŝŶĨĞĐƚĞĚƐƵƉĞƌŶĂƚĂŶƚĐŽŶƚƌŽů**

**Figure 1.** Successful inactivation of 1373-infected monocyte-derived macrophages (MDM) supernatant by UV-irradiation. (**A**,**C**) irradiated 1373-supernatant that show no evidence of intra-cytoplasmic replication by immune-peroxidase (IP) and immune-fluorescence (IF). (**B**,**D**) intra-cytoplasmic replication signal of the infected MDM supernatant, indicated by black and white arrows in the IP and IF assay, respectively (**B**,**D**).

#### *3.2. E*ff*ect of Supernatant from BVDV Infected MDM on Macrophage Phagocytic Activity Compared to the Direct BVDV E*ff*ect*

At 24 h post-treatment, the supernatant from Ncp 1373 significantly reduced MDM phagocytic activity by 38.5%, which continued to decrease to 51% at 48 h post-treatment compared to the supernatant from Ncp 28508-5, which was similar to the mock-infected control at any timepoint (Figure 2). Direct infection of MDM with the highly virulent 1373 strain, but not 28508-5, diminished phagocytic activity of MDM as early as 12 h post-treatment by 27.5% (Figure 2). We also found that heat-treated supernatant from 1373, or 28508-5 BVDV-infected supernatant had no effect on MDM phagocytic ability (data not shown).

**Figure 2.** The effect of BVDV strains and the corresponding infected-MDM supernatants on macrophage phagocytic activity. Phagocytic index is defined as the % of MDMs that contains >20 TRITC-labeled *C. albicans*. h: hours post-treatment, V: direct virus infection, no supernatant, Sup: BVDV-infected macrophage supernatant, non-infected Sup: non-infected macrophage supernantant, Mock: culture medium \*: *p* < 0.05 (>95% confidence), \*\*: *p* < 0.01 (>99% confidence).

#### *3.3. E*ff*ect of Supernatant from BVDV-Infected MDM on Macrophage Bactericidal Activity Compared to the Direct BVDV E*ff*ect*

No significant effect was observed at 6 or 12 h post-treatment, however, the supernatant from Ncp 1373 strain significantly (*p* < 0.05) reduced MDM bactericidal activity by 37.7% and 51.4% at 24 and 48 h post-treatment, respectively (Figure 3). Direct infection of MDM with 1373 decreased the bactericidal activity by 22%, which began as early as 12 hpi, unlike the late indirect effect. The inhibition of bactericidal activity was time-dependent, and continued to increase to reach 54.7% by 48 hpi. There was no significant effect with 28508-5 strain, its corresponding infected-MDM supernatant, mock-infected, or the non-infected supernatant on bactericidal activity (Figure 3).

**Figure 3.** The effect of BVDV strains and the corresponding infected-MDM supernatants on macrophage bactericidal activity for 48 h post-treatment. Sup: supernatant (indirect effect), V: virus (direct effect). \*: *p* < 0.05 (>95% confidence), \*\*: *p* < 0.01 (>99% confidence).

#### *3.4. E*ff*ect of Supernatant from BVDV Infected MDM on Macrophage Nitric Oxide Production Compared to the Direct BVDV E*ff*ect*

The direct infection of MDM with both BVDV strains significantly increased NO production (158–164 μmol/mL), while neither 1373- nor 28508-5-infected MDM supernatant affected NO production in uninfected MDM (Figure 4). Both supernatant treatments, as well as non-infected supernatant and mock-infected controls, showed low nitric oxide production (93.3 μmol/mL) (Figure 4).

**Figure 4.** Comparison between the direct effect of BVDV strains and the corresponding infected-MDM supernatants (the indirect effect) on macrophage nitric acid production 24 h post-treatment. Sup: supernatant (indirect effect), V: virus (direct effect). \*: *p* < 0.05 (>95% confidence).

#### *3.5. E*ff*ect of Supernatant on MHC II Expression Compared to the Direct BVDV E*ff*ect*

At 24 h post-treatment, cells treated with 1373 supernatant had significant reduction in MHC II expression to 61.0% (Figure 5A). At 48 h, MHC II expression was further reduced to 50.0% (Figure 5A) In contrast, 28508-5 or mock-infected supernatant had no significant reduction in MHC II expression and the expression level was greater than 94.0% at 24 or 48 h (Figure 5A).

#### *3.6. E*ff*ect of Supernatant on CD14 Expression Compared to the Direct BVDV E*ff*ect*

At 24 h post-treatment, the CD14 expression decreased in MDM infected with Ncp 1373 strain or it's supernatant (Figure 5B). At 36 h post-treatment, CD14 expression further decreased to 44.6% and 49.7% for Ncp 1373 and its supernatant, respectively (Figure 5B). The heat-treated supernatant or supernatant of 28508-5 did not show a significant decrease in the percentage of cells expressing CD14 and the average percentage was over 93% at 24 or 36 h (Figure 5B).

**Figure 5.** The effect of BVDV strains and the corresponding infected-MDM supernatants (indirect effect) on surface marker expression. (**A**) MHC II and (**B**) CD14 surface marker expression. Sup: supernatant (indirect effect), V: virus (direct effect). \*: *p* < 0.05 (>95% confidence), \*\*: *p* < 0.01 (>99% confidence).

#### *3.7. E*ff*ect of Supernatant from BVDV Infected Macrophages on MDBK Cells*

Chromatin condensation was detected after 24 h in treated MDBK with 1373 supernatant, while no chromatin condensation was detected in MDBK treated with 28508-5 nor non-infected supernatant (Figure 6A). The percentage of apoptotic cells was not significant until 24 h post-infection using Annexin V assay. It increased from 36.3% (24 h post-treatment) to reach 51.3% (48 h post-treatment) in contrast to less than 3% in case of mock-infected or treated MDBK with either 28508-5 or non-infected supernatant (Figure 6B).

(**B**)

**Figure 6.** The indirect (infected MDM 24 h supernatant) effect of different BVDV strains on MDBK apoptosis. (**A**) Chromatin condensation by 1373 strain in contrast to 28508-5 and non-infected supernatant. (**B)** The % of MDBK apoptosis in response to different supernatants. 1373 Sup: BVDV 1373-infected macrophage supernatant, 28508-5 Sup: BVDV 28508-5-infected macrophage supernatant Non-infected Sup: non-infected macrophage supernantant, Mock: culture medium \*\*: *p* < 0.01 (>99% confidence).

#### *3.8. E*ff*ect of Supernatant from BVDV-Infected Macrophages on Di*ff*erent Lymphocyte Populations*

The 12 h supernatants did not induce apoptosis compared to the mock-infected control (Figure 7). Both 24 and 48 h supernatants of the highly virulent 1373 resulted in an apoptotic effect on either freshly isolated (data not shown) or BL-3 lymphocytes. All supernatants of the low virulent Ncp 28508-5 at different timepoints did not induce apoptotic changes compared to the mock-infected control for either BL-3 lymphocytes or peripheral blood total lymphocyte population (Figure 7A,B).

**Figure 7.** The indirect (infected MDM 12, 24 and 48 h supernatant) and the direct effect of different strains of BVDV on lymphocyte apoptosis. (**A**) The indirect effect 24 h post-treatment (**B**) The indirect effect 48 h post-treatment 1373 Sup: BVDV 1373- infected macrophage supernatant, 28508-5 Sup: BVDV 28508-5-infected macrophage supernatant non-infected Sup: non-infected macrophage supernatant, Mock: culture medium, \*: *p* < 0.05 (>95% confidence).

#### *3.9. E*ff*ect of Direct Infection of Lymphocyte with BVDV Strains*

Only the Cp 296C strain was able to induce significant lymphocyte apoptosis as early as 12 h post-infection compared to the non-infected/mock control and the Ncp BVDV strains (Figure 8A) The direct infection of lymphocytes with 1373 was compared to the indirect effect of this virulent BVDV strain. Interestingly, the direct infection of lymphocytes did not induce significant changes compared to the mock-infected control until 48 h post-infection with 21.8% as the maximum apoptotic effect. The indirect effect of 1373 was tow times greater than the direct infection of lymphocyte at 24 and 48 hpi, with a maximum apoptotic effect of 35% at 48 h post-treatment (Figure 8B).

**Figure 8.** (**A**) The direct effect of BVDV strains on lymphocyte apoptosis (**B**) Comparison between the direct infection and the indirect (infected MDM 24 and 48 h supernatant) effect of the highly virulent 1373 BVDV strain on lymphocyte apoptosis for up to 48 pi. 296C V: direct lymphocyte infection with Cp 296C strain, 1373 V: direct lymphocyte infection with Ncp 1373 strain, 28,508 V: direct lymphocyte apoptosis with Ncp 28508-5 strain, Mock: non-infected lymphocyte, 1373 (48 h) Sup: supernatant from 1373-infected macrophage at 48 h post-treatment, 1373 (24 h) Sup: supernatant from 1373-infected macrophage at 24 h post-treatment, 1373 Virus: direct infection of macrophage with 1373 strain (not supernatant), Non-infected: supernatant from macrophage that was not infected, \*: *p* < 0.05 (>95% confidence), \*\*: *p* < 0.01 (>99% confidence).

#### *3.10. Supernatant Protein Analysis*

No viral proteins were detected in the supernatants at 6 hpi (Figure 9A), however, faint bands of 40–45 KD started to appear in the 1373-infected MDM supernatant at 12 hpi (Figure 9B). By 24 hpi, bands of 30 and 120 KD were observed in supernatant from Ncp1373. MDM culture infected with 28508-5 showed only one dense band of about 40–45 KD (Figure 9C). At 48 hpi, the band numbers and

densities increased in MDM infected with the Ncp 1373. In contrast, only two bands of 40 to 45 KD were observed in supernatant from 28508-5 Ncp BVDV strains by that time (Figure 9D). No bands were observed in supernatant from mock-infected MDM at 6, or 12 h. At 24 or 48 hpi, only one band of 40–45 KD was observed in the mock-infected cells supernatant (Figure 9).

**Figure 9.** Protein electrophoresis of supernatants from BVDV infected and mock-infected MDM at (**A**) 6 hpi, (**B**) 12 hpi, (**C**) 24 hpi and (**D**) 48 hpi. Marker: protein marker that range from 25 to 120 KD, Mock: non-infected macrophage control, 28508: supernatant from 28508-5-infected macrophage at different timepoints, 1373: supernatant from 1373-infected macrophage at different timepoints, 6, 12, 24, 48 hpi: the corresponding timepoints post treatment of macrophage with supernatant, arrows refer to the different band intensity between strains.

#### *3.11. Role of Cytokines in the Indirect Lymphocyte Apoptosis*

Apoptosis-related cytokines including TNF-α, IL-1β and IL-6, did not show significant change at any of the three timepoints compared to mock-infected control. Other cytokines like IFN-γ, IL-10, IFN-α, IFN-β, IL-12, IL-4, IL-8 and TGF-β have also shown no significant difference between groups. The positive control mixture of Concavalin A and LPS upregulated all of the cytokines (data not shown).

#### *3.12. Role of Viral Factors in the Indirect Lymphocyte Apoptosis*

There was no significant difference in the percentage of lymphocyte apoptosis between the neutralized supernatant with either mAb or polyclonal Ab and the un-neutralized supernatant (data not shown).

#### **4. Discussion**

This study sheds light on the role of macrophages in the pathogenesis of BVDV infection. Phagocytosis is an important immune response of the macrophage and constitutes one of the first defenses against microbial invasion. Successful phagocytic activity leads to the initiation of successful immune response. In this study, supernatant from the highly virulent 1373 strain reduced macrophage phagocytic activity. In a previous study, Ncp BVDV infection of alveolar macrophage decreased expression of complement and Fc receptors and chemokine production, reducing their ability to engulf opsonized pathogens [8]. IL-2 inhibitory substances are produced in BVDV-infected PBMCs cultures that bind to receptors on lymphocytes, macrophages and dendritic cells that activate protein

kinase C, resulting in the phosphorylation of proteins and inhibition of basic metabolic activities [41], that may reduce cell movement and the capacity to engulf organisms. A reduction in macrophage phagocytic capacity may be due to the presence of chemokine inhibitory mediators or the decrease in cytokine-induced chemotaxis [42] that lead to the decreased capacity of phagocytic cells for chemotaxis and phagocytosis.

Microbicidal activities are one of the main functions of macrophage to control and clear microbial infection. In vivo experiments suggest that BVDV immunosuppression resulted in subsequent secondary microbial infection. The supernatant from low virulent 28508-5-infected macrophage had no effect on bactericidal activity, while the highly virulent 1373 supernatant decreased the bactericidal activity of macrophages significantly with more than 50% by 48 hpi (*p* < 0.05). Considerable evidence has accumulated that BVDV may be a key component in multiple-etiology diseases [43].

Nitric oxide production was not affected by the supernatant exposure. Direct infection of MDM with highly virulent 1373 strain significantly induced NO production; however, 1373-infected macrophage supernatant (indirect) had no significant effect on NO production in MDM following stimulation with LPS. Another flavivirus, the hepatitis C virus (HCV) ERNS protein and HIV infection was associated with increased NO synthase (NOi) production in mice livers [44,45]. Type I IFN is a potent inducer of NO; however, it is downregulated by Ncp BVDV strains [25,46]. Taking these results together, Ncp BVDV might induce NO production directly through ERNS protein activation to NOi, but not indirectly due to the lack of secreted IFN in the MDM supernatant.

Microbial recognition, processing and presentation to lymphocytes is one of the unique functions of antigen-presenting cells for initiation of immune response. The recognition of pathogens is achieved through surface markers, which initiate the immune response. Our results indicated a significant reduction in surface marker expression of MHC II and CD14, associated with the highly virulent 1373-infected MDM supernatant. MHC II is an important marker for the initiation of immune response and antigen presentation. The current study indicated that the highly virulent 1373 reduced the surface expression of MHC II molecules. A previous study demonstrated that MHC II expression decreased in PBMCs infected with Ncp BVDV [47]. Another study reported that BVDV infection downregulated MHC I and II in monocytes [48]. This is consistent with what we found but in macrophage.

CD14 is important for toll-like receptors and gram-negative recognition and phagocytosis. Our findings indicated that low virulent 28508-5 supernatant had no effect on CD14 expression, while the highly virulent 1373 supernatant significantly decreased CD14 expression. CD14 downregulation was associated with the reduction in the phagocytic activity of neutrophils in HCV patients [49]. In human and mouse models, downregulation of CD14 was associated with suppression of antigen-specific lymphocyte proliferation [50]. Impairment of macrophage inflammatory function is a likely consequence of this surface reduction with subsequent secondary microbial invasion and the immunosuppressive effect of BVDV on the infected animals.

The findings of this study indicate the importance of soluble factors induced in BVDV-infected macrophages in viral pathogenesis. These results demonstrate that only the highly virulent Ncp BVDV strain induced the production of soluble factors capable of impairing macrophage inflammatory functions and surface marker expression as early as 12 h post-infection. These mediators and factors also induced epithelial cell and lymphocyte apoptosis. The release of several BVDV induced-soluble factors release has been reported [8,51–53]. On gel electrophoresis, several proteins ranging from 30 to 120 kDa were produced in 1373-infected MDM supernatants that were either less intense or not present at all in 28508-5-infected MDM supernatant. In a previous study, inflammatory mediators of 40–75 KD play a pivotal role in viral infection and pathogenesis [54].

Another study indicated that PMNC infected with Ncp or Cp released unknown factor(s) in the culture media that could be of cellular or viral origin [52,55]. Previously, secreted Erns viral glycoprotein played an important role in lymphocyte apoptosis induced by BVDV and CSFV [19]. In our study, we found no association of Erns glycoprotein or any other viral factors in the induction of apoptosis. Neither anti-ERNS mAb or a polyclonal BVDV Ab treatment of BVDV-infected MDM

supernantant reduced lymphocyte apoptosis. The difference between BVDV and CSFV may be due to the use of a cloned Erns glycoprotein of CSFV that would be expressed at higher levels versus constitutive expression of BVDV Erns in BVDV-infected cells.

In the current study, we found that a supernatant from the highly virulent 1373 infected MDM induced apoptosis in MDBK as early as 24 h post-treatment, while low virulent 28508-5 infected MDM supernatant did not induce apoptosis. Previous work has shown that Ncp 28508-5 BVDV strain did not cause an inflammatory response in vivo, while Ncp 1373 caused inflammation and depletion of lymphocytes [56]. Interestingly, the Ncp 28508-5 strain is much more successful at establishing persistent infection. This lack of an inflammatory response may be one evasion mechanism to avoid the immune response and establish persistent infection in the developing fetus.

Furthermore, we found that the virulence of the BVDV strain is very important in determining the severity of the corresponding lymphoid depletion and subsequent immune suppression. The low virulence of the 28508-5 supernatant failed to induce apoptosis on the freshly isolated lymphocytes, while the highly virulent 1373 supernatant induced significant lymphocyte and somatic cell apoptosis. These results support the association between virulence and the degree of immune suppression that is associated with lymphoid depletion in vivo [1–3,57].

The role of macrophages in lymphocyte apoptosis was investigated in this study. Our results suggested that the supernatant of infected macrophages with only virulent BVDV strains induced lymphocyte apoptosis. These results provide an explanation for the observation that there is an increased number of macrophages in lymph nodes 3 days post-infection prior to lymphoid depletion [13]. Moreover, this indirect effect mediated by macrophages was not significant until 24 h post-infection with virulent BVDV strains, suggesting that rapid treatment strategies may help to limit the infection and would decrease or prevent the severe immune suppression induced by BVDV.

Our data support the results of Pedrera et al., 2009 [13] that did not relate the apoptotic effect to the biotype of the strain but to its virulence. Interestingly, we noted that the highly virulent 1373 strain of BVDV induced a more marked and faster apoptotic effect on lymphocytes indirectly as compared to the direct effect. These results suggest that there is a more severe and rapid immune suppression associated with the highly virulent 1373 that is likely due to the indirect effect mediated by macrophages.

Apoptosis-related cytokines produced by BVDV-infected macrophages did not play a role in lymphocyte apoptosis. There was no significant change in TNF-α, IL-1α, IL-1β or IL-6 transcriptional levels. These results were different from those described for CSFV, where TNF-a was involved in the indirect lymphocyte apoptosis activated by the virus [36]. Although TNF-α induction appears to be included in the mechanism of CSFV-induced lymphocyte apoptosis, this does not appear to be true for BVDV. The lack of TNF-α induced apoptosis is consistent with previous research with BVDV lymphocyte apoptosis [13]. Another study suggested that a role of IFN-γ in inducing apoptosis, unlike our research, was done using monocytes and a different BVDV strain [26].

The BL-3 cell line gave similar results, as seen with the peripheral blood lymphocytes. BL-3 cell findings were consistent with previously reported apoptotic effect of the highly virulent BVDV strains on the BL-3 cell line [30]. The degree of apoptosis induced in the lymphocytes of both peripheral blood or BL-3 cell line was similar. This finding leads us to speculate that B cells are more susceptible than T lymphocytes, but this hypothesis needs to be tested.

Another study suggested the identity of some possible factors that mediate the lymphocyte apoptosis post-BVDV infection, like micro RNA and programmed death-1 (PD-1) [58,59]. Finally, the identification and characterization of these factors and the molecular mechanisms involved in their induction and modes of action will be important goals for future studies.

#### **5. Conclusions**

BVDV causes immune dysfunction that leads to vaccination failure and secondary bacterial infection with lymphoid depletion associated with BVDV clinical cases. It was important for our study to focus on immune dysfunction caused by BVDV through investigating the indirect effect of an infected macrophage supernatant of various BVDV strains on macrophage inflammatory function as well as lymphocyte apoptosis, these two cells that are important for clearance of the virus from the body of infected animals. Our data suggested that the immune dysfunction associated with the highly virulent Ncp 1373 strain was mainly due to the indirect effect mediated by macrophage-secreted mediators. These mediators may impair macrophage surface marker expression, which in turn disrupt the general macrophage inflammatory function, including phagocytic and bactericidal activity but not NO production. These soluble factors also enhance lymphocyte apoptosis with subsequent lymphoid depletion associated with BVDV cases. Unlike CSFV, neither TNF-a nor secreted Erns glycoprotein induces BVDV lymphocyte apoptosis. This led us to the hypothesis that host but not virus factors likely mediated the indirect lymphocyte apoptotic effect.

Further investigations of the factors present in the supernatant of the infected macrophages with the highly virulent strains need to be conducted so that we have a better understanding of the disease mechanism. This can lead to finding a way to block these factors that can be used as a therapeutic practice in acutely infected animals to help improve recovery. Understanding these factors better would also help to improve the current control strategies and decrease the risk and the economic losses of BVDV. Future studies that include other BVDV strains need to be conducted to further confirm our findings.

**Author Contributions:** Conceptualization, K.A., G.E., and C.C.L.C.; writing-original draft preparation, K.A., M.R.; writing-review and editing, K.A., G.E., and A.A.H.A.; visualization N.T., J.S., and H.A. All authors have read and agreed to the published version of the manuscript. K.A. and M.R. have the same contribution as first author.

**Funding:** This research received no external funding.

**Acknowledgments:** We would like to thank the support from Shollie Falkenberg for providing the BL-3 cell line. Robert Fulton for providing the polyclonal antibodies for BVDV testing. Adam Hoppe help and guidance with using the flow cytometry machine. The Department of Veterinary and Biomedical Sciences/Animal. Disease Research and Diagnostic Laboratory, SDSU, Brookings, SD, USA; SDSU. Functional Genomic Core Facility (FGCF), SDSU.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Bovine Viral Diarrhea Virus: Recent Findings about Its Occurrence in Pigs**

**Luís Guilherme de Oliveira 1,\*, Marina L. Mechler-Dreibi 1, Henrique M. S. Almeida <sup>1</sup> and Igor R. H. Gatto <sup>2</sup>**


Received: 23 April 2020; Accepted: 26 May 2020; Published: 31 May 2020

**Abstract:** Bovine viral diarrhea virus (BVDV) is an important pathogen belonging to the *Pestivirus* genus, *Flaviviridae* family, which comprises viral species that causes an economic impact in animal production. Cattle are the natural host of BVDV and the main source of infection for pigs and other animal species. Due to its antigenic and genetic similarity with other important pestiviruses such as Classical Swine Fever Virus (CSFV), several studies have been conducted to elucidate the real role of this virus in piglets, sows, and boars, not only in the field but also in experimental infections, which will be discussed in this paper. Although BVDV does not pose a threat to pigs as it does to ruminants, the occurrence of clinical signs is variable and may depend on several factors. Therefore, this study presents a survey of data on BVDV infection in pigs, comparing information on prevalence in different countries and the results of experimental infections to understand this type of infection in pigs better.

**Keywords:** BVDV; experimental infection; natural infection; pigs

#### **1. Updates on BVDV Infection in Swine**

Bovine viral diarrhea (BVD) is an infection caused by the bovine viral diarrhea virus (BVDV), belonging to the genus *Pestivirus*, family *Flaviviridae*, with single-stranded positive polarity RNA [1]. Viruses belonging to the *Pestivirus* genus infect hosts of several animal species and include viral agents of great impact for animal production [2,3]. The *Pestivirus* species have been recently named from A to K and, among them, the *Pestivirus* A (BVDV-1), *Pestivirus* B (BVDV-2), *Pestivirus* C (Classical Swine Fever Virus), and *Pestivirus* K (atypical porcine pestivirus) are the main viral species related to swine [4]. BVDV has two genotypes, type 1 and type 2, which are classified into sub-genotypes: BVDV-1 (1a to 1u), adding up to 21 sub-genotypes, and BVDV-2 (2a to 2d), with four sub-genotypes [5]. BVDV-1 is related to most reference strains, is commonly used for vaccine production, and was most frequently isolated from mild to moderate clinical cases in cattle. Conversely, BVDV-2 was isolated from acute disease outbreaks, also presenting strains of mild and moderate virulence [6].

Based on the effect of replication on cell culture, BVDV isolates can be divided into cytopathic (cp) and non-cytopathic (ncp), with the ncp isolates being responsible for most natural infections and persistent fetal infections, and cp isolates constituting a minority, which are isolated almost exclusively from cattle with mucosal disease [6].

Cattle are natural hosts of BVDV, considered the major source of infection for pigs and other animal species [7,8]. Usually, positive pig herds for BVDV occur when cattle and pigs are raised on the same farm, and the direct contact between these animal species is considered the main source of BVDV transmission

for pigs [7]. Infection caused by BVDV in pigs has been reported in China [9], the Netherlands [10], Brazil [11–13], Austria [14], Germany [15], Norway [16], Ireland [17], Denmark [18] and others. These data were found not only in domestic pigs but also in wild boars [19], which raise concerns about risk factors involved in BVDV infection, the clinical form of the disease, and the existence of accurate diagnostic tests. In Brazil, BVDV-1d was frequently reported in cattle [20]. Mósena et al. [11] states that by the phylogenetic analysis of sequenced samples collected from backyard pigs, classified as BVDV-1d and BVDV-2a, it is possible that one of the obtained sequences originated from contact between cattle and pigs.

It is known that all pestiviruses are genetically and antigenically related, and BVDV infection in pigs may be presented with a great variability of clinical signs [21]. Even though BVDV infections in pigs are not as problematic as Classical Swine Fever Virus (CSFV) infections, it is believed that distinguishing these two diseases could be difficult due to the similar clinical signs when considering low pathogenicity strains [22]. Reports of clinical signs associated with the infection consisting of anemia, delayed development, rough hair, polyarthritis, congenital tremors (CT), petechiae on the skin, diarrhea, conjunctivitis, and cyanosis [23]. Clinical signs similar to CSF, and sudden death [24] were also observed when the BVDV strain was isolated from both pigs and cattle from the same farm [24]. On the other hand, several recent studies with experimental infection did not report the presence of clinical signs of infection [25–34]. This may occur due to an inadequate level of viremia or a low virulence strain, biotype of the virus, host adaptation and/or route of inoculation [31–35]. A possible explanation is that cases in which BVDV infection-induced large numbers of lesions in adult pigs have been caused by viral strains that passed along previous adaptations in this species [23].

BVDV has a predilection for replication in defense cells, mainly lymphocytes, but it also infects monocytes and dendritic cells. As antigen presenters, dendritic cells play an important role in cellular immunity by initiating the nonspecific immune response against various pathogens [35]. Its infection promotes lysis of monocytes as a mechanism for evading the immune system, affecting the recognition and subsequent development of a specific immune-humoral response [36].

BVDV can contaminate cell cultures and fetal calf serum [37]. In countries that promote CSFV vaccination, the BVDV prevalence found in swine herds has been associated with the widespread use of live vaccines for Classical Swine Fever (CSF), which were produced with bovine sera from positive Chinese bovine herds [9]. Batches of live CSFV vaccines used in China confirmed five BVDV-contaminated samples out of 23 collected for testing [38].

Serological diagnosis by enzyme-linked immunosorbent assay (ELISA) is more efficient, cheaper, and faster than molecular techniques [37]. The neutralizing antibody titers in the serum of animals previously exposed to a Pestivirus member are usually medium to high regarding the homologous viral species, and low (or non-reactive) regarding other species [6]. Anti-BVDV antibodies were shown to be able to protect pigs against CSFV infection and the manifestation of clinical signs, even though the anti-CSFV antibody titers were low, which could hinder CSFV outbreaks in herds with a high prevalence of anti-BVDV antibodies [22]. The same condition could occur in the presence of anti-Border Disease Virus (BDV) antibodies, as cross-reactions could affect the transmission of CSFV and should be evaluated for an accurate diagnosis of a CSFV infection and for implementing specific surveillance protocols in cases of outbreaks [10].

Reverse transcription-polymerase chain reaction (RT-PCR) is widely used for detecting the viral agent for differential diagnosis [39], since samples of blood, milk, saliva, and tissue can be successfully tested [40], and can be stored for a long time with minimal losses [41]. Researchers have adapted numerous variations of PCR methods for the detection of infectious agents, using DNA templates as well as RNA templates after an RT step [42], which has enabled more accurate, sensible and specific diagnostics. Direct sequencing of the RT-PCR product for fragments of 5 UTR and N-terminal autoprotease (Npro) may also provide accurate differential diagnosis [19].

Given the antigenic and genetic similarity and the improvement in laboratory diagnostic methods, the comparison between results from recent and former studies should be cautious. Several studies that examine data collection in the field, as well as experimental infection with BVDV, have been conducted, and the results that contrast with the former data in the literature will be further discussed.

#### **2. Data Collection from Backyard and Intensive Pig Herds**

Aiming at collecting data on the occurrence of anti-BVDV antibodies in Brazilian swine herds, cross-sectional studies were carried out [12,13] in the backyard and intensive pig herds, respectively, located in the CSF-free zone of Brazil. For the first study, 56 pig herds from the northwest region of the state of São Paulo were evaluated, which are part of 11 municipalities. Blood samples were collected for serological testing by virus neutralization (VN), and titers higher than ten were considered positive. Out of the 360 serum samples, 4.72% (17/360) were reactive to BVDV in VN, which is 1.94% (7/360) reactive to BVDV-1 (Singer strain), with antibody titers ranging from 10 to 640, and 3.06% (11/360) reactive to BVDV-2 (VS-253 strain), with antibody titers ranging from 10 to 80, and only one reactive sample against both genotypes. Regarding herds, 27% (15/56) presented at least one animal positive to any of the genotypes. The prevalence of BVDV in bovine herds in the same region where this study was conducted was 56.49% [43], which may have resulted in the highest prevalence values of swine in the region when compared to previous reports [10]. Most of the farms evaluated in this study had cattle and pigs in close contact. As ruminants are the main source of BVDV infection for pigs [7,8], the prevalence of the disease in cattle herds is closely related to the presence of infections and influences the prevalence of the disease in pigs [9,10,44].

On the other hand, in a cross-sectional study carried out in 33 commercial pig herds, collected 1705 blood samples for analysis [13]. Samples were also tested by VN, and 5.34% (91/1705) of the samples were sero-reactive to BVDV with antibody titers ranging from 10 to 80. Of these, 3% (51/1075) were positive for reference strains of BVDV-1 (Singer strain) and 2.35% (40/1705) for reference strains of BVDV-2 (VS-253 strain), with 0.1% (2/1705) of samples with cross-reactions between both genotypes. Herds were sampled from 27 municipalities, located in seven Brazilian states, which are part of three different regions (South, Southeast, and Midwest). In 64% of herds (21/33), there was at least one positive sample for any of the BVDV genotypes in VN. As the presence of anti-BVDV antibodies in swine serum can lead to false-positive results in serological tests for the diagnosis of CSFV, the positive samples from both studies were sent for anti-CSFV antibodies detection, and were both negative. A survey carried out in The Netherlands [10] on commercial farm animals found a prevalence of 2.5% for gilts and 0.42% for finishing animals via ELISA. The difference in prevalence found in finishing animals from these two studies can be explained not only by the sensitivity of the techniques used but also by the different levels of biosecurity in the farms studied.

In a more recent study [11], swine sera were collected from 320 backyard pig herds in southern Brazil. Serum samples were tested by VN against BVDV-1a, -1b, and -2 strains, resulting in 4.2% (27/639) positive samples. Of those, 16 samples presented the highest titers against BVDV-1a (2 samples), BVDV-1b (5 samples), and BVDV-2 (9 samples). These studies confirm that ruminant Pestiviruses have been circulating in swine herds and must be considered in future Pestivirus control programs conducted in Brazil.

The low prevalence of BVDV in pigs was also found by other authors in Norway, Ireland, Denmark, and The Netherlands [16–18] (Table 1). Not only domestic pigs are a concern when it comes to Pestivirus infections since wild boar have been described as important reservoirs or transmitters of pathogens in nature due to their ability to reach long distances and transmit diseases to domestic swine. Other studies [45–47] have reported a low prevalence of BVDV in wild boar in Germany, the Czech Republic, and Eastern Serbia, respectively. Weber [19] was the first to detect BVDV RNA in wild boars' blood samples, and Gatto [48] first reported the presence of anti-BVDV antibodies in Thayasuids. In general, the low prevalence of anti-BVDV antibodies in pigs may be associated not only with the level of interaction between pigs and ruminants but also with the host-pathogen specificity, which seems to be lower in pigs compared to cattle. In intensive pig farming, biosecurity measures reduce or eliminate the presence of some infectious agents. Some researchers [10,44] attributed the low prevalence of BVDV in swine herds in their studies to the high specialization of agriculture, in which interspecific contact was reduced due to single zootechnical breeding on the property.


*Viruses* **2020** , *12*, 600

In China, swine samples with clinical signs such as diarrhea, miscarriage, and death, between 2007 and 2010, were tested for BVDV by nested RT-PCR. Unlike that heretofore described, the observed prevalence was 23.1% in 2007, 27.7% in 2008, 33.6% in 2009, and 23.6% in 2010, showing a high prevalence of BVDV-1 infection [9] when compared to the abovementioned study. These numbers should be analyzed carefully since only samples from clinical cases compatible with BVDV infection were analyzed. As stated by these authors, the use of live vaccines against CSFV may also be directly related to this higher prevalence of BVDV in pig herds from China, since, in general, the prevalence of BVDV in pig herds is low.

#### **3. Experimental Infection with BVDV in Pigs: Routes of Transmission and Disease Development**

The lack of studies concerning the routes of transmission of BVDV between piglets highlighted the need to develop researches clarifying this information (Table 2). Three studies were separately conducted with three groups of two weaned piglets separated by isolation cabinets: the challenged, sentinel, and control groups. The isolation cabinets were arranged to allow only a specific route of transmission, namely airborne, nose-to-nose [25], and by back pond water [26]. Although all challenged piglets shed the virus and seroconverted, only transmission by the back-pond water was confirmed since sentinel animals also shed the virus and seroconverted in this study. An interesting fact regarding the viral shedding observed in these studies was the intermittent pattern of nasal shedding. Challenged piglets shed the virus between 5- and 24-days post-inoculation, intermittently, detected by RT-PCR of nasal swabs. In the challenged groups, clinical signs such as diarrhea, rough hair, and oculo-nasal discharge were observed about 15 dpi, when the piglets started seroconversion. Despite other pathogens that have not been searched for, these clinical signs may be suggestive of BVDV infection since it was not observed in piglets from the control group.



These studies proved that BVDV can infect weaned piglets, which shed the virus by the nasal route, presented clinical signs, and seroconverted. The presence of BVDV in nasal secretions indicates that pigs can be a source of infection for other animals, especially if piglets become infected by high virulent strains of BVDV. In the literature, BVDV shedding by a persistently infected boar has also been reported [49], with the detection of the virus in oropharyngeal fluid, urine, and semen.

Although BVDV infection in young pigs can occur without clinical manifestation of the disease [32], other authors have reported the appearance of reproductive problems in pregnant gilts, such as abortions, birth of small piglets, stillborn and congenital persistently infected animals (PI) [22]. In cattle, during acute infection, viremia and viral shedding are usually transient and at low titers, but even so, they can result in vertical transmission [50]. BVDV-2 infection can lead to the occurrence of fetal malformations in cattle; however, if the fetus survives the infection long enough, non-specific changes in maturation may occur in the lymphoid tissues [51]. In bovine herds infected by BVDV, several fetal malformations were described comprising cerebellar hypoplasia, poor myelinization of the spinal cord, hydrocephalus, microcephaly, retinal atrophy or dysplasia, and many others [6]. Diseases affecting myelin sheath formation or nerve synapses alter electrical impulses in neurons, which may lead to tremors [52].

Congenital infection of piglets born from gilts infected with BVDV has been reported in some studies [22,23], in which piglets died between the 2nd and the 16th week of life with signs similar to that of CSF, in addition to showing growth retardation. In a study in which pregnant sows were inoculated with BVDV on the 35th and 45th day of gestation by intrauterine and intranasal route, respectively, nine piglets born from females infected by the intrauterine route and five born from animals infected by intranasal route were born persistently infected [22]. Other studies have also shown transplacental infection, with the virus isolated in at least one of the fetuses [32,53,54].

Conversely, Pereira [27] inoculated BVDV-2 in groups of pregnant gilts at different stages of gestation and before artificial insemination (AI). Seroconversion and a transient viremia were detected, but reproductive losses and clinical signs of the disease in gilts and piglets were not observed. Other recent studies analyzed groups of gilts inoculated by BVDV-2 by oronasal [28,29] and intrauterine [28] routes on the 45th day of pregnancy, before the fetal immunocompetence period. No transplacental transmission was observed since piglets from oro-nasally inoculated gilts were born BVDV-free; no anti-BVDV antibodies were detected in piglets at birth but were acquired by colostral passive transfer. Congenital persistent infection was not observed since piglets did not shed BVDV at any moment. Rates of BVDV transmission between pigs under field conditions is very low, and under experimental conditions it would be even more limited [33].

Regarding intrauterine inoculation [28], piglets were born with no clinical signs of infection and no signs of hypomyelination or CT. Surprisingly, high anti-BVDV-2 antibody titers were found. Serological investigations in bovine fetuses experimentally inoculated with the virus also indicated the development of specific immune competence before the period already established in the literature [51]. The average period of seroconversion of the gilts challenged with the virus was 20 days [28], varying between 12 and 33 days [27]. Other studies have described BVDV inducing viremia seven days post-infection and seroconversion three weeks after experimental inoculation in pigs [55–57]. BVDV was discarded as an etiological agent of CT [28], differently from atypical porcine pestivirus (APPV), which was linked to CT-disease in experimental and natural infection conditions [58–61].

Understanding the role of BVDV in the reproductive system of boars is valuable information, considering that biotechnological procedures have the expressive potential of spreading diseases to free herds [62]. When it comes to boars, the presence of agents of the genus Pestivirus has been confirmed in porcine semen. Shedding of CSFV in porcine semen was the first to be reported under natural and experimental infection [63], as well as virus transmission to sows and fetuses by AI and transplacentally, respectively [64]. Recently, APPV was also detected in the semen and preputial fluid of naturally infected boars, with a high viral load in semen [65]. Experimental infection with BVDV-2 did not result in changes in the post-period of pre-inoculation in most of the seminal characteristics

evaluated, and no viral shedding was detected in semen or preputial fluid, but lymphocytosis and monocytopenia were observed [30]. Considering a mild and transient viremia, the likelihood that the circulating virus in the blood reaches different organs was low. Also, the blood-testis barrier would decrease the chance of reaching semen, which may explain the absence of viral RNA detection in the reproductive tract of the inoculated boars [30]. A BVDV persistently infected boar presented viral shedding in the ejaculate, which contained no sperm cells [49]. Possibly, BVDV transmission by semen occurs in atypical cases of congenital persistent infection in pigs [49].

#### **4. Final Consideration**

The course of BVDV infection in pigs will depend on the virulence of the viral strain and the pig immune response [66] and may be limited [44]. Even so, the presence of the virus in the nasal secretions of infected animals demonstrated that pigs could act as a source of infection, thus facilitating the spread in the herd [26,27]. Although BVDV does not pose the same threat to pig herds as it poses to ruminants, it may lead to the development of a range of clinical signs and culminate in a serological cross-reaction with the CSFV, interfering negatively in classical swine fever monitoring and surveillance programs, and misleading diagnosis of the disease [10].

**Author Contributions:** All authors have made substantial contributions to this paper, including manuscript organization, writing, editing, and approving the final version. All authors have read and agreed to the published version of the manuscript.

**Funding:** We are grateful for the grants #2014/13590-3 and #2016/21421-2, São Paulo Research Foundation (FAPESP); grant 409435/2016-3 of National Council for Scientific and Technological Development (CNPq) and an M.L.M-D Master's scholarship #2016/02982-3, São Paulo Research Foundation (FAPESP).

**Acknowledgments:** The authors appreciate the support provided by Prof. Eduardo Furtado Flores Federal University of Santa Maria (UFSM) and Dra. Edviges Maristela Pituco (Biological Institute of São Paulo).

**Conflicts of Interest:** The authors declare no conflict of interest. At the time the studies were conducted, IRHG was part of the Graduate Program in Veterinary Medicine at the School of Agricultural and Veterinarian Sciences, and started working at Ourofino Animal Health after completing his Ph.D.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Clinical and Serological Evaluation of LINDA Virus Infections in Post-Weaning Piglets**

**Alexandra Kiesler 1,**†**, Kerstin Seitz 1,**†**, Lukas Schwarz 2,**†**, Katharina Buczolich 1, Helga Petznek 1, Elena Sassu 2, Sophie Dürlinger 2, Sandra Högler 3, Andrea Klang 3, Christiane Riedel 1, Hann-Wei Chen 1, Marlene Mötz 1, Peter Kirkland 4, Herbert Weissenböck 3, Andrea Ladinig 2, Till Rümenapf <sup>1</sup> and Benjamin Lamp 1,\*,**‡


Received: 17 September 2019; Accepted: 21 October 2019; Published: 23 October 2019

**Abstract:** The novel pestivirus species known as lateral-shaking inducing neuro-degenerative agent (LINDA) virus emerged in 2015 in a piglet-producing farm in Austria. Affected piglets showed strong congenital tremor as a result of severe lesions in the central nervous system. Here, we report the results of a controlled animal infection experiment. Post-weaning piglets were infected with LINDA to determine the susceptibility of pigs, the clinical consequences of infection and the humoral immune response against LINDA. No clinically overt disease signs were observed in the piglets. Viremia was hardly detectable, but LINDA was present in the spleen and several lymphatic organs until the end of the experiment on day 28 post-infection. Oronasal virus shedding together with the infection of one sentinel animal provided additional evidence for the successful replication and spread of LINDA in the piglets. Starting on day 14 post-infection, all infected animals showed a strong humoral immune response with high titers of neutralizing antibodies against LINDA. No cross-neutralizing activity of these sera with other pestiviral species was observed. According to these data, following postnatal infection, LINDA is a rather benign virus that can be controlled by the pig's immune system. However, further studies are needed to investigate the effects of LINDA on the fetus after intrauterine infection.

**Keywords:** Linda virus; serological profile; virus neutralization assay; virus pathogenicity; humoral immune response

#### **1. Introduction**

The genus *Pestivirus* within the family Flaviviridae currently comprises 11 different species recently termed *Pestivirus A–K* [1]. In addition to the long known classical swine fever virus (CSFV, *Pestivirus C*), a number of other pestivirus species have been identified in the porcine host in recent years, such as border disease virus (BDV, *Pestivirus D*), bovine viral diarrhea virus (BVDV-1, *Pestivirus A*), Bungowannah virus (BUNGO, *Pestivirus F*), and atypical porcine pestivirus (APPV, *Pestivirus K*) [1–5]. In 2015, we detected a yet unknown pestivirus species in a piglet-producing farm in Austria, which was termed lateral-shaking inducing neuro-degenerative agent (LINDA) virus [6]. Since the nucleotide sequence of LINDA shows a significant divergence of over 30% compared to the accepted pestivirus species, we proposed it as the new species *Pestivirus L* [1,6].

Pestiviruses are small enveloped viruses with a positive-sense, single-stranded, non-segmented RNA genome with a length of about 12 to 13 kilobases (kb) [7]. The genome consists of one large open reading frame (ORF), flanked by 5 - and 3 -non-coding regions [7]. This single ORF encodes a hypothetical polyprotein, that is co- and post-translationally processed into non-structural and structural proteins by viral and cellular proteases [8]. The three structural glycoproteins, termed Erns, E1 and E2, and the nucleocapsid protein named Core are generated by cellular proteases [9,10]. The generation of the non-structural proteins Npro, p7, NS2, NS3, NS4A, NS4B, NS5A and NS5B is very complex. Multiple processing steps mediated by autoproteases (Npro and NS2) and the major NS3/4A protease yield partially processed precursors, mature proteins and enzymatically active protein fragments [8,11–13]. The presence of the autoprotease Npro and the envelope glycoprotein Erns are recognized as characteristic of the genus *Pestivirus* [1,7]. Since the corresponding proteins have been found in the genome of LINDA, it can undoubtedly be classified in the genus *Pestivirus* [6].

CSFV is listed by the World Organization for Animal Health (OIE) as an economically important pig pathogen [14]. The clinical signs of classical swine fever (CSF) vary significantly depending on the virulence of the virus strain as well as the age and susceptibility of the infected pigs. CSF is usually characterized by fever, skin lesions, convulsions and, especially in young animals, death within a few days [15]. BUNGO emerged on a pig farm in Australia in 2003, causing an increased rate of stillbirths, mummification and sudden deaths of piglets [2,16]. Experimental studies were conducted to investigate the pathogenicity of BUNGO in weaner pigs and porcine fetuses under laboratory conditions. Despite the low pathogenicity of the virus in weaned piglets, a long-lasting viremia, efficient virus shedding and rapid seroconversion were detected [17]. In contrast, a multifocal non-suppurative myocarditis with myonecrosis was observed following direct fetal exposure to BUNGO mimicking intrauterine infection [18]. APPVs were discovered in the United States in 2015 by next-generation sequencing [4], and subsequently detected in many countries around the world [19–23]. A close correlation between intrauterine APPV infections and the occurrence of congenital tremor (CT) type A-II in newborn piglets was reported [24]. The simultaneous detection of nucleic acids of APPV and hypomyelination in the central nervous system of these piglets implied a causative role of APPV for the appearance of the so-called shaking piglet syndrome [20]. This causal relationship is further supported by the birth of shaking piglets after inoculation of pregnant sows with APPV-containing material [24].

LINDA was discovered during the investigation of an outbreak of CT in a piglet-producing farm. We identified the agent, isolated the virus, sequenced its genome and established a RT-PCR assay as well as serological reagents for its detection [6]. Since then, LINDA has not been found in any other farm in Austria or elsewhere in the world [25]. To gain a deeper insight into the biology of this virus, we infected weaned piglets with LINDA under controlled experimental conditions. The aim of this small-scale animal experiment was the determination of susceptibility, pathogenicity and virulence of LINDA in the immunocompetent porcine host. Sera from the experimentally infected piglets were further used to characterize the humoral immune response against LINDA and to study the induction of cross-neutralizing antibodies against other pestiviruses.

#### **2. Materials and Methods**

#### *2.1. Cells and Viruses*

SK-6 cells [26] and MDBK cells (ATCC® CCL-22™) [27] were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FCS, Bio and Sell GmbH, Feucht, Germany; negatively tested for the presence of pestiviruses), 100 U/mL penicillin and 100 μg/mL streptomycin. All cells were maintained at 37 ◦C and 5% CO2. Cell culture-derived LINDA was used for the experiments. After initial cell culture isolation from a clinical case (passage 1, P1), a primary LINDA stock was generated containing a 50% tissue culture infectious dose (TCID50) of 1.1 <sup>×</sup> 107 (P2, GenBank® KY436034) [6]. The virus was titrated in an endpoint dilution assay and supernatant from a single focus was harvested (P3) to ensure freedom from other pathogens. A master stock (P4) was prepared and characterized by RT-PCR and subsequent Sanger sequencing. In direct comparison with the consensus sequence of the original isolate no mutations were detected. All LINDA infection doses used for animal inoculations were recovered from the master stock and thus represent cell culture passage 5 of LINDA. BVDV-1a strain NADL and BVDV-1b strain NCP7 were obtained from E. Dubovi (Cornell University College of Veterinary Medicine, Ithaca, NY) [28]. CSFV 2.3 Alfort-Tübingen [29], BDV-1 X818 [30], BVDV-2 strain 890 [31], and BVDV-3 (unpublished strain isolated from FCS, South American origin) were obtained from the virus collection of the Institute of Virology in Giessen (Justus-Liebig-University, Giessen, Germany). Pestivirus strain Giraffe-1 [32] was a gift from D. J. Paton, Animal Health and Veterinary Laboratory Agency (AHVLA, Weybridge, United Kingdom). BUNGO was obtained from stocks of the Elizabeth Macarthur Agricultural Institute (Department of Primary Industries, Menangle, New South Wales, Australia) [2].

#### *2.2. Virus Infection and Titration*

Infections of MDBK and SK-6 cells with various strains of the different pestivirus species were performed with the indicated multiplicity of infection (MOI). Virus stocks for the experiments were generated using 10-cm cell culture dishes infected with a MOI of 0.1. At 72 h post-infection, the cell culture supernatant was harvested, filtered through a 0.45 μm cellulose filter (Sartorius, Göttingen, Germany), aliquoted and stored at −80 ◦C until use. The TCID50 of viral supernatants was determined in three replicates by an end-point dilution assay (EPDA). The virus titer was calculated using the Spearman-Kaerber algorithm [33]. Re-isolation of LINDA was performed using SK-6 cells and fresh sample material.

#### *2.3. RT-PCR Detection*

RNA was extracted from serum and tissue samples, saliva, feces, cultured cells or virus cell culture supernatant using the QIAamp Viral RNA Mini Kit and the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. RNA was eluted in 60 μL RNase free distilled water and directly used for RT-PCR or stored at –80 ◦C for subsequent analysis. RT-PCR was carried out using the OneTaq One-Step RT-PCR Kit (NEB, Ipswich, USA) or the One Step RT-PCR Kit (Qiagen) using the oligonucleotides PPF 5 -GTKATHCAATACCCTGARGC-3 and PPR 5 -GGRTTCCAGGARTACATCA-3 [6]. PCR amplicons were subjected to gel electrophoresis and purified with the peqGOLD Gel Extraction Kit (Peqlab, Erlangen, Germany), if needed.

#### *2.4. Calibration-Curve-Estimated Copy Number of LINDA Genome Equivalents*

For the quantification of virus excretion, viremia and virus loads of organ homogenates, qRT-PCRs were performed on an ABI 7500 cycler (Applied Biosystems, Foster City, USA) using the LINDA and BUNGO specific primers LVqRTfor196 (5 -CACTGGWAAGGATCACCCACT-3 ) and LVqRTrev351 (5 -AATYACAACGGATAWTMTTTATACTGG-3 ) and the FAM/TAMRA labeled probe, LVqRTprobe322 (5 -Fam-ATAGGATGCCGGCGGATGCCCTGT-TamRa -3 ). For the generation of a calibration curve, a T-vector plasmid fragment harboring the cDNA target sequence was produced using EcoRI digest, gel-purified and spectrophotometrically quantified. The copy number of recombinant plasmid fragment was calculated following the formula: N (molecules per μL) = (C

(DNA concentration in <sup>μ</sup>g/μL) / K (fragment size in bp)) <sup>×</sup> 185.5 <sup>×</sup> 1013. The factor results from the weight of the genome-equivalent ssDNA molecules (330 daltons per base), the volume projection factors and the Avogadro constant (6.02 <sup>×</sup> 1023 mol–1). In order to obtain a standard curve, a ten-fold dilution series of the DNA control was included in the qRT-PCR setup [34]. Cycling conditions were 50 ◦C for 10 min, 95 ◦C for 1 min and 40 cycles of 95 ◦C for 10 sec, 60 ◦C for 1 min (amplification and fluorescence detection step). Semi-quantitative copy number estimation was calculated by 7500 System SDS Software (Applied Biosystems) based on the calibration curve. The amplification of the dsDNA standard does not include reverse transcription, which is why minor deviations may occur between RNA samples and dsDNA standard amplification depending on the efficiency of the cDNA synthesis. The qRT-PCR assay presented here is linear up to 10 copies of the dsDNA template per reaction (Ct value of 38). Since 10 copies per reaction were included as the lowest template amount in the standard series, this value represents the limit of quantification and our lower assay cutoff. The genome equivalents of 1.0 μL of purified RNA were converted to copies per 1 mL liquid sample, 1 g tissue sample or copies per swab using appropriate projection factors. The samples from the animal experiment were measured once due to high numbers and limited financial resources. However, the qRT-PCR data were validated by virus isolation experiments carried out in parallel with all samples obtained during the experiment.

#### *2.5. Virus Neutralization Assay*

All virus neutralization assays (VNA) were prepared in triplicate in 96-well microtiter plates. Viruses used in the VNA were diluted in DMEM without FCS from stock solutions generating a final titer of 100 to 300 TCID50 per 0.1 mL. Initial two-fold dilution series of the serum samples were prepared with DMEM without FCS generating a final serum dilution of 1:256 in the last wells. Highly reactive sera were further diluted in five-fold series to a final serum dilution of 1:105.6 in the last wells. 100 μL of each serum dilution were mixed with 100 μL of the respective virus solution containing between 100 to 300 TCID50. After the virus was added, the VNA was incubated for 2 h at 37 ◦C in 5% CO2. One hundred microliters of this virus/serum mixtures were added to 96-well flat bottom plates containing confluent cell monolayers and incubated for 48 h at 37 ◦C in 5% CO2. Viral infections were detected by indirect immunofluorescence using murine monoclonal antibodies (MAbs) as indicated below. Back titration of each virus solution was performed in parallel. Defined positive and negative sera against the respective virus or groups were used as controls. The titers obtained from the VNAs were calculated using the Spearman-Kaerber algorithm and reported as the reciprocal of the serum dilution that inhibited infection of 50% of the cells (neutralization dose 50%, ND50).

#### *2.6. Indirect Immunofluorescence Assay and Antibodies*

The immunofluorescence assays were performed as previously described [8]. Briefly, the cells were fixed with 4% paraformaldehyde for 20 min at 4 ◦C, permeabilized with 1% (vol/vol) Triton-X 100 (Merck, Darmstadt, Germany) in PBS and stained with the mouse MAb 6A5 [35] and A18 [36]. The monoclonal antibody 6A5 was used to detect the E2 molecule of BVDV-1, BVDV-2, BVDV-3, BDV, BUNGO, giraffe pestivirus and LINDA infections. CSFV E2 was detected by MAb A18. Goat anti-mouse IgG conjugated with Cy3 (Dianova, Hamburg, Germany) was used as a secondary antibody. A porcine BUNGO antiserum (748-09.10-1) collected from a naturally infected sow, which had produced an abnormal litter, was kindly provided by the Elizabeth Macarthur Agricultural Institute.

#### *2.7. Animal Experiment*

The animal experiment was approved by the ethics committee of the University of Veterinary Medicine, Vienna and the Federal Ministry of Science, Research and Economy according to the §§ 26ff. of the Austrian Animal Experiments Act from 2012 (Permission code: BMWF-68.205/0130-WF/V/3b/2017, permission date: 07.07.2017). Post-weaning crossbreed piglets (*Sus scrofa domestica*) were used for the study of LINDA pathogenesis and virulence in immunocompetent hosts. The pigs came from the breeding farm Medau of the University of Veterinary Medicine, Vienna. The entire herd of the

farm was negative tested for swine pathogenic influenza A viruses (IAV; H3N2, H1N1, H1N2 and H1N1pan) and porcine reproductive and respiratory syndrome viruses (PRRSV). The mother sows of the experimental animals were vaccinated against parvovirus and erysipelas during lactation according to the manufacturer's instructions (Parvoruvac; Merial GmbH, Hallbergmoos, Germany). The piglets themselves were protected on day 21 with a combined vaccination against *Mycoplasma hyopneumoniae* and porcine circovirus-2 (Circoflex and Mycoflex; both from Boehringer Ingelheim Vetmedica GmbH, Ingelheim, Germany). No further special diagnostic tests were carried out as there were neither pathological nor clinical signs of disease in these piglets.

A week prior to the beginning of the trial (study days –7 to –1) 21 weaned piglets in the 13th week of life were housed in a biological safety unit (BSL-2) for adaptation. At the beginning of the animal experiment, six piglets were housed separately in order to later serve as sentinel animals. The remaining piglets were divided into three groups of five animals each and housed in separate units. One group was not infected and served as a negative control, one group was inoculated intramuscularly (i.m.) with 1 <sup>×</sup> 10<sup>7</sup> TCID50 LINDA and the last group was infected intranasally (i.n.) with 1 <sup>×</sup> 10<sup>7</sup> TCID50 LINDA (study day 0). One day after infection (study day 1), three sentinel animals were added to each of the infection-groups. A daily clinical score was determined for each individual animal. The general condition, behavior, body temperature, feed intake and weight gain of all animals were assessed and measured after the infection over a period of 28 days. Particular attention was paid to the occurrence of signs of disease. Each animal was assigned a daily clinical score, which included individual values for behavior, feed intake, dyspnea, ocular and nasal discharge, coughing and diarrhea in the range from 0 (physiological) to 3 (severe clinical symptoms) according to an established evaluation scheme of the University Clinic for Swine of the University of Veterinary Medicine, Vienna [37]. While the body temperature of all animals was also monitored daily, the body weight was assessed at the time point of arrival and on study days 0–7, 9, 14, 21 and 28. Blood and fecal samples as well as nasal and oral swabs were taken on the study days 0, 3, 5, 7, 14, 21 and 28/29. Urine samples (spontaneous urine samples or collected via cystocentesis) were obtained on study day 3 from most animals. All animals were euthanized with T61 (5.0 mg/mL tetracaine hydrochloride, 50 mg/mL mebezonium iodide and 200 mg/mL embutramide; 1 mL/10 kg) on study day 28 or 29 under general anesthesia (1.3 mg/kg azaperone and 10 mg/kg ketamine hydrochloride). During necropsy, organ samples were taken for molecular and pathohistological analysis. In particular, the LINDA RNA loads were analyzed in samples taken from the kidney, bladder, cerebellum, cerebrum, spinal cord, dorsal root ganglia, thymus, spleen, tonsils, lymph nodes (Lnn. inguinales, mesenteriales and mandibularis), parotid and sublingual glands, heart, lung, liver and intestinal segments from all animals.

#### *2.8. Pathological Examinations*

Tissue samples of the brain and spinal cord, dorsal root ganglia, liver, spleen, kidney, urinary bladder, thymus, spleen, tonsils, inguinal lymph node and mandibular gland were taken from all animals for histological examination. Additional samples of the coeliac ganglion, sciatic nerve, mesenteric and mandibular lymph nodes and parotid gland were included from the experimentally infected animals and sentinels. Five coronary sections of the brain were taken at the levels of telencephalon, diencephalon, cerebellum, mesencephalon and metencephalon. Transversal sections of the spinal cord were taken from cervical, thoracic and lumbar regions. The organ samples were fixed in 10% neutral buffered formalin, embedded in paraffin wax, sectioned at 2 μm and stained with hematoxylin and eosin (HE).

#### **3. Results**

#### *3.1. Pathogenicity and Virulence of LINDA in Weaned Piglets*

The animal experiment was performed with weaned piglets to investigate the clinical effects of LINDA infection in the immunocompetent host. The animals of all three groups (infected intranasally, intramuscularly or mock), including the sentinel animals, showed normal physiological parameters

and good general condition throughout the 28-day infection period. Neither the individually examined clinical values nor the additive clinical score showed major pathological changes in individual animals or significant changes between the different groups (Figure 1). Mild ocular and nasal discharge and cough were observed early after infection (study days 0 to 7) in most animals of both LINDA infected groups. However, these changes were also observed in sentinel animals of these groups that did not show seroconversion to LINDA (n.i. sent., described below). Mild fever (maximum rectal temperature of 40.7 ◦C) occurred in all i.m. infected animals within the first three days after inoculation. The single sentinel animal that later seroconverted showed an elevated body temperature and mild diarrhea. No differences were found between the groups in necropsy and histological examination. Gross examination revealed alveolar emphysema of the lung in animals of all the groups. Alveolar edema and pulmonary hyperemia were detected in some animals of all groups. Macroscopically, the central nervous system of all animals appeared normal. However, in the histological examination the majority of the animals from all groups (*n* = 17) showed mild, oligofocal, randomly distributed perivascular, mononuclear infiltrations and some glial nodules in the brain and/or spinal cord. A slight follicular hyperplasia of the spleen was observed in one sentinel animal of the i.n. and i.m. group. Stomach lesions such as ulcerations, hyperkeratinization or follicular hyperplasia were found in almost all animals. Mild, predominantly mononuclear, sometimes suppurating, interstitial nephritis or cortical infarction were found in many animals (*n* = 13) including animals from the non-infected control group. An interstitial lymphocytic infiltration of the mandibular gland was evident in one sentinel animal each of intranasally and intramuscularly infected group. An abscess was observed in the snout of an intranasally infected animal. The initial body weight of each animal was defined as 100% and the relative weight gain for each piglet and the average weight gain of the groups were calculated for each study day. The non-infected control group showed a slightly higher weight gain compared to the infected groups and the sentinel animals. However, the differences in weight gain between the experimental groups were neither pronounced nor statistically significant (Figure 2).

**Figure 1.** Pathogenicity of lateral-shakinginducing neuro-degenerative agent (LINDA)inimmunocompetent piglets. A total of 21 piglets were divided in five groups: five negative control, five intramuscularly (i.m.) infected and five intranasally (i.n.) infected animals as well as three sentinel animals for the i.m. and three sentinel animals for the i.n. group. The animals from the i.m. and i.n. infection groups were inoculated with LINDA (1 <sup>×</sup> 107 TCID50/mL) on study day 0. Body temperature, behavior, feed intake, dyspnea, ocular discharge, coughing, nasal discharge, diarrhea and sneezing were assessed daily and symptoms were classified by a scoring system with scores from 0 (physiological) to 3 (severe clinical symptoms). The mean of gathered parameters was calculated for the control, i.m. infected, i.n. infected, infected sentinel (inf. sent.) and non-infected sentinel (n.i. sent.) animals. No severe LINDA virus associated clinical signs were observed comparing infected and non-infected animals. However, mild fever and other signs of disease were seen in some infected animals, such as the infected sentinel animal within the i.m. group.

**Figure 2.** Average relative daily weight gain. Piglets were weighed on study days 0–7, 14, 21 and 28/29. The initial body weight was set as 100% value and relative weight gain was calculated for every individual at the indicated time-points. The mean value was calculated for every trial group. Infected groups show a slightly reduced weight gain compared to the negative control group after 7 days post-infection. However, the absolute differences between the groups were not significant.

#### *3.2. Replication of LINDA in the Immunocompetent Porcine Host*

Before the beginning of the experiment, blood samples of all animals were taken and tested for the presence of LINDA RNA and LINDA neutralizing antibodies. Serum samples were taken on study days 0, 3, 7, 14 and 28/29 of the experiment and analyzed in a LINDA-specific qRT-PCR assay as well as in virus isolation studies. No infectious LINDA or viral RNA were detected in the mock infected group at any time of the experiment. Most i.m. infected and all i.n. infected animals did not show a detectable viremia after LINDA inoculation. We observed a low level of viremia in two i.m. infected animals (5.0 <sup>×</sup> 105 GE/mL for animal 8 i.m. and 5.5 <sup>×</sup> 105 for animal 9 i.m. on study day 7) and a higher level in one sentinel animal of this group (2.3 <sup>×</sup> 107 GE/mL on study day 14, Figure 3). LINDA could be isolated from each of these qRT-PCR positive serum samples. Virus shedding was assessed using oral and nasal swabs as well as fecal and urine samples. Most of these samples gave negative results in the LINDA virus-specific qRT-PCR and virus isolation experiments. The oral swabs of one of the viremic piglets from the i.m. group (study day 7) and the oral and nasal swabs from three animals of the i.n. group gave signals below assay cutoff in the qRT-PCR assay (study days 3, 7 and/or 14). The RNA loads in these swabs were very low and cell culture virus isolation was not successful. An additional conventional RT-PCR was performed on the questionable samples as described before [6]. The amplification of LINDA-specific RT-PCR products was verified by nucleotide sequencing (Figure S1). LINDA RNA was not detectable in the fecal samples obtained on study days 3, 7 and 14 from the experimentally infected animals. However, the viremic sentinel animal (number 11) from the i.m. infected group showed substantial LINDA RNA loads in the feces (1.31 <sup>×</sup> <sup>10</sup><sup>5</sup> GE/<sup>g</sup> on study day 14) that also allowed successful virus isolation. All urine samples gave negative results in qRT-PCR and virus isolation. Multiple organs were sampled during necropsy on days 28/29 post-infection. Most organ samples gave negative results for the presence of LINDA RNA in the qRT-PCR. However, LINDA genomes were detectable in several lymphoid organs, such as the inguinal lymph nodes (*n* = 4), spleen (*n* = 2) and tonsils (*n* = 4) of animals of the i.m. infected group reaching values between 4.0 <sup>×</sup> 103 and 2.3 <sup>×</sup> 106 GE/g. LINDA RNA was also found in the infected sentinel animal of the i.m. group (animal 11) in the inguinal lymph nodes and spleen. Interestingly, the virus was detectable in the tonsils of all i.n. infected animals (*n* = 5), but only found in the inguinal lymph node of one of these animals (*n* = 1, Figure 3).

**Figure 3.** qRT-PCR results of the animal experiment. Multiple samples were analyzed for the presence of LINDA RNA before, during and after the experimental infection of piglets with LINDA. Blood and fecal samples as well as nasal and oral swabs were taken on study days 0, 3, 5, 7, 14, 21 and 28/29. Fecal samples were collected on study days 3, 7 and 14, while urine samples were only taken from most animals on day 3. Kidney, bladder, cerebellum, cerebrum, spinal cord, dorsal root ganglia, thymus, spleen, tonsils, different lymph nodes (Lnn. inguinales, mesenteriales and mandibularis), parotid and sublingual glands, heart, lung, liver and intestinal segments were sampled from all animals during necropsy on days 28/29. Most samples gave negative results for LINDA RNA. Only two i.m. infected animals showed detectable LINDA RNA loads in serum samples taken on study day 7 and one sentinel animal of the i.m. group on day 14. These serum samples also allowed the re-isolation of LINDA. Viral RNA was not detectable in urine samples and only detected in one fecal sample (animal 11, i.m. group, infected sentinel). However, the RNA of LINDA was detected in multiple samples from lymphoid organs demonstrating the presence of LINDA in the experimentally infected animals as well as in the infected sentinel animal until the end of the experiment.

#### *3.3. Humoral Immune Response against LINDA*

All serum samples obtained before, during and at the end of the experiment were tested for LINDA-specific antibodies using a LINDA virus neutralization assay (Figure 4). No virus neutralizing activity was measured in the serum samples taken before the start of the experiment (ND50 < 1/2, below limit of detection). Sera from the mock infected group and sera from the LINDA RNA negative sentinel animals showed no virus neutralizing activity at the end of the trial. A gradual onset of humoral immune responses was observed in all infected animals (i.m. and i.n.) between days 7 and 14 post-infection, reaching peak values of up to 1/8,640 ND50/mL. A comparably strong reactivity of 1/1,028 was also seen in the serum sample of the i.m. infected animal 6, in which no LINDA RNA replication was detected throughout the experiment. Interestingly, the development of the humoral immune response was delayed to study day 21 in the LINDA infected sentinel animal 11.

**Figure 4.** Virus neutralization activities of LINDA immune sera. The serum samples taken on study days 0, 3, 7, 14, 21 and 28/29 were tested in VNAs against LINDA. All VNAs were performed in triplicate and ND50 was calculated using the Spearman-Kaerber algorithm. The reciprocal ND50 value is presented for each serum sample including error bars for positive and negative standard deviation. All 1/ND50 values were less than 1 until study day 7, when the first weak neutralizing activities were measured in single animals. Significant neutralizing activities (ND50 > 1:10) were detected in all infected animals starting on day 14. Note the later onset of humoral immune response in the infected sentinel animal from the i.m. group on day 21.

#### *3.4. Cross Neutralization of LINDA-Immune Sera with Other Pestivirus Species*

To assess antigenic relations to other pestiviruses, we characterized the LINDA-immune sera from study days 28/29 for their cross-neutralizing activities against multiple pestiviral strains. In particular, we performed VNAs with the pestivirus species BVDV-1a (strain NADL), BVDV-1b (strain NCP7), BVDV-2 (strain 890), CSFV 2.3 (strain Alfort-Tübingen), BDV-1 (strain X818), pestivirus giraffe (strain giraffe 1), BUNGO and an unpublished BVDV-3 strain. We found no neutralizing activity of the LINDA-immune sera (ND50 < 1/2) against any of these viruses. Additionally, we tested a porcine BUNGO convalescence serum (748-09.10-1), initially obtained for immunodetection of BUNGO infections of cultured cells. This antiserum efficiently neutralized BUNGO (BUNGO ND50 1/3200) as well as LINDA in our VNAs (LINDA ND50 1/1,600) but had no effect on the infection of BVDV-1, BVDV-2 and CSFV (all ND50 < 1/2).

#### **4. Discussion**

The aim of this study was to investigate the pathogenicity of LINDA during post-natal infections and to characterize the humoral immune response against LINDA in order to obtain important basic data for sero-surveillance studies. Therefore, a small-scale animal experiment was set up using LINDA infections in post-weaning piglets. In this experiment, we found no evidence of severe acute disease in weaners caused by LINDA infection. Only mild clinical signs, such as mild fever, nasal discharge and mild changes in fecal consistency, were observed in single animals without significant influence on the growth rates of these piglets. Inflammatory infiltrates and glial nodules occurred in the brain and/or spinal cord of animals from all groups including the control group. Hence, these lesions were not associated with LINDA infection and were consequently interpreted as non-specific experimental background. Other findings of necropsy could be interpreted as random findings or might represent

stress-related diseases, such as the typical gastric lesions. However, productive LINDA infections were observed in all animals, which were experimentally inoculated with 1 <sup>×</sup> 107 TCID50, regardless of the infection route used. LINDA virus excretion was detected using qRT-PCR in oronasal fluids of some infected animals and excretion via feces was documented for a single animal. Viremia was hardly detectable in the serum of most animals, while all except one experimentally infected animal (animal 6 i.m.) showed a long-lasting presence of the virus in the tonsils and/or lymphatic organs. Unfortunately, no peripheral mononuclear blood cells (PBMCs) were preserved from the experiment that could have been used for a potentially more sensitive virus detection in the blood [38]. Therefore, future studies, also including pregnant sows, will investigate whether LINDA virus is associated with PBMCs and whether the analysis of isolated PBMCs allows a more sensitive diagnosis. The infection of one sentinel animal in the i.m. group confirmed that infectious loads of virus were secreted in this simulated acute infection scenario. When we consider the positive detection of LINDA in serum, nasal secretions and feces of some infected animals, blood contacts or oronasal uptake of the pathogen may be considered as possible pathways of infection as also demonstrated by the successful artificial i.m. and i.n. infection. Since the minimum infection dose of LINDA for piglets is unknown, it is up to future studies to clarify the natural routes of infection. The clinical data are in accordance with clinical and epidemiological data about other pestiviruses, such as BVDV, BDV or CSFV, showing that mononuclear cells and lymphoid organs are the primary targets of pestivirus infections [39,40]. Most pestiviruses are well-adapted to their host species and the acute infection of immunocompetent animals usually leads to mild to subclinical disease with limited virus replication [41,42]. Severe pestiviral disease, such as abortion, malformation or neurological disorders, is mostly a consequence of intrauterine infections and responsible for the high economic losses following pestiviral infections [43]. This circumstance also explains the differences in the clinical picture between the animal experiment with immunocompetent piglets presented here and the LINDA virus outbreak in a piglet breeding farm from which LINDA virus was originally isolated [6]. Similar results were obtained in infection studies with BUNGO in weaner piglets, where few clinical signs, a short phase of viremia (3 to 10 days post-inoculation) and low levels of virus excretion were observed [17]. We conclude that LINDA is not only the closest genetic relative of Bungowannah virus but also shows similar pathogenicity in pigs. The results of our post-natal experimental infection demonstrate that the clinical picture of BUNGO and LINDA in immunocompetent animals is distinct from symptoms of piglets affected by CSFV strains of high and moderate virulence but might be similar to low virulence CSFV strains [44]. Future studies evaluating LINDA infections of the unborn fetus will be necessary to assess the potential hazard of LINDA [18,45].

The acute infection of piglets with LINDA led to the development of a strong humoral immunity starting at about seven days post-infection (Figure 4). Despite high titers of neutralizing antibodies, viral RNA persisted in the tonsils and lymphoid organs as has also been shown for other pestiviruses. This phenomenon has been described for several pestiviruses and studied in detail for BVDV pointing to a potential risk of virus transmission from convalescent animals [46]. We used the generated LINDA-antisera to evaluate antigenic cross-reactivity between LINDA and other pestiviruses. Unfortunately, we could not include APPV in these tests, because no APPV strain was available that showed the necessary infectivity in the cell culture [19]. Our VNA data clearly demonstrate that antibodies from acute LINDA infections do not provide protection against infections with classical pestiviruses. Therefore, a serological interference with established VNAs for CSFV diagnosis is unlikely. However, further studies are needed to evaluate possible false positive reactions in serological CSFV tests using a larger sample size of LINDA-antisera as well as highly reactive immune sera obtained from sows infected during pregnancy with viremic, persistently infected piglets. Our data support the hypothesis that LINDA forms an independent species (*Pestivirus L*) within the genus *Pestivirus* with a highly divergent antigenic profile. Interestingly, the high titer BUNGO-antiserum showed a strong cross-neutralization activity in VNA against LINDA, while low to absent neutralization profiles of BUNGO-antisera against other pestiviruses were observed in previous studies [47]. This result is puzzling, because it is in conflict with the results obtained with the LINDA-antisera. A possible

explanation could be that the species BUNGO and LINDA form an antigenic group within the genus *Pestivirus*, sharing conserved antigenic motifs important for virus neutralization. Again, future studies may answer this question by analyzing a larger sample size of BUNGO- and LINDA-antisera in cross-protection VNAs and, even more importantly, in controlled animal experiments.

#### **5. Patents**

The authors B.L., L.S. and T.R. are inventors on a patent on Linda pestivirus (PCT/EP2017/084453; Isolation of a novel pestivirus causing congenital tremor).

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1999-4915/11/11/975/s1, Figure S1: Detection of LINDA virus by conventional RT-PCR in oro-nasal fluids.

**Author Contributions:** Conceptualization, K.S., L.S. and B.L.; Data curation, A.K. (Alexandra Kiesler), H.P., S.D, H.-W.C. and B.L.; Formal analysis, A.K. (Alexandra Kiesler), K.S., L.S., K.B. and B.L.; Funding acquisition, T.R. and B.L.; Investigation, A.K. (Alexandra Kiesler), K.S., L.S., K.B., H.P., E.S., S.D., S.H., A.K. (Andrea Klang), C.R., H.-W.C., M.M., P.K., H.W., A.L., T.R. and B.L.; Methodology, K.S., L.S., K.B., H.P., E.S., S.D., S.H., A.K. (Alexandra Kiesler), C.R., H.-W.C., M.M. and T.R.; Project administration, B.L.; Resources, P.K., H.W., A.L., T.R. and B.L.; Supervision, H.W., A.L. and B.L.; Validation, A.K. (Alexandra Kiesler), L.S., H.P. and B.L.; Visualization, A.K. (Alexandra Kiesler) and B.L.; Writing—original draft, A.K. (Alexandra Kiesler), T.R. and B.L.; Writing—review and editing, A.K. (Alexandra Kiesler), L.S., T.R. and B.L.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors acknowledge Markus Cihar who was responsible for animal care and supporting during sampling. The authors thank Boehringer Ingelheim Vetmedica GmbH (Binger Strasse 173, 55216 Ingelheim am Rhein, Germany) for their financial support (Contract number: 396806) and Eric Vaughn (Boehringer Ingelheim Animal Health USA Inc., 2412 S Loop Dr, Ames, IA, United States) for valuable scientific discussions.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Review*

### *Pestivirus K* **(Atypical Porcine Pestivirus): Update on the Virus, Viral Infection, and the Association with Congenital Tremor in Newborn Piglets**

### **Alais M. Dall Agnol 1,2,3, Alice F. Alfieri 1,2,3 and Amauri A. Alfieri 1,2,3,\***


Received: 4 July 2020; Accepted: 23 July 2020; Published: 18 August 2020

**Abstract:** The atypical porcine pestivirus (APPV) belongs to the species *Pestivirus K* of the genus *Pestivirus* and the family *Flaviviridae*, and it has been associated with congenital tremor (CT) type A-II in newborn piglets. Although APPV was discovered in 2015, evidence shows that APPV has circulated in pig herds for many years, at least since 1986. Due to the frequently reported outbreaks of CT on different continents, the importance of this virus for global pig production is notable. Since 2015, several studies have been conducted to clarify the association between APPV and CT. However, some findings regarding APPV infection and the measures taken to control and prevent the spread of this virus need to be contextualized to understand the infection better. This review attempts to highlight advances in the understanding of APPV associated with type A-II CT, such as etiology, epidemiology, diagnosis, and control and prevention measures, and also describes the pathophysiology of the infection and its consequences for pig production. Further research still needs to be conducted to elucidate the host's immune response to APPV infection, the control and prevention of this infection, and the possible development of vaccines.

**Keywords:** APPV; pestiviruses; congenital tremor type A-II; persistent infection; pigs

#### **1. Introduction**

In 2015, the atypical porcine pestivirus (APPV) was first described and identified by next-generation sequencing (NGS) in pig serum samples [1]. Initially, the virus was not believed to be associated with any clinical manifestations; later, an experimental inoculation study conducted in the United States (US) demonstrated that APPV was associated with the occurrence of congenital tremor (CT) in newborn piglets [2]. At present, the occurrence of APPV associated with CT is frequently reported by several studies on different continents, including the Americas (North and South), Europe, and Asia [3,4].

Since 2015, several studies have been conducted to clarify the association between APPV and CT. However, several findings regarding the infection and the measures taken to control and prevent the spread of the virus need to be contextualized to characterize the infection better. This review aims to address advances in understanding APPV associated with CT type A-II and further describe the pathophysiology of this virus and its consequences for pig production.

#### **2. Congenital Tremor Syndrome**

CT is a neurological disorder that affects newborn piglets and is characterized by muscle spasms in the head and body, which can be localized or generalized [5,6]. The intensity of the tremors is variable, and in more severe cases, they include generalized tremors that result in difficulty in standing or walking and, consequently, inability to suckle, resulting in death due to starvation. The clinical manifestation of CT is also called shaker pig syndrome, trembling pigs, or congenital myoclonus [6,7].

CT syndrome is classified into A and B according to the presence or absence of histopathological lesions in the central nervous system (CNS). In CT type A, histopathological changes are found in the brain or spinal cord. When histopathological lesions are not observed, it is classified as CT type B. Based on the etiology of CT type A, this type is subdivided into five subgroups (I-V) [5]. CT type A-I is characterized by cerebellar hypoplasia, dysplasia, hypomyelination of the spinal cord, and its etiology associated with other porcine pestivirus infections in the classical swine fever virus (CSFV) [8].

For a long time, the cause of CT A-II was not defined, although its infectious etiology has always been related to this type of tremor [9]. Since 2016, the etiology of CT A-II has been attributed to the newly described APPV, which has already been demonstrated by experimental infections [2,10]. Other viruses also were detected coinfection with APPV, such as porcine circovirus type 2 [11], astrovirus [12], porcine circovirus type 3 [13], linda virus [14], porcine circovirus-like virus P1 [15], and *Teschovirus A* [16], however, the role played by these three viruses as the primary cause of CT A-II is debatable.

CT type A-III is related to a genetic defect presented only by the Landrace breed, in which a lack of myelin is observed together with a reduction in the number of oligodendrocytes [17]. CT type A-IV is caused by a recessive genetic defect in the Saddleback swine and is characterized by hypomyelination of the brain and spinal cord [18,19]. Type A-V is caused by metrifonate (trichlorfon) intoxication. The occurrence of this type of CT may be associated with the use of metrifonate as an antiparasitic medication, and when administered during pregnancy, CT type A-V can cause cerebellar hypoplasia and cause piglets to be born with ataxia and tremor [20,21]. Finally, CT type B has no known specific cause, and changes in the CNS are not observed [5].

#### **3. Atypical Porcine Pestivirus**

#### *3.1. Classification and Etiology*

Until 2017, only four "classical" species belonged to the genus *Pestivirus*: bovine viral diarrhea virus 1 (BVDV-1); bovine viral diarrhea virus 2 (BVDV-2); border disease virus (BDV); and CSFV [22]. However, a large number of viruses linked to pestiviruses have been described [1,23–32], and Smith et al. [22] proposed that the study group of the *Flaviviridae* family of the International Committee on Viral Taxonomy (ICTV) revise the taxonomy of the genus *Pestivirus*. Thus, the genus *Pestivirus* was divided into 11 species, *Pestivirus A* (BVDV-1) [33], *Pestivirus B* (BVDV-2) [34], *Pestivirus C* (CSFV) [35], *Pestivirus D* (BDV) [36], *Pestivirus E* (pronghorn antelope pestivirus) [29], *Pestivirus F* (porcine pestivirus) [26], *Pestivirus G* (giraffe pestivirus) [23], *Pestivirus H* (HoBi-like pestivirus) [27], *Pestivirus I* (Aydin-like pestivirus) [24], *Pestivirus J* (rat pestivirus) [25], and *Pestivirus K* (atypical porcine pestivirus) [1], and one unclassified virus, bat pestivirus [30].

APPV is the single species of *Pestivirus K*, to the genus *Pestivirus*, and to the family *Flaviviridae* [37]. The viral particle is spherical, with a diameter of approximately 60 nm [38] and enveloped. Viral genomes are single-stranded, positive-sense RNA, exhibiting a genome size of approximately 11 to 11.6 kb. The genome comprises a 5 -noncoding region (NCR), one single open reading frame region encoding a single polyprotein with 3635 amino acids, and a 3 -NCR region [1,39]. The polyprotein is processed into 12 proteins, C (capsid protein), Erns, E1, E2 (envelope proteins), and nonstructural proteins Npro, p7, NS2, NS3, NS4A, NS4B, NS5A, and NS5B [40].

Phylogenic analyzes of the complete or partial genome of APPV strains detected in different countries and years, and in pigs from commercial farms and wild boars, demonstrate a high degree of genetic variability, subdividing into different clusters [3,41–44].

#### *3.2. Pathogenesis and Clinical-Pathological Manifestation Associated with APPV*

Two studies of experimental infection were carried out to associate APPV infection with CT and elucidate the pathogenesis. The first experimental inoculation was performed by Arruda et al. [2] and occurred in the US. Five pregnant sows, a group at 45 days and another at 62 days of gestation, were inoculated in the fetal amniotic vesicle, intranasally, and intravenously with known serum samples positive for APPV. In the period following the inoculation, no inoculated sow showed detectable viremia by qRT-PCR and no observation of clinical signs suggestive of CT. After birth, piglets were monitored for CT, and 57% to 100% of piglets showed signs of tremor. In the second experimental study carried out in the Netherlands by de Groof et al. [10], APPV-positive sera were inoculated intramuscularly in three gilts at 32 days of gestation. At 10 days postinfection, the sera from three gilts were RT-PCR-positive for the virus, and one of the gilts presented a relatively lower viral load than the other gilts. At birth, two of the three litters contained piglets showing clinical signs of CT, ranging from mild to severe signs in almost all piglets. No piglets born from the gilt that had low viremia showed signs of CT.

In both experimental studies, clinical signs were classified according to intensity; the most severe cases were characterized by intense tremors throughout the body, while the mild ones were characterized by muscle fasciculation in the limbs. Another clinical sign frequently observed was the presence of piglets with the splay leg [2,10]. This clinical manifestation is a syndrome characterized by temporary dysfunctionality of the posterior leg muscles, occurring shortly after birth, and resulting in difficulty or inability to stand and walk [45]. The presence of APPV RNA was detected in different organs of all piglets that showed signs of CT, and in some of the piglets not affected by CT [2,10]. These experiments were essential to elucidate the association of APPV with CT. However, Koch's postulate was not fulfilled by the two studies, probably due to the difficulty of viral isolation in cell cultures.

In the various studies that describe natural and experimental infections, the virus can be detected by conventional RT-PCR, qRT-PCR, and immunohistochemical in a wide variety of organs. The highest viral loads are observed primarily in the CNS and lymphoid tissues. In the CNS, the virus is found in greater quantities in the cerebellum and lymphoid tissues, mainly in the inguinal lymph nodes, submandibular lymph nodes, and thymus, suggesting that the cerebellum and lymph nodes are target organs of APPV [46,47]. In other organs, such as the brain, brain stem, spinal cord, heart, liver, spleen, lungs, and kidneys, the viral load is relatively lower [42,46,47].

To date, no significant macroscopic lesions have been described in piglets with CT [48–50]. Histopathological findings in APPV infections primarily include vacuolization of the white matter in the cerebellum [48,50,51]. Studies using Luxol® Fast Blue staining have shown a reduction in the intensity of myelin staining in the white matter of the spinal cord, cerebrum, cerebellum, and sciatic nerve, which implies a decrease in the amount of myelin in these tissues [16,48,50,52,53]. Gliosis has also been observed in the cerebral cortex and, to a lesser extent, in the spinal cord [16,49,50]. Other histopathological lesions observed in affected piglets were neuronal necrosis, neuronophagia with satellitosis, particularly at the cerebral cortex and spinal cord, Wallerian degeneration of the spinal cord, and necrosis of Purkinje cells of the cerebellum [49]. Transmission electron microscopy performed in two CNS tissues revealed not only hypomyelination of the cerebellar white matter but also more severe changes, such as interruption and breakdown of myelin, in animals affected by APPV [50]. These histopathological lesions described in piglets with CT, primary demyelination, and/or hypomyelination, can be considered a neuropathological characteristic associated with APPV infection [49]. The high viral loads of APPV in the CNS, along with the lesions observed in these tissues, explains the intensity of the neurological symptoms caused by APPV [46,47].

#### *3.3. Viral Shedding and Transmission*

Piglets that developed CT and recovered were monitored by some studies to assess the possibility of viral excretion and duration. Therefore, de Groof et al. [10] followed five pigs that presented CT and

recovered at 8.5 months of age; these animals continued to shed APPV in their feces. Excretion in feces was also suggested by Postel et al. [53] since high viral loads were detected in the duodenum, pancreas, and colon. APPV is found in salivary glands; therefore, the virus is shedding by saliva until at least the sixth month of life [50]. In these same animals, the virus was also found in semen. Another study evaluated semen collection from three commercial boar herds and detected the presence of the viral genome in swab and fluid preputial and in the semen from 15% of boars from a commercial boar stud, suggesting that artificial insemination may be an important form of transmission [54].

Regarding transmission, the horizontal and vertical forms of transmission were described, with the latter being the most important and responsible for the development of CT in newborn piglets from naturally infected sows. Transplacental transmission was demonstrated by experimental inoculation. Pregnant sows were infected with serum samples known to be positive for APPV, and the birth of piglets with CT of different levels of severity was observed [2,10].

In a controlled exposure, using fetal fluid from litters that had exhibited CT, a batch of 91 gilts was submitted to oral exposure to the antigen approximately 54 days before breeding. Signs of CT were observed in 45.0% of the litters and 30.8% of all piglets born [51]. These data suggest probable horizontal transmission, due to oral exposure, and the development of CT cases.

In piglets, horizontal transmission was described after the monitoring of piglets that showed signs of CT and obtained a confirmed diagnosis for APPV. After weaning, naive piglets were mixed with APPV-positive piglets. Due to close contact, infection of susceptible piglets occurred, suggesting that transmission occurs after mixing infected with susceptible piglets. No manifestation of disease was observed in this type of late infection; moreover, these animals were transiently infected [55]. It is likely that this type of transmission is related to environmental contamination and implies the spread of the virus. However, little is known about the forms of transmission and their possible effects; therefore, further investigations are necessary for a satisfactory clarification on this topic.

#### *3.4. Epidemiology and Implications in Pig Production*

In the United States, APPV was first identified by NGS in pig serum samples collected in 2014 from five different states [1]. In subsequent years, several studies from different countries reported the circulation of APPV in the US [1,2,51,54,56], Germany [53,55,57–61], Netherlands [10], Sweden [62], Austria [50], England [63], China [38,42,44,46,47,61,64–73], Spain [41,74], South Korea [75], Brazil [16,48,49,76], Great Britain [61], Italy [43,61], Serbia [61], Switzerland [61,77], Taiwan [61], Canada [52], and Hungary [78].

Epidemiological surveys demonstrate a high number of the pigs positive for APPV, depending on the region evaluated and the assay used. Postel and colleagues [61] used qRT-PCR to analyze 1460 serum samples from asymptomatic pigs from Germany, Italy, Serbia, Great Britain, Switzerland, China, and Taiwan and found a prevalence of 8.9% (130/1460) of APPV-positive samples. Additionally, in this same study, the highest (17.5%; 35/200) prevalence of positive samples was in Italy. Michelitsch et al. [59], by indirect immunofluorescence test (IFI), found an antibody prevalence of 16.3% (182/1115) in pig serum samples in Germany. In Switzerland, 1080 sera from pigs for slaughter, which were obtained between 1986 and 2018, were evaluated by qRT-PCR, and the APPV genome was detected in 13% [77].

Although this virus was discovered recently, some studies have shown that APPV has been circulating in pig herds for a long time. A study conducted in Spain detected the presence of the viral genome in pig serum samples collected in 1997 [41]. In Switzerland, another investigation analyzed 1080 pig sera collected between 1986 and 2018 and revealed that 7% (6/87) of the samples collected in slaughter pigs in 1986 were APPV-positive [77]. These findings correspond to reports of the long-term presence of APPV in the world.

The presence of APPV in wild boars has been demonstrated in Germany, Spain, and Italy, and the prevalence of the virus varies according to the country. In Germany, 456 wild boar serum samples were analyzed, and 19% of samples revealed the presence of the viral genome, while 52% of the serum samples had antibodies against the Erns protein [50]. In Spain and Italy, the prevalence

observed was lower, 0.23% (1/437), and 0.69% (3/430) of the evaluated sera had the virus genome, respectively [43,74]. Transmission of APPV from wild boars to pigs or vice versa has not been demonstrated to date. However, wild boars are sources of various pathogens that infect domestic pigs, including other pestiviruses, such as CSFV [79], and can contribute to their transmission. Due to the importance of the transmission of viral pathogens, these wild animals are a challenge for pig health and must be considered for effective control plans to be devised.

Regarding the occurrence of CT, the litters from first parity sows are most affected by APPV infection, and this infection rarely occurs in higher birth order [10,48,49]. During an outbreak on a farm in the Netherlands, 48 litters with CT were born from gilts and were monitored, the number of piglets affected within each litter ranged from <10 to 100%. Furthermore, the total mortality of piglets, reached 26%, with 60% of these deaths being attributed to CT [10]. In Brazil, an outbreak of CT observed in piglets born to gilts, lasted for three weeks, and the mortality rate reached 30% [49].

#### *3.5. APPV Infection and Immunity*

The dynamics of APPV infection can be hypothesized in two ways: persistently infected and transiently infected animals. These forms of infection are known to other pestiviruses, as is the case with CSFV [79] and BVDV in cattle [80]. In APPV infection, the dynamics have not been fully elucidated; however, this phenomenon can be explained by two studies [50,55]. Schwarz et al. [50] monitored the health status of two piglets (one female and one male) aged up to six months; shortly after birth, these two animals showed CT. Specific antibodies to APPV NS3 were detected at birth and at up to eight weeks of age. The tremor symptoms decreased and disappeared completely until 14 weeks; however, both piglets still presented viremia, antibody titer, and shedding of the virus by saliva. At six months the male piglet reached sexual maturity, and a high viral load was detected in saliva and semen; on the other hand, viremia was reduced.

Cagatay et al. [55], using direct and indirect tests, monitored 20 piglets from unaffected and affected litters by CT from birth to slaughter. In the vertically infected and symptomatic piglets, viremia was detected from the first days of life until slaughter. For the presence of antibodies, these piglets showed high levels of antibodies at six days of age, and these antibodies were undetectable at 21 and/or 48 days of age. It is possible that these antibodies came from the sows and disappeared with the drop in passive immunity. On the other hand, piglets infected horizontally after being mixed at weaning with those infected vertically showed viremia at 48 days of age and high titers of specific antibodies to E2 when evaluated at 69 and 161 days of life, suggesting the induction of protective immunity against infection.

Based on piglets infected horizontally, the immune response was higher for the E2 protein, and neutralizing antibody titers correlated with the presence of E2-specific antibodies, while a correlation with Erns-specific antibodies was not observed [58].

Due to the longevity of viral shedding, detectable viremia, together with the disappearance of specific antibodies over time, may suggest that persistent infection (PI) can be attributed to piglets that are intrauterine-infected. Soon after birth, these piglets can show signs of CT, which usually regresses over time [41,50,61]. On the other hand, piglets infected horizontally, through contact with persistently infected animals, develop a transient infection, with viremia detected for several weeks, but over time the piglet develops active immunity against APPV, and the virus becomes undetectable [55]. These studies provide evidence of the dynamics of infections and the immune response; however, the studies examine a small number of animals, meaning that further studies are required for complete elucidation.

#### *3.6. Diagnostic Methods of APPV*

Currently, a wide variety of diagnostic techniques are available for use in elucidating APPV infection in pigs. Due to practicality, speed, sensitivity, and specificity, molecular tests are the most commonly used. Among these tests, both conventional RT-PCR [16,48,49] and qRT-PCR [2] have been described in viral detection in several studies. Different clinical samples can be subjected to viral detection. In clinically affected animals, the CNS and the lymphoid organs are specimens of choice, since higher viral loads are found in these tissues, especially the cerebellum and lymph nodes [46,47]. Another available technique is NGS [2,10,51], which has been used since the first viral description in 2015 [1], and NGS is still employed as a diagnostic tool, assisting in the detection of the viral genome. In addition, NGS provides data on possible coinfections and primarily obtains larger fragments of the viral genome, which favors the phylogenetic study of circulating strains.

Histological tests performed from tissues fixed in paraffin are of great diagnostic importance, enabling the visualization of lesions caused by APPV. Luxol®® Fast Blue staining helps to observe the demyelination caused by the APPV in the CNS, primarily located in the cerebellum and spinal cord [16,48,50,52,53]. Histopathology, together with an immunohistochemical technique [46], and in situ hybridization [53,60] enables the detection of the viral agent (protein or nucleic acid) at the lesion site.

Other important methods that are used to assist in the diagnosis of infectious diseases are serological tests. The tests described for APPV infection are IFI [59], virus neutralization [55], and indirect enzyme-linked immunosorbent assay (ELISA) to the NS3, E2, and Erns proteins [50,55,58,61]. These tests have many important applications, and they can be used for population-based epidemiological studies and monitoring of infection in the herd; the tests also feature easy execution and low cost.

#### *3.7. Control and Prevention*

To date, there are no effective drugs or vaccines available to treat or prevent APPV infection. Zhang et al. [81] constructed a recombinant baculovirus of APPV glycoprotein E2, which induced a robust humoral and cellular immune response in mice. Based on these studies and knowing that the E2 protein is responsible for inducing neutralizing antibodies [55], this vaccine appears to be a promising tool as likely prevention of APPV in pigs. However, further viral challenge studies are needed to demonstrate an effective immune response.

Aslitters of gilts aremost affected by APPVinfection [10,49], itis probable that theintroduction of naive gilts in the herd is an important issue in the epidemiology of the disease [49,55]. Therefore, preventive measures are necessary to address this issue. The use of acclimatization for replacement gilts is an interesting tool that can be used in diseases that do not have vaccines and/or when available are not completely effective for prevention. Gatto et al. [3] suggest the protocol similar to what is used for the control of enzootic pneumonia caused by *Mycoplasma hyopneumoniae*. However, once horizontal transmission in gilts after oral exposure to fetal fluid antigen positive for APPV has been demonstrated, followed by the induction of CT in the litters of these gilts [51], this measure needs to be considered carefully. There is a need to conduct studies focused on this tool and, therefore, the development of protocols aimed at infection by APPV.

Another important point to be considered when designing control programs is the possibility of PI animals. Sensitive diagnostic methods that are able to detect PI animals, such as qRT-PCR, especially in the case of breeders, are highly important. After identifying these animals, they must be removed from the herd due to viral shedding by semen, feces, and oral fluid, which favors viral transmission. Another issue to be monitored is the semen used in the practice of artificial insemination; one possible measure is to include testing for APPV in the routine tests performed on these boars and their semen.

Despite specific measures to control APPV, biosecurity measures cannot be overlooked. When replacing breeders, animals from herds with a good health strategy must be purchased, and quarantining must be performed before introduction on the farm. Traffic control of people and vehicles on farms must be considered in addition to the execution of adequate cleaning and disinfection programs in the facilities. Finally, the control of wild animals, rodents, and insects contributes to the health of the herd in general.

#### **4. Conclusions and Perspectives**

The new porcine pestivirus (APPV) is strongly related to cases of CT type A-II, which until 2016, had an undefined etiology. Despite the recent discovery of APPV, evidence has shown that this virus has been circulating in pig herds for many years, since at least 1986, along with longstanding reports of CT, and the importance of this virus for global pig production is notable. In addition, APPV belongs to the genus *Pestivirus*, presenting important biological characteristics for the epidemiology of the disease, such as viral persistence, which can represent viral maintenance and a constant source of both horizontally and vertically transmitted infection in pig herds. These points are highly important for viral prevention. However, these issues and others regarding this disease still need to be elucidated, and further research should investigate the host's immune response, the control and prevention of APPV infection, and the development of vaccines.

**Author Contributions:** Conceptualization, A.M.D.A. and A.A.A.; Writing Original Draft Preparation, A.M.D.A.; Writing Review and Editing, A.F.A. and A.A.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the National Council of Technological and Scientific Development (CNPq grant number 305.062/2015-8).

**Acknowledgments:** The authors thank the following Brazilian Institutes for financial support: the National Council of Technological and Scientific Development (CNPq), the Brazilian Federal Agency for Support and Evaluation of Graduate Education (CAPES), the Financing of Studies and Projects (FINEP), and the Araucaria Foundation (FAP/PR). A.A.A. and A.F.A. are recipients of CNPq fellowships. A.M.D.A. is recipients of INCT-Leite/CAPES fellowship (grant number 88887.495081/2020-00).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### **Molecular Survey and Phylogenetic Analysis of Atypical Porcine Pestivirus (APPV) Identified in Swine and Wild Boar from Northern Italy**

#### **Enrica Sozzi \*, Cristian Salogni, Davide Lelli, Ilaria Barbieri, Ana Moreno, Giovanni Loris Alborali and Antonio Lavazza**

Istituto Zooprofilattico Sperimentale della Lombardia e dell'Emilia Romagna "Bruno Ubertini" (IZSLER), Via Antonio Bianchi 7/9, 25124 Brescia, Italy; cristian.salogni@izsler.it (C.S.); davide.lelli@izsler.it (D.L.); ilaria.barbieri@izsler.it (I.B.); anamaria.morenomartin@izsler.it (A.M.); giovanni.alborali@izsler.it (G.L.A.); antonio.lavazza@izsler.it (A.L.)

**\*** Correspondence: enrica.sozzi@izsler.it; Tel.: +39-030-2290361

Received: 14 November 2019; Accepted: 8 December 2019; Published: 10 December 2019

**Abstract:** Atypical porcine pestivirus (APPV) is a newly recognized member of the *Flaviviridae* family. This novel porcine pestivirus was first described in 2015 in the USA, where it has been associated with congenital tremor type A-II in new-born piglets. APPV is widely distributed in domestic pigs in Europe and Asia. In this study, a virological survey was performed in Northern Italy to investigate the presence of APPV using molecular methods. Testing of 360 abortion samples from pig herds revealed two APPV strains from distinct provinces in the Lombardy region and testing of 430 wild boar blood samples revealed three strains, one from Lombardy and two from Emilia Romagna. The nucleotide sequencing of a fragment of the nonstructural protein 3-coding region revealed a high similarity to the previously detected European strains (Spanish, German, and Italian) of APPV.

**Keywords:** pestivirus; pig; APPV; phylogenetic analysis; Italy

#### **1. Introduction**

Pestiviruses are highly variable single-stranded RNA genome viruses, belonging to the *Flaviviridae* family. Actually, based on molecular and epidemiological evidence, the genus Pestivirus includes eleven species, indicated with progressive letters from A to K. Thus, the "classical" species are A (Bovine viral diarrhea virus 1), B (Bovine viral diarrhea virus 2), C (Classical swine fever virus), and D (Border disease virus), whereas the new species are E to K [1]. In addition, other pestiviruses were described, likely as three additional species, respectively, in bats (bat-derived pestivirus) [2,3], sheep and goats (Tunisian sheep pestiviruses) [4], and pigs (Linda pestivirus) [5]. In the swine species only, in addition to Pestivirus C, three pestiviruses have been identified to date: (1) Pestivirus F (Bungowannah virus) reported only in Australia, as a cause of reproductive disorders, fetal death, and sudden death in piglets, [6]; (2) The Linda virus described in association with congenital tremors (CTs) in piglets in Austria and thereafter only occasionally reported, leaving its geographical spread and clinical relevance in pigs undefined [5]; and (3) Pestivirus K, commonly known as atypical porcine pestivirus (APPV), which is the most relevant due to the frequency of identification, clinical findings, and economic importance. In fact, it has been identified several times in North America [7,8], South America [9,10], Europe [11–13], and Asia [14,15], and it should be considered as stably present for a long time in domestic and wild pig populations [8,10,15,16]. This is true also for Italy, since in a previous survey, at least four APPV isolates were found between 2015–2017 and a quite high seroprevalence was detected in pig sera [15].

Although APPV has been repeatedly identified in asymptomatic animals, there is clear evidence that it is associated with CT syndrome type A-II (CT A-II) in newborns [7]. Clinically healthy pigs and wild boars may have an epidemiological role as vehicles of APPV, but, considering the different frequency of detection in wild boars, which was quite high in Germany and Serbia [17] and very low in Spain [18], the epidemiology of APPV may vary considerably from country to country with increases in livestock and wild populations, animal breeding, and world trade [18]. While one study described the economic losses caused by a 10% drop in the number of weaned piglets per sow [19], the full economic consequences of APPV outbreaks remain to be determined.

In this study, the presence of the APPV genome in pig fetuses and wild boars from both the Lombardy and Emilia Romagna regions of Northern Italy was determined, and the genetic characterizations of the identified strains are described.

#### **2. Materials and Methods**

#### *2.1. Pigs*

From 2016 to 2018, 360 fetuses of pigs from pig farms in the Lombardy region were examined at the IZSLER Diagnostic Laboratory in Brescia. All of the samples examined originated from field cases of spontaneous abortions in pig farms and sent to the general diagnostic laboratory of IZSLER to determine the presence of any infectious agent. In none of these cases was the presence of clinical signs referable to CT syndrome specifically reported. During necropsy, samples of organs (brain, lung, spleen, liver, and kidney) were taken from each aborted fetus, then collected into a single farm-specific pool and homogenized (10% w/v) in minimum essential medium (MEM; Gibco, Life Technologies, Paisley, UK) supplemented with an antibiotic (1000 U/mL penicillin, 1 mg/mL streptomycin; Gibco, Life Technologies, Paisley, UK) and anti-mycotic (2.5 μg/mL amphotericin B; Gibco, Life Technologies, Paisley, UK). After centrifugation, the supernatant was analyzed to identify any agents that cause abortions in swine. For bacteriological agents such as Brucella spp., Listeria spp., and Mycoplasma spp., the screening and pathogen identification were conducted according to Office International des Epizooties (OIE) standardized protocols [20]. The presence of Chlamydophila spp. was investigated by real-time Polymerase Chain Reaction (PCR) directly in biological samples [21] and typing by the PCR-restriction fragment length polymorphism (RFLP) assay, targeting the 16S ribosomal gene [22]. For Mycoplasma spp., the PCR method described by van Kuppeveld et al. [23] was used. Virological analyses for detecting the more common pig pathogens were conducted using a panel of PCR methods including porcine reproductive and respiratory syndrome virus (PRRSV) (AgPath-ID™ NA and EU PRRSV Multiplex© Applied Biosystems), porcine circovirus type 2 (PCV-2) [24], porcine parvovirus (PPV) [25], and porcine circovirus type 3 (PCV-3) [26]. The presence of pestiviruses was determined by using a pan-pestivirus real-time RT-PCR [27], and, considering its limited capacity in detecting APPV, by a APPV-specific real-time RT-PCR [19]. In addition, all the tested samples were inoculated on cell cultures (primary embryonic swine kidney cells, swine alveolar macrophages, and monkey kidney cell line MARC-145), which allow the isolation of a broad range of swine viruses. The inoculated cell monolayers were observed daily for 5–7 d for the appearance of a cytopathic effect and then sub-cultured to the second passage, at which time they were independently tested using an "in-house" sandwich enzyme-linked immunosorbent assay (ELISAs) for the presence of pestivirus [28] and PRRSV antigens [29]. Only those samples that were positive for APPV at the initial screening test using the APPV-specific real-time RT-PCR were further subcultured to improve the chance to isolate the APPV until the fifth passage, at which time, even in the absence of a cytopathic effect, they were assessed again with the APPV-specific real-time RT-PCR.

#### *2.2. Wild Boars*

In total, 430 blood samples of wild boars, killed during the 2017–2018 hunting season, were collected in the framework of the Lombardy and Emilia Romagna wildlife monitoring plans for classical swine fever (CSF) and Aujeszky disease and transferred to IZSLER for examination. Serum samples were tested for antibodies against swine vesicular disease virus [30,31], encephalomyocarditis virus [32], glycoprotein E of Aujeszky's disease virus [33], pestivirus (A–D) [34], swine influenza virus type A, subtypes H1N1, H1N2, and H3N2 [35,36], and finally for Brucella spp. (Svanovir Brucella–Ab C-ELISA©). Serological analyses were conducted with the methods currently in use at the IZSLER.

#### *2.3. Identification and Genomic Characterization of Atypical Porcine Pestivirus (APPV)*

For the investigation of the APPV genome, all samples, both the homogenates from fetal organs and the wild boar sera, were screened using the NS5B gene-specific real-time RT-PCR method [19]. Samples that tested positive were characterized by Sanger sequencing using RT-PCR that amplified a fragment of the NS3 region. The nucleotide sequences were aligned using the ClustalW method and compared with sequences present in GenBank [37] using MEGA6 software [38]. The maximum likelihood phylogenetic tree was constructed using IQ-tree software [39] by applying the TIM2+F+G4 model identified using ModelFinder selection [40].

#### **3. Results**

#### *3.1. Pigs*

The examination of the NS5B gene by real-time RT-PCR in the homogenized samples of pig fetuses identified two (0.6%) positive samples in two distinct farrow-to-finish farms, one from the province of Brescia in a pool containing the organs of three fetuses and the other from the province of Mantua in a pool containing the organs of two fetuses (Figure 1).

**Figure 1.** Geographical distribution of the pig farms and wild boar hunting sites where atypical porcine pestivirus (APPV) was identified.

At necropsy, none of these fetuses showed internal macroscopic lesions, and bacteriological investigations consistently produced negative results. Moreover, the molecular examinations for PRRSV, PCV-2, PPV, and pan-pestivirus were all negative, and only the real-time PCR for PCV-3 produced a positive result from a pool of fetuses that originated from the province of Brescia. The two sequences obtained, APPV\_Italy\_SW\_BS341729\_2017 and APPV\_Italy\_SW\_MN212160\_2016, had a nucleotide similarity of 92.5% between them and clustered with APPVs previously identified in Europe. In particular, the phylogenetic tree constructed using the NS3 region (Figure 2) revealed that the two identified strains belong to distinct groups: (1) the APPV\_Italy\_SW\_MN212160\_2016 strain is related to both the 98/Sp06 strain identified in Spain in 2006 [16] (99.2% identity) and the German strain Bavaria S5/9 identified in 2015 [11] (96.8% identity); and (2) APPV\_Italy\_SW\_BS341729\_2017 forms a separate and closely related clade with two sequences identified in 2015 in pigs from Italy, Italy-164 and Italy-181 [15], with a nucleotide similarity of 95.5% and 99.2%, respectively.

**Figure 2.** Phylogenetic tree based on a 645-nt fragment of the nonstructural protein 3-encoding region of the atypical porcine pestivirus (APPV) genome present in GenBank. A phylogenetic analysis using the maximum likelihood method including 1.000 bootstrap iterations was performed. Only bootstrap values ≥60 are indicated. Sequences in bold were generated in this study.

Despite the positive results of the real-time RT-PCR test, virological examinations of the organs of swine fetuses by inoculating cell cultures up to the fifth passage produced consistent negative results. Consequently, we did not succeed in isolating any APPV strains.

#### *3.2. Wild Boars*

All of the wild boar samples examined in this study including those that tested APPV-positive, originated from hunting activity, and thus, they were considered healthy based on their behavior before shooting and the absence of lesions on carcasses examined at slaughter. Out of the 430 blood samples examined, three (0.69%), originating from the Cremona Province in Lombardy (APPV\_Italy\_WB\_CR264058\_2018) and from two neighboring provinces, Rimini and Forlì in Emilia Romagna (APPV\_Italy\_WB\_RN262773\_2018 and APPV\_Italy\_WB\_FC132781\_2018, respectively) (Figure 1), screened positive for APPV. In agreement with the genomic characterization of the strains identified in pigs, the sequences of those identified in wild boars could be divided into two distinct clusters. The first included the two strains detected in Emilia Romagna (APPV\_Italy\_WB \_RN262773\_2018 and APPV\_Italy\_WB \_FC132781\_2018) and the sequence identified in the pig farm from Mantua Province in Lombardy (APPV\_Italy\_SW\_MN212160\_2016), which had identities of 95.6% and 96%, respectively; plus the previously characterized strains BavariaS5/9 and Spain 98/Sp06, with which the WB strains had a higher nucleotide identity of 98–99.2%.

The second clusterincluded the APPV\_Italy\_WB\_CR264058\_2018, the APPV Italy\_SW\_ BS341729\_2017 (96.4% identity), the Italian pig strains, Italy-164 and Italy-181, which had nucleotide of 96.4% and 96%, respectively, and the strains identified in 2016 in Lower Saxony, Germany (94.3% identity).

For the collateral serological examinations, all APPV positive samples were negative for all serological tests employed except for: (a) one serum (APPV\_Italy\_WB\_CR264058\_2018) that tested positive for anti-Brucella spp. antibodies and (b) one (APPV\_Italy\_WB\_FC132781\_2018) that tested positive for glycoprotein E antibodies to the Aujeszky's disease virus.

The nucleotide sequences of the five APPV isolates were deposited in GenBank (NCBI) with the accession numbers MN736974–MN736978.

#### **4. Discussion**

The examination of 360 swine fetuses and 430 hunted wild boar blood samples identified APPVs in the area, two strains from the former and three strains from the latter. Based on the phylogenetic analysis performed, all five of these strains clustered with APPV strains previously identified in Spain and Germany. Although all of the sequences obtained can be grouped into the putative Cluster I described by Muñoz-González et al. [16], which should include viruses that have a common origin, the strains identified here could be further divided in two distinct sub-clusters.

Based on the absence of lesions on both fetuses and wild boars, we were keen to exclude a definite pathogenic role for APPV in these specific cases. However, it remains to be clarified whether the PCV-3 plus APPV (APPV\_Italy\_SW\_BS341729\_2017) co-infection, which was detected in the pool of fetuses originating from Brescia Province, may have had a synergistic effect in infected piglets, resulting in a clinical form characterized by abortion and natimortality. The very low prevalence of APPV found in the examined pig samples (0.6%) is largely different from the sole previous study on APPV in Italy [15], which reported a higher prevalence of viral detection (17.5%). This could be likely due to the sample types and selection, since we included in the survey only aborted pig fetuses with the aim of trying to establish if any correlation existed between the clinical case and the detection of APPV. A more specific monitoring program focused on the two APPV positive farms, and in general, to have systematically reported and analyzed cases of CT are anticipated in order to better define the prevalence of APPV in pig farms in Italy and clarify its effects.

Wild boars are susceptible to APPV infection, but their role in the epidemiology of the virus remains unknown. The low prevalence of APPV in the examined wild boar population is well correlated with the numbers reported by Colom-Cadena et al. [18] in Spain, but is in contrast to the high prevalence found in wild boars from northern Germany [17], where APPV seems to be endemic among wild boars in many areas. Considering the small number of strains isolated from the large geographical territory represented by the two regions (Lombardy and Emilia Romagna) and the lack of any particular geographic distribution of the clustered strains, it is impossible to draw epidemiological interpretations on APPV diffusion or on transmission between domestic pigs and wild boars.

#### **5. Conclusions**

APPV appears to be well established in the domestic swine populations of different countries in Europe, America, and Asia. Indeed, although identifying APPV has only become possible during the last few years due to the progressive refinement of diagnostic techniques, retrospective studies have indicated its circulation for many years. Based on the available genomic characterization data in the literature and the prevalence in domestic pigs, APPV exhibits a high genetic diversity among viral strains detected in different countries and tends to form independent clusters according to geographic locations. This study confirms the presence and distribution of APPV in populations of domestic and wild pigs present in the Lombardy and Emilia Romagna regions of Italy, which are under the jurisdiction of IZSLER. The high homology levels with strains identified in Germany and Spain reinforce the hypothesis that Italian strains have a European origin, and they confirm the likely determining role in the spread of infection being the commercial trade in pigs among different countries. The detection of new pestiviruses indicates the need to monitor for their presence and distribution using a systematic surveillance and diagnostic approach. In fact, the accurate and constant characterization of circulating strains is necessary to update the serological and virological tests, which in turn may be used to collect more detailed epidemiological information regarding APPV such as routes of entry and dissemination, and genetic evolution.

**Author Contributions:** C.S. and G.L.A. performed necropsy and routine laboratory examinations; E.S., I.B., D.L., and A.M. performed the laboratory work and analyzed the data; E.S., A.M., D.L., and A.L. interpreted the results and designed the figures; E.S. and A.L. wrote the manuscript. All the authors critically analyzed, revised, and approved the manuscript.

**Funding:** This study was supported by grants from the project PRC 2015019 of the Italian Ministry of Health.

**Acknowledgments:** The authors would like to thank Loredana Zingarello and Debora Campagna for their excellent technical assistance.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### **Prevalence and Genetic Diversity of Atypical Porcine Pestivirus (APPV) Detected in South Korean Wild Boars**

**SeEun Choe 1, Gyu-Nam Park 1, Ra Mi Cha 1, Bang-Hun Hyun 1, Bong-Kyun Park 1,2 and Dong-Jun An 1,\***


Received: 30 May 2020; Accepted: 22 June 2020; Published: 24 June 2020

**Abstract:** Atypical porcine pestivirus (APPV), currently classified as *pestivirus K*, causes congenital tremor (CT) type A-II in piglets. Eighteen APPV strains were identified from 2297 South Korean wild boars captured in 2019. Phylogenetic analysis of the structural protein E2 and nonstructural proteins NS3 and Npro classified the APPV viruses, including reference strains, into Clades I, II and III. Clade I was divided into four subclades; however, the strains belonging to the four subclades differed slightly, depending on the tree analysis, the NS3, E2, and Npro genes. The maximum-likelihood method was assigned to South Korean wild boar APPV strains to various subclades within the three trees: subclades I.1 and I.2 in the E2 tree, subclade I.1 in the Npro tree, and subclades I.1 and I.4 in the NS3 ML tree. In conclusion, APPV among South Korean wild boars belonging to Clade I may be circulating at a higher level than among the South Korean domestic pig populations.

**Keywords:** APPV; wild boar; ML tree; Clade; pestivirus

#### **1. Introduction**

Pestiviruses are highly variable single-stranded RNA genome viruses belonging to the family *Flaviviridae*. The genus *Pestivirus* includes animal pathogens that are of worldwide socioeconomic significance—these include bovine viral diarrhea virus (BVDV, *pestivirus A–B*), classical swine fever virus (CSFV, *pestivirus C*), and border disease virus (BDV, *pestivirus D*) [1]. Other *Pestiviruses* include *Pestivirus E* (pronghorn pestivirus), *Pestivirus F* (Bungowannah virus), *Pestivirus G* (giraffe pestivirus), *Pestivirus H* (Hobi-like pestivirus), *Pestivirus I* (Aydin-like pestivirus), and *Pestivirus J* (rat pestivirus) [1].

A novel genetically distinct strain of *pestivirus K*, named atypical porcine pestivirus (APPV), was first identified in the USA in 2015 [2]. It was also identified in Germany, the Netherlands, Austria, Spain, China, Hungary, Brazil, and Sweden in 2016–18 [3–8]. In South Korea, the first case of APPV, which causes congenital tremor (CT) type A-II in suckling piglets, was identified in 2017 [9]. Although APPV has been found in animals with no clinical signs, there is clear evidence that it is associated with CT type A-II in newborns [10]. The presence of APPV in domestic pigs was confirmed repeatedly in several countries in Europe, North and South America, and Asia [2,11], and in wild boars from countries in Europe [10,12,13]. Clinically, healthy pigs and wild boars may have an epidemiological role as vehicles for APPV.

*Pestivirus* genomes can be classified into genogroups and sub-genogroups [14,15]. Phylogenetic analysis of the E2 and Npro proteins divided APPV into at least four (A–D) [5,16] or five (A–E) different genogroups [17].

In this study, we describe the detection and genetic characterization of APPV in wild boars from South Korea.

#### **2. Material and Methods**

#### *2.1. Wild Boars*

To satisfy the OIE requirements for the surveillance of wild boars and feral pigs in Classical Swine Fever-free countries, wild boars have been hunted (in co-operation with the Korean Pork Producers Association and the South Korea government) since 2010. Immediately after hunting, blood samples were collected from wild boars captured in nine provinces of South Korea and transported to the APQA. Blood samples from 2297 wild boars (1126 males, 1045 females, and 126 unknown), captured in 2019, were screened for APPV.

#### *2.2. Reverse Transcription-Polymerase Chain Reaction (RT-PCR) of APPV*

RT-PCR was performed to detect APPV [12,16]. Briefly, total RNA was extracted from 100 μL of whole blood using the QIAamp viral RNA mini kit (Qiagen, Cat. No. 52904. Hilden, Germany). Extracted RNA was reverse-transcribed using SuperScript III (Invitrogen, Cat. No. 18080093. Carlsbad, CA, USA). For APPV screening and sequencing, PCR was performed using primers designed to target areas of the conserved NS3-encoding region, as described previously [12]. Primers targeting the E2 and Npro genes were used to amplify the complete nucleotide sequences, as described previously [16]. The amplification products were purified using the QIAquick Gel Extraction Kit (Qiagen, Cat. No. 28704. Hilden, Germany) and used directly for sequencing (Cosmogentech Co., Seoul, Korea). PCR and serum neutralization tests were used to detect viral antigens and antibodies specific for CSFV and BVDV in APPV-positive samples, as described previously [18].

#### *2.3. Phylogenetic Analysis of APPV*

Multiple nucleotide sequence alignment was carried out by the Clustal X alignment program [19] using APPV sequences available in GenBank as references, and BLAST software (NCBI, Bethesda, Rockville, MD, USA). Outgroup strains comprised *pestiviruses A–H*. The partial sequences of NS3, derived from 18 APPVs detected in South Korean wild boars, were compared with 86 reference sequences (including eight outgroup strains) from Asia, North America, and Europe. The complete E2 sequence of four APPVs and the complete Npro sequence data of five APPVs detected in South Korean wild boars were compared with 70 (including ten outgroup strains) and 69 reference sequences (including ten outgroup strains), respectively. Nucleotide sequences of the NS3, E2, and Npro regions were analyzed phylogenetically using the maximum-likelihood (ML) method, with the Tamura–Nei model and bootstrap analysis (*n* = 1000) within MEGA 7.0 software (State College, PA, USA) with default parameters [20]. The ML tree was based on rates among sites (Gamma distributed with invariant sites (G+I)) and the ML heuristic method (Nearest-neighbor-interchange (NNI)). The partial NS3 sequences of 18 APPV strains (accession numbers: MT501737–MT501754), the complete E2 sequences of four APPV strains (accession numbers: MT501733–MT501736), and the complete Npro sequences of five APPV strains (accession numbers: MT501555–MT501759) detected in South Korean wild boars were deposited in GenBank.

#### *2.4. Ethical Approval*

The authors confirm that the work complies with the ethical policies of the journal. The work was approved by the Institutional Animal Care and Use Committee of the Animal and Plant Quarantine Agency (APQA) (Approval Number: 2019-448).

#### **3. Results**

#### *3.1. Geographic Prevalence of APPV*

Eighteen APPV strains were identified in 2297 blood samples collected from wild boars in 2019, suggesting that the prevalence of APPV is 0.78%. Of the APPV-positive wild boars, 15 were male (15/18, 83.3%), two were female (2/18, 11.1%), and one was of unknown sex (1/18, 5.6%). APPV strains were detected in wild boars from six provinces and of various weights (seven <30 kg; seven 30–60 kg; four >60 kg). Among the 18 APPVs detected in South Korean wild boars, five were detected in Gyeongnam (GN, 5/292; 1.71%), four in Gangwon (GW, 4/609; 0.66%), three each in Gyeonggi (GG, 3/452; 0.66%) and Chungnam (CN, 3/288; 0.35%), two in Chungbuk (CB, 2/204; 0.98%), and one in Gyeongkuk (GB, 1/275; 0.36%), as shown in Figure 1. All APPV-positive samples were negative for anti-CSFV and BVDV antibodies and antigens.

**Figure 1.** Locations in which APPV-positive wild boars were captured. Locations in which wild boars were captured are marked by a red dot. GW: Gangwon; GG: Gyeonggi; GN: Gyeongnam; GB, Gyeongbuk; JN: Jennam; JB: Jenbuk; CN: Chungnam; CB: Chungbuk; JJ: Jeju.

#### *3.2. ML Trees Based on NS3 Sequences*

The NS3 sequences (767 nucleotides (nt)) of the 18 APPVs detected from wild boars were 87.7–99.9% identical at the nt level and 96.5–100% identical at the amino acid (aa) level. ML tree analysis of NS3, E2, and Npro sequences revealed that *Pestivirus* strains were clearly divided into two groups: *Pestivirus K* (APPVs) and Other *Pestiviruses* (*A*–*H*), as shown in Figures 2–4). All APPVs were classified into three large Clades (I, II, and III) and four smaller subclades (I.1, I.2, I.3, and I.4), as shown in Figures 2–4). ML analysis of the NS3 sequences of the 18 APPVs from South Korean wild boars were included in Clade I (16 in subclade I.1 and two in subclade I.4), as shown in Figure 2. The nt sequence identity between the South Korean APPVs in subclades I.1 and I.4 was 86.8–88.7%; however, identity at the aa sequence level was 95.7–97.6%.

**Figure 2.** Phylogenetic tree of South Korean wild boar APPVs, based on NS3 sequences. The phylogenetic tree was constructed using the ML method (based on the Tamura–Nei model), with bootstrap analysis

(*n* = 1000), in MEGA 7.0 software. The 767 nt NS3 sequences of 18 APPVs from South Korean wild boars were compared with 86 reference sequences (including eight outgroup strains: *pestiviruses A–G*) from Asia, North America, and Europe. The Log likelihood (Log L) is −10,191.77, and only bootstrap values ≥70 are indicated on the nodes. South Korean wild boar APPV strains and South Korean domestic pig APPV strains are denoted by red bold and black bold letters. The scale bar indicates the number of nucleotide substitutions per site.

**Figure 3.** ML tree of South Korean wild boar APPVs, based on E2 sequences. The phylogenetic tree (Log L, −11,438.12) was constructed using the ML method (based on the Tamura–Nei model), with bootstrap analysis (*n* = 1000). The complete E2 sequences of four APPVs detected in South Korean wild boars were compared with 70 reference sequences (including ten outgroup strains: *pestiviruses A–H*).

**Figure 4.** ML tree of South Korean wild boar APPV, based on Npro sequences. The phylogenetic tree (Log L: −5113.82) was constructed using the ML method (based on the Tamura–Nei model), with bootstrap analysis (*n* = 1000). The complete Npro sequences of five APPV detected in South Korean wild boars were compared with 69 reference sequences (including ten outgroup strains: *pestiviruses A–H*).

The 767-nt NS3 sequences of the 16 South Korean wild boar APPVs (Subclade I.1) and 8 APPVs isolated from German wild boars (Subclade I.2), as shown in Figure 2, were 90.9–92.0% identical at the nt level and 97.6–99.2% identical at the aa level. Moreover, NS3 sequences from the two South Korean wild boar APPVs (subclade I.4) and of the eight German wild boar APPVs (subclade I.2) were 88.1–88.9% identical at the nt level and 96.1–97.3% identical at the aa level.

#### *3.3. ML Trees for the E2 Sequences*

The E2 sequences (723 nt) from four South Korean wild boar APPVs (wbKOR 13079, -13188, -14105, and -13370) were 91.3–94.3% identical at the nt level and 95.4–97.1% identical at the aa level. ML analysis of the E2 sequences categorized them as Clade I (three in subclade I.1 and one in subclade I.2), as shown in Figure 3. By contrast, six APPVs from South Korean domestic pigs belonged to three subclades (one to subclade I.1, three to subclade I.2, and two to subclade I.4), as shown in Figure 3. The four South Korean wild boar APPVs and six South Korean domestic pig APPVs were 87.4–94.3% identical at the nt level and 94.2–98.8% identical at the aa level. The E2 sequences from the four South Korean wild boar APPVs and the 17 other APPVs in Clade II were 84.4–85.1% identical at the nt level and 94.2–96.3% identical at the aa level, whereas the four South Korean wild boar APPVs and 11 other APPVs in Clade III were 81.5–82.7% identical at the nt level and 90.0–92.5% identical at the aa level.

#### *3.4. ML Trees for the Npro Sequences*

The Npro sequences (540 nt) of five APPVs from South Korean wild boars (wbKOR 13079, -13109, -13188, -14105, and -13370) were 92.6–98.7% identical at the nt level and 93.9–99.4% identical at the aa level. ML analysis of the Npro sequences categorized them as Clade I (all five belonged to subclade I.1), as shown in Figure 4. The six APPVs from South Korean domestic pigs were distributed among all subclades (one in subclade I.1, two in subclade I.2, one in subclade I.3, and one in subclade I.4), as shown in Figure 4. The nt sequence identity between the five South Korean wild boar APPVs and the six South Korean domestic pig APPVs was 85.0–95.4%, whereas the aa sequences were 89.4–95.0% identical. The Npro sequences of the five South Korean wild boar APPVs and the 16 APPVs in Clade II were 79.3–80.5% identical at the nt level and 80.6–83.9% identical at the aa level, whereas the sequences from the five South Korean wild boar APPVs and the 12 APPVs in Clade III were 77.0–80.4% identical at the nt level and 80.6–81.1% identical at the aa level.

#### **4. Discussion**

Some retrospective studies suggest that APPVs were circulating widely for decades before the recent reports describing their detection [21,22]. An early study identified an APPV strain from samples obtained from piglets with CT; qRT-PCR of samples from a pig herd in the USA revealed an APPV prevalence of 6% [2]. Another study reported APPVs in CT-affected piglets from many countries, which further supports the potential relationship between APPV and CT type A-II [3]. In South Korea, APPV sequences from the domestic pig samples collected in 2016 had been submitted to GenBank (accession numbers: MF979135, MH509410), and the first identification of APPV was reported in suckling piglets with CT in 2017 [9]. Moreover, an APPV detection rate of 2.4–22% was reported in apparently healthy pigs in the USA and Germany [6,12,23]. A recent study of serum samples from apparently healthy pigs revealed that the prevalence of APPV in Europe was 2.3–17.5%, and that in China it was 5–11% [11]. All of these results demonstrate that APPV is highly prevalent in both apparently healthy pigs and CT-affected pigs, suggesting that the virus may have spread worldwide.

Wild boars are susceptible to APPV infection, although the role of this species in the epidemiology of the virus is unknown [13]. A very recent finding is that the prevalence of APPV in serum samples from wild boars in Europe is 0.23–19%, suggesting that wild boars may be a reservoir for APPV [10,12,13]. All wild boars sampled in this study, including APPV-positive wild boars, were considered healthy (based on their behavior) before hunting. The prevalence of APPV among South Korean wild boars is low, as in Spain (0.23%) and Italy (0.69%) [10,13]; however, this is in contrast to the high prevalence

detected in northern Germany (19%) [12]. Furthermore, serological investigations in wild boars revealed an antibody detection rate of 52% in northern Germany, where APPV seems to be endemic among wild boars in many areas [12].

Phylogenetic analysis revealed that APPV sequences (complete or partial polyprotein) exhibit high genetic diversity between strains detected in different countries and species (domestic pigs or wild boars), and that they form independent clusters according to geographic location [24].

Recently, phylogenetic analysis in China revealed a high level of genetic variation among APPVs: three Clades (I–III), with four subgroups (1–4) in Clade I [25]. Sequence analysis of the APPV NS3, E2, and Npro genes, followed by the construction of ML trees, classified them into three Clades—APPVs isolated from China were assigned to all Clades, whereas APPVs isolated from Asia, North America, and European countries belonged only to Clade I [25,26]. APPV strains isolated from South Korean wild boars and South Korean domestic pigs belonged to Clade I, without a preference for any particular subclade. This means that APPVs may have high the potential for spreading between wild boars and domestic pigs. In the three ML trees (NS3, E2, and Npro) constructed for this study, Clade I contained four subclades; however, the APPV strains contained within these subclades were not clearly distinguished by the three ML trees. A previous study suggests that recombination events occur between Clades (Clades II and III) or within a Clade (Clade I) [25]. Therefore, further research is needed to determine whether the difference in subclade of South Korean APPV strains in three ML trees are because of the recombination events within Clade I.

#### **5. Conclusions**

Here, we present the first report of APPV detected in South Korean wild boars. We found that the overall prevalence of APPV in South Korean wild boars was low (0.78%). South Korean wild boars harbor genetically diverse APPV strains belonging to Clade I. Wild boars may be an important virus reservoir for APPV. More epidemiological information will help to establish effective control measures and to eradicate the virus from affected pig herds in the future.

**Author Contributions:** Conceptualization, S.C. and B.-K.P.; methodology, G.-N.P. and R.M.C.; writing—review and editing, B.-H.H. and D.-J.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This project was supported by grants (Project Code No. B-1543083-2019-21-02) from the Animal and Plant Quarantine Agency, Republic of Korea.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
