Stability of the Resistance to the Thiosemicarbazone Derived from 5,6-Dimethoxy-1-Indanone, a Non-Nucleoside Polymerase Inhibitor of Bovine Viral Diarrhea Virus

Bovine viral diarrhea virus (BVDV) is the prototype Pestivirus. BVDV infection is distributed worldwide and causes serious problems for the livestock industry. The thiosemicarbazone of 5,6-dimethoxy-1-indanone (TSC) is a non-nucleoside polymerase inhibitor (NNI) of BVDV. All TSC-resistant BVDV variants (BVDV-TSCr T1–5) present an N264D mutation in the NS5B gene (RdRp) whereas the variant BVDV-TSCr T1 also presents an NS5B A392E mutation. In the present study, we carried out twenty passages of BVDV-TSCr T1–5 in MDBK cells in the absence of TSC to evaluate the stability of the resistance. The viral populations obtained (BVDV R1–5) remained resistant to the antiviral compound and conserved the mutations in NS5B associated with this phenotype. Along the passages, BVDV R2, R3 and R5 presented a delay in the production of cytopathic effect that correlated with a decrease in cell apoptosis and intracellular accumulation of viral RNA. The complete genome sequences that encode for NS2 to NS5B, Npro and Erns were analyzed. Additional mutations were detected in the NS5B of BVDV R1, R3 and R4. In both BVDV R2 and R3, most of the mutations found were localized in NS5A, whereas in BVDV R5, the only mutation fixed was NS5A V177A. These results suggest that mutations in NS5A could alter BVDV cytopathogenicity. In conclusion, the stability of the resistance to TSC may be due to the fixation of different compensatory mutations in each BVDV-TSCr. During their replication in a TSC-free medium, some virus populations presented a kind of interaction with the host cell that resembled a persistent infection: decreased cytopathogenicity and viral genome synthesis. This is the first report on the stability of antiviral resistance and on the evolution of NNI-resistant BVDV variants. The results obtained for BVDV-TSCr could also be applied for other NNIs.


Introduction
Bovine viral diarrhea virus (BVDV), along with classical swine fever virus and border disease virus, is a member of the genus Pestivirus. This genus belongs to the family Flaviviridae, which also includes the genera Hepacivirus (hepatitis C virus) and Flavivirus (yellow fever virus, Dengue fever virus and West Nile virus).
BVDV infection is distributed worldwide and tends to be endemic in many countries [1], [2]. In Argentina, the prevalence of BVDV antibodies in adult cattle is around 70% [3], [4]. BVDV infections lead to a reduction in milk production, lower conception rates, respiratory disorders and death of the animals that acquire the acute infection, thus causing serious economic losses for the livestock industry. Moreover, fetal infection leads to persistently infected cattle, which are generally smaller, are more susceptible to other infections and eventually develop a lethal mucosal disease.
BVDV is an enveloped single-stranded (+) RNA virus. Its genomic RNA is about 12.5 kb long and consists of a single open reading frame (ORF) flanked by 59 and 39 non-translated regions (NTRs). The ORF encodes a polyprotein of approximately 3,900 amino acids that is co-and post-translationally processed to mature viral proteins by cellular and viral proteases. The viral proteins are sequentially designated Npro, C, Erns, E1, E2, p7, NS2, NS3, NS4A, NS4B, NS5A and NS5B. The autoprotease Npro is a non-structural (NS) protein that plays a role in blocking IFN-a/b induction [5][6][7][8]. The capsid protein (C) is followed by three virion glycoproteins: Erns, E1, and E2. Erns encodes an RNase that is secreted in nonvirion forms [9] and targets extracellular RNA, a major viral signal that triggers IFN synthesis [10], [11]. The rest of the polyprotein is processed to NS proteins, of which only NS3 through NS5B are required for RNA replication [12], [13]. NS2, together with the amino terminus of NS3, functions as an autoprotease that cleaves the NS2-NS3 junction of the polyprotein. This cleavage is required for RNA replication and is linked to BVDV cytopathogenicity and pathogenesis [14], [15]. NS3 is a multifunctional protein with a helicase/nucleoside triphosphatase and serine protease activity responsible for all downstream polyprotein cleavages [16][17][18][19]. NS4A is tightly associated with NS3 and serves both as a cofactor for NS3 protease activity and as a tether that localizes NS3 to membranes [17], [20]. NS4B is believed to function as an integral membrane scaffold upon which the replicase complex (RC) assembles [21]. NS5A is a phosphoprotein that plays an essential role in BVDV replication and pathogenesis [22][23][24]. Finally, NS5B is the viral RNA-dependent RNA polymerase that catalyzes viral RNA synthesis [25][26][27][28][29].
BVDV isolates are categorized into two biotypes according to their effect on cultured cells: noncytopathic (ncp) isolates, which infect permissive host cells without causing cell death, and cytopathic (cp) isolates, which produce rapid cytopathic effects (CPE) and kill cells [30]. Only ncp isolates are able to establish the persistent infection. This difference is associated with distinct interactions between each biotype and the host innate immune response against the viral infection, which is active early during gestation. Cp biotypes emerge from ncp biotypes exclusively in persistently infected animals and are invariably associated with the mucosal disease [12], [31]. The induction of apoptosis plays a significant role in the pathology of the cp isolates both in infected cell cultures and in the clinical manifestations of the mucosal disease [32][33][34][35].
Studies on the resistance to an antiviral agent are highly important for the development and therapeutic application of such antiviral agent. Given that the impact of resistance is difficult to predict, it is important to evaluate not only the emergence of resistance but also its stability and the effect of the associated mutations on the viral replicative fitness in an antiviral-free environment. Therefore, the aim of the present work was to evaluate the stability of the resistance to TSC. To this end, we carried out 20 passages of BVDV-TSC r T 1-5 in the absence of TSC. We also describe the molecular and biological characterization of the viral populations obtained, in terms of infectious virus production and cytopathogenesis.
Passages of BVDV-TSC r in the absence of TSC MDBK cells were infected with BVDV-TSC r T 1-5 at a multiplicity of infection (MOI) of 0.01. Cell cultures were incubated in infection medium (infection medium: EMEM supplemented with 2.5% Donor Horse Serum -DHS-Gibco) in a 5% CO 2 incubator at 37uC until cell monolayers presented 80-90% of viral CPE. The supernatant was collected and clarified by centrifugation for 10 min at 1,0006g at 4uC. After titration of the virus in MDBK cells, the virus was used to infect new cell monolayers to generate the next passage. This procedure was repeated twenty times, and the virus obtained from each BVDV-TSC r T 1-5 was named BVDV R 1-5 .

Plaque reduction assay for BVDV
To evaluate the antiviral activity of TSC against BVDV R 1-5 , MDBK cells were seeded in growing medium at a density of 1.3610 5 cells per well in a 24-well plate and incubated for 24 h at 37uC in a 5% CO 2 incubator. Then, cells were infected with approximately 100 plaque-forming units (PFUs) of BVDV R 1-5 or wt BVDV p0. Following 1 h of adsorption at 37uC, the inoculum was removed, cell monolayers were washed twice with phosphatebuffered saline (PBS), and 1 ml of overlay medium (2.5% DHS and 0.8% methylcellulose) containing 80 mM of TSC was added in each well. Mock-infected cells with and without TSC, and infected cells without TSC were included as controls. After 72 h of incubation at 37uC with 5% CO 2 , cells were fixed with formalin 10%, stained with 0.75% crystal violet, and the viral plaques counted.

Apoptosis measurements
MDBK cells were seeded in a 24-well plate as described above. After 24 h in a 5% CO 2 incubator at 37uC, cells were infected with wt BVDV p0 or R 1-5 (MOI 0.01). Following 1 h of adsorption at 37uC, the inoculum was removed, cell monolayers were washed twice with PBS and 1 ml of infection medium was added in each well. Mock-infected cells were included as controls. After 48 h of incubation at 37uC with 5% CO 2 , apoptosis was measured in infected and mock-infected cell cultures.
For apoptosis measurements, PE Annexin V Apoptosis Detection Kit I (BD Pharmigen, NJ, USA) was used following the manufacturer's instructions. Briefly, supernatants from infected and mock-infected cells were harvested and cells in the monolayer were collected by trypsinization and resuspended in 0.5 ml of infection medium. Adherent and floating cells were pooled and pelleted by centrifugation at 5006g for 10 min. Pelleted cells were washed twice with 0.5 ml of cold PBS. Before the second centrifugation (5006g, 10 min), cells were counted in a Neubauer chamber. To obtain the dead-cell control for apoptosis measurements, one mock-infected cell suspension was heated at 65uC for 30 min. Then, cells were resuspended in binding buffer (4610 6 cells/ml) and 100 ml of each cell suspension was stained with 5 ml Annexin V (conjugated with Phycoerythrin-PE-) and 5 ml 7-Amino-Actinomycin (7AAD) and incubated for 15 min at room temperature in the dark. Finally, 400 ml of binding buffer was added and the cells were analyzed by flow cytometry (PARTEC, PAS-III). PE Annexin V and 7AAD fluorescence was collected at FL 2 (550-600 nm) and FL4 (640-700 nm) channels, respectively. Histograms and density plots of 20,000 events were obtained and analyzed with WinMDI 2.9 software [39].
At 48 h p.i. some infected cultures already presented viral CPE, which is caused by cell apoptosis. The number of cells that would have died by this mechanism and could no longer be detected was

Multiple-step growth curve
To quantify the infectious virus produced by BVDV R 1-5 in the absence of TSC, MDBK cells were seeded in a 24-well plate as stated above and infected with BVDV R 1-5 or wt BVDV p0 (MOI 0.01). Following 1 h of adsorption at 37uC, the inoculum was removed, cell monolayers were washed twice with PBS, and 1 ml of infection medium was added to each well. The culture supernatant was collected every 24 h for 7 days and clarified by centrifugation for 10 min at 1,0006g at 4uC. The infectious virus in each sample was quantified by the PFU method in MDBK cells.
Given that BVDV R 4 and R 5 showed CPE and levels of cell apoptosis similar to those of BVDV R 1 and R 2 , respectively (see results section), they were not included in the following assays (unless otherwise indicated).

Fluorescent microscopy
MDBK cells were seeded in a 24-well plate at a density of 1.3 610 5 cells per well containing a circular slide cover of 10 mm in diameter. After 24 h in a 5% CO 2 incubator at 37uC, cells were infected with BVDV R 1 , R 2, R 3 or wt BVDV p0 (MOI 0.01). Following 1 h at 37uC, the inoculum was removed, cell   Viral RNA production MDBK cells were seeded in a 24-well plate as stated above. After 24 h in a 5% CO 2 incubator at 37uC, cells were infected with BVDV R 1 , R 2, R 3 or wt BVDV p0 (MOI 1). Following 1 h of adsorption on ice, the inoculum was removed, cell monolayers were washed twice with PBS, and 1 ml of infection medium was added to each well. After 24 h of incubation at 37uC, intracellular RNA was extracted with TRIzol reagent (Gibco-BRL, CA, USA). RNA was retrotranscribed with random hexamers (Biodynamics)

Real-time PCR
To measure the intracellular viral RNA (ivRNA), a 108-bp fragment from the 59NTR was amplified with the following oligonucleotides: 59NTR forward 59GAGGCTAGC-CATGCCCTTAGT39 and 59NTR reverse 59TCGAACCACT-GACGACTACCCT39. The reaction was carried out in 2X Power SYBR Green PCR Master Mix (Applied Biosystems, UK) in an ABI 7500 apparatus (Applied Biosystems), using the following experimental run protocol: 10 min of activation at 95uC, followed by 45 cycles of amplification and quantification (15 sec at 95.0uC, 1 min at 60.0uC and 35 sec at 78.5uC) during which the SYBR GREEN signal was measured. To normalize target amplification data, a 91-bp fragment of bovine b-actin mRNA was amplified simultaneously in each sample and used as endogenous control (forward oligonucleotide: 59CCCACACGGTGCCCATCTAT39 and reverse oligonucleotide: 59 CCACGCTCCGTGAG-GATCTTC39). Each sample was tested in triplicate and both positive and negative controls were included.
Data were analyzed by means of the relative standard curve method [40] to quantify the ivRNA at 24 h p.i. in the cell cultures infected with BVDV R 1 , R 2 and R 3 in relation with those infected with wt BVDV p0.

Sequencing
Total RNA was extracted with TRIzol reagent (Gibco-BRL) from wt BVDV p0, BVDV-TSC r T 1-5 and BVDV R 1-5 virus suspensions obtained from clarified supernatants of infected cells. The RNA obtained was denatured at 70uC for 5 min and primed with the designed specific oligonucleotides ( Table 1). The reverse transcription reaction was performed at 42uC for 90 min, using 10 U/ml of ArrayScript TM Reverse Transcriptase (Ambion, Inc.). The PCR fragments that cover the complete coding region of all the non-structural proteins (Npro, NS2-NS5B) and the glycoprotein Erns were obtained using specific oligonucleotides (Table 1) by PCR amplification with 0.025 U/ml of PFU DNA polymerase (Promega). Both DNA strands were sequenced using an ABI ABI3130XL sequencer 131 (Applied Biosystems; Unidad de Genómica, Instituto de Biotecnología, INTA, Castelar, Buenos Aires, Argentina). The sequences were aligned using ClustalX 1.83

Statistical analysis
The results were analyzed by Student's t test using InfoStat version 2011 [43]. A statistically significant difference was considered if p values were less than 0.050.

Results
The resistance to TSC is conserved after 20 passages BVDV R 1-5 were obtained after 20 passages of BVDV-TSC r T 1-5 in the absence of TSC. BVDV R 1-4 were not significantly inhibited by TSC, whereas wt BVDV p0 was completely inhibited (TSC vs untreated: R 1 p = 0.560; R 2 p = 0.054; R 3 p = 0.449; R 4 p = 0.382; p0 p = 0.003) (Figure 1). However, with the number of PFUs analyzed, BVDV R 5 showed a reduction in the number of viral plaques of approximately 44%, suggesting a slight decrease in drug resistance (TSC vs untreated: R 5 p = 0.002).

A delay in CPE production is observed in some viral populations during passages
The infection of MDBK cells with wt BVDV NADL cp at a MOI of 0.01 required 48 h of incubation at 37uC to reach 80-90% of CPE in the cell monolayer. However, along the passages of BVDV-TSC r T 1-5 in the same conditions, some viral populations showed a delay in producing CPE. To reach 80-90% of CPE, the infections with BVDV-TSC r T 2 and T 5 required 72 h of incubation from the 1 st to the 3 rd passage and then 120 h of incubation from the 4 th to the 20 th passage. Similarly, BVDV-TSC r T 3 required 120 h of incubation from the 13 th to the 20 th passage. This delay in producing CPE was also observed in the growth of viral plaques when the infected cultures were incubated in semi-solid medium. BVDV R 2 , R 3 , and R 5 needed 72-96 additional hours of incubation to properly exhibit the viral plaques.
The delay in producing CPE is correlated with a reduced induction of apoptosis in MDBK cells BVDV cp isolates kill cells by triggering apoptosis [30], [32], [44]. In accordance with the microscopic observations of the viral CPE, the infection of MDBK cells with BVDV R 1 or R 4 produced levels of apoptosis similar to those produced by wt BVDV p0 (13.1%, 10.8%, and 11.0% of PE Annexin-positive cells, respectively), whereas the infections with BVDV R 2 and R 5 produced lower level of apoptosis (5.7% and 5.5% of PE Annexinpositive cells, respectively). BVDV R 3 showed an intermediate value (9.1% of PE Annexin-positive cells) (Figure 2).
Due to viral CPE, the number of remaining cells at 48 h p.i. in the cultures infected with wt BVDV p0, R 1 or R 4 was lower than that in the cultures infected with BVDV R 2 , R 3 or R 5 (2.4-3.0610 5 and 4.8-5.5610 5 , respectively) and that in mock-infected cultures (6.4610 5 ). Taking into account that apoptotic cells could be quantified only in the remaining adherent and floating cells, the total number of cells that would die by viral CPE in each infection was estimated (see Materials and Methods Section). Having considered the differences in CPE between each viral infection, the results showed that the number of cells that would die by BVDV R 2 , R 3 and R 5 infections was markedly lower than that in wt BVDV p0 infection (15.6%, 27.5%, 15.5% and 48.6% of estimated dead cells, respectively) ( Figure 3).

The delay in CPE is not caused by a decreased infectious virus production
Although some viral strains produced no CPE, all viral populations showed the highest titers of infectious virus at 48 h p.i. (Figure 4). At that time, most BVDV R (R 1 and R 3-5 ) showed slightly higher virus titers (approximately three times higher) than wt BVDV p0, but, notably, BVDV R 2 showed viral titers approximately 30 times higher.
On the other hand, wt BVDV p0, R 1 and R 4 showed a marked decrease in viral titers at 72 h p.i., consistent with the lack of viable host cells as a result of the viral CPE. However, MDBK cells infected with R 2, R 3 and R 5 continued producing infectious virus for at least four additional days.
In addition, expression of the NS3 antigen was monitored by indirect immunofluorescence. In accordance with the results described above, cell monolayers infected with BVDV R 2 or R 3 were still confluent until 72 h p.i and maintained a continuous production of NS3 ( Figure 5).

Viruses that have a delayed CPE exhibit less accumulation of viral RNA
One of the factors that contribute to cell apoptosis and viral cytopathogenicity is the accumulation of viral RNA in the cells infected with cp BVDV [45]. In line with this, our results showed that the ivRNA levels in BVDV R 2 and R 3 infections were lower than those of wt BVDV p0 and R 1 . BVDV R 2 , which together with BVDV R 5 exhibited the lowest level of cell apoptosis in the infected cultures, also presented the lowest level of ivRNA accumulation (approximately three times lower than wt BVDV p0). On the other hand, BVDV R 3 , which showed an intermediate level of decrease in cell apoptosis, also presented intermediate levels of ivRNA ( Figure 6).

Discussion
We have previously reported the selection of TSC-resistant BVDV mutants (BVDV-TSC r T 1-5 ) with mutations in NS5B (T 1-5 : N264D; T 1 : A392E) [38]. In the present study, we performed 20 passages of BVDV-TSC r T 1-5 in the absence of TSC to analyze their genetic stability.
The resistance to TSC remained stable and mutations in NS5B associated with resistance were conserved in all the viral populations obtained (BVDV R 1-5 ). However, BVDV R 1-5 produced higher titers of infectious virus than wt BVDV p0 and, along the passages in the antiviral-free environment, three out of the five viral populations (BVDV R 2 , R 3 and R 5 ) showed a delay in producing CPE.
It is known that drug-resistant mutants may have fitness costs in the absence of the selecting drug, but the occurrence of compensatory mutations may allow them to partially or even completely regain fitness [46][47][48]. A previous work has shown that BVDV variants that carry the NS5B N264D mutation, which confers resistance to TSC and to other non-nucleoside polymerase inhibitors (NNI) [49], [50], present a replication disadvantage when competing with the wild-type (wt) virus in an antiviral-free environment [51]. In the present work, the replication of BVDV-TSC r in the absence of TSC led to the fixation of different mutations along the NS region that could compensate for the reduced fitness caused by the N264D mutation and confer some stability to the TSC-resistant phenotype. Specifically, mutations were detected in NS2 (BVDV R 1 ), NS4B (BVDV R 3 ), NS5A (BVDV R 2-5 ) and NS5B (BVDV R 3 and R 4 ). Particularly, mutation NS5B I261M (BVDV R 3 ) has been reported to compensate for the reduced fitness caused by the N264D mutation and to confer resistance to other BVDV NNI [49][50][51]. Additional mutations suggest that the effect of the NS5B N264D mutation on viral fitness could be also counterbalanced by a change in other NS proteins. Because BVDV-TSC r were not replicated under a selective pressure, these additional mutations may have resulted from the genetic drift during the passages and could have also been detected in the wild type virus if it had been passaged in conditions similar to those of BVDV-TSC r .
As mentioned above, BVDV R 2 , R 3 , and R 5 showed a delay in producing CPE. The cp biotypes of BVDV induce cell death by apoptosis [44], [52], triggered by the accumulation of viral doublestrand RNA [45], [53] and the continuous expression of NS3, the cleaved form of the nonstructural protein NS2-3 [14], [15]. Although viral RNA production in cp BVDV infections is higher than that in ncp ones, both show similar production of infectious viral particles [18], [53][54][55][56]. Accordingly, cell cultures infected with BVDV R 2, R 3 and R 5 showed a decrease in the induction of apoptosis and in the accumulation of viral RNA. However, BVDV R 1-5 produced higher infectious virus titers than wt BVDV p0, and due to host cell survival, BVDV R 2, R 3 and R 5 produced viral antigens and infectious viral progeny for at least four additional days.
BVDV NADL cytopathogenicity is determined by an in-frame insertion in the NS2 coding region of a 270-nucleotide cellular mRNA sequence (called cIns), which modulates NS3 production and upregulates RNA replication [18]. As for other cp BVDV strains, the deletion of the inserted host-derived sequences could lead to reversion to the ncp phenotype [54], [57], [58]. In this work, BVDV R 1-5 showed no deletion (or insertion) in the cIns or in any of the viral sequences. In addition, BVDV R 2, R 3 and R 5 showed no amino acid changes in NS2 or NS3, whose bond cleavage is required for RNA replication and is linked to BVDV cytopathogenicity [14], [15]. No amino acid changes were detected in NS3-cofactor NS4A or in Npro and Erns, which may play a role in blocking IFN induction and therefore could also affect the viral CPE [5][6][7][8], [10], [11]. This suggests that the delay of BVDV R 2 , R 3 and R 5 in producing CPE may not result from an alteration in the production or function of these viral proteins. On the other hand, BVDV R 3 showed a mutation in NS4B (E28G). This protein would also play a role in BVDV cytopathogenicity [21], and substitutions for residue NS4B Y15 have been shown to attenuate viral CPE despite NS3 production [55].
Interestingly, most of the additional amino acid changes found in BVDV R 2 and R 3 and the only one detected in BVDV R 5 were in NS5A. BVDV NS5A is a zinc metalloprotein essential for the formation of a functional replicase complex [22], [59], [60] that seems to inhibit TNF-a and double-strand RNA-induced NF-kB activation, which is involved in inflammation, innate immune response, cell survival, and pathogenesis or persistence of BVDV infections [24], [61], [62]. BVDV NS5A is composed of three domains. Mutations were found in domain I (I9M in BVDV R 2 , V177A in BVDV R 5 and K195E in BVDV R 3 ) and domain II (N305Y in BVDV R 2 and E370K in BVDV R 3 ), both critical for RNA replication [22], [63]. Thus, the additional mutations in NS5A could alter its function during viral RNA replication and/or interfere with its interaction with cellular proteins involved in the induction of apoptosis, thus leading to the observed reduction in viral cytopathogenesis.
To our knowledge, this is the first report on the stability and evolution of non-nucleoside polymerase inhibitor-resistant BVDV. The study of the TSC-resistant viruses that showed reduced CPE and prolonged virus production may be an interesting tool to learn more about the virus-cell interaction, viral cytopathogenesis, attenuation and persistence. Further studies and reverse genetic experiments would be needed to identify the specific phenotype of each of the additional mutations observed.