**Real Time Analysis of Bovine Viral Diarrhea Virus (BVDV) Infection and Its Dependence on Bovine CD46**

### **Christiane Riedel 1,\*, Hann-Wei Chen 1, Ursula Reichart 2, Benjamin Lamp 3, Vibor Laketa 4,5 and Till Rümenapf <sup>1</sup>**


Received: 16 December 2019; Accepted: 14 January 2020; Published: 17 January 2020

**Abstract:** Virus attachment and entry is a complex interplay of viral and cellular interaction partners. Employing bovine viral diarrhea virus (BVDV) encoding an mCherry-E2 fusion protein (BVDVE2-mCherry), being the first genetically labelled member of the family *Flaviviridae* applicable for the analysis of virus particles, the early events of infection—attachment, particle surface transport, and endocytosis—were monitored to better understand the mechanisms underlying virus entry and their dependence on the virus receptor, bovine CD46. The analysis of 801 tracks on the surface of SK6 cells inducibly expressing fluorophore labelled bovine CD46 (CD46fluo) demonstrated the presence of directed, diffusive, and confined motion. 26 entry events could be identified, with the majority being associated with a CD46fluo positive structure during endocytosis and occurring more than 20 min after virus addition. Deletion of the CD46fluo E2 binding domain (CD46fluoΔE2bind) did not affect the types of motions observed on the cell surface but resulted in a decreased number of observable entry events (2 out of 1081 tracks). Mean squared displacement analysis revealed a significantly increased velocity of particle transport for directed motions on CD46fluoΔE2bind expressing cells in comparison to CD46fluo. These results indicate that the presence of bovine CD46 is only affecting the speed of directed transport, but otherwise not influencing BVDV cell surface motility. Instead, bovine CD46 seems to be an important factor during uptake, suggesting the presence of additional cellular proteins interacting with the virus which are able to support its transport on the virus surface.

**Keywords:** *Pestivirus*; BVDV; CD46; life cell imaging; attachment; surface transport

#### **1. Introduction**

Before entering a host cell, viruses have to establish contact with the cell and travel to sites that are suitable for entry. This process of attachment can involve different receptor molecules and transport mechanisms, which can be classified into diffusion, drifts, and confinement [1]. Diffusive motion can be caused by interaction of viruses with a number of receptor molecules that is too low to cause confinement or can support the screening of the cell surface for suitable sites of endocytosis. Directed motion is the result of interaction of the virus with cellular proteins or lipid structures that are linked to F-actin. Understanding the extracellular movement of virus particles, as well as the kinetics of entry and protein expression, provides important insights into the key players involved in attachment and entry and in infection dynamics in general.

For the family *Retroviridae*—facilitated by the ease of generation of genetically labelled virus particles—attachment and entry dynamics have been studied excessively. Virus surfing—the directed, actin-dependent transport of virus particles on the outside of filopodia, cytonemes, and retraction fibres—was also discovered employing a member of the *Retroviridae*, namely murine leukemia virus (MLV) [2]. This mode of extracellular, directed transport has since been described for members of the *Adenoviridae* [1], *Herpesviridae* [3], and *Papillomaviridae* [4].

For the medically relevant family *Flaviviridae*, the lack of genetically labelled virus particles has been overcome by the utilization of lipophilic dyes or covalently linked fluorophores [5–8]. These studies demonstrated the diffusive movement of dengue virus along the cell surface to clathrin coated pits and the subsequent association with specific markers of endocytosis [8] and helped in the identification of T cell immunoglobulin mucin-1 as a dengue virus receptor [9]. Also, labelled hepatitis C virus (HCV) particles demonstrated the involvement of actin in cell surface and intracellular virus transport [7].

The genus *Pestivirus* is also part of the *Flaviviridae* and characterized by the presence of three viral surface glycoproteins—Erns, E1, and E2—and an additional N-terminal protease, Npro. Members of this genus are pathogens of cloven-hoofed animals, including the highly economically relevant pathogens bovine viral diarrhea virus (BVDV) and classical swine fever virus (CSFV). In recent years, genetically labelled BVDV and CSFV clones have been constructed based on N-terminal fusion of luciferase or fluorophores to either Erns or E2 [10–13]. For a BVDV E2-fluorophore fusion protein, visualization of purified particles in fluorescence microscopy based on the specific fluorescence signal has been reported [13]. BVDV enters host cells after initial interactions of Erns with heparan sulphates [14] and subsequent binding of E2 to its cellular receptor, bovine CD46 [15,16], by clathrin mediated endocytosis [17,18] and fusion occurs after endosomal acidification [18,19]. Interestingly, recent evidence implies that BVDV is preferably transmitted by direct cell-to-cell spread in a CD46 independent manner [12], indicating the involvement of additional factors in BVDV spread. The BVDV-1 clone employed in [12] also encodes for an mCherry-E2 fusion protein, in the backbone of strain NADL, demonstrating the utilization of different transmission modes in the presence of an E2-fusion protein.

Bovine CD46 is an ubiquitously expressed, type I transmembrane glycoprotein and exists as different splice variants, affecting the length of its heavily O-glycosylated, membrane proximal regions (STP) and its cytoplasmic C-terminus. The membrane distal, extracellular part of the protein consists of four complement control protein modules (CCP) which have been implicated in the binding of a variety of pathogens, leading to its description as a "pathogens magnet" [20]. Physiologically, CD46 is a cofactor of the inactivation of complement components C3b and C4b and also involved in T cell regulation, modulation of autophagy and reproductive biology (reviewed in [21]). Its surface levels are regulated by clathrin-dependent endocytosis [21] or micropinocytosis after cross-linking [22]. BVDV E2 binding to CD46 is mediated by the 30 C-terminal amino acids of CCP-1 [16]. Interestingly, CD46's physiological ligands mainly interact with the membrane proximal CCPs 3–4, while pathogens mostly interact with the membrane distal CCPs 1–2 [23].

While the function of bovine CD46 as a receptor for BVDV is well documented in the literature, it is currently unclear what its exact functions are during virus attachment and entry. Also, it has been shown that bovine CD46 is not sufficient for entry. To determine which stages of the entry process—attachment, surface motion, or uptake—are affected by CD46, we conducted life cell imaging experiments employing purified, fluorophore-E2 labelled BVDV particles and SK6 cell lines inducibly expressing fluorophore labelled CD46 with and without the virus interacting CCP-1 module [16].

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

#### *2.1. Viruses and Cells*

For propagation, cells were cultured in DMEM (Capricorn, Ebsdorfergrund, Germany) supplemented with 10% FCS (Bio & Sell, Feucht/Nürnberg, Germany) and penicillin/streptomycin (Merck, Darmstadt, Germany) at 37 ◦C, 5% CO2. SK6 tet-on cells inducibly expressing bovine CD46 labelled with mCherry or mClover (CD46fluo) have been described in [13]. The E2 binding CCP of CD46—as reported by [16]—was deleted by PCR (Vazyme Biotech, Nanjing, China) in the previously described CD46fluo-encoding construct employing the following primers (Eurofins, Ebersberg, Germany): forward: cgaCGGTGTCCTACCCTAGCTGATC; reverse: GGCATCGGAGGACGTGGGCAG, resulting in CD46fluoΔE2bind. SK6 tet-on cells were transfected with CD46fluoΔE2bind by electroporation and clonally selected as described in [13].

For the determination of susceptibility of SK6 CD46fluo or CD46fluoΔE2bind cells to BVDVE2-mCherry, 2 <sup>×</sup> 10<sup>5</sup> MDBK, SK6 CD46fluo or CD46fluoΔE2bind cells or SK6 cells inducibly expressing GFP were seeded in each well of a 24-well plate 24 h before the start of the experiment. SK6 cells are of porcine origin and porcine CD46 is not a receptor for BVDV [16]. Cells were induced 16 h before infection with 2.5 μg/mL doxycycline (Merck, Darmstadt, Germany). Cells were infected with ten-fold dilutions of a BVDV-1 strain expressing mCherry fused to the N-terminus of E2 (BVDVE2-mCherry) [13] for 4h. Thereafter, medium was exchanged to medium containing 1% carboxymethylcellulose to prevent virus transfer via the culture medium. 48 h after infection, infected foci were quantified by fluorescence microscopy (Olympus IX-70, Tokyo, Japan) to determine the titer and the susceptibility was calculated in % by dividing the titer determined on a given cell line through the titer determined on MDBK cells. BVDVE2-mCherry was propagated on MDBK cells in 5-layer tissue culture flasks (Corning, Corning, NY, USA) and purified and concentrated as described in [13]. The sequence of the plasmid encoding the full sequence of BVDVE2-mCherry is provided as supplementary data.

For life cell imaging, 2 <sup>×</sup> 10<sup>4</sup> SK6 CD46fluo or CD46fluoΔE2bind cells were seeded in each well of an IBIDI μ slide 8 well chamber slide 24 h before the start of the experiment in medium devoid of phenol red. 16 h before the start of the experiment, expression of CD46fluo or CD46fluoΔE2bind was induced by addition of 2.5 μg/mL doxycycline (Merck, Darmstadt, Germany).

#### *2.2. Life Cell Imaging*

All data was acquired on an Andor Revolution spinning disk confocal microscope (Oxford Instruments, Abingdon, UK) based on a Yokogawa CSU W1 with Nikon Ti2 stand equipped with two Andor DU-888 cameras. Imaging was performed in a humidified chamber at 37 ◦C and 5% CO2. An APO TIRF NA 1.49 100× magnification oil immersion objective was used, resulting in a lateral pixel size of 0.13 μm. For the detection of CD46fluo and CD46fluoΔE2bind, mClover was excited by a 515 nm laser for 0.1 s and the emitted signal was bandpass filtered with a 540/30 nm filter. E2-mCherry was excited by a 594 nm laser for 0.1 s and the emitted light was bandpass filtered with a 647/57 nm filter before detection. Excitation and detection of each fluorophore was performed consecutively.

For the examination of virus attachment, cell surface transport, and entry, one frame/10 s was recorded starting directly, 5, 10, 15, 20, or 25 min after virus addition (multiplicity of infection (MOI) = 10) and data was acquired for 5 or 10 min. Three focal planes (*Z*-step size = 0.8 μm) were acquired for each time point to compensate for potential focusing errors and to increase the area accessible for analysis. The *z*-level was chosen close to the cover slip to allow imaging of filopodia, retraction fibers, lamellipodia, and the lamella. In order to judge intracellular background levels during E2-mCherry excitation, one *z*-stack was acquired of each field of view to be employed in the experiment before addition of virus.

To follow the time course of infection, cells were imaged at one *z*-level every 10 min starting 60 min after infection until 16 h after infection. Two different MOIs (1 and 10) were employed to assess the effect of different MOIs on the development of E2-mCherry signal. Due to phototoxicity, it was not possible to acquire several *z*-levels.

#### *2.3. Image Processing and Analysis*

All image processing and particle tracking was performed in ImageJ [24]. Raw frames were filtered with a Gaussian filter (sigma = 1.4) to improve data visualization. For initial particle tracking, maximum intensity projections along the *z*-axis were generated and particles tracked employing ImageJ's [24] manual tracking plug-in. Subsequently, tracks were verified on the full z-stack and their localization (lamellipodium, lamella, cell body) as well as their direction of movement (towards, tangential, away, random, with reference to the cell body) and the association with CD46fluo or CD46fluoΔE2bind was documented. A python script (supplementary information) was used to analyze the output of the manual tracking plug-in with regard to direct and relative distance covered by a given particle, directionality (ratio direct to relative distance), average, minimum and maximum speed as well as the standard deviation of speed for a given track. The mean squared displacement (MSD) was calculated in R employing a customized script based on the code developed by [25]. Fits of the MSD curves were employed to calculate the velocity (V) of directed movement in case of an exponential slope using the following formula: *y* = *V*2*x*<sup>2</sup> + *D*0*x* + *a* and the diffusion coefficient (D0) was calculated using *y* = 4*D*0*x* + *a* in case of a linear slope. A two-tailed Student's *t*-test was performed to assess the likelihood of significant differences.

#### **3. Results**

#### *3.1. BVDV Entry Is a Slow Process*

To assess the infection cycle of BVDV in real time, we employed a previously described labelling strategy of the viral E2 surface glycoprotein [12,13]. SK6 cells inducibly expressing the fluorophore labelled, cellular surface receptor, bovine CD46 were chosen as system to study the infection cycle, as these cells demonstrated a high susceptibility to infection with BVDV after induction. SK6 cells not expressing bovine CD46 display a low susceptibility to BVDV infection, which is not mediated by porcine CD46, as porcine CD46 has previously been shown not to be involved in BVDV invasion [16]. After initial experiments employing either a system of mClover labelled BVDV and mCherry labelled CD46 or vice versa, we decided to employ mCherry labelled BVDV (BVDVE2-mCherry) and mClover labelled CD46 (CD46fluo), as strong photobleaching of mClover labelled BVDV rendered it inapplicable for life cell imaging.

Time series of 5 or 10 min duration with a frame rate of one frame/10 s were acquired at different time points (0–25 min in 5 min steps) after addition of BVDVE2-mCherry (MOI 10). Association of mCherry positive particles with the surface of SK6 CD46fluo cells could readily be observed and 801 particles were tracked by hand (*n* cells = 160). Due to the low signal to noise ratio, automatic tracking approaches did not improve the ease of analysis and were therefore not utilized. Entry events were rarely observed (*n* = 26, 2.8% of total particles tracked) and the majority occurred more than 20 min after virus addition (Figure 1A). In 65.5% of these entry events, the signal of the virus was associated with a CD46fluo positive vesicular structure. Association with such a vesicle was maintained after endocytosis in 76% of events. After endocytosis, particles could on average be tracked for 2.8 min and migrated on average 3 μm inside the cell (Figure 1A). Two examples of mCherry positive particles entering a cell are shown in Figure 1B,C and movies 1 and 2.

**Figure 1.** Characterization of BVDVE2-mCherry entry into SK6 CD46fluo cells. (**A**) Characterization of 26 entry events of BVDVE2-mCherry in SK6 CD46fluo cells with regard to occurrence after virus addition, association with CD46fluo signal, intracellular persistence of the E2-mCherry signal and intracellular distance travelled. Outliers (Q3 + 1.5-times interquartile range) are indicated by dots. (**B**,**C**) Examples of entry events of BVDVE2-mCherry (red) into SK6 CD46bov (green) cells. The full field of view at the time of the start of acquisition (time after virus addition is specified in the top right corner) is shown and the area of interest is indicated by grey squares. Frames as depicted in the detail images were acquired every 10 s for up to 10 min after the start of acquisition. Times indicated in s refer to the start of acquisition. The length of the scale bar in the detail images in (**B**) = 2.5 μm and in (**C**) = 5 μm.

Due to the intracellular background level in the mCherry emission range, the amount of intracellular, mCherry positive foci and their migration behavior was not analyzed as an unambiguous differentiation between background and specific signal was not possible.

Particles travelled with an average speed of 0.062 μm/s (s.d. 0.036 μm/s), an average maximum speed of 0.147 μm/s (s.d. 0.15) and an average directionality—defined as the ratio of direct distance versus real distance covered by a particle—of 0.34 (s.d. 0.22) on SK6 CD46fluo cells.

Directed transport of viruses along the outside of filopodia, cytonemes, and retraction fibers has been described for enveloped and non-enveloped viruses. The potential usurpation of this transport route was hence also examined in the context of BVDVE2-mCherry. Virus surfing could be observed for 16 particles (2.0% of total tracks, movie 3). These particles migrated with an average velocity of 0.070 μm/s and an average directionality of 0.446, indicating an advantage of this type of transport regarding movement in a given direction. The average maximum speed of these particles was 0.176 μm/s, indicating a potential coupling to retrograde actin flow [2,26].

#### *3.2. First Detection of BVDV E2-mCherry Signal Is Depending on MOI*

In order to further analyze the progression of infection and to visualize the intracellular distribution of E2-mCherry over time, time series starting 60 min after virus addition and running for 16 h (1 frame every 10 min) were recorded. In all cells observed, an evenly distributed, slightly granular E2-mCherry signal was present from the initial detection of a specific signal (movies 4 and 5). This intracellular distribution pattern was reminiscent of an ER staining pattern, which is in good accordance with the localization of E2 in the ER lumen as already reported by [27,28]. Interestingly, the time after which an E2-mCherry signal could be resolved correlated with the MOI employed in a given experiment. For an MOI of 1, E2-mCherry could be detected on average 645 min after virus addition (*n* = 13) (Figure 2A, movie 4), whereas E2-mCherry could already be resolved on average 195 min (*n* = 16, movie 5) after infection if an MOI of 10 was used.

To gain further insights into the dynamics of E2-mCherry trafficking 20 h after infection, cells were imaged for 5 min with a frame rate of 1 frame/10 s. In addition to the diffuse, granular staining pattern of E2-mCherry, point-shaped, high signal intensities could be observed (Figure 2B, movie 6). They were partially associated with high CD46fluo signal intensities and this association was maintained during the whole course of the experiment. High intensity E2-mCherry foci could both be stationary or highly mobile (up to 0.5 μm/s), especially in the cell periphery (movie 6).

**Figure 2.** E2-mCherry signal development but not signal distribution is depending on the MOI. (**A**) Occurrence of E2-mCherry signal in min after addition of BVDVE2-mCherry at an MOI of 1 (blue) or 10 (green) to SK6 CD46fluo cells. SK6 CD46fluo cells were imaged recording one frame every 10 min at one z-level for 16 h. The mean is indicated by x and outliers (Q3 + 1.5-times interquartile range) are indicated by dots. (**B**) Distribution of E2-mCherry (red) and CD46fluo (green) signal 20 h after infection with BVDVE2-mCherry in SK6 CD46fluo cells. SK6 CD46fluo cells were imaged for 10 min recording 3 z-levels (0.8 μm) every 10 s (movie 6). Points of high E2-mCherry intensity that are colocalizing with CD46fluo are indicated by white circles.

#### *3.3. CD46fluo Decreases the Speed of Directed Surface Motion of BVDVE2-mCherry*

In order to elucidate the effect of specific receptor binding on virus entry, SK6 cells inducibly expressing fluorophore-labelled CD46 with a deleted E2-binding domain (CD46fluoΔE2bind) were generated based on previous results by [16], reporting the CCP-1 as interaction partner. These cells showed a susceptibility to BVDV comparable to the susceptibility of SK6 cells inducibly expressing GFP (Figure 3). GFP expressing cells were chosen as control to account for a potential effect of induced expression on virus replication. The susceptibility of SK6 CD46fluoΔE2bind cells was more than 250-fold reduced in comparison to SK6 cells expressing unmodified CD46fluo after induction of expression.

This system was chosen over experiments employing inhibition of virus attachment by antibodies targeting neutralizing epitopes of E2 or functionally important domains of CD46 as it—in our opinion—provided the most direct and defined system to study the role of E2-CD46 interaction in the viral life cycle by life cell imaging. An inhibition of infection by BVDV tagged with a fluorophore at the E2 N-terminus employing before mentioned antibodies has already been demonstrated by [12] and this inhibition is comparable to BVDVs not bearing a tag at the E2 N-terminus. This clearly demonstrates the importance of the same domains in CD46-E2 interactions.

**Figure 3.** Deletion of CCP-1 reduces CD46fluo effect of the susceptibility of SK6 cells to the level of GFP-expressing controls. Susceptibility of SK6 CD46fluo (blue), CD46fluoΔE2bind (green) and GFP-expressing (grey) SK6 cells to infection with BVDVE2-mCherry in % of MDBK cells (=100%) with and without the induction of expression by the addition of doxycycline (Doxy) (*n* = 6). Cells were infected with serial dilutions of BVDVE2-mCherry for 4 h. Subsequently, medium was exchanged to medium containing 1% carboxymethycellulose and E2-mCherry positive foci were detected 48 h after infection by fluorescence microscopy to determine the focus forming units (ffu)/mL. Susceptibility of a given cell line was calculated as the percentage of ffu/mL determined for MDBK cells, which was set to 100%.

To assess the effect of receptor binding on the surface movement of BVDVE2-mCherry, 1081 mCherry-positive particles were tracked on SK6 CD46fluoΔE2bind cells (*n* cells = 297). Of those 1081 particles, only 2 particles (0.2%) could be observed entering a cell, which is 7% of the entry events found for SK6 CD46fluo cells. One of these events was associated with a CD46fluoΔE2bind after uptake.

In order to better understand the movement of particles on the surface of SK6 CD46fluo and SK6 CD46fluoΔE2bind cells, particle trajectories were further analyzed. The average direct distance covered on the cell surface during a given observation period was 3.15 μm (s.d. 2.71 μm) for SK6 CD46fluo and 2.83 μm (s.d. 2.18 μm) for SK6 CD46fluoΔE2bind cells (Figure 4), while the average real distance covered was 10.9 μm (s.d. 8.64 μm) for CD46bov and 9.14 μm (s.d. 6.06 μm) for CD46bovΔE2bind. The average directionality—defined as the quotient of direct versus real distance—was 0.34 (s.d. 0.22) for SK6 CD46bov and 0.37 (s.d. 0.23) for SK6 CD46bovΔE2bind cells and therefore seems unaffected by CD46 binding. Particles on SK6 CD46fluo cells travelled with an average velocity of 0.062 μm/s (s.d. 0.036 μm/s) and an average maximum speed of 0.147 μm/s (s.d. 0.15). Particles on SK6 CD46fluoΔE2bind cells were characterized by an average velocity of 0.063 μm/s (s.d. 0.028 μm/s) and an average maximum speed of 0.136 μm/s (s.d. 0.063 μm/s), indicating no effect of CD46 binding on observed velocities.

**Figure 4.** Movements of BVDVE2-mCherry on the surface of SK6 CD46fluo (blue) or CD46fluoΔE2bind cells (green) are comparable. Box and Whisker blots of maximum and mean velocity, directionality, real distance and direct distance of particles tracked on the surface of SK6 CD46fluo (blue, *n* = 801) or CD46fluoΔE2bind cells (green, *n* = 1081), respectively. The mean is indicated by x and outliers (Q3 + 1.5-times interquartile range) are indicated by dots.

On both, SK6 CD46fluo and SK6 CD46fluoΔE2bind cells, the majority of particles was localized on the lamella or lamellipodium up to 10 min after infection; beginning with 15 min after infection, the majority of particles was bound to the cell body (Supplementary Figure S1). On average 51% of the particles on SK6 CD46fluo cells were associated with an increased CD46 signal intensity for at least 3 consecutive frames. On SK6 CD46fluoΔE2bind cells, this percentage increased to an average of 74%. To assess potential differences in particle travelling directions, the overall direction of movement of a given particle was described as moving random or away from, tangential to or towards the cell body. Such a direction could unambiguously be assigned to 423 tracks on SK6 CD46fluo cells and 705 tracks on SK6 CD46fluoΔE2bind cells (Supplementary Figure S2). 3% of particles tracked on SK6 CD46fluo cells and 12% of particles tracked on SK6 CD46fluoΔE2bind cells showed a movement away from the cell body, whereas 57% and 42% respectively exhibited random movement. Tangential movement was observed for 17% of particles on SK6 CD46fluo cells and 18% of SK6 CD46fluoΔE2bind cells. 23% and 29% of particles exhibited movement towards the cell body on SK6 CD46fluo cells and SK6 CD46fluoΔE2bind cells, respectively. These results indicate that the major difference in the direction of particle transport is a four times reduced number of particles moving away from the cell body if CD46 is functional.

To further characterize the number of tracks following a certain type of motion on the cell surface and to determine transport velocities and the diffusion coefficients (D0), MSD analysis was performed. Within both datasets, tracks exhibiting characteristics of directed motion, as well as limited diffusion or purely diffusive motility, could be identified. For SK6 CD46fluo cells, 86 tracks (10.7% of total tracks, 47.5% of tracks in MSD analysis) exhibited linear slopes, indicative of purely diffusive motion, while 46 tracks (4.3% of total tracks, 31.7% of tracks in MSD analysis) were identified for SK6 CD46fluoΔE2bind cells (Figure 5A). The calculated diffusion coefficients were 0.020 μm2/s (s.d. 0.017 μm2/s) for SK6 CD46bov cells and 0.015 μm2/s (s.d. 0.019 μm2/s) for SK6 CD46fluoΔE2bind cells and were not significantly different (*p* = 0.103). 70 tracks exhibiting an exponential slope could be identified for SK6 CD46bov cells (8.7% of total tracks, 38.7% of tracks in MSD analysis) and 61 for SK6 CD46bovΔE2bind cells (5.6% of total tracks, 42.1% of tracks in MSD analysis) (Figure 5B). The calculated average particle velocity was 0.028 μm/s (s.d. 0.014 μm/s) for SK6 CD46fluo cells and 0.039 μm/s (s.d. 0.018 μm/s) for SK6 CD46fluoΔE2bind cells, suggesting a significantly increased directional transport speed on SK6 CD46fluoΔE2bind cells (*p* < 0.001). Twenty-five tracks on SK6 CD46fluo cells (3.1% of total tracks, 13.8% of tracks in MSD analysis) and 38 tracks on SK6 CD46fluoΔE2bind cells (3.5% of total tracks, 26.2% of tracks in MSD analysis) showed characteristics indicative of limited diffusion in MSD analysis.

**Figure 5.** The average velocity of directed motion of BVDVE2-mCherry particles is increased on the surface of SK6 CD46fluoΔE2bind cells. MSD analysis of particle movement on SK6 CD46fluo and CD46fluoΔE2bind cells. (**A**) Slopes of tracks on SK6 CD46fluo (blue) and CD46fluoΔE2bind (green) cells displaying a linear increase and distribution of the calculated diffusion coefficient D0 depicted as box and whisker blot. The mean is indicated by x and outliers (Q3 + 1.5 times interquartile range) are depicted as dots. Average slopes are depicted as thick black lines. (**B**) Slopes of tracks on SK6 CD46fluo (blue) and CD46fluoΔE2bind (green) cells displaying an exponential increase and distribution of the calculated particle velocities V, depicted as box and whisker blot. The mean is indicated by x and outliers (Q3 + 1.5 times interquartile range) are depicted as dots. Average slopes are depicted as thick black lines.

#### **4. Discussion**

Virus attachment and entry are governed by various affinities and highly dynamic processes. For BVDV, our findings suggest the presence of different types of motility on the cell surface, being diffusion, directed motion, and confined diffusion. This is different from observations of Dengue virus,

as the main motion type until the association with a clathrin-coated pit is diffusion [8]. Virus surfing, as a movement frequently utilized by other viruses (e.g., MLV), was rarely observed, albeit the presence of filopodia, cytonemes, and retraction fibers, implying a minor role of this transport mechanism for BVDV, at least in this experimental setup. All motion types could be observed independent of the presence or absence of the BVDV-binding competent cellular receptor, CD46 and track properties were comparable. Similarly, the proportions of cellular compartments—lamellipodium, lamella, and cell body—with associated virus particles were not affected by CD46's ability to bind E2 when comparing different time points after infection. However, when analyzing directed motion in more detail by MSD analysis, track velocities were significantly increased if CD46 was not able to interact with E2. This might suggest that another interaction partner on the cellular surface—for example heparan sulphate or an as of yet unidentified attachment factor—might be responsible for coupling to the underlying actin network and that CD46 is rather involved in deceleration of particle transport on the cell surface or the anchoring of particles at potential sites of endocytosis. The importance of attachment factors such as heparan sulphate has also been demonstrated for other members of the *Flaviviridae* (reviewed in [29]). Also, CD46 is not sufficient to render cell lines derived from non-cloven-hoofed animals or humans susceptible to infection with BVDV. Instead, it increases the amount of cell-associated virus [15]. This indicates that additional cellular factors are required for successful virus infection after binding to CD46. Interestingly, overexpression of CD46fluoΔE2bind resulted in a further drop of susceptibility to BVDV, which might indicate the sequestration of a potential additional entry factor by CD46fluoΔE2bind.

In general, the number of observable entry events into SK6 CD46fluo was very low (2.8% of tracked particles). This might indicate that cellular uptake of BVDV is a rather inefficient process. However, during data acquisition, we were not able to sample the whole cellular surface due to the short frame interval and phototoxicity. The focus of acquisition was on the cellular surface in contact with the growth support, to clearly depict cellular protrusions, the lamellipodium and the lamella. Particles could frequently be observed exiting the acquisition planes to regions of the cell body that were inaccessible in the chosen acquisition scheme. It is therefore possible that the low amount of observed entry events is—at least partially—due to a preference of BVDV to enter cells at specific areas of the cell body. Also, the specific infectivity of BVDVE2-mCherry is below 1:10, meaning that more than 10 genome equivalents are needed for successful establishment of infection. It is currently unclear what particle defects are responsible for this phenomenon, but it has to be taken into consideration that a certain number of particles are already unable to evoke cellular uptake. In order to clarify these open questions and to also examine the entry process of BVDV into primary host cells, MDBK CD46 knock-out cell lines would be a valuable tool for future elucidation of the entry process.

Contrasting the increased speed of particle transport, cell entry events were very rare if CD46 was E2-binding incompetent. This, in combination with the rather rare CD46—virus co-migration on the cell surface, might implicate that CD46 is important for the signaling to evoke endocytosis, but not for surface movements towards potential sites of endocytosis. The frequently observed association of virus particles with CD46 positive vesicles after endocytosis and their comigration for several frames further supports this hypothesis, but additional experiments will be required for verification. Still, BVDVE2-mCherry is able to enter host cells independent of a functional CD46 molecule. This strongly indicates the potential to use other entry receptors, but with much lower efficiency. The importance of factors different from CD46 in BVDV propagation is also highlighted by recent results reporting that BVDV cell-to-cell spread is independent of CD46 [12].

The CD46 splice variant employed in this study is identical to the bovine CD46 variant initially reported by [15]. Unfortunately, we were not able to elucidate the effects of different allelic versions of CCP-1 and different CD46 splice variants on the dynamics of BVDV entry due to the huge efforts of data analysis. A different permissiveness to infection of these different allelic and splice variants has however already been demonstrated [30] and their effect on the dynamics of BVDV entry will be an interesting topic of future studies.

*Viruses* **2020**, *12*, 116

Taking advantage of the easy tracking of virus protein production in infected cells, the effect of different MOIs on the earliest time point of specific fluorescence detection was examined. The highly significant difference observed leads to the conclusion that signal development is depending on the virus dose applied and that superinfection exclusion is not acting fast enough to prevent the infection of one cell with several particles in the context of this experimental setup [31]. Surveillance of the signal development of BVDVE2-mCherry after infection revealed an initially finely granular, likely ER-associated [27,28] distribution throughout the whole cell, which intensified over time and finally also developed strong signal foci. Therefore, it seems likely that E2 is quickly distributed throughout the cell and not restricted to and radially expanding from a potential initial replication site. To better understand viral replication sites, additional studies are warranted to elucidate how the intracellular distribution of E2 is reflecting the overall distribution of non-structural proteins like NS3 and NS5B as part of the replication complex. This, in combination with ultrastructural studies, will provide further insights into the morphology of replication sites and allow comparison to the detailed morphological studies already performed on the membranous replication compartments of HCV [32] and flaviviruses [33,34]. The occasional association of CD46 positive intracellular foci with E2 mCherry positive foci suggests that CD46 is usually excluded from E2 mCherry containing compartments or only present in amounts below the detection limit of this experimental setup. Whether this association is unfavorable for the virus or serves an as of yet unknown purpose—like facilitation of exocytosis by taking advantage of CD46 transport to the cell surface—will need to be elucidated in the future.

In the presented study a large variety of BVDV surface motions has been documented. Interestingly, apart from a decrease in the velocity of directed motion, no effect of CD46fluo on the surface transport of BVDV could be identified, indicating a minor role during this process. First insights into the progression of virus infection indicate the quick distribution of BVDV protein throughout the ER in an initial virus dose dependent manner. It will be highly interesting to identify and characterize the additional factors orchestrating BVDV entry to develop a clear picture of this process.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1999-4915/12/1/116/s1, Figure S1: Particle localization on the cell surface depending on the time after virus addition, Figure S2: Directions of movements of virus particles on the cell surface with reference to the cell body. supplementary information: python script used for track analysis, legend of the movie files, nucleotide sequence of the plasmid encoding the genome of BVDVE2-mCherry; Movie 1: Example 1 of an mCherry positive particle entering a cell, Movie 2: Example 2 of an mCherry positive particle entering a cell, Movie 3: Surfing of a BVDVE2-mCherry particle on the surface of a retraction fibre., Movie 4: Development of E2-mCherry (red) signal after infection of SK6 TO CD46fluo cells with an MOI of 1, Movie 5: Development of E2-mCherry (red) signal after infection of SK6 TO CD46fluo cells with an MOI of 10, Movie 6: Colocalization of E2-mCherry and CD46fluo inside the cell.

**Author Contributions:** C.R. designed the study. B.L., C.R., H.-W.C. and V.L. performed the experiments. C.R., T.R., U.R. and V.L. analysed the data. C.R. wrote the manuscript and all authors commented on it. All authors have read and agreed to the published version of the manuscript.

**Funding:** C.R. was supported by a Vetmeduni Start-up grant (PP23016276) and a CORBEL grant (PID 2375).

**Acknowledgments:** We would like to acknowledge the microscopy support from the Infectious Diseases Imaging Platform (IDIP) at the Center for Integrative Infectious Disease Research, Heidelberg, Germany. The authors thank Tobias Rasse and Christian Tischer for their support and the fruitful discussions. We would like to thank the University of Veterinary Medicine Vienna for Open Access Funding.

**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/).

### **A CRISPR**/**Cas9 Generated Bovine CD46-knockout Cell Line—A Tool to Elucidate the Adaptability of Bovine Viral Diarrhea Viruses (BVDV)**

#### **Kevin P. Szillat, Susanne Koethe, Kerstin Wernike, Dirk Höper and Martin Beer \***

Institute of Diagnostic Virology, Friedrich-Loeffler-Institut, Federal Research Institute for Animal Health, 17493 Greifswald-Insel Riems, Germany; Kevin.Szillat@fli.de (K.P.S.); Susanne.Koethe@fli.de (S.K.); Kerstin.Wernike@fli.de (K.W.); Dirk.Hoeper@fli.de (D.H.)

**\*** Correspondence: Martin.Beer@fli.de; Tel.: +49-38351-71200

Received: 29 June 2020; Accepted: 2 August 2020; Published: 6 August 2020

**Abstract:** Bovine viral diarrhea virus (BVDV) entry into a host cell is mediated by the interaction of the viral glycoprotein E2 with the cellular transmembrane CD46 receptor. In this study, we generated a stable Madin–Darby Bovine Kidney (MDBK) CD46-knockout cell line to study the ability of different *pestivirus A* and *B* species (BVDV-1 and -2) to escape CD46-dependent cell entry. Four different BVDV-1/2 isolates showed a clearly reduced infection rate after inoculation of the knockout cells. However, after further passaging starting from the remaining virus foci on the knockout cell line, all tested virus isolates were able to escape CD46-dependency and grew despite the lack of the entry receptor. Whole-genome sequencing of the escape-isolates suggests that the genetic basis for the observed shift in infectivity is an amino acid substitution of an uncharged (glycine/asparagine) for a charged amino acid (arginine/lysine) at position 479 in the ERNS in three of the four isolates tested. In the fourth isolate, the exchange of a cysteine at position 441 in the ERNS resulted in a loss of ERNS dimerization that is likely to influence viral cell-to-cell spread. In general, the CD46-knockout cell line is a useful tool to analyze the role of CD46 for pestivirus replication and the virus–receptor interaction.

**Keywords:** bovine viral diarrhea virus (BVDV); pestivirus; escape mutant; CD46; ERNS; adaptation; CRISPR; knockout; MDBK; cell entry

#### **1. Introduction**

The genus *Pestivirus* belongs to the family *Flaviviridae* and contains several veterinary-relevant virus species with major animal welfare and economic importance like *pestivirus A* and *B* (bovine viral diarrhea virus types 1 and 2, BVDV-1 and -2), *pestivirus D* (border disease virus, BDV) and *pestivirus C* (classical swine fever virus, CSFV) [1–4]. The host range of the classical pestivirus species includes cloven-hoofed animals. However, in recent years, a continuously growing diversity of pestiviruses has been seen worldwide with atypical pestiviruses being isolated from host species like rat, bat and whale [5–8].

Pestiviruses are positive-sense, enveloped, single-stranded RNA viruses with a genome size of 12.3–13 kb [9]. The genome encodes eight nonstructural and four structural proteins in a single open reading frame (ORF) [10], of which the glycoproteins ERNS, E1 and E2 play an important role in the initiation of BVDV uptake by the host cell [11,12]. It has been shown that ERNS interacts with the cell surface heparan sulfate and E1–E2 heterodimers bind the cellular receptor CD46 and mediate clathrin-dependent endocytosis [11,13,14]. CSFV ERNS interacts with the heparan sulfate of porcine cells and both CD46 and heparan sulfate, are major factors for the attachment of CSFV in vitro [15,16]. Interestingly, a shift to a dominant role of the ERNS-mediated binding in CSFV was connected with a drastic in vivo attenuation [15]. Furthermore, dimerization of ERNS is an important virulence factor

and abrogation leads to attenuation [17]. ERNS also plays a role in the control of the activation of beta interferon induced upon viral infection by inhibiting the double stranded RNA-induced response of cells [18]. Heparan sulfate has been further described to be important for cellular binding of different viruses, e.g., Schmallenberg virus, hepatitis E virus and rabies virus [19–22].

The binding partner of the pestivirus envelope protein E2, cellular CD46, is a type 1 transmembrane glycoprotein expressed on all nucleated cells that protects host cells from damage by the complement system by the inactivation of C3b and C4b complement products (reviewed in [23]). CD46 is known to serve as a binding partner for several human pathogens like certain adenoviruses, as well as for animal viruses like BVDV and CSFV [16,24]. In particular the two peptide domains E66QIV69 and G82QVLAL87 of the complement control protein module 1 (CCP1) are crucial for the binding of BVDV—preincubation of MDBK cells with an anti-CD46 serum leads to a strong reduction in infection efficiency [25]. In vitro studies have shown that BVDV spreads by direct cell-to-cell transmission from infected to uninfected cells, even in the presence of neutralizing antibodies against the virus and also if CD46 receptors are blocked by antibodies [26].

Interestingly, there is to our knowledge no bovine CD46-knockout cell line available, allowing e.g., the analysis of the interaction of bovine pestiviruses with receptor molecules. Due to rapid developments in the field of CRISPR/Cas9-mediated gene editing, genetically modified in vitro models like knockout cell lines can be generated in a straightforward manner in a relatively short amount of time [27]. By using ribonucleoprotein (RNP) complexes, target-specific guide RNAs (gRNAs) are complexed with the Cas9 protein and transfected into the target cell. Unlike delivery of mRNA or DNA, RNP-mediated editing does not depend on the host cell for the synthesis of Cas9 and gRNAs. Furthermore, the RNP-based approach reduces the risk of off-target effects and cell death due to the shortened half-life of the proteins inside the target cell [28,29]. We used the CRISPR/Cas9 RNP approach in this study to knockout the cellular receptor CD46 in the Madin–Darby Bovine Kidney (MDBK) cell line to establish a stable in vitro model. We used the generated cell line to passage different BVDV type 1 and 2 strains to study adaptation mechanisms of these viruses in a CD46-negative cellular environment.

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

#### *2.1. Cells and Viruses*

All cell lines used in this study were obtained from the Collection of Cell Lines in Veterinary Medicine (CCLV) at the Federal Research Institute for Animal Health, Insel Riems, Germany (FLI). Cells were cultured in minimal essential medium (MEM), supplemented with 10% fetal calf serum (FCS), at 37 ◦C and 5% CO2. A Madin–Darby Bovine Kidney cell line (MDBK, RIE0261) was used to generate the CD46-knockout cell line, further referred to as CD46-MDBK. Bovine esophagus cells (KOP-R, RIE0244) were used to propagate and titrate virus stocks of the BVDV-1 strains Paplitz (1b, cytopathic), NADL (1a), D02/11-2 (1d) and BVDV-2 strain CS8644 (2a). All pestiviruses were obtained from the German National Reference Laboratory for BVD/MD (FLI).

#### *2.2. Generation of a CD46-knockout Cell Line*

#### 2.2.1. Transfection of MDBK Cells and Clone Selection

The CRISPR/Cas9 RNP-mediated editing approach was used in this study to knockout the BVDV binding domains E66QIV69 and G82QVLAL87 of the cellular receptor CD46 in MDBK cells. CRISPR RNAs (crRNAs) were designed to introduce double-strand breaks (DBS) inside the CCP1 domain, spanning these binding domains (overview in Figure S1). Suitability of the crRNAs was confirmed by CRISPOR [30] and CHOPCHOP [31]. The crRNAs (crRNA-1: GGCTTCATAGAGACAAATCT and crRNA-3: TCATACACAATCTGCTCCCC) and all transfection reagents were purchased from Thermo Fisher Scientific (Waltham, MA, USA). Annealing of the crRNAs with the trans-activating

crRNA (tracrRNA) was done following the manufacturer's protocol. MDBK cells were seeded one day prior to transfection and subsequently transfected according to the manufacturer's instructions, using Lipofectamine CRISPRMAX Cas9 transfection reagents (Thermo Fisher Scientific) and TrueCut Cas9 Protein v2 (Thermo Fisher Scientific). The nontarget control cell line (NTC) was generated under identical conditions, using nontarget control gRNAs (Thermo Fisher Scientific). Single cell dilution and subsequent polymerase chain reaction (PCR) of the monoclonal cell colonies were used to screen for a cell clone with the intended deletion.

#### 2.2.2. Isolation of DNA, PCR and Sequencing

DNA was isolated using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The PCR to confirm the knockout was performed using the QuantiTect Multiplex-PCR Kit (Qiagen) in combination with CD46-specific primers (forward primer CD46\_CCP1\_F: 5 -GAT GCT GTC TCT TCC ATT TAC T-3 ; reverse primer CD46\_CCP1\_R: 5 -GCC TGA ATG CAT GGC TAT CT-3 ) and the following conditions: 15 min, 95 ◦C; 45× (60 s, 95 ◦C; 30 s 58 ◦C and 30 s 72 ◦C); 5 min, 72 ◦C; 12 ◦C storage. The PCR amplicon of the wild-type MDBK cell line is 545 nucleotides long and covers the entire CCP1 domain.

PCR products were analyzed by gel electrophoresis and bands were extracted from the gel using the QIAquick Gel Extraction Kit (Qiagen) according to the manufacturer's instructions. Single cell colonies with the intended knockout mutation display a truncated amplicon with a size of around 400 nucleotides. The PCR gel extract was further sequenced using the IonTorrent platform, as described later.

#### *2.3. Immunofluorescence (IF) Staining, Plaque Assay and Fluorescence-Activated Cell Sorting (FACS)*

Expression of the CD46 receptor by NTC and CD46- MDBK was visualized by IF staining, using antibodies BVD/CA 26/2/5 and BVD/CA 17/2/1(1:16 dilution in tris-buffered saline with tween (TBST)). The antibodies are directed against CD46 and were kindly provided by Prof. Becher, Institute of Virology, University of Veterinary Medicine, Hannover [32]. Antimouse Alexa Fluor™488 F(ab')2 (1:1000 in TBST, Life Technologies, Carlsbad, CA, USA) was used as conjugate. FACS and IF was used to study the impact of the CD46-knockout on pestivirus entry into the host cell. In brief, CD46- MDBK and NTC cells were seeded one day prior to infection. On the day of infection, the cells were incubated with the different pestivirus isolates for 1 h at 37 ◦C and 5% CO2, multiplicity of infection (MOI) of 1. Confluent cells were used at this point to study the initial infectivity of the virus isolates and at later time points the cell-to-cell spread. Uninfected cells were used as a negative control and incubated with maintenance medium only (supplemented with penicillin and streptomycin). After the incubation period, all cells were washed and cultured for 24 h at 37 ◦C and 5% CO2 in maintenance medium.

For the IF-staining of pestiviruses, cell supernatant was discarded, and cells were fixed at 80 ◦C for 2 h. Subsequently, fixed cells were incubated with a 1:500 dilution of antibody WB103/105 (APHA Scientific, Addlestone, UK) in TBST for 1 h at room temperature (RT). Cells were washed thrice with TBST and incubated with a 1:1000 dilution of antimouse Alexa Fluor™488 F(ab')2 (Life Technologies) in TBST for 1 h at RT. Cells were washed thrice and finally covered with 1,4-Diazabicyclo [2.2.2]octane (DABCO) fluorescence preservation buffer (Sigma-Aldrich, St. Louis, MO, USA) containing propidium iodide (Sigma-Aldrich).

For the IF plaque assay, CD46- MDBK cells were seeded one day prior to infection into a 24-well plate. Subsequently, cells were infected with the stock virus (passage 0) and passage 15 of virus isolate Paplitz (BVDV-1b) at MOI = 0.1 and incubated for 1 h at 37 ◦C, 5% CO2. Thereafter, the cells were washed once with maintenance medium and overlayed with a 1:1 mix of 4% agarose (Lonza, Basel, Switzerland) in aqua dest. and MEM (2× concentrated) supplemented with penicillin/streptomycin and gentamicin. The medium layer was removed after 48 h, cells were washed once with TBST and fixed with 4% paraformaldehyde (Sigma-Aldrich) for 20 min. Cells were washed once and permeabilized with 0.5% Triton X-100 in phosphate-buffered saline (PBS) for 5 min. Cells were washed again and

treated as previously described for the IF staining of pestiviruses. The size of 50 virus plaques was measured manually by using the Nikon Eclipse Ti-U inverted microscope (Nikon GmbH, Düsseldorf, Germany) and NIS-Elements software (v. 4.50). The assay was performed in triplicate. The difference between the groups was statistically evaluated with a Mann-Whitney rank sum test as implemented in SigmaPlot (Version 11.0; Systat Software Inc., San Jose, CA, USA).

For FACS analysis, cell supernatant and trypsinized cells were collected in a FACS tube (Sarstedt, Nümbrecht, Germany) and centrifuged for 5 min at 800× *g*. Supernatant was discarded and cells fixed with 4% paraformaldehyde (Sigma-Aldrich) for 15 min at 4 ◦C. Cells were further permeabilized with 0.01% Digitonin (Sigma-Aldrich) for 10 min at 4 ◦C and washed with PBS. Cells were incubated with antibody WB103/105 (1:500 dilution in PBS) for 15 min at 4 ◦C. The cells were washed with PBS prior application of the anti-mouse Alexa Fluor™488 F(ab')2 (1:1000 dilution in PBS) for 15 min at 4 ◦C. Cells were washed and resuspended in PBS. A total of 10,000 events were acquired from each sample, using the FACSCalibur (BD Biosciences, Franklin Lakes, NJ, USA). Uninfected cells were stained and used as a negative control. They were included in each run separately to allow subtraction of the background signal (<2.0%). Experiments were performed in three independent replicates.

#### *2.4. Pestivirus Passaging*

The different pestivirus isolates were passaged on CD46- MDBK and NTC cells to study the ability of the viruses to escape CD46-dependent cell entry and as a control, respectively. Cells were seeded one day prior to infection into a 24-well plate and infected with passage 0 (MOI = 1) for 1 h at 37 ◦C and 5% CO2. Passage 0 represents the virus stock grown from the initial virus isolates on KOP-R cells. Cells were washed once and incubated in 1 mL maintenance medium for 72 h before they were frozen at −20 ◦C. For the next virus passage, crude cell extract (noncleared mix of cells and supernatant) were thawed and 100 μL per well were used for infection of the next passage of CD46- MDBK and NTC cells. Virus titers of passage 15 were determined by endpoint titration on KOP-R cells and further tested in FACS and IF regarding their infectivity on CD46- MDBK.

#### *2.5. Sequencing, Sequence Assembly and Sequence Comparison*

Virus passages 0 and 15 of the virus isolates passaged on CD46- MDBK were processed using a modification of the protocol published by Wylezich and colleagues [33] and sequenced using the Ion Torrent S5XL platform. Briefly, RNA was extracted from the freeze-thawed crude cell extract using the RNAdvance kit (Beckman Coulter, Fullerton, CA, USA) following the manufacturer's instructions and further concentrated using the Agencourt RNAClean XP magnetic beads (Beckman Coulter). Afterwards, cDNA was synthesized using a combination of the SuperScript™ IV First-Strand cDNA Synthesis System (Thermo Fisher Scientific) and the NEBNext®Ultra™ II Non-Directional RNA Second Strand Synthesis Module (New England Biolabs, Ipswich, MA, USA). The generated cDNA was fragmented to 500 bp using the Covaris M220 Focused-Ultrasonicator (Covaris, Brighton, UK) before Ion Torrent-compatible libraries were generated using the GeneRead L Core Kit (Qiagen) and IonXpress Barcode Adapter (Thermo Fisher Scientific). After size selection, library quality was checked using the Agilent 2100 Bioanalyzer system (Agilent Technologies, Santa Clara, CA, USA). The library concentration was measured with the KAPA Library Quantification kit (Roche, Mannheim, Germany). Using an Ion Torrent S5 XL, libraries were sequenced on an Ion 530 chip in 400-bp mode according to the manufacturer's instructions (Thermo Fisher Scientific).

For sequence assembly, a random subset of 50,000 to 500,000 reads was assembled (Newbler v.3.0; 454/Roche). The complete data set was mapped against the assembled genome (Newbler v.3.0; 454/Roche) to confirm the sequence. Consensus sequences were annotated based on the NADL (BVDV-1a) reference strain (NC\_001461) and amino acid (aa) sequences were annotated accordingly (ERNS: aa 271-497, E1: aa 498-692 and E2: aa 693-1066). Sequences were aligned using the MAFFT alignment of the bioinformatic software Geneious Prime (v.2020.1.2; Biomatters Ltd., Auckland, New Zealand). The variant analysis for the detection of viral quasispecies was performed in Geneious Prime

(v.2020.1.2; Biomatters Ltd., Auckland, New Zealand), using the 454Contigs.bam from the mapping of passage 0 and passage 15 reads against the consensus sequence of passage 0 (Newbler v.3.0./Roche). Single-nucleotide-polymorphisms were detected with a minimum variant frequency of 0.1.

#### *2.6. Immunoblotting*

Cell lysates of passage 15 (on CD46- MDBK cells) and passage 0 (on NTC cells) were prepared 72 h after infection of the cells. For this purpose, cells were harvested and incubated for 30 min on ice in lysis buffer (Na2HPO4, 1% Triton), supplemented with protease inhibitor (Roche). Proteins were separated under nonreducing conditions on a 10% SDS-polyacrylamide gel for 75 min at 130 mA and subsequently blotted onto a GE Healthcare Amersham™ Protan™ nitrocellulose membrane (Thermo Fisher Scientific). The membrane was blocked with Roti®Block (Carl Roth, Karlsruhe, Germany) for 1 h. For the detection of pestivirus ERNS, the membrane was incubated with monoclonal antibodies HC/TC169/2/3 and BVD/C46/2/1 for 1 h (1:100 in Roti®Block, kindly provided by Prof. Becher, Institute of Virology, University of Veterinary Medicine Hannover). The POD-antimouse antibody (Dianova, Hamburg, Germany) was used as a secondary antibody, diluted 1:20,000 in phosphate buffered saline with tween (PBST). Proteins were detected by chemiluminescence, using SuperSignal™ West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific).

#### **3. Results**

#### *3.1. Characterization of CD46- MDBK*

After single cell dilution of the transfected cells, colonies were screened via PCR and the knockout cell was identified to carry a homozygous deletion in the target region. Deep sequencing of the PCR product confirmed the successful biallelic knockout of 120 and 134 nucleotides in the target region, spanning the BVDV binding domains E66QIV69 and G82QVLAL87 (Figure 1b). CD46- MDBK were propagated and the deletion was confirmed by PCR up to the last passage used for passaging of the virus isolates (CD46- MDBK passage 54) (Figure 1a). IF staining with the anti-CD46 antibodies BVD/CA 26/2/5 and BVD/CA 17/2/1 (data not shown) showed no fluorescence signal on CD46- MDBK, compared to NTC cells (Figure 1c).

**Figure 1.** Characterization of CD46- MDBK cell line. (**a**) PCR amplification of knockout target region of MDBK wild-type (wt), nontarget control (NTC), CD46- MDBK passage 1 (P. 1) and passage 54 (P. 54) (**b**) Sequence comparison of wt and CD46- MDBK reveals a biallelic deletion of 120 and 134 nucleotides in the CCP1 region (black box), including the bovine viral diarrhea virus (BVDV) binding sides E66QIV69 and G82QVLAL87 (orange box). crRNAs are depicted in green (**c**) Staining of CD46 receptor using the anti-CD46 BVD/CA 26/2/5 antibody and conjugate only control on NTC (top) and CD46- MDBK (bottom). Magnification: 20×.

The functional knockout of the receptor was further confirmed via FACS and IF. The different BVDV isolates grew successfully on the NTC cells, whereas growth was strongly inhibited on the CD46- MDBK cells (Figure 2b,c). Based on the FACS data, the isolates used in this study (passage 0) infected 70–95% of the NTC cells, compared to less than 5% of the CD46- MDBK (Figure 2a).

**Figure 2.** Detection of BVDV in the initial and the 15th passage on the indicated cell lines after incubation for 24 h. (**a**) Diagram depicting the percentages of positive cells in the fluorescence-activated cell sorting (FACS) analysis, conducted in triplicates. Immunofluorescence (IF) staining of (**b**) isolate passage 0 on non-target control (NTC), (**c**) passage 0 on CD46- MDBK (**d**) passage 15 on NTC and (**e**) passage 15 on CD46- MDBK. Scale bar 100 μm. Cells were incubated with the different pestivirus isolates (MOI = 1) for 1 h at 37 ◦C and subsequently washed and incubated for 24 h before analysis. Negative control: cells were incubated with maintenance medium only.

#### *3.2. Adaption of Pestivirus Isolates by Passaging on CD46- MDBK Cells*

The generated CD46- MDBK cell line was subsequently used to adapt different BVDV isolates to a host cell lacking the CD46 receptor. After 15 passages on the knockout cell line, the isolates NADL, D02/11-2 (BVDV-1d)and CS8644 (BVDV-2a) showed a strong increase in infectivity, from less than 5% positive cells (passage 0) to 40–60% (passage 15) on CD46- MDBKas measured by FACS analysis (Figure 2a). For all viruses, infectivity analyzed by IF staining was in line with the FACS data (Figure 2).

The cytopathic virus isolate Paplitz (cytopathic effect becomes apparent 3 to 5 days post infection) did not show a significant increase in CD46- MDBK cell infectivity after passaging based on the FACS results (from 1% positive cells with passage 0 virus to 4% with passage 15 virus, Figure 2a). Similarly, the IF staining did not show a strong increase in infectivity after incubation for 24 h on CD46- MDBK, when compared to the other isolates (Figure 2e). However, when the CD46- MDBK cells were incubated for 72 h, a complete infection of the cell layer was visible for all isolates (passage 15) (Figure 3d), that differed markedly from incubation for 72 h with the passage 0 virus (Figure 3b).

**Figure 3.** Comparison of the initial passage and passage 15 on non-target control (NTC) and CD46- MDBK after incubation for 72 h. Passage 0 on (**a**) NTC and (**b**) CD46- MDBK, passage 15 on (**c**) NTC and (**d**) CD46- MDBK. Cells were incubated with the designated pestivirus isolates (MOI = 1) for 1 h at 37 ◦C and subsequently washed and incubated for 72 h before analysis. Negative control: cells were incubated with maintenance medium only.

Therefore, passage 0 and 15 of virus isolate Paplitz (BVDV-1b) were further tested in a plaque assay, to investigate the underlying mechanism that led to a densely infected cell layer after 72 h. Virus plaques of passage 0 had an average area of 28,462 μm2 (standard deviation (SD): 11,230 μm2, median: 25,489 μm2). Virus plaques of passage 15 were on average twice as big, with an average area of 55,985 μm2 (SD: 21,349 μm2, median: 53,725 μm2) (Figure 4).

The difference between the plaque sizes of both groups was statistically significant (*p* < 0.001). In comparison, the plaque size of the other passage 0 virus isolates measures for NADL (BVDV-1a): 25,243 μm<sup>2</sup> (SD: 11,850 μm2, median: 23,132 μm2), D02/11-2 (BVDV-1d): 24,655 μm2 (SD: 9181 μm2, median: 24,011 μm2) and for CS8644 (BVDV-2a): 52,072 μm<sup>2</sup> (SD: 21,839 μm2, median: 50,417 μm2). Based on these measurments, the Paplitz (BVDV-1b) passage 0 isolate is comparable to the plaque size of the other tested BVDV-1 strains.

**Figure 4.** Comparison of plaque sizes of the initial passage and passage 15 of virus isolate Paplitz (BVDV-1b) on CD46- MDBK. Cells were incubated with a MOI = 0.1 for 1 h at 37 ◦C and subsequently washed and incubated for 48 h. The experiment was conducted in triplicate and 50 plaques were measured for each virus passage in each experiment (total *n* = 150 plaques/isolate). Whiskers indicate the 10% and 90% percentiles, respectively, and each outlier is depicted by a dot. The difference between the plaque sizes is statistically significant (\*\*\* *p* < 0.001).

#### *3.3. Sequence Analysis after Virus Passaging on CD46- MDBK*

In order to identify genetic adaptations that led to the observed shift in infectivity and virus-growth characteristics between the different virus passages, they were sequenced using next-generation sequencing. The genomic regions coding for the ERNS, E1 and E2 proteins of the initial virus passage were compared to the 15th passage, since these regions are known to be involved in pestivirus entry into the host cell [11,14]. The graphical overview of the aligned sequences is depicted in Figure S2. After 15 passages, all virus isolates gained at least one aa substitution in the ERNS-protein (Table 1).

**Table 1.** Overview of the amino acid substitution in ERNS, E1 and E2 proteins based on the consensus sequences of the original BVDV isolates and their 15th passage on the CD46-deficient cell line. Amino acid positions with known effects are highlighted in bold.


Virus isolates NADL (BVDV-1a), D02/11-2 (BVDV-1d) and CS8644 (BVDV-2a) displayed an aa exchange at the same position, namely 479 in the ERNS. NADL (BVDV-1a) and CS8644 (BVDV-2a) exchanged the uncharged glycine with a charged arginine and D02/11-2 (BVDV-1d) exchanged the uncharged asparagine with a charged lysine. The cytopathic strain Paplitz (BVDV-1b) exchanged an uncharged cysteine for a charged arginine at position 441 in the ERNS (Table 1). From the 15 mutations listed, two were already detected in the viral sequence reads from passage 0 as a minor population. Amino acid mutation in NADL (BVDV-1a) at position 677 and in D02/11-2 (BVDV-1d) at position 479 were identified with a sequence prevalence of 36% and 42%, respectively (Table 1). An important finding is that the minor population at aa position 479 of D02/11-2 (BVDV-1d) passage 0 is linked with another minor population in passage 0 that substitutes the original lysine at position 480 with an asparagine with a frequency of 39.9% (Table S1). During analysis of the actual reads of passage 0, it becomes apparent that reads are coding either for a lysine-asparagine or an asparagine-lysine (aa position 479 and 480) but not for a lysine-lysine or asparagine-asparagine (visualized in Figure S3).

#### *3.4. ERNS Dimerization of Virus Isolate Paplitz (BVDV-1b)*

The formation of ERNS dimers of virus isolate Paplitz (BVDV-1b) was analyzed, since the exchanged aa cysteine at position 441 was previously described to be important for dimerization of ERNS [17]. Immunoblot analysis confirmed formation of ERNS dimers (88-96 kDa) in virus passage 0, compared to monomer formation (48 kDa) of virus passage 15 (Figure 5).

**Figure 5.** ERNS formation of virus isolate Paplitz (BVDV-1b). Detection of ERNS in the lysates of infected cells 72 h post infection using immunoblotting. Proteins were separated under nonreducing conditions. ERNS of passage 0 (P.0) forms dimers (88–96 kDa), whereas passage 15 (P.15) forms monomers (44–48 kDa).

#### **4. Discussion**

The importance of bovine CD46 for pestivirus entry into a host cell has been already described in detail in the past and the two peptide domains E66QIV69 and G82QVLAL87 have been identified to be crucial [25,34,35]. However, for further analysis, bovine CD46-knock-out cells were needed. We therefore successfully knocked out the aforementioned domains and showed that this leads to a significant decrease in infectivity of the tested BVDV-1 and -2 isolates.

The strong reduction in infectivity on CD46- MDBK is consistent with previous findings of Krey et al. [25], who showed reduced binding of pestivirus isolates in vitro by at least 70% (in 29 out of 30 isolates tested) after preincubation of MDBK cells with an anti-CD46 serum. Interestingly, we could show that the knockout of CD46 did not result in a complete loss of infectivity. A small percentage of less than 5% of the cells were still able to get infected by the tested BVDV isolates (passage 0). Therefore, a small fraction of the viruses most likely uses a CD46-independent route to enter the respective host cells.

After passaging the four different virus isolates 15 times on CD46- MDBK, infectivity of the virus isolates NADL (BVDV-1a), D02/11-2 (BVDV-1d) and CS8644 (BVDV-2a) increased significantly when compared to passage 0 after incubation for 24 h. Sequencing of these passages showed that the isolates share the same aa exchange at position 479 after passaging. This aa exchange is of importance, since the same position (aa 476 in *pestivirus C* (CSFV) correlates to aa 479 in *pestivirus A*/*B*) has been shown to be crucial for CSFV interaction with membrane-associated heparan sulfate [36]. The exchange of an uncharged aa (glycine) to a positively charged aa (arginine) at this position has been described to increase virus replication in vitro of CSFV variants carrying this mutation [15,36]. It has been also suggested that this aa substitution (position 476 in CSFV; position 479 in BVDV-1 and -2) increases the positive charge of the ERNS region and that this particular aa is exposed to the surface and involved into direct binding to the negatively charged heparan sulfate [36]. Reimann and colleagues [37] also associated the increased virus infectivity in vitro with the same aa substitution of glycine to arginine at position 479 in CP7\_E2alf, a chimeric pestivirus constructed from a BVDV-1 backbone (strain CP7) and E2 from CSFV (strain Alfort). Overall, these findings indicate very clearly that the loss of the host cells CD46 favours the introduction of a positively charged aa at position 479 of the ERNS in the pestivirus isolates studied. Even though the aa position 479 is in the amphipathic C-terminal end of the ERNS that is important for membrane binding, it is likely that this aa substitution has no or only minor effects on lipid binding. Research has shown that the amphipathic helix has a robust lipid affinity that cannot be disturbed by mutating single amino acids [38].

The aa substitution at position 479 occurred most likely de novo in virus isolate NADL (BVDV-1a) and CS8644 (BVDV-2a), however, very low frequencies might be missed due to the sequencing depths. In case of virus isolate D02/11-2 (BVDV-1d), the minor variant of the aa substitution (asparagine for lysine) at position 479 can be detected already in passage 0 with a proportion of 42%. Therefore, we would expect the phenotype of this passage 0 virus isolate to show a higher infectivity than actually detected, since we postulate that this mutation is important for the change to a phenotype with higher infectivity. However, when we looked into the other minor variants of passage 0 virus isolate D02/11-2 (BVDV-1d), we found that a lysine at aa position 480 is substituted with an asparagine with a similar proportion of 39.9%. Furthermore, when looking into the actual reads of passage 0, we could see that the reads are coding either for a lysine-asparagine (in consensus) or an asparagine-lysine (in minor variant) at position 479–480 but not for a lysine-lysine or an asparagine-asparagine. We therefore assume that in case of D02/11-2 (BVDV-1d), both lysine at position 479 and 480 are necessary to increase the infectivity of the virus. This has probably been selected for since neither the lysine-asparagine nor the asparagine-lysine variant could be detected in passage 15.

We suggest that the aa exchange at position 479 may allow the virus isolates NADL (BVDV-1a), D02/11-2 (BVDV-1d) and CS8644 (BVDV-2a) to compensate for the loss of the potential binding side CD46 by an increased binding of heparan sulfate. This finding provides therefore additional evidence for previous assumptions concerning the role of the heparan sulfate binding and CD46 independent entry of pestiviruses [25]. Furthermore, it would be interesting to passage these mutated virus isolates (passage 15) in vivo, to study if the mutation at position 479 is selected against in vivo like it was shown for CSFV [15]. We would speculate that this is the case, considering the similarities of BVDV and CSFV cell entry in vitro regarding their dependency on CD46 and heparan sulfate [11,16,34].

Nevertheless, the cytopathic strain Paplitz (BVDV-1b) is of special interest, since the virus shows an altered and unexpected growth behaviour after passaging. It seems that the selection pressure from CD46-MDBK cell passaging does not result in an improved initial cell entry of the virus but rather changed the growth behaviour of the virus towards a markedly improved cell-to-cell spread after initial cell entry. The lower infectivity of Paplitz (passage 15) compared to Paplitz (passage 0) on the NTC cells might be explained by the special way of adaption of this strain to the CD46 deficient environment and the subsequent loss of ERNS dimers. It can be hypothesized that this trade-off favours cell spread in CD46-deficient cells for the cost of reduced binding to CD46 and growth in NTC cells. Interestingly, one mutation identified in the ERNS of Paplitz (BVDV-1b) results in the exchange of a cysteine at position 441, which is important for the formation of a intermolecular disulphide bond in the ERNS of CSFV [17]. Mutation of this cysteine, that occurred most likely de novo, leads to a loss of dimerization of ERNS

that results in case of CSFV in an attenuated virus that is still able to grow in cell culture [17] but also with larger plaque sizes due the reduced binding to heparan sulfate [39]. We could also demonstrate here the loss of ERNS dimerization in the passage 15 variant. Paplitz (BVDV-1b) passages 0 and 15 are not able to enter the host cell faster after passaging, but once inside the cell, the virus grows differently than the initial passage 0 and shows an enhanced cell-to-cell spread. This change is likely influenced by substitution of the cystein at position 441, that leads to monomeric ERNS. Further research should study if the adaption described for virus isolate Paplitz (BVDV-1b) is typical for cytopathic pestiviruses and if the other mutations identified in E1 and E2 might also play a role. It could be hypothesized that dimerizaton of ERNS could have a function concering cell-to-cell spread of pestiviruses like BVDV-1.

In summary, we describe here the succesful generation of a stable CD46-knockout cell line and we demonstrated that the infectivity of different BVDV isolates strongly depends on CD46. At the same time, we showed that a fraction of the virus particles is still able to enter the host cell even in the absence of the CD46 receptor. The study also shows that forced adaption of the pestiviruses studied leads to compensatory mutations in the ERNS, affecting the virus-host interplay. Forced adaption of virus isolates NADL (BVDV-1a), and CS8644 (BVDV-2a) and D02/11-2 (BVDV-1d) led to a mutation at aa position 479 probably increasing heparan sulfate binding. In contrast, isolate Paplitz (BVDV-1b) substituted a cystein at position 441 that led to a loss of ERNS-dimer formation, that is further suggested to increase cell-to-cell spread of the virus. The newly generated cell line is now available for future research to elucidate the role of the CD46 receptor in the context of the pestivirus biology and growth cycle.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1999-4915/12/8/859/s1. Supplementary Figure S1. Overview of the PCR amplicon, including the complement control protein 1 domain (CCP1, depicted in grey) with BVDV binding sides E66QIV69 and G82QVLA87 (both in orange), CRISPR RNAs (crRNA-1 and -3, depicted in yellow) and PCR primer (CD46\_CCP1\_F and \_R, depicted in green). Supplementary Figure S2. Amino acid sequence alignment of passage 0 and 15 of the BVDV isolates. Differences between passages are depicted as a black gap in the alignment. The red box indicates the shared amino acid exchange at position 479 in virus passage 15 of NADL (BVDV-1a), D02/11-2 (BVDV-1b) and CS8644 (BVDV-2a). Supplementary Figure S3. Comparison of passage 0 and passage 15 sequence reads of D02/11-2 (BVDV-1b) at aa position 479 and 480 compared to consensus sequence passage 0. Two minor variants resulting in either a AAA-AAC or AAC-AAA codon (aa lysine-asparagine or asparagine-lysine). In comparison, passage 15 reads assemble only AAA-AAA (aa lysine-lysine) at position 479 and 480. The initial consensus sequence of passage 0 assembles AAC-AAA (aa lysine-asparagine). Supplementary Table S1. Overview of the variant analysis for passage 0 and 15 of virus isolate D02/11-2 (BVDV-1b). The variant of aa 479 and 480 is represented by nucleotide position 1437 and 1440 respectively and highlighted in bold.

**Author Contributions:** Conceptualization, M.B.; methodology, S.K., K.W., D.H. and M.B.; formal analysis, K.P.S., S.K., K.W., D.H. and M.B.; investigation, K.P.S., S.K., D.H. and K.W.; resources, D.H. and K.W.; writing—original draft preparation, K.P.S.; writing—review and editing, S.K., K.W., D.H. and M.B.; visualization, K.P.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the European Union's Horizon 2020 research and innovation program HONOURs, under grant agreement No 721367.

**Acknowledgments:** The authors are grateful to Andrea Aebischer, Patrick Zitzow, Doreen Schulz and Bianka Hillmann for excellent technical assistance and support.

**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/).
