De Novo Cholesterol Biosynthesis and Its Trafficking in LAMP-1-Positive Vesicles Are Involved in Replication and Spread of Marek’s Disease Virus

MDV disrupts lipid metabolism and causes atherosclerosis in MDV-infected chickens; however, the role of cholesterol metabolism in the replication and spread of MDV is unknown. MDV-infected cells do not produce infectious cell-free virus in vitro, raising the question about the mechanism involved in the cell-to-cell spread of MDV. In this report, we provide evidence that MDV replication depends on de novo cholesterol biosynthesis and uptake. Interruption of cholesterol trafficking within multivesicular bodies (MVBs) by chemical inhibitors or gene silencing reduced MDV titers and cell-to-cell spread. Finally, we demonstrated that MDV gB colocalizes with cholesterol and LAMP-1, suggesting that viral protein trafficking is mediated by LAMP-1-positive vesicles in association with cholesterol. These results provide new insights into the cholesterol dependence of MDV replication.

transformed into lymphoma cells (1)(2)(3)(4). We have recently shown that MDV infection modulates cell metabolism and specifically enhances fatty acid synthesis and the formation of lipid droplets (5,6). It has been suggested that the alteration of lipid metabolism by MDV causes atherosclerosis in infected chickens (7,8). The influence of cholesterol synthesis on the pathogenesis of Marek's disease was confirmed in series of experiments showing that the use of a 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMG-CoA) inhibitor reduces atherosclerotic plaque areas and mortality rates in MDV-infected chickens (9). This study examines the mechanism involved in these in vivo observations. Cholesterol synthesis occurs in the cytosol and endoplasmic reticulum (ER), and the synthesized cholesterol is then rapidly transported via multivesicular bodies (MVBs) to other organelles, including the plasma membrane. As part of de novo cholesterol biosynthesis, HMG-CoA catalyzes the conversion of HMG-CoA to mevalonic acid. Thereafter, squalene epoxidase (SqE), in a rate-limiting step for sterol biosynthesis, catalyzes the first oxygenation step in the biosynthesis of lanosterol (10). The final stage involves the conversion of lanosterol to cholesterol, which is distributed via lysosome-associated membrane protein 1 (LAMP-1)-positive vesicles between intracellular compartments, including cell membranes (11)(12)(13). The majority of LAMP-1 is directly transported to lysosomes via the trans-Golgi network (TGN), while some LAMP-1 is initially transported out of the cell membrane, and it is then internalized and delivered to lysosomes (14). It has been suggested that herpesvirus particles are transported to the extracellular space by the fusion of the plasma membrane with vesicle-containing virus particles (15). MVBs are a suitable microenvironment for herpes simplex virus 1 (HSV-1) replication, and glycoprotein B (gB) of HSV-1 is colocalized with the MVB marker LAMP-1 (16). Interestingly, the final envelopment of varicella-zoster virus, a highly cell-associated alphaherpesvirus similar to MDV, in the cytoplasm occurs at TGN-derived vesicles (17,18).
Very little information is available relating to the mechanism involved in the transportation of MDV particles and the role of cholesterol trafficking in MDV-infected cells. This study examines the influence of MDV infection on cholesterol metabolism in chicken embryonated fibroblasts (CEFs) and demonstrates that MDV replication and spread are dependent on cholesterol synthesis and uptake. Intriguingly, a key relationship between de novo cholesterol biosynthesis and MDV cell-to-cell spread was established. Interruption of trafficking from MVBs by chemical inhibitors or gene silencing reduced MDV titers and cell-to-cell spread. Finally, we demonstrated that MDV gB was colocalized with cholesterol and LAMP-1, suggesting that viral protein trafficking is mediated by LAMP-1-positive vesicles in association with cholesterol.

Induction of de novo cholesterol biosynthesis in MDV-infected CEFs.
The relative levels of three metabolites (squalene, desmosterol, and cholesterol) involved in cholesterol biosynthesis were determined in cell lysates with a total of 6 technical replicates per biological replicate obtained from mock-and MDV-infected primary CEFs at 48 and 72 h postinfection (hpi) using comprehensive two-dimensional gas chromatography-mass spectrometry (GC/GC-MS). The results demonstrated that the levels of desmosterol were lower in MDV-infected cells than in mock-infected cells at both 48 and 72 hpi. In contrast, the cholesterol level was significantly higher in MDV-infected cells than in mock-infected cells at both 48 and 72 hpi (Fig. 1A), suggesting that higher cellular cholesterol contents may have reduced cholesterol synthesis at 48 and 72 hpi. The metabolites and the key enzymes involved in the cholesterol biosynthesis pathway are shown in Fig. 1B. HMG-CoA, farnesyltransferase (FTase), SqE, and 24-dehydrocholesterol reductase (DHCR24) are involved in cholesterol biosynthesis, while CYP27A1 is a gene encoding a cytochrome P450 oxidase that is involved in the degradation of cholesterol. On the other hand, the ATP-binding cassette transporter 1 (ABCA1) gene, expressing the cholesterol efflux regulatory protein, is involved in cellular cholesterol homeostasis (19). Gene expression analyses of the key enzymes involved in cholesterol biosynthesis by real-time PCR (RT-PCR) demonstrate that HMG-CoA, FTase, SqE, DHCR24, CYP27A1, and ABCA1 were upregulated in MDV-infected CEFs at 72 hpi. The upregulation of HMG-CoA and SqE proteins in MDV-infected CEFs was confirmed using a Western blot assay ( Fig. 1D and E). The results indicate that cholesterol biosynthesis is upregulated in MDV-infected cells. presented as relative fluorescence intensities over ␣-tubulin (E). Nonparametric Wilcoxon tests (Mann-Whitney) were used to assess normal distribution and test significance, with the results shown as means Ϯ SD. *** (P ϭ 0.0005) and **** (P ϭ 0.0001) indicate a statistically significant difference compared to the control. The experiment was performed in biological triplicates with six technical replicates per biological replicate.
Going forward, we present a comprehensive analysis of small pharmacological inhibitors and short hairpin RNAs (shRNAs) targeting key enzymes within the cholesterol biosynthesis pathway to elucidate the relative contribution of de novo cholesterol to MDV replication and spread. All inhibitors were titrated to determine nontoxic concentrations of lovastatin ( Fig 2E) based on the confluence of the cells using light microscopy and the presence of live-cell (7-aminoactinomycin D negative [7-AAD Ϫ ]) staining using flow cytometry. In addition, shRNAs targeting the knockdown (green fluorescent protein-positive [GFP ϩ ] cells) of the respective enzymes HMG-CoA, SqE, and LAMP-1 were used to validate the efficiency of the small pharmacological inhibitors utilized (Fig. 2F).
MDV infectivity is dependent on de novo mevalonic acid biosynthesis. Mevalonic acid is a key metabolite in the synthesis of cholesterol, and HMG-CoA, a ratecontrolling enzyme of the mevalonate pathway, is an essential enzyme within the first stage of cholesterol biosynthesis (19). Here, we examined the role of HMG-CoA in the replication of MDV using a small pharmacological inhibitor, lovastatin, and gene silencing by shRNA (Fig. 3A). To this end, we initially determined nontoxic concentrations of lovastatin and mevalonic acid based on the confluence of the cells using light microscopy and 7-AAD staining using flow cytometry. MDV titers were determined in MDV-infected CEFs in the presence of different concentrations of lovastatin. The results demonstrated that nontoxic concentrations of lovastatin (2.5, 10, 25, and 100 ng/ml) reduced MDV titers in a dose-dependent manner (Fig. 3B). Treatments of the cells with 2.5, 10, 25, and 100 ng/ml lovastatin reduced MDV titers in a dose-dependent manner. Exogenous mevalonic acid (2.5 to 7 g/ml) rescued the inhibitory effects of lovastatin (100 ng/ml) on MDV titers (Fig. 3C), while it did not have any effect on MDV titers in the absence of lovastatin (Fig. 3D). RNA silencing reduced HMG-CoA protein levels to  nondetectable levels (Fig. 3E) and reduced MDV titers (Fig. 3F), MDV plaque sizes ( Fig.  3G and H), and MDV copy numbers (Fig. 3I).
De novo lanosterol synthesis supports MDV replication in CEFs. The second stage in cholesterol biosynthesis is the lanosterol pathway (Fig. 1B), which preferentially utilizes mevalonic acid to generate lanosterol. In this stage, squalene is converted to lanosterol by SqE (Fig. 4A), while FTase, a cellular enzyme essential for the prenylation of cellular factors, is involved in a posttranslational protein modification process. To examine the role of SqE and FTase in MDV replication, we utilized both small pharmacological inhibitors and gene silencing using the shRNA system. Nontoxic concentrations of lonafarnib, an inhibitor of FTase, did not alter MDV titers (Fig. 4B) or plaque sizes ( Fig. 4C and D). In contrast, nontoxic concentrations of BIBB 515, an inhibitor of SqE, significantly reduced MDV titers (Fig. 4E). The silencing of SqE by shRNA reduced protein levels (Fig. 4F), MDV titers (Fig. 4G), plaque sizes ( Fig. 4H and I), and MDV copy numbers (Fig. 4J). Taken together, the results demonstrated that de novo synthesis of cholesterol is important for MDV replication and spread.
Exogenous cholesterol is also required for MDV replication in CEFs. To determine the importance of exogenous cholesterol in MDV replication, we examined the cell viability of mock-or MDV-infected CEFs in cell culture medium with different concentrations of cholesterol (ranging from 0 to 2,000 nM). High levels of cholesterol (1,500 and 2,000 nM) were toxic to mock-infected CEFs, while these levels of cholesterol were nontoxic to MDV-infected CEFs as determined based on cell viability using 7-AAD staining (Fig. 5A), suggesting that MDV infection alters the susceptibility of the cells to exogenous cholesterol. To further interrogate the role of exogenous cholesterol in MDV replication, we analyzed MDV titers in MDV-infected CEFs that were cultured in medium containing different concentrations of cholesterol. Lower MDV titers (Fig. 5B) and smaller plaque sizes (Fig. 5C) were observed in MDV-infected CEFs that were cultured in cell culture medium lacking exogenous cholesterol. Furthermore, the result demonstrated that addback of cholesterol increased MDV titers (Fig. 5B). Furthermore, the fate of exogenous cholesterol added to infection medium was determined by adding fluorescent TopFluor cholesterol to culture media of RB1B-and mock-infected CEF cells. Both RB1B-and mock-infected CEF cells showed uptake of TopFluor cholesterol from the surrounding medium (Fig. 5E). Furthermore, quantification of the mean fluorescence intensity (MFI) showed significantly higher intensity and uptake of TopFluor cholesterol preferentially in CEF cells infected with RB1B (average MFI of 50.9) than in mock-infected cells (average MFI of 28.4) (Fig. 5F). Also, RB1B-infected CEF cells cultured in the presence of exogenous cholesterol showed significantly higher accumulation of lipid droplets (average of 15 droplets per cell) than in RB1B-infected CEF cells cultured without exogenous cholesterol (average of 7 droplets per cell) ( Fig. 5G and H). Taken together, our data demonstrate that exogenous and de novo-synthesized cholesterol are involved in MDV replication and spread.
Modulation of cholesterol trafficking impairs MDV replication and spread. U18666A, a chemical inhibitor that inhibits late endosomes (MVBs) and cholesterol transport (Fig. 6A), was utilized to study the role of cholesterol trafficking in MDV replication and spread. The results demonstrate that inhibition of cholesterol trafficking by nontoxic concentrations of U18666A significantly reduced MDV titers (Fig. 6B) and plaque sizes (Fig. 6C). Exogenous cholesterol (628 nM) did not rescue the inhibitory effects of U18666A (10 ng/ml) on MDV spread ( Fig. 6C and D). Furthermore, a significant reduction in MDV genome copy numbers was observed in U18666A-treated MDVinfected CEFs compared to vehicle-treated cells (Fig. 6E).
The colocalization of filipin III, a fluorescent probe for cholesterol, and MDV (gB of MDV) was analyzed using confocal immunofluorescence microscopy. Vehicle-treated MDV-infected CEFs showed a diffuse distribution of cholesterol ( Fig. 6F and G), and the peak relative fluorescence intensity for filipin III was associated with gB of MDV (Fig. 6G). Manders' coefficient for filipin III and gB of MDV confirms the colocalization of cholesterol and the MDV viral protein (Fig. 6H). The results also demonstrated that The relative expression levels and colocalization of LAMP-1, a lysosomal marker, and gB of MDV were analyzed in MDV-infected CEFs using confocal microscopy. The results demonstrate that MDV-infected CEFs have higher expression levels of LAMP-1 as identified by the higher fluorescence intensity in MDV-infected cells than in mockinfected CEFs. Moreover, LAMP-1 and gB of MDV are colocalized in infected cells ( Fig.  6I and J). Manders' coefficient for LAMP-1 and gB of MDV confirms these observations (Fig. 6K), suggesting the involvement of the MVB pathway in MDV replication. The results also demonstrate a relationship between cholesterol (filipin III), LAMP-1, and gB in MDV-infected CEFs ( Fig. 6H and K to N).
LAMP-1 is essential for efficient replication and spread of MDV. To corroborate the importance of LAMP-1 in MDV infection, we initially verified the shRNA silencing of LAMP-1 and showed a downregulation of LAMP-1 protein in CEFs using confocal microscopy ( Fig. 7A and B). Interestingly, shRNA silencing of LAMP-1 in MDV-infected CEFs reduced the expression of gB in MDV-infected CEFs, while transfection with pScramble shRNA had no effect on gB expression in these cells ( Fig. 7C and D). The results also demonstrate that shRNA silencing of LAMP-1 significantly reduced MDV titers (Fig. 7E), plaque sizes ( Fig. 7F and G), and viral copy numbers (Fig. 7H). Taken together, the results demonstrate that LAMP-1 is required for the efficient replication and spread of MDV, which is consistent with the involvement of the MVB pathway in MDV replication.

DISCUSSION
The salient findings of this study are that (i) MDV infection increases the cholesterol content of infected cells and activates de novo cholesterol synthesis; (ii) exogenous and de novo cholesterol biosynthesis are required for efficient MDV replication and spread; (iii) MDV induces the upregulation of LAMP-1, involved in cholesterol trafficking, within multivesicular bodies (MVBs); and (iv) glycoprotein B of MDV, an essential viral protein for replication and egress, is associated with cholesterol and LAMP-1 within MVBs. Interestingly, gene silencing of LAMP-1 reduced both MDV replication and spread, highlighting the importance of LAMP-1 as a protein for the transportation of gB. The results suggest that MDV hijacks cellular cholesterol biosynthesis and cholesterol trafficking to facilitate cell-to-cell spread in a LAMP-1-dependent mechanism. The results from this study could be used to design control strategies to use chemical or physiological inhibitors targeting the cholesterol pathway to reduce virus replication and shedding. Alternatively, novel vaccines such as herpesvirus of turkeys (HVT) vaccines expressing shRNA for silencing the cholesterol pathway could be developed to reduce MDV replication in vivo.
MDV is a highly cell-associated alphaherpesvirus that infects chickens and causes deadly lymphoma and immunosuppression (1,2,20). Although MDV infection disturbs lipid metabolism and causes atherosclerosis (7-9), very little is known about the role of MDV-induced cholesterol metabolism in MDV replication and spread. MDV can infect various cell types; however, it transforms only CD4 ϩ T cells (1,21). In vitro, MDV infection of lymphocytes occurs only after the activation of B and T cells (22), which can mask a metabolic alteration under MDV infection (2,23). Therefore, here, we analyzed the role of cholesterol in MDV infection in CEFs, which can be infected with an inoculum without any external activation. CEFs contain various cell populations, including immune cells and fibroblasts, which are used for the propagation of MDV. The exact proportion of immune cells within CEFs is unknown as there are no specific    (24). Moreover, cholesterol and lipid rafts can also be involved in virus replication by influencing the intracellular membrane structures (25). MDV is a highly cell-associated virus, and the role of cholesterol in MDV cell-to-cell spread and replication needs to be further explored. This study demonstrated that genes involved in cholesterol synthesis and trafficking are upregulated and that the levels of cholesterol in MDV-infected cells are increased, suggesting that MDV infection increases cholesterol biosynthesis. Interestingly, MDV-infected cells became resistant to cytotoxicity, which is generally exerted by excessive exogenous cholesterol. Cell apoptosis is triggered by the enrichment of free cholesterol in the ER (26), which plays a major role in protein homeostasis and cholesterol production. It is possible that the usage and trafficking of free cholesterol remove free cholesterol from the ER in MDV-infected cells. Our confocal microscopy data demonstrating the colocalization of cholesterol and gB of MDV within LAMP-1 MVBs support this notion. To our knowledge, this is the first report confirming the crucial role of cholesterol synthesis and trafficking in MDV replication and spread.
LAMP-1 shuttles between lysosomes, endosomes, and the plasma membrane (27) and binds to cholesterol in association with Niemann-Pick disease type C1 (NPC1) and NPC2 proteins that export cholesterol from lysosomes (28). Our results demonstrated that MDV infection enhanced the expression of LAMP-1, a marker for the MVB compartment, reflecting the augmentation and involvement of the MVB pathway in MDV replication. The mechanism involved in the upregulation of LAMP-1 during MDV infection is unknown, and further work is required to establish any possible association between cholesterol synthesis and LAMP-1 expression in MDV-infected cells. Confocal microscopy showed the colocalization of gB with cholesterol and LAMP-1 within MVB membranes. MDV-encoded gB is one of the essential MDV glycoproteins, which is conserved in most herpesviruses, and here, we show that gB may be sorted to the MVB compartment, and confocal microscopy showed that gB was colocalized in part with the MVB membranes, which therefore represent a site of gB accumulation. Herpesvirus glycoproteins play an important role in the assembly and production of infectious  particles, and gB of herpesviruses recirculates between the plasma membrane and MVBs (16,29,30). RNA interference with LAMP-1 reduced MDV titers and plaque sizes, highlighting the importance of LAMP-1 and MVBs in MDV replication and cell-to-cell spread. It has been suggested that avian LAMP-1 is involved in the degradation of the avian reovirus nonstructural p10 protein, involved in the induction of cell syncytium formation and apoptosis, through interacting with both p10 and the E3 ligase Siah-1 (31,32). We are currently examining whether LAMP-1 is also involved in the degradation of MDV viral proteins, which are required for the generation of infectious cell-free MDV. Feather follicle epithelial cells are the only cells capable of producing infectious cell-free MDV in vivo, and we have recently generated feather follicle epithelial cell lines that will be used to differentially examine the role of LAMP-1 in the degradation of MDV proteins. In conclusion, the results suggest that MDV activates LAMP-1 expression and hijacks cellular cholesterol biosynthesis and cholesterol trafficking to facilitate MDV trafficking in a LAMP-1-dependent mechanism.

MATERIALS AND METHODS
Ethics statement. Ten-day-old mixed-sex specific-pathogen-free (SPF) embryonated chicken eggs, which were purchased from Valo Biomedia GmbH, were used to generate primary CEFs. All embryonated chicken eggs were handled in strict accordance with the guidance and regulations of European and United Kingdom Home Office regulations under project license number 30/3169. The experiments were approved by the ethics committee at The Pirbright Institute. CEF culture and virus preparations. CEFs were generated from mixed-sex SPF Valo eggs (Valo Biomedia GmbH) incubated in a Brinsea Ova-Easy 190 incubator at 37°C until 10 days in ovo. CEFs were seeded at a rate of 1.5 ϫ 10 5 cells/ml in 24-well plates with growth medium (E199 medium supplemented with 10% TBP, 5% fetal calf serum [FCS], 2.8% Milli-Q-filtered water, amphotericin B [0.01%], penicillin [10 U/ml], and streptomycin [10 g/ml]) and incubated overnight (38.5°C at 5% CO 2 ). The next day, 80% confluent monolayers were observed, and growth medium was replaced with maintenance medium (E199 medium supplemented with 10% tryptose phosphate broth (TPB) 2.5% FCS, 3.5% SQ water, amphotericin B [0.01%], penicillin [10 U/ml], and streptomycin [10 g/ml]).
Cells and MDV infection: metabolomics. CEFs were either mock infected or infected with the very virulent RB1B strain (100 PFU per 1.5 ϫ 10 5 cells or a multiplicity of infection [MOI] of 0.0006) in triplicates and harvested at 48 and 72 h postinfection (hpi). The cells were washed and counted; after protein quantification using the Bradford assay (33), the samples were sent for metabolome analysis using GC/GC-MS (Target Discovery Institute, University of Oxford); and data analyses were performed as previously described (34). In brief, the cells were homogenized using a bead beater in methanol-water (1:1), and t-butyl methyl ether was then added for phase separation. The organic phase was dried under a vacuum, while methanol was added to the remaining sample and mixed in the bead beater. After incubation at Ϫ80°C for 1 h, phase separation occurred after centrifugation, and the liquid layer was replicates per biological replicate. The data were adjusted and normalized based on the protein content. MDV infection did not change the size of the cells as determined using light microscopy.
(ii) Knockdown of HMG-CoA, SqE, and LAMP-1 expression. Small short hairpin RNAs (shRNAs), both silencing (pshRNA) and control scrambled (pScrambled) RNA sequences, were designed using siRNA Wizard (InvivoGen, Toulouse, France) ( Table 1) and chemically synthesized (Sigma-Aldrich/Merck, Dorset, UK). Construction of the shRNA-harboring plasmid was performed according to the manufacturer's instructions using the psiRNA-h7SKGFPzeo plasmid. The shRNAs were annealed and ligated into the BbsI-digested psiRNA-h7SK-GFPzeo plasmid vector (catalog no. ksirna4-gz21). The plasmid constructs with shRNAs were transformed into chemically competent GT116 Escherichia coli cells for blue/white screening and zeocin selection and amplification. The positive colonies were propagated, and the (iii) Viral titer. MDV-infected cells were titrated onto fresh CEFs. Cells were fixed (1:1 ice-cold acetone-methanol for 5 min) and blocked with blocking buffer (phosphate-buffered saline [PBS] plus 5% FCS) for 1 h at room temperature (RT). Cells were subsequently incubated with anti-gB mAb (HB-3) and then with horseradish peroxidase-conjugated rabbit polyclonal anti-mouse IgG. After the development of the plaques using the 3-amino-9-ethylcarbazole (AEC) substrate, the cells were washed with Super Q water, and viral plaques were counted using a light microscope.
(iv) Determining nontoxic concentrations of the inhibitors. To identify nontoxic concentrations of the chemicals, mock-infected and MDV-infected CEFs were exposed to the chemicals or vehicles, and cell morphology and adherence/confluence were monitored by light microscopy at different time points posttreatment. Moreover, CEFs were trypsinized, stained with 7-AAD (BD Bioscience, Oxford, UK), and acquired using a MACSQuant flow cytometer and FlowJo software for analysis of the data. Nontoxic concentrations of the inhibitors and chemicals were selected based on flow cytometry data and confluence.
qPCR to amplify MDV genes. DNA samples were isolated from 5 ϫ Standard curves generated for both Meq (10-fold serial dilutions prepared from the plasmid construct with a Meq target) and the ovo gene (10-fold serial dilutions prepared from the plasmid construct with an ovo target) were used to normalize DNA samples and quantify MDV genomes per 10 4 cells. All reactions were performed in triplicates to detect both Meq and the chicken ovotransferrin (ovo) gene on an ABI7500 system (Applied Biosystems) under standard conditions. MDV genomes were normalized and reported as viral genomes per 10 4 cells.
Plaque size measurement. There were approximately 70 to 100 plaques in each well, and many plaques were fused with their adjacent plaques or were located at the edge of the wells, which made measuring their sizes very difficult. Therefore, in previous studies, we standardized our methods and measured the sizes of all the plaques that could be measured correctly and compared them with the results from 20 plaques from each well. The data indicated that the size of 20 plaques from each well represents accurate average sizes of measurable plaques. Therefore, at 72 hpi, the treated CEFs were fixed, and virus plaques were visualized using the AEC substrate (as described above). Viral plaques were imaged (10ϫ lens objective) with an inverted light microscope, and the pictures were processed using Adobe Photoshop software. All viral plaques were measured using the ImageJ software area tool (NIH, USA). The freehand tool was used to measure the defined plaque area (square micrometers) based on anti-gB staining as described previously (5,6). Data were corrected and exported as plaque size areas (square millimeters).
Real-time PCR. (i) RNA extraction and cDNA. Total RNA was extracted from mock-and MDVinfected CEFs using TRIzol (Thermo Fisher Scientific, Paisley, UK) according to the manufacturer's protocol, and the purified RNA was reverse transcribed to cDNA using a Superscript III first-strand synthesis kit (Thermo Fisher Scientific, Paisley, UK) and oligo(dT) primers according to the manufacturer's recommended protocol.
(ii) SYBR green real-time PCR. Quantitative real-time PCR using SYBR green was performed on cDNA using the LightCycler 480 II system (Roche Diagnostics GmbH, Mannheim, Germany) as previously described (6). Briefly, each reaction involved a preincubation step at 95°C for 5 min, followed by 40 cycles of 95°C for 20 s, 55°C to 64°C (primer annealing temperature) for 15 s, and elongation at 72°C for 10 s. Subsequent melt curve analysis was performed by heating to 95°C for 10 s, cooling to 65°C for 1 min, and heating to 97°C. Primer sequences and accession numbers are outlined in Table 2. Relative expression levels of all genes were calculated relative to the housekeeping gene ␤-actin using LightCycler 480 software (Roche Diagnostics GmbH, Mannheim, Germany). Data represent means from 6 biological replicates in duplicates.
Western blotting. Samples were lysed in lysis buffer in the presence of protease inhibitors (Thermo Fisher Scientific, Paisley, UK). The lysates were suspended in sample loading buffer (Sigma-Aldrich/Merck, USA) and loaded in a 10% SDS-PAGE gel. After semidry transfer of SDS-PAGE gels, nitrocellulose membranes were blocked (5% skimmed milk powder in PBS for 2 h at RT). Membranes were incubated overnight (4°C) with primary antibodies, rabbit monoclonal antibody to chicken tubulin, rabbit monoclonal antibody to SqE, and rabbit monoclonal antibody to HMG-CoA, in the respective blots. The next day, membranes were washed (0.5% skimmed milk powder in PBS) and incubated (5% skimmed milk powder in PBS for 2 h at RT) with a secondary antibody, donkey pAb to goat IgG Red, in the respective blots. Finally, blots were probed with the Odyssey CLx imaging system (Li-Cor, USA), bands were quantified with Image Studio Lite software, and images were processed using Adobe Photoshop software.
Flow cytometry. The viabilities of CEFs treated with the pharmacological inhibitors or diluent (vehicles) were determined to identify their nontoxic concentrations. Each pharmacological inhibitor was titrated, and CEFs were exposed to the inhibitors for 72 h.
Viability assay. Trypsinized CEFs were stained with 7-AAD (BD Biosciences, Oxford, UK) and acquired using a MACSQuant flow cytometer, and cell viability was analyzed using FlowJo software. Nontoxic concentrations of the inhibitors and chemicals were selected based on cell viability determined using flow cytometry data and confluence. CEFs were acquired in triplicates from 4 independent experiments using a MACSQuant flow cytometer, and the results were analyzed using FlowJo software.
Fluorescence confocal microscopy. Seventy-two hours after mock infection or infection with the RB1B virus in the presence/absence of U18666A or shRNA and pScrambled-overexpressing CEFs targeting LAMP-1, the samples were prepared for imaging. CEFs were trypsinized and seeded in 24-well plates that contained 12-mm-diameter round coverslips at a rate of 1.0 ϫ 10 5 cells per well. In brief, mock-infected or infected CEFs were fixed with 4% formaldehyde for 30 min at RT and washed twice with PBS. Cells were subsequently permeabilized (where indicated) with 0.1% Triton X-100 buffer solution, blocked (1 h in 0.5% BSA-PBS), and incubated overnight (4°C) with anti-gB (mouse IgG2b; 1 g/500 l in 0.5% BSA-PBS) or anti-LAMP-1 (mouse IgG1; 1 g/ml in 0.5% BSA-PBS). Cells were washed twice with PBS and again incubated overnight (4°C) with anti-IgG2b-568 nm (goat anti-mouse, 0.5 g/500 l in 0.5% BSA-PBS) or goat anti-IgG1-568 nm (goat anti-mouse, 0.5 g/500 l in 0.5% BSA-PBS). The next day, nuclei were labeled with DAPI. Coverslips were mounted in Vectashield mounting medium for fluorescence imaging.
(i) Cholesterol staining. Cellular localization of cholesterol based on filipin III staining was performed according to the manufacturer's recommendations (Abcam Ltd., Cambridge, UK). In brief, CEFs were fixed in the cell-based assay fixative solution for 10 min (RT in the dark). Cells were subsequently washed and stained (1 h at RT in the dark) with filipin III (1:100 dilution of a filipin III stock solution in cholesterol detection assay buffer). After a final wash, coverslips were mounted in Vectashield mounting medium for fluorescence imaging (excitation of 340 to 380 nm and emission of 385 to 470 nm).
(ii) TopFluor cholesterol staining. CEF cells were either mock infected or infected with RB1B virus in maintenance medium containing 628 nM TopFluor cholesterol for 72 h, and samples were prepared for confocal imaging.
(iii) Lipid droplet staining. CEF cells were either mock infected or infected with RB1B virus in maintenance medium containing 628 nM cholesterol for 72 h, and samples were prepared for confocal microscopy. TopFluor cholesterol and lipid droplet groups were stained with Alexa Fluor 568 goat anti-mouse IgG2b and Alexa Fluor 488 goat anti-mouse IgG2b (Thermo Fisher Scientific, Paisley, UK) secondary antibodies, respectively. Further lipid droplets groups were stained with HCS LipidTOX red neutral lipid stain.
(iv) Visualization. Cells were viewed using a Leica SP2 laser scanning confocal microscope, and optical sections were recorded using either 663 or 640 nM with a numerical aperture of 1.4 or 1.25, respectively. All data were collected sequentially to minimize cross talk between fluorescence signals. The data are presented as maximum projections of z-stacks (23 to 25 sections; spacing of 0.3 mm). Maximum projections of z-stacks were analyzed with LAS AF Lite software for localization of the relative fluorescence intensity across a straight line.
(v) Manders' colocalization coefficient. All images were processed using Fiji software, and Manders' colocalization coefficient was calculated using the colocalization (COLOC) function for a specified region of interest (ROI). Manders' M1 and M2 weighted coefficients were calculated to determine the extent of colocalization between a pair of fluorescent signals and imaged in two channels. Manders' M1 determines the degree of channel colocalization (gB with filipin II, gB with LAMP-1, and LAMP-1 with filipin III), and M2 determines the reverse. The uptake of fluorescent cholesterol by mock-and RB1B-infected CEFs was analyzed using ImageJ and estimated as the mean fluorescence intensity (MFI) per cell. For estimation of lipid droplet formation, in total, 15 z-stack cells from mock-and RB1B-infected cells were analyzed using IMARIS (Bitplane Scientific Software).
Statistical analysis. All data are presented as means Ϯ standard deviations (SD) from at least three independent experiments. Quantification was performed using GraphPad Prism 7 for Windows. The differences between groups, in each experiment, were analyzed by one-way analysis of variance (ANOVA) followed by a Tukey multiple-comparison test to identify those groups that differed if the ANOVA result was significant. Results were considered statistically significant at a P value of Ͻ0.05 (*).