We examined the transcriptional responses of Drosophila S2 cells to FHV RNA replication using genome-wide microarray analyses. To focus on cellular responses related to RNA replication rather than those associated with other steps in the viral life cycle, such as virion attachment, entry, uncoating, encapsidation, and virion release, we conducted parallel analyses of cells either infected with FHV or transfected with an RNA1 replicon-expression plasmid. The RNA1 segment of the FHV genome encodes the viral RNA-dependent RNA polymerase (Fig. 1A) and is the only genome segment required for RNA replication (Fig. 1C). Thus, FHV RNA replication can be initiated in S2 cells by the introduction of a self-replicating RNA1 via the inducible plasmid pS2F1 (Figs. 1B and 1C) 22. To maximize the number of S2 cells expressing the replicon, we first generated stable cell lines containing pS2F1 or a control plasmid (pS2F1_fs, see Fig. 1B legend) containing a translation-defective RNA1 segment. Although there was a low level of baseline FHV RNA replication in pS2F1-expressing cells due to leaky metallothionein promoter activity, induction with copper sulfate resulted in a dramatic increase in subgenomic RNA3 and protein B2 accumulation, indicative of highly active viral RNA replication complexes (data not shown).

We analyzed transcript levels from three independent experiments with either FHV-infected cells or cells expressing an RNA1 replicon, and identified cellular genes whose transcription was significantly up- or downregulated in response to these stimuli. We identified 790 genes that were significantly upregulated and 392 genes that were significantly downregulated in FHV-infected S2 cells (Additional File 1), and 305 genes that were significantly upregulated in pS2F1-expressing cells (Additional File 2). There were no significantly downregulated genes in pS2F1-expressing cells. Full details regarding the molecular functions, biological processes, and cellular components for these genes are provided in the additional files. We further analyzed these datasets for transcripts that were upregulated with both stimuli and identified 73 co-regulated genes, which encoded proteins involved in numerous aspects of cellular physiology, such as metabolism, defense responses, and signal transduction (Table 1 and Additional File 3). Based in part on the well recognized importance of cellular membranes in positive-strand RNA virus replication 345, we focused on those genes that encoded proteins with known or hypothesized roles in lipid metabolism. We found that infection with FHV upregulated eight lipid metabolism-associated genes, whereas RNA1 replicon expression upregulated thirteen lipid metabolism-associated genes (Fig. 2A). Four of these genes were upregulated in both datasets, including Cct1 and Cct2, which encode CTP:phosphocholine cytidylyltransferases. These enzymes are essential for the salvage pathway of PC synthesis but have non-overlapping functions in Drosophila cells 23242526. We validated both Cct1 and Cct2 mRNA upregulation by RT-PCR after FHV infection (Fig. 2B) and RNA1 replicon expression (Fig. 2C). These results demonstrated that FHV RNA replication in Drosophila S2 cells was associated with distinct alterations in cellular gene transcription, including the transcription of genes associated with lipid metabolism.

The conclusion that FHV RNA replication was associated with transcriptional upregulation of lipid metabolism genes predicts that quantifiable changes in lipid constituents should also be observed, particularly with respect to PC content. To examine this prediction we measured total PC levels in S2 cells after infection or replicon expression using a modified phospholipase D-based biochemical assay 27 (Fig. 3A). As a control for these experiments we used cells treated with 1-hexadecylphosphorylcholine (miltefosine), which is a specific CTP:phosphocholine cytidylyltransferase inhibitor 28. S2 cells infected with FHV had a 35% increase in total PC levels, whereas treatment with miltefosine reduced PC levels by 40% (Fig. 3A, left graph). Furthermore, although baseline PC levels in stably transfected cells were increased compared to untransfected cells, expression of an FHV RNA1 replicon was associated with a 70% increase in PC levels (Fig. 3A, right graph). These results indicated that FHV RNA replication-associated transcriptional upregulation of PC biosynthetic genes in S2 cells (Fig. 2) was associated with a concomitant increase in cellular PC levels.

To evaluate potential lipid changes associated with FHV replication in S2 cells in more detail we used electrospray ionization tandem mass spectrometry (ESI-MS/MS) to quantitate polar phospholipids in extracts from FHV-infected cells. This analysis allowed the determination of relative individual phospholipid species content as a molar percentage of total recovered polar phospholipids but also provided details on total acyl chain length and saturation levels. As controls for these experiments we used cells treated with miltefosine to decrease PC levels or cells cultured with 1 mM oleic acid (cis-9-octadecenoic acid), which increases the production of phospholipids with individual C18:1 acyl chains and hence total sum acyl chain compositions of C36:2. Complete detailed results from the ESI-MS/MS analyses are provided in Additional File 4. Control membrane extracts from S2 cells contained predominantly PC and phosphatidylethanolamine (PE) as the primary phospholipids, with the latter representing the majority (~40%) of all phospholipids (Fig. 3B). Although PC is typically the most prominent phospholipid in most eukaryotic cell membranes, Diptera species such as Drosophila contain relatively high levels of PE 29. As expected, miltefosine dramatically reduced PC content with compensatory increases in the percentages of PE and phosphatidylinositol (PI). In contrast, total cellular membranes from FHV-infected S2 cells showed a trend towards an increase in the molar percentage of PC and phosphatidylserine (PS) with an associated decrease in PE content, but these differences were not statistically significant (Fig. 3B). However, when we analyzed PC species based on total acyl chain length or saturation level there were modest but statistically significant changes in the percentage of PC molecules with total acyl chain lengths of 32 or 34 carbons (Fig. 3C) and 0, 1, or 2 total double bonds (Fig. 3D), where FHV infection increased the fraction of PC molecules with longer unsaturated acyl chains. Furthermore, the selective increase in longer unsaturated acyl chains in FHV-infected cells was seen specifically with PC and not PE, PI, or PS (Fig. 3E and data not shown). These quantitative lipid analysis results suggested that FHV infection induces selective changes in PC metabolism, and are consistent with the observation that FHV RNA replication complex activity in isolated membrane fractions analyzed in vitro is preferentially stimulated by phospholipids with increased acyl chain length and decreased saturation 21.

We next examined the potential functional significance of lipid metabolism gene upregulation on FHV RNA replication in S2 cells using a targeted genetic approach that employed RNAi-mediated knockdown 17. We identified thirty-one cDNAs within the O'Farrell RNAi collection 30 that were targeted toward genes involved in glycerophospholipid metabolism in Drosophila cells, generated 300-600 bp dsRNAs, and performed RNAi knockdown experiments with infected S2 cells using dsRNA targeted against FHV RNA1 and miltefosine as positive controls (Fig. 4A). To quantitate FHV RNA replication, we used an optimized capture enzyme-linked immunosorbant assay (ELISA) that detected protein B2, which is a viral protein only produced during active RNA replication (see Fig. 1C). Although protein B2 is an RNAi suppressor, it functions during the assembly of RNAi-induced silencing complexes 31, and therefore would have less impact on the activity of functional silencing complexes formed during the 48 h dsRNA incubation period prior to FHV infection. We found that dsRNAs directed against five cellular genes, Ace, Cct1, Cct2, fu12, and san, all significantly suppressed FHV RNA replication in infected S2 cells (Fig. 4A). Although the level of suppression was 20-40% with individual knockdown of these five genes, the positive control FHV dsRNA only reduced protein B2 levels by ~45%, suggesting that this medium throughput assay provided a conservative estimate of the functional impact of cellular glycerophospholipid metabolism gene knockdown. The functions of Cct1 and Cct2 have been described above. Ace encodes acetylcholine esterase 32, fu12 encodes an enzyme with 1-acylglycerol-3-phosphate acyltransferase activity 33, and san also encodes an enzyme with acyltransferase activity 34. The enzymes encoded by Cct1, Cct2, Ace, fu12, and san are all involved in PC synthesis, either directly or indirectly (Fig. 4B). The main glycerophospholipids in the cell, PC and PE, are synthesized through two metabolic pathways, the de novo synthesis pathway that uses PS as a precursor for both, and the salvage or CDP pathway, also referred to as the Kennedy pathway, which recycles choline and ethanolamine to produce PC and PE, respectively 35. In higher eukaryotes the salvage or CDP pathway of PC synthesis normally predominates, where the conversion of phosphocholine to CDP-choline, which is catalyzed by CTP:phosphocholine cytidylyltransferases such as those encoded by Cct1 and Cct2, is the rate limiting step 35. Thus, we chose to focus further on these two Drosophila glycerophospholipid synthesis genes.

To verify the functional impact of Cct1 and Cct2 activity on FHV RNA replication in S2 cells, we generated alternative dsRNAs targeting different coding regions of these two genes and repeated RNAi-mediated knockdown experiments, including co-knockdown of both Cct1 and Cct2 mRNA (Fig. 5). We validated knockdown of Cct1 and Cct2 mRNA expression by RT-PCR, where we obtained an approximate 80-90% gene-specific reduction in steady state levels (Fig. 5A). Furthermore, quantitative analysis demonstrated a 30-35% reduction in total PC levels with individual knockdown of either Cct1 or Cct2 mRNA, and a ~60% reduction with knockdown of both genes (Fig. 5B). The incomplete knockdown of Cct1 and Cct2 mRNA, the presence of a Cct-independent de novo PC synthesis pathway (see Fig. 4B), and the essential role that PC plays in cellular structure and metabolism all likely contributed to the lack of synergistic suppression of total PC levels with an RNAi-based approach that targeted one step in the biosynthetic pathway. We also examined cell viability by MTT assay, as disruption of CTP:phosphocholine cytidylyltransferase expression or function in mammalian cells results in enhanced susceptibility to apoptosis 363738. Individual knockdown of either Cct1 or Cct2 mRNA had no significant impact on cell viability, whereas co-knockdown of both genes resulted in a small but reproducible 15% reduction in cell viability as measured by MTT assay (Fig. 5C). We infected S2 cells with FHV after RNAi-mediated suppression of Cct1, Cct2, or both genes, and measured viral RNA and protein accumulation at 12 h after infection. Downregulation of either Cct1 or Cct2 individually suppressed FHV subgenomic (+)RNA3 accumulation by ~30%, but did not significantly reduce the accumulation of genomic (+)RNA1, (-)RNA1, or protein A (Figs. 5D and 5E). However, downregulation of both Cct1 and Cct2 resulted in a 40-65% reduction in the accumulation of all viral products examined, similar to the levels of suppression seen with the positive controls miltefosine and dsRNA directly targeting FHV RNA1. Although we cannot exclude the possibility that the small reduction in cell viability with co-knockdown of both Cct1 and Cct2 (Fig. 5C) adversely affected viral RNA or protein accumulation, potentially via apoptosis-mediated reduction in synthesis or stability, published studies have demonstrated that FHV replication is not sensitive to the antiviral effects of apoptosis induction in Drosophila cells 39. These results indicated that genetic or pharmacologic disruption of PC synthesis inhibited FHV RNA replication in infected S2 cells.

To further examine the functional impact of Cct1 and Cct2 on FHV RNA replication in the absence of infectious virus, we repeated experiments using RNAi-mediated downregulation of these genes in S2 cells expressing an RNA1 replicon (Fig. 6). We could not use stably transfected cells for genetic knockdown experiments, as the low level of FHV RNA replication in these cells in the absence of copper sulfate induction resulted in protein B2 production that inhibited RNAi-mediated downregulation of cellular genes (data not shown). Therefore, we used a transient transfection approach. We treated S2 cells with dsRNAs for 48 h, transfected cells with pS2F1 (see Fig. 1B) and control plasmid pS2LacZ, and measured FHV RNA replication by northern blotting and β-galactosidase activity by enzymatic assay 18 h after induction. We used pS2LacZ as a transfection control, as modulation of PC levels may influence transfection efficiency. Knockdown of either Cct1 or Cct2 suppressed FHV (+)RNA3 accumulation by 20-40% in pS2F1-transfected cells, whereas knockdown of both genes increased suppression to 65%, similar to levels seen with control dsRNA targeted against FHV RNA1 (Figs. 6A and 6B). We did not quantitate genomic (+)RNA1 or protein A accumulation levels in transfected cells, as these viral products accumulated to low and almost undetectable levels in pS2F1-transfected cells (Fig. 6A and data not shown), such that attempts at quantitation would have been unreliable. Nevertheless, these results supported the conclusion that disruption of PC synthesis by genetic modulation of Cct1 or Cct2 inhibited FHV RNA replication.

Finally, we examined whether disruption of Cct1 or Cct2 expression directly suppressed FHV protein A accumulation or membrane association (Fig. 7), as we have previously demonstrated that cellular factors can impact viral RNA polymerase production 1722 and hence facilitate an early step in RNA replication complex assembly (see Fig. 1C). Although protein A levels were decreased in FHV-infected cells with RNAi-mediated downregulation of both genes (Figs. 5D and 5E), protein A accumulation in the setting of infectious virus or an FHV replicon is directly linked to RNA replication (see Fig. 1C). Thus, we used a protein A expression vector, called pS2FA-HA (Fig. 7A), which is designed to optimize translation and prevent RNA replication via modification of 5' and 3' untranslated sequences that contain essential cis elements 1722. In contrast to the effects of Cct1- or Cct2-specific dsRNAs on FHV RNA replication (Figs. 5 and 6), knockdown of either gene had no effect on RNA polymerase accumulation in cells transfected with pS2FA-HA (Fig. 7B), whereas dsRNAs against Hsp83 suppressed polymerase accumulation by 60%, consistent with published studies demonstrating the important role of hsp90 chaperones in FHV protein A synthesis 1722. Furthermore, knockdown of Cct1, Cct2, or both genes did not alter the ability of protein A to associate with intracellular membranes, as determined by differential centrifugation (Fig. 7C). These results indicated that PC synthesis was important for FHV RNA replication but not viral polymerase production or membrane association.

Drosophila S2 cells were cultured in Schneider's Drosophila media (SDM) and transfected with inducible expression plasmids as previously described 17. Sucrose gradient-purified FHV 11 was used for all infection experiments.

Standard molecular biology procedures were used for all cloning steps. The FHV RNA1 replicon expression plasmid pS2F1 (Fig. 1B), the protein A expression plasmid pS2FA-HA, and the control plasmid pS2LacZ have been previously described 17. The control plasmid pS2F1_fs, which contains an early frameshifting mutation in the protein A coding region (Fig. 1B) and hence will not produce a functional cis replicon transcript 612, was generated by cloning the ScaI/BsrGI fragment from pF1_fs 6 into the MscI/Acc65I sites of pMT-V5/HisA (Invitrogen, Carlsbad, CA). Rabbit polyclonal antibodies against FHV protein A have been previously described 11, and antibodies against FHV protein B2 were generously provided by Paul Ahlquist (University of Wisconsin-Madison). Rabbit polyclonal antibodies against the hemagglutinin (HA) epitope tag or the voltage-dependent anion channel porin were purchased from Santa Cruz Biotechnology (Santa Cruz, CA) or Affinity Bioreagents (Golden, CO), respectively, while monoclonal antibodies against tubulin were purchased from the Developmental Studies Hybridoma Bank (University of Iowa, Iowa City, IA). All secondary reagents for ELISA and immunoblot analyses were purchased from Jackson Immunoresearch (West Grove, PA). The CTP:phosphocholine cytidyltransferase inhibitor miltefosine (1-hexadecylphosphorylcholine) was purchased from Calbiochem (San Diego, CA) and stored as a 100 mM stock solution in dimethylsulfoxide at -20°C. Oleic acid (cis-9-octadecenoic acid) was purchased from Sigma (St. Louis, MO) and stored as an undiluted stock solution at -20°C.

Cells were either infected with FHV at a multiplicity of infection of 10 as previously described 11 and harvested at 12 to 24 h after infection, or transiently transfected with pS2F1 and pS2LacZ as previously described 17 and harvested at 18 h after induction with 0.5 mM copper sulfate. For microarray and phospholipid analyses, S2 cells stably transfected with pS2F1 or pS2F1_fs were induced with 1 mM copper sulfate and harvested at 18 h after induction. Total RNA was isolated using TRIzol reagent (Invitrogen) per the manufacturer's instruction. For microarray and RT-PCR analyses, RNA was further processed by digestion with RQ1 DNAse and subsequent purification with RNAsy columns (Qiagen, Valencia, CA). Total RNA purity, integrity, and concentration were checked by spectrophotometry and agarose gel electrophoresis, and samples were stored at -80°C until analysis.

Affymetrix Drosophila Genome 1.0 or 2.0 microarrays were used for FHV infection or replicon experiments, respectively, and biotinylated amplified cRNAs were produced using Affymetrix GeneChip target labelling kits per the manufacturer's instructions. Microarrays were scanned with an Affymetrix Scanner 3000, and complete original data files for all microarray experiments have been deposited in the Gene Expression Omnibus (GEO) database http://www.ncbi.nlm.nih.gov/geo/ under the accession number GSE15469.

Genomatix ChipInspector software package was used for primary microarray data analyses. This program uses a single probe method with an enhanced statistics package based on a SAM algorithm 59 that incorporates a t-test with a permuted artificial background to reduce false-positives. The following parameters were chosen to identify sets of up- or downregulated transcripts: (i) false-discovery rate of 1%; (ii) three probe minimum coverage; and (iii) 1.5-fold change from control. Similar results were obtained when microarray data were analyzed with the Affymetrix package of Bioconductor 60. The FlyBase database (FB2008_09; http://flybase.org/) was used to identify Gene Ontology (GO) terms, genetic interaction partners, and yeast-human orthologs for Drosophila genes identified in the microarray analyses.

First-strand cDNA synthesis was completed with equal amount of total RNA using oligo-dT primers and the Superscript II First Strand synthesis kit (Invitrogen) per the manufacturer's instructions. Ten-fold dilutions of cDNA were used for PCR reactions with gene-specific primer sets (Table 2) under the following conditions: 200 nM primers, 58°C annealing temperature, and 30 cycles. RT-PCR products were analyzed by agarose gel electrophoresis and ethidium bromide staining.

Total PC content was determined using a modified phospholipase D-based enzymatic method initially developed for measurement of plasma PC levels 27. Briefly, S2 cells were pelleted at 1,000 × g for 5 min at 4°C, washed twice with Tris-buffered saline (TBS) (100 mM sodium chloride, 50 mM Tris, pH 7.2), lysed with 1% Triton X-100, and incubated with reaction buffer containing 50 mM Tris (pH 7.2), 0.64 M calcium chloride, 0.73 M N-ethyl-N-(2-hydroxy-3-sulfopropyl)-3,5-dimethoxyanaline, 0.73 M 4-aminoantipyrine, 120 U/ml PC-specific phospholipase D, 0.5 U/ml choline oxidase, and 20 U/ml peroxidase. Reactions were incubated at 37°C for 2 h and the absorbance at 595 nm was measured using a FLUOstar Omega microplate reader (BMG Labtech, Durham, NC). Quantitative PC concentrations were calculated using a standard curve generated with purified PC. Parallel total protein concentrations were determined with cell lysates prepared using TBS and 1% n-octyl-β-D-glucopuranoside and BioRad total protein reagent with bovine serum albumin as the standard. For analysis of individual cellular phospholipids, S2 cells were pelleted, washed twice with TBS, and lipids were extracted with chloroform:methanol using the method of Bligh and Dyer 61. Final extract samples were dried under vacuum and analyzed for polar lipid content by ESI-MS/MS at the Kansas Lipidomics Research Center http://www.k-state.edu/lipid/lipidomics/index.htm.

The O'Farrell RNAi library was purchased from the Drosophila Genome Research Center (DGRC). This collection contains 300-600 bp double-stranded cDNAs corresponding to greater than 7,000 individual phylogenetically conserved Drosophila genes 30, which can all be amplified by PCR using a common oligonucleotide primer (S2-RNAi-T7) that contains a GC-rich linker and T7 RNA polymerase promoter (Table 2). PCR products were generated using Eppendorf Mastermix with 200 nM primers, annealing temperature of 62°C, and 40 cycles. If no product was obtained after the first amplification, a second PCR was done using an aliquot of the first reaction as a template. PCR products were analyzed by non-denaturing gel electrophoresis and ethidium bromide staining and purified using a PCR cleanup kit (Promega) per the manufacturer's instruction. To generate the cDNA templates for Cct1 and Cct2 validation studies, candidate RNAi targets that differed from the O'Farrell library constructs were identified using the E-RNAi program 62. To generate control FHV RNA1-specific dsRNA we targeted nucleotides 40-735, which include the initiator AUG codon. RT-PCR with gene-specific primer sets that incorporated a T7 RNA polymerase promoter (Table 2) were used to generate PCR templates for dsRNA generation. Control LacZ and Hsp83 cDNAs were produced as previously described 17.

RNAs were generated using a T7 Megascript kit (Ambion) per the manufacturer's instruction, purified by phenol-chloroform extraction and isopropanol precipitation, and resuspended in RNAse-free water. dsRNAs were generated by heating samples to 65°C for 30 min and cooling slowly to room temperature. Samples were analyzed by non-denaturing agarose gel electrophoresis and ethidium bromide staining to ensure the formation of properly sized dsRNA products, quantitated by spectrophotometry, and stored at -20°C until used for RNAi studies. RNAi was induced in cultured S2 cells as previously described 17. Briefly, 10 μg dsRNA was added to 10⁶ cells in 0.5 ml serum-free SDM, incubated at 25°C for 1 h, 0.5 ml SDM with 20% fetal bovine serum was added and cells were incubated for an additional 48-72 h prior to FHV infection or replicon transfection.

To facilitate the analysis of FHV RNA replication in a medium throughput format, an FHV protein B2 capture ELISA was developed. FHV protein B2-specific polyclonal antibodies for detection were purified from rabbit antisera by Staphylococcus aureus protein A affinity chromatography and biotinylated with 6-(biotinamidocaproylamido)caproic acid N-hydroxysuccinimide ester (Sigma) per the manufacturer's instructions. For capture antibodies, total immunoglobulins from FHV protein B2 antisera were isolated by saturated ammonium sulfate precipitation. Flexible microassay plates were coated with 5 μg/ml capture antibody in phosphate-buffered saline (PBS) (100 mM sodium chloride, 50 mM sodium phosphate, pH 7.4) overnight at 4°C, blocked with 1% non-fat milk, and S2 cell lysates in PBS with 0.5% Triton X-100 were incubated in duplicate wells for 2 h at room temperature. Plates were washed extensively with TBS and 0.1% Tween 20, incubated with 1.25 μg/ml biotinylated detection antibody followed by streptavidin-alkaline phosphatase. Plates were developed with 1 mg/ml p-nitrophenyl phosphate in 50 mM sodium carbonate buffer (pH 10.0) with 1 mM magnesium chloride and the absorbance at 405 nm was measured with the microplate reader described above. Initial optimization experiments demonstrated a high sensitivity for this assay, where lysates from less than ten S2 cells expressing a pS2F1-based replicon produced a positive signal (data not shown). Northern blot analyses for FHV-specific RNAs, immunoblot analyses for protein A accumulation, saponin-mediated permeabilization and differential centrifugation, and β-galactosidase assays were done as previously described 1722.

Viability assays were performed using 3- [4,5-dimethylthizol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) as previously described 63.

Microarray statistical analyses are described above. For additional statistical analyses a two-tailed t test assuming equal variances was used and a p value of < 0.05 was considered significant. All results are representative or a composite of at least three independent experiments, where quantitative data represent the mean ± standard errors of the mean.