Efficient Parvovirus Replication Requires CRL4-Targeted Depletion of p21 to Prevent Its Inhibitory Interaction with PCNA

Summary

Many DNA viruses induce and then exploit host cellular DNA damage responses to generate a suitable environment for their continued replication. Parvoviruses, important disease agents in both humans and animals, rely on host DNA polymerases to replicate their genomes in cell-cycle arrested cells. We show that efficient parvovirus replication requires the recruitment to viral replication compartments of a host cellular E3-ubiquitin ligase, CRL4Cdt2, to target the potent cell cycle regulator p21 for subsequent degradation. The DNA polymerase-δ cofactor PCNA provides a molecular platform for initial substrate recognition by this ligase, and subsequent p21 depletion prevents its continued interaction with PCNA which otherwise inhibits efficient viral replication. Virally-induced p21 degradation represents another way of promoting efficient replication of DNA polymerase-δ-dependent viruses.

Introduction

Minute Virus of Mice (MVM) is an autonomously-replicating parvovirus which induces a DNA damage response resulting in substantial p53 activation which persists throughout the course of viral replication [1]. p53 is a well-established activator of p21^WAF1/Cip1 (hereafter referred to as p21) expression. Transient expression of the MVM NS1 protein alone also leads to an increase in p21 levels [2], [3]. However, surprisingly, while these signals lead to an increase in p21 RNA accumulation, p21 protein levels remain low throughout the course of infection, including during the prolonged G2 phase in which the viral genome is replicated [2]. p21 can be a potent antiviral factor and possesses several potentially inhibitory activities including cyclin-dependent kinase (CDK) inhibition and repression of E2F1-mediated expression [4]. In addition, p21 has been shown to be an effective inhibitor of the DNA polymerase δ cofactor PCNA [5]–[7], and it has been shown to inhibit MVM replication in vitro [8]. p21 depletion during MVM infection was shown to be proteasomally mediated, suggesting that an E3 ubiquitin ligase was involved in targeting p21 for degradation [2].

Viruses often make use of the ubiquitin conjugation machinery to target for degradation cellular proteins that might otherwise negatively affect viral replication [9]. The Cullin-RING Ligase (CRL) CRL4^Cdt2 consists of the scaffold protein Cullin 4 and the homo-trimeric protein DDB1 which serves as an adaptor for the putative substrate recognition protein Cdt2. This ligase has been shown to program the ubiquitination and subsequent degradation of p21 in response to DNA damaging agents such as UV treatment in order to ensure low p21 levels during S-phase [10]–[12]. Upon DNA damage or S-phase entry CRL4^Cdt2 is recruited to chromatin via PCNA interaction where it targets substrate proteins for degradation [13].

We show here that efficient MVM replication in S/G2 arrested cells required the targeting for proteasomal degradation of p21 by the CRL4^Cdt2 E3-ubiquitin ligase which was re-localized to viral chromatin within active MVM replication centers. PCNA provides a molecular platform that aids substrate recognition by the CRL4^Cdt2 E3-ubiquitin ligase, and p21 targeting to this ligase during MVM infection required its interaction with PCNA. PCNA is also an important co-factor for DNA polymerase δ-dependent MVM replication which can be antagonized by p21 in vitro. Expression of a stable p21 mutant that retained interaction with PCNA inhibited MVM replication, while a stable p21 mutant which could no longer interact with PCNA did not. Introduction of a p21-derived peptide that bound to PCNA also substantially decreased viral replication. Our results suggest that interaction with PCNA was important for targeting p21 to the re-localized CRL4^Cdt2 ubiquitin ligase, yet subsequent depletion of p21 was required to prevent its sustained interaction with PCNA which otherwise inhibited efficient viral replication.

Results

The CRL4^Cdt2 ligase mediates p21 degradation during MVM infection

The CRL4^Cdt2 E3 ubiquitin ligase has been implicated in targeting p21 for proteasomal degradation upon S-phase entry and after cellular DNA damage [10]–[12]. Was this ubiquitin ligase also enlisted to target p21 at late times during MVM infection when cells were blocked at the G2/M border? To test this possibility, DDB1 and Cdt2, components of this ligase which are not present in other E3 ubiquitin ligases known to modify p21 [13], [14], were targeted via RNAi in the protocol illustrated in Figure 1A.

For these experiments cells were parasynchronized prior to infection to maximize the number of cells progressing uniformly through S-phase. This both synchronized infection and the characterization of p21 depletion. At the time of release from the synchronization procedure (as the cells progressed from G0 to G1) there were high levels of p21 expression (Mock T0, Figures 1B and 1C, lanes 1), which were reduced 24 hr post MVM infection (Figures 1B and 1C, lanes 2), and remained low up to 40 hr pi (Figures 1B and 1C, lanes 4). Targeting of endogenous DDB1 (panel 1B) or Cdt2 (panel 1C) by RNAi, which led to significant depletion of these proteins (Figure 1B and 1C, lanes 3 and 5), substantially prevented the loss of p21 both at 24 hr pi (Figures 1B and 1C, lanes 3), and also 40 hr pi (Figure 1B and 1C, lanes 5) – the latter being a point well past S-phase when infected cells are known to be arrested in G2 [1]. Expression of cyclin A indicated unperturbed entry into S-phase (Figures 1B and C, lanes 2–5), and expression of the viral NS1 protein indicated that the initiation of viral infection was unaffected by the RNAi protocol (Figures 1B and C, lanes 2–5). Similar results were also obtained following knockdown of Cullin 4A, a component of the CRL4^Cdt2 ligase also present in several other ubiquitin ligases (data not shown). Taken together, these results indicated that the CRL4^Cdt2 ligase was responsible for p21 degradation throughout virus infection.

CRL4^Cdt2 also targets Set8 and Cdt1 for degradation during S-phase and in response to DNA damaging agents [15]–[17]. We also observed loss of Set8 in response to MVM infection (Figure S1). Cdt1 levels were not reduced for reasons not yet clear.

CRL4^Cdt2 ligase function is required for efficient MVM replication

We next examined the functional consequence of CRL4^Cdt2 E3 ligase depletion for viral replication using the protocol illustrated in Figure 2A. DDB1 knockdown resulted in an approximate 2-fold decrease in accumulated viral replicative forms at each time point compared to application of negative control siRNA (Figure 2B, compare lanes 1 to 2, 3 to 4, 5 to 6). Cdt2 knockdown during infection resulted in an approximate 2.5-fold decrease in accumulated viral replicative forms at each time point when compared to negative control siRNA (Figure 2C, compare lanes 1 to 2, 3 to 4, 5 to 6). Importantly, as mentioned above, expression of NS1 (Figure 1), and flow cytometric analysis (Figure S2), confirmed normal entry into S-phase following this synchronization and RNAi protocol which thus did not affect the S-phase dependent initiation of infection. These results demonstrated that the activity of the CRL4^Cdt2 E3 ubiquitin ligase was necessary for efficient viral replication.

The CRL4^Cdt2 ligase is recruited to viral replication compartments

MVM replicates in nuclear compartments termed autonomous parvovirus associated replication (APAR) bodies which have been shown to be enriched for cellular proteins such as DNA polymerase δ, RPA, cyclin A, and PCNA, which are also essential for parvovirus replication [18]. These nuclear bodies have been shown, via BrdU staining, to serve as sites of ongoing viral replication in infected cells and can be visualized by staining for the viral replicator protein NS1 [19]. Importantly, whereas punctate staining distributed throughout the nucleus was observed in mock treated cells, we detected recruitment of both DDB1and Cdt2 to NS1-containing viral replication compartments. Resistance to detergent pre-extraction prior to immunofluorescence (Figure 3A and 3B, note the merged images for each) also suggested that it may be bound to viral chromatin. These results indicated that MVM infection redirected the viral replication-enhancing CRL4^Cdt2 ligase complex to APAR bodies. The recruitment was specific for the CRL4^Cdt2 ligase because the APC/C^Cdc20 E3 ligase, which targets p21 for degradation after mitotic entry [20], was not similarly recruited (Figure S3). This is the first demonstration of the specific recruitment of a cellular E3 ubiquitin ligase to parvoviral replication compartments.

Depletion of p21 during MVM infection requires its interaction with the CRL4^Cdt2 ligase and PCNA

The DNA polymerase δ cofactor PCNA is known to target the CRL4^Cdt2 E3 ubiquitin ligase to cellular chromatin via its interaction with Cdt2 during normal cell division [13], [21]. Ubiquitin modification of p21 by the CRL4^Cdt2 ligase requires Cdt2 interaction with the p21 degron motif, as well as interaction between p21 and PCNA via its PCNA-interaction (PIP) box [22]. p21 is recruited to UV-induced DNA lesions via interaction with PCNA, and a p21 mutant defective in binding to PCNA was resistant to degradation following UV treatment [10], [12].

To investigate the importance of these interactions for the MVM-dependent targeting of p21 by the CRL4^Cdt2 ligase we generated stable murine cell lines via lentivirus transduction that conditionally expressed FLAG-tagged wild-type or mutant p21 (p21^WT, p21^ΔDegron, p21^ΔPCNA, mutations shown in Figure 4A) in a doxycycline-responsive manner. As expected, MVM infection of a p21^WT expressing cell line resulted in degradation of the tagged p21 (Figure 4C and 4E, compare lanes 2 to 4), which could be prevented by treating cells with the proteasome inhibitor MG132 (Figure S4, panel A), or via siRNA knockdown of CRL4^Cdt2 components (Figure S4, panels B and C). These results suggested that the depletion of p21 in these cell lines occurred via a similar mechanism to that of the endogenous p21. The lysine at the p21 +4 position, 3′ to its PIP box, has been reported to be required for interaction of p21 with the CRL4^Cdt2 ligase complex via Cdt2. Mutation of the +3 to +5 amino acids KRR to AAA (p21^Δdegron) abolished p21 interaction with the CRL4^Cdt2 complex following its transient transfection as reflected by loss of interaction with DDB1 (Figure 4B, compare lanes 2 to 3). Murine cell lines that expressed the p21^Δdegron mutant were generated, and when infected with MVM, in contrast to cell lines expressing wild-type p21 (p21^WT), the p21^Δdegron protein was resistant to degradation (Figure 4C, compare lanes 6 and 8 to lanes 2 and 4). This suggested that MVM-induced p21 degradation required interaction of p21 with the CRL4^Cdt2 ubiquitin ligase complex.

The CRL4^Cdt2 ligase also requires PCNA as a cofactor for targeting of its substrates [13]. Mutation of the p21 PIP box (p21^ΔPCNA, Figure 4A) abrogated interaction with PCNA (Figure 4D, compare lane 3 to 2), and the p21^ΔPCNA protein expressed in murine cells was not degraded as efficiently as the wild-type protein (p21^WT) following infection (Figure 4E, compare lanes 8 and 6 to 4 and 2). These results suggested that PCNA-binding was also essential for effective CRL4^Cdt2 targeting of p21 during MVM infection.

Stable p21 that retains interaction with PCNA inhibits MVM replication

p21 interaction with PCNA has been shown to interfere with DNA polymerase δ−mediated cellular DNA replication [5], [6]. PCNA is an important cofactor for MVM replication; it contributes to MVM replication in vitro [8], and is recruited to APAR bodies during infection. Thus, we investigated whether MVM-dependent depletion of p21 during infection, mediated by the CRL4^Cdt2 E3 ubiquitin ligase, promoted efficient replication of the MVM genome by abrogating its inhibition of PCNA.

Unexpectedly, the p21^ΔDegron mutant, although stable, interacted poorly with PCNA for reasons not yet clear (data not shown). As a result, we could not use cell lines expressing this mutant to determine whether stabilized, p21 affected MVM replication via PCNA binding. Thus we generated a murine cell line conditionally expressing a mutant p21 in which all seven lysines in p21 were changed to arginine [p21^K7R, a similar mutation has been reported by others [23]]. Similar to the p21^ΔDegron mutant, the p21^K7R protein was resistant to degradation following MVM infection (Figure 5A, compare lanes 6 and 8 to 2 and 4), yet retained substantial interaction with PCNA in transient transfection assays (Figure 5B, compare lane 3 to lanes 2 and 4). Whereas induction of p21^WT expression for 8 hrs after infection had little effect on MVM replication (Figure 5C, compare lanes 1 and 2), p21^K7R expression reduced replication by up to 3 fold (Figure 5C, compare lanes 3 and 4). Importantly, in contrast, the p21^ΔPCNA mutant-expressing cell line, which expressed a stable p21 which no longer could interact with PCNA (Figure 4D and 4E), failed to inhibit MVM replication upon induction (Figure 5D, compare lanes 3 and 4). This was also the case with the p21^ΔDegron mutant-expressing cell lines (data not shown). Furthermore, cell lines expressing a mutant of p21^K7R in the PIP box-mutated background (p21^K7RΔPIP) (see Figure 5B, lane 4) also failed to inhibit MVM replication (Figure S5), demonstrating that absent PCNA binding, the K7R mutation itself had no deleterious effect on MVM replication. All the mutants tested were recruited to APAR bodies during MVM infection (Figure S6).

To confirm that the p21-PCNA interaction mediated the inhibitory role of p21 during infection, we made use of a previously described peptide containing 20 residues derived from sequences comprising the p21 PIP box [24] fused to a 16-mer penetratin motif to facilitate cellular entry [25]. A scrambled version of the p21 peptide linked to the penetratin peptide was used as control. Following application to murine cells, the wild-type (Figure 5E, lane 4), but not the scrambled version (Figure 5E, lane 3), disrupted the interaction between over-expressed FLAG-tagged p21 and endogenous PCNA, demonstrating that the peptide could competitively interact with PCNA. Subsequently, while treatment of cells with the scrambled peptide had no effect on MVM replication (Figure 5F, compare lanes 2 and 4), treatment of cells with the wild-type p21 PCNA-binding peptide significantly inhibited MVM replication (Figure 5F, compare lanes 2 and 3). These results indicated that a stabilized p21 interaction with PCNA was detrimental to viral replication. Thus, while interaction with PCNA was important for targeting p21 to the re-localized CRL4^Cdt2 ligase, efficient viral replication required subsequent depletion of p21 and consequent abrogation of its inhibition of PCNA.

Discussion

Parvoviruses and other small DNA viruses rely on host polymerases to replicate their genomes. How the replication machinery is exploited to sustain parvovirus replication in G2-arrested cells, which normally contain potentially inhibitory cellular proteins such as p21, is not fully understood. p21 levels are high in G1, low in S-phase, restored in G2 phase, and are regulated by proteasomal degradation during cell cycle progression [4]. We and others have previously reported that expression of the parvoviral NS1 protein leads to increases in p21 levels [2], [3]. Additionally, p53, a transcriptional activator of p21, is significantly up-regulated and activated throughout MVM infection. Remarkably however, while these signals lead to increased p21 RNA accumulation, p21 protein levels remain low throughout infection [2]. Here we have identified the mechanism by which p21 was degraded upon parvovirus infection, and identified the consequence of this for virus replication.

Degradation of p21 during S-phase and in response to DNA damaging agents such as UV treatment is programmed by the CRL4^Cdt2 ubiquitin ligase [13] and in this manuscript we have demonstrated that the same ligase targets p21 for degradation during MVM infection. The circumstances surrounding p21 degradation and the signals leading to it in the context of parvoviral infection, however, differ from how it occurs during S-phase. During MVM infection p21 loss persists for extended periods of time while virus is replicating in G2 arrested cells in the presence of high amounts of activated p53 and NS1 [2]. Additionally, whereas ATR activity is important for p21 degradation in response to various DDR-inducing agents [23], [26], the ATR substrate Chk1 is not activated during MVM replication (Adeyemi and Pintel, in preparation), suggesting that p21 degradation during infection may occur independently of ATR activity.

During MVM infection the CRL4^Cdt2 ligase is recruited to viral replication centers. Recently, the CRL4^Cdt2 ligase was shown to be recruited to cellular chromatin via direct PCNA interaction [21]. It is not yet clear whether similar mechanisms mediate CRL4^Cdt2 recruitment to MVM APAR bodies; however, it appears that PCNA recruitment to MVM chromatin may represent a critical step leading to viral hijacking of the CRL4^Cdt2 ligase. Neither interaction with PCNA nor the ligase was required for recruitment of p21 to replication centers, as all the mutants tested were relocalized to APAR bodies.

Stabilization of p21 via CRL4^Cdt2 depletion led to reduced MVM replication subsequent to the initiation of genome replication following S-phase entry. This is the first published demonstration of the requirement and re-localization to replication centers of a specific cellular ubiquitin ligase during autonomous parvovirus replication. p21 is a potent inhibitor of CDKs and PCNA. Exactly how stabilized p21 inhibited MVM replication is not fully clear; however, interaction with PCNA mediated its inhibitory role. Using inducible cell lines expressing wild-type and mutant p21 proteins we demonstrated that p21 degradation during infection required motifs that mediate interaction with both Cdt2 and PCNA. PCNA is recruited to MVM chromatin [18] and is essential for parvovirus replication [27], [28]. Overexpression of a stable mutant p21 that retained interaction with PCNA, but could not be targeted for degradation by the CRL4^Cdt2 ligase, led to reduced virus replication. Stable p21 mutants unable to bind PCNA did not affect MVM replication, indicating that other potential inhibitory functions of p21, such as CDK binding and promoter repression, were not detrimental to viral replication absent PCNA interaction. Additionally, introduction of a p21-derived peptide which maintained PCNA interaction abolished viral replication. Thus, p21 interaction with PCNA (and Cdt2) was necessary for targeting of p21 to the co-localized CRL4^Cdt2 ligase, and its subsequent depletion also prevented its sustained interaction with PCNA that would otherwise be inhibitory to viral replication. Although our earlier work had suggested that Cdk2 activity was required for MVM replication and that p21 degradation might be necessary to prevent inhibition of CDKs [2], we have recently shown that Chk2 activation during MVM infection results in CDC25A degradation leading to partial CDK2 inhibition, independent of p21 [29]. Thus, while some level of CDK activity is required for MVM replication, prevention of CDK inhibition is not likely to be the critical reason for p21 degradation.

p21 has been shown to inhibit MVM replication in vitro, and this effect was shown to be overcome by the addition of increasing amounts of PCNA [8]. p21 binds to PCNA via its PIP box, a conserved motif shared by substrates of the CRL4^Cdt2 ligase, as well as cellular proteins such as DNA polymerase δ, that are essential for replication but escape ubiquitin targeting due to the absence of a PIP degron [13]. The p68 subunit of DNA polymerase δ binds to PCNA via a similar hydrophobic pocket recognized by the p21 PIP box. Thus, while the mechanism of p21 inhibition of the DNA polymerase δ/PCNA complex has not been clearly resolved, a stabilized high-affinity interaction of p21 with the DNA polymerase δ binding pocket within PCNA could directly inhibit viral replication by competing with DNA polymerase δ for binding to its cofactor [30].

Due to its myriad effects on cell cycle progression and cancer, several viruses make use of different strategies to inactivate p21 during infection. For example, several oncogenic DNA viruses indirectly inhibit p21 by targeting p53 [31]. Additionally, papillomaviruses HPV E7 proteins sequester p21 thereby preventing its interaction with PCNA and CDKs [32], [33]. KSHV encodes a microRNA that down-regulates p21 in order to prevent cell cycle arrest [34]. p21 restricts HIV in myeloid cells of certain patients [35], [36], and a recent report demonstrated that this restriction may occur via ribonucleotide reductase-2 repression resulting in inhibition of dNTP biosynthesis [37]. Thus, cellular components required for viral genomic replication, such as dNTPs and polymerase cofactors, may be critically dependent on p21 depletion. Our findings present first a novel mechanism of p21 antagonism by a virus, namely, the use of a cellular E3 ligase recruited to viral replication factories. Further, we shown that efficient virus replication depends on the depletion of p21 to prevent its inhibition of PCNA. PCNA is an important cofactor for the replication of several DNA viruses. PCNA-mediated degradation of p21 in the context of viral infection may emerge as an important paradigm for allowing sustained viral replication of DNA polymerase-δ dependent viruses in infected cells.

Materials and Methods

Cell lines and drug treatments

Murine A9 and Human 293T cell lines were propagated as previously described [38], [39]. Stable doxycycline-inducible A9 cell lines were generated by infection of A9 cells with pseudotyped virus using the pINDUCER20 lentiviral system [40]. Cell lines were selected with 800 µg/ml of geneticin (Gibco) and maintained like regular A9s except for addition of geneticin. A9 cells were parasynchronized in G0 by isoleucine deprivation as previously described [1]. pINDUCER20 lentiviral transformed cell lines were induced with 500 ng/mL doxycycline hydrochloride (MP Biomedical). MG132 (Calbiochem) was added at a final concentration of 10 µM.

Viruses and infections

Wild-type MVMp was propagated and titered in murine A9 cell lines as described [1]. Pseudotyped viruses were generated by co-transfecting equal concentrations of HIV Gag/Pol, VSV-G and pINDUCER20 plasmids into 293T cells. Supernatants were collected and used to infect A9 cells.

Plasmids and DNA transfection

Murine wild-type p21 cDNA (Origene) was cloned into p3XFLAG-CMV 7.1 (Sigma). Additionally, p21 was tagged with a 3× HA tag by PCR mutagenesis. p21^K7R was generated by PCR mutagenesis. p21^ΔPCNA, p21^Δdegron and p21^K7RΔPIP (Q139A, L142A, F145A, Y146A) mutants were generated by site-directed mutagenesis (Agilent). FLAG and HA-tagged wild-type and mutant p21 were cloned into pDONR221 (Invitrogen) and pINDUCER20 using BP and LR clonase kits (Invitrogen) respectively. pINDUCER reagents were a gift from Guang Hu at NIH/NIEHS. DNA transfection was performed using LipoD293 (Signagen) or Lipofectamine (Invitrogen).

RNA interference

A9 cells were transfected twice with 40 nM Control siRNA (Qiagen #1022076), Cdt2 (DTL) siRNA (Dharmacon #L-045921-01-0005), Cul4A (Dharmacon #L-052208-00-0005) or DDB1 siRNA (Dharmacon #L-043146-01-0005). siRNA transfections were performed using HiPerfect (Qiagen).

Peptides

Wild-type p21 (KRRQTSMTDFYHSKRRLIFSRQIKIWFQNRRMKWKK) and scrambled p21 (KSTARHTKLSAQRYIRSFARRQIKIWFQNRRMKWKK) were purchased from Peptide 2.0 Inc. The peptides consist of p21-derived or scrambled sequences fused to penetratin – a 16 amino acid nuclear internalization sequence derived from the Antennapedia homeodomain [25]. Peptides were added to cells at 25 µM.

Antibodies

Commercially available antibodies used in this study were obtained from Bethyl (Cdc20, cat# A301-180A), Cell Signaling (Set8, cat# 2996S), Invitrogen (DDB1, cat# 399901; p21, cat# 556430), Millipore (Cdt1, cat# 06-1295; PCNA, cat# CBL407), Novus (Cullin 4A, cat# NB100-2267), Pierce (Actin, cat# MA515739), Sigma (FLAG, cat# F1804; HA, cat# H9658; Tubulin, cat# T4026), and Upstate (Cyclin A, cat# 06-138). Cdt2 antibody was a generous gift from Anindya Dutta (University of Virginia). All secondary antibodies were from Invitrogen. NS1 (CE10) and NS1/2 (M55) were previously described [1].

Immunoblot analysis

Immunoblots were performed as described previously [1]. Protein concentrations were quantified by Bradford assay and equal amounts of lysates were run.

Co-immunoprecipitation

FLAG and HA-tagged p21 was immunoprecipitated from 293T cells as previously described [12]. FLAG beads (Sigma) and HA antibodies were used for IP.

Immunofluorescence (IF)

For IF, A9 cells were grown on glass coverslips in 35 mm dishes and infected with MVMp at an MOI of 10. At 24 hr pi, cells were washed twice with cold phosphate-buffered saline (PBS) solution and then with cytoskeleton buffer [10 mM piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES), pH 6.8, 100 mM NaCl, 300 mM sucrose, 1 mM MgCl₂, 1 mM EGTA]. Afterwards cells were pre-extracted for in cytoskeletal buffer containing 0.5% Triton X-100, protease and phosphatase inhibitors for 3 minutes on ice, washed, fixed with 4% paraformaldehyde and stained for the indicated proteins. Nuclei were visualized by staining with To-Pro-3 (Invitrogen). The coverslips were mounted in Fluoromount-G (Southern Biotech) and images were acquired using a Zeiss LSM 510 META confocal microscope. All images were captured using an objective of 63×.

Analysis of viral DNA

Southern blots were carried out as previously described [38], using whole MVM genome probes. Loading of DNA samples was normalized using a nanodrop spectrophotometer. This procedure was verified using probes on Southern blots against mitochondrial DNA. Unless otherwise indicated, infections were carried out at a low MOI in order to maximize effects of siRNA knockdowns and overexpression analyses on viral replication. Representative Southern Blots are shown; quantifications in the text reflect multiple knockdown experiments.

Cell cycle analysis

A9 cells were fixed in 70% ethanol overnight at 4°C. Cells were then pelleted, washed in PBS and resuspended in PBS containing 0.2 mg/ml RNAse A for 1 hr at 37°C, then propidium iodide was added at 40 µg/ml. Flow cytometry was performed using FACScan (BD biosciences). Data were analyzed using Summit software (Beckman Coulter).

Supporting Information

Attachment 1

Attachment 2

Attachment 3

Attachment 4

Attachment 5

Attachment 6