Interaction of phospholipid scramblase 1 with the Epstein-Barr virus protein BZLF1 represses BZLF1-mediated lytic gene transcription

Human phospholipid scramblase 1 (PLSCR1) is strongly expressed in response to interferon (IFN) treatment and viral infection, and PLSCR1 has been suggested to play an important role in IFN-dependent antiviral responses. In this study, we showed that the basal expression of PLSCR1 was significantly elevated in Epstein-Barr virus (EBV)–infected nasopharyngeal carcinoma (NPC). PLSCR1 was observed to directly interact with the EBV immediate-early transactivator BZLF1 in vitro and in vivo, and this interaction repressed the BZLF1-mediated transactivation of an EBV lytic BMRF1 promoter construct. In addition, PLSCR1 expression decreased the BZLF1-mediated up-regulation of lytic BMRF1 mRNA and protein expression in WT and PLSCR1-knockout EBV-infected NPC cells. Furthermore, we showed that PLSCR1 represses the interaction between BZLF1 and CREB-binding protein (CBP), which enhances the BZLF1-mediated transactivation of EBV lytic promoters. These results reveal for the first time that PLSCR1 specifically interacts with BZLF1 and negatively regulates its transcriptional regulatory activity by preventing the formation of the BZLF1-CBP complex. This interaction may contribute to the establishment of latent EBV infection in EBV-infected NPC cells.

Epstein-Barr virus (EBV) 2 is linked to the development of several malignancies, including endemic Burkitt's lymphoma (BL), posttransplantation lymphoma, Hodgkin's disease, nasopharyngeal carcinoma (NPC), and some types of gastric cancers (1,2). EBV can cause both latent and lytic infections, typically establishing latent infections in B cells and lytic infections in epithelial cells (3)(4)(5). However, the latent form of EBV can be detected in EBV-associated malignancies of both lymphoid and epithelial origin (3)(4)(5). B cell-specific transcription factors have been reported to promote the latent EBV infection of B cells by directly interacting with and inhibiting the function of BZLF1 (also termed Z, Zta, or ZEBRA), an EBV-encoded immediate-early lytic gene product (6,7). The establishment of latent infections and/or the inhibition of lytic infections may play an important role in the development of EBV-associated epithelial malignancies. However, the specific cellular factors required for the negative regulation of lytic infections in EBV-associated epithelial malignancies remain poorly understood.
BZLF1 plays a crucial role in the switch from latent to lytic EBV infections (8,9). BZLF1 contains an N-terminal transactivation domain and a C-terminal basic leucine zipper (bZIP) motif. As a bZIP transcription factor that is homologous to c-Jun and c-Fos, BZLF1 binds as a homodimer to consensus AP-1 sites and AP-1-like motifs known as BZLF1-responsive elements (10,11). The binding of BZLF1 to BZLF1-responsive elements and AP-1 sites results in the transactivation of viral and specific cellular promoters, leading to an ordered cascade of viral gene expression in which the expression of early genes involved in DNA replication and metabolism is followed by that of late genes encoding viral structural proteins (8,11).
Human phospholipid scramblase 1 (PLSCR1) was identified as an enzyme involved in the calcium-dependent, nonspecific, rapid redistribution of phospholipids (12). However, further studies have suggested that PLSCR1 is not involved in phospholipid redistribution when a perturbation of plasma membrane asymmetry is required following cell activation or apoptosis (13,14). Furthermore, the results of recent studies suggest that PLSCR1 inhibits tumorigenesis, promotes apoptosis, and facilitates the differentiation of myeloid cells through its interaction with several signaling molecules (15,16). Human PLSCR1 expression is robustly induced in response to IFN treatment and viral infection (17)(18)(19). In addition, PLSCR1 has been reported to enhance the IFN-dependent induction of IFNstimulated gene expression and antiviral activity, and PLSCR1 has been suggested to play an important role in IFN-dependent antiviral responses (18). However, PLSCR1 is not involved in the IFN-␣-mediated induction of IFN-stimulated gene 15 expression in our CRISPR/Cas9-generated PLSCR1-KO HEK-293 cells (Fig. S1), and the precise mechanisms of PLSCR1mediated antiviral activity remain unclear. Recently, we and other groups reported that PLSCR1 directly interacts with and affects the function of several viral proteins (20 -23), which may be important for PLSCR1-mediated antiviral activity.
In this study, we showed that the basal expression of PLSCR1 is significantly elevated in EBV-infected NPC cells. PLSCR1 directly and specifically interacts with the bZIP motif of BZLF1 through its 1-163 and 160 -250 amino acid regions. PLSCR1 overexpression was observed to repress the BZLF1-mediated transactivation of an EBV early BMRF1 promoter reporter construct in an interaction-dependent manner. In addition, PLSCR1 overexpression decreased the BZLF1-mediated expression of lytic BMRF1 at the mRNA and protein levels in the WT and PLSCR1-KO EBV-infected NPC cell line C666-1. Furthermore, PLSCR1 was observed to repress the interaction between BZLF1 and the transcriptional co-activator CREB-binding protein (CBP), which is required for the efficient transactivation of EBV early promoters by BZLF1.
These results reveal for the first time that PLSCR1 specifically interacts with BZLF1 and negatively regulates its transcriptional regulatory activity by preventing its interaction with CBP. This interaction may contribute to the establishment of latent EBV infection in EBV-infected NPC cells.

Basal PLSCR1 expression is significantly elevated in EBV-infected NPC cells
Human PLSCR1 expression has been shown to be induced in response to IFN treatment and viral infection (17,18). The expression of PLSCR1 was determined in the presence and absence of IFN treatment in the EBV-infected NPC cell line C666-1 and in EBV-negative epithelial cells. The basal expression of PLSCR1 was significantly lower in HeLa, SW480, and MCF-7 cells in the absence of IFN treatment.
However, PLSCR1 expression was significantly induced in these cell lines in the presence of IFN-␣-2b (Fig. 1A, lanes 3, 4, and 7-10), consistent with our previous report (21). Interestingly, the basal expression of PLSCR1 was robustly higher in C666-1 and A431 cells in the absence of IFN treatment (Fig. 1A, lanes 1 and 5). In addition, the expression of PLSCR1 was weakly induced by the IFN treatment in these cell lines (Fig. 1A, lanes 1, 2, 5, and 6).
Next, to assess whether the basal expression of PLSCR1 was also increased in EBV-infected B cells, we evaluated PLSCR1 expression in EBV-negative and EBV-positive B cell lines. Consistent with previous observations of other lymphoid cells (21,24), the basal expression of PLSCR1 in EBV-negative Burkitt's lymphoma BJAB cells was markedly high, similar to that observed in the C666-1 cells, in the absence of IFN. In contrast with the results obtained in C666-1 cells, PLSCR1 expression was significantly induced by the IFN treatment in BJAB cells (Fig. 1B, lanes 1-3). Notably, PLSCR1 expression was difficult to detect in all EBV-positive B cell lines tested in the absence of IFN treatment. However, PLSCR1 expression was significantly induced by the IFN treatment, except in B95-8 cells (Fig. 1B, lanes 4 -11). Although B95-8 cells originate from marmoset B cells, the sequences of human PLSCR1 and marmoset PLSCR1 are highly homologous, with 90% identity and 95% similarity, and marmoset PLSCR1 is likely to be detected by the anti-human PLSCR1 polyclonal antibody. However, human IFN-␣-2b may be ineffective at inducing IFN signaling in marmoset B cells.
To confirm whether PLSCR1 expression was also induced in other EBV-infected epithelial cells, we analyzed the basal PLSCR1 levels in EBV-infected NPC xenograft C15 and C17 tumors. The basal expression levels of PLSCR1 were signifi-

Figure 1. PLSCR1 expression is highly induced in EBV-infected NPC cells in the absence of IFN-␣-2b treatment.
A-C, cells were treated with or without 3000 units/ml of IFN-␣-2b for 16 h. Total cell lysates were prepared using RIPA buffer (Sigma), and a total of 15 g (A), 8 g (B), or 10 g (C) of total cell lysates were subjected to SDS-PAGE. Immunoblotting (IB) was performed using an anti-PLSCR1 antibody for endogenous PLSCR1 or an anti-actin antibody for endogenous actin. LCL, lymphoblastoid cell line.

PLSCR1 directly interacts with the EBV lytic transactivator BZLF1
PLSCR1 has been reported to repress the functions of viral transactivators, including the human T-cell leukemia virus (HTLV)-1 Tax, HIV-1 Tat, and human hepatitis B virus HBx proteins, through direct interaction (21,22,25). It is possible that PLSCR1 also interacts with and negatively regulates the EBV-encoding transactivator BZLF1, which stimulates the transcription of genes that causes the switch from latent to lytic EBV infection. To determine whether PLSCR1 directly interacts with BZLF1, full-length PLSCR1 encoding amino acids 1-318 was tagged with the trxA and S epitopes (pET-S-PLSCR1) ( Fig. 2A), and a BZLF1 truncation encoding amino acids 3-245 was tagged with a 3ϫFLAG epitope (pTE-3FG-BZLF1) (Fig. 3A). These constructs were expressed in Escherichia coli, after which the bacterial cell lysates were mixed, and pulldown assays were performed using S-protein beads to precipitate PLSCR1. Immunoblot analyses of the PLSCR1-con-taining complexes revealed that BZLF1 was efficiently co-precipitated with PLSCR1 (Fig. 2B, lane 5), whereas the empty vector-transformed lysates precipitated only trace amounts of PLSCR1 (Fig. 2B, lane 4). These results indicated that PLSCR1 directly interacts with BZLF1 in vitro.
PLSCR1 contains several functional domains, including three predicted intrinsically disordered (ID) regions (21) and the N-terminal-and central ID region-containing domains involved in interacting with target proteins (21,22,26). To identify which PLSCR1 regions are involved in this interaction in vivo, S epitope-tagged full-length PLSCR1 (S-PLSCR1) or its truncated mutants encoding amino acids 1-163 (S-PLSCR1(1-163)) or 160 -250 (S-PLSCR1(160 -250)) were expressed in HEK-293 cells together with a 3ϫFLAG epitope-tagged BZLF1 encoding amino acids 3-245 (3FG-BZLF1). Immunoblot analysis of the total cell lysates indicated that all constructs were expressed at similar levels ( Fig. 2C, lanes 2-4). However, immunoblot analysis of the complexes containing PLSCR1 or its truncated mutants revealed that BZLF1 was efficiently coprecipitated with PLSCR1 (Fig. 2C, lane 6) but weakly co-precipitated with PLSCR1(1-163) and PLSCR1(160 -250) (Fig. 2C,  lanes 7 and 8). These observations revealed that PLSCR1 contains two BZLF1-binding regions and that the amino acid regions 1-163 and 160 -250 of PLSCR1 are sufficient for this interaction. A, schematic representation of PLSCR1 and its truncated mutant expression constructs containing epitope tags at the N termini. The PLSCR1 structural motifs have been described previously (21) and annotated as follows: Palmitoylation, palmitoylation sites; NLS, nuclear localization signal; TM, transmembrane domain. B, bacterial lysates prepared in CHAPS lysis buffer were mixed and incubated with S-protein beads to precipitate PLSCR1 as follows: 100 g of pET32a(ϩ) vector or pET-S-PLSCR1 lysate and 30 g of pTE-3FG-BZLF1 lysate were used. Following pulldowns, 0.8 g of total bacterial lysates and the precipitated complexes were divided into two portions and subjected to SDS-PAGE. Immunoblotting was performed using an anti-S-tag antibody for PLSCR1 or an anti-FLAG M2 antibody for BZLF1. C, HEK-293 cells were transfected with 1 g of pcDNA3 or 3FG-BZLF1 and 2 g of pcDNA3, 0.7 g of S-PLSCR1, 2 g of S-PLSCR1(1-163) or 2 g of S-PLSCR1(160 -250). The total amount transfected was equalized by adding pcDNA3. A total of 400 g of total cell lysate prepared in CHAPS lysis buffer was incubated with S-protein beads to precipitate PLSCR1 and its mutants. Following pulldowns, 20 g of total cell lysate and the precipitated complexes were divided into two portions and subjected to SDS-PAGE. Immunoblotting was performed using an anti-FLAG M2 antibody for BZLF1 or an S-tag antibody for PLSCR1 and its mutants.
To further confirm this interaction, 3FG-BZLF1 and a 3ϫFLAG epitope-tagged C-terminal bZIP motif of BZLF1 encoding amino acids 160 -245 (3FG-BZLF1(160 -245)) ( Fig.  3A) were expressed in HEK-293 cells and assessed for their ability to bind to PLSCR1. Immunoblot analysis of the immunoprecipitated complexes containing BZLF1 or BZLF1 bZip motif revealed that PLSCR1 was co-precipitated with both BZLF1 and BZLF1(160 -245) (Fig. 3C, lanes 5 and 6). These results revealed that the C-terminal bZIP motif-containing region of BZLF1 is sufficient for its interaction with PLSCR1 in vivo.

PLSCR1 does not affect the nuclear localization of BZLF1
PLSCR1 is localized at the plasma membrane and in various internal membrane pools. However, PLSCR1 also localizes to  , or 3FG-BZLF1(3-169). A total of 300 g of total cell lysate prepared in CHAPS lysis buffer was incubated with anti-FLAG M2 beads to precipitate BZLF1 and its mutants. Following immunoprecipitation, 15 g of total cell lysate and the precipitated complexes were divided into two portions and subjected to SDS-PAGE. Immunoblotting was performed using an anti-myc antibody for PLSCR1 or an anti-FLAG M2 antibody for BZLF1 and its mutants. IP, immunoprecipitation. C, HEK-293 cells were transfected with 3.5 g of 3M-PLSCR1 and 1.5 g of pcDNA3, 3FG-BZLF1, or 3FG-BZLF1(160 -245). A total of 300 g of total cell lysate prepared in CHAPS lysis buffer was incubated with anti-FLAG M2 beads to precipitate BZLF1 and its mutants. Following immunoprecipitation, 20 g of total cell lysate and the precipitated complexes were divided into two portions and subjected to SDS-PAGE. Immunoblotting was performed using an anti-myc antibody for PLSCR1 or an anti-FLAG M2 antibody for BZLF1 and mutant.

PLSCR1 represses EBV BZLF1-dependent lytic transcription
the nucleus in response to cytokine stimulation (28), and BZLF1 has been reported to be primarily localized in the nucleus (29). To further investigate the functional relevance of PLSCR1-BZLF1 complex formation, the intracellular localization of BZLF1 and PLSCR1 was assessed through immunofluorescence analysis. Consistent with previous observations, BZLF1 was primarily detected when PLSCR1 was absent (Fig.  4A, top panel), whereas PLSCR1 was detected throughout the cytoplasm and nucleus (Fig. 4A, bottom panel) (21,22). This nuclear localization pattern of BZLF1 was not affected by the presence of PLSCR1 (Fig. 4B).
To confirm this nuclear co-localization of BZLF1 and PLSCR1 using subcellular fractionation, we expressed 3FG-BZLF1 and 3M-PLSCR1 in PLSCR1-KO HeLa (HeLa-PLSKO) cells. Immunoblotting of the whole cell lysates and the cytoplasmic and nuclear fractions indicated that the glyceraldehyde-3phosphate dehydrogenase (G3PDH) and histone H3 were barely detected in the nuclear and cytoplasmic fractions, respectively, which demonstrated the accuracy of the fractionation (Fig. 4C, third panel from the top and bottom panel). PLSCR1 was efficiently detected in the cytoplasmic and nuclear fractions, and the levels of PLSCR1 in the cytoplasmic fraction were slightly higher than levels in the nuclear fraction in the absence and presence of BZLF1 (Fig. 4C, lanes 1-3 and 7-9, second panel from top). In the absence of PLSCR1, most BZLF1 was localized in the nuclear fraction, and this nuclear localization of BZLF1 was not affected by the expression of PLSCR1 (Fig. 4C, lanes 4 -6 and 7-9, top panel). These data confirmed that this interaction does not affect the nuclear localization of BZLF1 and that PLSCR1 and BZLF1 co-localize in the nucleus in vivo.

PLSCR1 represses BZLF1-mediated transactivation in an interaction-dependent manner
BZLF1 binds as a homodimer to target DNA through its C-terminal bZIP motif, which interacts with PLSCR1. To deter-mine whether PLSCR1 affects BZLF1-mediated transactivation of early gene promoters, we measured the activity of a luciferase-reporter plasmid under the control of the lytic BMRF1 promoter (pBMRF1pro-4.10) in the presence of 3FG-BZLF1 and S-PLSCR1 in COS-1 cells. In the presence of BZLF1, luciferase activity was increased ϳ18-fold in the absence of PLSCR1 expression (Fig. 5A, lane 5). Although PLSCR1 slightly decreased the luciferase activity in the absence of BZLF1, PLSCR1 overexpression efficiently decreased this BZLF1-induced luciferase activity in a dose-dependent manner (Fig. 5A, lanes 6 -8).
To determine whether the binding of PLSCR1 to BZLF1 is sufficient for this repression, we measured the luciferase activity of pBMRF1pro-4.10 in COS-1 and HEK-293 cells containing empty vector or 3FG-BZLF1 and either 3M-PLSCR1, 3M-PLSCR1(1-163), or 3M-PLSCR1(160 -250). In COS-1 cells, BZLF1 expression increased the luciferase activity ϳ3-fold compared with that observed in the empty vectortransfected cells (Fig. 5B, lane 2). PLSCR1 and PLSCR1(1-163) efficiently decreased this BZLF1-induced luciferase activity to near basal levels (Fig. 5B, lanes 3-6), whereas this activity was reduced to ϳ60% of that observed in the empty vectortransfected cells in the presence of PLSCR1(160 -250), even when an excess amount of BZLF1 was expressed by transfecting four times more DNA (Fig. 5B, lanes 7 and 8). In HEK-293 cells, BZLF1 expression increased the luciferase activity ϳ8-fold compared with that observed in the empty vector-transfected cells (Fig. 5B, lane 10). PLSCR1 efficiently decreased this BZLF1-induced luciferase activity to lower than 45% of that observed in the empty vector-transfected cells (Fig. 5B, lane  12), whereas this activity was reduced to ϳ60 and 75% of that observed in the empty vector-transfected cells in the presence of PLSCR1(1-163) and PLSCR1(160 -250), respectively, even when an excess amount of BZLF1 was expressed by transfecting four times more DNA (Fig. 5B, lanes 14 and 16). These obser- FLAG M2-and myc-stained fluorescence images were acquired using a FV500 CLSM instrument and merged using FLUOVIEW. Yellow areas indicate the co-localization of the two proteins. Scale bar, 20 m. C, HeLa-PLSKO cells were transfected with 150 ng of 3FG-BZLF1 and/or 2 g of 3M-PLSCR1. The total amount of plasmid transfected was equalized by adding pcDNA3. Whole cell extractions and subcellular fractionations were performed as described in "Experimental procedures." Then, 5 g of the whole cell lysate and the cytoplasmic and nuclear fractions were subjected to SDS-PAGE. Immunoblotting (IB) was performed with an anti-FLAG M2 antibody for BZLF1, an anti-myc antibody for PLSCR1, an anti-G3PDH antibody for G3PDH (as a cytoplasmic marker) and an anti-histone H3 antibody for histone H3 (as a nucleus marker). W, whole cell lysate; C, cytoplasmic fraction; N, nuclear fraction.

PLSCR1 represses EBV BZLF1-dependent lytic transcription
vations revealed that the BZLF1-binding regions of PLSCR1 (amino acids 1-163 and 160 -250) are sufficient to repress the BZLF1-dependent transactivation of the BMRF1 promoter. However, the N-terminal amino acids 1-163 of PLSCR1, which have been suggested to contain a long ID region, produce a stronger inhibition than amino acids 160 -250.

PLSCR1 expression negatively regulates lytic gene expression in EBV-infected NPC cells
To investigate the PLSCR1-mediated repression of BZLF1dependent transactivation at the endogenous mRNA level in EBV-infected NPC cells, lytic gene expression was induced by transfecting C666-1 cells with 3FG-BZLF1 in the presence and absence of 3M-PLSCR1. Immunoblot analysis of the total cell lysates indicated that all constructs were expressed at similar levels ( Fig. 6A, lanes 1-4, top panel). The results of semiquantitative RT-PCR assays demonstrated that the levels of G3PDH mRNA were almost identical in the presence and absence of BZLF1 and PLSCR1 (Fig. 6A, bottom panel). BMRF1, an early gene in the EBV lytic cycle, was efficiently transcribed in the presence of BZLF1 (Fig. 6A, lane 3, third panel from the top). However, the levels of BMRF1 mRNA were significantly decreased in the presence of PLSCR1 to ϳ50% of that observed in the empty vector-transfected cells (Fig. 6A, lanes 3 and 4,  third panel from the top). These results indicated that PLSCR1 negatively regulates the BZLF1-dependent transactivation of the lytic gene in EBV-infected NPC cells.
Next, we assessed whether a high level of PLSCR1 expression represses endogenous BZLF1-dependent transactivation in EBV-infected NPC cells. PLSCR1 expression was decreased in C666-1 cells transfected with nontarget control or PLSCR1specific shRNA plasmids (21). In contrast, immunoblot analysis of total cell lysates revealed that PLSCR1 was highly expressed in nontarget shRNA-transfected cells and that PLSCR1-specific shRNA transfection significantly repressed PLSCR1 expression (Fig. 6B, top panel). The results of semiquantitative RT-PCR assays demonstrated that the levels of G3PDH mRNA in the PLSCR1-specific shRNA-transfected cells were slightly lower than those observed in the nontarget shRNA-transfected cells (Fig. 6B, bottom panel). However, the levels of BMRF1 mRNA were increased 2.5-fold in the PLSCR1-specific shRNAtransfected cells compared with that observed in the nontarget shRNA-transfected cells (Fig. 6B, third panel from the top). These results indicated that a high level of PLSCR1 expression represses the BZLF1-dependent transactivation of the lytic genes in EBV-infected NPC cells.
To confirm this PLSCR1-mediated repression of BZLF1-dependent transactivation at the protein level, BMRF1 protein expression was induced in C666-1 cells by transfection with 3FG-BZLF1 in the presence and absence of PLSCR1-specific shRNA transfection. Immunoblot analysis of the total cell lysates indicated that PLSCR1-specific shRNA transfection efficiently decreased endogenous PLSCR1 expression (Fig. 6C,  lanes 1-4, top panel) and that the levels of BZLF1 were almost

PLSCR1 represses EBV BZLF1-dependent lytic transcription
identical in the presence and absence of PLSCR1-specific shRNA (Fig. 6C, lanes 3 and 4, second panel from the top). Unfortunately, we were unable to detect endogenous BMRF1 protein expression, although the levels of BZLF1-induced BMRF1 protein were significantly increased in the PLSCR1specific shRNA-transfected cells (Fig. 6C, lanes 3 and 4, third  panel from the top). Taken together, these observations revealed that PLSCR1 represses BZLF1-dependent lytic gene expression in EBV-infected NPC cells.

PLSCR1 expression negatively regulates lytic gene expression in PLSCR1-KO EBV-infected NPC cells
To confirm the PLSCR1-mediated repression of BZLF1-dependent transactivation of lytic gene expression in PLSCR1-KO EBV-infected NPC cells, we generated the PLSCR1-KO C666-1 cells by using the CRISPR/Cas9 genome editing system. In contrast with parental C666-1 cells, PLSCR1 protein expression was barely detected in our PLSCR1-KO C666-1 cells (Fig. 7A,  top panel). The results of semiquantitative RT-PCR assays demonstrated that the levels of G3PDH mRNA were almost identical in C666-1 and PLSCR1-KO C666-1 cells (Fig. 7A, bottom panel). However, the levels of BMRF1 mRNA increased approximately 2-to 4-fold in the PLSCR1-KO C666-1 cells compared with parental C666-1 cells (Fig. 7A, third panel from top).
Next, to examine whether the high levels of BMRF1 mRNA in PLSCR1-KO C666-1 cells were caused by the KO of the PLSCR1 gene, PLSCR1 was overexpressed in PLSCR1-KO C666-1 cells. The results of immunoblot analysis revealed that the epitope-tagged and untagged forms of PLSCR1 were efficiently expressed in cells transfected with 3M-PLSCR1 (Fig. 7B,  top panel). The results of semiquantitative RT-PCR assays demonstrated that the levels of G3PDH mRNA in the 3M-PLSCR1-transfected cells were slightly lower than those observed in the empty vector-transfected cells (Fig. 6B, middle  panel). However, the levels of BMRF1 mRNA were significantly decreased in the PLSCR1-overexpressed cells to ϳ40% of that observed in the empty vector-transfected cells. These results indicated that PLSCR expression decreased the levels of lytic BMRF1 mRNA in PLSCR1-KO C666-1 cells.
To confirm the PLSCR1-mediated repression of BZLF1-dependent transactivation at the protein level in PLSCR1-KO C666-1 cells, lytic BMRF1 protein expression was induced by transfecting cells with 3FG-BZLF1 in the presence or absence of 3M-PLSCR1. Consistent with previous observations, immunoblot analysis of the total cell lysates indicated that 3FG-BZLF1 and with 3 g of pcDNA3 or 3M-PLSCR1. After 48 h of transfection, total cell lysates and total RNA were prepared. Total cell lysates (5 g) were subjected to SDS-PAGE, and immunoblotting was performed using an anti-FLAG M2 antibody for BZLF1 or an anti-myc antibody for PLSCR1. PCR was performed as follows: 21 cycles of amplification for G3PDH using Hot Start TaqDNA Polymerase (HSTaq) (New England Biolabs) and 23 cycles of amplification for BMRF1 using KAPATaq EXtra HotStart ReadyMix with Dye (KAPATaqEx) (Kapa Biosystems). The intensities of the RT-PCR products were quantified using ImageJ, and the levels of G3PDH or BMRF1 mRNA from the empty vector and 3FG-BZLF1-transfected cells were defined as 1.0. ND, not detected. B, C666-1 cells were transfected with 3.5 g of a nontarget shRNA control or PLSCR1-specific shRNA. After 72 h of transfection, total cell lysates and total RNA were prepared. Total cell lysates (8 g) were subjected to SDS-PAGE, and immunoblotting was performed using an anti-PLSCR1 antibody for PLSCR1 or an anti-actin antibody for actin. PCR was performed as follows: 20 cycles of amplification for G3PDH using HSTaq and 35 cycles of amplification for BMRF1 using KAPATaqEx. The intensities of the immunoblotted bands and the RT-PCR products were quantified using ImageJ, and the intensities of the products obtained using nontarget shRNA-transfected cells were defined as 1.0. C, C666-1 cells were transfected with 5 g of a nontarget shRNA control or PLSCR1-specific shRNA. After 36 h of transfection, shRNA-transfected C666-1 cells were transfected with 1 g of pcDNA3 or 3FG-BZLF1. After 48 h of the second transfection, total cell lysates were prepared. Total cell lysates (5 g for BZLF1, PLSCR1, and actin and 15 g for BMRF1) were subjected to SDS-PAGE, and immunoblotting was performed using an anti-FLAG M2 antibody for BZLF1, an anti-PLSCR1 for endogenous PLSCR1, an anti-actin antibody for endogenous actin, or an anti-BMRF1 antibody for endogenous BMRF1. The intensities of the immunoblotted bands were quantified using ImageJ, and the intensities of the products obtained using 3FG-BZLF1 and nontarget shRNA-transfected cells were defined as 1.0.

PLSCR1 represses EBV BZLF1-dependent lytic transcription
3M-PLSCR1 transfection efficiently expressed the epitopetagged and untagged forms of PLSCR1 (Fig. 7C, lanes 2 and 4,  second panel from the top). The levels of BZLF1 expression were almost identical in the absence and presence of 3M-PLSCR1 (Fig. 7C, lanes 3 and 4, top panel), whereas the levels of BZLF1-induced BMRF1 protein were significantly decreased in the PLSCR1-overexpressed cells compared with the empty vector-transfected cells (Fig. 7C, lanes 3 and 4, third panel from the top). Taken together, these observations confirmed that PLSCR1 expression also represses BZLF1-dependent lytic gene expression in PLSCR1-KO EBV-infected NPC cells.
To confirm this observation in other PLSCR1-KO cells, PLSCR1-KO HeLa cells stably expressing empty vector or 3M-PLSCR1 were transfected with 3FG-CBP(1-721) and S-BZLF1. Immunoblot analysis of the total cell lysates indicated that all constructs were expressed at similar levels in the presence and absence of PLSCR1 expression (Fig. 8B, lanes 1-4). However, pulldown analysis using S-protein beads to precipitate BZLF1 revealed that PLSCR1 expression significantly decreased CBP(1-721) co-precipitation with BZLF1 (Fig. 8B,  lanes 7 and 8). These results revealed that PLSCR1 represses the interaction between BZLF1 and CBP in vivo. to SDS-PAGE, and immunoblotting was performed using an anti-PLSCR1 antibody for PLSCR1 or an anti-actin antibody for actin. PCR was performed as follows: 17 cycles of amplification for G3PDH using HSTaq and 27 cycles of amplification for BMRF1 using KAPATaqEx. Gel electrophoresis and quantification of RT-PCR products were performed as described in Fig. 6A. The levels of G3PDH or BMRF1 mRNA obtained from C666-1 cells were defined as 1.0. B, PLSCR1-KO C666-1 cells were transfected with 4 g of pcDNA3 or 3M-PLSCR1. After 72 h of transfection, total cell lysates and total RNA were prepared. Total cell lysates (5 g) were subjected to SDS-PAGE, and immunoblotting was performed using an anti-PLSCR1 antibody for PLSCR1. PCR was performed as follows: 20 cycles of amplification for G3PDH using HSTaq and 30 cycles of amplification for BMRF1 using KAPATaqEx. Gel electrophoresis and quantification of RT-PCR products were performed as described in Fig. 6B. The levels of G3PDH or BMRF1 mRNA obtained from empty vector-transfected cells were defined as 1.0. C, PLSCR1-KO C666-1 cells were transfected with 0.25 g of pcDNA3 or 3FG-BZLF1 and 4 g of pcDNA3 or 3M-PLSCR1. After 72 h of transfection, total cell lysates were prepared. Total cell lysates (5 g for BZLF1, PLSCR1, and actin and 15 g for BMRF1) were subjected to SDS-PAGE, and immunoblotting was performed using an anti-FLAG M2 antibody for BZLF1, an anti-PLSCR1 for endogenous PLSCR1, an anti-actin antibody for endogenous actin, or an anti-BMRF1 antibody for endogenous BMRF1. The intensities of the immunoblotted bands were quantified using ImageJ, and the intensities of the products obtained for the empty vector and 3FG-BZLF1-transfected cells were defined as 1.0.

Discussion
This work reveals for the first time that PLSCR1 directly interacts with EBV BZLF1 in vitro and in vivo (Figs. 2 and 3) and that amino acids 1-163 and 160 -250 of PLSCR1 (Fig. 2) and the C-terminal bZIP region of BZLF1 (Fig. 3) are involved in this interaction. Both BZLF1-binding regions of PLSCR1 contain an ID region (21) and interact with the ID regions of HTLV-1 Tax and HIV-1 Tat (21,22). ID regions are known to confer conformational flexibility, which facilitates posttranslational modifications and enables a protein to functionally interact with many cellular partners (31). Interestingly, the bZIP region of BZLF1 exhibited more effective binding to PLSCR1 than did fulllength BZLF1 (Fig. 3). These data are similar to the previous observation that the N-terminal 133 amino acid-truncated BZLF1 region more effectively interacts with the NF-B p65 subunit than does the full-length molecule (32). Notably, DIS-PROT VSL2P analysis results indicated that the bZIP region of BZLF1 also contains a long ID region (169 -216 amino acids) (33). This BZLF1 ID region may also alter its conformation so that it can interact with distinct regions of PLSCR1, and the N-terminal region of BZLF1 may affect the conformational change of the ID region of the bZIP region of BZLF1. The ID regions of both proteins likely play a key role in this proteinprotein interaction.
Because BZLF1 plays a crucial role in the switch from latent to lytic EBV infection (8,9), the functional consequences of the PLSCR1⅐BZLF1 interaction for BZLF1-dependent transactivation were assessed in vivo. Reporter analysis demonstrated that PLSCR1 overexpression effectively represses the BZLF1-dependent transactivation of the lytic BMRF1 promoter and that the BZLF1-binding regions of PLSCR1 are sufficient for this repression (Fig. 5). However, the N-terminal (1-163 amino acids)-binding region of PLSCR1 exhibited more effective repression than the 160 -250-amino acid region (Fig. 5). Furthermore, in parental and PLSCR1-KO EBV-infected NPC cells, PLSCR1 overexpression decreased the expression of BZLF1 up-regulated lytic BMRF1 expression at the mRNA and protein levels, and the knockdown or KO of endogenous PLSCR1 expression increased the levels of BMRF1 mRNA (Figs. 6 and 7). However, we could not isolate virus particles from parental and PLSCR1-KO C666-1 cells, and we could not determine whether PLSCR1 expression also reduces virion production in EBV-infected NPC cells. Further investigation is warranted to confirm whether this elevated expression of PLSCR1 not only represses lytic gene expression but also reduces EBV virion production.
The interaction of BZLF1 with CBP through its dimerization domain is known to effectively enhance the BZLF1-mediated transactivation of early gene promoters (30). In this study, the results of pulldown assays revealed that PLSCR1 efficiently represses BZLF1-CBP complex formation in PLSCR1-KO HEK-293 and HeLa cells (Fig. 8). Notably, the PLSCR1-binding region of BZLF1 is included within the DNA-binding domain of the bZIP motif (Fig. 3). However, PLSCR1 competes with the C-terminal bZIP motif of BZLF1 for its interaction with CBP. Because the PLSCR1-binding region of BZLF1 is located next to the dimerization domain of BZLF1, PLSCR1 may sterically interfere with BZLF1-CBP complex formation. Furthermore, BZLF1 interacts with a number of host regulatory molecules, including the p53, NF-B p65, and C/EBP proteins (32,34,35), and affects their functions. Further investigations are needed to determine whether PLSCR1 affects BZLF1-mediated regulation of host target molecules through protein-protein interactions in EBV-infected cells.
Although EBV can infect both B cells and epithelial cells, it typically establishes only latent infections in B cells. B-cell-specific transcription factors have been suggested to promote viral latency through the negative regulation of BZLF1 in B cells (6,7). However, latent EBV infection occurs in EBV-associated epithelial malignancies, such as NPC and gastric cancers, and the negative regulatory mechanism(s) of BZLF1 may be involved in the development of EBV-associated epithelial or S-BZLF1, and 3.6 g of pcDNA3 or 3M-PLSCR1. A total of 500 g of total cell lysate prepared using Nonidet P-40 lysis buffer, which was subsequently incubated with S-protein beads to precipitate BZLF1. Following pulldowns, 15 g of total cell lysate and the precipitated complexes were divided into three portions and subjected to SDS-PAGE. Immunoblotting was performed using an anti-S-tag antibody for BZLF1, an anti-FLAG M2 antibody for CBP(1-721), or an anti-myc antibody for PLSCR1. B, PLSCR1-KO HeLa cells stably expressing empty vector or 3M-PLSCR1 were transfected with 1.6 g of 3FG-CBP(1-721) and 0.8 g of pcDNA3 or S-BZLF1. A total of 250 g of total cell lysate prepared by using Nonidet P-40 lysis buffer was incubated with S-protein beads to precipitate BZLF1. Following pulldowns, 7.5 g of total cell lysate and the precipitated complexes were divided into three portions and subjected to SDS-PAGE. Immunoblotting was performed using an anti-S-tag antibody for BZLF1, an anti-FLAG M2 antibody for CBP  or an anti-myc antibody for PLSCR1.

PLSCR1 represses EBV BZLF1-dependent lytic transcription
malignancies. The results of this study show for the first time that the basal expression of PLSCR1 was significantly elevated in EBV-infected NPC cells. (Fig. 1, A and C). However, the basal expression of PLSCR1 was significantly lower in EBV-infected BL cells, and IFN treatment strongly induced PLSCR1 expression these cells (Fig. 1B), consistent with our previous report on EBV-negative human epithelial cells (21). The precise induction mechanism of PLSCR1 expression in EBV-infected NPC cells remains unclear. Interestingly, the basal expression of PLSCR1 was also significantly elevated in the human epidermoid carcinoma cell line A431 in the absence of IFN (Fig. 1A). Because C666-1 and A431 cells originate from epidermal cells, we assessed the basal expression of PLSCR1 in EBV-negative human immortalized primary keratinocyte cells. Notably, the basal expression of PLSCR1 was also increased in two immortalized human epidermal keratinocyte cell lines, normal oral keratinocyte and HaCaT, similar to those observed in three EBV-infected NPC cells and an IFN-␣-treated HeLa cell line (Fig. S3, lanes 2-7). The normal oral keratinocyte cell line is a human primary oral epithelial keratinocyte cell line immortalized by human telomerase, and the HaCaT cell line is a spontaneously immortalized human primary skin keratinocyte cell line. A previous report showed that human telomeraseimmortalized primary keratinocytes exhibit similar properties to primary human keratinocytes (36). Thus, the basal expression of PLSCR1 may be elevated in human primary epidermal keratinocytes, similar to that in both immortalized human epidermal keratinocyte cell lines. This elevated expression of PLSCR1 in EBV-infected NPC cells may not arise from EBV infection but from these cells' epidermal origin. c-Myc has been suggested to up-regulate the expression of PLSCR1 by directly binding to its promoter in HEK-293 cells (37). c-Myc is also known to be predominantly expressed in the basal cell layers of the epidermis and is constitutively expressed in mouse primary keratinocytes (38,39). This constitutive expression of c-Myc in the epidermis may play a key role for in high level of PLSCR1 expression in epidermal cells. However, further investigation is warranted to confirm whether the basal expression of PLSCR1 is significantly elevated in epidermal cells by using primary epidermal tissues and/or cells. Furthermore, EBV infection increases the expression of c-Myc through the STAT3 and NF-B pathways in human nasopharyngeal epithelial cells (40), and c-Myc protein expression is negatively regulated in EBV latently infected type ⌱ and type ⌱⌱⌱ BL cells (41,42). The levels of c-Myc protein may also play an important role in PLSCR1 expression in EBV-infected BL and NPC cells. Interestingly, IFN-␣ treatment has been shown to inhibit lytic gene expression in lytic replication-induced Daudi cells (43). The results of this study revealed that IFN-␣ treatment significantly induces PLSCR1 expression in EBV-positive BL cells, including Daudi cells (Fig. 1B). This IFN-␣-induced high level of PLSCR1 expression may also contribute to the IFN-mediated repression of EBV lytic replication in EBV-infected BL cells through the negative regulation of BZLF1. Taken together, PLSCR1 expression may be enhanced by the EBV-dependent and -independent c-Myc up-regulation in EBV-infected NPC cells. Further investigation is warranted to determine whether c-Myc affects PLSCR1 expression in human epidermal cells. Furthermore, this high level of PLSCR1 expression may contribute to promoting viral latency through the negative regulation of BZLF1 by preventing BZLF1-CBP complex formation in EBV-infected NPC cells.

Materials
The EBV-infected NPC cell line C666-1 (44) and EBV-infected C15 and C17 NPC xenograft frozen tumors (45) were generously provided by N. Raab-Traub. A431, SW480, and MCF-7 cell lines were obtained from the Cell Resource Center for Biomedical Research, Cell Bank, Tohoku University. The anti-EBV EA-D-p52/50 (BMRF1) and histone H3 (96C10) antibodies used in this study were purchased from Chemicon and Cell Signaling Technology, respectively. Immunoblotting, cDNA preparation, PCR amplification for cloning, and sequencing of all DNA constructs were performed as described previously (46,47). Other materials used in this study were obtained as described previously (21).

Bacterial protein expression and pulldown assays
Bacterial protein expression and pulldown assays were performed as described previously (21) Cell culture and transfection HEK-293, COS-1, HeLa, A431, and MCF-7 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with penicillin and streptomycin (PS) (Biological Industries) and 10% fetal bovine serum (FBS) (Gibco BRL) as described previously (21,48). SW480 cells were maintained DMEM/F-12 medium containing PS and 10% FBS. C666-1, BJAB, B95-8, Namalwa, P3HR1, and Daudi cells were main-PLSCR1 represses EBV BZLF1-dependent lytic transcription tained in RPMI 1640 medium containing PS and 10% FBS. For pulldown and immunoprecipitation assays, HEK-293 and PLSCR1-KO HEK-293 cells (1 ϫ 10 6 ) were plated in 60-mm cell culture plates and were transfected with the indicated amounts of DNA using jetPEI (Polyplus Transfection) as suggested by the manufacturer. For the luciferase assay and immunofluorescence analysis, COS-1 cells were transfected with the indicated amounts of DNA using jetPEI or TransIT-LT1 (Mirus) as suggested by the manufacturer. HeLa and PLSCR1-KO HeLa cells (5 ϫ 10 5 ) were plated in 60-mm cell culture plates and transfected with the indicated amounts of DNA using jetPRIME (Polyplus Transfection) as suggested by the manufacturer. C666-1 and PLSCR1-KO C666-1 cells (9 ϫ 10 5 ) were plated in fibronectin-coated 60-mm cell culture plates and were transfected with the indicated amounts of DNA using Tran-sIT-X2 Dynamic Delivery System (Mirus) as suggested by the manufacturer.

Preparation of empty vector or stable PLSCR1-expressing PLSCR1-KO HeLa cells
PLSCR1-KO HeLa cells were transfected with empty vector (pIRES-puro3) or pIRp3-3M-PLSCR1 as described previously. Next, 48 h after transfection, the PLSCR1-KO HeLa cells were selected with 2 g/ml of puromycin (InvivoGen) for 2 weeks. Subsequently, puromycin-resistant cells were harvested and 3ϫmyc epitope-tagged PLSCR1 expression was confirmed by immunoblot analyses using an anti-myc antibody.

Immunofluorescence analysis
Immunofluorescence analysis was performed as described previously (46).

Subcellular fractionation
The subcellular fractionation was performed as described previously (21).

Dual luciferase assay
Dual luciferase assays were performed as described previously (46).

Semiquantitative RT-PCR
Total cell lysates and total RNA were prepared as described previously (47). First-strand cDNA was synthesized from 0.5 g of total RNA from each sample as described previously (49). PCR was performed using 1 l of a 4-fold dilution of each cDNA reaction mix and primer sets for human G3PDH (47) and BMRF1 (5Ј-CTAGCCGTCCTGTCCAAGTGC-3Ј and 5Ј-AGCCAAACAGCTCCTTGCCCA-3Ј) (50). PCR was performed using the indicated DNA polymerases for the indicated number of amplification cycles, as suggested by the manufacturer. PCR products were subjected to agarose gel electrophoresis, and DNA bands were visualized using Gel Green Nucleic Acid Gel Stain (Biotium).