Main

Epstein-Barr virus (EBV) is a γ-herpesvirus that causes infectious mononucleosis in normal adolescents and lymphoproliferative disease in immunocompromised individuals1, and is also associated with human cancers including Burkitt lymphoma, nasopharyngeal carcinoma and Hodgkin disease1,2. More recently, and controversially, EBV was shown to be associated with invasive breast carcinoma3,4. Typically, infection of resting primary B-cells by EBV results in expression of six nuclear proteins (EBNAs), three latent membrane proteins (LMPs) and two early RNAs (EBERs) and the development of lymphoblastoid cell lines (LCLs)1,2,5,6,7. Genetic recombination studies revealed that EBNA-1, -LP, -2, -3A, -3C and LMP1 are critical for growth transformation whereas EBNA-3B, LMP2A, LMP2B and the EBERs are dispensable for the immortalization of B cells5,6,7,8. EBNA-1 is important for episome maintenance, whereas the other EBNAs appear to function primarily as regulators of transcription1,9.

The EBNA-3C is a large transcription factor encoded by EBV and is a member of the EBNA-3 family of proteins arranged in tandem on the viral genome. These proteins have similar structural motifs in that they have binding sites for RBP-Jκ, a leucine zipper, acidic domains, proline and glutamine rich repeats and several arginine or lysine residues (Fig. 1b) responsible for nuclear translocation10. Recombinant EBV with a stop codon inserted after amino acid 365 in EBNA-3C results in EBV incapable of immortalizing primary B-lymphocytes indicating that interactions that occur downstream of amino acid 365 are critical events in this process7. Although EBNA-3C does not specifically bind DNA11,13, it upregulates expression of CD21 (ref. 1) and of LMP1 in G1-arrested, EBV-infected Raji cells1,14. EBNA-3C can interact with RBP-Jκ and modulate EBNA2 transactivation of viral promoters, but also can repress transcription directly when targeted specifically to promoters in transient transfection assays15,17. Thus, this dual role exhibited by EBNA-3C in transcriptional activation and repression may be critical for transformation of B-lymphocytes.

Figure 1: Structural motifs of Nm23-HI and EBNA-3C.
figure 1

a and b, Schematic representation of the structure and expression of Nm23-H1 (a) and EBNA-3C (b). a, Secondary structure of Nm23-H1 shows potential α-helical regions, β-sheets26 and the head rich in leucines (Head, 43–69aa) involved in interactions with nucleotides. In the region 94–114aa, the K-pn loop is shown. Serines at positions 44, 120, 122 and 125 are the autophosphorylation sites indicated by asterices. b, The open-boxed region shows the RBP-Jκ site (Jκ). The proline-rich repeats (P Rich), the acidic domain (AD), the glutamine-rich repeats (Q Rich) and the nuclear localization signal (NLS) present in EBNA-3C are indicated. c, Northern-blot analysis of PolyA RNA from EBV infected LCLs and Burkitt lymphoma cell lines using [32P]-labeled Nm23-H1 probe and GAPDH DNA probe. d, Western-blot analysis of B-cells indicating the expression of EBNA-3C and intracellular Nm23-H1 and EBNA-3C are shown in panel along with the same blot stained with Ponceau S to show protein-loading control levels.

Full size image

Here we report that the carboxy terminal region of EBNA– 3C (aa365–992) critical for immortalization of B lymphocytes interacts with the human metastatic suppressor protein Nm23-H1 (ref. 18), a cellular protein which belongs to a family of genes highly conserved in eukaryotes. Eight distinctly different genes of this family have now been identified in humans: NM23-H1, NM23-H2, DR-nm23, NM23-H4, NM23-H5, NM23-H6, NM23-H7 and NM23-H8 (ref. 19). Interestingly, expression of the NM23 gene family is linked to suppression of tumor metastasis, differentiation, apoptosis, proliferation and DNA mutation20.

NM23-H1 and -H2 are nucleoside diphosphate (NDP) kinases21 and suppression of metastasis was observed in several tumor cell lines transfected with Nm23/NDP kinase-A (refs. 22–25). Moreover, a murine Nm23 gene (Nme6) was identified by differential hybridization of cDNAs from high- and low-metastatic melanoma cell lines18. Based on the crystal structure of Nm23-H2 (ref. 26), the predicted structure of Nm23-H1 includes α-helical regions, β sheets, the K-pn loop and a region involved in interactions with nucleotides (Fig. 1a). Typically, in cancers of the breast, ovaries, cervix, gastric epithelia, hepatocellular carcinoma and melanomas, there is an inverse relationship between Nm23-H1 expression and metastasis20,27. However, in other tumors like neuroblastomas, mutations in the leucine-rich motif of Nm23-H1 directly correlate with Nm23 expression and tumor cells metastasis23,28. Therefore, Nm23-H1 is implicated in the regulation of metastasis in a variety of human cancers29.

In this report we demonstrate interaction of Nm23-H1 with EBNA-3C and show that expression of EBNA-3C reverses the ability of Nm23-H1 to inhibit migration of Burkitt lymphoma and breast carcinoma cells. This represents the first evidence that a DNA tumor virus known to infect humans, targets a cellular protein associated with metastasis to promote migration of human cancer cells.

Nm23-H1 interacts with the carboxy terminus of EBNA-3C

To better understand the role of the functionally important carboxy terminus of EBNA-3C, we performed a yeast two-hybrid study using an EBNA-3C (aa365-992) bait to screen a LCL-derived cDNA library30. Two His+/LacZ+ clones were isolated which encoded the metastasis suppressor protein Nm23-H1 (ref. 18). In yeast two-hybrid interaction studies, interaction between the EBNA-3C carboxy terminus and Nm23-H1 was equivalent to interaction between full length EBNA-3C and RBP-Jκ, as measured by β-galactosidase activity in a filter assay (data not shown).

Expression of NM23-H1 in EBV-negative and -positive B-cell lines

Northern-blot analysis of multiple tissue blots showed a single mRNA species hybridizing with the NM23-H1 cDNA probe, expressed in a tissue-dependent manner. Lowest expression was seen in peripheral blood lymphocytes (PBL) and lung tissue, and increased expression was in placenta, smooth muscle, kidney, spleen, thymus, colon and uterus. The highest levels were in heart, brain, liver, pancreas, prostate, testis and small intestine (data not shown).

Northern-blot analysis from EBV-positive and -negative cell lines showed low but detectable levels of the 0.8-kb NM23-H1 RNA transcript in cell lines (Fig. 1c). We did not see major differences in expression levels in EBV-infected BL41/B958 or in EBV-transformed B-lymphoblastoid cell lines (LCL1 and IB4); this indicated that similar levels of expression of Nm23-H1 in EBV-positive and -negative cell lines.

Western-blot analysis with an Nm23-H1 antibody on lysates from LCL1, LCL2 and BL41/B958 showed the presence of Nm23-H1 in all cell lines (Fig. 1d). The results indicated slightly lower amounts of Nm23-H1 in uninfected BL41 compared with EBV-infected LCLs expressing the latent viral genes (BL41/B958). We stripped the blot and probed it with a monoclonal antibody specific for EBNA-3C and showed that EBNA-3C was expressed in LCL1 and LCL2 and to a lesser extent in the BL41/B958 cell line.

EBNA-3C interacts with Nm23-H1 in vitro and in vivo

In similar binding assays as above, we observed no binding of GST-Nm23-H1 to in vitro-translated EBNA-3C containing amino acids 1-365 (Δ3′), whereas in vitro-translated EBNA-3C containing amino acids 366-992 (Δ5′) bound with similar affinity as that of full-length EBNA-3C (Fig. 2b). We did not see binding to in vitro-translated EBNA-3B, a protein with similar structural motifs and pI as EBNA-3C (Fig. 2b, right panel). Together, these results confirmed the binding specificity of Nm23-H1 to EBNA-3C.

Figure 2: Epstein-Barr virus nuclear protein EBNA-3C associates with NM23-H1 in vitro.
figure 2

a, GST, GST-NM23-H1 and GST-NM23-H2 fusion proteins were incubated with in vitro-translated [35S]-labeled EBNA-3C full-length (E3C-FL) and luciferase as control. b, E3CΔ5′ (aa366–992), E3CΔ3′ (aa1–365) and EBNA-3B full-length represented as E3B-FL. c, A GST pull-down assay using BJAB cell lines transfected with a eukaryotic expression vector containing EBNA-3C cDNA insert or empty vector pZIPneo47 (BJAB) or EBNA-3C cDNA insert (E3C.7 or E3C.10.).

Full size image

To corroborate these results, we incubated GST-Nm23-H1 with cell lysates from EBNA-3C negative (BJAB) and EBNA-3C positive cell lines (EBNA-3C.7 and EBNA-3C.10). EBNA-3C was detectable by western blotting in the the protein complexes using antibody against EBNA-3C. We observed strong binding only in lysates from EBNA-3C expressing cells (Fig. 2c).

We expressed EBNA-3C and MYC-tagged Nm23-H1 by transient transfection in 293 cell lines to investigate whether these proteins interact in vivo. Immunoprecipitation of MYC-tagged Nm23-H1 using anti-MYC antibody resulted in co-immunoprecipitation of EBNA-3C detected by western blotting using EBNA-3C antibody only in cells transfected with both expression constructs (Fig 3a). Similarly, immunoprecipitation of EBNA-3C co-immunoprecipitated MYC-tagged NM23-H1 detected by western blotting using MYC antibody (Fig. 3b).

Figure 3: Nm23-H1 and EBNA-3C interact in human 293 cells in vivo.
figure 3

293 cells were transfected with various Nm23-H1 and EBNA-3C constructs shown above the lanes. a, Immunoprecipitation of MYC-tagged Nm23-H1 using anti-MYC monoclonal antibody results in co-immunoprecipitation of EBNA-3C detected by western blotting using EBNA-3C monoclonal antibody (top panel). Equal amounts of cell lysates from each transfection were probed with anti-EBNA-3C antibody to verify expression (bottom panel). b, Immunoprecipitation of EBNA-3C using anti-EBNA-3C monoclonal antibody results in co-immunoprecipitation of MYC-tagged NM23-H1 detected by western blotting using anti-MYC monoclonal antibody (top panel). Equal amounts of cell lysates from each transfection were probed with anti-MYC antibody (middle panel) or anti-NM23-H1 antibody (bottom panel).

Full size image

Interaction in human B-lymphoblast cell lines

We assayed various human B-lymphoblastoid cell lines for EBNA-3C–Nm23-H1 complexes to verify interaction between endogenous proteins. Immunoprecipitates of EBNA-3C contained Nm23-H1 and correspondingly, immunoprecipitation of Nm23-H1 yielded EBNA-3C in EBV positive B lymphoma cell lines (EBNA3C.7 and EBNA3C.10) but not in the EBV negative B lymphoma cell line BJAB (Fig. 4a and b). Similar results were obtained in recently immortalized LCLs expressing the entire repertoire of EBV latent genes (Fig. 4c). Approximately 10 to 15% of the EBNA-3C co-immunoprecipitated with Nm23-H1 and nonspecific association was not observed in the precleared control lanes. The multiple bands were due to posttranslational modification15,17. Repeat studies using an isotype-specific IgG antibody control confirmed the specificity of the interaction (Fig. 4d). These results verify the interaction of endogenous NM23-H1 and EBNA-3C in human B lymphoblastoid cell lines.

Figure 4: Nm23-H1 co-immunoprecipitates with EBNA-3C in stably transfected EBNA-3C expressing cell lines and in EBV-transformed lymphoblastoid cell lines.
figure 4

a and b, BJAB cell lines expressing EBNA-3C or pZIPneo vector control. Proteins were immunoprecipitated using Nm23-H1 or EBNA-3C antibodies and western blotted using A10 monoclonal antibody or mouse monoclonal Nm23-H1 antibody for the detection of EBNA-3C (121 kD) and Nm23-H1 (23 kD), respectively. c, LCLs used to determine the interaction of intracellular Nm23-H1 with EBNA-3C in EBV-transformed cells. d, To determine the specificity of the interactions the Nm23-H1 immunoprecipitation was done as above and the control immunoprecipitation was done using a matched isotype-specific IgG control. Protein A (PA), Immunoprecipitation (IP).

Full size image

EBNA-3C colocalizes with Nm23-H1 in the nucleus of B-cells

Immunofluorescence analysis of B cells expressing EBNA-3C demonstrated that Nm23-H1 colocalizes with EBNA-3C in the same subnuclear structures. In an EBV-negative BJAB cell line, Nm23-H1 predominantly stained the cytoplasm (Fig. 5a, top panels). However, in cells that are transfected with EBNA-3C, Nm23-H1 signals were pronounced in the nucleus in specific subnuclear structures (Fig. 5a, bottom panels). This observation indicates that EBNA-3C is capable of increasing the levels of Nm23-H1 in the nucleus and may regulate Nm23-H1 function. We counted sets of 200-cell fields from different wells probed with antibodies independently for Nm23-H1 and EBNA-3C and plotted the number of cells showing predominant nuclear staining for Nm23-H1 (Fig. 5b). The number of cells showing Nm23-H1 staining correlated with the transfection efficiency of EBNA-3C (20%) indicating that EBNA-3C expression contributed to the increased translocation of Nm23-H1 to subnuclear structures. Notably, 1% of control transfected cells showed mostly nuclear staining for Nm23-H1. Although the basis for this is not yet known, we speculate that these molecules may translocate to the same nuclear compartment and are involved in specific cellular functions.

Figure 5: Immunofluorescence analysis to determine the cellular localization of Nm23-H1 and EBNA-3C.
figure 5

a, BJAB cells transfected were fixed and incubated with primary antibodies against Nm23-H1 (rabbit anti-NM23-H1) and EBNA-3C (A10 monoclonal), as indicated. Localization of the EBNA-3C and Nm23-H1 proteins are shown for BJAB cells transfected with and without EBNA-3C (a, left and middle panels). Colocalization of the proteins was demonstrated by merging the left and middle panels and presented as panels on the right. b, Localization of Nm23-H1 in the nucleus was quantified by counting immunopositive cells with predominantly nuclear staining in each transfected sample from 5 sets of 200 cells per field.

Full size image

EBNA-3C reverses suppression of tumor cell migration

As increased cell motility is correlated with increased metastatic potential, we investigated the migratory ability of the human breast carcinoma cell line MDA-MB-435 and the human Burkitt lymphoma cell line BJAB in the presence of EBNA-3C and Nm23-H1. The cells were maintained under selection and analyzed for stable expression before they were assayed for migration (see Fig. 6g and h). The breast carcinoma cell line MDA-MB-435, which was stably transfected with vector control, demonstrated greater migration in the three assays performed at varying concentrations. In comparison, the cell lines stably expressing Nm23-H1 showed approximately 50–75% reduction in migration in response to different concentrations of chemo-attractants. When EBNA-3C was co-expressed with Nm23-H1, the reduction in cell migration associated with Nm23-H1 expression was reversed (Fig. 6a, top panels). These results demonstrate that EBNA-3C reverses the Nm23-H1–mediated reduction in cell motility of MDA-MB-435 cells.

Figure 6: Motility of MDA-MB-435 breast carcinoma and BJAB cell lines in the presence of different chemoattractants.
figure 6

a, The breast carcinoma MDA-MB435 and the Burkitt lymphoma, BJAB cell lines were stably transfected and the expression of Nm23-H1 and EBNA-3C determined before the cells were used in the assay. The number of cells attached to the underside of the 0.8-micron membrane was determined with a ×40 bright field microscopy. Each experiment was carried out in triplicate twice and the error bars indicate the standard deviation calculated from 6 such data points. Results for the MDA-MB435 are presented in upper panels and for the BJAB assays in lower panels. ▪, 0.25% serum; □, 0.5% serum. b, To detect Nm23-H1 and EBNA-3C western blots of cell lysates were performed using Nm23-H1 monoclonal antibody or EBNA-3C antibody as indicated.

Full size image

The initial motility assays performed on MDA-MB-435 cells demonstrated that EBNA-3C inhibits the antimetastatic effects of Nm23-H1. However, the presence of EBV in breast carcinoma cells is somewhat controversial, and EBNA-3C is not typically detected in EBV associated epithelial cancers2. Nevertheless, EBV can infect breast carcinoma cells31 and EBNA-3C may be involved in initiation and establishment of EBV in epithelial cancers. Given these factors, we selected BJAB cells as in the MDA-MB-435 cell line for stable expression of EBNA-3C and Nm23-H1. EBNA-3C inhibited the effects of Nm23-H1 in BJAB cells (Fig. 6a, bottom panels) but had no effect on migration if expressed alone. Representative blots are shown for EBNA-3C and Nm23-H1 (Fig. 6b) expression from MDA-MB435 cell lines. As expected, levels of Nm23-H1 expression were slightly higher in the stably transfected cells expressing exogenous levels of Nm23-H1 from a heterologous promoter compared with vector controls. All results were repeated multiple times in triplicate, normalized and plotted. We demonstrate that EBNA-3C inhibits the ability of Nm23-H1 to suppress the migration of breast carcinoma as well as Burkitt lymphoma cells in an in vitro migration assay.

Discussion

EBNA-3C is one of the six latent proteins essential for EBV transformation of B-lymphocytes1,2. EBNA-3C can interact with cellular transcription factors RBP-Jκ/CBF1 and Sp-1 (refs. 15,16,32). Here we have identified the metastatic suppressor protein Nm23-H1 as a cellular target for EBNA-3C. Nm23-H1 is a 152–amino-acid 17-kD nuclear protein33 which is expressed in a number of primary tumors like neuroblastoma, colorectal carcinoma, breast carcinoma and renal carcinoma, and is sometimes mutated which alters the structure and function of the protein23,34.

Previous PCR studies have identified the presence of EBV-encoded RNA (EBER1) in malignant cells of breast tumors, whereas other studies have not been able to detect EBER1 (refs. 35,36). More recently, EBNA-1 was detected in malignant breast tissue from a large subset of patients3 and EBV was shown to infect a number of breast carcinoma cell lines. Furthermore, the relationship between the presence of EBV and several poor prognostic factors such as estrogen receptors in these tumors was investigated3,31. Although these previous studies have demonstrated the association between EBV and human cancers, there have not been any previous studies linking a known etiologic agent associated with human malignancies and a cellular molecule with anti-metastatic potential.

This is the first demonstration of a specific interaction between Nm23-H1 and any viral oncoprotein. Using a variety of assays we demonstrated a physical interaction between Nm23-H1 and EBNA-3C in vitro and in vivo. We have shown that Nm23-H1 is predominantly localized in the cytoplasm in B-lymphocytes. However, in EBV-transformed LCLs and in B cells transfected with EBNA-3C, Nm23-H1 is predominantly nuclear and colocalizes with EBNA-3C. These results indicate that EBV may influence the cellular localization and function of Nm23-H1 in infected cells—probably through EBNA-3C and other essential EBNA proteins.

A number of studies have shown a tight correlation between cell motility and metastatic potential in a number of related tumor cell lines25,29,33,37,38. Motility assays were performed using the MDA-MB-435 breast carcinoma and BJAB Burkitt lymphoma cell lines stably expressing Nm23-H1 and EBNA-3C. Our results corroborated previous studies in which increased expression of Nm23-H1 resulted in a decrease in motility of the breast carcinoma cells when chemo-attractants like serum, insulin growth factor (IGF) and platelet-derived growth factor (PDGF) were added to the medium39. EBNA-3C expressed with Nm23-H1 resulted in the inhibition of the ability of Nm23-H1 to suppress cell motility in all cases, and is probably due to direct interaction with EBNA-3C.

We present strong evidence to support the interaction of Nm23-H1 and EBNA-3C in vitro and in human cells. However, EBNA-3C by itself had a lesser effect on cell migration when overexpressed in these cells in the absence of exogenous expression of Nm23-H1. The endogenous Nm23-H1 may be sequestered by cellular partners competing for intracellular pools of Nm23-H1. Alternatively, the assay system may have a threshold such that overexpressing EBNA-3C on a background level of endogenous Nm23-H1 would not be expected to dramatically affect Nm23-H1 function.

Studies carried out in human 7721-hepatocarcinoma cell lines showed upregulation of Nm23-H1 in all-trans retinoic acid (ATRA)-treated cell lines and downregulation of Nm23-H1 in epidermal growth factor (EGF)-treated and HER-2/neu (c-erbB-2)-transfected cell lines40. EBV increases EGF (ref. 41) and this may in turn downregulate Nm23-H1 expression40 leading to a decrease in cell adhesion and an increase in cell migration and invasion. Several lines of evidence have indicated that αv-integrins may have a role in tumor development and that expression of αv-integrins correlates with metastasis of melanomas and breast carcinomas, and with cell proliferation and extracellular matrix invasion42,43,44. αv-integrin is expressed in LCLs and its expression is associated with the presence of EBV in B-lymphocytes. Moreover, the motility data presented here indicates that EBNA-3C reverses the ability of Nm23-H1 to suppress motility of human cancer cells, possibly by modulating expression of αv integrins.

Our results indicate a specific interaction between the EBV essential nuclear antigen EBNA-3C and Nm23-H1. The association of these two molecules in cells may result in increased migration of breast carcinoma and Burkitt lymphoma cells. Whether this association may be linked to the upregulation of adhesion molecules including αv-integrins44 and E-cadherin45 is not yet known. Nonetheless, these results provide an initial molecular framework for elucidating a mechanism by which EBV nuclear antigens or EBNA-3Cs, specifically, trigger migration and thus metastasis in human malignancies.

Methods

Constructs and cell lines.

The GST-NM23-H1 fusion construct was cloned from a PCR amplified product into pGEX-2TK vector46. The forward primer 5′-TTAGGATCCCATATGGCCAACTGTGAGCG-3′ and reverse primer 5′-TTACCCGGGCATGGATCCTCCTCCTGTCATTCA-3′ contained BamH1 sites at the ends. The forward and reverse primer for cloning NM23-H2 are 5′-TTAGGATCCATGGCCAACCTGGAGCGCA-3′ and 5′-TATGAATTCTTATTCATAGACCCAGTCATGA-3′ forward and reverse primers, respectively amplified using VentR DNA polymerase (NEB, Beverly, Massachusetts)). The MYC-fusion protein was generated as above (except for use of the reverse primer 5′-GCTGCGGCCGCTTCATAGATCCAGTTCTGAGC-3′) and cloned into the pA3M vector prepared with BamHI and NotI (ref. 47). The EBV-negative Burkitt lymphoma BJAB cell line was obtained from E. Kieff1. EBNA-3C–expressing cell lines were as described16,47. The BL41/B958 cell line is an EBV-infected Burkitt lymphoma cell line48 and LCL1 and LCL2 are recently transformed B cell lines47. B-cell lines were maintained in RPMI 1640 medium (Gibco-BRL, Rockville, Maryland) supplemented with 10% fetal bovine serum, 2 mM glutamine, 25 U/ml penicillin/streptomycin and 20 μg/ml gentamicin (Gemini Bioproducts). BJAB-neo and EBNA-3C–expressing cell lines were maintained in the medium with 200 ng/ml neomycin (Gibco-BRL, Rockville, Maryland) added. For cotransfection and motility experiments, we used the 293-embryonic kidney cell line47 and the human breast carcinoma cell line MDA-MB-435, respectively39 cultured in DMEM (Gibco-BRL, Rockville, Maryland) and supplemented as above.

Northern-blot analysis.

The presence of NM23-H1 transcripts in different tissues was analyzed by northern blot of polyA RNA isolated from BL41, BL41/B958, IB4 and LCL1. The [32P]dCTP-labeled control β-actin and NM23-H1 probes were made using a random primer kit (Stratagene, La Jolla, California)) and were purified using a Nuc-Trap purification column (Stratagene, La Jolla, California). Hybridization and washing was done according to manufacturer's instruction (Clonetech, Palo Alto, California). The blot was exposed using the PhosphorImager system (Molecular Dynamics, Sunnyvale, California).

In vitro protein-binding studies.

GST-Nm23-H1 and GST-Nm23-H2 beads used for the in vitro binding experiments and the GST-Nm23-H1 used in the GST pull-down assays were prepared as described previously16,47. For binding experiments, EBNA-3C full-length (E3C-FL), E3CΔ3′ (aa1–365), E3CΔ5′ (aa366–992) and control luciferase were translated in vitro (Promega, Madison, Wisconson) using [35S]Met/Cys express label (NEN-Dupont, Boston, Massachussets). Proteins bound to both the GST and GST fusion proteins were denatured using SDS-β-mercaptoethanol lysis buffer and fractionated on SDS-PAGE. The gel was dried and analyzed using a PhosphorImager and signals quantified using the ImageQuant program (Molecular Dynamics, Sunnyvale, California). GST pull-down assays were carried out as described16,47. As before, the proteins were denatured and fractionated on SDS-PAGE. The proteins were transferred to 0.45-μm nitrocellulose filters and identified by western blotting using the A10 EBNA-3C mouse monoclonal antibody47.

Immunoprecipitation experiments.

5 × 107 cells were collected and washed with PBS and were lysed in RIPA buffer and immunoprecipitation was performed as described47. Briefly, the lysates were incubated with mouse monoclonal Nm23-H1 antibody or a matched isotype-specific control antibody (Santa Cruz, Santa Cruz, California) or EBNA-3C rabbit polyclonal antibody overnight along with the Protein A-Sepharose beads at 4 °C. The membrane was incubated with EBNA-3C mouse monoclonal antibodies followed by incubation with HRP-conjugated secondary antibody against mouse and detected using chemiluminescence detection protocols (Amersham, Piscataway, New Jersey).

Transfection and colocalization.

20 μg of pA3M empty vector, pA3M-NM23-H1, pSG5-EBNA-3C and pSG5-EBNA-3C + pA3M-NM23-H1 were transfected into 1 × 107 BJAB cells by electroporation (210 V and 975 μF), collected after 24 h and immunoprecipitated using Nm23-H1 mouse monoclonal antibody or EBNA-3C rabbit polyclonal antibody and analyzed by western blotting for the detection of EBNA-3C and Nm23H1. Expression of MYC-tagged Nm23-H1 was detected using antibody against MYC ascites. BJAB, EBNA-3C-expressing BJAB and LCL1 cells were fixed onto slides using 1:1 methanol and acetone and then blocked using 20% goat serum. EBNA-3C rabbit polyclonal antibody A10 and Nm23-H1 mouse monoclonal antibody or rabbit anti-Nm23-H1 were used as primary antibodies. FITC-conjugated anti-mouse and Texas Red-conjugated anti-rabbit secondary antibodies were added and the colocalized proteins were visualized on an Olympus BX60 fluorescent microscope. The photographs were captured using an Olympus digital camera and the Esprit program, version 1.2.

Cell motility assays.

MDA-MB-435 or BJAB cells were transfected with 10 μg of pCMV vector, pCMV-Nm23-H1, pSG5EBNA-3C and pCMV-Nm23-H1 using Superfect (Qiagen, Valencia, California). Cells were selected in 2 mg/ml neomycin and stable cell lines were maintained in use in the motility assays. Cell motility was determined using 96-well Boyden chemotaxis chambers (NeuroProbe, Gaitherburg, Maryland). The chemoattractants used were serum (Gemini-Bioproducts, Calabasas, California), EGF and fibronectin (Boehringer-Roche, Indianapolis, Indiana). Dilution of the chemoattractants was done in DMEM medium (for MDA-MB435 cells) or RPMI 1640 medium (for BJAB cells) containing 0.1% BSA, 10 mM HEPES and 100 U/ml penicillin/streptomycin. The concentrations of the chemoattractants used are 0.25% and 0.5% serum; 2.5 ng/ml and 5 ng/ml of fibronectin or EGF. 7.5 × 104 selected cells were added to the upper wells of the chamber and the chamber was incubated for 2 h at 37 °C in a humidifying CO2 incubator39. The membranes were removed and stained using Gill's hematoxylin stain. The cells were counted at ×40 magnification with an Olympus BX40 light microscope. The data presented for each concentration represents the mean of 3 separate experiments.