The Epstein-Barr virus encoded latent membrane protein-1 (LMP-1) and vimentin (VIM) are well known as nasopharyngeal carcinoma (NPC) biomarkers, which are reported to be highly expressed in NPC tissues. The exact mechanism between LMP-1 and vimentin proteins is still unclear, although some studies have reported that LMP-1 increases the expression of VIM messenger ribonucleic acid (mRNA) and proteins, and VIM level reductions decrease ERK activation in LMP-1-positive NPC cells. We aim to investigate the interaction between LMP-1 and VIM proteins, and the relationship between LMP-1 and VIM interaction, cell apoptosis, and integrity of lipid rafts. Quantitative fluorescence resonance energy transfer (FRET) is adopted in our study, which is the only non-intrusive method for dynamically monitoring protein-protein interaction in living cells.

We synthesize molecular biosensors for monitoring LMP-1 and VIM proteins in live cells by connecting a kind of cyan fluorescence protein Cerulean and a yellow fluorescence protein Venus to carboxy terminals of VIM and LMP-1 respectively. VIM cellular distribution differences between cells only expressing VIM or co-expressing VIM and LMP-1 are analyzed by employing VIM-Cerulean and LMP-1-Venus molecular biosensors. Additionally, quantitative FRET is utilized to investigate the interaction between LMP-1 and VIM by transfecting VIM-Cerulean and LMP-1-Venus into CNE1 cells, one of the well-differentiated nasopharyngeal squamous carcinoma cell lines. First, the FRET imaging platform is performed on a wide-field microscope with three imaging channels, including donor [AT435/425?445 nm (excitation), AT455DC, ET480/465?495 nm (emission), Chroma], acceptor [AT495/485?505 nm (excitation), AT515DC, ET540/525?555 nm (emission), Chroma], and FRET [AT435/425?445 nm (excitation), AT515DC, ET540/525?555 nm (emission), Chroma]. Next, we calibrate two acceptor bleedthrough coefficients (a and b), two donor bleedthrough coefficients (c and d), and the G factor, defined as the ratio of the sensitized emission to the corresponding amount of donor recovery in the donor imaging channel after acceptor photobleaching, by adopting partial acceptor photobleaching assays and standard FRET constructs, including Venus (Addgene #27794), Cerulean (Addgene #15214), and C32V (Cerulean-32-Venus, Addgene #29396). Fluorescence images of Venus and Cerulean constructs obtained from the donor, acceptor, and FRET channels are employed for bleedthrough coefficients a, b, c, and d. Fluorescence images of C32V constructs obtained from the donor, acceptor, and FRET channels before or after photobleaching are leveraged for the G factor. To validate the FRET imaging platform, we measure the FRET efficiency of the standard FRET construct VCV (Venus-5-Cerulean-5-Venus, Addgene #27788) by the 3-cube FRET (E-FRET) method via mapping fluorescence images of VCV constructs on three imaging channels. Furthermore, E-FRET imaging is adopted to map the interaction between LMP-1 and VIM in single living CNE1 cells. The fluorescence images of a single CNE1 cell co-expressing LMP-1-Venus and VIM-Cerulean from different imaging channels are obtained, with FRET efficiency from different regions of single cell calculated pixel by pixel based on per-pixel fluorescence intensity on different channels. Additionally, Methyl-β-Cyclodextrin (MβCD) is utilized to disrupt lipid rafts and induce cell death in our study. Three regions of interest (ROIs) are selected within a single cell co-expressing LMP-1-Venus and VIM-Cerulean, and then E-FRET imaging is employed to map FRET efficiency from three ROIs. FRET efficiency difference from 30 ROIs within ten cells before or after treatment with 40 mol/mL MβCD for 20 min is analyzed by employing the Student’s t-test via Microsoft Excel. Meanwhile, the 3-(4,5-dimethylthiazol-2-yl)-2, 5-diphenyl-tetrazolium bromide (MTT) assay is also adopted to evaluate the activity of cells involvement of LMP-1 and VIM interaction. Time-lapse FRET imaging is leveraged to evaluate FRET efficiency from five ROIs within a single cell during the 30 min treatment period.

We employ VIM-Cerulean and LMP-1-Venus molecular biosensors to monitor the cellular distribution of VIM proteins. Fluorescence images show that VIM proteins are observed in the plasma membrane, cytoplasm, and nuclear. The majority of VIM is localized in perinuclear rings in LMP-1-negative cells and reorganized into a single patch in cytoplasm in LMP-1-positive cells. Additionally, the distribution of LMP-1 is similar to that of VIM, and LMP-1 patches partially colocalize with VIM patches. These results indicate an interaction between LMP-1 and VIM may occur in NPC cells. Four bleedthrough coefficients are a=(0.13±0.005), d=(0.50±0.015), b≈c≈0 from 40 cells, and G factoris G=(3.0±0.28) from 40 cells by our imaging platform. Next, the FRET efficiency of VCV constructs measured by our imaging platform is (66.24%±1.6%) from 45 cells. Our data is similar to that from previous studies. Furthermore, direct interaction between LMP-1 and VIM can be observed by E-FRET imaging in live cells, and the region of LMP-1 and VIM interaction colocalizes with that of VIM assembling. These observations demonstrate that the localization of VIM proteins may be affected by LMP-1 and VIM interaction. FRET efficiency of ROIs in CNE1 cells expressing VIM-Cerulean and LMP-1-Venus after treatment with 40 mol/mL MβCD for 20 min significantly is stronger than that before treatment (p<0.001), and statistic data is obtained from 30 ROIs in 10 cells. Data from time-lapse E-FRET imaging validates this conclusion, but responses of various regions from a single cell to MβCD are somewhat different. Therefore, the interaction between LMP-1 and VIM is not dependent on the integrity of lipid rafts. Data from MTT assays shows that the activity of LMP1-1 and VIM interaction-negative cells (CNE1) is markedly weaker than that of LMP1-1 and VIM interaction-positive cells (CNE1-LMP-1) after treatment with 40 mol/mL MβCD for 30 min (p<0.001). Therefore, it is strongly suggested that interaction between LMP-1 and VIM can inhibit cell death.

We analyze the interaction between LMP-1 and VIM in NPC cells by the quantitative FRET method combined with molecular fluorescence biosensors. Some new points associated with LMP-1 and VIM are presented as follows. The interaction between LMP-1 and VIM can change the cellular distribution of VIM and prevent cell death. These results indicate that protein-protein interaction during cell death can be accurately demonstrated by monitoring FRET efficiency, and protein distribution in different types of cells can also be analyzed by this method. Finally, our study provides some novel insights into the anti-death mechanisms of LMP-1 proteins.