Rho kinase phosphorylation promotes ezrin-mediated metastasis in hepatocellular carcinoma

Abstract

During progression of hepatocellular carcinoma, multiple genetic and epigenetic alterations act to posttranslationally modulate the function of proteins that promote cancer invasion and metastasis. To define such abnormalities that contribute to liver cancer metastasis, we carried out a proteomic comparison of primary hepatocellular carcinoma and samples of intravascular thrombi from the same patient. Mass spectrometric analyses of the liver cancer samples revealed a series of acidic phospho-isotypes associated with the intravascular thrombi samples. In particular, we found that Thr567 hyperphosphorylation of the cytoskeletal protein ezrin was tightly correlated to an invasive phenotype of clinical hepatocellular carcinomas and to poor outcomes in tumor xenograft assays. Using phospho-mimicking mutants, we showed that ezrin phosphorylation at Thr567 promoted in vitro invasion by hepatocarcinoma cells. Phospho-mimicking mutant ezrinT567D, but not the nonphosphorylatable mutant ezrinT567A, stimulated formation of membrane ruffles, suggesting that Thr567 phosphorylation promotes cytoskeletal-membrane remodeling. Importantly, inhibition of Rho kinase, either by Y27632 or RNA interference, resulted in inhibition of Thr567 phosphorylation and a blockade to cell invasion, implicating Rho kinase-ezrin signaling in hepatocellular carcinoma cell invasion. Our findings suggest a strategy to reduce liver tumor metastasis by blocking Rho kinase-mediated phosphorylation of ezrin.

Figure 1

Proteomic analyses of HCC progression. A, cellular proteins from primary and metastatic HCC were labeled with Cy5 and Cy3, respectively, mixed, and fractionated by 2D SDS-PAGE. Proteins from metastatic HCC are labeled in green whereas the proteins from primary HCC are seen in red. Merging shows that the proteins from 2 stages of distinct tumor cells share some similarity and also exhibit some dramatic difference in protein biochemical properties. Among a list of proteins whose expression and modification levels changed between HCC and paired cancer embolus, the 2D separation patterns seen among 21 paired samples were exemplified here. B, in the metastatic tumor sample, ezrin contains multiple acidic isotypes appearing as a total of 8 separate spots. Four of these 8 spots are superimposed onto that of ezrin from primary HCC, which seemed as yellow in the merge. Our mass spectrometric analysis indicated that the acidic spots of ezrin contain phospho-Thr567. C, Western blotting analyses of ezrin, phospho-Thr567 and actin confirmed a specific hyperphosphorylation of ezrin-Thr567 in metastatic HCC. D, protein samples from cancer embolus were separated by 2D SDS-PAGE and transferred onto a nitrocellulose membrane. The region containing ezrin spots was probed with ezrin monoclonal antibody 4A5 followed by phospho-Thr567–specific rabbit antibody. The phospho-Thr567 antibody marks spots 2 and 3 of ezrin from the cancer embolus preparation, validating that they contain phospho-Thr567. In general, spot #2 represents the most heavily phosphorylated Thr567 spots among 5 well-separated spots. Relative spot densities of phospho-Thr567 over ezrin protein from 5 well-separated spots were quantified from 7 paired samples and expressed as mean ± SE. *, P < 0.001 compared with those of spot #1, spot #4, and spot#5, respectively.

Figure 2

Phospho-Thr567 is highly expressed in metastatic HCC. A, sections of normal tissue (a), HCC tissue (b), and metastatic HCC sample (c) were probed for phospho-Thr567 antibody. In normal liver, little positive staining was observed in the epithelial cells (a), consistent with early observation in mice. Phospho-Thr567 staining is visualized in a few HCC tissue cells (b). However, positive staining of Phospho-Thr567 is prominently seen in most cells in the invasive tumor sample (c). Bar, 30 μm. B, confocal microscopic analyses of double immunofluorescence of phospho-Thr567 (green) versus ezrin expression (red) in normal and HCC samples. C, quantitation of phospho-ezrinT567 levels in normal liver and HCC tissues. The pixel intensities of phospho-Thr567 (normalized to the ezrin signal) in normal, primary, and metastatic HCC cells were measured. Values represent the means ± S.E. of at least 250 different cells from each categories of 5 different paired samples. *, P < 0.001 compared with those of embolus samples.

Figure 3

Phospho-Thr567 promotes invasion of HepG2 cells. A, proteins from HepG2 cells transfected with one of several GFP-ezrin constructs, including wild-type ezrin, T567A substitution mutant, and T567D substitution mutant. 30–36 hours posttransfection, the cells were collected and the cellular proteins were solubilized and fractionated by SDS-PAGE followed by Western analyses of CFP-ezrin and actin. Note that both wild-type and mutant proteins were expressed at comparable levels. B, HepG2 cells infected with adenoviral constructs for overexpressing CFP-ezrin proteins were examined in the invasion assay. a, wild-type CFP-ezrin; b, phospho-mimicking ezrin-T567D; c, nonphosphorylatable ezrin-T567A; d, CFP tag only. Image is a representative of at least 4 independent experiments. C, phospho-Thr567 is required for cell invasion. The number of CFP-positive HepG2 cells passing through the chamber is expressed as a percentage of those cells bearing wild-type CFP-ezrin. In addition, parental cells (uninfected), CFP-infected (CFP) along with ROCK siRNA-treated (ROCK siR) cells were also evaluated. Bars, mean±SE from at least 4 independent experiments. **, P < 0.01 compared with parental cells. *, P > 0.01 compared with parental cells. D, Western blotting analyses of phospho-Thr567 of exogenously expressed CFP-ezrin in scramble siRNA-treated (Ezrin), ROCK siRNA-treated parental cells (ROCK siRNA), CFP-ezrin infected and ROCK siRNA-treated (Ezrin+ROCK siRNA), and parental cells infected with CFP-ezrinT567A. Bars, mean±SE from 4 independent experiments. **, P < 0.01 compared with wild-type CFP-ezrin-infected samples.

Figure 4

ROCK-mediated phospho-Thr567 promotes HCC invasion. A, HepG2 cells infected with adenoviral CFP-ezrin were examined in the invasion assay as described. CFP-ezrin-expressing HepG2 cells were scored in response to different concentrations of Y27632 and then expressed as a percentage of that treated with same volume of vehicle. Bars, mean±SE from three independent experiments. *, P < 0.01. B, aliquots of HepG2 cells were treated with different concentrations of Y27632 before being harvested for SDS-PAGE and Western blotting analyses. Cell extracts were probed with anti–phospho-Thr567 antibody (top panel; phosphor-T567) and anti-ezrin antibody 6H11 as internal controls (bottom panel). C, HepG2 cells were transfected with the Rho kinase siRNA oligonucleotides (25 nmol/L) or scramble oligonucleotides and subjected to SDS-PAGE and immunoblotting analyses. Cell extracts were probed for ROCK (top; Rho kinase), phospho-Thr567 (middle; phospho-T567), and ezrin (internal controls; bottom). D, HepG2 cells infected with adenoviral constructs for overexpressing CFP-ezrin proteins were examined for membrane ruffles using FITC-phalloidin staining. a, CFP vector; b, wild-type ezrin; c, ezrin-T567A; d, ezrin-T567D. Note that T567D stimulates membrane ruffles (d, arrows). Image is a representative of 4 independent experiments.

Figure 5

Phosphorylation of Thr567 promotes HCC metastasis in mouse xenografts. A, representative images of the livers of tumor-bearing mice infected with CFP, wild-type ezrin (WT), ezrin-T567A (T567A), and ezrin-T567D (T567D), respectively. B, H&E staining for the xenografts showing the tumor margin (C; parental cells, arrow in 100X). Ezrin-T567D (TD) expression led to massive metastasis in the livers of the mice as compared with ezrin (WT) and ezrin-T567A (TA) expression. Magnified images (400X) are also shown. C, the number of metastatic nodules under gross inspection in the liver of each mouse bearing MHCC97-H xenografts expressing CFP, wild-type (wt), ezrin-T567D (T567D), and ezrin-T567A(T567A). **, P < 0.001 compared with the number of metastatic nodules seen in the livers of CFP vector, wild-type CFP-ezrin, and CFP-ezrinT567A-infected xenografts. D, body weight of tumor-bearing mice measured 9 weeks after tumor xenografts. Bars, mean ±SD.