Advertisement

Expression of Intercellular Adhesion Molecule 1 by Hepatocellular Carcinoma Stem Cells and Circulating Tumor Cells

Published:February 04, 2013DOI:https://doi.org/10.1053/j.gastro.2013.01.046

      Background & Aims

      Intercellular adhesion molecule 1 (ICAM-1) is believed to be involved in metastasis of hepatocellular carcinoma (HCC) cells. Cancer stem cells promote tumor relapse and metastasis. We investigated whether ICAM-1 is a marker of HCC stem cells.

      Methods

      Sphere formation and tumor formation assays were performed to investigate the stem cell properties of ICAM-1+ cells in vitro and in vivo. A specific targeting system that inhibits ICAM-1 expression and hepatitis B virus transgenic mice (M-TgHBV) were used to investigate whether inhibition of ICAM-1 reduced tumor incidence and metastasis in vivo. We used real-time polymerase chain reaction and immunoblot analysis to assess ICAM-1 and Nanog expression in tumor cell lines, and flow cytometry analysis was used to investigate ICAM-1 expression in HCC and blood samples.

      Results

      ICAM-1 was expressed on a minor cell population in HCC tumor cell lines, as well as in tumor tissues and circulating tumor cells isolated from patients and transgenic mice. ICAM-1+ tumor cells had greater sphere-forming and tumorigenic capacities and increased expression of stemness-related genes compared with ICAM-1 tumor cells. The specific inhibition of ICAM-1 reduced formation and metastasis in M-TgHBV mice. ICAM-1 was found to be a marker of circulating tumor cells from patients and M-TgHBV mice. Increased numbers of CD45ICAM-1+ cells in blood samples of patients with HCC correlated with worse clinical outcomes. The stem cell transcription factor Nanog regulated expression of ICAM-1 in HCC stem cells.

      Conclusions

      ICAM-1 is a marker of HCC stem cells in humans and mice; ICAM-1 inhibitors slow tumor formation and metastasis in mice. ICAM-1 expression is regulated by the stem cell transcription factor Nanog.

      Keywords

      Abbreviations used in this paper:

      AFP (α-fetoprotein), ChIP (chromatin immunoprecipitation), CSC (cancer stem cell), CTC (circulating tumor cell), HCC (hepatocellular carcinoma), ICAM-1 (intercellular adhesion molecule 1), K19 (cytokeratin 19), mRNA (messenger RNA), PCR (polymerase chain reaction), P# (patient number), Sc (pAFP–ICAM-1–scramble), SD (standard deviation), Sh (pAFP–ICAM-1–short hairpin RNA), shRNA (short hairpin RNA)
      Hepatocellular carcinoma (HCC) is the 5th most common cancer in the world and accounts for more than 90% of human liver cancers. Hundreds of thousands of deaths result from HCC worldwide every year, and as many as 90% of these cancer-associated deaths are related to metastasis.
      • Jemal A.
      • Bray F.
      • Center M.M.
      • et al.
      Global cancer statistics.
      Intercellular adhesion molecule 1 (ICAM-1; CD54), a 90-kilodalton cell surface glycoprotein of the immunoglobulin superfamily, is believed to be responsible for HCC metastasis.
      • van de Stolpe A.
      • van der Saag P.T.
      Intercellular adhesion molecule-1.
      Previous studies have shown that ICAM-1 is expressed on hepatocytes in cancerous areas but not on hepatocytes in noncancerous areas.
      • Momosaki S.
      • Yano H.
      • Ogasawara S.
      • et al.
      Expression of intercellular adhesion molecule 1 in human hepatocellular carcinoma.
      The expression of ICAM-1 has been reported to mediate adhesion-dependent cell-cell interactions and facilitate the movement (or retention) of cells through the extracellular matrix,
      • Lawson C.
      • Wolf S.
      ICAM-1 signaling in endothelial cells.
      • Zimmerman T.
      • Blanco F.J.
      Inhibitors targeting the LFA-1/ICAM-1 cell-adhesion interaction: design and mechanism of action.
      • Duperray A.
      • Languino L.R.
      • Plescia J.
      • et al.
      Molecular identification of a novel fibrinogen binding site on the first domain of ICAM-1 regulating leukocyte-endothelium bridging.
      and it has been shown to be positively correlated with tumor size and poor prognosis in HCC.
      • Sun J.J.
      • Zhou X.D.
      • Liu Y.K.
      • et al.
      Invasion and metastasis of liver cancer: expression of intercellular adhesion molecule 1.
      • Sun J.J.
      • Zhou X.D.
      • Zhou G.
      • et al.
      Expression of intercellular adhesive molecule-1 in liver cancer tissues andliver cancer metastasis.
      Recently, it was observed that ICAM-1 is expressed on stem cells, such as bone marrow mesenchymal stem cells, adipose stem cells, periodontal ligament stem cells, and placenta mesenchymal stem cells,
      • Assis A.C.
      • Carvalho J.L.
      • Jacoby B.A.
      • et al.
      Time-dependent migration of systemically delivered bone marrow mesenchymal stem cells to the infarcted heart.
      • De Francesco F.
      • Tirino V.
      • Desiderio V.
      • et al.
      Human CD34/CD90 ASCs are capable of growing as sphere clusters, producing high levels of VEGF and forming capillaries.
      • Brooke G.
      • Tong H.
      • Levesque J.P.
      • et al.
      Molecular trafficking mechanisms of multipotent mesenchymal stem cells derived from human bone marrow and placenta.
      • Sununliganon L.
      • Singhatanadgit W.
      Highly osteogenic PDL stem cell clones specifically express elevated levels of ICAM1, ITGB1 and TERT.
      in addition to endothelial cells and epithelial cells.
      • van de Stolpe A.
      • van der Saag P.T.
      Intercellular adhesion molecule-1.
      Based on these findings, ICAM-1 is now considered a mesenchymal stem cell and periodontal ligament stem cell marker.
      • Sununliganon L.
      • Singhatanadgit W.
      Highly osteogenic PDL stem cell clones specifically express elevated levels of ICAM1, ITGB1 and TERT.
      • Strakova Z.
      • Livak M.
      • Krezalek M.
      • et al.
      Multipotent properties of myofibroblast cells derived from human placenta.
      Although it is known that cancer stem cells (CSCs) have crucial roles in cancer relapse and metastasis, whether ICAM-1 is expressed on HCC CSCs remains unclear.
      CSCs, a subset of cancer cells with features of stem cells such as self-renewal ability and pluripotency, are believed to be responsible for tumor relapse and metastasis.
      • Dalerba P.
      • Cho R.W.
      • Clarke M.F.
      Cancer stem cells: models and concepts.
      • Al-Hajj M.
      Cancer stem cells and oncology therapeutics.
      • Wicha M.S.
      Cancer stem cells and metastasis: lethal seeds.
      In HCC, CSCs were first defined as a side population. Haraguchi et al identified side population cells that possess stem cell properties such as self-renewal, pluripotency, and chemoresistance.
      • Haraguchi N.
      • Utsunomiya T.
      • Inoue H.
      • et al.
      Characterization of a side population of cancer cells from human gastrointestinal system.
      Recently, CSCs were characterized based on their expression of various surface molecules, such as CD133, CD90, EpCAM, and CD44.
      • Ma S.
      • Chan K.W.
      • Hu L.
      • et al.
      Identification and characterization of tumorigenic liver cancer stem/progenitor cells.
      • Yang Z.F.
      • Ho D.W.
      • Ng M.N.
      • Lau C.K.
      • et al.
      Significance of CD90+ cancer stem cells in human liver cancer.
      • Yamashita T.
      • Ji J.
      • Budhu A.
      • et al.
      EpCAM-positive hepatocellular carcinoma cells are tumor-initiating cells with stem/progenitor cell features.
      • Zhu Z.
      • Hao X.
      • Yan M.
      • et al.
      Cancer stem/progenitor cells are highly enriched in CD133+CD44+ population in hepatocellular carcinoma.
      Using these cell surface molecules, HCC CSCs were sorted from HCC cell lines, tissues, and blood samples and found to be capable of self-renewal and tumor initiation and to be resistant to chemotherapeutic drugs. Experimental procedures targeting cells expressing CSC markers have been shown to reduce tumor incidence and metastasis in vivo, indicating that eliminating CSCs may be an efficient therapeutic strategy.
      In the present study, we detected ICAM-1 expression in a minor cell population found in HCC tumor cell lines, tumor tissues, and circulating tumor cells. ICAM-1+ tumor cells displayed enhanced sphere-forming and tumorigenic capacities and elevated expression of stemness-related genes, including Nanog and Oct4, compared with ICAM-1 tumor cells. Moreover, the inhibition of ICAM-1 reduced tumor initiation and metastasis in vivo. Additionally, we found that Nanog transcribes ICAM-1 expression in CSCs. Based on these findings, we propose that ICAM-1 is a functional CSC surface marker in HCC that is regulated by the stem cell transcription factor Nanog.

      Materials and Methods

       Samples

      Human tumor tissues and blood samples were obtained from patients with HCC at the Eastern Hepatobiliary Surgery Hospital after obtaining informed consent. The follow-up procedures applied to these patients have been described in a previous report.
      • Shuqun C.
      • Mengchao W.
      • Han C.
      • et al.
      Tumor thrombus types influence the prognosis of hepatocellular carcinoma with the tumor thrombi in the portal vein.
      Overall survival and disease-free survival were defined as previously described.
      • Zhu X.D.
      • Zhang J.B.
      • Zhuang P.Y.
      • et al.
      High expression of macrophage colony-stimulating factor in peritumoral liver tissue is associated with poor survival after curative resection of hepatocellular carcinoma.
      Female nude mice (4–6 weeks old) were purchased from the Transgenic Animal Research Center, Second Military Medical University. All mice were maintained in a pathogen-free facility and used in accordance with the institutional guidelines for animal care.

       Quantitative Real-Time Polymerase Chain Reaction

      Total RNA was isolated from cell lines and clinical samples with TRIzol reagent according to the manufacturer's protocol (Invitrogen, Carlsbad, CA). Reverse-transcription reactions were conducted with oligo(dT) 18 primers and random primers according to the instructions of the manufacturer of the M-MLV Reverse Transcriptase Kit (Invitrogen). Real-time polymerase chain reaction (PCR) was performed with SYBR Premix Ex Taq (Takara Bio Inc, Otsu Shiga, Japan) using the StepOnePlus Real-Time PCR System (Applied Biosystems, Foster City, CA). The primers used in these assays are listed in Supplementary Table 1. The gene expression levels were calculated relative to the expression of β-actin in tumor cell lines or clinical samples using the 2−ΔΔCt method.

       Western Blot Analysis

      The total soluble proteins (100 μg) extracted from the samples were resolved in 10% sodium dodecyl sulfate/polyacrylamide gels and transferred electrophoretically to a polyvinylidene fluoride membrane. The blots were blocked with 5% skim milk and then incubated with primary antibodies (Supplementary Table 2). The blots were then incubated with an anti-rabbit secondary antibody (Santa Cruz Biotechnology, Santa Cruz, CA) and visualized by enhanced chemiluminescence.

       Single-Cell Sorting

      Single-cell suspensions were obtained from clinical samples by digesting tumor tissues with type IV collagenase (Gibco BRL, Grand Island, NY) for 1 to 3 hours at 37°C, followed by filtration through a 100-μm cell strainer (BD Biosciences, San Jose, CA). To isolate ICAM-1+ cell populations, single cells from cell lines or clinical samples were stained with a PE-conjugated ICAM-1 antibody (eBioscience, San Diego, CA) and with the corresponding isotype control. The samples were analyzed and sorted on a FACSAria cell sorter (BD Biosciences). The positive and negative populations were selected for the following experiments.

       Flow Cytometric Analysis

      The antibodies used in these analyses are listed in Supplementary Table 2. The cells were incubated with the antibodies in phosphate-buffered saline containing 1% bovine serum albumin and 0.1% sodium azide. The corresponding isotype immunoglobulins were used as controls. The data were analyzed with a FACSCalibur flow cytometer and CellQuest software (BD Biosciences).

       Chromatin Immunoprecipitation Assay

      Cells were processed for chromatin immunoprecipitation (ChIP) assays using a chromatin immunoprecipitation assay kit (Millipore Corp, Billerica, MA) according to the manufacturer's protocol. Briefly, the cells were cross-linked with 1% formaldehyde for 10 minutes at 37°C and lysed with sodium dodecyl sulfate lysis buffer. The lysate pellets were sonicated with a sonic dismembrator (Fisher Scientific, Waltham, MA). Protein-DNA complexes were immunoprecipitated with the appropriate antibodies (Supplementary Table 2). The immunoprecipitates were dissolved in 20 μL water for PCR analysis. Standard PCR amplifications were performed using Taq PCR Master Mix (Takara Bio Inc) with the specific primers listed in Supplementary Table 1.

       Statistical Analysis

      Student t tests were used to compare 2 groups unless otherwise indicated (χ2 test). Categorical data were analyzed using Fisher exact test, and quantitative variables were analyzed using t tests or Pearson's correlation test. Survival was calculated with the log-rank test. The Cox regression model was used to perform multivariate analysis. P < .05 was considered statistically significant.
      For a description of other materials and methods used in this study, see Supplementary Materials and Methods.

      Results

       ICAM-1+ Tumor Cells Possess Characteristics of Stem/Progenitor Cells

      To investigate whether ICAM-1 can be used as a CSC marker, we first determined whether an ICAM-1+ cell population was present in tumor cell lines. Flow cytometry analysis showed that approximately 5% of Huh7 cells and 7% of Hep3B cells expressed ICAM-1 (Figure 1A). We next examined whether the ICAM-1+ cells exhibited intrinsic properties of stem cells. For this purpose, ICAM-1+ cells were isolated from tumor cell lines, and the expression of stemness-related genes, including sox2, nanog, oct4, and β-catenin, was assessed at the messenger RNA (mRNA) level. Compared with ICAM-1 cells, ICAM-1+ cells expressed higher levels of these 4 genes, of which β-catenin was increased the least (Figure 1B, upper left panel). This finding was further confirmed with real-time PCR assays (Figure 1B, right panel) and Western blotting (Figure 1B, lower left panel).
      Figure thumbnail gr1
      Figure 1ICAM-1+ tumor cells possess stem cell properties. (A) ICAM-1+ tumor cells were observed among Huh7 cells (4.82% ± 0.45%) and Hep3B cells (7.27% ± 0.29%). iso, isotype control for ICAM-1 antibody. The data are shown as the mean ± SD from at least 3 independent experiments. (B) Reverse-transcription PCR (upper left panel), Western blot (lower left panel), and real-time PCR (right panel) analyses showed that ICAM-1+ tumor cells sorted from Huh7 cells overexpressed several stemness-related genes (sox2, oct4, nanog, and β-catenin) compared with ICAM-1 tumor cells. *P < .05; **P < .01. Error bars represent the standard deviation (SD) of data obtained from at least 3 independent experiments. (C) A representative image of spheres formed by ICAM-1+ or ICAM-1 tumor cells sorted from Huh7 and Hep3B cell lines. Sphere formation assays showed that ICAM-1+ tumor cells sorted from both Huh7 and Hep3B cells exhibited an enhanced sphere-forming capacity compared with corresponding ICAM-1 tumor cells. **P < .01. Error bars represent the SD of data obtained from at least 3 independent experiments (scale bar = 100 μm). (D) Representative image showing that the ICAM-1+ cells sorted from Huh7 and Hep3B cells induced tumor formation. See also . The right flanks of mice were injected with ICAM-1+ cells, whereas the left flanks were injected with ICAM-1 tumor cells (right). (E) Similar histologic features were detected via H&E staining in tumor xenografts generated by total tumor cells (the cultured heterogeneous tumor cells) and those generated by isolated ICAM-1+ cells. Scale bars = 50 μm. The low-power and wider histologic views are provided as .
      To investigate the CSC properties of ICAM-1+ cells in vitro, a sphere formation assay was performed. Isolated cells were cultured in serum-free epidermal growth factor/basic fibroblast growth factor–supplemented medium. One week later, many hepatospheres were observed in cultures of ICAM-1+ cells isolated from both Huh7 and Hep3B cells, whereas very few spheres formed in the ICAM-1 cell cultures (Figure 1C). We next inoculated nude mice subcutaneously with ICAM-1+ cells and ICAM-1 cells to investigate their tumorigenicity in vivo. A significant difference in tumor incidence was observed between the mice inoculated with ICAM-1+ cells and those inoculated with ICAM-1 cells (Table 1). Three months postinoculation, as few as 1000 ICAM-1+ cells were sufficient to generate tumors in nude mice, whereas the ICAM-1 cells did not induce tumor formation, even when 1 × 104 cells were injected (Figure 1D and Table 1). Histologic staining identified similar histologic features in tumor xenografts generated by ICAM-1+ cells and those generated by total tumor cells (Figure 1E).
      Table 1Tumorigenic Capacity of ICAM-1+ Cells From Cell Lines
      Cell linePhenotypesNo. of injected cellsNo. of mice showing tumor formation/total No. of mice injected with cells
      2 mo3 mo4 mo
      Huh7ICAM-1+5000/50/50/5
      10000/52/54/5
      50000/53/54/5
      1040/54/55/5
      ICAM-15000/50/50/5
      10000/50/50/5
      50000/50/50/5
      1040/50/51/5
      Total cells
      Heterogeneous tumor cells, from which the ICAM-1+ and ICAM-1− cell subpopulations were sorted, respectively.
      1065/5
      Hep3BICAM-1+5000/50/50/5
      10000/51/53/5
      50000/52/54/5
      1040/55/5
      ICAM-15000/50/50/5
      10000/50/50/5
      50000/50/50/5
      1040/50/52/5
      Total cells
      Heterogeneous tumor cells, from which the ICAM-1+ and ICAM-1− cell subpopulations were sorted, respectively.
      1065/5
      a Heterogeneous tumor cells, from which the ICAM-1+ and ICAM-1 cell subpopulations were sorted, respectively.
      We next assessed the expression of known liver CSC markers, including CD24, CD90, CD133, EpCAM, and CD44, on ICAM-1+ cells sorted from Huh7 and Hep3B cells by flow cytometry. As shown in Supplementary Figure 1, ICAM-1 expression overlapped with the expression of these markers. Almost all CD24, CD44, and CD90 were expressed on ICAM-1+ cells sorted from Huh7 cells, and approximately half of the EpCAM+ or CD133+ cells were ICAM-1+ (Supplementary Figure 1A). In Hep3B cells, these markers were mostly expressed on ICAM-1+ cells (Supplementary Figure 1B).
      Collectively, these results suggest that ICAM-1+ cells display CSC properties.

       Clinical Significance of ICAM-1+ HCC Cells

      Having shown that ICAM-1+ cells derived from HCC cell lines possess CSC characteristics, we next investigated the role of ICAM-1+ HCC cells in patients with HCC. Because ICAM-1 has been reported to be expressed in certain other cell types, such as endothelial cells, epithelial cells, and fibroblasts,
      • van de Stolpe A.
      • van der Saag P.T.
      Intercellular adhesion molecule-1.
      single-cell suspensions were first prepared using fresh HCC tumor tissues from patient 30 (P#30) and patient 50 (P#50) and cultured in vitro. HCC cells were harvested from these primary cultures, and the ICAM-1 expression pattern was then evaluated by flow cytometry. As shown in Figure 2A, the ICAM-1+ cell subpopulations were found to represent less than 2% of the cultured primary HCC cells. The positive expression of cytokeratin 19 (K19) and α-fetoprotein (AFP) ensured that these ICAM-1+ cells were HCC cells (Supplementary Figure 2). To investigate the CSC characteristics of ICAM-1+ tumor cells, a sphere formation assay was performed. As shown in Figure 2B (left panel), the ICAM-1+ cells isolated from P#30 and P#50 showed higher sphere-forming capacities than the ICAM-1 cells from the same patients. Next, the properties of cells in the hepatospheres were further evaluated. The expression of high levels of hepatic stem cell markers (TACSTD1, MYC, and hTERT) and low levels of mature hepatocyte markers (UGT2B7 and CYP3A4) indicated that the cells in the hepatospheres might be hepatic stem cells (Supplementary Figure 3A). The positive immunostaining of hepatic stem markers (AFP and K19) further supported these findings (Supplementary Figure 3B). Subsequently, tumor formation assays were performed. As few as 2500 ICAM-1+ tumor cells were sufficient to induce tumor formation in nude mice, whereas no tumors were observed after 90 days when equivalent numbers of ICAM-1 cells were used (Table 2 and Figure 2B, middle panel). H&E staining confirmed the histologic features of the tumor xenografts (Figure 2B, right panel). The expression of known liver stem cell markers was also assessed in ICAM-1+ cells from tumor tissues. Except for CD90, which was not observed in the HCC tissues of these 2 patients, most of the results were consistent with those in the tumor cell lines (Supplementary Figure 4).
      Figure thumbnail gr2
      Figure 2Clinical relevance of ICAM-1 expression. (A) Flow cytometry analysis revealed ICAM-1+ cell subpopulations in single-cell suspensions from tumor tissues from patients with HCC (1.57% ± 0.02% in P#30, 0.99% ± 0.03% in P#50). The data are shown as the mean ± SD from at least 3 independent experiments. (B) Sphere formation assays showed that ICAM-1+ tumor cells from patients' tumor tissues displayed an enhanced sphere-forming capacity compared with the corresponding ICAM-1 tumor cells (left panel). **P < .01. Error bars represent the SD of data obtained from at least 3 independent experiments. Representative image showing that ICAM-1+ cells from tumor tissues induced tumor formation. The right flanks of mice were injected with ICAM-1+ cells, whereas the left flanks were injected with ICAM-1 tumor cells (middle panel). Representative image of H&E staining showing that the tumor xenografts displayed histologic features characteristic of HCC (right panel) (scale bar = 50 μm). (C) Representative image showing that CD45ICAM-1+ cells (0.3% ± 0.02%) were found in the blood from a patient with HCC (P#30) but not (0%) in the normal control blood (Con). iso, isotype control for ICAM-1/CD45 antibody. The data are shown as the mean ± SD from at least 3 independent experiments. (D) Sphere formation assays showed that CD45ICAM-1+ tumor cells from patients' blood displayed an enhanced sphere-forming capacity compared with the corresponding CD45ICAM-1 tumor cells (left panel). **P < .01. Error bars represent the SD of data obtained from at least 3 independent experiments. Representative image showing that CD45ICAM-1+ cells from patients' blood induced tumor formation. The right flanks of mice were injected with CD45ICAM-1+ cells, whereas the left flanks were injected with CD45ICAM-1 tumor cells (middle panel). H&E staining revealed that tumor xenografts displayed histologic features characteristic of HCC (right panel) (scale bar = 50 μm). (E) Kaplan–Meier curves for disease-free and overall survival were compared based on the frequency of CD45ICAM-1+ cells in the patients' blood. Patients with a high frequency of CD45ICAM-1+ cells (⩾0.157%) exhibited significantly shorter disease-free survival (P < .0001, log-rank test) and shorter overall survival (P = .013, log-rank test) than those with a low frequency of CD45ICAM-1+ cells (<0.157%).
      Table 2Tumorigenic Capacity of ICAM-1+ Cells and CD45ICAM-1+ Cells From Tumor Specimens and Blood Samples From Patients With HCC
      PatientsPhenotypesNo. of injected cellsNo. of mice showing tumor formation/total No. of mice injected with cells
      2 mo3 mo4 mo
      P#30ICAM-1+5000/50/50/5
      10000/50/50/5
      25000/51/52/5
      50000/51/53/5
      ICAM-15000/50/50/5
      10000/50/50/5
      25000/50/50/5
      50000/50/505
      P#50ICAM-1+5000/50/50/5
      10000/50/505
      25000/50/51/5
      50000/51/52/5
      ICAM-15000/50/50/5
      10000/50/50/5
      25000/50/50/5
      50000/50/50/5
      P#30CD45ICAM-1+5000/51/51/5
      10000/51/51/5
      25000/52/52/5
      50000/52/53/5
      CD45ICAM-15000/50/50/5
      10000/50/50/5
      25000/50/50/5
      50000/50/50/5
      P#50CD45ICAM-1+5000/50/51/5
      10000/50/51/5
      25000/51/52/5
      50000/53/53/5
      CD45ICAM-15000/50/50/5
      10000/50/50/5
      25000/50/50/5
      50000/50/50/5
      Because ICAM-1 expression was previously reported to be correlated with HCC relapse and metastasis,
      • Sun J.J.
      • Zhou X.D.
      • Liu Y.K.
      • et al.
      Invasion and metastasis of liver cancer: expression of intercellular adhesion molecule 1.
      we next sought to determine whether ICAM-1+ tumor cells were present in the blood of patients with HCC. The existence of ICAM-1+ tumor cells was analyzed by flow cytometry. Because ICAM-1 has been reported to be expressed in some lymphocytes,
      • van de Stolpe A.
      • van der Saag P.T.
      Intercellular adhesion molecule-1.
      CD45 staining was performed, and CD45ICAM-1+ cells were defined as nonlymphatic ICAM-1+ cells in the blood.
      • Yang Z.F.
      • Ho D.W.
      • Ng M.N.
      • Lau C.K.
      • et al.
      Significance of CD90+ cancer stem cells in human liver cancer.
      As shown in Figure 2C, 0.30% of the cells in a blood sample from a patient with HCC (P#30) were CD45ICAM-1+ cells, whereas no CD45ICAM-1+ cells were detected in the control blood sample (Con). After confirming that the CD45ICAM-1+ cells isolated from the blood of the patient with HCC were hepatocytes (Supplementary Figure 5), sphere formation and tumor formation assays were performed to investigate whether these cells possessed CSC properties. Isolated CD45ICAM-1+ cells and CD45ICAM-1 cells were cultured in serum-free epidermal growth factor/basic fibroblast growth factor–supplemented medium. More hepatospheres were observed among the CD45ICAM-1+ cells than among the CD45ICAM-1 cells (Figure 2D, left panel). To perform the tumor formation assays, CD45ICAM-1+ cells were purified from the blood of patients with HCC via flow cytometry sorting and then injected subcutaneously into nude mice. Three months after the injections, mice injected with 500 circulating CD45ICAM-1+ cells developed tumor nodules, whereas no tumor formation was detected in mice injected with CD45ICAM-1 cells (Table 2 and Figure 2D, middle panel). Subsequent H&E staining confirmed that tumor xenografts and tumors induced by CD45ICAM-1+ cells from HCC tumors shared similar histologic features (Figure 2D, right panel).
      Next, the circulating CD45ICAM-1+ tumor cells from 60 patients with HCC were quantified by flow cytometry to determine whether their numbers were correlated with the clinical outcome of patients with HCC. The median CD45ICAM-1+ cell frequency (0.157%) was used as the cutoff value. A total of 30 (50.0%) of the 60 cases exhibited more than 0.157% CD45ICAM-1+ cells in their blood. The correlation of the CD45ICAM-1+ tumor cell frequency with clinicopathologic features is summarized in Table 3. The frequency of CD45ICAM-1+ cells was only correlated with age (P = .040, t test). A Kaplan–Meier survival analysis was then performed using the median CD45ICAM-1+ cell frequency as the cutoff value. Patients with a high CD45ICAM-1+ cell frequency (≥0.157%) exhibited a significantly shorter disease-free survival period (P < .0001, log-rank test) and shorter overall survival period (P = .013, log-rank test) than those with a low CD45ICAM-1+ cell frequency (<0.157%) (Figure 2E). A univariate analysis of the CD45ICAM-1+ cell frequency and clinicopathologic factors revealed prognostic significance in all 60 patients. These significant variables (P < .05) were further entered into a multivariate Cox model, and the results showed that a high frequency of CD45ICAM-1+ cells in the blood of patients with HCC was an independent risk factor for a poor outcome, portal vein tumor thrombus, and ascites (Supplementary Table 3). In summary, higher frequencies of CD45ICAM-1+ cells in the blood of patients with HCC correlated with more aggressive tumor behavior and worse clinical outcomes.
      Table 3Relationship Between the Frequency of CD45ICAM-1+ Cells and Clinicopathologic Features
      Clinical featureLow (n = 30)High (n = 30)P value
      Age (y).04
       Median45.555
       Range31–7536–74
      Sex.42
       Male28 (93.3)25 (83.3)
       Female2 (6.7)5 (16.7)
      Hepatitis B e antigen.60
       Negative21 (70.0)23 (76.7)
       Positive9 (30.0)7 (23.3)
      Hepatitis B surface antigen1
       Negative2 (6.7)2 (6.7)
       Positive28 (93.3)28 (93.3)
      Cirrhosis.61
       No3 (10.0)1 (3.3)
       Yes27 (90.0)29 (96.7)
      Tumor size (cm).77
       ≤59 (30.0)8 (26.7)
       >521 (70.0)22 (73.3)
      Tumor no.1
       Single27(90.0)26 (86.7)
       Multiple3 (10.0)4 (13.3)
      Microvascular invasion.52
       Absent25 (83.3)23 (76.7)
       Present5 (16.7)7 (23.3)
      Tumor encapsulation.20
       No9 (30.0)10 (33.3)
       Incomplete4 (13.3)9 (30.0)
       Complete17 (56.7)11 (36.7)
      Micrometastases.27
       Absent22 (73.3)18 (60.0)
       Present8 (26.7)12 (40.0)
      Edmondson grade.49
       II6 (20.0)4 (13.3)
       III24 (80.0)25 (83.3)
       IV0 (0.0)1 (3.3)
      Portal vein tumor thrombus.17
       No27 (90.0)23 (76.7)
       Yes3 (10.0)7 (23.3)
      Ascites.71
       No27 (90.0)25 (83.3)
       Yes3 (10.0)5 (16.7)
      AFP (ng/mL).14
       Negative, ≤2010 (33.3)5 (16.7)
       Positive, >2020 (66.7)25 (83.3)
      HBV DNA (copies/mL).60
       ≤10,00017 (56.7)19 (63.3)
       >10,00013 (43.3)11 (36.7)
      Serum bilirubin (μmol/L)1
       ≤1721 (70.0)21 (70.0)
       >179 (30.0)9 (30.0)
      Serum albumin (g/L).18
       ≤408 (26.7)13 (43.3)
       >4022 (73.3)17 (56.7)
      Serum prealbumin (mg/L).43
       ≤17011 (36.7)14 (46.7)
       >17019 (63.3)16 (53.3)
      Alanine aminotransferase (U/L).20
       ≤4113 (43.3)18 (60.0)
       >4117 (56.7)12 (40.0)
      Aspartate aminotransferase (U/L).60
       ≤3714 (46.7)12 (40.0)
       >3716 (53.3)18 (60.0)
      NOTE. All values are expressed as n (%) unless otherwise indicated. A P value of less than .05 was considered to indicate statistical significance. P values were calculated using Fisher exact test except for age, which was calculated with an unpaired t test.

       Reduction of the Number of ICAM-1+ Cells Inhibits Tumor Incidence and Metastasis In Vivo

      Because higher frequencies of CD45ICAM-1+ cells in the blood of patients with HCC correlated with more aggressive tumor behavior and worse clinical outcomes, we next asked whether the inhibition of ICAM-1 in HCC cells would reduce tumor incidence and metastasis in vivo. To reduce ICAM-1 expression specifically in HCC cells, the AFP core promoter was cloned into the plasmid pAFP–ICAM-1–short hairpin RNA (shRNA) to initiate the expression of shRNA targeting ICAM-1 (Supplementary Figure 6A). After validating the inhibitory effect of pAFP–ICAM-1–shRNA (Sh) on ICAM-1 expression in vitro (Supplementary Figure 6B and C), pAFP–ICAM-1–shRNA and pAFP–ICAM-1–scramble (Sc) were administered to M-TgHBV transgenic mice, which can be used as a model that mimics development of human HCC.
      • Zhang X.
      • Liu S.
      • Hu T.
      • et al.
      Up-regulated microRNA-143 transcribed by nuclear factor kappa B enhances hepatocarcinoma metastasis by repressing fibronectin expression.
      Six months later, the frequency of CD45ICAM-1+ cells in the blood of the mice and the frequency of ICAM-1+ HCC cells in their livers were analyzed by flow cytometry. As shown in Figure 3A (upper panel), the frequency of CD45ICAM-1+ cells in the blood of mice treated with Sh was significantly decreased compared with that in the blood of mice treated with Sc (0.12% vs 0.95%, respectively). The proportion of ICAM-1+ HCC cells in the liver was reduced to 0.93% from 3.36% by treatment with Sh (Figure 3A, lower panel).
      Figure thumbnail gr3
      Figure 3Reduction in the frequency of ICAM-1+ cells inhibits tumor incidence and metastasis in vivo. (A) Inhibition of ICAM-1 expression reduced the frequency of CD45ICAM-1+ cells in the blood (from 0.95% ± 0.03% [Sc] to 0.12% ± 0.003% [Sh]) and livers (from 3.36% ± 0.94% [Sc] to 0.93% ± 0.03% [Sh]) of M-TgHBV transgenic mice. iso, isotype control for ICAM-1/CD45 antibody. The data are shown as the mean ± SD from at least 3 independent experiments. (B) Representative images of H&E staining showing fewer nodal tumors in the livers and lungs of mice treated with Sh compared with those treated with Sc. No lung metastasis was observed in mice treated with Sh (scale bars = 60 μm). (C) The incidence of tumors and local and distant metastases in M-TgHBV transgenic mice treated with Sh was reduced compared with those treated with Sc.
      The necropsies performed on all mice that were killed showed that the mice exhibited regional, intrahepatic, and pulmonary metastases. These metastases comprised small visible colonies distributed around local tumors (Figure 3B). In the Sh group, 5 of 20 transgenic mice developed HCC tumors, whereas 15 mice were found to exhibit HCC in the Sc group (n = 20). Furthermore, in the Sh group, 3 of 5 mice showed liver metastasis and no lung metastases were found, whereas in the Sc group, all 15 mice with HCC tumors displayed liver metastasis and 53.3% (8/15) of the mice exhibited lung metastasis (Figure 3C).
      The results show that inhibition of ICAM-1 expression in tumor cells reduced HCC development and metastasis in vivo, which indicates that ICAM-1 could be used as a therapeutic target.

       Nanog Directly Regulates ICAM-1 Expression in CSCs

      Having documented that ICAM-1 was a potential CSC marker, we sought to elucidate the mechanism regulating ICAM-1 expression in CSCs. Because Sox2, Oct4, and Nanog were reported to have crucial roles in stem cell maintenance and because the ICAM-1+ cells displayed elevated expression of these genes (Figure 1B), we performed a bioinformatic analysis to find the binding sites for these transcription factors in the ICAM-1 promoter. As shown in Figure 4A and Supplementary Figure 7A, 2 Sox2/Oct4 binding sites and 4 Nanog binding sites were found in the promoter of ICAM-1. ChIP experiments were thus performed with CSCs enriched from clinical tumor tissues (P#30 and P#50) and tumor cell lines to determine whether Sox2, Oct4, and Nanog bind to these sites. As shown in Figure 4B and Supplementary Figure 7B, Nanog bound to site 4 in the ICAM-1 promoter, whereas Sox2/Oct4 did not bind the ICAM-1 promoter, indicating that Nanog bound to the ICAM-1 promoter and transcribed ICAM-1 expression.
      Figure thumbnail gr4
      Figure 4ICAM-1 is directly transcribed by Nanog in CSCs. (A) Nanog binds directly to the human ICAM-1 promoter sequence. Bioinformatics analysis found 4 binding sites in the 10-kilobase genomic sequence upstream (−10 kb) of the ICAM-1 gene. (B) ChIP assays in spheres produced from clinical samples from patients (P#30 and P#50) and tumor cell lines (Huh7 and Hep3B). PCR was performed with primers specific for site 4 and the negative site (N site, nonbinding site of Nanog). (C) Overexpression of nanog (Nanog) up-regulated ICAM-1 in Huh7 cells at both the mRNA (left panel) and protein levels (right panel) compared with control (EGFP). **P < .01. Error bars represent the SD of data obtained from at least 3 independent experiments. (D) Decreased nanog levels (Sh) resulted in the down-regulation of ICAM-1 expression in Huh7 cells at both the mRNA (left panel) and protein levels (right panel) compared with control (Sc). **P < .01. Error bars represent the SD of data obtained from at least 3 independent experiments. (E) Immunofluorescence staining revealed the presence of Nanog (green) and ICAM-1 (red) double-positive cells in the spheres produced from Huh7 and Hep3B cells (scale bars = 20 μm). (F) Flow cytometry analysis detected Nanog+ICAM-1+ cell subpopulations in single-cell suspensions from tumor tissues (1.1% ± 0.16% in P#30, 0.84% ± 0.02% in P#50) (left panel) and CD45ICAM-1+Nanog+ cell subpopulations (1.24% ± 0.02%) in patients' blood (right panel). The data are shown as the mean ± SD from at least 3 independent experiments.
      To further test whether ICAM-1 expression was enhanced by Nanog expression, the plasmid pEGFP-nanog containing the human nanog open reading frame was constructed and transfected into Huh7 cells. After validating that Nanog overexpression was induced by pEGFP-nanog transfection (Supplementary Figure 8), ICAM-1 expression was analyzed by real-time PCR. The expression of ICAM-1 was up-regulated approximately 5-fold in the cells transfected with pEGFP-nanog (Nanog) relative to the expression level in cells transfected with the empty pEGFP vector (EGFP) (Figure 4C, left panel). Subsequent Western blot analysis validated the increase in ICAM-1 expression induced by overexpression of Nanog at the protein level (Figure 4C, right panel). We also determined whether the ICAM-1 expression level could be reduced by inhibiting Nanog expression. We constructed the plasmid pLKO-Sh, which expresses shRNAs targeting human nanog, and the plasmid pLKO-Sc, which was used as a control. Compared with the cells transfected with Sc, ICAM-1 expression was significantly reduced at both the mRNA and protein levels in cells transfected with Sh (Figure 4D). Next, immunofluorescence staining was performed to validate the colocalization of Nanog and ICAM-1 in spheres from tumor cell lines. Double immunofluorescence staining showed that ICAM-1 (red) was located near Nanog (green), which localized to the nuclei of the cells (Figure 4E). Furthermore, flow cytometry analysis showed the existence of Nanog+ICAM-1+ tumor cells in tumor samples and blood samples from patients with HCC, indicating that Nanog and ICAM-1 are coexpressed in tumor cells in vivo (Figure 4F).
      Collectively, these results indicate that ICAM-1 expression is directly transcribed by Nanog in CSCs enriched through sphere formation assays.

      Discussion

      In this study, we showed that ICAM-1+ cells isolated from cell lines and clinical samples were capable of inducing tumors in vivo and forming spheres in vitro, indicating that the ICAM-1+ cells exhibited CSC properties and that ICAM-1 could be used as a potential marker for CSCs. The specific inhibition of ICAM-1 expression in vivo reduced HCC development and metastasis, which indicated that ICAM-1 could be used as a therapeutic target. Moreover, ICAM-1 was found to be transcribed directly by Nanog in CSCs.
      Because ICAM-1, identified as a CSC marker, was found to be regulated by Nanog under our experimental conditions, ICAM-1+ tumor cells may represent CSCs with a phenotype involving Nanog expression. Emerging studies have detected the general expression of stemness-related genes, including nanog, in CSCs isolated based on their expression of various surface molecules.
      • Yang Z.F.
      • Ho D.W.
      • Ng M.N.
      • Lau C.K.
      • et al.
      Significance of CD90+ cancer stem cells in human liver cancer.
      • Lee T.K.
      • Castilho A.
      • Cheung V.C.
      • et al.
      CD24(+) liver tumor-initiating cells drive self-renewal and tumor initiation through STAT3-mediated NANOG regulation.
      In these reports, the expression of stemness-related genes was used to characterize the CSC properties of tumor cells positive for the expression of potential CSC markers. Little has been reported about the relationship of these potential CSC markers with the expressed stemness-related genes except that CD24 activated the Nanog promoter in a STAT3-dependent manner.
      • Lee T.K.
      • Castilho A.
      • Cheung V.C.
      • et al.
      CD24(+) liver tumor-initiating cells drive self-renewal and tumor initiation through STAT3-mediated NANOG regulation.
      In addition, the fact that even CSCs isolated from the same tumor can express several different surface markers and exhibit distinct cellular behaviors indicates that tumors may contain distinct subpopulations of CSCs.
      • Yang Y.M.
      • Chang J.W.
      Current status and issues in cancer stem cell study.
      In this study, we found that the expression of ICAM-1, a potential CSC marker, was transcribed by Nanog in tumor cells (Figure 4). Because of the crucial role of Nanog in maintaining stem cells, ICAM-1+ tumor cells may represent the primary CSC subpopulation. In the present study, ICAM-1 expression in HCC cells overlapped with the expression of reported HCC CSC markers, such as CD24, CD44, CD90, CD133, and EpCAM (Supplementary Figure 1, Supplementary Figure 4), and ICAM-1+ tumor cells also expressed hepatic stem cell markers (AFP, K19) (Supplementary Figure 2, Supplementary Figure 5). These data further suggest that ICAM-1+ tumor cells may generate heterogeneous subpopulations that display stem/progenitor cell features and express other reported CSC markers. Under our experimental condition, however, very few ICAM-1+ cells expressed NCAM, a potential biomarker for liver cancers,
      • Tsuchiya A.
      • Kamimura H.
      • Tamura Y.
      • et al.
      Hepatocellular carcinoma with progenitor cell features distinguishable by the hepatic stem/progenitor cell marker NCAM.
      which indicates that ICAM-1 cannot be used as a marker for all CSCs in liver cancer. Additionally, the specific inhibition of ICAM-1 expression in HCC cells in vivo reduced tumor development and metastasis, indicating that ICAM-1 could be used as a therapeutic target. The expression of ICAM-1 in CSCs from other types of tumors should be analyzed; we plan to address this question in the future.
      Finding CD45ICAM-1+ cells in the blood of patients with HCC provided support for the existence of circulating tumor cells (CTCs) and suggested that CTCs also possess CSC properties. CTCs, which possess the antigenic and/or genetic characteristics of a specific tumor type, have been found in the blood of patients with different types of tumors, including breast cancer, colorectal cancer, and lung cancer. Emerging evidence has shown a correlation between CTCs and patient prognosis.
      • Mavroudis D.
      Circulating cancer cells.
      Consistent with previous studies, we detected <0.5% CD45ICAM-1+ tumor cells in patients' blood and none in the blood from healthy subjects (Figure 2C). Furthermore, statistical analysis showed that the frequency of CD45ICAM-1+ cells correlated with patient prognosis (Figure 2E). In addition, CTCs were found to express different genes than primary tumor cells and to possess a metastatic nature.
      • Meng S.
      • Tripathy D.
      • Shete S.
      • et al.
      HER-2 gene amplification can be acquired as breast cancer progresses.
      • Smirnov D.A.
      • Zweitzig D.R.
      • Foulk B.W.
      • et al.
      Global gene expression profiling of circulating tumor cells.
      In the present study, we found that CD45ICAM-1+ circulating cells exhibit tumor induction and sphere-forming capacities, indicating that the circulating tumor cells may possess CSC properties. Yang et al also reported that tumor cells detected in the blood based on their expression of the surface molecules CD90 and CD44 display CSC properties.
      • Yang Z.F.
      • Ho D.W.
      • Ng M.N.
      • Lau C.K.
      • et al.
      Significance of CD90+ cancer stem cells in human liver cancer.
      The CSC theory suggests that this result might be attributable to the significant genetic and phenotypic heterogeneity of CTCs. Additionally, the mesenchymal characteristics of CD45ICAM-1+ cells (E-cadherin/Vimentin+; Supplementary Figure 9) further suggested that this subpopulation in the blood of patients with HCC possessed CSC properties and supported the potential relationship between the frequency of these cells and a poor prognosis.
      Although ICAM-1+ CSCs have been shown to play a crucial role in hepatocarcinogenesis, our study also raises many critical questions. How is ICAM-1 expression regulated in tumor cells? What are the ICAM-1–mediated signaling networks in CSCs? How do we exclude the possibility that these are stellate cell precursors that express ICAM-1
      • Hellerbrand
      • Wang S.C.
      • Tsukamoto H.
      • et al.
      Expression of intracellular adhesion molecule 1 by activated hepatic stellate cells.
      ? Comprehensive investigation will be helpful in revealing the mechanisms involved in the maintenance of ICAM-1+ liver CSCs. In addition, although transcription factors including Nanog and Oct4 are known to be crucial for embryonic stem cell self-renewal and have been shown to contribute much to tumorigenesis,
      • Hu T.
      • Liu S.
      • Breiter D.R.
      • et al.
      Octamer 4 small interfering RNA results in cancer stem cell-like cell apoptosis.
      • Shan J.
      • Shen J.
      • Liu L.
      • et al.
      Nanog regulates self-renewal of cancer stem cells through the insulin-like growth factor pathway in human hepatocellular carcinoma.
      • Jeter C.R.
      • Badeaux M.
      • Choy G.
      • et al.
      Functional evidence that the self-renewal gene NANOG regulates human tumor development.
      little is known about the underlying mechanisms. It should be noted that Nanog is derived from a transcribed pseudogene (NANOGP8) in tumor cells, not from NANOGP1 as in ES cells. This indicates that a distinct network mediated by Nanog might be involved in tumorigenesis. The investigation of the pathways and molecules associated with Nanog in CSCs is necessary to reveal the regulatory circuitry of CSCs. Much of this work is in progress in our laboratory.
      In summary, we identified ICAM-1 as a CSC marker regulated by Nanog, which is generally expressed in CSCs in HCC. Our investigation showed the presence of ICAM-1+ CSCs in HCC cell lines and clinical samples. Cohort analysis confirmed the relevance of circulating ICAM-1+ cells for the prognosis and survival of patients with HCC. Our results indicate that ICAM-1 can be used as a potential CSC marker and may therefore be helpful in developing an effective treatment against cancer. Supplementary Figure 10

      Acknowledgments

      The authors thank Baoyang Hu and Xiaoqing Zhang for helpful discussions and critical readings.

      Supplementary Materials and Methods

       Cell Lines and Cell Culture

      The human HCC cell lines Huh7 and Hep3B were maintained in Dulbecco's modified Eagle medium with high glucose levels (Gibco BRL, Grand Island, NY) supplemented with 10% fetal bovine serum (Gibco BRL), 100 mg/mL penicillin G, and 50 μg/mL streptomycin (Gibco BRL) at 37°C in a humidified atmosphere containing 5% co2.

       Sphere Formation Assay

      A total of 1000 single HCC cells were plated onto 6-well polyHEMA (Sigma, St. Louis, MO)-coated plates. The cells were grown in Dulbecco's modified Eagle medium/F12 (Gibco BRL, Grand Island, NY) supplemented with 20 ng/mL insulin-like growth factor (PeproTech, Rocky Hill, NJ), 1.0 ng/mL basic fibroblast growth factor (Gibco), and 20 ng/mL epidermal growth factor (AbD Serotec Kidlington, UK) for 1 week.

       Transient Transfection

      Transfections were performed using the Lipofectamine 2000 Kit (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. The cells (1–3 × 105) were grown to 50% to 60% confluence in 6-well plates and transfected with 4 μg of plasmids. The cells were harvested to perform subsequent real-time PCR or Western blot analyses 48 hours after transfection.

       Plasmid Construction

      To construct an expression vector for human Nanog (pEGFP-nanog), the nanog open reading frame was cloned using nanog expression primers (nanog-ORF; Supplementary Table 2) and inserted into pIRES2-EGFP at the SacI/EcoRI site.
      To construct an RNA interference expression vector for human nanog (pLKO-Sh), an oligonucleotide encoding human nanog shRNA was inserted into the AgeI/EcoRI site of PLKO.1puro. A corresponding scrambled sequence was inserted into the AgeI/EcoRI site of PLKO.1puro to construct the control plasmid pLKO-Sc.
      • Zaehres H.
      • Lensch M.W.
      • Daheron L.
      • et al.
      High-efficiency RNA interference in human embryonic stem cells.
      To construct a vector specifically expressing ICAM-1 shRNA in mouse HCC cells, the AFP promoter was cloned and inserted into pEGFP-C1 at the AseI/BglII site. An shRNA oligonucleotide targeting ICAM-1 was then inserted into the recombinant vector at the EcoRI/BamHI site downstream of the AFP promoter to yield the plasmid Sh. The corresponding scrambled sequence was used to construct the control plasmid Sc.
      • Yu F.
      • Yao H.
      • Zhu P.
      • et al.
      let-7 regulates self renewal and tumorigenicity of breast cancer cells.
      • Hirano Y.
      • Sakurai E.
      • Matsubara A.
      • et al.
      Suppression of ICAM-1 in retinal and choroidal endothelial cells by plasmid small-interfering RNAs in vivo.

       Treatment, Necropsy, Histopathology, and Immunohistochemistry of M-TgHBV Transgenic Mice

      Male M-TgHBV transgenic mice (12 months old) were randomly divided into 2 groups (n = 20). The mice were injected with Sh or Sc via the tail vein once per week (100 μg each mouse). After 6 months of injections, the mice were killed. Their livers and lungs were analyzed through histopathology and immunohistochemistry.
      Supplementary Table 1Primers and Short hairpin RNA Sequences
      NameSequence
      sox2
       Sense5′CGAGATAAACATGGCAATCAAAAT 3′
       Antisense5′AATTCAGCAAGAAGCCTCTCCTT 3′
      oct4
       Sense5′CAAAGCAGAAACCCTCGTG 3′
       Antisense5′TTACAGAACCACACTCGGACC3′
      Nanog
       Sense5′AATACCTCAGCCTCCAGCAGATG3′
       Antisense5′TGCGTCACACCATTGCTATTCTTC3′
      β-catenin
       Sense5′ATGGCTTGGAATGAGACYGC3′
       Antisense5′ATGCTCCATCATAGGGTCCA3′
      UGT2B7
       Sense5′ AGAATTTCATCATGCAACAG3′
       Antisense5′GTTATGTCACCAAATATTG3′
      CYP3A4
       Sense5′ATTCAGCAAGAAGAACAAGGACA3′
       Antisense5′TGGTGTTCTCAGGCACAGAT3′
      hTERT
       Sense5′GGAGCAAGTTGCAAAGCATTG3′
       Antisense5′TCCCACGACGTAGTCCATGTT3′
      MYC
       Sense5′TCAAGAGGTGCCACGTCTCC3′
       Antisense5′TCTTGGCAGCAGGATAGTCCTT3′
      TACSTD1
       Sense5′ CGCAGCTCAGGAAGAATGTG 3′
       Antisense5′ TGAAGTACACTGGCATTGACG 3′
      Human ICAM-1
       Sense5′GGAGCTTCGTGTCCTGTATGGC3′
       Antisense5′CAGTGATGATGACAATCTCATACCG3′
      Mouse ICAM-1
       Sense5′CCAGATCCTGGAGACGCAGAG3′
       Antisense5′CCAGCCGAGGACCATACAGC3′
      β-actin
       Sense5′GCACCACACCTTCTACAATGAG3′
       Antisense5′ACAGCCTGGATGGCTACGT3′
      Nanog open reading frame
       Sense5′ATGAGTGTGGATCCAGCTTGTC 3′
       Antisense5′TCACACGTCTTCAGGTTGCATG 3′
      Mouse AFP promoter
       Sense5′ATGGTGAAGAACATTTGCAGC3′
       Antisense5′TTCGAGCAGTCCTGCTGAAGTCCTT3′
      Nanog ChIP primer
       Site 1
        Sense5′ CCTGGCGACAGAGCTAGACTG 3′
        Antisense5′ GGCAAGAGGAAGAGTGATGAGG 3′
       Site 2
        Sense5′ CTGCGTCATACCTGTCAGGAAAC 3′
        Antisense5′ TGCCTGTTTTTATCCTCAATCTGG 3′
       Site 3
        Sense5′ ACCGAGCCTGGCCCCAA 3′
        Antisense5′ CTGACCTCGTGTTCCACCTGC 3′
       Site 4
        Sense5′ AGAGCGAGAATCCGTCTCG 3′
        Antisense5′ TGACCAACATGCAGTTCTCTAA 3′
      Sox2/Oct4 ChIP primer
       Site 1
        Sense5′CCAGCCTTGTTGGGGTTGAA3′
        Antisense5′CAGTCAGGCAGGCCAAAGAAT3′
       Site 2
        Sense5′TGTGCACCACCACACCCG3′
        Antisense5′CTGTAATCCCAGCATTTTGGG3′
       N site
      Nonbinding site of Nanog, Sox2, and Oct4.
        Sense5′ CAGCGACAGGCAGGGATTT3′
        Antisense5′TCTACACCCGAGGGCACTCAC3′
      shRNA
       Nanog shRNA
        Sense
      • 5′CCGGTAAGGGTTAAGCTGTAACATACC
      • TCGAGGTATGTTACAGCTTAACCCTTTTTTTG 3′
        Antisense
      • 5′AATTCAAAAAAAGGGTTAAGCTGTAACA
      • TACCTCGAGGTATGTTACAGCTTAACCCTTA 3′
       Nanog scramble
        Sense
      • 5′CCGGTAACGTACGAATACTTCGACTCGA
      • GTCGAAGTATTCGTACGTTTTTTTG 3′
        Antisense
      • 5′AATTCAAAAAAACGTACGAATACTTCGACT
      • CGAGTCGAAGTATTCGTACGTTA 3′
       Mouse ICAM-1 shRNA
        Sense
      • 5′ AATTCATCGTCACGGCGATTTATACTC
      • GAGTATAAATCGC CGTGACGATTTTTTG 3′
        Antisense
      • 5′ GATCCAAAAAATCGTCACGGCGATTTAT
      • ACTCGAGTATAAATCGC CGTGACGATG 3′
       Mouse ICAM-1 scramble
        Sense
      • 5′AATTC AACGTACGAATACTTCGACTCGAG
      • TCGAAGTATT CGTACGTT TTTTTG 3
        Antisense
      • 5′GATCCAAAAAAACGTACGAATACTTCGAC
      • TCGAG TCGAAGTATT CGTACGTTG3′
      a Nonbinding site of Nanog, Sox2, and Oct4.
      Supplementary Table 2Antibodies Used in the Study
      AntibodyCatalog no.IsotypeManufacturer
      Anti-human ICAM-1 APC17-0549mIgG1KeBioscience
      Anti-human ICAM-1 PE12-0549mIgG1KeBioscience
      Anti-human CD45 APC17-9459mIgG1KeBioscience
      Anti-human CD24 PE12-0247mIgG1KeBioscience
      CD326 (EpCAM) FITC130-080-301mIgG1Miltenyi Biotec
      Bergisch Gladbach, Germany.
      Anti-human CD44 PE12-0441Rat IgG2bKeBioscience
      Anti-human CD90 FITC11-0909mIgG1KeBioscience
      CD133/1(AC133) PE130-080-801mIgG1Miltenyi Biotec
      AFP antibody (C19) FITCsc-8108 FITCGoat IgGSanta Cruz Biotechnology
      Anti-CD324 (E-cadherin) Alexa Fluor 48853-3249Rat IgG1eBioscience
      Anti-human CD56 (NCAM) PE-Cy725-0567mIgG1KeBioscience
      Alexa Fluor 488 mouse anti-human Nanog560791mIgG1KBD Pharmingen
      San Jose, CA.
      Anti-mouse CD45 APC17-0451Rat IgG2bKeBioscience
      Anti-mouse CD54 (ICAM-1) PE12-0542Rat IgG2aKeBioscience
      Anti-human AFP purified14-6583mIgG1eBioscience
      Anti-human cytokeratin 19 purified14-9898mIgG1KeBioscience
      Anti-vimentin purified14-9897mIgG1KeBioscience
      Anti-human/mouse β-catenin purified14-2567mIgG1KeBioscience
      Anti-mouse CD54 (ICAM-1) purified14-0542Rat IgG2aKeBioscience
      β-Actin rabbit monoclonal antibody4970Rabbit IgGCell Signaling
      Beverly, MA.
      Nanog(1E6C4) mouse monoclonal antibody4893mIgG1Cell Signaling
      ICAM-1 antibodyAP8656bRabbit IgGABGENT
      Rabbit polyclonal to Nanog-ChIP gradeAb21624Rabbit IgGAbcam
      Cambridge, England.
      Anti-Oct4 antibody - ChIP gradeab19857Rabbit IgGAbcam
      Anti-Sox2 antibody - ChIP gradeAb59776Rabbit IgGAbcam
      Anti-mouse IgG1 PE-Cy725-4015Rat IgGeBioscience
      Anti-mouse IgG1 FITC11-4015Rat IgGeBioscience
      Normal rabbit IgGsc2027Santa Cruz Biotechnology
      Ig, immunoglobulin.
      a Bergisch Gladbach, Germany.
      b San Jose, CA.
      c Beverly, MA.
      d Cambridge, England.
      Figure thumbnail gre1
      Supplementary Figure 1HCC CSC marker expression on ICAM-1+ cells from Huh7 and Hep3B cell lines. (A) Flow cytometry analysis of CD24, EpCAM, CD90, CD133, and CD44 expression on ICAM-1+ cells sorted from Huh7 cells. iso, isotype controls for the corresponding antibodies. The data are shown as the mean ± SD from at least 3 independent experiments. (B) Flow cytometry analysis of CD24, EpCAM, CD90, CD133, and CD44 expression on ICAM-1+ cells sorted from Hep3B cells. iso, isotype controls for the corresponding antibodies. The data are shown as the mean ± SD from at least 3 independent experiments.
      Figure thumbnail gre2
      Supplementary Figure 2Characterization of hepatic stem cell marker expression on ICAM-1+ cancer cells from tumor tissues. Representative image of AFP and K19 expression on ICAM-1+ cells from patients' tumor tissues as analyzed by flow cytometry. The data are shown as the mean ± SD from at least 3 independent experiments.
      Figure thumbnail gre3
      Supplementary Figure 3Characterization of the properties of cells in hepatospheres formed by ICAM-1+ cells from tumor tissues. (A) Representative image of the reverse-transcription analysis of hepatic stem cell marker (TACSTD1, MYC, and hTERT) and mature hepatocyte marker (UGT2B7 and CYP3A4) expression in hepatospheres. (B) Immunofluorescence staining analysis of AFP and K19 expression in hepatospheres (scale bar = 20 μm).
      Figure thumbnail gre4
      Supplementary Figure 4HCC CSC marker expression on ICAM-1+ cells from tumor tissues. Flow cytometry analysis of CD24, EpCAM, CD90, CD133, and CD44 expression on ICAM-1+ cells from patients' tumor tissues. iso, isotype controls for the corresponding antibodies. The data are shown as the mean ± SD from at least 3 independent experiments.
      Figure thumbnail gre5
      Supplementary Figure 5Characterization of hepatic stem cell marker expression on ICAM-1+ cancer cells from patients' blood. Representative image of AFP and K19 expression on ICAM-1+ cells from patients' blood as analyzed by flow cytometry. The data are shown as the mean ± SD from at least 3 independent experiments.
      Supplementary Table 3Multivariate Cox Regression Analysis of the Effect on Disease-Free Survival and Overall Survival
      VariableHazard ratio (95% confidence interval)P value
      Disease-free survival
       CD45ICAM-1+ rate (high vs low)7.15 (2.99–17.09).000
       α-fetoprotein (>20 vs ≤20 ng/mL)6.05 (1.67–21.91).006
       Tumor size (>5 vs ≤5 cm)3.55 (1.40–8.98).007
       Tumor number (multiple vs single)4.59 (1.69–12.44).003
       Portal vein tumor thrombus (yes vs no)2.40 (1.06–5.45).037
       Ascites (yes vs no)26.08 (6.99–97.40).000
      Overall survival
       CD45ICAM-1+ rate (high vs low)2.28 (0.95–7.82).062
       Portal vein tumor thrombus (yes vs no)3.90 (1.23–12.36).021
       Ascites (yes vs no)5.58 (1.79–17.34).003
       Serum prealbumin (>170 vs ≤170 mg/L)0.38 (0.15–0.10).049
      NOTE. A P value of less than .05 was considered to indicate statistical significance. P values were calculated with a Cox regression test.
      Figure thumbnail gre6
      Supplementary Figure 6Construction of a plasmid expressing ICAM-1 shRNA through the AFP promoter. (A) An shRNA oligonucleotide targeting ICAM-1 was inserted downstream of the AFP promoter. (B) shRNA controlled by the AFP promoter (Sh) specifically down-regulated ICAM-1 in Huh7 and Hepa1–6 HCC cells but not in the MCF-7 and 4T1 breast cancer cell lines at the mRNA level. **P < .01. Error bars represent the SD of data obtained from at least 3 independent experiments. (C) shRNA controlled by the AFP promoter (Sh) down-regulated ICAM-1 in Huh7 cells and Hepa1–6 cells at the protein level.
      Figure thumbnail gre7
      Supplementary Figure 7Sox2/Oct4 did not bind to the promoter of ICAM-1. (A) A bioinformatics analysis revealed 2 binding sites of Sox2/Oct4 in the 10-kilobase genomic sequence upstream (−10 kb) of ICAM-1. (B) Representative image of ChIP assays performed in spheres from tumor cell lines (Huh7 and Hep3B) showed that Sox2/Oct4 did not bind to these sites. N site, nonbinding site of Sox2/Oct4.
      Figure thumbnail gre8
      Supplementary Figure 8Detection of overexpression of Nanog. Transfection of pEGFP-nanog (Nanog) into Huh7 cells induced significant up-regulation of nanog at the mRNA and protein levels compared with cells transfected with the control vector (EGFP). **P < .01. Error bars represent the SD of data obtained from at least 3 independent experiments.
      Figure thumbnail gre9
      Supplementary Figure 9Characterization of EMT gene expression on circulating ICAM-1+ cancer cells. Representative image of E-cadherin and vimentin expression on CD45ICAM-1+ cells from patients' blood as analyzed by flow cytometry.
      Figure thumbnail gre10
      Supplementary Figure 10Immunohistochemistry and In situ hybridization images. (A and B) Images of H&E staining corresponding to E (scale bar = 50 μm). (C) Image of H&E staining corresponding to B (scale bar = 50 μm). (D) Image of H&E staining corresponding to D (scale bar = 50 μm). (E) Image of the negative controls for the immunofluorescence staining corresponding to E (scale bar = 20 μm).

      References

        • Jemal A.
        • Bray F.
        • Center M.M.
        • et al.
        Global cancer statistics.
        CA Cancer J Clin. 2011; 61: 69-90
        • van de Stolpe A.
        • van der Saag P.T.
        Intercellular adhesion molecule-1.
        J Mol Med (Berl). 1996; 74: 13-33
        • Momosaki S.
        • Yano H.
        • Ogasawara S.
        • et al.
        Expression of intercellular adhesion molecule 1 in human hepatocellular carcinoma.
        Hepatology. 1995; 22: 1708-1713
        • Lawson C.
        • Wolf S.
        ICAM-1 signaling in endothelial cells.
        Pharmacol Rep. 2009; 61: 22-32
        • Zimmerman T.
        • Blanco F.J.
        Inhibitors targeting the LFA-1/ICAM-1 cell-adhesion interaction: design and mechanism of action.
        Curr Pharm Des. 2008; 14: 2128-2139
        • Duperray A.
        • Languino L.R.
        • Plescia J.
        • et al.
        Molecular identification of a novel fibrinogen binding site on the first domain of ICAM-1 regulating leukocyte-endothelium bridging.
        J Biol Chem. 1997; 272: 435-441
        • Sun J.J.
        • Zhou X.D.
        • Liu Y.K.
        • et al.
        Invasion and metastasis of liver cancer: expression of intercellular adhesion molecule 1.
        J Cancer Res Clin Oncol. 1999; 125: 28-34
        • Sun J.J.
        • Zhou X.D.
        • Zhou G.
        • et al.
        Expression of intercellular adhesive molecule-1 in liver cancer tissues andliver cancer metastasis.
        World J Gastroenterol. 1998; 4: 202-205
        • Assis A.C.
        • Carvalho J.L.
        • Jacoby B.A.
        • et al.
        Time-dependent migration of systemically delivered bone marrow mesenchymal stem cells to the infarcted heart.
        Cell Transplant. 2010; 19: 219-230
        • De Francesco F.
        • Tirino V.
        • Desiderio V.
        • et al.
        Human CD34/CD90 ASCs are capable of growing as sphere clusters, producing high levels of VEGF and forming capillaries.
        PLoS One. 2009; 4: e6537
        • Brooke G.
        • Tong H.
        • Levesque J.P.
        • et al.
        Molecular trafficking mechanisms of multipotent mesenchymal stem cells derived from human bone marrow and placenta.
        Stem Cells Dev. 2008; 17: 929-940
        • Sununliganon L.
        • Singhatanadgit W.
        Highly osteogenic PDL stem cell clones specifically express elevated levels of ICAM1, ITGB1 and TERT.
        Cytotechnology. 2012; 64: 53-63
        • Strakova Z.
        • Livak M.
        • Krezalek M.
        • et al.
        Multipotent properties of myofibroblast cells derived from human placenta.
        Cell Tissue Res. 2008; 332: 479-488
        • Dalerba P.
        • Cho R.W.
        • Clarke M.F.
        Cancer stem cells: models and concepts.
        Annu Rev Med. 2007; 58: 267-284
        • Al-Hajj M.
        Cancer stem cells and oncology therapeutics.
        Curr Opin Oncol. 2007; 19: 61-64
        • Wicha M.S.
        Cancer stem cells and metastasis: lethal seeds.
        Clin Cancer Res. 2006; 12: 5606-5607
        • Haraguchi N.
        • Utsunomiya T.
        • Inoue H.
        • et al.
        Characterization of a side population of cancer cells from human gastrointestinal system.
        Stem Cells. 2006; 24: 506-513
        • Ma S.
        • Chan K.W.
        • Hu L.
        • et al.
        Identification and characterization of tumorigenic liver cancer stem/progenitor cells.
        Gastroenterology. 2007; 132: 2542-2556
        • Yang Z.F.
        • Ho D.W.
        • Ng M.N.
        • Lau C.K.
        • et al.
        Significance of CD90+ cancer stem cells in human liver cancer.
        Cancer Cell. 2008; 13: 153-166
        • Yamashita T.
        • Ji J.
        • Budhu A.
        • et al.
        EpCAM-positive hepatocellular carcinoma cells are tumor-initiating cells with stem/progenitor cell features.
        Gastroenterology. 2009; 136: 1012-1024
        • Zhu Z.
        • Hao X.
        • Yan M.
        • et al.
        Cancer stem/progenitor cells are highly enriched in CD133+CD44+ population in hepatocellular carcinoma.
        Int J Cancer. 2010; 126: 2067-2078
        • Shuqun C.
        • Mengchao W.
        • Han C.
        • et al.
        Tumor thrombus types influence the prognosis of hepatocellular carcinoma with the tumor thrombi in the portal vein.
        Hepatogastroenterology. 2007; 54: 499-502
        • Zhu X.D.
        • Zhang J.B.
        • Zhuang P.Y.
        • et al.
        High expression of macrophage colony-stimulating factor in peritumoral liver tissue is associated with poor survival after curative resection of hepatocellular carcinoma.
        J Clin Oncol. 2008; 26: 2707-2716
        • Zhang X.
        • Liu S.
        • Hu T.
        • et al.
        Up-regulated microRNA-143 transcribed by nuclear factor kappa B enhances hepatocarcinoma metastasis by repressing fibronectin expression.
        Hepatology. 2009; 50: 490-499
        • Lee T.K.
        • Castilho A.
        • Cheung V.C.
        • et al.
        CD24(+) liver tumor-initiating cells drive self-renewal and tumor initiation through STAT3-mediated NANOG regulation.
        Cell Stem Cell. 2011; 9: 50-63
        • Yang Y.M.
        • Chang J.W.
        Current status and issues in cancer stem cell study.
        Cancer Invest. 2008; 26: 741-755
        • Tsuchiya A.
        • Kamimura H.
        • Tamura Y.
        • et al.
        Hepatocellular carcinoma with progenitor cell features distinguishable by the hepatic stem/progenitor cell marker NCAM.
        Cancer Lett. 2011; 309: 95-103
        • Mavroudis D.
        Circulating cancer cells.
        Ann Oncol. 2010; 21: vii95-vii100
        • Meng S.
        • Tripathy D.
        • Shete S.
        • et al.
        HER-2 gene amplification can be acquired as breast cancer progresses.
        Proc Natl Acad Sci U S A. 2004; 101: 9393-9398
        • Smirnov D.A.
        • Zweitzig D.R.
        • Foulk B.W.
        • et al.
        Global gene expression profiling of circulating tumor cells.
        Cancer Res. 2005; 65: 4993-4997
        • Hellerbrand
        • Wang S.C.
        • Tsukamoto H.
        • et al.
        Expression of intracellular adhesion molecule 1 by activated hepatic stellate cells.
        Hepatology. 1996; 24: 670-676
        • Hu T.
        • Liu S.
        • Breiter D.R.
        • et al.
        Octamer 4 small interfering RNA results in cancer stem cell-like cell apoptosis.
        Cancer Res. 2008; 68: 6533-6540
        • Shan J.
        • Shen J.
        • Liu L.
        • et al.
        Nanog regulates self-renewal of cancer stem cells through the insulin-like growth factor pathway in human hepatocellular carcinoma.
        Hepatology. 2012; 56: 1004-1014
        • Jeter C.R.
        • Badeaux M.
        • Choy G.
        • et al.
        Functional evidence that the self-renewal gene NANOG regulates human tumor development.
        Stem Cells. 2009; 27: 993-1005