Hepatic cancer stem cell marker granulin-epithelin precursor and β-catenin expression associate with recurrence in hepatocellular carcinoma

Abstract

Granulin-epithelin precursor (GEP) has been demonstrated to confer enhanced cancer stem-like cell properties in hepatocellular carcinoma (HCC) cell line models in our previous studies. Here, we aimed to examine the GEP-expressing cells in relation to the stem cell related molecules and stem-like cell properties in the prospective HCC clinical cohort. GEP protein levels were significantly higher in HCCs than the paralleled non-tumor liver tissues, and associated with venous infiltration. GEPhigh cells isolated from clinical HCC samples exhibited higher levels of stem cell marker CD133, pluripotency-associated signaling molecules β-catenin, Oct4, SOX2, Nanog, and chemodrug transporter ABCB5. In addition, GEPhigh cells possessed preferential ability to form colonies and spheroids, and enhanced in vivo tumor-initiating ability while their xenografts were able to be serially subpassaged into secondary mouse recipients. Expression levels of GEP and pluripotency-associated genes were further examined in the retrospective HCC cohort and demonstrated significant correlation of GEP with β-catenin. Notably, HCC patients with high GEP and β-catenin levels demonstrated poor recurrence-free survival. In summary, GEP-positive HCC cells directly isolated from clinical specimens showed β-catenin elevation and cancer stem-like cell properties.

Keywords: cancer stem cells; granulin-epithelin precursor; hepatocellular carcinoma; β-catenin.

Figure 1. GEP positive HCC cells express stem cell related molecules

Fresh HCC and paralleled non-tumor liver tissues were collected. After enzymatic digestion, cell viabilities were assessed by trypan blue staining, and only cases with high cell viability (viability ≥ 70%) (n = 42/90, 47%) were subject to subsequent characterization. A. Cells isolated from fresh HCC and non-tumor tissues were stained for total cellular GEP and analyzed by flow cytometry (n = 42). B. Cells were sorted according to cell surface GEP. Sorted GEP^high and GEP_low HCC cells were then permeabilized and stained for cellular GEP using anti-GEP antibody recognizing different epitope from that of the sorting antibody, and analyzed by flow cytometry. Percentage of GEP+ cells and mean fluorescence intensity (MFI) of the unsorted and sorted populations were shown. C. Sorted GEP^high and GEP_low cells and unsorted control were stained for hepatic surface CSC markers CD133 and EpCAM, pluripotency-associated signaling molecules β-catenin, Oct4, Nanog and SOX2, and drug transporter ABCB5, and analyzed by flow cytometry (n > 7 for each marker). The lines in scatter plots indicated the median values.

Figure 2. GEP^high cells possess CSC properties in vitro

A. Sorted GEP^high cells isolated from freshly resected HCC were able to form more colonies than GEP_low cells (n = 6). In brief, 1000 cells of each freshly sorted subpopulation were seeded onto 6-well plate and allowed to grow for a month. B. Upper panel: GEP^high cells, but not GEP_low cells, isolated from freshly resected tumors were able to generate primary (1°) spheroids (n = 3). 1° spheroids were collected and dissociated, and 1000 disaggregated cells were allowed to grow for 1 month for generating secondary (2°) spheroids (n = 3). GEP^high cell-derived, but not GEP_low cells-derived 1° spheroids, were able to generate 2° spheroids. Following induced differentiation, disaggregated cells generated from GEP^high cell-derived 2° spheroids differentiated and grew as adherent cells. Lower panel: Flow cytometric analysis showed the enrichment of GEP expression in mechanically dissociated GEP^high cells-derived spheroids as compared to the original resected tumors (clinical specimens) and the differentiated adherent counterpart. Briefly, 1000 cells of each freshly sorted subpopulation were seeded into ultra-low attachment 24-well plate, and allowed to grow for 1 month to generate spheroids. C. After exposure to doxorubicin (0.5μg/ml) for 24h, GEP^high cells retained significantly less doxorubicin than GEP_low cells and unsorted control (n = 3). Data are expressed as mean percentage + SD.

Figure 3. GEP^high cells possess CSC properties in vivo.

GEP^high and GEP_low cells were sorted from freshly resected tumors of HCC patients and injected into NOD/SCID mice. A. The upper panel shows the NOD/SCID mice injected subcutaneously with 1× 10⁴ GEP^high (right flank) and GEP_low (left flank) cells at 8-14 weeks after inoculation. Arrow indicated the site of tumor formation at the right flank of mice. The lower panel shows the corresponding subcutaneous tumors derived from GEP^high cells. B. Flow cytometric analysis showed the total cellular GEP expression in mechanically dissociated GEP^high cells derived-xenograft tumors (1° and 2°) as compared to the original clinical specimens. In brief, to demonstrate the in vivo self-renewal ability of the cells, GEP^high and GEP_low subpopulations were sorted from xenografts growing from initial inoculation (1° xenografts) and then transplanted into secondary mouse recipients. GEP^high, but not GEP_low subpopulations, were able to generate secondary (2°) xenograft tumors. Data are expressed as mean percentage + SD.

Figure 4. Clinical significance of GEP and β-catenin in HCC clinical specimens

A. β-catenin transcript level was significantly up-regulated in HCC tumor (HCC, n = 77) compared with the paralleled tumor-adjacent non-tumor liver tissues (non-tumor, n = 77) and normal livers from healthy individuals (normal, n = 10). The lines indicate the median values. B. Expression levels of GEP significantly correlated with that of β-catenin in HCC tumor tissues (HCC), and in the paralleled non-tumor tissues. The tumor / non-tumor (T/NT) ratio showed the same trend. C. Kaplan–Meier recurrence-free survival plot according to β-catenin levels (log-rank test, p = 0.053). There were 56 patients with low β-catenin expression and 21 patients with high β-catenin expression (median recurrence-free survival of 24.5 months and 12.8 months, respectively). D. Patients (n = 77) were segregated into the low expression group (either one or both low in GEP and β-catenin) and the high expression group (both high in GEP and β-catenin). There were 61 patients in the low expression group (median recurrence-free survival, 24.5 months) and 16 patients in the high expression group (median recurrence-free survival, 12.6 months). Patients with high GEP and β-catenin levels were found to have poor recurrence-free survival (log-rank test, p = 0.038). When the patients were segregated into early and late tumor stages, patients with high GEP and β-catenin levels also demonstrated poor recurrence-free survival (log-rank test, p = 0.022).