Specimens from 54 patients with glioma who underwent surgical resection at the Department of Neurosurgery, Peking University People's Hospital between November of 2009 and May of 2012 were collected. Of the patients, 29 were male and 25 were female. Ages of the patients at the time of surgery ranged from 21 to 75 years [mean age ± standard deviation (SD), 45.8±14.5 years]. According to the revised WHO criteria for the central nervous system (18), tumors were categorized into grade I (n=5), grade II (n=13), grade III (n=13) and grade IV (n=23). All tumor tissues were obtained from the initial surgery, and none of the patients had been subjected to chemotherapy or radiation therapy prior to tumor excision. The histologic subtypes and pathologic grades of all glioma samples were confirmed by two pathologists independently. The present study was approved by the Institutional Review Board, and all participants provided written informed consent.

All paraffin-embedded sections were deparaffinized followed by washing in xylenes and serial dilutions of ethanol. Endogenous peroxidase was blocked by 3% H₂O₂ for 12 min. After antigen retrieval, blocks for avidin and biotin and the Fc receptor were applied. The rabbit anti-LGR5 polyclonal antibody was used at a 1:150 dilution (Abcam, Cambridge, UK) or the rabbit anti-human Ki-67 polyclonal antibody was used at a 1:200 dilution (Santa Cruz Biotechnology, Santa Cruz, CA, USA) overnight at 4°C in a humidified chamber. The primary antibodies were then detected using the appropriate labeled streptavidin-biotin (LSAB) kit (Fuzhou Maixin Biotechnology, Fuzhou, China) according to the manufacturer's instructions. Immunolabeled sections were visualized with 3′,3′-diaminobenzidine tetrahydrochloride (DAB; Sigma, St. Louis, MO, USA) and counterstained with hematoxylin. As control, phosphate-buffered saline (PBS) was used instead of the primary antibody.

The staining results for immunohistochemistry were evaluated by two independent neuropathologists who were blinded to clinical information. Brown-yellow staining in the cytoplasm and/or membrane was considered positive for LGR5. Brown-yellow staining in the nucleus was positive for Ki-67. To measure the LGR5 immunoreactivity score (IRS) and proliferative index (PI), 10 high-power (x400) fields (~1,000 cells) were randomly chosen for quantification in the most strongly stained tumor area of each section. The LGR5 staining intensity (LGR5-SI), the percentage of LGR5-positive tumor cells (LGR5-PP), and the resulting LGR5 immunoreactivity score (LGR5-IRS) were evaluated by a modified method as previously described (19,20). Briefly, the immunoreactivity score (LGR5-IRS: negative, 0; weak, 1–3; moderate, 4–6; strong, 8–12) was determined by multiplication of the value for LGR5 staining intensity (LGR5-SI: 0, no staining; 1, weak staining; 2, moderate staining; 3, strong staining) and the value for the percentage of LGR5-positive tumor cells (LGR5-PP: 0, <1%; 1, 1–25%; 2, 26–50%; 3, 51–75%; 4, >75%). Due to the heterogeneous staining intensity of the tumor cells, the SI was determined according to the staining intensity noted in the majority of the cells. The percentage of Ki-67-positive cells was regarded as the PI of each tumor tissue sample, respectively.

Human malignant glioma cell lines (U118, U87 and U251) and normal human astrocytes (1800) were obtained from the Cell Library of the Chinese Academy of Sciences (Shanghai, China). U118, U87 and U251 cells were cultured at 37°C in 5% CO₂ in Dulbecco's modified Eagle's essential medium (DMEM) (Gibco, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS; HyClone, Logan, UT, USA), 2 mM L-glutamine and 100 U/ml penicillin-streptomycin (Gibco). The normal astroctyes (1800) were cultured at 37°C in 5% CO₂ in modified RPMI-1640 (HyClone) supplemented with 10% FBS, 2 mM L-glutamine and 100 U/ml penicillin-streptomycin (Gibco). The medium was changed every 3–4 days, and cultures were split using 0.25% trypsin. U87, U87-NC and U87-KD fluorescent (EGFP)-labeled cells were developed (Shanghai GeneChem Co, Ltd., Shanghai, China) and an in vivo optical imaging technique was used.

Total RNA was isolated from cells using the RNeasy Mini kit, including DNase treatment (Qiagen K.K., Tokyo, Japan). cDNA was synthesized using the PrimeScripts RT reagent kit (Perfect Real-Time; Takara Bio, Shiga, Japan) and qPCR was performed on a Thermal Cycler Dice Real-Time System using SYBR Premix Ex Taq™ (Perfect Real-Time). The primer sequences for qPCR were as follows: GAPDH forward, 5′-ATCATCCCTGCCTCTACTGG-3′ and GAPDH reverse, 5′-TTTCTAGACGGCAGGTCAGGT-3′; LGR5 forward, 5′-GAGGATCTGGTGAGCCTGAGAA-3′ and LGR5 reverse, 5′-CATAAGTGATGCTGGAGCTGGTAA-3′. GAPDH was used as a reference. Fold induction values were calculated using the 2^−ΔΔCt method. All experiments were performed in triplicate and repeated at least three times in separate experiments; representative data are shown.

Cell lysates or the tissues dissolved in SDS sample buffer were separated by SDS-PAGE and transferred to nitrocellulose membranes. β-actin was used as a control. Membranes were probed for the LGR5 antibody (1:1,500; Abcam) or β-actin (1:2,000; Abcam) in Tris-buffered saline (TBS) containing 1% milk and 0.05% Tween-20 overnight at 4°C. The secondary antibody was horseradish peroxidase (HRP)-goat anti-rabbit or anti-mouse IgG (1:2,500; Cell Signaling) and incubation ws carried out for 1 h at room temperature. Blots were developed with Amersham ECL Western Blotting Detection reagent (GE Healthcare, Chalfont St. Giles, UK).

For knockdown of human LGR5, small hairpin RNA (shRNA) of the human LGR5 lentivirus gene transfer vector encoding green fluorescent protein (GFP) sequence was constructed by Shanghai GeneChem Co., Ltd. (Shanghai, China). The targeting sequence of the shRNA was 5′-GTCTGCAATCAGTTACCTA-3′. The recombinant lentivirus of small interference RNA targeting LGR5 (LGR5-RNAi-Lentivirus) was prepared and titered to 10⁹ TU/ml (transfection units/ml). A scrambled short-hairpin RNA (shRNA) was used as a negative control.

An MTT assay was performed to detect the anti-proliferative effect of LGR5 RNA interference on U87 cells. U87 cells were seeded in 96-well plates at a density of 2×10³/well. After 24 h of incubation, cells were serum starved overnight. In another experiment, U87-NC-shRNA and U87-LGR5-shRNA cells were seeded in 96-well plates at a density of 2×10³/well and incubated for 24, 48, 72, 96 or 120 h. At each time-point, 20 ml of 5 mg/ml MTT (Sigma) solution was added to each well. After 4 h of incubation, the medium was removed from the wells by aspiration, and the formazan crystals were dissolved in 150 ml of dimethyl sulfoxide (DMSO; Sigma). Color intensity was measured at 490 nm with an enzyme linked immunosorbent assay plate reader (Bio-Rad Laboratories, Hercules, CA, USA). Cell growth curves were determined using the average absorbance at 490 nm from triplicate samples of three independent experiments.

Cells were cultured in 25-ml flasks and incubated until they reached 60–70% confluence in DMEM containing 10% FBS. The cells were collected and washed twice with ice-cold PBS, and then fixed overnight with 70% ethanol at 4°C. Following incubation with 50 mg/ml RNase A at room temperature for 30 min, the cells were stained with 20 mg/ml propidium iodide (PI; Sigma) for an additional 30 min. DNA content and cell cycle distribution were analyzed by flow cytometry (FACSCalibur; Becton-Dickinson, San Jose, CA, USA) and the results were interpreted using ModFit and CellQuest software. All of the samples were assayed in triplicate.

Cells (800 cells/plate) were cultured in 3 ml of DMEM supplemented with 10% FBS and 800 mg/ml G418 in a 6-well plate. After 2 weeks, colonies were rinsed with PBS, fixed with methanol for 5 min, and stained with Giemsa (Sigma) for 20 min. Clearly visible colonies (>50 mm in diameter) were counted as positive for growth.

For tumorsphere formation, single-cell suspensions were suspended in Dulbecco's modified Eagle's medium/F12 (DMEM/F12; HyClone) supplemented with B-27 (1X, Gibco), 20 ng/ml epidermal growth factor (EGF; PeproTech Inc., Rocky Hill, NJ, USA) and 20 ng/ml basic fibroblast growth factor (bFGF; Peprotech) and then plated in 24-well ultra-low attachment plates (Corning Incorporated, Corning, NY, USA) at a concentration of 1,000 cells/well. Plates were analyzed 7–10 days later for tumorsphere formation, which was quantified using an inverted microscope (Olympus) at ×100 magnification.

In an orthotopic model, 6-week-old BALB/C nude mice (Cancer Institute of the Chinese Academy of Medical Science) were randomly divided into three groups (three mice per group). All experiments were performed following approval of the Animal Studies Ethics Committee of the Peking University People's Hospital. A small burr hole, 2 mm in diameter was made (1 mm to the midline and 0.5 mm anterior to the bregma) using a microskull drill. A trochar packed with donor tissue was navigated to a depth of 2.5 mm via the skull hole. Cells (5×10⁵) (U87, Si-NC and Si-LGR5 cells) suspended in 5 μl PBS were slowly and smoothly injected into the subcortex of the mouse brain. The skull hole was sealed with bone wax and the scalp was sutured. Tumor growth in mice was detected by the Kodak In-Vivo FX Pro system (Kodak, Rochester, NY, USA). The tumor volume of the xenografts was detected every 5 days over the course of the study using fluorescence signaling.

Prior to the in vivo imaging, the mice were anesthetized with phenobarbital sodium. Fluorescence imaging was carried out with an excitation wavelength of 490 nm and emission wavelength of 535 nm. Exposure times ranged from 1 to 2 min. Mice were sacrificed and examined 5 weeks after the implantation. LGR5 was detected by immunohistochemistry.

The experiments were performed in triplicate and repeated three times independently. SPSS 16.0 was used for all statistical analysis. Comparisons among all groups were performed using one-way analysis of variance (ANOVA). The t-test was used for comparison of differences between two groups. Correlation coefficients of LGR5 IRS with the PI were evaluated using Pearson correlation analysis. Values of P<0.05 were considered to indicate statistically significant results in all cases.

In the present study, LGR5 protein was overexpressed in human glioma specimens. Positive tumor cells showed primarily cytoplasmic and/or membranous labeling under a light microscope. However, normal brain tissues had exceedingly weak immunoreactivity for this protein (data not shown). The LGR5 IRS was 4.60±2.57 for 54 cases of tumor specimens. The LGR5 IRS was positively and markedly correlated with increasing WHO grade (Fig. 1I; Table I). There were significant differences in LGR5 IRS between grade I and grade III (P<0.05), grade I and grade IV (P<0.005), and grade II and grade IV (P<0.05) gliomas, respectively. Representative images of LGR5 immunostaining are shown in Fig. 1A–D, and the related results are provided in Table I. The results of the western blot analysis and RT-PCR were coincident with that of immunohistochemistry. The results showed that mRNA and protein expression levels of LGR5 markedly increased with an increase in pathologic grade of the brain gliomas (P<0.05, Fig. 1K and L).

Pathologic grade has been linked to cell proliferation. Therefore, we sought to determine whether high expression of LGR5 might also be associated with proliferation. To address this we first compared LGR5 expression with proliferation. The proliferative index (PI) of tumor cells was evaluated by Ki-67 staining. The cell proliferation marker Ki-67 was expressed in all tumor specimens (Fig. 1E–H). With increasing pathologic grade of glioma, PI increased markedly (Fig. 1J; Table I). Moreover, the PI was positively correlated with LGR5 IRS (r=0.886, P<0.0001) (Fig. 1M).

The expression levels of LGR5 mRNA and protein in several high grade glioma-derived cell lines (U118, U87 and U251) cultured in vitro were compared to the expression levels in normal human cultured primary astrocytes by western blot analysis and qRT-PCR (Fig. 2A and C). The results showed that LGR5 was highly expressed in the three glioma cell lines when compared with that in the normal astrocytes; U87 cells were randomly used as an appropriate in vitro model for assessing LGR5 function in the subsequent experiments. To study the role of LGR5 in the malignant progression of glioma, we established a stably transfected U87 cell line expressing shRNA against LGR5. The protein level of LGR5 was confirmed to be dramatically downregulated in the U87-KD cells when compared to the parental U87 and U87-NC cells. In addition, there was no obvious difference in LGR5 expression between the control cell lines (Fig. 2B). These data indicate that transfection of LGR5 shRNA significantly and specifically inhibited the endogenous LGR5 expression in U87-KD human glioma cells.

To evaluate the effect of LGR5 on the growth of U87 cells, viability curves for U87, U87-NC and U87-KD cells were determined by MTT assay. As shown in Fig. 3, the growth of U87-KD cells was obviously inhibited when compared with that in the other groups, with this effect being most obvious from day 3 to day 7 (P<0.01). However, there were no significant differences in cell growth between the U87 and U87-NC cells (P>0.05). These results indicate that downregulation of LGR5 expression by RNAi markedly inhibited the growth of U87 cells.

To investigate the effect of LGR5 knockdown on cell cycle progression, flow cytometry was performed to determine the cell cycle distribution (Fig. 4A). Compared with parental U87 and U87-NC cells, U87-KD cells accumulated in the G0/G1 phase (P<0.05), whereas the percentage of cells distributed in the S phase was decreased sharply (P<0.05). There was no obvious difference in cell cycle distribution between the parental U87 and U87-NC cells (P>0.05; Fig. 4B). These results suggest that reduction in LGR5 expression in U87 cells by RNAi delays cell cycle progression and decreases cell proliferation.

To analyze the effect of LGR5 downregulation on the anchorage-dependent growth potential of U87 cells, plate colony formation assays were performed for parental U87, U87-NC and U87-KD cells (Fig. 5A). Compared to the control cells, the number and size of colonies for U87-KD cells were significantly decreased (P<0.05; Fig. 5B). In contrast, there were no obvious differences in colony-forming ability between the control cell lines (P>0.05). These data indicate that reduction in LGR5 expression decreased the colony-formation ability of U87 cells in vitro.

To examine the self-renewal potential of U87 cells with or without LGR5 knockdown, we undertook tumorsphere formation culture of shRNA-LGR5/U87, shRNA-Ctr/U87 and untransfected U87 cells in a special ultra-low attachment culture plate with conditional medium for tumorsphere formation. After 7–10 days, plates were analyzed for tumorsphere formation which was quantified using an inverted microscope. As shown in Fig. 6, significantly fewer and smaller tumorspheres were observed in the shRNA-LGR5/U87 cells than those in the U87 and shRNA-Ctr/U87 cells (P<0.05; Fig. 6).

To evaluate the effect of LGR5 on glioma in vivo, we injected shRNA-LGR5/U87, shRNA-Ctr/U87 and untransfected U87 cells into nude mice to develop orthotopic tumors. Tumor growth in the mice was monitored by a live imaging system detecting the fluorescent signals. During the course of the study, shRNA-LGR5/U87 cells demonstrated significantly inhibited tumor growth (Fig. 7A). All data were presented as the mean ± SD and as an average of three measurements from a representative experiment. The luciferase signal from shRNA-LGR5/U87 cells was significantly less than that from shRNA-Ctr/U87 and non-transfected U87 cells (P<0.05; Fig. 7B). Immunohistochemical analysis is shown in Fig. 7C)