Side population purified from hepatocellular carcinoma cells harbors cancer stem cell–like properties


  • Tetsuhiro Chiba,

    1. Department of Regenerative Medicine, Graduate School of Medical Science, Yokohama City University, Yokohama, Japan
    2. Department of Cellular and Molecular Medicine, Graduate School of Medicine, Chiba University, Chiba, Japan
    Search for more papers by this author
  • Kaoru Kita,

    1. Department of Regenerative Medicine, Graduate School of Medical Science, Yokohama City University, Yokohama, Japan
    Search for more papers by this author
  • Yun-Wen Zheng,

    1. Department of Regenerative Medicine, Graduate School of Medical Science, Yokohama City University, Yokohama, Japan
    Search for more papers by this author
  • Osamu Yokosuka,

    1. Department of Medicine and Clinical Oncology, Graduate School of Medicine, Chiba University, Chiba, Japan
    Search for more papers by this author
  • Hiromitsu Saisho,

    1. Department of Medicine and Clinical Oncology, Graduate School of Medicine, Chiba University, Chiba, Japan
    Search for more papers by this author
  • Atsushi Iwama,

    1. Department of Cellular and Molecular Medicine, Graduate School of Medicine, Chiba University, Chiba, Japan
    Search for more papers by this author
  • Hiromitsu Nakauchi,

    1. Laboratory of Stem Cell Therapy, Center for Experimental Medicine, Institute of Medical Science, University of Tokyo, Tokyo, Japan
    Search for more papers by this author
  • Hideki Taniguchi

    Corresponding author
    1. Department of Regenerative Medicine, Graduate School of Medical Science, Yokohama City University, Yokohama, Japan
    2. Research Unit for Organ Regeneration, Center for Developmental Biology, RIKEN, Kobe, Japan
    3. Biomaterials Center, National Institute for Materials Science, Tsukuba, Japan
    • Department of Regenerative Medicine, Graduate School of Medical Science, Yokohama City University, 3-9 Fukuura, Kanazawa-ku, Yokohama 236-0004, Japan
    Search for more papers by this author
    • fax: (81) 45-7878963

  • See Editorial on Page 23

  • Potential conflict of interest: Nothing to report.


Recent advances in stem cell biology enable us to identify cancer stem cells in solid tumors as well as putative stem cells in normal solid organs. In this study, we applied side population (SP) cell analysis and sorting to established hepatocellular carcinoma (HCC) cell lines to detect subpopulations that function as cancer stem cells and to elucidate their roles in tumorigenesis. Among four cell lines analyzed, SP cells were detected in Huh7 (0.25%) and PLC/PRF/5 cells (0.80%), but not in HepG2 and Huh6 cells. SP cells demonstrated high proliferative potential and anti-apoptotic properties compared with those of non-SP cells. Immunocytochemistry examination showed that SP fractions contain a large number of cells presenting characteristics of both hepatocyte and cholangiocyte lineages. Non-obese diabetic/severe combined immunodeficiency (NOD/SCID) xenograft transplant experiments showed that only 1 × 103 SP cells were sufficient for tumor formation, whereas an injection of 1 × 106 non-SP cells did not initiate tumors. Re-analysis of SP cell–derived tumors showed that SP cells generated both SP and non-SP cells and tumor-initiating potential was maintained only in SP cells in serial transplantation. Microarray analysis discriminated a differential gene expression profile between SP and non-SP cells, and several so-called “stemness genes” were upregulated in SP cells in HCC cells. In conclusion, we propose that a minority population, detected as SP cells in HCC cells, possess extreme tumorigenic potential and provide heterogeneity to the cancer stem cell system characterized by distinct hierarchy. (HEPATOLOGY 2006;44:240–251.)

Cancer is believed to be unicellular in origin,1, 2 although cancer cells generally exhibit functional heterogeneity in a variety of cancers.3 Colony formation assay in soft agar and spleen colony assay demonstrated that the ability to form colonies is restricted in a small population, but not in bulk cancer cells.4, 5In vivo transplant assay, the injection of a small subset of cancer cells, also resulted in tumor development.6, 7 These results indicated the possibility that only a minor population possesses the ability to imitate tumors, whereas most lack tumor-initiating activities. These experimental observations have led to two general theories8, 9: The stochastic model indicates that cancer is composed of a comparatively homogenous population and a few cells that undergo stochastic events and gain the potential to proliferate extensively and form new tumors. The alternative hypothesis, namely, the hierarchy model, postulates that a subpopulation of cancer stem cells generates a hierarchy organization containing varied downstream descendants, proliferates extensively, and initiates new tumors at high frequency.

Technical advancements in stem cell biology such as cell isolation using flow cytometry, cell culture, and transplantation into immunodeficient mice facilitate the identification of stem cells in tumors. A small subset of cells with the ability to initiate human acute myelogenous leukemia in non-obese diabetic/severe combined immunodeficiency (NOD/SCID) mice was identified from leukemic patients.10 They were isolated as CD34+CD38− cells and could initiate leukemia in serial transplantation. Moreover, it has been reported that common mechanisms exist that regulate the self-renewal of both hematopoietic stem cells and leukemic stem cells.11 Attempts are also being made to detect tumorigenic subpopulations in solid tumors using flow cytometry beyond the field of leukemic stem cells. Recently, CD44+CD24−/low epithelial-specific antigen–positive cells in breast cancer and CD133+ cells in medulloblastoma or glioblastoma were prospectively identified as “cancer stem cells.”12, 13 In these studies, a small subset of cells purified as cancer stem cells exhibited tumorigenetic potential, whereas most tumor cells did not in in vivo transplant assays. These observations obtained from in culture and in vivo transplant assays might advocate a hierarchy model.14

Hepatocellular carcinoma (HCC) is a common malignancy worldwide and still carries a high mortality rate.15 It is presumed that accumulation of genetic and epigenetic alterations contributes to multistep hepatocarcinogenesis.16–18 However, the whole mechanism underlying hepatocarcinogenesis has not been clearly documented; moreover, the cellular origin of HCC remains to be elucidated. The prospective identification of cancer stem cells and elucidation of the hierarchy in HCC cells might serve in the understanding of hepatocarcinogenesis and exploration of novel therapeutic approaches.

Side population (SP) cell sorting was initially applied for the identification of hematopoietic stem cells and has been used to enrich stem cell compartments in diverse tissues and organs.19–21 SP cells are detected by their own ability to efflux Hoechst 33342 dye through an adenosine triphosphate (ATP)-binding cassette (ABC) membrane transporter. Recently, SP cells were also used in an attempt to isolate a stem cell–like fraction in cancer cells.22 The approach seems reasonable and valuable, because it is documented that a variety of cancers, including HCC, highly express ABC transporters and closely contribute to multi-drug resistance.23

In this study, we extended SP cell analysis and sorting to established HCC cell lines. We further estimated whether the SP fraction possessed cancer stem cell–like properties in culture and in an NOD/SCID xenotransplant model.


NOD/SCID, non-obese diabetic/severe combined immunodeficiency; HCC, hepatocellular carcinoma; SP, side population; ATP, adenosine triphosphate; ABC, ATP-binding cassette; DMEM, Dulbecco's minimum essential medium; MTS, 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt; TUNEL, terminal deoxynucleotidyl transferase-mediated nick-end labeling; PBS, phosphate-buffered saline; AFP, alpha-fetoprotein; CK19, cytokeratin 19; RT, reverse transcription; PCR, polymerase chain reaction.

Materials and Methods

Cell Culture.

The human liver cancer cell lines HepG2, Huh6, Huh7, and PLC/PRF/5 were obtained from the Health Science Research Resources Bank (Osaka, Japan). These cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen Life Technologies, Carlsbad, CA) containing 10% fetal calf serum and 1% penicillin/streptomycin (Invitrogen), and incubated at 37°C in an atmosphere containing 5% CO2.

Side Population Analysis Using Flow Cytometry.

The cells were detached from the dishes with Trypsin-EDTA (Invitrogen) and suspended at 1 × 106 cells/mL in Hank's balanced salt solution supplemented with 3% fetal calf serum and 10 mmol/L Hepes. These cells were then incubated at 37°C for 90 minutes with 20 μg/mL Hoechst 33342 (Sigma Chemical, St Louis, MO), either alone or in the presence of 50 μmol/L verapamil (Sigma). After incubation, 1 μg/mL propidium iodide (BD Pharmingen, San Diego, CA) was added and then filtered through a 40-μm cell strainer (BD Falcon) to obtain single-suspension cells. Cell analysis and purification were performed using MoFlo carrying a triple-laser (DakoCytomation, Fort Collins, CO). Hoechst 33342 was excited with the UV laser at 350 nm and fluorescence emission was measured with 405/BP30 (Hoechst blue) and 570/BP20 (Hoechst red) optical filters. Propidium iodide labeling was measured through the 630/BP30 filter for the discrimination of dead cells.

Proliferation Assay and Apoptosis Detection.

SP cells and non-SP cells (5 × 103) per well were isolated to a 96-well microplate and 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt (MTS) assay was performed to examine cell viability with triplicate samples as previously described.24 Quantification of apoptotic cells was performed in triplicate by terminal deoxynucleotidyl transferase-mediated nick-end labeling (TUNEL) assay according to the manufacturer's instructions (Takara Bio Inc., Tokyo, Japan). The cells were labeled with fluorescein isothiocyanate using terminal deoxynucleotidyl transferase, and DNA free ends were stained with anti–fluorescein isothiocyanate horseradish peroxidase and diaminobenzidine. The proportion of TUNEL-positive cells was calculated by counting 1 × 103 cells randomly. MTS and TUNEL assays were conducted 24, 48, and 72 hours after cell sorting.

Immunocytochemistry Analysis.

Sorted SP cells and non-SP cells were cultured for 12 hours and washed in phosphate-buffered saline (PBS). The cells were fixed by methanol at −20°C for 20 minutes and washed by PBS containing 0.1% Tween 20 (Wako). After blocking by 10% goat normal serum for 1 hour, fixed cells were incubated with primary antibodies rabbit anti-alpha-1-fetoprotein (AFP) (DakoCytomation) and mouse anti-cytokeratin 19 (CK19) (DakoCytomation) in a moist chamber at 4°C at overnight. They were washed in PBS containing 0.1% Tween 20, blocked again for 30 minutes, and treated with Alexa 555–conjugated goat anti-rabbit IgG (Molecular Probes, Eugene, OR) and Alexa 488-conjugated goat anti-mouse IgG (Molecular Probes) at room temperature for 1 hour. After washing in PBS, the cells were coverslipped with a mounting medium containing DAPI (Vector Laboratories, Burlingame, CA) and examined under an Olympus IX70 microscope.

Non-obese Diabetic/Severe Combined Immunodeficiency Xenograft Transplantation.

NOD/SCID mice were purchased from Sankyo Laboratory Co. Ltd. (Tsukuba, Japan). Various numbers of SP and non-SP cells, ranging from 100 to 1 × 106, were suspended in 200 μL DMEM and Matrigel (BD) (1:1) and transplanted to the male NOD/SCID mice (6 to 10 weeks old) under anesthesia. SP and non-SP cells were injected into the subcutaneous space of the right and left back, respectively. Tumor formation was monitored weekly by observation for 16 weeks. Subcutaneous tumors were fixed in formalin and embedded in paraffin. Sections were subjected to hematoxylin-eosin staining. Next, SP cell–derived tumors were removed and minced in sterile PBS on ice. Obtained small-cut pieces of tumors were put in DMEM containing 5 mg/mL collagenase type II (Sigma) and digested at 37°C for 3 hours. After washing and filtration, cells were cultured for 7 days to exclude non-epithelial cells. Harvested cells were analyzed using flow cytometry and injected into NOD/SCID mice again as mentioned. These experiments were performed in accordance with the institutional guidelines for the use of laboratory animals.

RNA Extraction and Oligonucleotide Microarray Analysis.

Total RNA was extracted separately form SP cells and non-SP cells using Isogen reagent (Nippon Gene, Toyama, Japan) according to the manufacturer's instructions. Microarray analysis was performed according to a standard protocol (Affymetrix GeneChip Manual). Briefly, first and second cDNA were generated from 1 μg total RNA using oligo(dT)24-T7 primer and SuperScript kit (Invitrogen). Biotinylated cRNA was then synthesized using the Bioarray, High Yield RNA Transcript Labeling Kit (Enzo Diagnostics, Farmingdale, NY). Fifteen μg cRNA was hybridized to Human Genome U133A Plus 2.0 Arrays (Affymetrix), which represented more than 47,000 gene transcripts, including 38,500 characterized genes. Array images were scanned using GeneChip Scanner 3000 (Affymetrix), after amplification and detection of the hybridized signals through the streptavidin-conjugated phycoerythrin fluorescence. Obtained data was analyzed with GeneChip Operating Software. The detection P value was calculated using pairs of perfect match and mismatch probes, and probes scored as absent (A) call in both SP and corresponding non-SP cells were excluded from the following analysis. After experimental normalization, data mining including fold change measurement and categorization was conducted using GeneSpring system (Agilent Technologies, Palo Alto, CA).

Quantitative Real-time Reverse Transcriptase Polymerase Chain Reaction.

Reverse transcription (RT) was carried out using first-strand superscript (Invitrogen) after pretreatment with amplification-grade DNase I (Invitrogen). Polymerase chain reaction (PCR) was carried out with TaqMan technology using the ABI PRISM 7000 Sequence Detection System (Applied Biosystems, Foster City, CA). Gene expression assays primer and probe mixes were used for ITGB1, FMN2, ABCF2, USP9X, ZFX, TXNL, WTAP, ELAVL4, AF5Q31, ADARB1, FZD7, ABCB1, STATIP1, GCLM, XRCC5 and β-actin (assay IDs Hs00236976_m1, Hs00298949_m1, Hs00255026_m1, Hs00245009_m1, Hs00268739_m1, Hs00169455_m1, Hs00191727_m1, Hs00222634_m1, Hs00232683_m1, Hs00210562_m1, Hs00275833_s1, Hs00184500_m1, Hs00217448_m1, Hs00157694_m1, Hs00221707_m1, and Hs99999903_m1, respectively; Applied Biosystems). Thermal cycling was performed using 45 cycles of 95°C for 15 seconds and 60°C for 1 minute with triplicates of interest. β-Actin was used for normalization. Relative quantitation was carried out using the comparative cycle threshold (CT) method as described by the manufacturer.

Statistical Analysis.

Data are presented as the means ± SEM. Statistical differences between two groups were analyzed using the Mann-Whitney U test. P values less than .05 were considered significant.


Detection of Side Populations in Hepatocellular Carcinoma Cells.

Flow cytometry analysis with Hoechst 33342 staining demonstrated that Huh7 and PLC/PRF/5 cells included SP cells in 0.25% and 0.80%, respectively. These SP cells were practically diminished in the presence of Hoechst 33342 and verapamil, a calcium channel blocker. Conversely, SP cells were not detected in Huh6 and HepG2 cells. The SP and non-SP cells in Huh7 and PLC/PRF/5 cells were sorted separately and applied for further experiments (Fig. 1). There was no marked difference between SP and non-SP cells in morphology.

Figure 1.

SP cell analysis. HepG2 and Huh6 cells do not contain SP fractions; however, SP cells in Huh7 and PLC/PRF/5 cells were detected in 0.25% and 0.80%, respectively. The SP cells disappeared with Hoechst 33342 and verapamil cotreatment. SP, side population.

Proliferation Activity and Apoptosis Tolerance.

To examine the difference in the in vitro proliferation activity in SP and non-SP cells, we determined the viable cell number using MTS assay. Cell variability in SP cells was consistently higher than in non-SP cells in Huh7 cells (Fig. 2A) and PLC/PRF/5 cells (Fig. 2B). SP cells in 72 hours' culture in PLC/PRF/5 cells showed significantly (P < .05) higher cell viability than non-SP cells. The difference in apoptosis sensitivity in SP and non-SP cells was also investigated by TUNEL assay. The percentage of TUNEL-positive cells in SP cells was less than that of non-SP cells in Huh7 (Fig. 2C) and PLC/PRF/5 cells (Fig. 2D). TUNEL positivity in Huh7 SP cells in 24 and 72 hours' culture and PLC/PRF/5 SP cells in 72 hours' culture demonstrated statistical significance (P < .05) compared with those of non-SP cells.

Figure 2.

Cell viability and apoptosis sensitivity of Huh7 (A, C) and PLC/PRF/5 cells (B, D) at 24, 48, and 72 hours after isolation. SP cells purified from Huh7 and PLC/PRF/5 cells demonstrated higher viability than the corresponding non-SP cells. The number of TUNEL-positive cells in the SP fraction was consistently more than in the non-SP fraction in both cell lines. *, statistically significant. SP, side population; TUNEL, terminal deoxynucleotidyl transferase-mediated nick-end labeling.

Analysis of Lineage Marker Expression.

Dual immunocytochemistry analysis was conducted to compare the differentiation potential between SP and non-SP cells (Fig. 3). Huh7 and PLC/PRF/5 SP cells contained a large number of cells marked with both a hepatocyte-specific marker, alpha-fetoprotein (AFP), and a cholangiocyte-specific marker, cytokeratin 19 (CK19). The percentage of cells positive for both AFP and CK19 was 68.2 and 55.0% in Huh7 and PLC/PRF/5 SP cells, respectively. Conversely, most non-SP Huh7 and PLC/PRF/5 cells were labeled with either AFP or CK19. The percentage of cells that expressed only a single marker was 74.4 and 79.6% in Huh7 and PLC/PRF/5 non-SP cells, respectively. A small number of cells marked with neither AFP nor CK19 existed similarly in SP and non-SP cells in these cell lines.

Figure 3.

Immunocytochemical analysis of sorted Huh7 and PLC/PRF/5 cells. (A-D) AFP (red), and CK19 (green) expression was examined in Huh7 SP (A), Huh7 non-SP (B), PLC/PRF/5 SP (C), and PLC/PRF/5 non-SP cells (D). A large number of cells positive for both markers (yellow) were observed in SP cells (A, C). (E, F) Cells expressing both markers were detected in 68.2 % of Huh7 SP (E) and 55.0% of PLC/PRF/5 SP cells (F). In contrast, cells positive for either AFP or CK19 were preferentially found in Huh7 non-SP (74.4 %) (E) and PLC/PRF/5 non-SP cells (79.6%) (F). Scale bar = (a, b) 20 μm; (c, d) = 50 μm. AFP, alpha-fetoprotein; CK19, cytokeratin 19; SP, side population.

Tumor Initiation Caused by Side Population Injection.

To address the issue of whether tumorigenic activity differs between SP and non-SP cells, various numbers of SP and non-SP cells in Huh7 and PLC/PRF/5 cells were injected into NOD/SCID mice (Table 1). Subcutaneous tumor formation required at least 1 × 106unsorted Huh7 and PLC/PRF/5 cell injection. As low as 1 × 103 Huh7 and PLC/PRF/5 SP cell injection could initiate tumors in eight of nine and eight of eight mice, respectively (Fig. 4A). However, 1 × 106non-SP cells in Huh7 and PLC/PRF/5 cells injection consistently failed to form tumors in all mice injected. These SP cells could be enriched at least 1,000-fold for tumorigenic cells. Histological analysis of Huh7 and PLC/PRF/5 SP cell–originated tumors showed similar features to tumors formed by unsorted cells injection (Fig. 4B-C).

Table 1. Tumorigenicity of SP Cells in NOD/SCID Xenotransplant Assay
 Cell Numbers for Injection
1001 × 1031 × 1041 × 1051 × 1061 × 107
  • NOTE. SP and non-SP cells were isolated separately and injected into the subcutaneous space of NOD/SCID mice. Tumor formation was observed for 16 weeks after injection.

  • *

    A mouse became sick and died without tumor formation 4 weeks after injection.

Unsorted cells   0/55/55/5
SP cells0/58/98/88/8 
Non-SP cells0/50/50/80/80/8
Unsorted cells  0/50/55/55/5
SP cells0/58/88/9*8/8
Non-SP cells0/50/50/90/80/8 
Figure 4.

Tumorigenicity in SP cells. (A) Representative subcutaneous tumors (arrows) due to the injection of 1 × 103 Huh7 SP cells. Hematoxylin-eosin staining of tumors derived from Huh7 SP (B) and PLC/PRF/5 SP cells (C). Transplantation of non-SP cells could not initiate tumors for 16 weeks. Scale bar = 50 μm. SP, side population.

Re-analysis of Side Population–Derived Tumors and Serial Transplantation.

To elucidate whether SP cells self-renew and generate nontumorigenic non-SP cells through asymmetrical cell division in vivo, flow cytometry analysis of SP cell–derived subcutaneous tumors was conducted. SP analysis demonstrated that Huh7 and PLC/PRF/5 cells also include 0.34% and 0.90% SP cells, respectively (Fig. 5). These analysis patterns resembled those of preisolated HCC cells. In culture, Huh7 and PLC/PRF/5 SP cells also generated both SP and non-SP cells after 4 weeks' culture (data not shown). Subsequently, secondary transplantation of SP and non-SP cells into NOD/SCID mice was performed (Table 2). As few as 1 × 103SP cells in both cell lines resulted in tumor formation in four of five NOD/SCID mice, although the injection of non-SP cells failed to form new tumors.

Figure 5.

Re-analysis of SP cell–derived tumors. Huh7 SP cell- and PLC/PRF/5 SP cell-derived tumors consisted of both SP and non-SP cells. The percentages of SP cells in Huh7 SP cell– and PLC/PRF/5 SP cell–derived tumor cells were 0.35% and 0.90%, respectively. The SP analysis patterns are similar to those of original Huh7 and PLC/PRF/5 cells demonstrated in Fig. 1.

Table 2. Tumorigenicity of Resorted SP Cells in Serial Transplantation
 Cell Numbers for Injection
1001 × 1031 × 1041 × 105
  1. NOTE. SP and non-SP cells were isolated separately from digested SP cell–derived tumors and injected into NOD/SCID mice. Transplantation and observation of tumor formation was performed as in the first transplantation.

SP cells0/55/55/54/5
Non-SP cells0/50/50/50/5
SP cells0/44/54/45/5
Non-SP cells0/50/50/40/5

Gene Expression Profile Based on Microarray Analysis.

To clarify differential gene expression profiles between SP and non-SP cells, microarray analysis was performed. In total, 26,764 probes in Huh7 cells and 25,755 probes in PLC/PRF/5 cells were analyzed according to the call algorithms mentioned. The numbers of upregulated (ratio > 2.0) genes in Huh7 and PLC/PRF/5 SP cells compared with corresponding non-SP cells were 1,015 and 1,542, respectively. The genes were functionally categorized using available websites as follows: Gene Ontology ( and GeneCards ( The percentages of characterized genes showing upregulation in Huh7 and PLC/PRF/5 SP cells were 43.3% (440/1,015) and 51.2% (789/1,542), respectively. They were categorized into 11 groups according to their function as follows: transcriptional regulation, nucleic acid binding, signaling, transport, metabolism, protein synthesis, ubiquitination/proteolysis, receptor activity, apoptosis/cell cycle, cytoskeleton/membrane, and others (Fig. 6A). Approximately one third of these genes was categorized into “Transcriptional regulation” and “Nucleic acid binding” groups. Sixty-two upregulated genes were commonly found in both cell lines (Supplementary Table 1; Supplementary material for this article can be found on the Hepatology website ( Conversely, 940 and 941 genes showed downregulation (ratio < 0.5) in Huh7 and PLC/PRF/5 SP cells, respectively, and 51 genes overlapped in both cell lines.

Figure 6.

Microarray-based gene expression profile. (A) Among upregulated (ratio > 2.0) genes in Huh7 and PLC/PRF/5 SP cells, 440 and 789 genes were characterized, respectively. They were also categorized into 11 groups. (B) Overlap of stemness genes among the gene expression profiles in HCC SP cells and stem cell–enriched genes reported by Ramalho-Santos et al.25 and Ivanova et al.26 (C) Fold changes of SP/Non-SP calculated by real-time RT-PCR are comparatively consistent with those of microarray data. ESC, embryonic stem cell; NSC, neural stem cell; HSC, hematopoietic stem cell; SP, side population; RT-PCR, reverse transcription polymerase chain reaction.

Next, the profiling of upregulated genes in SP cells was compared with previous microarray-based profiling of so-called “stemness genes,” which are commonly expressed in stem cells; embryonic stem cells, neural stem cells, and hematopoietic stem cells.25, 26 These profiling data were converted to those of the human homologue using available websites and compared with our present data (Fig. 6B; Table 3). As a result, ITGB1, USP9X, and ZFX in Huh7 SP cells and five genes, including TXNL, WTAP, ABCB1, STATIP1, and GCLM in PLC/PRF/5 SP cells were concordant with the expression profiling of stemness genes reported by Ramalho-Santos et al.25 In addition, XRCC5, which is upregulated in both Huh7 and PLC/PRF/5 SP cells, was documented. FMN2 and ABCF2 in Huh7 SP cells, and four genes including AF5Q31, ADARB1, ELAVL4, and FZD7 in PLC/PRF/5 SP cells, were included in the stemness gene profiling documented by Ivanova et al.26

Table 3. Overlap of Stemness Genes in Huh7 SP and PLC/PRF/5 SP Cells
Gene SymbolGene NameFold ChangeProbe SetUnigeneFunction
  • *

    Overlapped genes in profiling reported by Ramalho-Santos et al.25

  • Overlapped genes in profiling reported by Ivanova et al.26

Huh7 SP cells   
ITGB1*Integrin beta 16.038216178_x_atHs.287797Receptor activity
FMN2Formin 23.1311555471_a_atHs.24889Others
ABCF2ATP-binding cassette, sub-family F, member 22.556214232_atHs.438823Transport
XRCC5*X-ray repair complementing defective repair in Chinese hamster cells 52.279233007_atHs.257082Nucleic acid binding
USP9X*Ubiquitin specific protease 9, X-linked2.207230543_atHs.77578Ubiquitination/Proteolysis
ZFX*Zinc finger protein, X-linked2.062207920_x_atHs.2074Transcriptional regulation
PLC/PRF/5 SP cells
TXNL*Thioredoxin-like 14.211243664_atHs.114412Apoptosis/Cell cycle
WTAP*Wilms tumor 1 associated protein4.0121560274_atHs.446091Others
ELAVL4ELAV (embryonic lethal, abnormal vision, Drosophila)-like 43.128206051_atHs.75236Nucleic acid binding
XRCC5*X-ray repair complementing defective repair in Chinese hamster cells 52.79233007_atHs.257082Nucleic acid binding
AF5Q31ALL1 fused gene from 5q312.64232865_atHs.446658Transcriptional regulation
ADARB1Adenosine deaminase, RNA-specific, B12.6203865_s_atHs.148822Nucleic acid binding
FZD7Frizzled homolog 7 (Drosophila)2.587203706_s_atHs.173859Signaling
ABCB1*ATP-binding cassette, sub-family B (MDR1)2.267243951_atHs.489033Transport
STATIP1*Signal transducer and activator of transcription 3 interacting protein 12.105235623_atHs.8739Others
GCLM*Glutamate-cysteine ligase, modifier subunit2.064236140_atHs.315562Metabolism

Validation of Microarray Data Using Quantitative Real-Time Reverse-Transcription Polymerase Chain Reaction.

We analyzed the mRNA expression of six genes including ITGB1, FMN2, ABCF2, XRCC5, USP9X, and ZFX in Huh7 SP and non-SP cells by quantitative real-time RT-PCR. Ten genes, namely TXNL, WTAP, ELAVL4, XRCC5, AF5Q31, ADARB1, FZD7, ABCB1, STATIP1, and GCLM in PLC/PRF/5 SP and non-SP cells were also examined. Real-time RT-PCR showed that these 17 genes showed higher levels of mRNA expression in SP cells than in non-SP cells, which was closely concordant with the microarray data (Fig. 6C).


The capacity to proliferate extensively, self-renew, and differentiate are the most characteristic properties of stem cells. In this study, we estimated whether SP cells purified from four HCC cell lines, showing different clinical characteristics,27 possess stem cell–like properties in culture and in xenotransplant assay. HepG2, Huh7, and PLC/PRF/5 cells are originated from HCC, and Huh6 cells are from hepatoblastoma. In contrast to Huh6 and PLC/PRF/5 cells, Huh7 and HepG2 cells lack the hepatitis B virus integration. The p53 function is impaired in these cells except for HepG2 cells. SP cells accounting for less than 1% of total cells could be detected in Huh7 and PLC/PRF/5 cells; however, HepG2 and Huh6 cells lacked the subpopulation. HepG2 and Huh6 cells scarcely showed changes in flow cytometry analysis patterns, even when stained with an increased dose of Hoechst dye. Detecting stem-like subpopulations based on high efflux capability might not be desirable, because they are detected in fewer than 30% of the cancer cell lines examined.28

At first, we showed a higher proliferative capacity of SP cells in Huh7 and PLC/PRF/5 cells than that of corresponding non-SP cells in culture, which was accompanied by tolerance to apoptosis. It has been reported that ormal stem/precursor cells of various tissue origins express a high level of anti-apoptotic protein such as Bcl-2.29, 30 Taken together, an anti-apoptotic function might be a preferentially prepared mechanism in both normal and cancer stem cells. Next, we examined the tumorigenic potential and self-renewal ability of SP cells in NOD/SCID xenotransplant assay. Injection of 1 × 103 SP cells purified from Huh7 and PLC/PRF/5 cells could develop tumors; however, non-SP cells, which constitute the majority of the cancer cells, could not form tumors in spite of an increased number of injected cells. Additional markers of SP phenotype might facilitate the enrichment of cancer stem cells as CD44+CD24-/lowepithelial-specific antigen–positive cells in breast cancers.12 These results indicate that only the SP cells possess tumor-initiating capability, and functional and phenotypical heterogeneity also exists in HCC cell lines. The current analysis of SP cells in culture for 4 weeks showed the existence of both SP and non-SP cells. Moreover, SP cell–derived tumors also generated SP and non-SP cells, which indicated both the self-renewal and differentiation capacity of SP cells. The proportion of SP and non-SP cells in SP cell-derived tumors was similar to those of pre-isolated cells in both Huh7 and PLC/PRF/5 cells. It seems that proportions in SP cells and non-SP cells are tightly regulated and maintained in culture and in vivo. Moreover, the secondary transplantation of SP cells also led to tumor formation, and tumorigenic capability was well preserved in serial transplantation.

It is well-known that Huh7 and PLC/PRF/5 cells are both AFP-producing and CK19-positive cell lines.31, 32 It is believed that transformed cells usually succeed in differentiation potential from their origin.33 Analysis of surgical specimens also demonstrated biliary marker–positive HCC, which indicates the possibility that at least some HCC arise from subpopulations with bipotential stem/progenitor properties.34 Our immunocytochemistry analysis obviously showed disproportions in marker expressions between SP and non-SP cells. Cells expressing both AFP and CK19 were highly enriched in these SP cells, whereas most non-SP cells were positive for either AFP or CK19. In consideration of the bipotency of oval cells or hepatoblasts, which are considered putative hepatic stems/progenitors,35 SP cells might exist within the upstream hierarchy of HCC cells and retain a latent capacity to differentiate into hepatocytes and cholangiocytes under appropriate culture conditions.

Comprehensive genomic screening using oligonucleotide array showed distinct gene expression profiling between SP and non-SP cells in Huh7 and PLC/PRF/5 cells. A couple of thousand genes in these SP cells showed altered gene expression, which indicated distinct heterogeneity in HCC cells from the genetic aspects. Among 62 upregulated genes in both Huh7 and PLC/PRF/5 SP cells, only HOXA13 and CYP2E1 genes have been documented to be involved in HCC previously.36, 37 It should be emphasized that SP cells exhibit quite differential mRNA expression patterns compared with those of bulk HCC cell lines31, 37 and surgical samples.38, 39

We also compared data showing a subset of upregulated genes in SP cells with the profiling of so-called “stemness genes.”25, 26 These genes are considered extremely important to maintain functional and phenotypical properties in stem cells. In this analysis, six genes upregulated in Huh7 SP cells and 10 genes upregulated in PLC/PRF/5 SP cells were overlapped with “stemness genes” documented in previous microarray studies. ITGB1 (CD29), upregulated in Huh7 SP cells, is recognized as one of the stem cell markers in the liver. The monoclonal antibody to ITGB1 is often used for hepatic stem cell isolation using flow cytometry.40 FZD7, upregulated in PLC/PRF/5 SP cells, is one of the receptors of the Wnt signaling pathway, which is closely associated with the regulation of stem cell self-renewal. Moreover, WNT5A, WNT6, CTNNB1, and JUN, which are important components of the Wnt pathway, is also upregulated in PLC/PRF/5 SP cells. The dysregulated Wnt pathway sometimes allows tumor cells to proliferate extensively,41 and the oncogenic role of FZD7 has already been reported in HCC.42 Taken together, the results might indicate that these genes are deeply involved in the maintenance of both normal stem cells and cancer stem cells in the liver. Hereafter, it should be examined whether these gene expressions are critically involved in the definition of cancer stem cell-like properties, including tumorigenicity, in SP cells.

It has been reported that the SP phenotype is determined by ABCG2/BCRP1 because a remarkable decrease in SP cells was observed in the bone marrow of ABCG2/BCRP1 null mice43, 44; however, whether other ABC transporters such as ABCB1 (MDR1) are similarly associated with the phenotype is controversial.45 Our microarray analysis showed the upregulation of ABCG1 and ABCF2 in Huh7 SP cells and ABCB2, ABCC7, ABCA5, and ABCB1 in PLC/PRF/5 cells, but not ABCG2/BCRP1 in both SP cells. It was also documented that ABCA3 is more highly expressed in neuroblastoma SP cells than in non-SP cells.46 Taken together, different mechanisms to export Hoechst dye might be operating between normal cells and transformed cells. Although it is not technically feasible to detect SP cells in HCC samples in situ, immunohistochemical examination of these ABC transporters might help understanding the localization of cancer stem cells and their special microenvironments in HCC. Approaches such as this would be of use to elucidate the origin of HCC and clarify the molecular mechanisms underlying hepatocarcinogenesis.

In conclusion, we were able to define tumorigenic subpopulations using SP cell sorting in Huh7 and PLC/PRF/5 cells. These data indicate the heterogeneity characterized by distinct hierarchy from cancer stem cells down to their progenies in HCC cell lines. Most of recent cancer research and therapeutic modalities appear to consider tumors as masses consisting of comparatively homogenous cells, but it is necessary to recognize heterogeneity in tumors and seek a therapeutic approach for targeting cancer stem cells. That SP cells exhibit low sensitivity to anti-cancer reagents in culture has been documented.46, 47 Further studies for the identification and characterization of cells with stem cell–like properties in primary HCC samples might contribute to the establishment of novel therapeutic strategies.


The authors thank Yohei Morita, Yuji Yamazaki, and Daiki Matsubara for technical support in flow cytometry. We also thank Naoko Sasaki for laboratory assistance.