Annals of Surgical Oncology

, Volume 15, Issue 10, pp 2927–2933

CD133+CD44+ Population Efficiently Enriches Colon Cancer Initiating Cells

Authors

  • Naotsugu Haraguchi
    • Department of Gastroenterological Surgery, Graduate School of MedicineOsaka University
  • Masahisa Ohkuma
    • Department of Surgery, Medical Institute of BioregulationKyushu University
  • Hiroyuki Sakashita
    • Department of Surgery, Medical Institute of BioregulationKyushu University
  • Shinji Matsuzaki
    • Department of Surgery, Medical Institute of BioregulationKyushu University
  • Fumiaki Tanaka
    • Department of Surgery, Medical Institute of BioregulationKyushu University
  • Koshi Mimori
    • Department of Surgery, Medical Institute of BioregulationKyushu University
  • Yukio Kamohara
    • Department of Surgery, Medical Institute of BioregulationKyushu University
  • Hiroshi Inoue
    • Department of Surgery, Medical Institute of BioregulationKyushu University
    • Department of Gastroenterological Surgery, Graduate School of MedicineOsaka University
Gastrointestinal Oncology

DOI: 10.1245/s10434-008-0074-0

Cite this article as:
Haraguchi, N., Ohkuma, M., Sakashita, H. et al. Ann Surg Oncol (2008) 15: 2927. doi:10.1245/s10434-008-0074-0

Abstract

Background

Previous reports have demonstrated that CD133+ cells or CD44+ cells might be cancer initiating cells (CIC) of colon cancer. However, the association between the two cell types is unclear. In this study, we evaluated the tumorigenicity of each population of human colon cancer divided by CD133 and CD44 using non-obese diabetic/severe combined immunodeficient (NOD/SCID) mice.

Methods

Using the colon cancer cell lines HT29 and Caco2 we evaluated the change of expression status of CD133 or CD44 by a treatment with sodium butyrate (NaBT) that can induce cellular differentiation. Next, we prepared ten clinical samples of colon cancer and analyzed the expression and tumorigenicity of CD133 and CD44.

Results

With NaBT treatment, CD44 expression was greatly downregulated in both HT29 and Caco2 (HT29: nontreatment versus treatment; 77.8% versus 0.6%, Caco2: 14.0% versus 0.4%, respectively), more than CD133 expression (HT29: nontreatment versus treatment; 90.1% versus 67.7%, Caco2: 98.9% versus 76.3%, respectively). In clinical samples, the percentages of CD133+ cells and CD44+ cells varied from 0.3% to 82.0% (mean 35.5%), and from 11.5% to 58.4% (mean 30.0%), respectively. Subcutaneous injection of CD133+ or CD44+ cells made a tumor in all mice (3/3 and 4/4, respectively). The combined analysis of CD133 and CD44 revealed that only the CD133+CD44+ population had the ability to produce a tumor (3/3).

Conclusion

The findings demonstrate that, at present, the CD133+CD44+ population may be the best to identify tumor initiating cells of human colon cancer.

Keywords

CD133CD44Colon cancerColon cancer initiating cellsSodium butyrate

Recently the cancer stem cell theory has been introduced not only in leukemia1 but also in several kinds of solid cancers.26 According to this theory, only cancer stem cells that exist as a minor population in a tumor can produce a tumor and maintain it.7 CD133+ cells or CD44+ cells were reported to be cancer stem cells or tumor initiating cells that produce a tumor of human colon cancer.811 However, there have been few studies demonstrating the correlation between the two populations identified by CD133 or CD44, and evaluating the ability of tumor formation using both markers. In this study we therefore aimed to clarify the correlation between these two populations and whether it is possible to better identify tumor initiating cells using both CD markers. The results indicate that both CD133+ and CD44+ cells might be real tumor initiating cells of human colon cancer.

MATERIALS AND METHODS

Tumor Cell Preparation and Cell Culture

Human colorectal cancer cell lines, HT29 and Caco2, were used in the following analysis. HT29 was cultured in McCoy’s medium 5A (Invitrogen Corporation, CA, USA)/10% fetal bovine serum (FBS; Equitech-Bio, TX, USA) with 2 mM L-glutamine (Invitrogen Corporation, CA, USA), 100 μg/ml penicillin G, and 100 unit/ml streptomycin (Invitrogen Corporation, CA, USA). Caco2 was cultured in Roswell Park Memorial Institute (RPMI) 1640 medium (Invitrogen Corporation, CA, USA)/10% FBS with 100 μg/ml penicillin G and 100 unit/ml streptomycin. The cells were cultured at 37°C in a humidified atmosphere containing 5% CO2. HT29 was obtained from American Type Culture Collection (ATCC), and Colo201 was obtained from Japanese Collection of Research Bioresources Cell Bank (JCRB).

Ten samples from cases of colorectal cancer were obtained from Kyushu University at Beppu and Oita Red Cross Hospital after obtaining patients’ informed consent and approval by the Research Ethics Board at Kyushu University in Japan. Surgical specimens were subsequently subjected to mechanical and enzymatic digestion as described below. Digested tumor cells were prepared for flow cytometric analysis and cell sorting, or were injected into non-obese diabetic/severe combined immunodeficient (NOD/SCID) mice to obtain further large numbers of tumor cells.

Sodium Butyrate Treatment and Alkaline Phosphatase Assay

Cells of HT29 and Caco2 colon cancer cell lines were dissociated with 0.25% trypsin and 0.02% ethylenediamine tetraacetic acid (EDTA) and then 1 × 106 cells were seeded in 10-cm plastic flasks (BD FALCON, CA, USA) at day 0. At day 1, sodium butyrate (NaBT; Wako, Osaka, Japan) was added at concentration of 0, 3, 5, and 8 mM and incubated for 72 h.1214 The change in alkaline phosphatase13 level and activity13 was detected by enzyme-linked immunosorbent assay (ELISA) with SensoLyte™ pNPP alkaline phosphatase assay kit (AnaSpec, CA, USA) according to the manufacturer’s protocol.

Transplantation of Human Colon Cancer Cells into NOD/SCID Mice

Colon cancer cells were suspended in a 1:1 mixture of media and Basement Membrane Matrix High Concentration (BD Biosciences, CA, USA) and injected subcutaneously into the flank of NOD/SCID mice (5 weeks of age) with anesthesia. After 12–16 weeks all mice were sacrificed by cervical dislocation, and tumors were removed and analyzed by flow cytometry or histology. All of the xenograft lines were originally implanted into NOD/SCID mouse subcutaneously and never cultivated and expanded in vitro. To study the tumorigenecity of CD44+ and CD133+ cells, 1 × 104 cells from each population were injected subcutaneously into NOD/SCID mice. Tumorigenecity was evaluated at 6 weeks after NOD/SCID transplantation.

Tumor Digestion

Colon cancer tissues were cut into small fragments, further minced with a sterile scalpel, and washed twice with Dulbecco’s modified Eagle medium (DMEM)/10% FBS with 100 μg/ml penicillin G and 100 unit/ml streptomycin (Invitrogen Corporation, CA, USA), then placed in DMEM/10% FBS with 2 mg/ml collagenase A (Roche Diagnostics, Basel, Switzerland) solution. The mixture was incubated at 37°C for up to 2 h to allow complete digestion. Every 15 min, the solution was mixed through a 10-ml pipette to encourage dissociation. Cells were filtered through 40-μm nylon mesh and washed twice, and then cell fragments and debris were eliminated by using Ficoll (GE Healthcare, Buckinghamshire, UK) density gradient centrifugation. Cells were stained for flow cytometry or following transplantation into NOD/SCID mouse.

Flow Cytometric Analysis and Cell Sorting

To characterize colon cancer initiating cells, the following antibodies were used: allophycocyanin (APC)-conjugated anti-human CD133/1 (clone AC133, mouse IgG2a, Miltenyi-Biotec, Bergisch Gladbach, Germany), phycoerythrin (PE)-conjugated anti-human CD44 (clone G44-26, mouse IgG2b, BD Pharmingen, CA, USA). Fluorescein isothiocyanate (FITC)-conjugated anti-human CD326 (EpCAM; clone HEA-125, mouse IgG1, Miltenyi-Biotec, Bergisch Gladbach, Germany).

To isolate human cells from mouse xenografts, biotinylated anti-mouse H-2Kd (clone SF1-1.1, mouse IgG2a, BD Pharmingen, CA, USA) and biotinylated anti-mouse CD45 (clone 30-F11, mouse IgG2b, BD Pharmingen, CA, USA) were used. Streptavidine-APC-Cy7 (BD Pharmingen, CA, USA) was used as secondary antibody. Doublet cells were eliminated by forward scatter-hight/forward scatter-width (FSC-H/FSC-W) and side scatter-hight/side scatter-width (SSC-H/SSC-W). Dead and dying cells were eliminated with 7-amino actinomycin D (7-AAD) (BD Pharmingen, CA, USA). FcR blocking was applied with FcR blocking reagent (Miltenyi-Biotec, Bergisch Gladbach, Germany).

Samples were analyzed and sorted with FACSVantage SE flow cytometers (Becton Dickinson, CA, USA) and data were analyzed with Diva software (Becton Dickinson, CA, USA).

RESULTS

NaBT-Induced Differentiation and Changes of Cell Surface Markers

In colon cancer, especially in HT29 and Caco2 colon cancer cell lines, cell differentiation of colonic epithelium stimulated by NaBT is well known.12,14 The expression of CD44 and CD133 were analyzed in untreated and NaBT-treated lines. The effect of NaBT was estimated by alkaline phosphatase (ALP) expression,13 a well-known differentiated colonocyte marker, with ELISA assay. The expression of ALP was induced mostly at 5 mM concentration of the NaBT in both HT29 and Caco2 cell lines (Fig. 1).
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-008-0074-0/MediaObjects/10434_2008_74_Fig1_HTML.gif
Fig. 1

NaBT-induced differentiation in colon cancer cell lines. HT29 and Caco2 were treated with various concentrations of sodium butyrate (NaBT). The optimum density of the NaBT was evaluated by the change of alkaline phosphatase (ALP) level with ELISA. The expression level of ALP was the highest in both HT29 and Caco2 at the 5 mM NaBT concentration.

CD44 expression was greatly downregulated in both HT29 and Caco2 by the NaBT treatment (HT29: nontreatment versus treatment; 77.80% versus 0.60%, Caco2: 14.00% versus 0.40% respectively). On the other hand, CD133 expression was 90.10% and 67.70% in untreated and treated lines, respectively, in HT29, and 98.90% and 76.30% in Caco2, respectively (Table 1). These findings confirm that CD44 and CD133 were useful markers to identify undifferentiated or primitive colon cancer cells.
Table 1

NaBT-induced differentiation and changes of cell surface markers

Cell Surface marker

Cell samples

HT29 untreated (%)

HT29 NaBT treated (%)

Caco2 untreated (%)

Caco2 NaBT treated (%)

CD44

77.80

0.60

14.00

0.40

CD133

90.10

67.70

98.90

76.30

Flow Cytometric Analysis in Primary Colon Cancer

The expression of CD44 or CD133 was studied in four primary colon cancers and six NOD/SCID xenografted lines. The expression of CD44 varied from 11.5% to 58.4% (mean 30.04 %), and that of CD133 varied from 0.3% to 82% (mean 35.5 %). Then CD44 and CD133 expression was analyzed by multicolor flow cytometry. In all ten cases, CD133+CD44+ population was identified, ranging from 0.2% to 50.5% (mean 15.73 %) (Table 2). To confirm whether the population derived from epithelial cells or not, FITC-CD326 (EpCAM) was used. CD133+CD44+ population also expressed CD326, thus, this population was certainly derived from human epithelial cell (Fig. 2A and B).
Table 2

Flow cytometric analysis in primary colon cancer

Sample

Origin

Expression of cell surface marker (%)

CD44+

CD133+

CD133+CD44+

CK070228*

Sigmoid colon

24.8

0.3

0.2

CK070228*

lymph node metastasis (Sigmoid colon)

18.8

0.9

0.4

CK070319*

Liver metastasis

46.4

9.2

3.5

CK070326

Liver metastasis

58.4

82.0

50.5

CK070326 NOD/SCID xeno*

Liver metastasis

24.7

4.3

3.5

CK070606 NOD/SCID xeno

Transverse colon

27.2

77.4

25.6

CK070426 NOD/SCID xeno*

Rectal cancer

43.3

39.5

25.7

CK070514 NOD/SCID xeno

Sigmoid colon

11.5

75.9

10.6

CK070611 NOD/SCID xeno

Rectal cancer

14.7

15.2

11.2

CK070317 NOD/SCID xeno

Liver metastasis

30.6

50.3

26.1

Mean

 

30.04

35.50

15.73

* Groups contain CD133CD44, CD133+CD44, CD133CD44+, and CD133+CD44+ populations.

 Groups contain CD133CD44, CD133+CD44, and CD133+CD44+ populations.

https://static-content.springer.com/image/art%3A10.1245%2Fs10434-008-0074-0/MediaObjects/10434_2008_74_Fig2_HTML.gif
Fig. 2

Flow cytometry analysis of NOD/SCID xenografted lines. NOD/SCID xenografted line was digested and analyzed by flow cytometry. Doublet and dead cells were eliminated by FSC-H/FSC-W, SSC-H/SSC-W, and 7-AAD (P2∼P4 gate). Anti-mouse H2-Kd and anti-mouse CD45-negative population (P5 gate) were used for further analyses. A combined analysis with CD133 and CD44 demonstrated three populations: CD133CD44, CD133+CD44 and CD133+CD44+. Anti-human EpCAM was used to confirm whether P5-gated population expressed epithelial cell marker or not. The P7 gate was EpCAM positive. A black arrow shows the P1-gated population and a white arrow shows the P5-gated population.

A combined analysis of CD133 and CD44 showed that the primary or the xenografted colon cancer cells could be classified into two groups: one containing CD133CD44, CD133+CD44, CD133CD44+, and CD133+CD44+ populations (5/10 cases) (Figs. 3A, 4, Table 2), and the other containing CD133CD44, CD133+CD44, and CD133+CD44+, but not CD133CD44+ (5/10 cases) (Fig. 3B, Table 2).
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-008-0074-0/MediaObjects/10434_2008_74_Fig3_HTML.gif
Fig. 3

Combined analysis of CD133 and CD44. A combined analysis revealed two groups. One group showed the four populations CD133CD44, CD133+CD44, CD133CD44+, and CD133+CD44+, and this group contained five cases (A). The other group showed the three populations CD133CD44, CD133+CD44, and CD133+CD44+. This group showed none or very low expression of CD133CD44+. This group contained the remaining five cases (B).

https://static-content.springer.com/image/art%3A10.1245%2Fs10434-008-0074-0/MediaObjects/10434_2008_74_Fig4_HTML.jpg
FIG. 4

CD133+CD44+ population shows the ability to form a tumor in NOD/SCID mouse. Ten thousand cells of CD133CD44, CD133+CD44, and CD133+CD44+ populations obtained from CK070611 xenografted line (Fig. 2) were injected into NOD/SCID mouse subcutaneously. Each arrow shows the injected area. Note that the CD133+/CD44+ population made a tumor in NOD/SCID mouse.

Tumorigenicity in NOD/SCID Mouse

CD44+ population formed a tumor in 4/4 cases, with a mean major axis of 0.8 ± 0.14 cm. Conversely, CD44 population formed no tumor (0/3 cases). CD133+ population also formed a tumor in 3/3 cases, with a mean major axis of 0.83 ± 0.28 cm. CD133 population did not form a tumor (0/2 cases). Ten thousand whole cancer cells were injected into NOD/SCID mouse as a control. The size of the control tumor was smaller (0.2 cm) than that of CD44+ or CD133+ populations (Table 3).
Table 3

Tumorigenicity of CD133 and CD44 in NOD/SCID mouse

 

Control

CD44

CD44+

CD133

CD133+

Tumor formation

1/1

0/3

4/4

0/2

3/3

Tumor size (cm)

0.2

0

0.8 ± 0.14

0

0.83 ± 0.28

To specify the population of NOD/SCID colon cancer initiating cells, CD133CD44, CD133+CD44, and CD133+CD44+ populations were isolated and 1 × 104 cells from each population were injected subcutaneously into NOD/SCID mouse. Interestingly, only CD133+CD44+ population made a tumor in NOD/SCID mouse (3/3 cases; tumor size 1.0 cm), but CD133+CD44 or CD133CD44 populations did not produce a tumor (Table 4, Fig. 3).
Table 4

Tumorigenicity of each combined populations in NOD/SCID mouse

 

CD133CD44

CD133+CD44

CD133+CD44+

Tumor formation

0/3

0/3

3/3

Tumor size (cm)

0

0

1.2 ± 0.35

DISCUSSION

This study demonstrated that the CD133+ population varied from 0.3% to 82.0% with a mean of 35.5% in colon tumors. CD133+ cells showed significantly higher tumorigenicity in NOD/SCID mice than did CD133 cells. This result was very similar to the reports by O’Brien et al. and Ricci-Vitiani et al.9,10 On the other hand, the CD44+ population varied from 11.5% to 58.4% with a mean of 30.0%. CD44+ cells showed significantly higher tumorigenicity in NOD/SCID mice than did CD44 cells. Thus this study confirmed that CD133+ or CD44+ is an adequate marker for tumor initiating cells of human colon cancer.

Cancer initiating cells or cancer stem cells are considered to be immature cells, and they may occupy a small percentage of the tumor. It is difficult to identify appropriate markers that can detect such cells. One conceivable method is to use NaBT, which can induce differentiation in the cell.1217 Human colon cancer cell lines, HT29 and Caco2, were treated with NaBT, and the results demonstrated that both CD133+ and CD44+ cells decreased dramatically. The rate of decrease was much greater in CD44 than CD133, suggesting that CD44 may be more suitable as a marker of immature cells than CD133.

As mentioned above, CD133 and CD44 are adequate markers of cancer initiating cells of human colon cancer, however, there are few studies that described the association between the two markers. Dalerba et al. described that most CD44+ cells were CD133+. They thus suggested that CD44 may be more appropriate as a cancer initiating cell marker than CD133.11 This suggestion may be partly supported by our experiment using NaBT, which showed that CD44 may be a more suitable marker of immature cells than CD133.

Our combined analysis demonstrated that there are two different groups. One group of five tumors consisted of four populations including both positive (CD133+CD44+), both negative (CD133CD44), and one positive (CD133+CD44 or CD133CD44+). The other group of five tumors consisted of three populations, with CD133CD44+ lacking. In our study, CD133+ cells were detected more or less in all ten clinical samples.

We studied the tumor formation ability of the three populations CD133CD44, CD133+CD44, and CD133+CD44+, among which only the CD133+CD44+ population exhibited tumorigenicity. It was interesting that the CD133+CD44 population did not exhibit tumorigenicity. This result suggests that it would be much better to use the combined markers to identify cancer initiating cells of human colon cancer than either CD133 or CDE44 as a single marker. Our study is alsio interesting because it clarifies the significance of the combined use of the two major markers. We did not study the tumorigenicity of CD133CD44+ population because sufficient numbers of this cell population could not be captured. Our study demonstrated that CD133+ or CD44+ cells show tumorigenicity. On the other hand, CD133 or CD44 cells did not show tumorigenicity. The CD133+CD44 population also did not show tumorigenicity. Thus it would be imagined that the CD133CD44+ population may show no tumorigenicity. However, further study is needed to confirm this.

In summary, this study indicated that the CD133+CD44+ population may be real cancer initiating cells of human colon cancer.

Acknowledgements

This study was supported by CREST of Japan, Science and Technology Agency; a Grant-in-Aid for Scientific Research (S) (17109013) from the Japan Society for the Promotion of Science; a Grant-in-Aid for Scientific Research on Priority Areas (17015032) from the Ministry of Education, Culture, Sports, Science and Technology of Japan; a Grant for The Third Term Comprehensive Ten-year Strategy for Cancer Control, and a Grant for Cancer Research from the Ministry of Health, Labor and Welfare of Japan.

The authors would like to thank Miss T. Shimooka, Miss Y. Nakagawa, and Miss M. Kasagi for their excellent technical assistance.

Copyright information

© Society of Surgical Oncology 2008