The following were purchased: Histofine Simple Stain MAX PO (R) and (M) kits from Nichirei (Tokyo, Japan); mouse monoclonal anti-CD24 antibody from Santa Cruz Biotechnology (Santa Cruz, CA, USA); mouse monoclonal anti-CD44, anti-nestin antibody, mouse IgG1, and mouse IgG2A from R&D Systems Inc. (Westerville, OH, USA); mouse polyclonal ESA antibody from Abnova (Taipei City, Taiwan), rabbit polyclonal anti-CD133, anti-CXCR4 antibody, rabbit IgG, and mouse IgG from Abcam plc. (Cambridge, UK); mouse monoclonal anti-Ki 67 antibody and anti-PCNA antibody from Dako Corp. (Santa Barbara, CA, USA); High Capacity cDNA Reverse Transcription kit and TaqMan Gene Expression assay for CD24 (Hs02379687_s1), CD44 (Hs00153304_m1), ESA (Hs00901885_m1), CD133 (Hs01009238_m1), CXCR4 (Hs00607978_s1), nestin (Hs00707120_s1), and 18S ribosomal RNA (18S rRNA, Hs03928990_g1) from Applied Biosystems (Foster City, CA, USA); NucleoSpin RNA II from Takara Biotechnology (Tokyo, Japan); human serum from Lonza (Walkersville, MD, USA); Zenon Labeling kit from Invitrogen Corp. (Carlsbad, CA, USA); and SureLink Fluorescein (FITC) Labeling kit from KPL, Inc. (Gaithersburg, MD, USA). All other chemicals and reagents were purchased from Sigma Chemical Corp. (St. Louis, MO, USA).

Tissues from 105 patients with conventional PDAC were obtained for this study (Table I). These patients received treatment at Nippon Medical School Hospital (Tokyo, Japan) between 1995 and 2011. None of the patients received preoperative chemotherapy and radiotherapy. The patients consisted of 61 males and 44 females, with a median age of 68 years (range, 35–87 years). Clinicopathological stages were determined according to the TNM classification system of the Union for International Cancer Control (UICC). The mean follow-up period was 17.2 months (range, 0.4–153.1 months). Forty-six patients did not receive any postoperative chemotherapy, whereas 59 patients received adjuvant chemotherapy following surgery; 34 patients received gemcitabine, 14 received uracil/tegafur, 6 received gemcitabine and tegafur/gimeracil/oteracil potassium, 3 received uracil/tegafur and gemcitabine and 2 received fluorouracil.

Two PDAC subtypes, adenosquamous carcinomas (N=7) and anaplastic carcinoma (N=1), were separately analyzed. The normal pancreatic tissues (N=6) were obtained from patients who underwent surgery for ectopic spleen. PanIN tissues (N=41) were obtained from both PDAC and normal tissues. This study was carried out in accordance with the principles embodied in the Declaration of Helsinki, 2008; each patient gave informed consent for the use of the pancreatic tissues.

Paraffin-embedded sections (3.5 μm) were subjected to immunohistochemistry (IHC). After deparaffinization, antigen retrieval was performed (except for nestin) at 121°C for 15 min in a sodium citrate buffer solution (pH 6.0). Endogenous peroxidase activity was blocked by incubation for 30 min with 0.3% hydrogen peroxide in methanol. The tissue sections were then incubated with each antibody in phosphate-buffered saline (PBS) containing 1% bovine serum albumin (BSA) overnight at 4°C. The dilutions of each primary antibody are listed in Table II. Bound antibodies were detected with Simple Stain MAX PO (R) or (M) reagent, using diaminobenzidine tetrahydrochloride as the substrate. The sections were then counterstained with Mayer’s hematoxylin. Negative control tissue sections were prepared by omitting the primary antibody.

IHC staining was evaluated independently by three investigators (S.K. and Y.M. for PanIN and Y.M and T.I. for PDAC) who were blind to clinical and outcome data. Percentages of positive cells (0–100%) were evaluated for each stained sample. For PDACs, evaluation was performed within the tumor area of an individual slide. For PanINs and normal ducts, one or more ducts were evaluated from one slide (10) and each case or duct was subdivided into high- and low-expression groups according to the median percentage of positive cells for each marker.

KLM-1, PANC-1, MIA PaCa-2, PK-1, PK-45H and PK-8 PDAC cell lines were obtained from the Cell Resource Center for Biomedical Research, Institute of Development, Aging and Cancer, Tohoku University (Sendai, Japan), and Capan-1 was purchased from the American Type Culture Collection. The AsPC-1 cell line was obtained from Dainippon Sumitomo Pharma Co. Ltd. (Osaka, Japan). The cells were grown in RPMI-1640 medium containing 10% fetal bovine serum (FBS) at 37°C under a humidified 5% CO₂ atmosphere. Capan-1 cells were incubated in the same medium with 15% FBS.

All PDAC cells were seeded at 2.5×10⁵ cells per 60-mm dish, and were incubated for 48 h. Total RNA extraction was performed using NucleoSpin RNA II. Then, cDNA synthesis was performed using a High Capacity cDNA Reverse Transcription kit, following the manufacturer’s protocol. Quantitative real-time PCR (qRT-PCR) was performed using the StepOnePlus Real-time PCR systems (Applied Biosystems, Inc.). The PCR reaction mixture (18 μl), containing 2 μl of template cDNA, 10 μl of TaqMan Fast Universal PCR Master Mix, and 1 μl of the TaqMan Gene Expression assay for CD24, CD44, CD133, CXCR4, ESA, or nestin, was placed in a 96-well reaction plate. 18S rRNA, as the internal positive control, was amplified using the TaqMan gene expression assay. The optimized program involved denaturation at 95°C for 20 sec, followed by 40 cycles of amplification (95°C for 1 sec and 60°C for 20 sec). Results were expressed as target/18S rRNA, as an internal standard concentration ratio. Gene expression measurements were performed in triplicate.

For flow cytometry (FCM), we used the same antibodies as were used for IHC (Table II). CD24, CD44, CD133, CXCR4, and nestin antibodies were labeled with allophycocyanin (APC) using the Zenon mouse or rabbit IgG Labeling kit, and ESA antibody was labeled with FITC using the SureLink Labeling kit according to the manufacturer’s protocol. Cells were incubated for 20 min at 4°C in PBS containing 10% human serum. Cells (5×10⁵ cells) were then centrifuged briefly to remove the serum-containing medium, and incubated with antibody for 60 min in the dark at 4°C; 1 μg of propidium iodide was added to label dead cells. To stain nestin, cells were fixed with 4% paraformaldehyde, and incubated with the anti-nestin antibody and 0.1% Triton-X for 60 min in the dark at room temperature. We prepared isotype control-treated cells as a negative control (Table II). Each marker’s expression was determined using a BD FACSAria II flow cytometer (BD Bioscience, Franklin Lakes, NJ, USA) and analyzed using FlowJo version 7.6.4 (Tree Star, Inc., Ashland, OR, USA).

Results for cell proliferation are shown as mean ± SE, and the data between different groups were compared using the Student’s t-test. Whenever indicated, the Chi-square test and Fisher’s exact test were used to analyze the correlation between marker expression and clinicopathological features. Cumulative survival rates were calculated by the Kaplan-Meier method, and the significance of differences in survival rate was analyzed by the log-rank test. Data between multiple groups were compared using one-way ANOVA, then were analyzed using a post-hoc test. P<0.05 was considered significant in all analyses. Computations were performed using the StatView J version 5.0 software package (SAS Institute, Inc., Cary, NC, USA).

First, we conducted immunohistochemical analyses of various CSC markers, in order to confirm the localizations and differences among normal, PanIN, and PDAC tissues. Normal pancreatic tissues were mostly negative for CD24, CD44, CXCR4, and nestin, whereas CD133 and ESA were relatively highly expressed in ductal cells (Fig. 1). PanIN-3 lesions showed weakly positive expressions of CD44, CD133, and nestin in ductal cells (Fig. 2). CD24, CXCR4 and ESA were moderately to strongly expressed in the ductal cells of PanIN-3 lesions, with different intracellular localization patterns: CD24 in luminal surface of cell membrane, CXCR4 in cells with a granular staining pattern, and ESA diffusely distributed both in cell membrane and cytoplasm (Fig. 2). PDAC showed expressions of CD24, CD44, and CD133 in cell membranes of cancer cells; CXCR4 and nestin were expressed in cytoplasm, and ESA was expressed in both cytoplasm and cell membrane (Fig. 3).

Next, we compared the percentages of immunohistochemical expression of each CSC marker in normal, PanIN, and PDAC tissues (Table I; Fig. 4). The expression levels of CD24, CD44, CXCR4, ESA, and nestin showed tendencies to increase according to the malignancy grade of PanINs, but CD133 showed an opposite trend (Fig. 4). As compared with normal, PanIN-1, and PanIN-2 tissues, significantly increased CD24 expression was observed in PanIN-3 and PDAC (Fig. 4A; ^**P<0.01). Expressions of CD44 and CXCR4 were significantly increased in PanIN-2, PanIN-3 and PDAC (Fig. 4B and D; ^*P<0.05; ^**P<0.01). ESA staining was relatively stronger compared to the other markers, and all grades of PanINs and PDAC exhibited significantly higher positivity than normal tissues (Fig. 4E; ^*P<0.05; ^**P<0.01). PDAC showed significantly higher positivity for nestin compared to normal, PanIN-1 and PanIN-2 tissues (Fig. 4F; ^**P<0.01). In contrast to the other markers, CD133 showed highest positivity in normal epithelial duct (Fig. 4C; ^**P<0.01).

To confirm that the PanIN lesions selected in the present study possessed high proliferative ability and that they exist as pre-cancerous lesions for PDAC, we analyzed Ki 67- and PCNA-labeling indices. Both Ki 67- and PCNA-labeling indices increased along with the PanIN grades, and PDAC showed significantly higher proliferative activity than normal and PanIN tissues (data not shown). These findings suggest that CSC markers, including CD24, CD44, CXCR4, ESA and nestin, are involved in proliferative activity or carcinogenesis stages of PDAC via PanIN lesions.

Next, we examined the correlations between each CSC marker and the clinicopathological features of PDACs. Nestin exhibited the lowest expression, followed, in order of increasing expression, by CD133, CD44, CD24, CXCR4 and ESA (Table III). The CSC marker expression levels widely varied in PDAC tissues; thus, we analyzed clinicopathological aspects for the high- and low-expression groups, using each marker’s median as a cut-off. Correlations between the expression levels of each CSC marker and the clinicopathological characteristics of conventional PDAC were analyzed. A statistically significant correlation was found between CD133 expression levels in PDAC and pT stages (P= 0.0494), but these data may be biased by great differences in the number of cases for each T stage. As not expected, ESA and CXCR4 expressions in PDAC were inversely associated with histological grade, with well-differentiated PDAC tending to express CXCR4 and ESA (P=0.0112 and P=0.0058, respectively). Regarding metastatic features, more severe status of venous invasion was associated with higher CD133 expression (P= 0.0056) and lower ESA expression (P=0.0243). There were no significant differences between expression of CSC markers and gender, age and TNM stage. These results may indicate that the expressions of some CSC markers are associated in a different manner with differentiation and invasiveness of PDAC.

Overall survival was not significantly correlated with any of the six CSC markers (Fig. 5). Disease-free survival was not correlated with expression of CSC markers (data not shown). However, higher nestin expression tended to be associated with worse overall and disease-free survivals (P=0.0957 and P=0.0840, respectively). Previous reports have shown that analysis of co-expressions of some CSC markers can be more effective for detecting the CSC fraction (5,10); therefore, we also performed survival analysis using co-expressions of CD24, CD44, and ESA, or co-expression of CXCR4 and CD133. These analyses did not find any significant correlations with survival (data not shown). Furthermore, we performed the survival analyses with the cut-off for positivity set at 10 and 30%, but still found no statistically significant correlation (data not shown).

Next, we examined the mRNA levels of each CSC marker, and the corresponding protein levels in established PDAC cell lines. All CSC markers were expressed in the PDAC cell lines at various levels (Fig. 6). As in the immunohistochemistry results, CD133 was weakly expressed in all lines except for Capan-1 (Fig. 6C).

We also conducted FCM using PDAC cell lines, PANC-1, MIA PaCa-2 and KLM-1 (Fig. 7). Expression levels of each CSC marker varied depending on the cell line. Protein levels of CD24, CD44, CXCR4, and nestin were measured by FCM in PANC-1, MIA PaCa-2, and KLM-1 cells and were correlated with their mRNA levels. Consistent with IHC and qRT-PCR analysis, FCM showed very low CD133 expression (Fig. 7C). On the other hand, we observed relatively high expression levels of CD44 and nestin compared to other markers, which differed from the IHC results (Fig. 7B and F). The expression levels of each marker were detected in the following order of increasing percentage in PDAC cell lines: CD133 < nestin < CD24 < CXCR4 < ESA < CD44 (Table III). The order of increasing percentages of CSC markers in PDAC cell lines and PDAC tissues were not identical.

Each CSC marker showed a tendency of having a different expression pattern depending on histological types or differentiation of PDAC cells, and expressions widely varied between each case and cell line. These findings led us to hypothesize that histological variety of PDACs determines the CSC marker expression levels. Therefore, we analyzed two PDAC subtypes, adenosquamous carcinoma (N=7) and anaplastic carcinoma (N=1), which consist of histologically different cancer cells within an individual sample. In adenosquamous carcinoma, CD133 and ESA were expressed predominantly in the adenocarcinoma component, whereas CD44 was far more strongly expressed in the squamous cell carcinoma component (Fig. 8). CD24, CXCR4, and nestin were expressed at similar levels in each component. ESA showed significantly higher expression in the adenocarcinoma component than in the squamous cell carcinoma component (Table IV, P= 0.0048). In anaplastic carcinoma, CD24 and ESA were predominantly expressed in the adenocarcinoma component, while CD44 and nestin were predominantly expressed in the anaplastic component (Fig. 9). These results indicate that expression of each CSC marker widely varies in human tissues depending on histological type.