Differential expression of aldehyde dehydrogenase 1a1 (ALDH1) in normal ovary and serous ovarian tumors

Tissue was obtained from the Department of Pathology at Rush University Medical Center, Chicago, IL. All procedures followed an Institutional Review Board (IRB) approved protocol. Ovarian tissue was obtained from women with normal ovaries at hysterectomy (mean age 47.4 ± 3.4 years; n = 11), patients with benign serous ovarian tumors (mean age 56.1 ± 13.6 years; n = 9) and primary ovarian cancer patients with malignant serous ovarian tumors (mean age 58 ± 11.1 years; n = 8). The tumor histology and tumor grade were determined by diagnostic evaluation by a pathologist. Malignant serous tumors comprised Grade 3 (n = 6) and Grade 1 (n = 2) with Stage II (n = 3) and Stage III (n = 5) pathology. The criterion for inclusion in the study was women ≥ 40 years old (range 43-76 years; mean age 54.2 ± 11.6 years) and for the patients with benign or malignant ovarian tumors the inclusion criteria included primary serous ovarian tumors. The criteria for exclusion were previous history of any cancer and prior chemotherapy or radiation treatment.

Total RNA was isolated using TRIZOL reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's recommendation. RNA was measured at an optical density (OD) of 260 nm and the purity was evaluated using an OD 260/280 nm absorbance ratio ≥1.7. Before the first strand synthesis, 1 μg of total RNA was treated with DNase to remove trace genomic DNA. cDNA was synthesized using 500 ng of DNase treated RNA with a High-Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA) according to manufacturer's recommendation. Primer pairs were designed using Oligoperfect Designer software (Invitrogen) for ALDH1A1 [GenBank: NM_000689; in-between exon 6 and exon 7]. The Primer sequences were: ALDH1A1 Forward (5'- TTGGAATTTCCCGTTGGTTA-3') and Reverse (5'- CTGTAGGCCCATAACCAGGA-3'); Actin Forward (5'-CTGTGGCATCCACGAAACTA-3') and Reverse (5'- ACATCTGCTGGAAGGTGGAC -3'). The PCR amplifications were carried out in a 25 μl reaction volume containing 25 ng of cDNA using Platinum Taq DNA Polymerase (Invitrogen) according to manufacturer's recommendation. The mixture was denatured at 94°C (3 minutes) followed by 35 cycles at 94°C (30 seconds) and 54°C (30 seconds) to anneal and 72°C (1 minute) for extension followed by a final extension at 72°C (10 minutes) in a programmable Peltier Thermo Cycler (PTC-200, MJ Research Inc. Ramsey, MN). The PCR products were separated by electrophoresis in a 3% (W/V) agarose gel (Invitrogen) and visualized using ethidium bromide stain (Fischer Scientific, Pittsburg, PA). Amplicon from one positive sample each from normal ovary and ovarian serous carcinoma was purified using a QIAquick PCR purification kit (QIAGEN, Valencia, CA) and sequenced at DNA sequencing facility (University of Illinois at Chicago) using an ABI 3100 Genetic analyzer (Applied Biosystems). The amplicon sequences were blasted against the NCBI RefSeq human mRNA database and confirmed with a perfect match for ALDH1A1 gene [GenBank: NM_000689.3]. Quantitative Reverse Transcriptase-PCR (qRT-PCR) was carried out using SYBR green master mix in an ABI 7500 RT-PCR system and analyzed using the ΔCt method with human Actin as an internal control according to the manufacturer's recommendation (Applied Biosystems). The ΔΔCt was determined by subtracting ΔCt of each sample from the average ΔCt of normal ovary. The differences in ALDH1 mRNA expression levels were calculated as the fold change using the formula 2^-ΔΔCt as previously described [38].

Tissues were fixed in formaldehyde, embedded in paraffin and sectioned (6 μm thick). Sections were mounted on microscope slides (Fischer Scientific, Pittsburg, PA), dried (16 hours; 37°C), deparaffinized in xylene, rehydrated in graded alcohols and rinsed with tap water. Sections were examined for histopathology following routine staining with hematoxylin and eosin (H&E; Sigma-Aldrich, St. Louis, MO). ALDH1, CD44, CD117 and CD133 expression was visualized using mouse anti-human ALDH1 mAb (clone 44, BD Transduction Lab San Jose, CA), mouse anti-human CD44 mAB (clone IM7; BioLegend, San Diego, CA), rabbit anti-human CD117 polyclonal antibody (C-19; c-Kit; Santa Cruz Biotechnology, Santa Cruz, CA) and mouse anti-human CD133 mAb (clone EMK08; eBioscience, San Diego, CA) respectively. Staining was carried out according to the manufacturer's protocol (Vector Laboratories, Burlingame, CA). In brief, antigens were unmasked by treating with antigen Unmasking solution (Vector Laboratories) and boiling in a microwave. Endogenous peroxidase was inactivated using substrate (0.3% H₂O₂ in methanol; 20 minutes; 22°C). Sections were washed with phosphate buffer and non-specific binding sites were blocked with normal horse serum (30 minutes). The sections were then incubated with mouse anti-human ALDH1 antibody (1:200) diluted in phosphate buffer containing 1% Bovine Serum Albumin (BSA; Sigma-Aldrich, St. Louis, MO) in a humid chamber (2 hours, 22°C). The bound anti-human ALDH1 antibody was detected using ABC Universal kit and the antigen-antibody reaction was visualized with 3, 3-diaminobenzidine peroxide substrate (DAB; brown color). As a control for secondary antibody binding directly to sections, the ALDH1 antibody was omitted. Sections were briefly rinsed in water, counterstained with hematoxylin (Fischer Scientific) and rinsed in running water (15 minutes). Double label immunostaining was carried out according to the manufacturer's multiple labeling protocol (Vector Laboratories). In brief, the ALDH1 stained sections were further treated with normal horse serum (30 minutes) to block non-specific binding sites. Sections were then incubated with anti-human CD44 or CD177 or CD133 antibody (1:100, diluted in 1% BSA in phosphate buffer) and processed as described for anti-ALDH1 alone, except that the color was developed with DAB and Nickel peroxide substrate (gray/black color). Finally, the sections were dehydrated in graded alcohols and xylene, and covered using Permount (Fischer Scientific). Sections were examined by light microscopy (Olympus BX-41, Center Valley, PA) and images captured and evaluated with MicroSuite Five software (Olympus).

ALDH1 protein expression and localization was assessed using a unbiased cell counting stereology method with a microscope (Olympus BX60, Center Valley, PA) interfaced with a digital camera (CX9000; MBF Bioscience Williston, VT), motorized stage and image analysis software (StereoInvestigator 8.1, MBF Bioscience, Williston, VT). Cell estimation was performed using optical fractionator procedure [39]. Three sections/sample (triplicates) were evaluated. Briefly sections were outlined and scanned at low magnification (×12.5). The thickness of each section was measured at higher magnification (×600) in three separate areas, and the average thickness of each section was calculated. Cells were counted under higher magnification (×600) using an oil immersion objective. Cell counts were estimated within a dissector height of 7 μm, using an 800 × 800 μm² grid size and a 60 × 60 μm² counting frame size. The coefficient of error was calculated based on the Gundersen equation [40]. ALDH1 staining was quantified using average number of ALDH1 positive cells divided by the average number of Hematoxylin counterstained cells in each group and expressed as % mean ± standard deviation (SD).

Total protein was extracted from tissue and separated by one-dimensional Western blot using 10% gradient Tris-HCl gels (Bio-Rad, Hercules, CA; 10 μg total protein/lane) using standard procedures as described previously [1]. Proteins were transferred to a nitrocellulose membrane (0.45 μm; Bio-Rad). Recombinant ALDH1A1 (rALDH1; 1 μg/lane) produced in collaboration with Dr. Jim Dias (University of Albany, Albany, NY) was used as a positive control. Mouse anti-human ALDH1 (1:2000; clone 44, BD Transduction Lab San Jose, CA) and peroxidase-conjugated donkey anti-mouse IgG (1:5000; Jackson ImmunoResearch Laboratories, West Grove, PA) antibody was used to detect ALDH1. Human β-actin was used as a loading control and was detected with mouse anti-actin (1:2000; Sigma, St. Louis, MO). Antibodies were diluted in Blocker solution (Sigma) containing 0.05% Tween 20 (Bio-Rad). The membranes were washed after each step using Tris-buffered saline (10 mM Tris and 0.15 M NaCl, pH7.5) containing 0.05% Tween 20. The protein bands were detected using SuperSignal West Dura substrate (Thermo Scientific, Rockford, IL). MagicMark XP Western standards (Invitrogen, Carlsbad, CA) were used to estimate molecular weight. Digital images were obtained with a Chemidoc XRS Imaging System (BioRad) and analyzed by Quantity One software (Bio-Rad) according to manufacturer's recommendation. The relative density of each ALDH1 band was expressed as a ratio of the density of ALDH1 band and the corresponding β-actin band.

The tissue was dissociated mechanically and enzymatically using a solid human tissue dissociation protocol (Stemcell Technologies, Vancouver, BC) with minor modifications. In brief, tissue was minced, washed in cold Dulbecco's Phosphate Buffered Saline (DPBS; Invitrogen, Carlsbad, CA) and suspended in Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F12; Invitrogen) supplemented with 5% Fetal Bovine Serum (FBS; Invitrogen), collagenase type I (Worthington, Lakewood, NJ) and DNase 1 (Stemcell Technologies) followed by incubation with gentle agitation (2 hours; 37°C). The cell pellet and tissue fragments were separated by centrifugation (5 minutes; 100 × g) followed by a wash with DPBS. A single-cell suspension was obtained after filtering through 40 μm sterile nylon mesh (BD Falcon, San Jose, CA). The flow through was collected in a fresh tube, centrifuged (5 minutes; 100 × g), washed and suspended in DPBS. To remove and lyse red blood cells the cells were treated with ammonium chloride solution (BioLegend, San Diego, CA; 10 minutes; 4°C). Cells were then suspended in DPBS with 2% BSA and the cell count was determined using a Coulter Counter (Beckman, Brea, CA).

Aldehyde dehydrogenase enzyme activity in viable cells was determined using a fluorogenic dye based ALDEFLOUR assay (Stemcell Technologies) according to the manufacturer's instructions. In brief, cells were suspended (0.5 × 10⁶ cells/mL) in ALDEFLUOR assay buffer containing ALDH substrate (Bodipy-Aminoacetaldehyde) and incubated (45 minutes; 37°C). As a reference control, the cells were suspended in buffer containing ALDEFLUOR substrate in the presence of diethylaminobenzaldehyde (DEAB), a specific ALDH1 enzyme inhibitor. Propidium iodide (2 μg/mL; Sigma, St. Louis, MO) was used to exclude dead cells. The cells were analyzed using a FACSCalibur flow cytometer (BD Biosciences, Rockville, MD) and the data was analyzed using FlowJo 7.6.1 software (Tree Star, Ashland, OR).

The outcome variables were expressed as mean ± SD. SPSS (Student version 7.5, SPSS Inc., Chicago, IL) was used for statistics. The independent samples t-test was used to test the statistical difference between groups. Correlation was analyzed by calculating a Pearson correlation coefficient (r). P values < 0.05 were considered statistically significant.

ALDH1 mRNA expression was significantly lower in malignant ovarian tumors (n = 5) compared to normal ovary (p < 0.001; n = 5) and benign ovarian tumors (p = 0.008; n = 5) (Figure 1). There was no significant difference in ALDH1 mRNA expression between normal ovary and benign ovarian tumors (p = 0.18). The target amplified gene was confirmed as ALDH1A1 [GenBank: NM_000689.3] (data not shown).

The proportion of ALDH1 immunostained cells was significantly lower in malignant ovarian tumors (17.1 ± 7.6%; n = 5) compared to normal ovaries (37.4 ± 5.4%; p = 0.001; n = 5) and benign ovarian tumors (31.0 ± 6.7%; p = 0.015; n = 5) (Figure 2A). There was no significant difference between normal ovary and benign ovarian tumors (p = 0.11), thus confirming the mRNA data. The ALDH1 mRNA expression levels and the proportion of immunostained cells was positively correlated (r = 0.7; p < 0.01).

ALDH1 protein was detected as a single band at 55 kDa in all of the ovarian tissues tested by Western blot (Figure 2B). Densitometry analysis of the blots showed lower levels of ALDH1 in malignant tumors compared to normal ovary and benign tumors. Furthermore, a higher ALDH1 band intensity was detected in a well differentiated malignant tumor (lane 15; Figure 2B) compared to poorly differentiated tumors (lane 11 -14; Figure 2B). A strong positive correlation was observed between the levels of ALDH1 protein expression in Western blot and proportion of ALDH1 immunostained cells (r = 0.8; p < 0.01) among the tested samples.

ALDH1 immunostaining was observed in various cell types in normal ovary and serous ovarian tumors. In normal ovary, a diffuse ALDH1 staining pattern was observed in the stroma in fibroblasts-like cells and fibrous tissue. In addition, the surface epithelial cells stained intensely although there were occasional cells without stain (Figure 3A and 3B). The smooth muscle cells surrounding the blood vessels and the granulosa cell layer surrounding developing follicles did not stain for ALDH1; however, the stromal cells in the perivascular regions and in the developing theca layer of follicles showed ALDH1 staining (Figure 3C and 3D). The fibrous tissue between cords of luteal cells in the regressing corpus luteum (corpus albicans) also stained for ALDH1 but not the cells of corpus luteum (Figure 3E and 3F).

The staining pattern of ALDH1 in uninvolved areas adjacent to benign serous ovarian tumors was similar to that of normal ovary (Figure 4A). In contrast to normal ovary, strong ALDH1 expression was observed near some neo-angiogenic blood vessels in benign ovarian tumors (Figure 4C). In addition, staining of the surface epithelium was patchy compared to normal ovary (Figure 3A) and contained areas of intense staining adjacent to areas of no staining (Figure 4D - 4F).

In malignant serous ovarian tumors ALDH1 staining varied (strong to weak or no staining; Figure 5) and was seen primarily in the fibroblast like cells in the stroma and a few well differentiated tumor epithelial cells (Figure 5B). Well differentiated malignant tumor cells (Figure 5A - 5B) showed higher ALDH1 expression compared to poorly differentiated tumor cells (Figure 5C - 5F). Interestingly, ALDH1 staining differed in the same malignant tumor tissue based on cellular differentiation (Figure 6). Poorly differentiated regions of solid tumor cell nests (Figure 6B and 6C) had little or no ALDH1 expression compared to the adjacent, highly stained differentiated regions with micro-papillary tumor architecture (Figure 6A and 6B).

To further investigate ALDH1 expression in high grade malignant serous ovarian tumors, sections were co-stained with CD44, CD117 and CD133 to determine if there was an association with CSC markers (Figure 7). ALDH1 and CSC markers were expressed in different cell populations. CD44 was expressed in lymphocytes in and near blood vessels as expected. CD117 (Figure 7C & 7D) and CD133 (Figure 7E & 7F) expression was localized in tumor epithelial cells while ALDH1 immunostaining occurred in the tumor stroma. We evaluated sections from 3 poorly differentiated malignant ovarian tumors and found occasional cells (< 1%) that were co-stained with ALDH1 and CD44. However, it is not clear whether they were tumor cells or infiltrating lymphocytes. We did not find any visible co-staining with ALDH1 and CD117 or CD133 markers.

The mean fluorescence intensity (MFI) was significantly decreased in malignant ovarian tumors (15 ± 8.8 MFI; n = 3) compared to normal ovary (92.3 ± 24 MFI; p = 0.02; n = 3) and benign ovarian tumors (74 ± 18.7 MFI; p = 0.018; n = 3) (Figure 8). While no significant difference in MFI was observed between normal ovary and benign tumors (p = 0.3). In addition, the proportion of ALDH^Bright cells was lower in malignant ovarian tumors (6.4 ± 2.9%) compared to normal ovary (22.8 ± 6.4%) and benign tumors (16.3 ± 5.6%). The ALDEFLUOR assay was positively correlated with the proportion of cells expressing ALDH1 by semi-quantitative immunohistochemistry (r = 0.77; p < 0.01). Overall, the estimation of enzyme activity in ovarian cells was consistent with ALDH1 mRNA and protein expression levels.