Supplemental Data. ALDH1 Is a Marker of Normal and Malignant. Human Mammary Stem Cells. and a Predictor of Poor Clinical Outcome

Transcription

1 Cell Stem Cell, Volume 1 Supplemental Data ALDH1 Is a Marker of Normal and Malignant Human Mammary Stem Cells and a Predictor of Poor Clinical Outcome Christophe Ginestier, Min Hee Hur, Emmanuelle Charafe-Jauffret, Florence Monville, Julie Dutcher, Marty Brown, Jocelyne Jacquemier, Patrice Viens, Celina Kleer, Suling Liu, Anne Schott, Dan Hayes, Daniel Birnbaum, Max S. Wicha, and Gabriela Dontu

2

3 Figure S1. Overlap between the normal mammary epithelial cell population with ALDH activity detected by the ALDEFLUOR assay and the cell population expressing ALDH1 protein or expressing cytokeratin 18 protein, as detected by immunostaining. ALDEFLUOR-negative and ALDEFLUOR-positive cells from normal breast epithelium were separated by FACS using the ALDEFLUOR assay. Cells were then fixed in RNA later, immunostained with ALDH1 antibody and analyzed by flow cytometry (A-C), or cytospun on poly-lysined slides, fixed in methanol-acetone (1:1) and immunostained with ALDH1 or CK18 antibodies (D-G). ALDEFLUOR-negative cells did not have ALDH1 protein at levels detectable by immunostaining (M1=.88%) (A), whereas ALDEFLUOR-positive cells contained the entire cell population detected by the ALDH1 antibody (B). Overlay showing a direct comparison between ALDH1 immunostaining of ALDEFLUOR-negative and ALDEFLUOR-positive cells (C). Immunostaining with ALDH1 antibody on cytospun sorted cells confirmed the FACS analysis showing that ALDEFLUOR-negative cells are negative for ALDH1 staining whereas ALDEFLUOR-positive cells comprised 15% of ALDH1-positive cells (D-E). Immunostaning with CK18 antibody on cytospun sorted cells confirmed the results of the double immunofluorescent staining in situ, showing that only ALDEFLUOR-negative cells contained CK18-positive cells (F-G)

4

5 Figure S2. ALDH1 staining and duct morphology in a sequence of four consecutive sections of normal breast epithelium. A-D. Four consecutive sections of normal breast epithelium were stained with H&E (A) and immunostained for ALDH1 (red cytoplasmic staining) (B-C). The sequence of serial sections shows the formation of three side branches from the initial duct. The ALDH1-positive cells are located at the separation points, where side branches emerge. D. A schematic representation of the branching structure showing the position of the four consecutive sections (blue arrows).

6 UM1 UM2 UM3 Supplementary Figure 3 With DEAB Without DEAB With DEAB Without DEAB With DEAB Without DEAB 1K R1 R2 1K R1 R2 1K R1 R2 1K R1 1K R2 R1 R2 1K R1 R Side Scatter 6 4 Side Scatter 6 4 Side Scatter 6 4 Side Scatter 6 4 Side Scatter 6 4 Side Scatter BAAA BAAA BAAA BAAA BAAA BAAA Region Events %Gated R R Region Events %Gated R R Region Events %Gated R R2 9.9 Region Events %Gated R R Region Events %Gated R R Region Events %Gated R R Tum or size ( cm ) Days after injection 5, cells ALDEFLUOR + 5, cells ALDEFLUOR + 5 cells ALDEFLUOR + 5, cells ALDEFLUOR - 5, cells ALDEFLUOR - 5 cells ALDEFLUOR - 5, unsorted cells 5 unsorted cells Tumor size ( cm ) Days after injection 5, cells ALDEFLUOR + 5, cells ALDEFLUOR + 5 cells ALDEFLUOR + 5, cells ALDEFLUOR - 5, cells ALDEFLUOR - 5 cells ALDEFLUOR - 5, unsorted cells 5 unsorted cells Tumor size (cm) Days after injection 5, cells ALDEFLUOR + 5, cells ALDEFLUOR + 5 cells ALDEFLUOR + 5, cells ALDEFLUOR - 5, cells ALDEFLUOR - 5 cells ALDEFLUOR - 5, unsorted cells 5 unsorted cells

7 Figure S3. The ALDEFLUOR positive cell population has properties of cancer stem cells. Representative flow cytometry analysis of ALDH activity in cells derived from human breast tumors, orthotopically xenotransplanted in NOD/SCID mice (UM1, left panel; UM2, central panel; UM3 right panel). Cells incubated with ALDEFLUOR substrate (BAAA) and the ALDH specific inhibitor, DEAB, were used to establish the baseline fluorescence of these cells (R1) and to define the ALDEFLUOR-positive region (R2). Incubation of cells with ALDEFLUOR substrate in the absence of DEAB induced a shift in BAAA fluorescence defining the ALDEFLUOR-positive population. All the ALDEFLUOR analyses of human breast tumor cells were first gated on PI negative cells (viable cells) that represented 73.6 ± 1.8% of the total population. Tumor progression curves were plotted for the numbers of cells injected in NOD/SCID mice (5, cells; 5, cells; 5 cells) and for each cell population (ALDEFLUOR-positive, ALDEFLUOR-negative, Unseparated).

8

9 Figure S4. Overlap between the cell populations identified by the ALDEFLUOR assay and the CD44+/CD24 - /lin - phenotype in human breast tumors. ALDEFLUOR-negative and -positive cells from three human breast tumor xenotransplants (MC1, UM1, UM2) were separated by FACS using the ALDEFLUOR assay. In all experiments cells were first gated on PI negative cells (viable cells) that represented 73.6 ± 1.8% of the total population. Cells were then fixed in RNA later, immunostained with a CD24-PE antibody, a CD44 APC antibody, and antibodies for lineage markers, labeled Pe-Cy5. In all the flow cytometry analyses cells were first gated on lin - markers, that represented 12.3 ± 1.1% of the total population. These cell populations were gated out in the flow charts shown on the left side of the figure. The diagrams in the right side show the representation of the cell fractions defined by the ALDEFLUOR and the CD44 + /CD24-/lin - combined phenotype in the total tumor cell population (PI negative).

10 A B D Gated on dead cells stained with 7AAD PI Alde - Alde + Alde - Alde + 14% 11% 16% 1% C Supplementary Figure 5

11 Figure S5. Example of FACS analysis procedure used for the ALDEFLUOR assay in the normal and malignant human breast epithelium. A. All samples (normal breast reductions and tumors) were first gated according to the side and the forward scatter to select epithelial cells and to eliminate contaminating non-epithelial cells, clusters, and debris (gate R). The cell fraction gated on R represented 65.4 ±4.2% of the total cell population. B. Subsequently, all tumor samples were gated according to PI staining and H2Kd staining. Only the cells negative for PI staining (viable cells) and negative for H2Kd staining (human cells) were selected for the ALDEFLUOR analysis (gate R1). The PI-negative/H2Kd-negative population represented 73.6 ±1.8% of the cell population gated on R. For the normal epithelium samples, only PI staining was performed for viability, since there were no contaminating cells of mouse origin. The PI-negative population represented 93.4 ±2.4% of the cell population gated on R. C. All samples (normal epithelium and tumors) were analyzed using the ALDEFLUOR assay on the cell population gated on R and R1. D. The percentage of non-viable cells detected using PI staining was similar to that detected using 7AAD. Two aliquots from normal mammary epithelial cells treated with BAAA according to the ALDEFLUOR assay were stained in the last step with 7AAD and PI respectively, to identify potential differences in the ability of the two stains to detect non-viable cells. The two viability assays appeared to identify the same cells.

12 Compensation and gate setup Analysis Control unstained ALDEFLUOR-negative Control APC only ALDEFLUOR-positive CD44-APC CD24-PE Control PE only CD44-APC CD24-PE Supplementary Figure 6

13 Figure S6. Analysis of the cell populations defined by the ALDEFLUOR and CD44CD24lin- phenotypes in the UM2 tumor. The ALDEFLUOR-positive population in UM2 showed a considerably higher fluorescence compared to the other tumors (Figure S3). This required compensation and different gate setup for the analysis of the markers CD44CD24lin, as shown in the flow chart on the left side of the figure.

14 Table S1. Tumorigenicity in the humanized fat pad of NOD/scid mice. MC1 UM1 UM2 UM3 Tumors/injections Cells injected/fat pad 5x x1 4 5x1 3 5 ALDEFLUOR-negative 2/4 --- /2 /4 ALDEFLUOR-positive 4/ /1 4/4 Unseparated 4/ /1 3/3 ALDEFLUOR-negative /4 --- /1 /3 ALDEFLUOR-positive 3/ /1 3/3 Unseparated 3/ /1 2/3 ALDEFLUOR-negative 1/3 /1 /1 /3 ALDEFLUOR-positive 3/3 1/1 1/1 3/3 Unseparated 3/3 1/1 1/1 3/3 ALDEFLUOR-negative 1/ /4 ALDEFLUOR-positive 3/ /1 4/4 Unseparated 3/ /1 3/3

15 Characteristics Table S2. Correlation between ALDH1 protein expression and histoclinical characteristics U.M. set I.P.C. set ALDH1 ALDH1 ALDH1 ALDH1 Negative Positive Negative Positive No. of (%) No. of (%) p-value No. of (%) No. of (%) p-value patients patients patients patients All cases 122 (81) 24 (19) 243 (7) 12 (3) Age (years) 5 33 (27) 6 (25) NS 79 (35) 25 (25) NS >5 89 (73) 18 (75) 164 (65) 77 (75) Pathological tumor size PT1 61 (6) 1 (45) NS 13 (42) 37 (37) NS PT2 33 (33) 7 (32) 18 (45) 46 (47) PT3 7 (7) 5 (23) 3 (13) 16 (16) SBR grade I 25 (22) 2 (9) <.5 8 (33) 24 (24) <.1 II 55 (49) 7 (3) 12 (49) 38 (38) III 33 (29) 14 (61) 43 (18) 39 (38) Lymph node metastasis Negative 53 (56) 9 (41) NS 134 (56) 43 (44) NS Positive 42 (44) 13 (59) 17 (44) 55 (56) Estrogen receptor Negative 38 (34) 14 (61) <.5 41 (17) 39 (39) <.1 Positive 75 (66) 1 (39) 198 (83) 61 (61) Progesterone receptor Negative 5 (44) 15 (62) <.5 61 (26) 51 (52) <.1 Positive 64 (56) 9 (38) 17 (74) 47 (48) Ki-67 Negative (<2) (91) 77 (82) <.5 Positive ( 2) (9) 17 (18) ERBB2 Negative (/1+) 94 (83) 15 (62) <.5 28 (94) 76 (79) <.1 Positive (2+/3+) 19 (17) 9 (38) 14 (6) 2 (21) Cytokeratin 18 (CK18) Negative (5) 9 (9) NS Positive (95) 85 (91) Cytokeratin 5/6 (CK5/6) Negative (72) 46 (58) <.5 Positive (28) 33 (42) Cytokeratin 14 (CK14) Negative (96) 68 (86) <.1 Positive (4) 11 (14)

16 Supplemental Experimental Procedures Normal breast tissue dissociation Normal breast tissue from reduction mammoplasties was dissociated mechanically and enzymatically, as previously described (Stingl et al., 1998). Tissue was minced and dissociated in Ham s F12/Dulbecco s modified Eagle s medium (F12:DMEM; 1:1; StemCell technologies, Durham, NC, USA) supplemented with 1 mm Hepes, 2% bovine serum albumin (BSA; Fraction V; GIBCO TM INVITROGEN), 5 mg/ml insulin,.5 mg/ml hydrocortisone, 1 ng/ml cholera toxin, 3 U/ml collagenase and 1 U/ml hyaluronidase (all from Sigma, St Louis, MO, USA) at 37 C for 16h. The epithelialcell-rich pellet (95-99% purity) was collected by centrifuging the cell suspension at 8 g for 4 min followed by one wash with F12/DMEM. The supernatant from the first centrifugation contained the mammary stromal cells (fibroblasts and endothelial cells). Epithelial organoids were further digested for 5 min in.5% trypsin (Gibco)-.25% EDTA (Sigma) solution to generate a single-cell suspension. An equal volume of F12/DME/H supplemented with 5% FBS was added to stop the digestion. The cell suspension was filtered twice through a 4-mm nylon mesh (BioDesign Inc., New York, N.Y., USA). Following centrifugation at 8 g, the pellet was resuspended in F12/DMEM with a reduced calcium concentration (.6 mm, StemCell technologies) supplemented with 5 U/ml dispase (Collaborative Biomedical Products, Bedford, Md., USA). To remove red blood cells the pellets were treated with ammonium chloride solution. Flow cytometry analysis CD44/CD24/Lin staining was performed as previously described (Al-Hajj., 23). Briefly, cells were stained with primary antibodies anti-cd44 labeled APC (dilution 1:1, BD Biosciences), anti-cd24 labeled PE (dilution 1:1, BD Biosciences), and lineage markers anti-cd31, CD64, CD14b (BD Biosciences), CD2, CD3, CD1, CD16, CD18 (all labeled with PE-

17 Cy5, Jackson Labs). Fresh cells were stained with 1µg/ml PI (Sigma) for 5 min. for viability. Examples of flow charts and more details on gate setup are shown in Figures S5 and S6. Mammosphere culture Mammosphere culture was performed as previously described (Dontu et al., 23). Single cells were plated in ultra-low attachment plates (Corning, Acton, MA, USA) or plates coated with 1% agarose in PBS, at a density of 2, viable cells/ml in primary culture and 5 cells/ml in subsequent passages. For mammosphere culture, cells were grown in a serum-free mammary epithelial basal medium (MEBM) (Cambrex Bio Science Walkersville, Inc, Walkerville, MD, USA) supplemented with B27 (INVITROGEN, Carlsbad, CA, USA), 2 ng/ml EGF (BD Biosciences, San Jose, CA, USA), antibiotic-antimycotic (1 unit/ml penicillin G sodium, 1 ug/ml streptomycin sulfate and.25 µg/ml amphotericin B), 2 ug/ml Gentamycin, 1 ng/ml Hydrocortisone, 5 µg/ml Insulin and 1 µm beta-mercaptoethanol (GIBCO TM INVITROGEN) in a humidified incubator (1% CO 2 : 95% air, 37 C for 7-1 days, as previously described (Dontu et al., 23).