Increased cell proliferation and Rb protein phosphorylation in p18^−/− mammary epithelium and tumors
(A) Immunostaining of Ki67 (brown) of 18-month old virgin p18^+/− and p18^−/− mammary glands.
(B) Immunostaining of BrdU (green), CK8 (red), and DAPI (blue) in mammary glands of 9-month-old virgin littermate mice. The boxed areas were enlarged in the insets to show BrdU incorporation (pointed by white arrows) in an individual gland. The percentages of BrdU positive cells were calculated from cells situated in clear duct/gland structure. Results represent the mean ± SD of 3 animals per group.
(C) Immunostaining of BrdU (green) in mammary glands of 8-week-old virgin littermate mice was performed to determine cell proliferation and nuclei were stained with DAPI (blue). The percentages of BrdU positive cells were calculated from cells situated in clear duct/gland structure. Results represent the mean ± SD of 3 animals per group.
(D) Sections of normal mammary glands and mammary tumors of WT and p18^−/− mice at around one year of age were examined for pRb protein phosphorylation at Ser608 by CDK4 and CDK6. Counter staining is blue and positive staining is brown. The percentages of pRb-S608 positive cells were calculated from cells situated in clear duct/gland structure in normal glands and from all cells in tumors. Results represent the mean ± SD of 3 animals per group.

p18 loss expands luminal progenitors
(A, B, C, D). Mammary cells from WT (or p18^+/−) and p18^−/− virgin littermate mice of Balb/c backgrounds at indicated ages were isolated, sorted, and analyzed by flow cytometry for CD24 together with CD29 (A); CD49f (B); SCA1 (C), or Ki67 (D). The bar graphs in (A) represent the mean ± SD of 3 animals per group.

p18 and GATA3 level is inversely related to luminal A type tumors and conversely predicts human breast cancer patient outcomes
(A) ANOVA and box plot analysis of gene expression levels in 295 breast cancer patients from the NKI patient series (Chang et al., 2005; van de Vijver et al., 2002) according to tumor subtype.
(B) Scatter plot analysis of INK4C and GATA3 expression for 295 patients from the NKI patient series. Correlation coefficient (r) = −0.379, p value<0.0001.
(C) 295 patients from the NKI patient series were divided into high and low groups based upon each gene’s rank order expression values. Probabilities for overall survival are shown for each gene and the p values calculated.

p18 null mice develop spontaneous luminal mammary tumors
(A) Mammary tumor-free survival of Balb/c mice of different genotypes.
(B) Spontaneous tumor development in Balb/c-p18 mutant mice.
(C) Representative H&E staining of primary mammary tumor (T) and lung metastasis (M) developed in p18^−/− mice. A WT mammary control at a similar age is shown.
(D) Immunofluorescent staining of CK5 and CK8 in age-matched (14–16 months of age) mammary tissue from WT mice or tumors from mutant mice. Note that the patched green signals inside the WT glands were autofluorescence caused by the secretion of milk.
(E) Immunohistochemical staining of ERα (brown) in mammary tumors from p18^−/− mice. Note the highly expressed ERα levels in normal mammary glands (arrows) surrounding tumors. Boxed areas are magnified in the insets.

Loss of p18 increases label-retaining luminal epithelial cells and cells with morphologic features of luminal progenitors
(A) and (B) 1-month-old littermate mice were injected intraperitoneally with BrdU twice a day for seven consecutive days, followed by chase for one (A) or 37 days (B) before sacrifice. Mammaries were dissected and stained with antibodies to BrdU, CK5, and CK8. The percentage of BrdU positive cells were calculated as described in the Experimental Procedures. Results represent the mean ± SD of 3 animals per group.
(C) After immunofluorescence staining and photographing of sections from the label retaining assay (37 days chase, top panels), the identical section was re-stained with H&E. Matched cells between two stainings were identified microscopically and label-retaining small light cells (SLCs) and undifferentiated large light cells (ULLCs) were counted (indicated by red and yellow arrows, respectively, bottom panels). Lower magnification pictures for these paired stainings were also shown in (B). Three additional types of histological distinct cell types—differentiated large light cells (DLLCs), large dark cells (LDCs), and myoepithelial cells (MYOs)—were also identified and quantified (see Table S1).
(D) Littermate WT or p18^−/− mice at 2–3 months of age were dissected and stained with H&E. SLCs and ULLCs are indicated by red and yellow arrows, respectively. 400–600 ductal and lobular cells from each genotype were counted, and the number and percentage of SLCs and ULLCs are presented. Three other histological types (DLLCs, LDCs, and MYOs) were also identified and quantified (Figure S1).

p18 deficiency increases luminal progenitor cell proliferation in vitro
(A) Mammary cells freshly isolated from both WT and p18^−/− mice were cultured in 3D Matrigel for one week, pulse labeled with BrdU, and acini were stained for antibodies to CK8 (red) and to BrdU (green). Two sections from two separate acini are shown.
(B) Mammary cells from WT and p18^−/− virgin littermate mice at 2–3 months of age were isolated and analyzed by mammosphere assay. Representative mammospheres in common size (left panel) and in large size (right panel) are shown. The sphere size was measured as described in the Supplemental Experimental Procedures, and the number of spheres larger than 60 μm was quantified. The assay was performed in triplicate for each animal. The bar graphs represent the mean ± SD of 4 animals per group.
(C) Freshly isolated mammary cells were plated in Matrigel-coated 24-well plate (1,000 cells per well). Nine days after culture, colonies were photographed and counted. Representative spherical and flat colonies are shown. The assay was performed in triplicate for each animal. The bar graphs represent the mean ± SD of 3 animals per group.
(D) Freshly isolated mammary cells were plated in Matrigel-coated 24-well plate (20,000 cells per well). Nine days after culture, colonies were stained with CK8 and CK5 on Matrigel and counted as described in the Supplemental Experimental Procedures. Representative luminal (red arrows) and myoepithelial colonies (green arrows) are indicated. The assay was performed in triplicate for each animal. The bar graphs represent the mean ± SD of 3 animals per group.

p18 is a direct target of GATA3 in mammary epithelium.
(A) Mammary cells from WT mice at seven weeks of age were isolated and sorted by flow cytometry for CD24 and CD29 cell populations. Total RNA from the indicated cell populations (R4, CD24⁻CD29⁻; R3, CD24⁺CD29⁻; R2, CD24⁺CD29⁺) was extracted and analyzed for the expression of Gata3 and p18 by Q-RT-PCR. Data are expressed relative to the corresponding values for R4 populations as mean ± SD from triplicates of each of the two independent mice.
(B) Mammary tissues from Gata3^f/f;WAP-cre⁺ and Gata3^f/f;WAP-cre⁻ dams that successfully produced and nursed pups were analyzed for Gata3 and p18 expression. Mammary tissues from 2-month old virgin WT and p18 null females were used as control. n.s., non-specific bands.
(C, D) Human MCF-7 or 293T cells were transfected with either scrambled siRNA (Ctrl) or siRNA targeting GATA3, and the cells were lysed for Western blot, (C) or the RNA were extracted for Q-RT-PCR (D) 48 hours after transfection. Similar results were derived from two different siRNA targeting different sequences of GATA3. Data are expressed relative to the corresponding values for control (Ctrl) cells as mean ± SD from triplicates of a representative experiment.
(E) MCF-10A cells were infected with pBabe-puro or pBabe-puro-GATA3, selected with puromycin for two to five days, and analyzed by Q-RT-PCR. Data are expressed relative to the corresponding values for vector-infected cells as mean ± SD from triplicates of a representative experiment.
(F) MCF-10A cells infected with pBabe-puro-empty or pBabe-puro-GATA3 were pulse-labeled with BrdU. Cells were stained with anti-BrdU antibody and PI and subjected to flow cytometry. BrdU positive cells are marked by the box and the percentage of BrdU positive cells are shown.
(G) GATA3 binds to p18 locus. The upper panel illustrates the human p18 locus which expresses multiple transcripts, including a longer 2.4-kb transcript from an upstream promoter and a shorter 1.2-kb transcript from a downstream promoter. A 40 bp region containing 10 GATA sequences in tandem is shown. The dotted line represents the 8.9 kb region screened for GATA3 binding by ChIP assay using 33 pairs of primers. Locations and results of representative amplicons are shown, MCF-7 cells were used for this assay.
(H) Working model for the function and mechanism of GATA3 and INK4c in regulating mammary stem, progenitor, and luminal cell differentiation.