CRIPTO/GRP78 signaling maintains fetal and adult mammary stem cells ex vivo

Abstract

Little is known about the extracellular signaling factors that govern mammary stem cell behavior. Here, we identify CRIPTO and its cell-surface receptor GRP78 as regulators of stem cell behavior in isolated fetal and adult mammary epithelial cells. We develop a CRIPTO antagonist that promotes differentiation and reduces self-renewal of mammary stem cell-enriched populations cultured ex vivo. By contrast, CRIPTO treatment maintains the stem cell phenotype in these cultures and yields colonies with enhanced mammary gland reconstitution capacity. Surface expression of GRP78 marks CRIPTO-responsive, stem cell-enriched fetal and adult mammary epithelial cells, and deletion of GRP78 from adult mammary epithelial cells blocks their mammary gland reconstitution potential. Together, these findings identify the CRIPTO/GRP78 pathway as a developmentally conserved regulator of fetal and adult mammary stem cell behavior ex vivo, with implications for the stem-like cells found in many cancers.

Graphical abstract

Figure 1

ALK4^L75A-Fc Inhibits Growth-Factor-like Effects of CRIPTO on Immortalized Human Mammary Epithelial Cells (A) ALK4^L75A-Fc (10 μg/ml) inhibits binding of ¹²⁵I-CRIPTO to native MCF10A cells. Data from one of two independent experiments using three technical replicates are shown, with SDs. (B) Phospho-AKT (Ser 473) or total AKT western blots on lysates from MCF10A cells treated with soluble CRIPTO (300 ng/ml), ALK4^L75A-Fc, and PI3K inhibitor LY 294002 (10 μM) as indicated. (C) Phospho-SMAD2 and SMAD2/3 western blots on MCF10A cells stably transduced with CRIPTO-Flag virus (MCF10A-CR) or control virus (MCF10A-V) and treated as indicated with ACTIVIN A (1 nM), NODAL (30 nM), ALK4^L75A-Fc, and the ALK4/5/7 inhibitor SB 431542 (10 μM). (D) Time course showing that MCF10A-CR cells accumulate more rapidly than MCF10A-V cells in 2D culture. Treatment with ALK4^L75A-Fc selectively inhibits growth of MCF10A-CR cells. ^∗p < 0.01. Triplicate wells in a representative experiment are shown, with SDs. (E) MCF10A-CR cells are more migratory than MCF10A-V cells. The CRIPTO-dependent increase in migrational velocity and persistence is blocked by ALK4^L75A-Fc treatment or a GRP78 neutralizing antibody (N-20). ^∗∗p < 0.01, two-way ANOVA. Data from ≥50 individual cells from three fields of view per condition in one experiment (of two performed) are shown, with SDs. (F) Suspension cultures of MCF10A-V and MCF10A-CR cells 11 days after seeding in 24-well plates. MCF10A-CR cells grow more rapidly than MCF10A-V cells in suspension culture, and this growth advantage is inhibited by ALK4^L75A-Fc treatment. (G) MCF10A-CR cells produce larger and more plentiful colonies than MCF10A-V cells when grown in immobilizing methylcellulose suspension cultures, and ALK4^L75A-Fc treatment blocks this CRIPTO-dependent growth. Bar represents 30 μm. (H) Colony growth was significantly reduced by either ALK4^L75A-Fc treatment or PI3K inhibition (LY 294002 10 μM), but not by treatment with SB 431542. ^∗p < 0.05. Two independent experiments with duplicate wells in each are tabulated, with SDs. SB 431542 treatment increased colony number, but not mean size, of MCF10A-V cells. See also Figures S1 and S2.

Figure 2

Expression of CRIPTO and GRP78 in the Fetal Mammary Rudiment and Responsiveness of fMaSCs to Soluble CRIPTO (A and B) Whole-mount E18.5 fetal mammary rudiment demarcated with immunofluorescence staining for KRT8 (blue) or DAPI (white) and costained for CRIPTO (green) and GRP78 (red) (right). (A, i–iii) control rudiment stained with KRT8 and isotype-matched controls for GRP78 and CRIPTO. Scale bar represents 200 μm (A) or 50 μm (B). (C) CRIPTO is expressed in the fMaSC population and stromal cells as measured by RT-PCR. A total of 150 rudiments were pooled and assayed in technical triplicates, with SDs. (D) CRIPTO staining colocalizes with F4/80 staining (arrows) adjacent to mammary epithelium demarcated by KRT8 staining (blue). (E) Flow cytometric analysis of fetal mouse mammary cells stained with CD24, GRP78, and isotype control antibodies as indicated. The percentage of the CD24^high population also positive for GRP78 is shown. (F) Phospho-AKT staining in serum-starved, fMaSC-derived organoids treated as indicated with soluble CRIPTO (300 ng/ml), ALK4^L75A-Fc (10 μg/ml), LY 294002 (10 μM), or N-20 (2 μg/ml). Scale bar represents 50 μm. (G) 2D fMaSC cultures were treated with CRIPTO (400 ng/ml) and/or ALK4^L75A-Fc (10 μg/ml) as indicated and stained with Ki67 antibody and DAPI. Scale bar represents 50 μm, and inset represents 10 μm.

Figure 3

CRIPTO Promotes Maintenance of fMaSCs (A) Sphere-bisecting confocal cross sections of day 5, fMaSC-derived colonies grown in Matrigel and stained with KRT8 (green) and KRT14 (red). Colonies were grown in differentiation media or maintenance media with soluble CRIPTO (0.33 μg/ml) or ALK4^L75A-Fc (10 μg/ml) as indicated. Yellow arrows indicate KRT8⁺KRT14⁺ double-positive cells. Red arrows indicate lineage-committed KRT14 single-positive cells. Control represents the same isotype. Scale bar represents 100 μm. (B) KRT8 and KRT14 staining was quantified from images and is represented as the percent KRT8⁺KRT14⁺ double-labeled area per organoid. The percentage of organoids containing KRT8⁺KRT14⁺ double-labeled cells is also indicated. Results of three independent representative experiments are shown. (C) Mammary gland repopulation efficiency of first-generation colonies grown in Matrigel-containing maintenance media culture with soluble CRIPTO and/or ALK4^L75A-Fc. Each circle represents one recipient mammary gland. Shaded circles represent ducts extending through >50% of the fat pad. (D) Colony-forming efficiencies (>100 μm) at first, second, and third passage for fMaSCs grown for 7 days per passage in differentiation or maintenance media with CRIPTO and/or ALK4^L75A-Fc. Results of three independent experiments (maintenance media) and a representative experiment (differentiation media) were quantified. Triplicate wells were quantified in each. Error bars represent SDs. ^∗p ≤ 0.01. See also Figure S3.

Figure 4

GRP78 Is Required for MaSC Activity and Marks a CRIPTO-Responsive Stem Cell-Enriched Population of Adult Mouse Mammary Epithelial Cells (A) Immunohistochemical localization of CRIPTO and GRP78 in adult mouse mammary tissue. Inset: secondary antibody alone (2^nd). (B) CRIPTO and GRP78 proteins colocalize in stromal areas including adipose (f, fat) and in epithelial structures (e), where a subset of cells display elevated staining for both markers (white arrows). (C) Repopulation rate of de-epithelialized mammary glands inoculated with 500 GRP78-floxed epithelial cells transduced with Ad-CRE/GFP- or Ad-GFP and sorted for GFP. Each circle represents one recipient mammary gland. Filled circles indicate >50% fat pad filling. (D) Flow cytometric analysis of wild-type adult mouse mammary cells stained for CD24 and GRP78 (lineage excluded) indicating the presence of cell-surface GRP78 in adult mammary epithelia and demarcating the GRP78^high and GRP78^low populations (left). Antibodies to CD49f were also incorporated (right), and cells positive for GRP78 and CD24 were gated to generate a population density graph (red lines) superimposed on the CD24/CD49f staining. The standard positions of the CFC, MYO, and MRU populations (Stingl et al., 2006) are indicated. (E) Colonies grown from single cells sorted into individual wells in 96-well plates. Luminal, organoid, and myoepithelial morphologies were distinguishable (left). The number of each type of clonal outgrowth deriving from single GRP78^low and GRP78^high cells in one representative experiment (of three similar experiments) was quantified according to morphology (right). (F) Confocal micrographs showing the keratin distribution observed in representative colonies grown from cells sorted using CD24 and GRP78. GRP78^low cells predominantly expressed either KRT8 or KRT14, while GRP78^high cultures frequently exhibited KRT8⁺KRT14⁺ outgrowths. Representative organoid (i), luminal (ii), and myoepithelial (iii) colonies are shown. Color panels show individual fluorophores separately for the organoid (i, insets). Scale bars represent 100 μm. (G) Mammary gland repopulation efficiencies of GRP78^high and GRP78^low cells represented as in (C) (the data fit a single-hit model; likelihood ratio test p = 0.938; ELDA). (H) GRP78^high cells are selectively responsive to soluble CRIPTO or ALK4^L75A–Fc treatment when serially passaged in Matrigel culture. Results of two independent experiments conducted in triplicate are shown, with SDs. Black bracket p < 0.05; red bracket p < 0.01. (I) Model depicting cooperativity of soluble CRIPTO (sCR) and cell-surface GRP78 in promoting mammary stem cell maintenance. Soluble CRIPTO reinforces a robust primitive epithelial stem cell state, whereas loss of soluble CRIPTO signaling leaves cells vulnerable to differentiation cues and eventual loss of stem cell characteristics such as bipotency (yellow cells) and self-renewal (circular arrow). See also Figure S4.