Final Report Summary - MG STEM (Stem Cells and Cancer Stem Cells in Mouse Mammary Glands)
We have developed a binary system (which we call binary Verde) for labeling and isolating LRCs from the mammary gland. In one of the mice, an rtTA gene is put under the control of the MMTV-LTR and knocked-in into the ROSA26 locus. The regulatory elements of the locus do not contribute to expression of the rtTA gene because the ROSA26 and rtTA genes are transcribed in opposite orientations. The rtTA is expressed at high levels and with significant specificity in the mammary gland without variability among individuals. The second mouse strain (Tumbar et al., 2004), formally designated Tg(tetO-HIST1H2BJ/GFP)47Efu/J, was obtained from the Jackson Labs repository. In this transgenic mouse, H2B/GFP is expressed under the control of tetO fused to cytomegalovirus (CMV) regulatory elements. For operation of the binary Verde system, we obtained progeny by crossing the LTR-rtTA and the tetO-H2B/GFP mouse strains. Pregnant bitransgenic mice were then pulse-labeled by adding Dox to the drinking water from day 11.5 of gestation until parturition. Alternatively, nursing mothers were placed on Dox from day 0 postpartum until weaning of the pups (on day 28, instead of the usual 21). The labeling was restricted to the epithelial and hematopoietic cells in which the MMTV-LTR is also trancriptionally active. The label was subsequently chased for a period of several weeks during which the number of green cells steadily declined (Figure 1).
Bitransgenic mice that were pulsed-labeled during embryogenesis were isolated after a six-week chase period. Four different populations were sorted directly into lysis buffer (RNeasy Micro Kit, Qiagen): LRCs and non-LRCs from the CD24highCD49flow which contain the luminal precursors (CFCs; Stingl et al., 2006); we also sorted LRCs and non-LRC from the CD24medCD49fhigh fraction which contains the MaSCs. The expression profiles of all four populations were generated using Affymetrix expression arrays (Mouse 430 2.0) and analyzed with the Genespring software. Of higher priority were genes which were differentially expressed in the MaSC LRCs in comparison to the other three populations (Table I).
We found that among other genes, the Notch ligand Delta-like 1 (Dll1) was 13X higher in MaSC LRCs in comparison to the MaSC non-LRCs, and approximately 20X higher in comparison to the luminal cells. Because it has been already reported that only 1 in 50 to 1 in 100 cells within the CD24+CD49fhigh or CD24+CD29high population is actually a stem cell (Stingl et al., 2006; Shackleton et al., 2006), we sought to test whether Dll1 can be used to further enrich for MaSCs within the aforementioned gates. The populations were separated as follows: Dll1+ positive cells are visualized with a goat (SC-12530) primary and a APC-Cy7-conjugated anti-goat secondary antibody and separated from Dll1- cells (P3 and P4, respectively). The two populations are subsequently plotted on the CD24/CD49f profile for the CD24+CD49fhighDll1+ (P5) and CD24+CD49fhighDll1- (P6) populations to be isolated (Figure 2). Interestingly, the majority of Dll1+ cells fall within the CD24+Cd49fhigh group, the presumptive location of MaSCs (Figure 2, bottom right panel, green and blue cells together). Using this approach we were able to isolate Dll1-positive (Dll1+) cells from CMV-GFP mice that constitutively express GFP in all cells. GFP is this case serves merely as a marker to distinguish between growths originating from transplanted cells from those that are the result of incompletely cleared fat pads. As shown in Figure 3, only CD24+/CD49f+/Dll1+ cells are able to reconstitute mammary glands whereas CD24+/CD49f+/Dll1- cells fail to do so in two independent experiments.
We then generated a “knockin” mouse line in which a CreERT2/IRESegfp cassette was grafted directly into the 5’ UTR of the DLL1 locus with the ATG of the endogenous transcript to serve as the initiation codon of CreERT2 recombinase. This way, we hope to achieve a CreERT2 and egfp expression pattern that recapitulates as closely as possible that of Dll1. We crossed this mouse line (Dll1-cre/egfp) with a ROSA-stop-tdTomato mouse line that expresses the fluorescent protein tdTomato upon Cre recombination. Bitransgenic virgin mice were administered Tamoxifen and mammary glands were dissected, fixed, embedded, sectioned and visualized by fluorescent microscopy. As already indicated from our flow sorting experiments, Tomato-positive cells were identified in both the luminal and myoepithelial layer (Figure 4).
Finally, we flow sorted Tomato-positive cells separately from the two layers and used them in fat pad assays. As shown in Figure 5, only basal Dll1-expressing cells are able to reconstitute the mammary gland indicating that only basal cells have bipotent potential.
Overall, our experiments show that Dll1-expressing cells with basal characteristics have properties of mammary stem cells. Whether Dll1-positive residing in the luminal layer are committed luminal progenitors or not, is under question and we hope that our current experiments will address this important question.