Brief report: Musashi1-eGFP mice, a new tool for differential isolation of the intestinal stem cell populations


  • Francesca Maria Cambuli,

    1. Centre de Génétique et de Physiologie Moléculaire et Cellulaire, Université Claude Bernard, France
    Search for more papers by this author
  • Amélie Rezza,

    1. Centre de Génétique et de Physiologie Moléculaire et Cellulaire, Université Claude Bernard, France
    Search for more papers by this author
  • Julien Nadjar,

    1. Centre de Génétique et de Physiologie Moléculaire et Cellulaire, Université Claude Bernard, France
    Search for more papers by this author
  • Michelina Plateroti

    Corresponding author
    1. Centre de Génétique et de Physiologie Moléculaire et Cellulaire, Université Claude Bernard, France
    • Correspondence: Michelina Plateroti, Ph.D., Centre de Génétique et de Physiologie Moléculaire et Cellulaire, Université Lyon 1, 16 Rue Raphael Dubois, 69622 Villeurbanne, France. Telephone: 33–472431595; Fax: 33–472432685; e-mail:

    Search for more papers by this author

  • Author contributions: F.C. and A.R.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing; J.N.: collection and assembly of data; M.P.: conception and design, financial support, assembly of data, data analysis and interpretation, manuscript writing, and final approval of the manuscript. All authors approved the manuscript. F.M.C. and A.R. contributed equally to this article.


The intestinal epithelium self-renews rapidly and continuously throughout life, due to the presence of crypt stem cells. Two pools of these cells have been identified in the small intestine, which differ in position (“+4” or the bottom of the crypts), expression of specific markers (Bmi1/mTert or Lgr5/Ascl2), and cell cycle characteristics. Interestingly, the RNA-binding protein Musashi1 is expressed in both populations and therefore a potential marker for both stem cell types. In order to locate, isolate, and study Musashi1-expressing cells within the intestinal epithelium, we generated transgenic mice expressing GFP fluorescent protein under the control of a 7-kb Msi1 promoter. The expression pattern of GFP in the intestinal crypts of both small and large intestines completely overlapped that of Musashi1, validating our model. By using fluorescence-activated cell sorting, cellular, and molecular analyses, we showed that GFP-positive Msi1-expressing cells are divided into two major pools corresponding to the Lgr5- and mTert-expressing stem cells. Interestingly, monitoring the cell cycle activity of the two sorted populations reveals that they are both actively cycling, although differences in cell cycle length were confirmed. Altogether, our new reporter mouse model based upon Musashi1 expression is a useful tool to isolate and study stem cells of the intestinal epithelium. Moreover, these mice uniquely enable the concomitant study of two pools of intestinal stem cells within the same animal model. Stem Cells 2013;31:2273–2278


Continuous renewal of the intestinal epithelium depends on stem cells located in the crypts of Lieberkühn [1]. Different studies have suggested that two pools of stem cells exist: one located at the very bottom of the crypts, the crypt basal columnar (CBC) stem cells, actively cycling and expressing Lgr5 and Ascl2 markers [2, 3], and one considered quiescent but more resistant to irradiation [4-6], located at the “+4” position from the crypt bottom and expressing m-Tert, Lrig1, and D-Camkl1 markers [2, 6-9]. The “+4”-stem cells are required to maintain intestinal crypts, and thus epithelial homeostasis [5]. However, studies showed that the best-characterized stem cell markers are expressed in gradient throughout a “stem zone,” and not exclusively in a single stem cell pool [10, 11]. In addition, a recent paper showed that the quiescent label-retaining cells (LRC) in the crypts express CBC-associated genes, can serve as reserve stem cells upon injury, and are Paneth and enteroendocrine precursors [12], but their precise link to the classically defined “+4”-stem cells remains unclear. Therefore, the exact cell hierarchy at the crypt bottom is still elusive, and the debate concerning the location and physiology of gut stem cells has been reopened [13]. Interestingly, several studies reported Musashi1 (Msi1) as an intestinal epithelial stem cell marker for both populations of small intestine stem cells, which also identifies colon stem cells [4, 11, 14]. Msi1 is an RNA-binding protein originally described to control stemness in Drosophila [15]. We have recently shown that its overexpression in intestinal epithelial progenitors enhances their proliferative capacity through activation of Wnt and Notch pathways [16], key regulators of gut cell fate and stem cell biology [17-19]. In order to study Msi1-expressing cells and their potential stem cell-like properties, we describe here the generation of Msi1-eGFP transgenic mice as a new and efficient tool to concomitantly label and study intestinal crypts stem cells.

Materials and Methods

Materials and Methods are described in the Supporting Information.

Results and Discussion

Msi1-eGFP Mice Express GFP in the Gut Stem Cell Compartment

We generated Msi1-eGFP mice (Supporting Information Fig. S1) to study Msi1-expressing cells and their associated stemness. Green fluorescent protein (GFP) immunostaining on sections revealed specific expression in small intestine at the level of the CBCs and around the +4 position (Fig. 1A, 1C, 1D; Supporting Information Fig. S2) and at the bottom of colonic crypts (Supporting Information Fig. S3). Importantly, GFP labeling strongly overlapped the expression pattern of Msi1 in all Msi1-eGFP lines (Fig. 1A; Supporting Information Fig. S2, S3). Coimmunolabeling experiments showed the presence of GFP+ cells between and above Paneth cells (Supporting Information Fig. S4A). Moreover, GFP colocalized with a restricted subset of β-catenin or CD44-expressing cells at the crypt bottom (Supporting Information Fig. S4B, S4C). To further characterize this new model, small intestinal epithelia of Msi1-eGFP mice were fractionated (Fig. 2; Supporting Information Fig. S5) and the presence of GFP in the fractions was evaluated by fluorescent microscopy. As expected GFP+ cells were detected only in F4 from F19 and F20 Msi1-eGFP mice (Fig. 2C; Supporting Information Fig. S5B). Moreover, F4 cells were enriched for GFP and Msi1 mRNAs and for the established stem cell markers Lgr5 and Bmi1 mRNAs [2, 5] (Fig. 2D; Supporting Information Fig. S5C). These data show that our model expresses GFP specifically in the stem cell compartment of the small intestinal epithelium.

Figure 1.

Analysis of GFP and Msi1 expression in small intestinal crypts of Msi1-eGFP mice. (A, B): Immunofluorescence analysis of Msi1 and/or GFP on small intestinal sections from F19 Msi1-eGFP (A) or WT (B) mice. Each panel represents a merged image of 4′,6′-diamidino-2-phénylindole (DAPI) (all nuclei), Msi1 and/or GFP staining as indicated. WT littermates displayed no GFP expression (B). (C): Immunohistoenzymatic staining for GFP confirmed the presence of GFP-positive cells at the bottom of the crypts. Black arrows point to a few labeled cells. (D): Frequency of Msi1/GFP-positive cells at specific position of the crypt vertical axis, as depicted in the upper scheme. Crypt basal columnar cells in position +1 are in brown, “+4” stem cells are in blue, Paneth cells are in yellow, green cells represent progenitors. It is worth noting that Msi1/GFP-positive cells are present at high frequency at the positions +1 and +4, while the Lgr5-GFP cells have a more restricted frequency around the +1 position [2]. Counts were performed on 50-well oriented crypts. Scale bar = 15 μm (A, B); enlargement = 7.5 μm; C = 0.5 μm. Abbreviations: GFP green fluorescent protein; WT, wild type.

Figure 2.

Presence of GFP+ cells in crypt cell preparations of Msi1-eGFP mice. (A, B): Intestinal epithelial cells from F19 were separated in four fractions from the top of the villi to the bottom of the crypts (A). Efficiency of fractionation was assessed by quantitative reverse transcriptase polymerase chain reaction (RTqPCR) for Cyclin D1 and I-Fabp (B) in fraction one (F1, differentiated cells) and fraction four (F4, crypt cells). (C): GFP+ cells were detected only in F4 of Msi1-eGFP mice. No GFP was present in F4 of WT mice. Pictures show bright-field (upper panels), GFP (middle panels), and merged images (lower panels). Scale bar = 100 μm. (D): RTqPCR experiments showed enrichment of GFP, Msi1, and stem cell markers Lgr5 and Bmi1 mRNAs in F4 compared to F1. Histograms represent mean ± SD, n = 4, after normalization with PPIA or PPIB mRNA. The results shown are representative of at least four independent experiments. ***, p < .001 by Student t test, comparing F1 versus F4 of the same genotype. Abbreviations: GFP, green fluorescent protein; WT, wild type.

Different Levels of GFP in Msi1-eGFP Mice Distinguish Two Intestinal Stem Cell Pools

To identify and study GFP+ cells from Msi1-eGFP small intestinal crypts, we used a fluorescence-activated cell sorting approach. Given that GFP+/GFP− populations showed strong similarities between founders (not shown), detailed analysis focused on F19 line. GFP levels and the common crypt-cell surface marker CD24 [12, 20] were used to isolate different cell populations. GFPHi and GFPLo cells were distinguishable (Fig. 3A), both being CD24Lo (Fig. 3B). Four populations were sorted: GFP−/CD24Hi, GFP−/CD24Lo, GFPLo/CD24Lo, and GFPHi/CD24Lo (Fig. 3B), and quantitative reverse transcriptase polymerase chain reaction (RTqPCR) was performed (Fig. 3C; Supporting Information Fig. S6). GFPHi/CD24Lo cells (from here GFPHi) expressed high levels of Msi1, mTert, Hopx, and Lrig1 mRNAs, and low levels of Lgr5, Ascl2, Olfm4, and Smoc2 mRNAs, suggesting an enrichment of “+4”-stem cells in this population. No specific enrichment of Bmi1 was observed, consistent with its expression throughout the crypts [10, 21, 22]. GFPLo/CD24Lo cells (from here GFPLo) expressed low levels of Msi1, mTert, Hopx, and Lrig1 but high amounts of Lgr5, Ascl2, Olfm4, and Smoc2 mRNAs, reflecting the presence of CBC-stem cells in this population. GFPLo cells also expressed high levels of Wnt-targets Axin2 and c-Myc (Supporting Information Fig. S6C), in accordance with high Wnt activity in CBCs [23]. Mmp7, recently described as a LRC marker [12], was absent from both GFP+ populations (Supporting Information Fig. S6D). GFP−/CD24Lo and GFP−/CD24Hi cells are clearly mixed populations (Fig. 3C; Supporting Information Fig. S6D). No contaminants from differentiated epithelial cells were detected in GFPHi or GFPLo cells (Supporting Information Fig. S6D). Analysis of Msi1 expression in intestinal crypts of Lgr5-GFP mice [2] confirmed a gradient of Msi1 expression in the different GFP+ cell populations (not shown), as reported [11].

Figure 3.

Different levels of GFP distinguish two pools of stem cells in the small intestine of Msi1-eGFP mice. (A): Fluorescence-activated cell sorting (FACS) analysis of small intestine epithelial crypt cells from F19 revealed the presence of two populations of cells expressing different levels of GFP. (B): FACS sorting of crypt cells from Msi1-eGFP small intestine. Staining of crypt cells with CD24 showed the presence of four distinguishable populations. (C): Quantitative reverse transcriptase polymerase chain reaction experiments on indicated populations showed preferential expression of the “+4” stem cell markers Msi1, mTert, Hopx, Lrig1 mRNAs in GFPHi cells. GFPHi cells express high levels of Hes1, previously associated with the intestinal stem cell signature [11], further confirming coexpression of Msi1 and Hes1 [26, 27]. Hes1 is also expressed by absorptive progenitors [24], suggesting the potential presence of these cells within the GFPHi population. Finally, these cells also express high levels of Frat1 mRNA, a target of Msi1 [16]. Lgr5, Ascl2, Olfm4, and Smoc2 mRNAs were highly expressed in GFPLo cells, suggesting enrichment of crypt basal columnar stem cells in this population. Intriguingly, Ascl2 mRNA was also highly detected in the GFP−/CD24Lo population. Histograms represent mean ± SD, n = 3, after normalization with PPIA or PPIB mRNA. Results are representative of at least seven independent experiments (Supporting Information Fig. S5B). Abbreviations: GFP, green fluorescent protein; SSC, side scatter cell.

Altogether, these results strongly indicate that Msi1 expression characterizes the whole crypt stem cell zone and suggest the existence of two distinct pools of stem cells, as originally proposed [25]. Moreover, the differential GFP expression in Msi1-eGFP mice distinguishes CBCs- and “+4”-stem cells, as also suggested by the frequency profile (Fig. 1D).

GFPHi and GFPLo Cells Have Different Cell Cycle Activity

Previous studies on cycling activity, evaluated by BrdU incorporation and immunohistochemistry, suggested that “+4”-stem cells are essentially quiescent while CBCs are actively cycling [2, 4, 6]. To quantitatively analyze the proliferative characteristics of the different populations, bromo-deoxy-uridine (BrdU) was administered to Msi1-eGFP mice and its incorporation was analyzed by flow cytometry (Fig. 4; Supporting Information Fig. S7). Three incorporation protocols were used: one pulse with a 2-hour chase to label cells in S-phase or with a 48-hour chase to follow crypt cell dynamics, and a continuous oral administration for 48-hour to label every replicating/replicated cell (Fig. 4A). After a 2-hour chase, 7.3% (±0.28, n = 2) of total crypt cells incorporated BrdU, whereas only 2.55% (±0.35, n = 2) remained BrdU+ after a chase period of 48 hours (p < .01). After continuous administration for 48 hours, 12.85% (±0.49, n = 2) of crypt cells were BrdU+ (p < .05). GFPHi and GFPLo populations were enriched in BrdU+ cells compared to total crypt cells after a 2-hour chase (Fig. 4B) but these percentages were not significantly different contrary to the assumption of “+4”-stem cell quiescence [5, 6]. After a 48-hour chase, the percentage of BrdU+ cells was significantly higher in GFPHi than in GFPLo cells, clearly indicating a slower cell cycle. This result was confirmed by the continuous BrdU administration, where the GFPHi population showed significantly fewer BrdU+ cells than the GFPLo population (Fig. 4B). These data show that the GFPHi population is enriched in cells with a longer cycle than GFPLo cells, suggesting that “+4”-stem cells, that is, Msi1-GFPHi cells, are slow cycling compared to rapidly cycling CBCs, that is, Msi1-GFPLo cells.

Figure 4.

GFPHi (“+4” stem cells) and GFPLo (crypt basal columnar stem cells) cells have different cell cycle length. (A): Three different BrdU administration protocols were used. BrdU was injected as a single pulse with a chase period of 2 or 48 hours or through oral administration over 48 hours before sacrifice. BrdU incorporation was analyzed by flow cytometry on sorted cells. (B): Percentages of BrdU+ cells in GFPHi and GFPLo sorted populations after application of the different labeling protocols. Histograms are representative of two independent experiments. **, p < .05 by Student t test. Abbreviations: BrdU, bromo-deoxy-uridine; GFP, green fluorescent protein.


We show here that newly generated Msi1-eGFP mice express different levels of GFP specifically within different pools of intestinal stem cells: GFPHi cells preferentially express “+4”-stem cell markers and GFPLo, CBC-stem cell markers. These populations are clearly distinct regarding gene expression and cell cycle characteristics. However, contrary to previous reports, Msi1-GFPHi “+4”-stem cells are not quiescent but do have a slower cell cycle than CBCs. Additionally, Msi1-GFPHi cells appear different from LRCs.

Altogether, Msi1-eGFP mice represent a useful model to study Msi1-expressing cells and their potential stem cell properties. Future investigations will aim to better characterize these cells and the specific hierarchy between CBCs, Msi1-GFPHi “+4”-stem cells and LRCs. This new model will also be useful to explore whether Msi1-expressing cells are involved in gut tumor development, as it has been shown for Lgr5- and Bmi1-expressing cells.


We thank Nadine Aguilera for animal handling, Sebastien Dussurgey and Thibault Andrieu for their invaluable help with fluorescence-activated cell sorting analysis. We are indebted with Dr. Philippe Jay for Lgr5-GFP animals and with Dr. Maria Sirakov and Rachel Sennett for critical reading of the manuscript. This work was supported by the Institut National pour le Cancer (Grant INCA-2009-175), the ANR Blanc ThRaSt (ANR-11-BSV2-019), and the Ligue contre le cancer Department du Rhone (N-074937). F.C. was supported by the Associazione Italiana per la Ricerca sul Cancro (AIRC) and the Association pour la Recherche sur le Cancer (ARC); A.R. was supported by the Ligue Nationale Contre le Cancer and the Fondation pour la Recherche Medicale. A.R. is currently affiliated with Black Family Stem Cell Institute, Department of Developmental and Regenerative Biology, Icahn School of Medicine at Mount Sinai, New York, NY.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflict of interest.