Requirement of Math1 for Secretory Cell Lineage Commitment in the Mouse Intestine

See allHide authors and affiliations

Science  07 Dec 2001:
Vol. 294, Issue 5549, pp. 2155-2158
DOI: 10.1126/science.1065718


The mouse small intestinal epithelium consists of four principal cell types deriving from one multipotent stem cell: enterocytes, goblet, enteroendocrine, and Paneth cells. Previous studies showed that Math1, a basic helix-loop-helix (bHLH) transcription factor, is expressed in the gut. We find that loss ofMath1 leads to depletion of goblet, enteroendocrine, and Paneth cells without affecting enterocytes. Colocalization ofMath1 with Ki-67 in some proliferating cells suggests that secretory cells (goblet, enteroendocrine, and Paneth cells) arise from a common progenitor that expresses Math1, whereas absorptive cells (enterocytes) arise from a progenitor that isMath1-independent. The continuous rapid renewal of these cells makes the intestinal epithelium a model system for the study of stem cell regeneration and lineage commitment.

The mouse gut begins developing at embryonic day 7.5 (E7.5). Invagination of the most anterior and posterior endoderm leads to the formation of the foregut and hindgut pockets, respectively, which extend toward each other and fuse to form the gut tube. By E15.5, the gut appears as a poorly differentiated, pseudostratified epithelium. From E15.5 to E19, nascent villi with a monolayer of epithelial cells develop in a duodenum-to-colon pattern. During the first two postnatal weeks, the intervillus epithelium, where proliferating and less differentiated cells reside, develops into the crypts of Lieberkühn. Stem cells in the intervillus epithelium (during embryogenesis) or crypts (in adulthood) give rise to four principle cell types: absorptive enterocytes or columnar cells, mucous-secreting goblet cells, regulatory peptide-secreting enteroendocrine cells in the large and small intestines, and antimicrobial peptide-secreting Paneth cells in the small intestine only. Enterocytic, goblet, and enteroendocrine cells continue to differentiate and mature while migrating up the villus, and are finally extruded into the lumen at the tip. This journey takes about 2 to 3 days. The Paneth cells migrate downward and reside at the base of the crypt for ∼21 days before being cleared by phagocytosis (1–3).

The epithelial-mesenchymal interaction has been shown to be critical in the proximal-distal, crypt-villus patterning during gut development. A number of signaling molecules and transcription factors are involved in these processes (4–7). Previous studies have suggested that all four epithelial cell lineages originate from a common ancestor (13, 8), but the mechanisms that control the epithelial lineage differentiation are not well understood. T cell factor-4 (Tcf4) plays a role in the stem cell maintenance in the small intestine but does not induce epithelial cells to differentiate into enterocytes or goblet cells (9). Because Math1 is expressed in the gut (10) and involved in cell fate determination in the nervous system (11, 12), we sought to determine its function during gut development.

We have two null alleles forMath1:Math1 –/– (with the coding region replaced by Hprt) andMath1 β-Gal/β-Gal (with the coding region replaced by the β-galactosidase gene, which is then expressed under the control of the Math1 promoter) (11).Math1 null mice die shortly after birth, butMath1 heterozygous mice survive to adulthood and appear normal. We previously showed that Math1/LacZ expression faithfully mimics the endogenous gene expression (11). Here we used Math1 β-Gal/– instead ofMath1 β-Gal/β-Gal null mice for X-gal staining experiments to ensure equal copy numbers of theLacZ gene in heterozygous and null animals.

Math1/LacZ expression within the gut is restricted to the intestinal epithelium starting at E16.5 and is sustained until adulthood (13). We detected noMath1/LacZ expression in the stomach, pancreas, or lung. In E18.5 heterozygous mice, LacZ-positive cells are sparsely scattered in the villi, the intervillus epithelium (Fig. 1A), and colonic crypts (Fig. 1C). InMath1 null littermates, however, LacZ-expressing cells are clustered in the intervillus region of ileum (Fig. 1B) and at the bases of the colonic crypts (Fig. 1D).Math1/LacZ expression persists throughout duodenum, jejunum, ileum, and colon [Fig. 1, E and F, and Web fig. 1 (14)] in adultMath1β-Gal/+ mice. In the villi, the scattered blue cells appear to have a goblet cell morphology (a spherical vacuole); at the base of the crypt, most apical granule-containing Paneth cells appear to be LacZ-positive. X-gal stained cells are also found in the mid-crypt region.LacZ expression in adult crypts suggests thatMath1 helps initiate cytodifferentiation of the epithelial cells.

Figure 1

Math1/LacZ expression detected by X-gal staining. Math1/LacZ expression in E18.5 intestines (A to D). Cross section ofMath1β-Gal/+ ileum (A) and colon (C),Math1β-Gal/– ileum (B) and colon (D); Arrows indicate sparse lacZ-positive cells in heterozygous animals. Math1/LacZ expression in 5-month-oldMath1β-Gal/+ mice (E andF). Longitudinal section of jejunum (E) and colon (F). Original magnification, ×400.

We detected no Math1/LacZ expression in the enteric nervous system (intestinal) from E14.5 to adult. An acetylcholinesterase activity assay (15) revealed no gross abnormalities in the enteric neurons (16), although subtle deficits may not be apparent at these resolutions.

The small and large intestines of Math1 null embryos (E14.5 to E18.5) showed normal villus architecture, lamina propria, and musculature, but no goblet cells (Fig. 2, A and B). In wild-type animals Alcian blue–positive goblet cells increased in number along the duodenal-colonal axis (Fig. 2C), but were not detected in Math1 β-Gal/– mice (Fig. 2D).

Figure 2

Lack of goblet and enteroendocrine cells in E18.5 intestines. H&E staining reveals several goblet cells in wild-type duodenum [arrowheads in (A)] and none in null mutant (B); Alcian blue staining shows positively stained goblet cells in wild-type ileum [(arrowheads in (C)] but none in Math1 β-Gal/– null ileum (D). Serotonin-positive enteroendocrine cells [red-stained cells in (E), the arrow points to the cell enlarged in the inset] are evident in wild-type (E) but notMath1 β-Gal/– (F) jejunum. Original magnification, ×200.

We then analyzed the enteroendocrine lineage in the gut epithelium. Neither pan-endocrine markers (chromogranin A, synaptophysin) or specific endocrine markers (glucagon, gastrin, somatostatin, neurotensin, and serotonin) (17) were detectable in any regions of Math1 β-Gal/– null mouse intestine (Fig. 2F; cf. wild type, Fig. 2E).

Electron microscopy (EM) on E18.5 embryos (17) revealed no granular or common goblet or enteroendocrine cells in any region of Math1 β-Gal/– null mouse intestines (Fig. 3B, cf. wild type in 3A). Null mouse enterocytes, however, had a normal microvillus brush border: strongly positive for alkaline phosphatase and lactase, ample endoplasmic reticulum, a few secondary lysosomes, and regular columnar height with uniform nuclei close to the inner aspect of the cell [Fig. 3B, Web fig. 2 (14)]. Some mutant enterocytes have abundant glycogen (16), like immature enterocytes, whereas wild-type enterocytes no longer have cytoplasmic clusters.

Figure 3

Electron microscopy, cryptdin RT-PCR and colocalization of Math1/LacZ and proliferation marker Ki-67. (A) EM of the ileum reveals goblet cells (G) and enteroendocrine cells (E) in wild-type mice; neither of these secretory cells is formed in the Math1 null mice (B). Enterocytes [arrowheads in (A) and (B)] appear normal inMath1 null mice (B). Cryptdin mRNA (C) was detected in wild-type duodenum, jejunum, and ileum, but not in colon, whereas the Math1 null mutant lacked cryptdin RNA in all intestinal tissues examined. G6PDH mRNA level was used as a control. X-gal and Ki-67 antibody staining (blue cytoplasmic and red nuclear, respectively) in sections from adult duodenum (D) and ileum (E) (no hematoxylin counterstaining was applied). Paneth cells with apical granules, located at the bottom of crypts (arrow), show no Ki-67 staining. A subset of Ki-67–positive cells are alsoMath1/LacZ-positive (arrowheads). Original magnification, ×2500 (A and B), ×1000 (D and E).

Electron microscopy cannot be used to evaluate Paneth cells inMath1 null animals, because their characteristic apical granules do not mature until after birth (18). But cryptdin-1 is one of the earliest markers expressed in Paneth cells, starting at E15.5 (19), so we examined its expression (20). Cryptdin-1 consensus primers were used to amplify a 272-bp product corresponding to nucleotides 80 to 352 (21). Cryptdin-1 expression was detected in wild-type duodenum, jejunum, and ileum but was completely absent in these three regions inMath1 null animals (Fig. 3C). As expected (19), we detected no cryptdin-positive cells in wild-type or Math1null colon (Fig. 3C). Neither EM nor tunnel assays revealed signs of premature cell death in Math1 null gut [Fig. 3B and (16)].

In adult crypts, epithelial stem cells and multipotent progenitor cells are proliferating and show nuclear staining for Ki-67, a cell proliferation marker (9) (Fig. 3, D and E). In theMath1 β-Gal/+ mice, theLacZ-expressing cells show a cytoplasmic blue staining pattern (11). This feature permits colocalization ofMath1/β-galactosidase and Ki-67. In the crypts, double-positive cells are scattered from the 4th to 13th cell position from the base of the small intestine and are in the 2nd to 4th position in the colon [Fig. 3, D and E, Web fig. 3 (14); (16)]. Clearly not all Ki-67–positive cells express Math1, suggesting that Math1-negative progenitors give rise to the enterocytes, whereasMath1-expressing progenitors become goblet, enteroendocrine, and Paneth cells. Upon deletion of Math1, the latter group of cells fail to differentiate, and their progenitors remain in the proliferating stage, thus accounting for the intense X-gal staining seen in the crypts of null embryos (Fig. 1, B and D). To ascertain the effects of Math1 deletion on the proliferative status of the secretory lineage progenitors, we examined 2500 Ki-67–positive cells in three pairs of E18.5 Math1 null and heterozygous mice for LacZ-positive staining. We scored double-positive cells as a fraction of the total cycling Ki-67–positive population in E18.5 Math1 heterozygous and null mouse intestines. The ratio of double-positive to Ki-67–positive cells in Math1null animals, from duodenum to colon, was roughly three times that seen in heterozygotes (25 to 68% versus 7 to 22%), supporting the hypothesis that cells lacking Math1 fail to exit the cell cycle and differentiate.

Previous studies have shown that members of the Notch signaling pathway (e.g., Mash1, Neurogenin 3, andNeuroD) are involved in endocrine cell differentiation (22–24). Deletion of Hes1, a Notch signaling component that represses bHLH transcriptional activators, leads to an increased number of enteroendocrine and goblet cells but fewer enterocytes, and elevates expression ofDelta1, Delta3, NeuroD, andMath1 in the small intestine (25).Hes1 also negatively regulates inner ear hair cell differentiation by suppressing Math1(26). These studies support the hypothesis thatMath1 controls cell fate determination via a Delta-Notch signaling pathway.

Quantitative reverse transcription–polymerase chain reaction (RT-PCR) analysis revealed that Delta3 was reduced to half of wild-type levels in Math1 null mice, andNeuroD expression was lost completely (Fig. 4A). In contrast, Delta1,Hes-1, and Notch1, 2, 3, and4 expression levels and cellular localization ofHes-1 appeared unaffected [Fig. 4A and Web fig. 4 (14)]. These observations are consistent with previous findings that Math1 is upstream ofNeuroD (27) and the notion thatMath1 has a positive feedback effect on Notch ligand (e.g.,Delta3) expression.

Figure 4

Expression of Notch components and model for epithelial cell lineage differentiation in mouse intestine. (A) E18.5 small intestines were subjected to RT-PCR using primers specific to the indicated genes. G6PDH mRNA level served as a control. (B) Math1 is essential for secretory cells. Whether Math1-expressing cells descend directly from stem cells or an intermediate progenitor remains unknown. Abbreviations: Sec, secretin; L, glucagon/peptide YY; CCK, cholecystokinin; SP, substance P; 5HT, serotonin; Som, somatostatin; GIP, gastric inhibitory peptide; Gas, gastrin.

Our findings provide new insight into the role of Notch-mediated lateral inhibition in controlling differentiation of intestinal epithelial lineages. Building on the model set forth by Bjerknes and Cheng (8), we propose that a single self-maintaining stem cell gives rise to two daughter cells directly or through intermediate progenitors (Fig. 4B). In one daughter cell, interaction between Delta and Notch homologs elevates Hes1 expression, inhibitingMath1 expression, and this cell adopts an enterocyte fate. In the other daughter cell, lack of Hes1 expression increases Math1 expression, and this cell becomes a committed multipotent progenitor that will differentiate into a secretory lineage cell (Fig. 4B). Further differentiation of the secretory lineage into goblet, enteroendocrine, and Paneth cells requires other factors. NeuroD has been shown to play a role in differentiation of the secretin and cholecystokinin enteroendocrine cells (24); early committed multipotential endocrine cells can branch into at least three lineages (Fig. 4B) (24). Rac1 is reported to play a positive role in goblet and Paneth cell differentiation but does not seem to have any impact on the enteroendocrine lineage (18), suggesting that goblet and Paneth cells share a closer relationship during later development. Constitutively activated Rac1causes precocious enterocyte growth, indicating its positive role in the absorptive cell lineage (18). These observations suggest that there is cross talk between the Notch and Rho GTPase pathways during formation of the gut epithelium. We cannot yet rule out other models for controlling the secretory and absorptive lineages: e.g., instead of arising from one Math1-positive progenitor, the goblet, enteroendocrine, and Paneth cells may differentiate from three distinct progenitors that each express Math1. Further study of Math1 in the mouse intestine will yield deeper insight into the mechanisms controlling production of the different cell types, which may in turn provide therapeutic tools for endocrine and colorectal cancers and regeneration of the intestinal epithelium after injury.

  • * To whom correspondence should be addressed. E-mail: hzoghbi{at}


View Abstract

Navigate This Article