Volume 132, Issue 7, Pages 2478-2488 (June 2007) Intestine-Specific Ablation of Mouse atonal homolog 1 (Math1) Reveals a Role in Cellular Homeostasis  Noah F. Shroyer, Michael A. Helmrath, Vincent Y.–C. Wang, Barbara Antalffy, Susan J. Henning, Huda Y. Zoghbi  Gastroenterology  Volume 132, Issue 7, Pages 2478-2488 (June 2007) DOI: 10.1053/j.gastro.2007.03.047 Copyright © 2007 AGA Institute Terms and Conditions

Figure 1 Construction of a conditional allele of Math1. (A) Targeting constructs and recombination events. Mouse genotypes are shown at the left, as referred to in the text. Genomic DNA fragments used for homologous recombination are shown as boxes, with the 5′ homology arm in gray and the 3′ homology arm in black. Coding sequences are in open boxes, transcription start sites as bent arrows, and recombinase target sequences as triangles (loxP sites) or diamonds (frt sites). EcoRI restriction endonuclease sites used for Southern blot genotyping are shown, as is the probe to the 3′ end of the Math1 locus. Fragment sizes detected by this probe are indicated by the double headed arrow below each allele. (1) We identified ES cells in which homologous recombination between pMath1-floxed and the endogenous Math1 locus occurred, giving rise to the Math1Flox-Neo allele. These cells were injected into blastocysts to generate chimeric mice carrying the Math1Flox-Neo allele. (2) Mice bearing the Math1Flox-Neo allele were bred to hACT-FlpE mice to remove the PGK-Neo cassette, giving rise to mice bearing the Math1Flox allele. (3) Mice bearing the Math1Flox allele were bred to EIIa-cre mice to delete the Math1 coding sequence, giving rise to mice bearing the Math1Δ allele. (B) Southern blot with the Math1 3′ probe on tail DNA showing the recombination events described in A. Fragment sizes from the marker are given at the left, genotypes of the mice are shown above each lane. Gastroenterology 2007 132, 2478-2488DOI: (10.1053/j.gastro.2007.03.047) Copyright © 2007 AGA Institute Terms and Conditions

Figure 2 Generation of intestine specific Math1-ablated mice. (A) Schematic showing Rosa26Reporter (Rosa26R) recombination. At the top, in cells that do not transcribe Fabpl4X AT – 132Cre, transcription from the Rosa26R locus is stopped by a transcriptional stop signal (octagon in the figure) that is flanked by loxP sites. Below, cells that express Cre delete the transcription stop signal, and express the lacZ gene. (B) Lumenal en face view of the intestine from [Rosa26Reporter; Fabpl4X AT − 132Cre] bitransgenic mice, opened along the mesenteric side and stained with Xgal to visualize cells that express lacZ. Individual crypts with active Cre (and thus lacZ expression) can be visualized as small dark spots. Recombination occurs primarily in the distal ileum and colon. (C) Schematic showing recombination in [Math1+/flox; Fabpl4X AT − 132Cre] mice. At the top, cells that do not express Cre retain 2 functional copies of Math1. Below, cells that express Cre delete the Math1 coding region from the Math1Flox allele. (D) Southern blot with the Math1 3′ probe on DNA extracted from intestinal tissues from a [Math1+/flox; Fabpl4X AT − 132Cre] mouse. The relative intensity of the unrecombined Math1flox allele (middle band) to the deleted Math1Δ allele (lower band) shows that Math1-deleted epithelium occurs primarily in the distal ileum, cecum, and colon. Gastroenterology 2007 132, 2478-2488DOI: (10.1053/j.gastro.2007.03.047) Copyright © 2007 AGA Institute Terms and Conditions

Figure 3 Mosaic deletion of Math1 in Math1Δintestine/lacZ mice. (A) Schematic drawing representing recombination events in Math1Δintestine/lacZ mice. At the top, cells that do not express Cre retain expression of Math1 from the Math1flox allele. Below, cells that express Cre delete the Math1 coding sequence to generate the Math1Δ allele and are therefore Math1-null. (B) Drawing representing the intestines of Math1Δintestine/lacZ mice, showing the location of Math1-mutant crypts in gray, primarily in the distal ileum, cecum, and colon. (C–E) Hematoxylin and eosin stained sections of jejunum (C), ileum (D), and colon (E) from Math1Δintestine/lacZ mice. The overall structure of the crypts and villi are normal. (F) In situ hybridization with a Math1 cRNA probe (in dark blue) on sections from Math1Δintestine/lacZ ileum demonstrates mosaic loss of Math1 expression in a crypt-by-crypt manner; wild-type crypts are underlined in gray, Math1-mutant crypts are underlined in black. Scale bars in C–E are 100 microns. Gastroenterology 2007 132, 2478-2488DOI: (10.1053/j.gastro.2007.03.047) Copyright © 2007 AGA Institute Terms and Conditions

Figure 4 Math1Δintestine/lacZ mice lack intestinal secretory cells in Math1-mutant crypts. (A) In situ hybridization of Math1Δintestine/lacZ ileum with a Lysozyme cRNA probe (in dark blue) to mark Paneth cells. Wild-type crypts produce abundant Lysozyme, whereas adjacent Math1-mutant crypts express none. (B) Alcian blue/periodic acid Schiff staining (dark purple) of Math1Δintestine/lacZ ileum to mark goblet cells. Wild-type crypts produce abundant mucus-producing goblet cells, whereas adjacent Math1-mutant crypts produce none. (C) Chromogranin A immunostaining (brown) of Math1Δintestine/lacZ ileum to mark enteroendocrine cells. Stained cells indicated with reddish-brown arrowheads. Wild-type crypts produce enteroendocrine cells, whereas adjacent Math1-mutant crypts produce none. (D–G) Xgal staining (blue) of Math1Δintestine/lacZ ileum indicates recent activation of the Math1 promoter. In wild-type crypts, Xgal staining is confined to secretory cell types (gray arrowheads), and is excluded from absorptive enterocytes. Most villus enterocytes are Xgal positive in Math1-mutant epithelium (black arrows). (E–G) Dipeptidyl peptidase IV (DPP-IV; brown) staining of Xgal-stained Math1Δintestine/lacZ ileum. DPP-IV is restricted to the brush borders of absorptive enterocytes. (F) High-power image of wild-type crypt-villus unit. (G) High-power image of Math1-mutant crypt-villus unit. In Math1-mutant epithelium, DPP-IV stained absorptive enterocytes are also stained by Xgal (black arrow in E). (H–I) Transmission electron microscopy images of Math1-mutant epithelium. No wild-type tissue was apparent in this sample. (H) Villus epithelium shows normal appearing absorptive enterocytes but no secretory cells. (I) High-power image of brush border of villus epithelium from H shows microvilli. (A–E) Underline below each image denotes genotype: black underline indicates Math1-mutant crypts, gray underline indicates wild-type crypts. Scale bars in A–E are 100 microns, in F–G are 20 microns, H is 10 microns, and in I is 500 nanometers. Gastroenterology 2007 132, 2478-2488DOI: (10.1053/j.gastro.2007.03.047) Copyright © 2007 AGA Institute Terms and Conditions

Figure 5 Abnormal proliferation in Math1Δintestine/lacZ mice. (A) Immunostaining for BrdU (brown) labels cells in the S-phase following a 2-hour BrdU pulse. Proliferating cells are confined to the crypt in both wild-type (underlined in gray) and Math1-mutant crypts (underlined in black). (B) Bar graphs of proliferation measurements in Math1Δintestine/lacZ mice. Quantitation of BrdU positive cells (left), mitoses (middle), and cell numbers (right) per crypt. Gray bars, wild-type crypts from Math1Δintestine/lacZ ileum; black bars, Math1-mutant crypts. Error bars show standard error, n = 11 mice, >10 crypts/genotype/mouse. Gastroenterology 2007 132, 2478-2488DOI: (10.1053/j.gastro.2007.03.047) Copyright © 2007 AGA Institute Terms and Conditions

Figure 6 Overall adaptation following SBR is normal in Math1Δintestine/lacZ mice. The top graph shows weight changes 7 days after surgery in wild-type (white bars) and Math1Δintestine/lacZ (black bars) mice. Significant weight loss was observed in mice that underwent SBR (solid bars), whereas no significant weight loss was observed in sham-operated animals (Sham, stippled bars); P < .005 for treatment effect. No differences were observed between Math1Δintestine/lacZ mice and their wild-type littermates that underwent the same operation; P > .1 for genotype and genotype × treatment interaction effects. The bottom graph shows ileal wet weight 7 days after surgery. Ilea weighed more in SBR versus sham-treated mice (P < .001); no difference between genotypes was observed. Error bars show standard error, n = 8–13 mice per group. Gastroenterology 2007 132, 2478-2488DOI: (10.1053/j.gastro.2007.03.047) Copyright © 2007 AGA Institute Terms and Conditions

Figure 7 Blunted adaptive response in Math1-mutant crypts 7 days after SBR. Hematoxylin and eosin stained sections of ileum from Math1Δintestine/lacZ mice (A) 7 days after sham operation or (B) 50% SBR. (C) Linear crypt depth measurements in microns. A significant (P < .0001) treatment effect and genotype × treatment interaction effect (P < .0001) was seen. (D) Counting of cell number per crypt. A significant (P < .0001) treatment effect and genotype × treatment interaction effect (P < .01) was seen. Gray bars, wild-type crypts from Math1Δintestine/lacZ ileum; black bars, Math1-mutant crypts. Sham and SBR treatments are indicated below. Error bars show standard error, n = 8–13 mice per group, >10 crypts/genotype/mouse. Gastroenterology 2007 132, 2478-2488DOI: (10.1053/j.gastro.2007.03.047) Copyright © 2007 AGA Institute Terms and Conditions

Figure 8 Blunted adaptive response in Math1-mutant crypts 36 hours after SBR. (A) Hematoxylin and eosin stained sections of ileum from Math1Δintestine/lacZ mice 36 hours after sham operation or (B) 50% SBR. (C) Linear crypt depth measurements in microns. Significant (P < .0001) genotype and treatment effects, and (P < .05) genotype × treatment interaction effect were seen. (D) Counting of cell number per crypt. No significant genotype × treatment interaction effect was seen. Gray bars, wild-type crypts from Math1Δintestine/lacZ ileum; black bars, Math1-mutant crypts. Sham and SBR treatments are indicated below. Error bars show standard error, n = 8–13 mice per group, >10 crypts/genotype/mouse. Gastroenterology 2007 132, 2478-2488DOI: (10.1053/j.gastro.2007.03.047) Copyright © 2007 AGA Institute Terms and Conditions