INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

Dimeric basic helix-loop-helix (bHLH)¹ transcription factors play important roles in cellular development. Their HLH domain promotes dimer formation, while their basic domain mediates binding to an E-box sequence (CANNTG) present in target genes. Homodimers of class B, tissue-specific bHLHs activate transcription of "master regulatory" bHLHs that allow cells to commit to a particular fate. Tissue-specific bHLHs can also form heterodimers with ubiquitously expressed class A bHLHs (also known as E-proteins). These heterodimers activate transcription of genes associated with differentiation (reviewed in Refs. 1-4).

The four known Ids (inhibitors of DNA binding; Ref. 5) are naturally occurring dominant-negative HLH proteins that lack the basic domain (1, 2, 4). Each Id forms high affinity heterodimers with E-proteins (6). These heterodimers cannot bind to DNA and therefore inhibit transcription of differentiation-associated genes by sequestering E-proteins (5-9). Forced expression of Ids has been used to show that tissue-specific bHLHs regulate differentiation in cultured cells and in specified cell lineages in vivo (10-19). Individual Ids have distinct affinities for tissue-specific bHLH proteins (6), suggesting that each Id could regulate expression of different subsets of bHLH target genes. Individual Ids also interact in distinct ways with non-bHLH proteins (20-23).

Ids and their affiliated proteins also affect proliferation. For example, Ids appear to be required by human diploid fibroblasts and NIH 3T3 cells for progression through G₁/S (24, 25). Interactions between Id-2 (but not Id-1 or Id-3) and pRB are associated with enhanced proliferation in cultured cells (21, 26). Yan et al. (27) recently reported that E2A^/ mice that escape neonatal lethality develop T-cell lymphomas. In situ hybridization studies of developing normal mouse embryos revealed high levels of Id expression in proliferating undifferentiated cells and decreased expression in committed differentiating cells (28, 29).

Although considerable attention has been paid to the roles of Ids in myogenesis, neurogenesis, and B-cell development (1-4), there have been few in vivo studies of their functions in, and effects on, epithelia. The mouse intestinal epithelium has two features that make it attractive for examining the effects of Id expression. First, it renews itself rapidly and continuously throughout life. Proliferation, lineage allocation, differentiation, and death are confined to readily identifiable regions in each of the adult intestine's crypt-villus units: i.e. cells at all developmental stages are represented. Second, this epithelium is known to produce several (b)HLH proteins (28, 30-32).

The self-renewing adult mouse small intestinal epithelium contains ~1 million flask-shaped structures known as crypts of Lieberkühn (33). One or more active multipotent stem cells are functionally anchored near the base of each long lived crypt (34-39). Stem cells give rise to daughters that undergo 4-6 rounds of cell division in the mid-portion of the crypt (40). Epithelial cells are allocated to four principal lineages through mechanisms that are poorly understood. Committed cells belonging to each lineage complete their differentiation during a highly organized, rapid migration. Differentiating absorptive enterocytes, mucus-producing goblet cells, and enteroendocrine cells exit the crypt and move up an adjacent finger-like structure (the villus) in coherent columns (41, 42). Cells are removed at the villus tip by apoptosis and/or by extrusion into the lumen (43). The entire sequence is completed in 3-5d (44-47). Paneth cells differentiate as they move down to the crypt base, where they secrete antimicrobial peptides and growth factors (34, 48-52). Their life span at the crypt base is ~20 days (48). One of the intestine's known bHLH proteins, BETA2/NeuroD, regulates expression of secretin, a peptide produced during terminal differentiation of a subset of enteroendocrine cells (32). Another bHLH protein, MATH-1, is related to atonal, a master regulator of neurogenesis and is expressed almost exclusively in the intestinal epithelium of adult mice (31). Its function is unknown.

There are several reasons why we decided to assess the effects of forced expression of Id-1 on intestinal epithelial homeostasis. Id-1 is normally expressed at a critical point in gut development, embryonic day 16, when the endoderm begins to undergo cytodifferentiation to an epithelial monolayer. At this time, the intestinal epithelium is one of the few sites where Id-1 mRNA levels are high (28). Moreover, forced expression of Id-1 in many cultured cell lines has been shown to inhibit bHLH-regulated differentiation (10-15). Finally, Id-1 has been used in transgenic mice to disrupt bHLH functions in nonepithelial cell lineages: B-cells (16) and skeletal muscle (53). In this report, we describe the use of chimeric mice to assess effects of Id-1 on the accumulation of other Ids and on intestinal epithelial homeostasis. Surprisingly, a subset of these mice developed intestinal adenomas.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

Generation of Chimeric Mice-- A DNA fragment derived from nucleotides 4-558 of mouse Id-1 mRNA (5) was produced by reverse transcriptase (RT)-PCR using small intestinal RNA from a 3-month-old C57Bl/6 (B6) animal. Each PCR primer contained a BamHI restriction site at its 5' end (5'-GAATTCGGATCCGTACAACCTTTCTCCAAC-3' and 5'-GAATTCGGATCCATCTGGTCCCTCAGTGC-3'). The PCR product was purified and subcloned into the T/A vector (InVitrogen). Sequence analysis confirmed that no errors had been introduced during the amplification. A 576-base pair BamHI fragment, containing the mouse Id-1 open reading frame and stop codon, was excised from the T/A vector and inserted into pL596hGHpNeoB₂ (42) at its unique BamHI site to produce pLF/Id-1. This resulted in placement of the Id-1 open reading frame at nucleotide +3 of the human growth hormone (hGH) gene, and under the control of nucleotides 596 to +21 of a rat fatty acid-binding protein gene (Fabpl; Refs. 54 and 55). Id-1 contains the first in-frame Met codon in this recombinant DNA so that Id-1, but not hGH, is produced from the mRNA transcript. A pgk-neomyocin resistance selection cassette is located just downstream of hGH in pLF/Id-1.

Preparation of Intestinal Whole Mounts-- Mice were given an intraperitoneal injection of an aqueous solution of 5'-bromo-2'-deoxyuridine (BrdUrd) and 5'-fluoro-2'-deoxyuridine (12 mg and 1.2 mg/g body weight, respectively) in order to label cells in S phase. Animals were sacrificed 90 min later. The entire small intestine was removed en bloc. The proximal third was snap frozen in liquid nitrogen and used as a source of total cellular RNA to verify transgene expression (see below).

Histochemistry-- Segments of X-Gal- or Bluo-Gal-stained intestinal whole mounts were embedded in paraffin, and 5-µm sections were cut and collected on glass slides. The sections were deparaffinized using xylene (one cycle of 10 min followed by two cycles of 5 min each), incubated in isopropyl alcohol (three cycles, 5 min each), and then rehydrated in distilled water (5 min). Typically, 300 serial sections were prepared from both the middle and distal thirds of the small intestine of each animal. These regions are referred to as jejunum and ileum, respectively. Sections were cut parallel to the cephalocaudal axis of the intestine: every 30th section was stained with nuclear fast red or hematoxylin and eosin (60) to assess cellular morphology and crypt-villus architecture. Members of specific cell lineages were visualized in adjacent serial sections using the following histochemical stains (60): (i) Alcian Blue (detects acidic mucins in goblet cells); (ii) phloxine/tartrazine (visualizes the apical secretory granules of Paneth cells); and (iii) periodate acid Schiff (PAS; detects neutral mucins present in goblet cell mucus globules and Paneth cell granules, as well as glycoconjugates associated with the apical brush-border membrane of enterocytes). Intracellular X-Gal and Bluo-Gal precipitates were not removed during these staining procedures, allowing us to distinguish -Gal-positive B6 ROSA26/+ epithelium from -Gal-negative 129/Sv epithelium.

RT-PCR Assays-- Frozen portions of the proximal third of the small intestine from a normal B6 ROSA26/+  129/Sv chimera or a B6 ROSA26/+ 129/Sv(Id-1) chimeric-transgenic mouse were added to RNAzol (Tel-Test; 1 ml of RNAzol/30 mg of tissue) and homogenized with a TissueMizer (Tekmar), and total cellular RNA was isolated according to the manufacturer's protocol. cDNA was synthesized using Avian Myeloblastosis Virus reverse transcriptase (Promega) and oligo(dT)_12-18 primers according to the supplier's protocol. An aliquot of each reaction was then subjected to PCR as follows. RNA transcripts from the Fabpl-Id-1-hGH transgene were amplified using two different sets of primers. In one set, the upstream and downstream primers (5'-TGAACTCGGAGTCTGAAGTC-3' and 5'-TTGGCGAAGACACTCCTGAG-3', respectively) spanned the Id-1/hGH boundary. In the other set, the upstream and downstream primers (5'-CTGCTGCTCATCCAGTCGTG-3' and 5'-GCAGCTAGAAGCCACAGCTG-3', respectively) spanned intron 4 of hGH. PCR conditions were as follows: annealing at 60 °C for 30 s; extension at 72 °C for 2 min; denaturation at 95 °C for 30 s, for a total of 34 cycles. DNAs representing mRNAs derived from endogenous mouse Id and -actin genes were also amplified, using primers and cycling conditions described by Riechmann et al. (9).

Immunohistochemistry-- A panel of previously characterized antibodies was used to define expression of various differentiation markers in each of the four principal small intestinal epithelial cell lineages. In addition, a set of anti-peptide antibodies specific for HLH proteins were used to characterize their cellular and intracellular patterns of accumulation. Staining was performed on paraffin-embedded sections prepared from X-Gal-stained whole mounts. Sections were deparaffinized, rehydrated, placed in blocking buffer (PBS containing 1% bovine serum albumin and 0.05% Tween 20) for at least 30 min at room temperature, and then incubated overnight at 4 °C with one of the following antibodies: (i) rabbit anti-mouse Id-1 (generated against residues 129-148 of the protein; final concentration of 0.7 µg/ml PBS-blocking buffer; Santa Cruz Biotechnology, Inc., Santa Cruz, CA); (ii) rabbit anti-mouse Id-2 (generated against residues 167-186; 0.7 µg/ml; Santa Cruz Biotechnology); (iii) rabbit anti-human Id-3 (residues 98-117; 0.7 µg/ml; Santa Cruz Biotechnology); (iv) rabbit anti-human E12/E47 (residues 632-649; 2 µg/ml; Santa Cruz Biotechnology); (v) goat anti-rat apolipoprotein B (1:500; specific for enterocytes in the small intestine; kindly provided by Patrick Tso, University of Cincinnati School of Medicine, Cincinnati, OH); (vi) goat anti-rat apolipoprotein AIV (1:1000; enterocytes; P. Tso); (vii) rabbit anti-serotonin (1:1000, specific for subpopulation of enteroendocrine cells (61); Incstar); (viii) rabbit anti-cryptdin (1:500, reacts with Paneth cell cryptdins 1, 2, 3, and 6 (49, 52, 62); kindly provided by Michael Selsted, University of California, Irvine); (ix) rabbit antibodies to the secreted phospholipase A₂ encoded by Pla2 g2a (1:40,000; Paneth cells (52, 63, 64); kindly provided by Rita Mulherkar, Cancer Research Institute, Bombay, India)); (x) rabbit anti-human lysozyme (1:1000; Paneth cells (49); Dako); and (xi) goat anti-BrdUrd (1:2000; cells in S-phase (54, 65); kindly provided by Steve Cohn, Washington University).

Lectin Staining-- Glycoconjugate production is a sensitive marker of the differentiation programs of mouse intestinal epithelial cell lineages (57, 66). Therefore, PLP-fixed and X-gal-stained sections of whole mounts were deparaffinized, rehydrated, and stained with horseradish peroxidase-conjugated lectins as described in an earlier report (66). Each lectin was used at a final concentration of 5 µg/ml PBS blocking buffer. Members of the lectin panel included (i) Ulex europaeus agglutinin 1 (Sigma; recognizes Fuc1,2Gal containing glycoconjugates synthesized by Paneth, goblet, and enteroendocrine cells); (ii) Anguilla anguilla agglutinin (EY Labs; reacts with -L-Fuc-containing glycoconjugates produced by Paneth, goblet, and enteroendocrine cells); and (iii) Maackia amurensis agglutinin (Boehringer Mannheim; recognizes Neu5Ac2,3Gal4Glc/GlcNAc-containing glycoconjugates present in Paneth cells, goblet cells, and enterocytes).

Light Microscopy-- Stained sections were viewed with a Zeiss Axiophot microscope and/or a Molecular Dynamics model 2001 confocal microscope. Scans in the confocal microscope were performed at 1-µm intervals.

Electron Microscopy-- Intestinal whole mounts were prepared exactly as described above except that PBS, 2% paraformaldehyde, 0.4% glutaraldehyde was used as the initial fixative. Following incubation with Bluo-Gal, intestines were washed with PBS, and regions containing adjacent 129/Sv and B6 ROSA26/+ patches were isolated by dissection. Pieces of tissue were then post-fixed overnight at 4 °C in 2% paraformaldehyde, 2% glutaraldehyde, washed in PBS, treated for 1 h in 2% osmium tetroxide at room temperature, dehydrated in graded alcohols, and embedded in Poly/Bed 812 (Electron Microscopy Sciences). Semithin (0.5-µm) sections were cut with a glass knife and examined using bright field microscopy until regions containing juxtaposed patches of B6 ROSA26/+ and 129/Sv crypt-villus units were located. Thin sections (0.1 µm) were then prepared and viewed with a JOEL model 100C electron microscope. The Bluo-Gal precipitate was retained in B6 ROSA26/+ epithelial cells after fixation and embedding, allowing the epithelium to be genotyped under the electron microscope.

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

Description of the Chimeric-Transgenic System Used to Assess the Effects of Id-1 on Intestinal Epithelial Homeostasis-- 129/Sv ES cells were stably transfected with a recombinant DNA consisting of Id-1 under the control of nucleotides 596 to +21 of a fatty acid-binding protein gene (Fabpl). These transcriptional regulatory elements are well suited to assess the effects of Id-1 on proliferation, commitment, and differentiation in the small intestinal epithelium. They have been used to express a variety of proteins in the region of crypt containing the multipotent stem cell. Expression is maintained in members of all four of the stem cell's descendant lineages as they proliferate, differentiate, and then complete their life on the villus (e.g. see Refs. 42, 55, 58, 67-69). Fabpl-reporter transgenes are expressed from embryonic day 15 through the period that crypt-villus units form (postnatal weeks 1-3) and during the subsequent 1.5-2 years of life. Highest levels of reporter mRNA and protein occur in the distal half of the small intestine (jejunum/proximal ileum).

View larger version (77K): [in this window] [in a new window]   Fig. 1.   Distal jejunum from a normal control B6 ROSA26/+  129/Sv chimeric mouse. A, whole mount preparation of the jejunum that has been opened, stained with X-gal, and photographed looking down on villi. The B6 ROSA26/+ epithelium expresses -Gal and appears blue after staining with X-gal. The 129/Sv component lacks -Gal and appears white. The arrows point to several representative polyclonal villi. Each polyclonal villus appears striped because it receives columns of -Gal-negative (white) epithelial cells from monoclonal 129/Sv crypts and -Gal-positive cells (blue) from monoclonal B6 ROSA26/+ crypts. B, PLP fixed section prepared from the same whole mount as shown in A but stained subsequently with nuclear fast red. The villus on the left is wholly B6 ROSA26/+; it is only supplied by monoclonal B6 ROSA26/+ crypts that contain an entirely -Gal-positive population of blue epithelial cells. The villus on the right is wholly 129/Sv because it is only supplied by monoclonal 129/Sv crypts composed of entirely -Gal-negative epithelial cells. The villus in the middle is polyclonal; it is supplied by the monoclonal B6 ROSA26/+ crypt (blue) and by the monoclonal 129/Sv crypt. Bar in B, 25 µm.

Id-1 Levels Are Elevated in Transgenic Epithelium-- Seventeen 3-12-month-old B6 ROSA26/+  129/Sv(Id-1) chimeric-transgenic mice were studied. They were produced from two different clones of stably transfected ES cells. Twenty-seven age-matched normal chimeras were generated using nontransfected 129/Sv ES cells and were employed as external controls to identify strain-specific differences between B6 ROSA26/+ and 129/Sv gut epithelium.

View larger version (70K): [in this window] [in a new window]   Fig. 2.   Evidence for expression of the Fabpl-Id-1 transgene. A, RT-PCR analysis of RNAs isolated from the proximal small intestine of a normal chimeric mouse generated using nontransfected ES cells (lane 1) and an age-matched chimeric-transgenic mouse with equivalent 129/Sv contribution (lane 2). The 500-base pair band in lane 2 corresponds to the expected sized product generated from the transgene's mRNA transcript. B and C, a single PLP-fixed section of a polyclonal villus from the jejunum of a B6 ROSA26/+  129/Sv(Id-1) mouse. The section was prepared from an X-gal-stained whole mount preparation of intestine and photographed under bright field in B. C, immunofluorescent image of the section after incubation with affinity-purified rabbit antibodies to Id-1, Cy3-tagged donkey anti-rabbit Ig and the nuclear stain bis-benzimide (dark blue). Levels of immunoreactive Id-1 (magenta) are higher in transgenic 129/Sv(Id-1) epithelial cells compared with their adjacent normal B6 ROSA26/+ neighbors. D, crypt-villus units from an aged matched normal control chimera. The section was stained with the same reagents as in C. The villus on the left is polyclonal. The villus on the right is wholly 129/Sv. Steady state levels of immunoreactive Id-1 in 129/Sv and B6 ROSA26/+ epithelial cells are equivalent, demonstrating that the differences observed in C are attributable to forced expression by the transgene. E, blocking control experiment using a polyclonal villus from the normal chimeric mouse studied in D. The section was incubated with the reagents used in C and D. However, the Id-1 antibody preparation was pretreated with the peptide used for its generation. F, confocal micrograph showing a transgenic 129/Sv(Id-1) crypt and the base of an associated villus. The section was incubated with rabbit anti-Id-1 and Cy3-tagged donkey anti-rabbit Ig. The arrow points to the crypt-villus junction. Id-1 (orange) is limited to the cytoplasm. Nuclei do not contain detectable levels of the protein and therefore appear black. G, higher power view of the boxed region in F. Bars, 25 µm.

Forced Expression of Id-1 Results in Marked Reductions in Cellular Id-2 and Id-3 Levels-- RT-PCR assays of RNA isolated from the proximal and distal halves of the small intestines of normal adult mice revealed that the endogenous Id-2, Id-3, and Id-4 genes are also expressed in this organ (data not shown). Moreover, immunohistochemical studies of normal control chimeras disclosed that Id-2 and Id-3 concentrations increase abruptly as B6 ROSA26/+ and 129/Sv epithelial cells exit the cell cycle and arrive at the crypt-villus junction. Id-2 and Id-3 levels do not change appreciably as cells complete their migration from the base to tips of villi (Fig. 3, A and B, and data not shown). Unlike Id-1, Ids 2 and 3 are located in the nucleus (Fig. 3B). Despite the positive RT-PCR result, we were unable to detect Id-4 in the 129/Sv or B6 intestinal epithelium of normal chimeras using commercially available antibodies and sensitive immunohistochemical detection techniques.

View larger version (59K): [in this window] [in a new window]   Fig. 3.   Forced expression of Id-1 produces a loss of immunoreactive Id-3 but has no detectable effect on E12/47. A and B, sections of a polyclonal villus from the jejunum of a normal chimeric mouse generated using nontransfected ES cells. The section in A was prepared directly from the X-gal-stained whole mount preparation of intestine. The same section is shown in B after incubation with affinity-purified rabbit antibodies to Id-3, Cy3-conjugated donkey anti-rabbit Ig, and bis-benzimide. The solid arrows in B point to the crypt-villusjunction. Id-3 levels rise abruptly as epithelial cells exit the crypt. Levels are sustained as differentiating cells migrate upward to the villus tip. Note that Id-3 is associated with nuclei (e.g. open arrows), and there are no appreciable differences in the cellular concentration or intracellular distribution of this protein between the B6 ROSA26/+ and 129/Sv components of this polyclonal villus. C and D, a polyclonal villus from an age-matched chimeric-transgenic mouse treated with the same reagents as in A and B. The column of 129/Sv(Id-1) epithelial cells has markedly diminished levels of immunoreactive Id-3 (compare with B). E, adjacent B6 ROSA26/+ and 129/Sv jejunal villi from the chimeric-transgenic animal shown in C and D. The section has been stained with affinity-purified rabbit antibodies that recognize the E12 and E47 forms of E2A, Cy3-labeled donkey anti-rabbit Ig, and bis-benzimide. Immunoreactive E12/E47 (magenta) is associated with lymphocytes in the mesenchyme (e.g. closed arrow) and villus epithelial cells. The open arrow highlights the granular pattern of staining observed in the apical cytoplasm of enterocytes. Blocking controls (not shown) revealed that pretreatment of the antibodies with the peptide antigen from which they were generated abolished immunostaining in both the epithelium and in the mesenchyme. Bars, 25 µm.

A Subset of Chimeric-Transgenic Mice Develop Paneth Cell Depletion and Adenomas-- Analysis of sections prepared from the distal two-thirds of the small intestine from 14 of the 17 chimeric-transgenic mice (age 3-10 months; 129/Sv contribution of <1 to 30%) failed to reveal any abnormalities in epithelial homeostasis (external controls were normal, age-matched chimeras). Transmission electron microscopic analysis of 129/Sv crypts indicated that forced expression of Id-1 did not produce detectable morphologic changes in their epithelial cell populations. This includes the region near the crypt base where stem cells reside (Fig. 4, A and B). Forced expression of Id-1 in these 14 mice did not result in any significant quantitative changes in proliferation within jejunal and ileal crypts, whether defined by pulse-labeling mice with BrdUrd 90 min prior to sacrifice and noting the number of S-phase cells/crypt section (Fig. 4C) or by surveying the number of M-phase cells per crypt section after hematoxylin and eosin staining (data not shown). In addition, there were no qualitative changes in proliferation: epithelial cells did not re-enter the cell cycle as they migrated up the villus (Fig. 4C). Paneth cell number and differentiation appeared normal, as judged by transmission electron microscopy (Fig. 4, A and B) and by their reaction with two histochemical stains (phloxine/tartrazine and PAS), three lectins that recognize fucosylated- or Neu5Ac2,3Gal4Glc/GlcNAc-containing glycoconjugates, and three antibodies that recognize gene products produced by this lineage (cryptdins, lysozyme, and a secreted phospholipase A₂) (Fig. 4D). The number and differentiation of goblet cells, as defined by their reaction with three lectins and two histochemical stains, were unaffected (Fig. 4E). The crypt-villus distributions of terminal differentiation markers in enterocytes and enteroendocrine cells were also unperturbed when cellular pools of Id-1 were increased (Fig. 4, F and G, and data not shown). Finally, forced expression of Id-1 did not affect the orderliness of cell migration: (i) polyclonal villi contained coherent vertical columns of wholly B6 ROSA26/+ or wholly 129/Sv cells; (ii) the borders between the cellular columns were well demarcated; and (iii) there was no "wandering" of cells of one genotype into columns of the opposite genotype (Fig. 4H). Orderly migration was also observed in the follicle-associated epithelium overlying Peyer's patches; the submucosal lymphoid aggregates that can serve as inductive sites for initiation of immune responses (Fig. 4H).

View larger version (63K): [in this window] [in a new window]   Fig. 4.   Analysis of proliferation and differentiation in the intestinal epithelium of chimeric-transgenic mice. A and B, transmission electron microscopic study of adjacent B6 ROSA26/+ and 129/Sv(Id-1) jejunal crypts. Sections were prepared from a Bluo-Gal-stained whole mount preparation of intestine. The crypt in A can be identified as B6 ROSA26/+ due to the presence of electron-dense Bluo-Gal precipitates within its epithelial cells (e.g. closed arrows). Paneth cells with their distinctive, prominent apical secretory granules (e.g. open arrow) are evident at the crypt base. B, The 129/Sv crypt has morphologic features similar to those in A except that the average diameter of the Paneth cells' electron-dense secretory granules is larger. The results of morphometric comparisons with age- and gender-matched normal chimeras indicated that this represents a strain-specific difference. C, section prepared from a mouse that had received an intraperitoneal injection of BrdUrd 90 min prior to sacrifice. The section was incubated with goat anti-BrdUrd, Cy3-donkey anti-goat Ig, and bis-benzimide. Cells labeled in S-phase (magenta) are confined to the crypts (e.g. arrows). The number of BrdUrd-positive cells in B6 ROSA26/+ and 129/Sv(Id-1) crypts is similar. D, comparison of cryptdin expression in B6 ROSA26/+ and 129/Sv(Id-1) Paneth cells. The base of a polyclonal villus is shown. The section was incubated with rabbit antibodies that recognize cryptdins 1, 2, 3, and 6; Cy3-donkey anti-rabbit Ig; and bis-benzimide. The number of cryptdin-positive cells (magenta) and the cellular levels of cryptdins are similar in B6-ROSA26/+ crypts (e.g. closed arrow), and 129/Sv crypts (e.g. open arrow). E, goblet cells identified in a polyclonal villus stained with X-Gal and PAS. The number of PAS-positive goblet cells (magenta, e.g. arrows) is similar in the B6 ROSA26/+ component (blue) and in the 129/Sv component of the villus. F and G, apolipoprotein B accumulation in the enterocytes of transgenic 129/Sv(Id-1) jejunal villi (F), and adjacent B6 ROSA26/+ villi (G). The sections were incubated with rabbit anti-apolipoprotein B, Cy3-donkey anti-rabbit Ig, and bis- benzimide. Forced expression of Id-1 does not perturb the levels or pattern of expression (i.e. crypt-villus distribution) of apolipoprotein B (magenta). H, a whole mount preparation of the ileum with a Peyer's patch located in the center of the photograph. A follicle-associated epithelium overlies the submucosal Peyer's patch. The follicle-associated epithelium is polyclonal; it receives cellular inputs from B6 ROSA26/+ crypts (e.g. arrowhead) as well as 129/Sv crypts. Epithelial cells migrate from these crypts to the center of the follicle-associated epithelium, forming wedge-shaped arrays that are wholly 129/Sv or wholly B6 ROSA26/+ (e.g. closed arrow). Some of these crypts also supply cells to adjacent villi. The sharply demarcated borders of these wedges and the sharply demarcated stripes in the polyclonal villi (e.g. open arrow) indicate that forced expression of Id-1 does not perturb the orderliness of cell migration. Bars in A-G, 25 µm; bar in H, 125 µm.

View larger version (98K): [in this window] [in a new window]   Fig. 5.   Adenomas and Paneth cell depletion in chimeric-transgenic mice. A, Bluo-Gal stained whole mount preparation of the distal jejunum. The arrow points to a small cluster of 129/Sv villi. At the center of the photograph is an adenoma composed of -Gal-positive and -negative cells. Thin black bar to the left of the adenoma, 450 µm. B, section prepared from the adenoma shown in A and subsequently stained with nuclear fast red. The adenoma contains -Gal-positive (light blue) and -negative cells. C, X-gal-stained section from an adenoma obtained from another mouse. D, same section shown in C after incubation with goat anti-BrdUrd, Cy3-donkey anti-goat Ig, and bis-benzimide. The animal had received an intraperitoneal injection of BrdUrd 90 min prior to sacrifice to label proliferating cells. Comparison of C and D reveals that -Gal-positive and -negative cells are in S-phase. E, a section from the adenoma has been incubated with affinity-purified rabbit antibodies to Id-3, Cy3-conjugated donkey anti-rabbit Ig, and bis-benzimide. Id-3 is absent from -Gal-positive and -negative cells. F, Paneth cell loss in another mouse containing an adenoma. The section has been stained with X-gal and PAS. The closed arrow points to a B6 ROSA26/+ crypt with Paneth cells at its base. Note the red apical granules in these Paneth cells. The two 129/Sv(Id-1) crypts (open arrows) lack Paneth cells. The inset shows another section stained with rabbit anti-cryptdin, Cy3-donkey anti-rabbit Ig, and bis-benzimide. The B6 ROSA26/+ crypt (closed arrow) contains Paneth cells with immunoreactive cryptdins. The 129/Sv(Id-1) crypts (open arrows) lack cryptdins, consistent with the loss of Paneth cells. Bars in B-F, 25 µm.

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

We have found that the normal adult mouse small intestinal epithelium supports a complex pattern of expression of three Id genes. Cellular Id-1 levels do not change appreciably as the descendants of the multipotent intestinal stem cell proliferate and then undergo terminal differentiation. In contrast, Id-2 and Id-3 increase markedly as postmitotic cells exit the crypt and differentiate. Id-2 and Id-3 are targeted to the nucleus, while Id-1 appears to be restricted to the cytoplasm. The cytoplasmic location of immunoreactive Id-1 and E12/E47 indicates that nuclear import of these transcription factors is normally highly restricted in this epithelium. Transient transfection studies of COS cells indicate that E-proteins function as nuclear chaperons for Ids (75). This raises the possibility that Id-1 lacks a functional chaperon within the cytoplasm of small intestinal epithelial cells and/or that Id-1 and E12/E47 are somehow prevented from interacting with one another within this cellular environment.

Forced expression of Id-1 produces a global change in Id accumulation: Id-1 levels increase, but the protein remains restricted to the cytoplasm while cellular Id-2 and Id-3 levels decline to below the limits of detection. These findings not only disclose the interrelated nature of Id accumulation within this epithelium but also indicate that forced expression of Id-1 in the mouse intestine should be viewed as producing both loss-of-function (abolition of Ids 2 and 3) as well as a "gain" of Id-1.

Modest changes in the activities of bHLH proteins can have a dramatic effect on cellular function. E2A regulation of B cell development provides one such example: a 2-fold reduction in bHLH activity is associated with an equivalent reduction in the likelihood of adopting this cellular fate (73). As noted in the Introduction, forced expression of Id-1 inhibits the differentiation of many cultured cell lines. In contrast, forced expression of Id-1 does not produce a detectable effect on intestinal epithelial cell differentiation in the majority of mice we examined. The cytoplasmic location of both Id-1 and E12/E47 could account for the general insensitivity of small intestinal epithelium to augmented Id-1 levels.

Since the loss of Id-2 and Id-3 was generally not accompanied by detectable perturbations in proliferation, lineage allocation, or differentiation, it appears that neither Id is absolutely required to control these processes or that their loss can be functionally complemented. Moreover, the fact that Id-2 and Id-3 are confined to the villus epithelium further suggests that their role is not to inhibit terminal differentiation of villus-associated lineages.

A recent report mentioned that mice homozygous for a null allele of Id-1 had no phenotype, based on a histologic survey that included the gastrointestinal tract (27). The same group noted that Id-3^/ mice were phenotypically normal, while compound homozygotes were not viable (27). These results led them to conclude that Id-1 and Id-3 have overlapping functions. Based on our findings, it would be interesting to explore whether loss of the one Id resulted in an increase in the other Id(s), thereby maintaining a net level of Id function in these Id-1^/ or Id-3^/ mice. The effects of knocking out Id-2 have not been reported, and thus we do not know the contribution of this Id to establishing and maintaining intestinal epithelial homeostasis.

A subset of our chimeric-transgenic mice exhibited Paneth cell loss in their 129/Sv epithelium plus adenomas. It is not clear why only a subset of mice exhibited these abnormalities or why the long-lived Paneth cell lineage was more responsive than other lineages to the effects of forced expression of Id-1. The Paneth cell depletion/adenoma phenotype is unlikely to be due to insertion of the Fabpl-Id-1 transgene at a particular site within the mouse genome, since it was evident in chimeras generated from two independent, stably transfected ES cell clones. Our analysis of 50 normal control chimeras indicated that they do not spontaneously lose Paneth cells or develop adenomas. Thus, we believe that the presence of this phenotype in 3 of 17 chimeric-transgenic mice with relatively low 129/Sv contribution in their ileum/jejunum (5, 7, and 15%) is significant. If forced expression of Id-1 was sufficient in itself to produce tumorigenesis, we would have expected adenomas to be present in mice with the highest fractional representation of 129/Sv(Id-1) epithelium. Such a correlation was not apparent, although the number of animals with adenomas was small. Thus, our findings suggest that elevated levels of cytoplasmic Id-1 and/or reduced levels of nuclear Id-2 and Id-3 are not sufficient to directly initiate tumorigenesis and that other factors are required.

Transgenic mice with a genetically engineered diphtheria toxin A fragment-mediated ablation of the Paneth cell lineage do not have detectable proliferative abnormalities in their crypts and do not develop intestinal adenomas (52). Although this result indicates that loss of Paneth cells per se is not sufficient to initiate tumorigenesis, the loss of at least one Paneth cell product has been shown to help promote tumorigenesis in mice with a predilection to initiation. Animals heterozygous for a mutant allele of the Adenomatous polyposis coli gene, Apc^Min, develop multiple intestinal adenomas (76). Mom1 is a semidominant modifier of tumor multiplicity in Min/+ animals. Genetic studies indicate that the secreted phospholipase A₂ encoded by Pla2g2a is a very likely candidate for Mom1 (77-80). Two Pla2g2a alleles are known (79). One contains a frameshift mutation that prevents production of the intact protein and is associated with a significantly greater number of intestinal adenomas (Mom-1^S). The other allele, Mom1^R, does not contain the mutation. A transgene that permits forced expression of a functional Pla2 g2a product reduces tumor multiplicity and size in Min/+ mice to the level associated with a single copy of Mom-1^R (80). The mechanism by which this secreted phospholipase A₂ influences tumorigenesis is unclear at present. Studies in intestinal isografts and aggregation chimeras indicate that Apc and Mom1 act locally to affect the (crypt) cell lineage from which adenomas arise (81, 82). B6 ROSA26/+ epithelium contains Mom1^S (79), and immunoreactive enzyme was not detectable in adenomas or in the 129/Sv(Id-1) crypts that lacked Paneth cells. Nonetheless, we have no direct evidence that loss of this phospholipase contributes to tumorigenesis in our chimeric-transgenic animals. However, the combination of (i) loss of this and other Paneth cell products, (ii) the effects of dysregulated Id expression on the crypt epithelium, and (iii) other as yet unspecified (environmental) factors may play a key role in determining whether tumorigenesis can proceed in some crypts.

Each of the solitary adenomas present in B6-ROSA26/+  129/Sv(Id-1) chimeras was composed of proliferating populations of both -Gal-positive and -Gal-negative epithelial cells. One possible explanation for this phenotype is that adenoma formation was initiated in the B6 ROSA26/+ (crypt) epithelium and that at some point during progression there was silencing or deletion of the ROSA26 locus. In other words, these tumors were not composed of B6-ROSA26 and 129/Sv cells but rather were amplified clones of B6-ROSA26^+/ and B6-ROSA26^/ cells. Studies by Dove and co-workers (82, 83) argue against this scenario. They never found mixed populations of -Gal-positive and -negative cells in adenomas arising in nonchimeric ROSA26/+, Min/+ mice or in aggregation chimeras that contain Apc^Min on ROSA26/+ lineages (Ref. 82; cf. Ref. 83). A more likely and intriguing possibility is that these adenomas arise through cooperative interactions between B6 ROSA26/+ and 129/Sv(Id-1) cells. Two recent studies of humans suggest that during early stages of adenoma formation, tumors are polyclonal. Patients with familial adenomatous polyposis inherit a mutant APC allele, predisposing them to develop hundreds to thousands of intestinal adenomas during their first several decades of life (84). Analysis of a XYXO mosaic patient with familial adenomatous polyposis revealed that this individual's small intestine and colon crypts were monoclonal but that three quarters of his microadenomas were polyclonal (85). Bjerknes and co-workers (86) also studied individuals with familial adenomatous polyposis. Immunohistochemical analyses of their crypts and early adenomas using APC antibodies, coupled with microdissection and APC genotyping also revealed polyclonality. Conversion to monoclonality at later stages of tumor progression was attributed to a proliferative advantage conferred on a subclone as a consequence of additional mutations.

Our findings suggest that in the chimeric mouse system used for this study, perturbations in Id gene expression are associated with juxtacrine or paracrine signaling between initiated 129/Sv(Id-1) cells and normal B6 ROSA26/+ cells and that such signaling may facilitate recruitment and subsequent retention of the B6 ROSA26/+ population in an evolving tumor. If this hypothesis is correct, then this chimeric system provides a model for studying the recruitment process. At a minimum, the system provides an opportunity to examine how perturbations in Id expression might affect vulnerability to tumorigenesis.