Examining the role of Paneth cells in the small intestine by lineage ablation in transgenic mice.

The Paneth cell lineage is one of four epithelial lineages derived from the adult mouse small intestine's multipotent stem cell. Mature Paneth cells secrete antimicrobial peptides (cryptdins), growth factors, as well as two gene products, a secreted phospholipase A2 and matrilysin, that has been implicated as modifiers of adenoma formation in mice containing a mutation in the tumor suppressor Apc. Immature Paneth cells are located just above and below the cell layer, in intestinal crypts, that has been proposed to contain the multipotent stem cell. Paneth cells differentiate during a downward migration to the crypt base. The location and direction of Paneth cell migration, their high density and long residency time at the crypt base, and the nature of their secreted gene products, suggest that they may influence the structure and/or function of the stem cell niche. Paneth cell ablation can therefore be viewed as an experimental manipulation of the cellular microenvironment that purportedly contains the stem cell and its immediate descendants. Two types of ablation experiments were performed in transgenic mice. Nucleotides -6500 to +34 of the mouse cryptdin-2 gene (CR2) were used to express an attenuated diphtheria toxin A fragment. Light and electron microscopic immunohistochemical analyses of several pedigrees of postnatal day 28 to 180 animals established that ablation of Paneth cells is accompanied by an increase in the proportion of undifferentiated crypt base columnar cells. These cells normally co-exist with Paneth cells. The ablation does not produce a detectable effect on the proliferation or terminal differentiation programs of the other three lineages or on host-microbial interactions. The last conclusion is based on the ability of crypts to remain free of microbes detectable by Gram and Warthin-Starry stains and by retention of the normal crypt-villus distribution of components of the diffuse gut-associated lymphoid tissue. CR2-directed expression of simian virus 40 large T antigen also results in a loss of mature Paneth cells but produces a marked amplification of crypt cells having a morphology intermediate between Paneth and granule goblet cells. EM immunohistochemical analyses suggest that intermediate cells can differentiate to mature goblet cells but not to Paneth cells, as they migrate up the crypt-villus axis. Our findings suggest that (i) stemness in the crypt is not defined by instructive interactions involving the Paneth cell; (ii) expressing a Paneth cell fate may require that precursors migrate to the crypt base; (iii) antimicrobial factors produced by Paneth cells are not required to prevent colonization of small intestinal crypts; and (iv) this lineage does not function to maintain the asymmetric crypt-villus distribution of components of the diffuse gut-associated lymphoid tissue.

The Paneth cell lineage is one of four epithelial lineages derived from the adult mouse small intestine's multipotent stem cell. Mature Paneth cells secrete antimicrobial peptides (cryptdins), growth factors, as well as two gene products, a secreted phospholipase A 2 and matrilysin, that has been implicated as modifiers of adenoma formation in mice containing a mutation in the tumor suppressor Apc. Immature Paneth cells are located just above and below the cell layer, in intestinal crypts, that has been proposed to contain the multipotent stem cell. Paneth cells differentiate during a downward migration to the crypt base. The location and direction of Paneth cell migration, their high density and long residency time at the crypt base, and the nature of their secreted gene products, suggest that they may influence the structure and/or function of the stem cell niche. Paneth cell ablation can therefore be viewed as an experimental manipulation of the cellular microenvironment that purportedly contains the stem cell and its immediate descendants. Two types of ablation experiments were performed in transgenic mice. Nucleotides ؊6500 to ؉34 of the mouse cryptdin-2 gene (CR2) were used to express an attenuated diphtheria toxin A fragment. Light and electron microscopic immunohistochemical analyses of several pedigrees of postnatal day 28

to 180 animals established that ablation of Paneth cells is accompanied by an increase in the proportion of undifferentiated crypt base columnar cells. These cells normally co-exist with Paneth cells. The ablation does not produce a detectable effect on the proliferation or terminal differentiation programs of the other three lineages or on host-microbial interactions. The last conclusion is based on the ability of crypts to remain free of microbes detectable by Gram and Warthin-Starry stains and by retention of the normal crypt-villus distribution of components of the diffuse gut-associated lymphoid tissue. CR2-directed expression of simian virus 40 large T antigen also results in a loss of mature Paneth cells but produces a marked amplification of crypt cells having a morphology intermediate between Paneth and granule goblet cells. EM immunohistochemical analyses suggest that intermediate cells can differentiate to mature goblet cells but not to Paneth cells, as they migrate up the crypt-villus axis. Our findings suggest that (i) stemness in the crypt is not defined by instructive interactions involving the Paneth cell; (ii) expressing a Pan-
eth cell fate may require that precursors migrate to the crypt base; (iii) antimicrobial factors produced by Paneth cells are not required to prevent colonization of small intestinal crypts; and (iv) this lineage does not function to maintain the asymmetric crypt-villus distribution of components of the diffuse gut-associated lymphoid tissue.
The structural and functional organization of the adult mouse small intestinal epithelium lends itself to studying both the regulation and integration of cellular proliferation, differentiation, and death programs. The epithelium contains four principal cell types: absorptive enterocytes (comprising Ͼ80% of the total population), enteroendocrine cells, mucus-producing goblet cells, and Paneth cells. All four lineages are derived from a multipotent stem cell that is functionally anchored near the base of each of the small intestine's 1.1 million crypts of Lieberkü hn (1)(2)(3)(4). Cell division is confined to these crypts (5). Enterocytes, enteroendocrine, and goblet cells migrate out of the crypt and up an adjacent villus. Migration is highly ordered and associated with terminal differentiation. Cell death occurs near the villus tip where cells are exfoliated into the lumen (6,7). Proliferation, differentiation, and death take place in a spatially well-organized continuum that extends from the crypt to the apex of a villus. This sequence is completed rapidly (2)(3)(4)(5) days in the case of enterocytes, enteroendocrine, and goblet cells; Refs. 1 and 8 -10) and is recapitulated throughout the lifespan of the mouse.
The Paneth cell lineage differs from the others in a number of notable ways. It is the only lineage that executes its terminal differentiation program during a downward migration from the stem cell zone to the crypt base (11). It is the longest lived lineage, and the only one that exists entirely within the proliferative compartment. Each crypt contains 30 -50 mature Paneth cells that survive for 18 -23 days before degenerating and undergoing phagocytosis by their neighbors (11)(12)(13). Paneth cell age correlates with position in the crypt; the most mature cells are located at or near the crypt base (2). The size of the Paneth cell's apical secretory granules also correlates with age; larger granules are produced by older cells (2,11).
The function of the Paneth cell has not yet been clearly defined. Residency at the crypt base places this lineage in a position to release products from its apical granules that could affect establishment and/or maintenance of the stem cell's niche or influence the properties of the stem cell's descendants. A number of factors exported by Paneth cells could regulate epithelial proliferation and differentiation programs. They include tumor necrosis factor-␣ (14), guanylin (15), and epidermal growth factor (16). Two Paneth cell products have been implicated as modifiers of adenoma formation in mice heterozygous for a mutation in the adenomatous polyposis coli gene, Apc Min (17). Production of matrilysin, a matrix metalloprotein-ase, is limited to the Paneth cell lineage in the adult mouse intestine (18). The protein is expressed in a high percentage of early stage human colorectal neoplasms and Min adenomas (19). Min/ϩ mice homozygous for a null allele of the matrilysin gene have 60% fewer adenomas than animals with the wild type allele, suggesting that the enzyme functions as suppressor of tumorigenesis (19). Pla2 g2a encodes a phospholipase A 2 that is secreted from Paneth cells (20 -22). This gene is a strong candidate for Mom1, a semi-dominant modifier of adenoma size and multiplicity in Min/ϩ animals (23)(24)(25).
Paneth cells also export lysozyme (26,27) and a family of defensin-related anti-microbial peptides known as cryptdins (28). The intestine contains a complex microflora. Components of this microflora are able to establish stable niches at particular positions along the duodenal-ileal axis (29). The fact that different cryptdins exhibit distinct developmental and spatial patterns of expression along this axis (30) suggests that Paneth cells could play a role in modulating the composition of the microbiota or contribute to mucosal barrier functions.
We have examined the contribution of this lineage to epithelial and microbial homeostasis by generating two types of transgenic mice in which mature Paneth cells have been eliminated.

EXPERIMENTAL PROCEDURES
Construction of Transgenes-A 2.7-kb 1 DNA fragment containing simian virus 40 large T antigen (SV40 TAg) was excised from pIF-TAg-hGH (31) with BamHI and subcloned into the BamHI site of pCR-H1 (32). This yielded pCR2-TAg which contained SV40 TAg under the control of nucleotides Ϫ6500 to ϩ34 of the mouse cryptdin-2 gene (33).
Complete digestion of pIF-TAg-hGH with BamHI, followed by partial digestion with EcoRI, allowed purification of a DNA fragment containing pBluescript SK ϩ (Stratagene) with nucleotides ϩ3 to ϩ2150 of the human growth hormone gene (hGH; Ref. 34). This fragment and a BamHI/EcoRI fragment containing cryptdin-2 Ϫ6500 to ϩ34 and SV40 TAg from pCR2-TAg were ligated together, producing pCR2-TAg-hGH. pCR2-TAg-hGH was subsequently cut with BamHI, treated with Klenow, and ligated to a 0.6-kb HincII fragment containing an attenuated diphtheria toxin A fragment (tox176; Ref. 35). The resulting plasmid, pCR2-tox176-hGH, contained tox176 immediately downstream of cryptdin-2 Ϫ6500 to ϩ34 and immediately upstream of hGH ϩ3 to ϩ2150 . tox176 was placed in exon1 of hGH to enhance the chances of efficiently expressing the toxin. hGH will not be produced from the RNA transcript of cryptdin-2 Ϫ6500 to ϩ34 -tox176-hGH ϩ3 to ϩ2150 : the initiator Met and the first stop codon are from the open reading frame of tox176, and there is no ribosomal re-entry site to re-initiate translation at the downstream initiator ATG of hGH.
Four CR2-hGH founders were identified from 38 live born mice, 2 CR2-tox176 founders from 67 mice, and 10 CR2-TAg founders from 87 animals. Pedigrees were established from each of the CR2-hGH and CR2-tox176 founders and from 8 of the CR-TAg founders. All pedigrees were maintained by crosses to normal FVB/N littermates. Pedigree 61, containing CR2-hGH, has been described in an earlier publication (32).
Maintenance of Animals-Mice were housed in microisolator cages under a strictly controlled light cycle (lights on at 0600 h and off at 1800 h) and given a standard irradiated chow diet ad libitum (Pico rodent chow 20, Purina Mills). Routine screens for hepatitis, minute, lymphocytic choriomeningitis, ectromelia, polyoma, sendai, pneumonia, and MAD viruses, enteric bacterial pathogens, and parasites were negative. Specific pathogen-free transgenic animals and their nontransgenic littermates were sacrificed between postnatal days 28 (P28) and P180.
Histochemical Stains-Immediately after sacrifice, the small intestine was removed en bloc, flushed with ice-cold phosphate-buffered saline (PBS), fixed in 10% buffered formalin (Fisher) for 4 -6 h, and then washed in 70% ethanol overnight at room temperature. The intestine was embedded in plastic (JB-4 Embedding Kit, Polysciences), and 1-2-m thick sections ("thin sections") cut from its proximal, middle, and distal thirds (these segments were arbitrarily designated duodenum, jejunum, and ileum, respectively). Alternatively, after washing in 70% ethanol, the intestine was cut open along its duodenal-ileal axis, rolled into a circle, and held in this circular configuration by mounting agar (2% agar (Sigma) in 5% buffered formalin). Each of the resulting "Swiss rolls" was then placed in a tissue cassette, embedded in paraffin, and 5 m-thick serial sections were prepared. Plastic-or paraffinembedded sections were stained with hematoxylin and eosin, phloxine/ tartrazine, or with Alcian blue and periodic acid Schiff (PAS) using standard protocols (37).
Goblet cells were quantitated by counting Alcian blue/PAS-positive cells in all well-oriented jejunal crypt-villus units present in at least two non-adjacent sections cut from Swiss rolls (sections were prepared from three P28 transgenic animals and three normal littermates per pedigree). Paneth cells were likewise quantitated by counting phloxine/ tartrazine-positive cells in jejunal crypts.
To analyze the distribution of components of the microflora along the crypt-villus units of specific pathogen-free transgenic animals and their normal littermates, mice from the various pedigrees were sacrificed at P28, P42, and P120 -P180. Their small intestines were fixed 4 -6 h in 10% buffered formalin without prior flushing and then cut into 1-2-cm segments. Each segment was embedded in paraffin, 4 -6-m thick sections were cut, and the sections treated with Warthin-Starry or Gram stains (37).
Two controls were performed for each experiment employing a given primary antiserum: (i) direct amplification of EPX alone (see above) and (ii) omission of primary antibodies. The latter control involved direct amplification of EPX followed by application of an HRP-conjugated secondary antibody and subsequent indirect tyramide signal amplification with biotinyl tyramide and Cy3-streptavidin. Alternatively, when biotinylated secondary antibodies were employed, there would be a control to establish whether there was any labeling of endogenous biotin. This control consisted of direct tyramide amplification of EPX with FITC-tyramide followed by HRP-streptavidin but without addition of the biotinylated secondary antibodies or the primary antibodies. (Note that adding Cy3-streptavidin alone to jejunal sections did not produce any cellular staining.) Light Microscopy-A Molecular Dynamics Multiprobe 2001 inverted confocal microscope system was used to scan sections subjected to single and/or multi-label immunohistochemistry. Sections were also viewed and photographed using a Zeiss Axioscope.
Identification of Apoptotic Cells-Apoptotic cells were scored in adjacent sections of Swiss rolls, prepared from normal, CR2-tox176, and CR2-TAg mice, using the terminal deoxynucleotidyltransferase-mediated, dUTP nick end-labeling assay, and by their morphologic appear-ance after staining with hematoxylin and eosin (6,7,44). Incorporation of digoxigenin-labeled dUTP was detected using peroxidase-conjugated sheep anti-digoxigenin Fab fragments (Boehringer Mannheim, diluted 1:500 in PBS-blocking buffer) and the Vector VIP kit (Vector Laboratories). Sections were counterstained with methyl green (Zymed).
EM Morphologic Analysis-Three-mm 2 fragments were obtained from the distal jejunum of CR2-tox176 and CR2-TAg transgenic mice plus their age-matched littermates (n ϭ two P28 animals/pedigree). (Note that "distal jejunum" was defined as two-thirds of the distance from the gastro-duodenal junction to the ileal-cecal junction.) The fragments were then fixed for 6 h at 4°C in 2% paraformaldehyde, 2% glutaraldehyde (prepared in PBS), washed in PBS, post-fixed for 1 h in 2% osmium tetroxide, and stained with a solution containing aqueous uranyl acetate and lead. Samples were dehydrated in graded alcohols and embedded in Poly/Bed 812 (Electron Microscopy Sciences). One hundred nanometer-thick sections were prepared and viewed with a JOEL model 100C electron microscope.

General Comments About Paneth, Goblet, and Intermediate
Cells-Paneth cells are distributed along the length of the duodenal-ileal axis in adult (postnatal day 28 to 180) FVB/N mice (32). They can be recognized based on staining of their characteristic apical secretory granules with tartrazine, by their reaction with antibodies directed against lysozyme, the secreted phospholipase encoded by Pla2g2a, and cryptdins, as well as by their production of fucosylated glycoconjugates detected by the lectins Urex europaeus agglutinin type 1 (UEA1), peanut (Arachis hypogaea) agglutinin (PNA), and Dolichos biflorus agglutinin (DBA) (32,42).
Approximately 10% of goblet cells in the normal adult mouse intestine contain very small electron dense cores within their mucin granules. As these "granule goblet cells" migrate up villi and differentiate, they secrete their dense core mucin granules which are then replaced by "common" mucin granules that lack electron-dense cores (8). A rare cell type has been observed in normal small intestinal crypts. Its granules contain electrondense cores that are intermediate in size between those in granule goblet cells and those in the apical granules of young Paneth cells (Fig. 1, A-C). The granules of these rare cells also contain small amounts of mucin. Because of their morphologic features, they have been termed "intermediate" (13), "granulomucous" (46), or "transitional" (47,48) cells. These cells have been proposed to be Paneth cells undergoing transformation to goblet cells, goblet cells in the process of being converted to Paneth cells, or a precursor of both lineages.
Nucleotides Ϫ6500 to ϩ34 of the Mouse Cryptdin-2 Gene Are Active in the Paneth, Granule Goblet, and Intermediate Cells of Adult Transgenic Mice-Light microscopic surveys of adult FVB/N small intestine disclosed that ϳ95% of all crypts present in a cross-section contain cryptdin-positive cells (average ϭ 3 cells/duodenal crypt section; 5 cells/ileal crypt section). EM immunohistochemical analysis using polyclonal antibodies that recognize several members of the cryptdin family revealed that cryptdins are present in the dense core granules of granule goblet cells, intermediate cells, and Paneth cells (Fig. 1, E and  F). The EM study also indicated that these cryptdins are not expressed in any other intestinal epithelial cell type, including mature goblet cells (data not shown).
We subsequently used light and EM immunohistochemistry to define the small intestinal patterns of expression of a human growth hormone reporter in several pedigrees of P28 to P180 transgenic mice containing a cryptdin-2 Ϫ6500 to ϩ34 /human growth hormone fusion gene (CR2-hGH). The results established that nucleotides Ϫ6500 to ϩ34 of the mouse cryptdin-2 gene are active in Paneth, granule goblet, and intermediate cells (e.g. Fig. 1D) and silent in all other epithelial cell types present in crypt-villus units.
Cryptdin-2 Ϫ6500 to ϩ34 -directed Expression of an Attenuated Diphtheria Toxin A Fragment Results in Paneth Cell Ablation-Fragment A of diphtheria toxin (DT-A) ADP-ribosylates and inactivates elongation factor 2, causing inhibition of protein synthesis and cell death. tox176 is an attenuated DT-A allele with a Gly 128 3 Asp substitution that reduces its potency 15-30-fold (35). The attenuated toxin was chosen for Paneth cell ablation in transgenic mice because it is less likely than wild type DT-A to cause death if expressed at low basal levels in nontarget cell populations (49).
Two pedigrees of FVB/N cryptdin-2 Ϫ6500 to ϩ34 /tox-176 (CR2-tox176) mice were produced. There were no statistically significant differences between the growth rates and adult body weights of CR2-tox176 mice and their normal littermates. Comparably aged members of each pedigree had identical phenotypes (see below). A 95% reduction in the number of Paneth cells was evident throughout the length of the small intestines of transgenic animals by P28, whether defined by a loss of staining of serial sections with tartrazine (Fig. 2, A-C), a loss of cellular reactivity with the lectins UEA-1, PNA, and DBA, or the failure of antibodies to detect cryptdins, lysozyme, and Pla2g2a (phospholipase A 2 ) in crypt-villus units (e.g. Fig. 3, A and B). The ablation of Paneth cells was verified using transmission EM (Fig. 4A).
The reduction in Paneth cell number within the small intestines of CR2-tox176 mice persists from P28 through at least P180, although the magnitude of the reduction is less at later time points (e.g. 82% at P42). Granule goblet and intermediate cells can only be defined using EM methods that are not useful for broad surveys of rare cell populations. Nonetheless, EM studies indicated that these cells were reduced in number, but not as markedly as Paneth cells.
The space normally occupied at the crypt base by Paneth cells was occupied in CR2-tox176 mice by "crypt base columnar cells" (Fig. 4, A-C). Crypt base columnar cells are normally interspersed among Paneth cells and constitute 60 -70% of the cells that populate the bottom three cell layers of duodenal, jejunal, and ileal crypts (9). Previous [ 3 H]thymidine labeling/EM radioautography studies indicated that: (i) their residence time at the crypt base after entering S-phase is just a few hours; (ii) they migrate up and out of duodenal, jejunal, and ileal crypts within 3-4 days; and (iii) they differentiate into enterocytes (9). The morphologic features of crypt base columnar cells present in CR2-tox176 transgenic mice and in their normal littermates were indistinguishable. In each case, they resembled the undifferentiated, proliferating transit cell population located in the mid-crypt (see Fig. 4C and legend).
To define the effects of Paneth cell ablation on epithelial cell proliferation, P28, P42, and P120 -180 CR2-tox176 mice and their age-matched normal littermates were pulse-labeled with BrdU 1.5 h prior to sacrifice. The number of S-phase cells was counted in sections of distal jejunal crypts (n ϭ 3 mice/pedigree/ time point). There were no statistically significant differences between the total number of S-phase cells in the middle and upper thirds of normal and CR2-tox176 transgenic crypts. Since the fractional representation of crypt base columnar cells was markedly increased in transgenic mice, it was not surprising that there was a greater number of BrdU ϩ cells in the lower third of transgenic crypts (Fig. 3B).
Surveys of sections of duodenal, jejunal, and ileal villi prepared from normal and CR2-tox176 mice that had received BrdU 1.5 h before sacrifice failed to reveal villus epithelial cells in S-phase (Fig. 3B). Transgenic mice and their normal littermates were subsequently sacrificed at later time points after receiving BrdU. There were no differences in the rate of clearance of BrdU-positive cells from normal or transgenic crypts; by 24 h, BrdU ϩ cells were no longer present in the lower two-thirds of duodenal, jejunal, or ileal crypts; by 36 h, crypts were free of these labeled cells; and by 48 h and 72 h, the location of the trailing edge of BrdU ϩ cells on duodenal, jejunal, or ileal villi was similar in normal and transgenic animals (e.g. Fig. 3, C and D; (n ϭ 2-6 animals/time point/pedigree)). These results indicate that the columnar cells that populate the base of transgenic crypts are not abnormally retained at this location in the absence of Paneth cells. The data are also consistent with the notion that the production and subsequent upward migration of crypt epithelial cells are not grossly perturbed by the lineage ablation.
Loss of Paneth cells had no demonstrable qualitative or quantitative effects on the other three small intestinal epithelial cell lineages. A comparison of CR2-tox176 mice and their age-matched normal littermates did not reveal any statistically significant differences in the number of their Alcian blue/PASpositive goblet cells per duodenal, jejunal, or ileal villus section (e.g. Fig. 5, A-C). The goblet cell lineage normally exhibits complex variations in its pattern of glycoconjugate production along the crypt-villus and duodenal-ileal axes. These variations are a sensitive marker of the lineage's differentiation program and are definable with a panel of lectins and in situ histochemical assays (42). Histochemical surveys using three of these lectins (UEA1, DBA, and PNA) indicated that terminal differentiation of goblet cells is unaffected by loss of Paneth cells (Fig. 3A, B plus data not shown). Based on results obtained with antibodies to chromogranin A and serotonin plus the three lectins, we concluded that there were no significant changes in enteroendocrine cell number or differentiation. The enterocytic lineage also appears unaffected, as judged by the number and distribution of villus epithelial cells that react with antibodies to intestinal fatty acid binding protein and with lectins that recognize fucosylated glyconjugates (data not shown).

Loss of Paneth Cells Does Not Have Any Appreciable Effect on the Crypt-Villus Distribution of Components of the Microflora-
Paneth cells are not present in adult FVB/N colonic crypts. Bacterial density increases along the duodenal-colonic axis of mice and man (50). The colonic crypts of many mammals, including mice, are colonized by bacteria. Microbial colonization is rarely seen in the small intestinal crypts of healthy animals. The secretion of cryptdins and lysozyme from the apical secretory apparatus of Paneth cells may help prevent colonization.
We used a prominent and easily detectable member of the normal FVB/N intestinal microflora to determine whether loss of Paneth cells affected the distribution of bacteria along the crypt-villus and duodenal-ileal axes. A segmented filamentous bacterium (SFB), thought to belong to the Gram-positive Clostridia (51), colonizes the normal small intestine after weaning (52) and reaches very high densities in the ileum of adult animals. The SFB cannot be recovered from the intestine and cultured in vitro (53,54), but can be easily seen in the ileum of normal FVB/N mice using the Warthin-Starry silver stain. SFB adheres to ileal enterocytes located in the upper half of the villus (Fig. 6, A and B). It is not present in crypts (Fig. 6A). Unperfused small intestines from P28 CR2-tox176 transgenic mice and their normal littermates (housed in the same microisolator cages) were subdivided into 1-2-cm segments along the length of the duodenal-ileal axis. Warthin-Starry stains of sections prepared from each segment indicated that loss of Paneth cells had no effect on the duodenal-ileal or crypt-villus distribution of SFB (Fig. 6, A and C). Gram stains provided independent confirmation that duodenal, jejunal, and ileal crypts of P28-P180 CR2-tox176 mice were free of detectable microbes (data not shown).

Ablation of Mature Paneth Cells Is not Associated with a Change in the Crypt-Villus Distribution of Components of the Diffuse Gut-associated Lymphoid Tissue (GALT)-Luminal antigens and microbes are delivered by M-cells to submucosal
lymphoid tissues (e.g. Peyer's patches). These components of the organized GALT can serve as inductive sites for initiation of immune responses (56). Analysis of intestines harvested from specific pathogen-free P28-P180 transgenic mice and their normal littermates indicated that loss of Paneth cells is not accompanied by changes in the size, number, distribution, or histochemical features of Peyer's patches (data not shown).
Complex, dynamic, and often subtle interactions occur between the gut epithelium and components of the diffuse GALT. For example, mice homozygous for a null allele of the ␥ subunit gene of the T-cell receptor (TCR) lack ␥␦ T-cells and exhibit a reduction in crypt cellularity as well as a reduction in epithelial cell migration rates up the villus. Mice homozygous for a null allele of the ␤ subunit gene of the TCR lack ␣␤ T-cells but do not manifest these abnormalities (57). The normal distribution of components of the diffuse GALT along the crypt-villus axis has not been extensively characterized because of the difficulty in identifying cellular markers with conventional immunohistochemical detection methods. We used a protocol employing tyramide signal amplification to identify these components in serial sections of jejunum prepared from age-matched FVB/N CR2-tox176 mice and their normal nontransgenic cagemates.
In normal P42 animals with a conventional microflora (n ϭ 3), CD4 ϩ T-cells are largely confined to the lamina propria and are present throughout the length of the crypt-villus axis (Fig.  7B). CD8 ϩ intraepithelial T-cells have been postulated to serve cytotoxic as well as immunosuppressive functions (58). In normal P42 animals, CD8 ϩ T-cells, unlike CD4 ϩ T-cells, are predominantly intraepithelial and restricted to the villus (Fig.  7D). ␣␤ T-cells are distributed along the length of the cryptvillus axis and populate the intraepithelial and lamina propria compartments (Fig. 7F). In contrast, ␥␦ T-cells are limited to the villus where they are predominantly located within the epithelium (Fig. 7H). B-cells, defined using CD45R/B220 as a marker, are confined to the lamina propria and are distributed from the base of the crypt to the villus tip (data not shown). These results are summarized in Fig. 7A.
Comparable immunohistochemical surveys of P42 CR2-tox176 mice (n ϭ 3) indicated that ablation of Paneth cells was not associated with any detectable alteration in the distribution of these components of the diffuse GALT (Fig. 7, C, E, G,  and H). Moreover, hematoxylin and eosin stains of sections prepared along the length of P28-P180 CR2-tox176 small intestines failed to disclose any evidence of acute or chronic inflammatory changes (n ϭ 40 animals). Together, these results suggest that Paneth cells do not have a direct or indirect organizing function for the diffuse GALT. Furthermore, they are consistent with the notion that loss of Paneth cells does not produce marked perturbations in host-microbial interactions in pathogen-free mice.
Cryptdin lineages has been exploited to generate transgenic mouse models of specific physiologic deficiency syndromes. Physiologic deficiencies result because differentiation of these progenitors is blocked; the "entrapped" progenitors do not have the functional capacities of their terminally differentiated descendants and therefore cannot compensate for their loss (59). SV40 TAgstimulated amplification of normally rare progenitors also provides an opportunity to study their intrinsic properties and/or the consequences of their increased representation (e.g. Refs. 59 and 60).
With these thoughts in mind, nucleotides Ϫ6500 to ϩ34 were used to direct expression of SV40 TAg in FVB/N transgenic animals. Three lines of CR2-TAg mice were analyzed from P28 to P180. All pedigrees had identical intestinal phenotypes. The growth rates and adult body weights of transgenic mice were not significantly different from those of their normal littermates.
Staining with tartrazine, UEA1, and antibodies to cryptdins, lysozyme, and enhancing factor revealed a decrease in the number of mature Paneth cells in CR2-TAg mice comparable to the decrease observed in similarly aged CR2-tox176 animals (e.g. 90 -95% at P28) (Figs. 2D and 3F). EM confirmed the Paneth cell ablation (Fig. 8A). The few mature Paneth cells observed in scattered duodenal, jejunal, and ileal crypts did not contain detectable levels of SV40 TAg (Fig. 3E). The mouse cryptdin gene family includes at least 17 members whose ex-pression varies both as a function of developmental stage and cellular position along the crypt-villus and duodenal-ileal axes (30). Although we do not have antibodies specific for cryptdin-2, it seems likely that this small population of residual mature SV40 TAg-negative Paneth cells was able to complete its terminal differentiation because its members do not support expression of the endogenous cryptdin-2 gene or transgenes under the control of cryptdin-2 Ϫ6500 to ϩ34 . SV40 TAg-positive epithelial cells were distributed along the length of the crypt-villus axis. These cells were most abundant in the crypts and lower half of the villus. SV40 TAg levels decreased as cells moved to the upper half of the villus (Fig. 3F). SV40 TAg-positive villus epithelial cells were also UEA-1-positive (Fig. 3F). The UEA-1/SV40 TAg-positive cells were members of the goblet cell lineage. Unlike CR2-tox176 mice, CR2-TAg animals exhibit a statistically significant 2-3-fold increase in the number of Alcian blue/PAS-positive goblet cells per duodenal, jejunal, or ileal villus section (p Ͻ 0.05; reference control ϭ age-matched normal littermates) (Fig. 5, D and E). Pulse labeling with BrdU 1.5 h prior to sacrifice revealed that production of SV40 TAg is associated with re-entry of these villus goblet cells into S-phase (Fig. 3F).
EM immunohistochemical studies provided further insights about the origins of this amplified goblet cell population. Analyses of distal jejunal crypts from CR2-TAg mice and their normal littermates disclosed a marked amplification of crypt- din-and phospholipase A 2 (Pla2g2a)-positive intermediate and granule goblet cells in transgenic crypts (Fig. 8, A-G). Intermediate cells were found in the lower two-thirds of these crypts. The diameter of the granule's electron-dense core diminishes, and the relative area occupied by its mucin increases as cryptdin-and phospholipase A 2 -positive cells occupy the upper regions of transgenic crypts, resulting in an amplified population of granule goblet cells positioned in the upper crypt/ lower villus (Fig. 8, A-G). SV40 TAg-and UEA1-positive goblet cells in the upper half of the villus lack dense core granules, do not contain the secreted phospholipase A or cryptdin, and have the morphologic, histochemical, and immunohistochemical features of normal mature common goblet cells (data not shown). These results suggest a sequence of "differentiation" involving transformation from intermediate to granule goblet to "mature" common goblet cell.
Even though the ablation of mature Paneth cells by CR2-TAg was accompanied by an amplification of intermediate and granule goblet cells, the crypt base in these mice resembled the crypt base in CR2-tox176 animals, i.e. the position normally occupied by mature Paneth cells was partially filled by crypt base columnar cells (Fig. 8A). These crypt base columnar cells lack SV40 TAg but incorporate BrdU (Fig. 3F).
Other Phenotypic Considerations-Macro-and microscopic surveys of the intestines of P28 -P180 mice from the three CR2-TAg pedigrees indicated that expression of this viral oncoprotein did not lead to the development of intestinal neoplasms (n ϭ 40 animals). The increased levels of apoptosis noted in the crypts of CR2-TAg mice (Figs. 3D and 8A) suggest that some differentiating Paneth cells could be cleared by this mechanism, induced as a consequence of their SV40 TAginduced proliferation. There is precedent for this elsewhere in the crypt-villus axis; forced expression of SV40 TAg in postmitotic FVB/N villus enterocytes using a different promoter results in their pRB-dependent re-entry into the cell cycle and the induction of a p53-independent apoptosis (61). Augmented apoptosis of SV40 TAg ϩ Paneth cells may explain, in part, why CR2-TAg mice do not develop intestinal neoplasms (see "Discussion").
As with CR2-tox176 animals, there were no apparent perturbations in host-microbial interactions. Gram and Warthin-Starry stains revealed no detectable crypt colonization and no change in the spatial distribution of segmented filamentous bacteria compared with their normal littermates or to CR2-tox176 animals (data not shown). CR2-TAg animals do not develop evidence of gut mucosal inflammation (n ϭ 40 mice, 28 -180 days old).

Paneth Cells Do Not Appear to Be Necessary to Establish and Maintain a Functional Stem Cell
Niche-The precise location of the multipotent stem cell in the adult mouse small intestinal crypt has not been established. However, tritiated thymidine labeling/radioautographic analyses of cell proliferation, movement, and differentiation programs have led to speculation that stem cells are positioned in the fifth cell stratum from the crypt base (1)(2)(3)5).
Immature Paneth cells are located just above and below the presumptive stem cell niche. They differentiate during a downward migration to the crypt base where they comprise ϳ50% of the cell population in the first through fourth cell layers (9). The location and direction of Paneth cell migration, the concomitant acquisition of the ability to produce growth factors, their potential for providing instructions to neighboring cells during their downward descent, and their high density and the long residency time at the crypt base raise the possibility that Paneth cells may influence the structure and/or function of the stem cell niche. Paneth cell ablation by an attenuated diphtheria toxin A fragment represents the first reported experimental manipulation of the cellular microenvironment that purportedly contains the stem cell and its immediate descendants.
The gut provides a unique system in which to perform these types of experimental manipulations because the stem cell is contained in a readily detectable anatomic unit, i.e. the crypt. Furthermore, each adult mouse crypt contains a monoclonal population of cells (62)(63)(64). The stem cell's descendants undergo 4 -6 rounds of rapid cell division, generating a steady state population of ϳ250 crypt epithelial cells of which ϳ150 are cycling at any given moment (5). Thus, perturbations that change the biological properties of the stem cell and its immediate descendants may be inferred by noting changes in the behavior and/or composition of their amplified progeny.
The results of tox176-mediated ablation of Paneth cells suggest that this lineage is not essential for establishment of a functional stem cell niche, at least based on the observations that (i) proliferative activity is maintained in crypts over the first 6 months of life and (ii) the composition of terminally differentiated members of the intestine's other self-renewing epithelial lineages is not perturbed. Further analysis of the longer term effects of Paneth cell loss in this model is limited by the gradual re-appearance of members of this lineage. This could reflect time-dependent changes in CR2-mediated expres- . Sections were treated with rabbit anti-cryptdin and gold-labeled goat anti-rabbit Ig. Note how the diameter of the electron dense cores diminishes as cells move up the crypt-villus unit. As long as these cores are present, cryptdin and enhancing factor are detectable. Control experiments using non-immune serum gave no signal (data not shown). Villus goblet cells with common mucin globules that lack electron dense cores do not contain detectable levels of the phospholipase A 2 or cryptdin (data not shown). Bars ϭ 1 m. sion of tox176. If the level rather than the generality of CR2-tox176 expression in Paneth cells is the cause of this phenomenon, then using wild type DT-A may produce a longer lasting ablation.
Changes at the crypt base in CR2-tox176 animals include loss of mature Paneth cells and an increase in the proportion of crypt base columnar cells. Therefore, we can also conclude that the augmented representation of crypt base columnar cells does not have a demonstrable effect on the functional properties of the gut stem cell.
We do not know whether ablation of Paneth cells is associated with a change in the physical location of stem cells in the crypt or whether that location is fixed even in normal animals. Although the properties of "stemness" in the crypt may not be defined by instructive interactions involving neighboring Paneth cells, signals derived from other neighboring epithelial cell populations and/or components of the underlying extracellular matrix/mesenchyme may play critical roles.
Directional Migration and the Differentiation of Paneth and Goblet Cell Lineages-The first through fourth cell strata in normal adult mouse crypts also contain a small percentage (1-2%) of oligomucous and intermediate cells (9). A relationship between Paneth and goblet cell differentiation has been suggested but never proven. The morphologic features of intermediate cells had suggested to others that they may be very young granule goblet cells or a precursor of differentiated Paneth and granule goblet cells (13). To explore this possible precursor-product relationship, various groups have attempted to detect lysozyme in intermediate cell granules by light microscopic immunohistochemistry. Some workers have reported success (65), others failure (66,67). Because of the high background of nonspecific cellular staining observed when commercially available antibodies to lysozyme were used for EM immunohistochemistry, we have also been unable to determine whether intermediate cells contain lysozyme.
Although lysozyme has long been considered part of the immunohistochemical definition of a Paneth cell, cryptdins and the secreted phospholipase A 2 encoded by Pla2g2a have been recently identified as markers of this lineage. We have now demonstrated that cryptdins are produced in the intermediate and granule goblet cells of normal adult FVB/N small intestinal crypts. In addition, cells with morphologic and immunohistochemical features of the ordinarily rare crypt intermediate cells abound in CR2-TAg crypts. These intermediate cells produce SV40 TAg, proliferate, and support expression of cryptdins as well as the phospholipase A 2 . As these SV40 TAg-positive cells emerge from the crypt and move up the villus, they undergo a sequence of cellular alterations that includes a decrease in the size and number of their electron dense granules and an increase in their mucin content. Immunoreactive cryptdins and the phospholipase A 2 are retained until these granules are lost and cellular SV40 TAg expression is suppressed in mature or common goblet cells.
These observations suggest that SV40 TAg is expressed in an intermediate cell precursor with shared features of immature Paneth and goblet cells. Terminal differentiation of this or other precursors into Paneth cells appears to be blocked throughout the crypt. In contrast, the upper crypt and lower half of the villus appear capable of supporting proliferation, survival, and differentiation of the intermediate cell to granule goblet and then mature common goblet cells but not to Paneth cells.
These contrasting responses of the Paneth and goblet cell lineages to SV40 TAg expression may reflect the effects of the microenvironment along the crypt-villus axis. Adopting and expressing a Paneth cell fate may require that precursors mi-grate to the crypt base rather than to the villus. Such a notion is consistent with a recent study in which epithelial cell migration from the crypt to the villus tip was slowed in transgenic and chimeric-transgenic mice by forced expression of E-cadherin (68). Despite the slowing of cell migration out of the crypt, terminal differentiation markers that are normally only expressed on the villus were not produced in the crypt. This finding suggests that terminal differentiation of the principal small intestinal epithelial lineages is largely cell nonautonomous and apparently dependent upon instructions obtained at specific positions along the crypt-villus axis. An intriguing question raised by these considerations is what determines whether a cell migrates out of a crypt or down to its base.
Paneth Cells and Intestinal Neoplasia-A Leu 3 Stop substitution at codon 850 in the 2845-residue mouse Apc protein is associated with the development of multiple intestinal neoplasms (Min; Refs. 17 and 69). Comparable germ line mutations in the human APC gene also leads to multiple gut adenomas (familial adenomatous polyposis). Mom1 is a semidominant modifier of tumor multiplicity in Min/ϩ animals located on mouse chromosome 4 (70). As noted in the Introduction, genetic studies indicate that Pla2g2a is a candidate gene for Mom1 (23)(24)(25). Two Pla2g2a alleles have been described. One allele contains a frameshift mutation (Mom-1 S ) and is encountered in C57BL/6J and 129/Sv-Pas mice. The other allele does not contain the mutation (Mom1 R ) and is found in AKR/J, MA/MyJ, BALB/cByJ, and Mus castaneus animals (25). Mom1 S is associated with a 4 -8-fold greater number of intestinal adenomas.
Current evidence indicates that a Min adenoma arises within a small intestinal crypt and that the initiated cell may be the stem cell or one of its immediate descendants (e.g. Ref. 71). The mechanism by which the secreted phospholipase A 2 encoded by Pla2g2a could influence initiation or progression is unclear at present. Our experiments indicate that neither a marked reduction in Pla2g2a in the crypt (CR2-tox176 mice) nor an expansion of the population of intermediate and granule goblet cells that produce this phospholipase A 2 (CR2-TAg mice) are associated with apparent changes in the properties of the crypt stem cell or its descendants.
Although the Mom1 allele of FVB/N mice has not been characterized, crosses of FVB/N CR2-tox176 and their nontransgenic littermates to Min/ϩ animals with Mom1 S or Mom1 R alleles illustrate how Paneth cell ablation can be used as a preliminary test of the effects of gene products produced in this lineage on tumorigenesis. 2 Targeting potential regulators of tumorigenesis to the apical secretory apparatus of Paneth cells using nucleotides Ϫ6500 to ϩ 34 of the mouse cryptdin-2 gene could result in their export to the stem cell zone, thereby testing their effects in Min/ϩ or other mouse models. These latter experiments illustrate how the Paneth cell can be used as a tool for delivering a variety of molecules to a critical region of the crypt where decisions about proliferative status and lineage allocation are made.
Ablation of the Paneth Cell Lineage Does Not Appear to Affect Host-microbial Interactions or the Spatial Organization of the Diffuse GALT-Small intestinal crypts are normally devoid of detectable micro-organisms. Surprisingly, our lineage ablation experiments suggest that the anti-microbial factors produced by Paneth cells may not be required to prevent colonization of these crypts. Nonetheless, Paneth cells may play other roles in host-microbial interactions. For example, the anti-microbial products released from Paneth cells could function to define the composition and/or density of the intestinal microflora. This idea is consistent with the fact that bacterial density in the small intestine is generally 3 to 4 orders of magnitude less than in the cecum and colon (50) which lack Paneth cells. Members of the microflora are able to establish niches at various positions along the duodenal-colonic axis even in the face of the epithelium's continuous replacement. Colonization of this open ecosystem begins soon after birth and progresses through a series of stages resulting in complex but stable climax communities predominated by Gram-negative anaerobes (29). "Autochthonous" members of the flora represent indigenous species that normally inhabit a given ecological niche. "Allochthonous" members are colonizers that do not normally occupy a given niche. Their existence in the ecosystem either results from a transient "passing through" or from conditions that significantly disrupt the stability of the autochthonous flora (e.g. starvation or treatment with antibiotics). The composition and flux of the autochthonous flora play a significant role in the ability of pathogens to gain a foothold in the ecosystem. Cephalocaudal differences in Paneth cell cryptdin expression together with the differential sensitivity of various microorganisms to different cryptdins may help define the gut's autochthonous (and allochthonous) flora. With these thoughts in mind, it will be important to evaluate the effects of Paneth cell ablation on the density, composition, and cephalocaudal distribution of components of the gut's microbiota. This can be done using conventional mice or animals that have been raised under germ-free conditions and then inoculated with one or more species of bacteria that normally reside in the gut.
CR2-tox176-mediated ablation of Paneth cells in mice that are free of pathogens had no demonstrable effect on the cryptvillus distribution of members of the diffuse GALT. This finding not only provides another piece of evidence that host-microbial interactions are not markedly deranged but also indicates that this lineage does not function, directly or indirectly, in establishing or maintaining the asymmetric distribution of some critical components of the diffuse GALT.
While previous genetic experiments have shown that ablation of components of the diffuse GALT (e.g. intraepithelial ␥␦ T-cells) affects epithelial homeostasis (57), reciprocal experiments designed to test the effects of ablating epithelial cell lineages on GALT homeostasis have not been described. This likely reflects prior difficulty in defining the spatial organization of the diffuse gut-associated immune system, a problem that can now be overcome by using tyramide signal amplification to increase the sensitivity of immunohistochemical surveys.
There are several notable asymmetries in the crypt-villus distribution of components of the diffuse GALT in adult FVB/N CR2-tox176 transgenic mice and their normal littermates. First, CD8 ϩ cells are principally intraepithelial and are restricted to the villus, whereas CD4 ϩ cells are distributed along the length of the crypt-villus axis where they reside principally in the lamina propria. Second, ␥␦ T-cells are limited to the villus and its intraepithelial compartment. In contrast, ␣␤ Tcells are not restricted to the intraepithelial compartment and are distributed along the length of crypt-villus units. The absence of CD8 ϩ and ␥␦ T-cells from the crypt may reflect, at least in part, the lack of microbial colonization of this region of the crypt-villus axis. If true, these cells may be useful markers to follow if and when colonization occurs.
Our studies suggest that further assessment of the contributions of Paneth cells to the regulation of host-microbial and microbial-immune interactions might require provocative tests. Possibilities include introduction of pathogens, with or without members of normal flora, into germ-free CR2-tox176 animals and their germ-free nontransgenic littermates or crosses between conventionally raised CR2-tox176 animals and mice that are genetically predisposed to develop inflammatory bowel disease. An example of the latter would be interleukin-10 knockout mice who develop small intestinal mucosal inflammation when housed in a conventional animal facility but not when they are raised in a specific pathogen-free state (73).