A Gnotobiotic Transgenic Mouse Model for Studying Interactions between Small Intestinal Enterocytes and Intraepithelial Lymphocytes* 210

The mouse intestinal epithelium undergoes continuous renewal throughout life. Intraepithelial lymphocytes (IELs) represent a significant fraction of this epithelium and play an important role in intestinal mucosal barrier function. We have generated a germ-free transgenic mouse model to examine the effects of a genetically engineered proliferative abnormality in the principal epithelial cell lineage (enterocytes) on IEL census and on IEL-enterocytic cross-talk. SV40 large T antigen (TAgWt) or a mutant derivative (TAgK107/8) that does not bind pRB was expressed in small intestinal villus enterocytes under the control of elements from the intestinal fatty acid binding protein gene (Fabpi). Quantitative immunohistochemical and flow cytometric analyses of conventionally raised and germ-free FVB/NFabpi-TAgWt,Fabpi-TAgK107/8, and nontransgenic mice disclosed that forced reentry of enterocytes into the cell cycle is accompanied by an influx of thymically educated αβ T cell receptor (TCR)+ CD4+ and αβ TCR+CD8αβ+ IELs and a decrease in intestinally derived γδ TCR+ CD8αα IELs. Real time quantitative reverse transcriptase-PCR studies of jejunal villus epithelium recovered from germ-free transgenic and normal mice by laser capture microdissection and γδ TCR+ jejunal IELs purified by flow cytometry disclosed that the proliferative abnormality is accompanied by decreased expression of enterocytic interleukin-7 as well as IEL interleukin-7Rα and transforming growth factor β3. The analysis also revealed that normal villus epithelium expresses Fms-like tyrosine kinase 3 (Flt3), a known regulator of hematopoietic stem cell proliferation and neuronal cell survival, and its ligand (Flt3L). Epithelial expression of this receptor and its ligand is reduced by the proliferative abnormality, whereas IEL expression of Flt3L remains constant. Together, these findings demonstrate that changes in the proliferative status of the intestinal epithelium affects maturation of γδ TCR+ IELs and produces an influx of αβ TCR+ IELs even in the absence of a microflora.


The mouse intestinal epithelium undergoes continuous renewal throughout life. Intraepithelial lymphocytes (IELs) represent a significant fraction of this epithelium and play an important role in intestinal mucosal barrier function. We have generated a germfree transgenic mouse model to examine the effects of a genetically engineered proliferative abnormality in the principal epithelial cell lineage (enterocytes) on IEL census and on IEL-enterocytic cross-talk. SV40 large T antigen (TAg Wt ) or a mutant derivative (TAg K107/8 ) that does not bind pRB was expressed in small intestinal villus enterocytes under the control of elements from the intestinal fatty acid binding protein gene (Fabpi). Quantitative immunohistochemical and flow cytometric analyses of conventionally raised and germ-free FVB/N
Fabpi-TAg Wt , Fabpi-TAg K107/8 , and nontransgenic mice disclosed that forced reentry of enterocytes into the cell cycle is accompanied by an influx of thymically educated ␣␤ T cell receptor (TCR) ؉ CD4 ؉ and ␣␤ TCR ؉ CD8␣␤ ؉ IELs and a decrease in intestinally derived ␥␦ TCR ؉ CD8␣␣ IELs. Real time quantitative reverse transcriptase-PCR studies of jejunal villus epithelium recovered from germ-free transgenic and normal mice by laser capture microdissection and ␥␦ TCR ؉ jejunal IELs purified by flow cytometry disclosed that the proliferative abnormality is accompanied by decreased expression of enterocytic interleukin-7 as well as IEL interleukin-7R␣ and transforming growth factor ␤3. The analysis also revealed that normal villus epithelium expresses Fms-like tyrosine kinase 3 (Flt3), a known regulator of hematopoietic stem cell proliferation and neuronal cell survival, and its ligand (Flt3L). Epithelial expression of this receptor and its ligand is reduced by the proliferative abnormality, whereas IEL expression of Flt3L remains constant. Together, these findings demonstrate that changes in the proliferative status of the intestinal epithelium affects maturation of ␥␦ TCR ؉ IELs and produces an influx of ␣␤ TCR ؉ IELs even in the absence of a microflora.
The adult mouse small intestine is a complex, spatially diversified ecosystem that maintains distinctive cephalocaudal differences in its various functions. This regional variation in function is accompanied by regional differences in the differentiation programs of its four continuously renewing epithelial cell lineages, in the composition of its mucosal immune system, and in the composition of its resident society of commensal/ symbiotic microorganisms (reviewed in Refs. [1][2][3]. A full understanding of how this ecosystem is organized and functions in health and how it is reorganized or disorganized in various disease states requires knowledge about the nature and regulation of interactions between its microflora, epithelium, and gut-associated lymphoid tissue (1,4,5). The molecular nature and significance of the signals exchanged between these components have been difficult to decipher because of the dynamic quality and complexity of the system. One way of approaching this problem is to simplify the ecosystem using inbred strains of mice with defined microbiological status (gnotobiotic animals). For example, comparative functional genomics studies of mice containing no resident microorganisms (germ-free), conventionally raised mice harboring an complete microflora, and germ-free animals that have been colonized with a single species from the normal microflora (ex-germ-free) have shown that indigenous commensal bacteria play an important role in regulating host nutrient processing, fortifying the epithelial barrier, and organizing/educating the mucosal immune system (5,6).
The intestine contains a large population of intraepithelial lymphocytes (IELs), 1 equivalent in size to the population of peripheral lymphocytes that resides in the spleen (7). IELs are distributed throughout the epithelium that overlies small intestinal villi (average of one IEL for every 6 -10 epithelial cells (8)). Virtually all small intestinal IELs are T cells, but they are heterogeneous with respect to their surface phenotype. The majority are CD3 ϩ and can be divided into ␣␤ T cell receptorpositive (TCR ϩ ) and ␥␦ TCR ϩ subsets (8). They can be further subdivided based on expression of CD8 (␣␣ homodimer or ␣␤ heterodimer) or CD4 coreceptors (i.e. nude mice or peripheral lymph node T cells from euthymic mice demonstrated that generation of ␣␤ TCR ϩ CD4 ϩ and CD8 ϩ IELs is thymus-dependent, whereas ␥␦ TCR ϩ CD8␣␣ ϩ IELs appeared in the absence of a thymus (9). One site of extrathymic maturation may be the crypts of Lieberkuhn. These distinct mucosal invaginations surround the base of each villus and contain long-lived multipotent stem cells (10) that give rise to the four epithelial lineages of the small intestine: enterocytes, goblet, and enteroendocrine cells, which differentiate as they migrate from the crypt up adjacent villi; and Paneth cells, which differentiate and remain at the crypt base (11)(12)(13)(14)(15). Crypts possess structures (cryptopatches) that contain clusters of c-Kit ϩ interleukin-7 receptor (IL-7R) ϩ Thy1 ϩ lymphocytes (16). Mice with a truncated mutation of the common cytokine receptor chain (17) lack these cryptopatches and do not have ␥␦ TCR ϩ CD8␣␣ ϩ IELs but contain thymus-dependent ␣␤ TCR ϩ CD4 ϩ and ␣␤ TCR ϩ CD8␣␤ ϩ IELs, suggesting a role for cryptopatches in maturation of extrathymically derived ␥␦ TCR ϩ IELs (18 -20).
The epithelium also appears to play a direct role in regulating IEL development. Epithelial cells produce stem cell factor (21), a ligand for the c-Kit receptor expressed on the surface of ␥␦ TCR ϩ IELs (22). Mice deficient in either stem cell factor or c-Kit have reduced numbers of ␥␦ TCR ϩ IELs (22). Furthermore, thyrotropin-releasing hormone stimulation of enterocytes results in local release of thyroid-stimulating hormone, which interacts with IEL-based thyroid-stimulating hormone receptor to promote IEL development (23) (e.g. hyt/hyt mice, which have a loss-of-function thyroid-stimulating hormone receptor mutation, have disrupted IEL maturation) (24,25).
Epithelium-based IL-7 provides another regulatory signal for IEL proliferation (26). Studies of mice that lack IL-7 or the IL-7R have demonstrated that IL-7R-mediated signaling is essential for ␥␦ TCR ϩ IEL development (26,27). Moreover, Laky et al. (28) used transcriptional regulatory elements from the rat intestinal fatty acid-binding protein (Fabpi) to express IL-7 in the villus enterocytes of Il-7 Ϫ/Ϫ mice. ␥␦ TCR ϩ IELs were restored in the intestinal epithelium but remained absent from all other tissues, indicating that local production of IL-7 was sufficient for proper development/survival of this IEL subset.
Interactions between intestinal epithelial cells and IELs are reciprocal; IELs can influence epithelial cell biology. One illustration of this reciprocity is provided by TCR␦ subunit-deficient mice. These animals have reduced numbers of dividing cells in their crypts of Lieberkuhn and reduced crypt cellularity (29) and exhibit more severe intestinal epithelial damage following infection with the parasite Eimeria vermiformis (30). ␥␦ TCR ϩ IELs produce keratinocyte growth factor, which affects epithelial cell growth and repair (31). These findings raise the question of whether ␥␦ TCR ϩ IELs form part of a homeostatic surveillance mechanism that can detect and respond to perturbations in intestinal epithelial proliferation in order to maintain steady state cellular census in crypts and their associated villi.
Some workers have proposed that IELs are key elements in a "mucosal intranet," where they function to control epithelial integrity and immunologic homeostasis (32). Recent comparative DNA microarray-based studies of gene expression in ␥␦ TCR ϩ IELs harvested from the small intestines of conventionally raised adult C57Bl6/J mice and ␣␤ TCR ϩ cells harvested from their mesenteric lymph nodes have provided a list of candidate factors, preferentially expressed by ␥␦ TCR ϩ IELs, that may support this mucosal intranet (33,34).
In the present study, we examine the cross-talk between IELs and epithelium using transgenic mice that express simian virus 40 large T antigen (TAg Wt ) in their villus-associated enterocytes. The rationale for our experimental approach was as follows. Fabpi-directed expression of TAg Wt produces a proliferative abnormality restricted to villus enterocytes: Fabpireporter transgenes are not expressed in the IELs. Expression of the viral oncoprotein in postmitotic enterocytes induces their reentry into the cell cycle (35) and an associated p53-independent apoptosis (36) but is not accompanied by evidence of dysplasia during the 1-2-day interval that they take to complete their migration to the cellular extrusion zone located at the villus tip (36,37). Fabpi-directed expression of a mutant TAg containing a Glu 3 Lys substitution at residues 107 and 108 (TAg K107/8 ) disrupts pRB binding that does not produce this proliferative abnormality. Thus, a three-way comparison of FVB/N Fabpi-TAg Wt and Fabpi-TAg K107/8 transgenic mice and their age-matched nontransgenic littermates would allow direct assessment of whether a proliferative abnormality limited to the predominant intestinal epithelial lineage is accompanied by changes in the fractional representation of extrathymically educated or thymically derived IEL subsets. By performing this analysis in conventionally raised and germ-free mice, we could also determine whether the microflora contributed to any observed changes in IELs. Finally, by using laser capture microdissection (LCM) of small intestinal cryosections to harvest villus epithelium, flow cytometry to retrieve their IELs, and the DNA microarray-based data sets of IEL gene expression to direct quantitative reverse transcription-PCR measurements of the levels of specified mRNAs in each cell population, we could use this environmentally well defined system to identify enterocytic gene products affected by proliferative status that may impact on IEL development/survival.
Our results show that the engineered proliferative abnormality is accompanied by a microflora-independent reduction in extrathymically educated ␥␦ TCR ϩ CD8␣␣ ϩ IELs. This change is accompanied by coordinate changes in the expression of enterocytic and ␥␦ TCR ϩ IEL gene products that probably help legislate the observed change in IEL representation.

EXPERIMENTAL PROCEDURES
Generation and Maintenance of Conventionally Raised and Germfree Transgenic Mice-FVB/N mice hemizygous for a transgene containing nucleotides Ϫ1178 to ϩ28 of rat Fabpi linked to TAg Wt or TAg K107/8 are described in earlier reports (35,36,38). Conventionally raised animals were maintained in microisolators in a specified pathogen-free state.
Normal and transgenic mice were rederived as germ-free by Caesarian section of transgenic mothers and transfer of their embryonic day 19 fetuses to plastic gnotobiotic isolators (Standard Safety Equipment Co.) containing germ-free foster mothers. The protocol used for this rederivation is described in a recent publication (6). Both conventionally raised and germ-free mice were given sterilized BeeKay Autoclavable Diet (B & K Universal Inc.) ad libitum. All animals were maintained under a strict light cycle (lights on at 0600 h and off at 1800 h). Animals were genotyped using primers, tail DNA, and PCR conditions described in Ref. 36. Some mice received an intraperitoneal injection of an aqueous solution of 5-bromo-2Ј-deoxyuridine (BrdUrd; 120 mg/kg) and 5-fluoro-2Ј-deoxyuridine (12 mg/kg) (Sigma) 90 min prior to sacrifice. Only male mice were studied.
Following incubation with these primary antibodies, sections were washed in TNT buffer (three cycles, each 5 min). Biotin-conjugated mouse anti-rat IgG1/IgG2a (BD PharMingen) or biotin-conjugated mouse anti-hamster IgG mixture (BD PharMingen) was added (final dilution of each ϭ 1:100 in TNB blocking buffer). After a 30-min incubation with the secondary antibodies at room temperature, sections were treated three times with TNT wash buffer (5 min/wash cycle). The sections were then incubated for 30 min at room temperature with streptavidin-horseradish peroxidase (PerkinElmer Life Sciences; 1:1000 in TNB) followed by three washes of 5 min each in TNT buffer. The final steps consisted of (i) adding biotinyl-tyramide (PerkinElmer Life Sciences; diluted 1:100 in 1ϫ amplification diluent from the same manufacturer) for 10 min; (ii) washing three times with TNT buffer (5 min/cycle); (iii) incubating the section with indocarbocyanine (Cy3)conjugated streptavidin (PerkinElmer Life Sciences; diluted 1:500 in TNB) for 30 min, and (iv) performing three final rinses in TNT buffer. Two controls were performed to verify the specificity of the signals produced: (i) direct amplification of endogenous peroxidase activity alone without the addition of primary or secondary antibodies but with the addition of biotinyl-tyramide; (ii) direct amplification of endogenous peroxidase activity followed by omission of each primary antibody but with inclusion of all other steps and reagents.
Only well oriented jejunal crypt-villus units were scored. "Well oriented" was defined as sectioned parallel to the crypt-villus axis with an unbroken epithelial column extending from the crypt base to the villus tip. The data were compiled as the number of IELs of a particular type per 1000 villus epithelial cells or per 100 crypt epithelial cells. A minimum of 100 jejunal crypt-villus units were scored per mouse. Data obtained with each antibody from all mice of a given genotype (germfree or conventional) were averaged.
FACS Analysis of IELs-6 -8-week-old transgenic mice and their normal littermates were sacrificed, and their jejunums were recovered (n ϭ 3 germ-free and 3 conventionally raised mice/genotype/experiment; three independent experiments). Peyer's patches were identified by inspecting the serosal surfaces of the jejunal segment and were then excised. Each jejunal segment was subsequently opened with a longitudinal incision, washed in PBS, and cut into 1-cm pieces that were placed in 40 ml of ice-cold sterile PBS. The pooled segments from all three animals/genotype/experiment were washed five times in PBS (vigorous shaking), allowed to settle by gravity, and resuspended in 25 ml of R2 medium (RPMI 1640 buffer containing 5% fetal calf serum (Sigma), 1 mM sodium pyruvate, 1 mM sodium bicarbonate, 1% nonessential amino acids (Sigma), and 0.1% 2-mercaptoethanol). The mixture was shaken gently for 30 min at 37°C and then rigorously for 2 min at room temperature. The intestinal segments were allowed to settle by gravity, and the supernatant was collected and passed through a Nytex filter (Becton Dickinson). The flow-through, containing IELs and epithelial cells, was passed over a column of dimethyldichlorosilanetreated glass wool fiber (0.5 g/10-ml syringe) preequilibrated in R2 medium. The flow-through was spun at 1500 ϫ g for 5 min, and the resulting cell pellet, highly enriched for IELs, was resuspended in 10 ml of R2 medium. The suspension was centrifuged at 1500 ϫ g for 5 min, and the pellet resuspended to a final concentration of 10 7 cells/ml of FACS staining buffer (RPMI, 1% bovine serum albumin (Sigma), 1 mg/ml human IgG (Sigma)).
Laser Capture Microdissection (LCM) of Jejunal Villus Epithelium-LCM was conducted using jejunal cryosections that had been stained briefly with eosin Y and methyl green. Dissection of villus epithelium was restricted to well oriented crypt-villus units and was accomplished using the PixCell II system (Arcturus; 7.5-m diameter laser spot), CapSure HS LCM Caps (Arcturus), and protocols described in Ref. 40. ϳ10,000 jejunal villus epithelial cells were harvested from each germfree normal and TAg mouse (n ϭ 3 animals/group). RNA was prepared from captured cells from each mouse in each group using the PicoPure RNA Isolation Kit (Arcturus). The concentration of each preparation was defined (RiboGreen RNA quantitation kit; Molecular Probes, Inc., Eugene, OR), and equally sized aliquots from members of a group of animals were pooled.
Analysis of Previously Published DNA Microarray Data Sets-Data sets of gene expression profiles from ␥␦ TCR ϩ IELs and the ␣␤ TCR ϩ cells were a generous gift from Aude Fahrer and Y-H. Chien (Dept. of Microbiology and Immunology, Stanford University) (33). These data sets were obtained using an early manufactured version of a high density, oligonucleotide-based DNA microarray containing probe sets representing 6352 mouse genes or expressed sequence tag clusters (Mu6K GeneChip; Affymetrix).
We used GeneChip software (version 4.0; Affymetrix) to compute an average fluorescence intensity across all probe sets on the GeneChips prior to conducting pairwise chip-to-chip comparisons (41) of ␥␦ TCR ϩ IEL versus ␣␤ T cell transcript levels (the ␣␤ TCR ϩ cell GeneChip was designated as "base line"). We then extracted all mRNAs fulfilling the following selection criteria: (i) called "present" in either the base-line or partner chip; (ii) Ն2-fold difference in transcript level in the two RNA populations (increased or decreased); and (iii) the increase or decrease was reproduced in duplicate GeneChip comparisons SYBR Green-based Real Time Quantitative PCR (qRT-PCR)-qRT-PCR was used to examine changes in levels of selected mRNAs in RNAs prepared from the intact jejunums, LCM villus epithelium, and/or sorted ␥␦ TCR ϩ IELs harvested from 6 -8-week-old male germ-free normal and TAg Wt mice. cDNAs were generated from each pooled RNA preparation (see above) using reagents and protocols described in Ref. 40. cDNA was added to 25 l of qRT-PCRs containing 12.5 l of 2ϫ SYBR Green master mix (Applied Biosystems), 900 nM gene-specific primers (see Table I in the supplemental material) and 0.25 units UDP-N-glycosidase (Invitrogen). A melting curve was used to identify a temperature where only the amplicon, and not primer dimers, accounted for the SYBR Green-bound fluorescence (6). Assays were performed in triplicate with an ABI Prism 7700 sequence detector (Applied Biosystems). All data were normalized to an internal standard (glyceraldehyde-3-phosphate dehydrogenase mRNA; ⌬⌬C T method, User Bulletin 2, Applied Biosystems). For the ␥␦ TCR ϩ IEL mRNA analysis, 18 S rRNA was used as the internal standard.

Forced Expression of TAg Wt in Villus Enterocytes Causes a Change in the Representation of IEL Subsets within the Small
Intestinal Epithelium-As noted in the Introduction, transcriptional regulatory elements from the Fabpi gene were used to direct expression of TAg Wt in small intestinal villus enterocytes of adult FVB/N transgenic mice (Fig. 1A). There was no detectable TAg Wt in the crypt epithelium, the mesenchyme underlying crypt-villus units (Fig. 1A), the organized gut-associated lymphoid tissue (Peyer's patch lymphocytes plus smaller submucosal lymphoid aggregates), or in the spleen and thymus (data not shown). Other than villus enterocytes, the only other site of transgene expression was the follicle-associated epithelium overlying Peyer's patches (Fig. 1B). An identical pattern of transgene expression was observed in FVB/N mice from the reference control pedigree containing Fabpi-TAg K107/108 (data not shown).
Age-matched 6 -8-week-old Fabpi-TAg Wt and Fabpi-TAg K107/108 male mice as well as their nontransgenic litter-mates were given an intraperitoneal injection of BrdUrd, 1.5 h prior to sacrifice (n ϭ 2-3 mice/genotype). Expression of the wild type viral oncoprotein induced villus enterocytes to reenter the cell cycle (Fig. 1A). In contrast, the jejunal villus epithelium and follicle-associated epithelium were not labeled with BrdUrd in either wild type or Fabpi-TAg K107/108 mice (data not shown). To determine whether the proliferative abnormality induced by TAg Wt caused a change in the composition or spatial organization of IELs, these cells were isolated from the jejunal epithelium of each group of conventionally raised mice and subjected to flow cytometry. There were no statistically significant differences in the purity of the lymphocyte preparations from each group of mice; Ͼ80% of the gated lymphocytes expressed the IEL-specific marker, CD103 (Fig. 2A). The total yield of lymphocytes was similar in each group (5-7 ϫ 10 7 ).
The majority of the IELs were also positive for CD45, a pan-lymphocyte marker (Fig. 2B). However, there was a statistically significant increase in the fractional representation of ␣␤ TCR ϩ IELs in Fabpi-TAg Wt mice compared with their normal littermate controls (p Ͻ 0.05; Student's t test) and a statistically significant decrease in ␥␦ TCR ϩ IELs (p Ͻ 0.05) (Fig.  2, C and D). In contrast, there were no differences in the percentages of these IEL subsets in Fabpi-TAg K107/8 versus normal animals (Fig. 2, C and D).
We performed a quantitative immunohistochemical study of jejunal crypt-villus units to determine whether the change in ␣␤ TCR ϩ and ␥␦ TCR ϩ IEL representation in Fabpi-TAg Wt mice was restricted to the villus epithelium, where the proliferative abnormality was evident, or whether the change extended to the crypt epithelium, where there was no change in proliferative status. An analysis of sections of jejunum indicated that there were no significant differences in the total number of CD103 ϩ IELs per 1000 villus epithelial cells between age-matched Fabpi-TAg Wt , Fabpi-TAg K107/108 , and normal FVB/N mice (Fig. 3A). However, there was a significant increase in the density of ␣␤ TCR ϩ IELs, and a significant reduction in the density ␥␦ TCR ϩ IELs in TAg Wt mice compared with the other two groups (p Ͻ 0.05) (Fig. 3, B and C). In the crypt epithelium of conventionally raised normal male 6 -8-week-old FVB/N mice, the densities of CD103 ϩ , ␣␤ TCR ϩ , and ␥␦ TCR ϩ lymphocytes are 10 Ϯ 1, 5 Ϯ 1, and 4 Ϯ 1 per 100 epithelial cells, respectively. There were no statistically significant differences in the numbers of these cells among the three groups of mice, indicating that the proliferative abnormality produced by TAg Wt had a "local" effect on villus IELs that did not extend to the crypt.
The Increase in ␣␤ TCR ϩ and Decrease in ␥␦ TCR ϩ IELs Observed in Conventionally Raised TAg Wt Transgenics Is Recapitulated in Germ-free Mice-One question raised by these findings is whether the intestinal microflora was exerting an influence on the composition of the villus IEL population, e.g. from a potential epithelial barrier disruption associated with the engineered proliferative abnormality, or as a direct consequence of a cross-talk between components of the microflora and the epithelium. To address this question, we rederived our pedigrees of Fabpi-TAg Wt and Fabpi-TAg K107/108 transgenic mice and their normal littermates as germ-free. The cellular patterns of expression of TAg Wt and TAg K107/108 were not affected when the microflora was removed. An epithelial proliferative abnormality extending from the base to the tips of the villi was evident in 6 -8-week-old germ-free FVB/N Fabpi-TAg Wt but not in Fabpi-TAg K107/108 or normal animals (Fig. 4A plus data not shown).
Quantitative immunohistochemical studies also disclosed that Fabpi-TAg Wt mice, like their conventionally raised counterparts, had a reduction in the density of their villus ␥␦ TCR ϩ IELs when compared with age-and gender-matched FVB/N Fabpi-TAg K107/108 or normal mice (p Ͻ 0.05; Fig. 4, B-D). There was also a modest increase in ␣␤ TCR ϩ IELs associated with the TAg Wt -induced proliferative abnormality, although the differences were not statistically significant compared with the other two groups of mice (Fig. 4E). The density of all IELs (i.e. CD103 ϩ cells) in the jejunal villus epithelium was similar in all three groups of germ-free mice (Fig. 3F) but severalfold less than in conventionally raised animals (compare Figs. 3A and 4F). As in conventionally raised mice, production of TAg Wt in the villus epithelium did not result in any changes in the number of crypt-associated CD103 ϩ , ␣␤ TCR ϩ , or ␥␦ TCR ϩ IELs (data not shown).
FACS analysis of germ-free jejunal IELs confirmed the results of our quantitative immunohistologic survey and established that TAg Wt expression produced a statistically significant increase in ␣␤ TCR ϩ and a significant decrease in ␥␦ TCR ϩ IELs (p Ͻ 0.05 in comparison with age-matched normal or TAg K107/108 mice (Fig. 5, A-C).
Based on these findings, we concluded that the alterations in these two IEL populations occurred independently of the microflora and were ascribable to the proliferative effects of TAg Wt rather than to other functions mediated by regions of the viral oncoprotein located outside of its pRB pocket protein binding domain.
Expression of TAg Wt Leads to a Decrease in Accumulation of Intestinally Derived ␥␦ TCR ϩ CD8␣␣ IELs and an Increase in Thymically Derived ␣␤ TCR ϩ CD8␣␤ IELs-As noted in the Introduction, intestinal IELs are derived from two sources. The vast majority of ␣␤ TCR ϩ CD4 ϩ and ␣␤ TCR ϩ CD8␣␤ ϩ IELs are thymically derived, whereas all CD8␣␣ ϩ cells, whether they express ␣␤ TCR or ␥␦ TCR, are derived from extrathymic sites (42). The phenotype produced by TAg Wt -induced proliferation of villus enterocytes in germ-free mice could reflect subtle disruptions of epithelial barrier function with resulting presentation of nonmicrobial luminal antigens (e.g. from the diet) to components of the gut-associated lymphoid tissue. If this were the case, one would expect an increased influx of thymically derived, antigen-induced ␣␤ TCR ϩ IELs.
We addressed this hypothesis in two ways. First, germ-free Fabpi-TAg Wt mice and their normal littermates were given an intraperitoneal injection of BrdUrd 1.5 h prior to sacrifice to label intestinal epithelial cells in S phase. Sections of jejunal crypt-villus units were then stained with antibodies to BrdUrd and E-cadherin, the principal epithelial cadherin and an important regulator of cell adhesion in this system (43,44). Expression of TAg Wt and/or entry of jejunal enterocytes into the cell cycle produced no detectable changes in the steady state cellular levels or intracellular compartmentalization of E-cadherin (data not shown; n ϭ 2 germ-free mice/genotype). Second, FACS analysis of jejunal IELs demonstrated that the TAg Wtassociated increase in ␣␤ TCR ϩ IELs involved both the CD4 and CD8␣␤ subsets (Fig. 5, D and E). There were no changes in the CD8␣␣ subtype of ␣␤ TCR IELs (data not shown). These findings confirm that the proliferative abnormality engineered FIG. 3. Quantitative immunohistochemical studies of ␣␤ TCR ؉ and ␥␦ TCR ؉ IELs in conventionally raised transgenic and normal mice. Sections, prepared from the jejunums of 6 -8-week-old male Fabpi-TAg Wt , Fabpi-TAg K107/108 , and normal animals, were stained with antibodies to CD103 and the ␤ chain or the ␦ chain of TCR. A, evidence that there are no significant differences in the total number of CD103 ϩ IELs per 1000 villus epithelial cells among the three groups. Mean values Ϯ S.E. are shown. B, data indicating that there is a statistically significant increase in the density of ␣␤ TCR ϩ IELs in Fabpi-TAg Wt mice (asterisk; p Ͻ 0.05 when compared with normal FVB/N mice). C, results showing a statistically significant reduction in ␥␦ TCR ϩ IELs in Fabpi-TAg Wt animals. in enterocytes is associated with an influx of thymically derived IELs.
FACS analysis also established that expression of TAg Wt , but not TAg K107/108 , in germ-free villus epithelium leads to a significant (p Ͻ 0.05) reduction in intestinally derived ␥␦ TCR ϩ CD8␣␣ IELs compared with normal littermate controls (Fig. 5F). Immunostaining of intestinally derived ␥␦ TCR ϩ CD8␣␣ IELs and thymically derived ␣␤ TCR ϩ CD4 ϩ and CD8␣␤ subsets obtained by flow cytometry revealed that they did not contain detectable levels of TAg (data not shown plus see below).
Taken together, these findings demonstrate that TAg Wt -dependent reentry of villus enterocytes into the cell cycle produces a specific decrease in the ␥␦ TCR ϩ CD8␣␣ IEL populations that normally develop in the intestine.
qRT-PCR Analysis of TAg Wt -dependent Regulation of IL-7 Expression-Previous studies have established that the majority of intestinal IELs are maintained in G 0 of the cell cycle (45). In addition, some reports have suggested that epithelial cells may act as antigen-presenting cells for induction and activation of these resting IELs (46,47). Thus, the proliferative abnormality produced by TAg Wt could result in suppression of critical trophic factors necessary for the appropriate development and activation of ␥␦ TCR ϩ CD8␣␣ IELs, leading to their diminution in the epithelium. IL-7 is one such trophic factor: it is produced by the epithelium and required for generation of mature ␥␦ TCR ϩ IELs (see Introduction). ␥␦ TCR ϩ IELs express the receptor for this cytokine, IL-7R (48). We tested the hypothesis that TAg Wtinduced reentry of villus enterocytes into the cell cycle is accompanied by reduced IL-7 expression by performing a qRT-PCR analysis of RNAs isolated from intact jejunum as well as LCM jejunal villus epithelium (Fig. 6A). The results revealed a 12-fold lower steady state concentration of IL-7 mRNA in the intact jejunum of germ-free Fabpi-TAg Wt mice compared with germ-free normal littermates and a 4-fold reduction in levels in their LCM villus epithelium (Fig. 6B). Control qRT-PCR assays of LCM epithelial RNA documented a 2-fold reduction in TCR␦ mRNA (Fig. 6C), consistent with the reduced representation of ␥␦ TCR ϩ IELs in transgenic mouse jejunum documented by quantitative immunohistochemical and flow cytometry analyses (Figs. 4D and 5F).
To address the question of whether the TAg Wt -induced proliferative abnormality in villus enterocytes produced changes in ␥␦ TCR ϩ IEL gene expression, we purified these cells, using flow cytometry, from the jejunums of 6 -8-week-old germ-free male Fabpi-TAg Wt and normal mice (n ϭ 50 mice/group). qRT-PCR studies indicated that the IELs from transgenic mice did not contain detectable levels of TAg mRNA, in agreement with the results of our immunohistochemical studies (see above). IL-7R␣ mRNA levels were significantly decreased in ␥␦ TCR ϩ IELs from transgenic compared with normal mice (5.5-fold; p Ͻ 0.05; Fig. 7). Fujihashi et al. (26) used IL-7 knockout mice to show that IL-7 signaling is necessary for IL-7R expression in ␥␦ TCR ϩ IELs and for their subsequent activation and cell division. Based on this observation and the findings described above, we concluded that TAg Wt expression in villus enterocytes results in decreased epithelial expression of IL-7, leading to a concomitant decrease in expression of the IL-7 receptor in ␥␦ TCR ϩ IELs, and impeded intestinal development of ␥␦ TCR ϩ CD8␣␣ IELs.
TAg Wt Expression in Enterocytes Is Associated with Reduced Levels of Other ␥␦ TCR ϩ IEL-derived Factors That May Affect IEL:Epithelial Cross-talk-Analysis of published DNA microarray-based expression profiles of ␥␦ TCR ϩ IELs purified from conventionally raised C57Bl/6 mice (33) allowed us to identify factors whose expression is enriched in ␥␦ TCR ϩ IELs relative to ␣␤ TCR ϩ cells and that may affect epithelial barrier functions and/or important interactions between the epithelium and its population of IELs.
TGF-␤3-Recent reports have shown that IL-7 regulates TGF-␤3 production in fibroblasts (49). Increased expression of TGF-␤3 leads to enhanced intestinal epithelial cell migration across wound edges in an in vitro model. Neutralizing antibodies to TGF-␤3 inhibit this process (50). TGF-␤3 also functions as a signaling factor that induces apoptotic cell death during involution of the mammary epithelium (51), suggesting that it may help regulate epithelial cell census.
qRT-PCR studies disclosed a 10-fold decrease in the steady state level of TGF-␤3 mRNA in LCM villus epithelium from germ-free Fabpi-TAg Wt compared with germ-free normal littermates (Fig. 6C). Furthermore, enterocytic expression of TAg Wt is associated with a 262-fold decrease in the concentration of ␥␦ TCR ϩ IEL TGF-␤3 mRNA (Fig. 7). Together, these results indicate that one consequence of reduced enterocytic IL-7 expression is reduced ␥␦ TCR ϩ IEL-derived TGF-␤3. Loss of TGF-␤3 may alter the integrity of the epithelial barrier, contribute to the observed influx of thymically derived ␣␤ TCR ϩ IELs, or help regulate the extent of the p53-independent apoptotic response that occurs in villus enterocytes undergoing unscheduled, TAg Wt -induced reentry into the cell cycle.
Flt3L-The DNA microarray studies revealed that the mRNA encoding the ligand for Fms-like tyrosine kinase-3 (Flt3) is enriched in ␥␦ TCR ϩ IELs compared with ␣␤ TCR ϩ cells (33). The function of Flt3L in ␥␦ TCR ϩ IELs is not known. Flt3 was initially identified in hematopoietic stem cells (52). It is a member of the class III receptor tyrosine kinases that includes c-Kit. Flt3 ligand stimulates proliferation of quiescent as well as cytokine-stimulated hematopoietic progenitors (e.g. see Refs. [53][54][55]. However, this proliferative response is not shared by other progenitors; Flt3 ligand inhibits EGF-and FGF2-stimulated division of neuronal stem cells (56). There is very little information about the regulation of expression of Flt3 ligand and its receptor or their functions in epithelia. One report indicated that Flt3 mRNA is present in mouse bile duct epithelium (57), whereas another identified the transcript in dividing neuroepithelial cells (56).
Our LCM/qRT-PCR studies revealed that the receptor is expressed in normal jejunal villus epithelium. Moreover, expression is down-regulated by the engineered proliferative abnormality; mRNA levels are reduced an average of 7.5-fold in LCM TAg Wt compared with nontransgenic epithelium (Fig.  6C).
The qRT-PCR/LCM analysis indicated that the mRNA encoding Flt3 ligand is also reduced in TAg Wt epithelium (Fig.  6C). qRT-PCR assays disclosed that TAg Wt expression in en- terocytes does not have a discernible effect on IEL Flt3 ligand expression (Fig. 7). Since the extent of the reduction in Flt3 ligand mRNA in TAg Wt epithelium was severalfold greater than the reduction of ␥␦TCR ϩ IEL number (5-versus 2-fold), and since IEL Flt3 ligand expression is unaffected by enterocytic TAg Wt expression, we concluded that the proliferative abnormality reduces epithelial expression of the ligand. The response of Flt3 and its ligand to changes in the proliferative status of enterocytes raises the possibility that signaling through this system may normally serve to help suppress cell division as members of this lineage execute their terminal differentiation program.
Prospectus-These studies reveal that an engineered proliferative abnormality in postmitotic enterocytes impedes intestinal development of ␥␦ TCR ϩ CD8␣␣ IELs and promotes accumulation of thymically educated CD4 and CD8␣␤ subsets of ␣␤ TCR ϩ IELs. Our findings highlight the interdependent contributions of enterocytes and ␥␦ TCR ϩ IELs to intestinal mucosal biology, a point illustrated by the diminution in enterocytic IL-7 expression associated with TAg Wt production. The resulting diminution in intestinal maturation of ␥␦ TCR ϩ IELs "robs" the epithelium of IEL-derived factors known or postulated to support epithelial barrier function (e.g. TGF-␤3). Gnotobiotic FVB/N Fabpi-TAg Wt mice provide an environmentally and genetically defined, "sensitized" model system for genetic or pharmacologic tests of the role of enterocyte-derived factors postulated to promote maturation of ␥␦ TCR ϩ IELs, of IELderived factors that may affect epithelial barrier function, and/or of microbes or microbially derived products that may influence mucosal biology.
Acknowledgments-We are indebted to Aude Fahrer and co-workers for generously providing the Mu6K GeneChip data sets of IEL gene expression, our colleague Jason Mills for help in analyzing the data sets, and David O' Donnell and Maria Karlsson for superb technical assistance in generating and maintaining the germ-free mice used in this study.