Use of normal and transgenic mice to examine the relationship between terminal differentiation of intestinal epithelial cells and accumulation of their cell cycle regulators.

A spatially well organized continuum of proliferation, differentiation, and death is displayed along crypt-villus units in the adult mouse small intestine. This continuum provides an opportunity to examine in vivo the mechanisms by which proliferative status changes as a function of cellular differentiation. Immunohistochemical studies of normal FVB/N mice revealed that as epithelial cells complete their terminal differentiation during a 48-72-h migration up villi, there is a marked and rapid fall in the levels of two important regulators of the G1/S transition, cyclin D1 and cyclin-dependent kinase (cdk) 2. However, cellular levels of their partners, cdk4 and cyclin E, remain unchanged as does the level of pRB. Adult FVB/N transgenic mice were studied that contained an intestinal fatty acid binding protein gene promoter (Fabpi) linked to wild type Simian virus 40 large T antigen (SV40 TAgWt) or a mutant TAg with Lys for Glu substitutions at residues 107 and 108 (SV40 TAgK107/8) that fails to bind pRB and related pocket proteins. Both transgenes are expressed only in villus enterocytes. SV40 TAgWt causes these terminally differentiated cells to re-enter the cycle. Re-entry is accompanied by a reduction in un/hypophosphorylated pRB, an induction of cyclin D1 and cdk2, but no change in cdk4, cyclin E, or E2F-1. In contrast, SV40 TAgK107/8 fails to induce re-entry and does not produce changes in un/hypophosphorylated pRB, cyclin D1, or cdk2 accumulation. These results suggest that un/hypophosphorylated pRB is an important mediator of the cell cycle arrest that normally occurs as enterocytes exit the crypt and complete their differentiation. Fabpi-directed expression of E2F-1 does not cause villus enterocytes to return to the cell cycle, alter their suppression of cyclin D1 or cdk2, or affect their state of differentiation, emphasizing the insensitivity of these cells to the effects of E2F-1. Analyses of p53−/− and p53+/+ mice containing Fabpi-SV40 TAgWt and Fabpi-SV40 TAgK107/8 established that the proliferation induced by SV40 TAgWt does not require p53 and is associated with increased (p53-independent) apoptosis. The presence of cyclin E and cdk4 in differentiating villus enterocytes emphasizes that these cells retain part of their proliferative heritage expressed 24-72 h earlier in the crypt. The data suggest that down-regulation of cdk2 and/or cyclin D1 expression may be important for control of proliferative status and/or execution of terminal differentiation.

A spatially well organized continuum of proliferation, differentiation, and death is displayed along crypt-villus units in the adult mouse small intestine. This continuum provides an opportunity to examine in vivo the mechanisms by which proliferative status changes as a function of cellular differentiation. Immunohistochemical studies of normal FVB/N mice revealed that as epithelial cells complete their terminal differentiation during a 48 -72-h migration up villi, there is a marked and rapid fall in the levels of two important regulators of the G 1 /S transition, cyclin D 1 and cyclin-dependent kinase (cdk) 2. However, cellular levels of their partners, cdk4 and cyclin E, remain unchanged as does the level of pRB. Adult FVB/N transgenic mice were studied that contained an intestinal fatty acid binding protein gene promoter (Fabpi) linked to wild type Simian virus 40 large T antigen (SV40 TAg Wt ) or a mutant TAg with Lys for Glu substitutions at residues 107 and 108 (SV40 TAg K107/8 ) that fails to bind pRB and related pocket proteins. Both transgenes are expressed only in villus enterocytes. SV40 TAg Wt causes these terminally differentiated cells to re-enter the cycle. Re-entry is accompanied by a reduction in un/hypophosphorylated pRB, an induction of cyclin D 1 and cdk2, but no change in cdk4, cyclin E, or E2F-1. In contrast, SV40 TAg K107/8 fails to induce reentry and does not produce changes in un/hypophosphorylated pRB, cyclin D 1 , or cdk2 accumulation. These results suggest that un/hypophosphorylated pRB is an important mediator of the cell cycle arrest that normally occurs as enterocytes exit the crypt and complete their differentiation. Fabpi-directed expression of E2F-1 does not cause villus enterocytes to return to the cell cycle, alter their suppression of cyclin D 1 or cdk2, or affect their state of differentiation, emphasizing the insensitivity of these cells to the effects of E2F-1. Analyses of p53 ؊/؊ and p53 ؉/؉ mice containing Fabpi-SV40 TAg Wt and Fabpi-SV40 TAg K107/8 established that the proliferation induced by SV40 TAg Wt does not require p53 and is associated with increased (p53-independent) apoptosis. The presence of cyclin E and cdk4 in differentiating villus enterocytes emphasizes that these cells retain part of their proliferative heritage expressed 24 -72 h earlier in the crypt. The data suggest that down-regula-tion of cdk2 and/or cyclin D 1 expression may be important for control of proliferative status and/or execution of terminal differentiation.
The adult mouse small intestinal epithelium undergoes continuous renewal throughout the lifespan of the animal. This renewal takes place in easily identifiable anatomic units. Each unit consists of a finger-like villus that is surrounded at its base by flask-shaped crypts of Lieberkü hn. Cellular proliferation is confined to crypts. Differentiation occurs during a rapid orderly migration up the villus. Cell loss is restricted to the villus tip. This precisely and perpetually maintained stratification of proliferation and differentiation along the crypt-villus axis makes the intestine an attractive model system for examining the relationships between execution of a terminal differentiation program and regulation of expression of mediators of the cell cycle.
The adult mouse small intestine has ϳ1 million crypts (1). The epithelial cell population that overlies each villus is derived from several crypts. Each crypt contains a steady state census of ϳ250 cells derived from one or several active multipotent stem cells (2). These stem cells are functionally anchored near the base of each crypt (3). The stem cell's daughters undergo four to six rounds of cell division (T c ϭ 12 h) in the mid-portion of the crypt forming a transit cell population of ϳ150 cells (3). These daughters adopt one of four fates. Terminal differentiation programs are expressed as nonproliferating cells undergo a bipolar migration. Columnar absorptive enterocytes, mucus-producing goblet cells, and enteroendocrine cells complete their differentiation as they exit each crypt and move up an adjacent villus in vertical coherent columns. The orderliness and rate of this migration is controlled, at least in part, by E-cadherin and its affiliated catenins (4,5). When cells approach the tip of the villus, they enter a death program and are exfoliated into the lumen (6). This entire sequence is completed within a 2-5-day period, e.g. enterocytes, which represent Ͼ80% of small intestinal epithelial cells, complete their terminal differentiation in 48 -72 h (7)(8)(9)(10)(11). Members of the Paneth cell lineage differentiate as they migrate downward from the stem cell zone to the base of crypts where they elaborate antimicrobial peptides, growth factors, and digestive enzymes (12,13). Paneth cells are eliminated through phagocytosis by neighboring cells (13).
Studies in cultured cells have shown that the G 1 /S transition of the mammalian cell cycle is particularly sensitive to external stimuli that regulate growth and differentiation (reviewed in Refs. 14 and 15). There have been a few in vivo studies in defined genetic systems of the relationship between execution of a terminal differentiation program and production of regulators of this (G 1 /S) transition. Loss of the retinoblastoma pro-tein (pRB) in the lens fiber cells of embryonic pRB Ϫ/Ϫ mice is associated with induction of cyclins A, B 1 , and E and their partners p34cdc2 and cdk2, 1 as well as cdk4, but not with any change in D-type cyclins (16). The cells exhibit alterations in differentiation, reenter the cell cycle, and subsequently undergo apoptosis. These findings raise the possibility that terminal differentiation of this lineage is associated with or may even require suppression of cyclins and cdks and that pRB coordinates this process (16). Most cell lineages differentiate normally in cyclin D 1 Ϫ/Ϫ mice, although there are a few reported abnormalities. Retinal cell number is reduced, but morphology appears normal. The mammary epithelium fails to expand during pregnancy but breast development and epithelial differentiation are unaffected (17,18). Studies of E2F-1 Ϫ/Ϫ mice indicate that this effector of entry into S phase is not generally required for normal differentiation. However, testicular atrophy develops in older animals and is associated with oligospermia and overgrowth of Leydig cells. The number of mature T-lymphocytes is increased due to a defect in apoptosis. Nuclear abnormalities are evident in pancreatic and salivary gland acinar cell populations (19,20). Finally, mice homozygous for null alleles of the cyclin-dependent kinase inhibitors p21 WAF1/Cip1/sdi1 or p16 INK4a develop normally (21)(22)(23).
Examining expression of regulators of the G 1 /S transition in vivo provides an opportunity for verifying results obtained in cultured cell lines and for determining whether there are cell lineage-specific differences in expression of these mediators during terminal differentiation. In this report, we describe the patterns of accumulation of pRB, cyclins D 1 and E, and cdks 2 and 4 along the crypt-villus axis of normal adult mice and transgenic animals with a genetically engineered perturbation in the function of pRB and related pocket proteins.

Generation of FVB/N Transgenic Mice
Fabpi-SV40 TAg Transgenics-pRSV-B-Neo3213 (kindly supplied by J. Pipas, University of Pittsburgh) contains a mutant simian virus 40 large T antigen DNA with a Glu 3 Lys substitution at residues 107 and 108 (SV40 TAg K107/8 ). The endogenous polyadenylation site in SV40 TAg K107/8 was eliminated by PCR mutagenesis and the expected substitution (AAATAA3 ATTGCA) spanning nucleotides 2657-2662 of the SV40 genome was confirmed by sequencing the PCR product. A 2.7kilobase pair DNA fragment, containing the SV40 TAg K107/8 open reading frame plus its mutated polyadenylation site, was excised from pRSV-B-Neo3213 with KpnI and BamHI, and introduced into the polylinker region of pSP73 (Stratagene), yielding pSP73-SV40 TAg K107/8 . SV40 TAg K107/8 was removed from this plasmid with BglII and BamHI and subcloned into the BamHI site of pI1178hGH⌬B 2 (4, 5), generating pI1178/SV40 TAg K107/8 /hGH. This placed the SV40 TAg K107/8 open reading frame just downstream from nucleotides -1178 to ϩ28 of the rat intestinal fatty acid binding protein gene (Fabpi Ϫ1178 to ϩ28 , Ref. 24) and linked to nucleotides ϩ3 to ϩ2150 of the human growth hormone (hGH) gene (25). hGH will not be produced from the RNA transcript of Fabpi Ϫ1178 to ϩ28 /SV40 TAg K107/8 / hGH ϩ3 to ϩ2150 , because the initiator Met and the first stop codon are from SV40 TAg K107/8 and because there is no ribosomal re-entry site to reinitiate translation at the downstream initiator ATG of hGH. A 6.2kilobase pair restriction fragment containing Fabpi Ϫ1178 to ϩ21 /SV40 TAg K107/8 /hGH ϩ3 to ϩ2150 was released with XbaI and XhoI, purified by agarose gel electrophoresis followed by glass bead extraction (Geneclean II, Bio101), and subsequently used for pronuclear injections into FVB/N oocytes. Injected eggs were transferred to pseudopregnant Swiss Webster females using standard techniques (26).
Liveborn mice were screened for the presence of the transgene using tail DNA and primers that recognize sequences flanking intron 2 of the hGH gene (5Ј-AGGTGGCCTTTGACACCTACCAGG-3Ј, which hybridizes to the antisense strand of exon 1, and 5Ј-TCTGTTGTGTTTCCTAC-CAGG-3Ј, which anneals to the sense strand of exon 3). The PCR mixture contained 50 mM KCl, 10 mM Tris, pH 8.4, 2 mM MgCl 2 , 2 mg/ml gelatin, 200 M dNTP, 10 M of each primer, 0.7 unit AmpliTaq (Perkin-Elmer), and 1 g of genomic DNA in a final volume of 20 l. The cycling conditions for amplification were: denaturation for 1 min at 94°C, annealing for 1.5 min at 55°C, and extension for 2 min at 72°C for a total of 25 cycles. The transgene generates a 360-base pair PCR product. Four Fabpi-SV40 TAg K107/8 founders were identified from the 90 liveborn mice that were screened. Pedigrees were established from two founders and maintained by crosses to normal FVB/N littermates (lines 7 and 61). Two FVB/N transgenic pedigrees containing Fabpi Ϫ1178 to ϩ28 linked to wild type SV40 TAg (SV40 TAg Wt ) are described in an earlier publication (27).
Fabpi-E2F-1 Transgenics-A 1.8-kilobase human E2F-1 cDNA (generously supplied by M. Rosenberg, Glaxo-Wellcome) was introduced into the unique BamHI site of pI1178hGH⌬B 2 , thereby embedding the E2F-1 open reading frame in exon 1 of the hGH gene and placing it under the control of Fabpi Ϫ1178 to ϩ28 . The fusion gene was separated from vector sequences by digestion with EcoRI and KpnI. The resulting 5.0-kilobase pair restriction fragment was purified as above and used for injection into FVB/N oocytes. Eight transgenic founders were identified among 81 liveborn animals using the same hGH primer pair and PCR protocol employed for Fabpi-SV40 TAg K107/8 mice. Pedigrees were established from four of these founders and maintained by crosses to normal FVB/N littermates.

Maintenance of Animals
All mice were housed in microisolator cages and kept under a strictly controlled light cycle (lights on at 0600 h and off at 1800 h). Animals were given a standard irradiated chow diet (Pico rodent chow 20, PMI Feeds) ad libitum. Routine screens for Hepatitis, Minute, Lymphocytic Choriomeningitis, Ectromelia, Polyoma, Sendai, Pneumonia, and MAD viruses, enteric bacterial pathogens, and parasites were negative. The specified pathogen-free animals were given an intraperitoneal injection of 5-bromo-2Ј-deoxyuridine (BrdU) (120 mg/kg body weight) and 5-fluoro-2Ј-deoxyuridine (12 mg/kg) 90 min prior to sacrifice at postnatal days 42 and 70.

RNA Analyses
Immediately after sacrifice, the entire gastrointestinal tract was removed and divided into stomach, small intestine (subdivided into equal thirds designated duodenum, jejunum, and ileum), cecum, and colon (subdivided into proximal and distal halves). Eleven other tissues were recovered: brain, lung, heart, liver, spleen, pancreas, kidney, spleen, gonads, skeletal muscle, and skin. All tissues were frozen in liquid N 2 and pulverized, and total cellular RNA was isolated using RNAzol (Tel-Test). The integrity of each RNA preparation was verified using denaturing formaldehyde-agarose gel electrophoresis (28).
The concentration of SV40 TAg mRNA was defined in samples of total cellular RNA prepared from the jejunum of Fabpi-SV40 TAg Wt and Fabpi-SV40 TAg K107/8 transgenics using a ribonuclease protection assay. The assay, which is described by Kim et al. (27), employed a 178-base, 32 P-labeled cRNA spanning nucleotides 4886 -5033 of the viral genome and generates a protected fragment that is 127 bases long.
The tissue-specific patterns of Fabpi-SV40 TAg K107/8 and Fabpi-E2F-1 expression were characterized using a reverse transcriptase-PCR assay. The hGH primer pair described above was used together with varying amounts of total cellular RNA and the GeneAmp® Thermostable Reverse Transcriptase RNA PCR kit (Perkin-Elmer) to produce a 150-base pair product. A hot start protocol was used with an initial incubation at 95°C for 2 min prior to beginning the polymerase chain reaction (cycling conditions: 1 min at 95°C, 1 min at 56°C for 35 cycles, followed by 7 min at 72°C).
Antigen-antibody complexes were visualized with alkaline phosphatase-conjugated secondary antibodies using the Western Light kit (Tropix). Blots were scanned with a laser densitometer. Only signals in the linear range of film sensitivity were used to quantitate changes in the steady state levels of immunoreactive proteins. Actin was used as an internal reference control to confirm that equivalent amounts of jejunal proteins were contained in each lane of a blot.

Single and Multilabel Immunohistochemical Studies
6 -10-week-old male and female transgenic mice and their normal littermates were studied (n ϭ 3 transgenic and 3 normal littermates/ pedigree). The entire gastrointestinal tract was removed en bloc immediately after sacrifice, flushed with ice-cold PBS, fixed in Bouin's solution for 6 -12 h, and then washed with 70% ethanol. The small intestine was opened with an incision along its cephalocaudal axis and then rolled up from its proximal to distal end. The resulting Swiss rolls were embedded in paraffin, and 5-m-thick serial sections were cut.
Two other fixation protocols were also used: (i) following a PBS wash, the opened intestine was placed in 10% buffered formalin (Fisher) and incubated at room temperature for 8 -10 h; (ii) 1-2-cm segments were taken from the mid-portion of the duodenum, jejunum, and ileum, quick frozen at Ϫ80°C with Cytocool spray (Stephens Scientific), embedded in OCT (Miles), and 5-8-m sections were cut in a cryostat and then fixed in ice-cold methanol for 10 min. E2F-1 was detected in frozen, methanol-fixed samples of adult human jejunum (30).
The methods used for single and multilabel immunohistochemical surveys are detailed in Refs. 4 and 31. Unmasking steps were employed to visualize pRB, cyclin D 1 , and cyclin E. To detect pRB and cyclin E, sections were placed in 1 N HCl for 20 min at 24°C followed by a 15-min incubation at 37°C in digestion buffer (0.5 mg of bovine pancreas ␣-chymotrypsin (Sigma)/ml of 7 mM CaCl 2 , pH 7.8). Detection of cyclin D 1 required that sections be incubated in Target Unmasking Solution (Pharmingen) for 10 min at 90°C.
A large panel of lectins were used to define the terminal differentiation programs of the four principal intestinal epithelial cell lineages in normal and transgenic mice. This lectin panel as well as the lectins' lineage-specific, differentiation-dependent, and duodenal-ileal patterns of reactivity in FVB/N small intestine are described in Ref. 31. Lectins were used at a final concentration of 5 g/ml blocking buffer.

Confocal Microscopy
A Molecular Dynamics Multiprobe 2001 inverted confocal microscope system was used to scan sections of Bouin's solution-or buffered formalin-fixed small intestine prepared from normal and transgenic mice and stained as described above. Section series scans were collected at 0.6-m intervals.

Quantitation of Apoptosis
Terminal deoxynucleotidyl transferase-mediated, dUTP nick end labeling (TUNEL) assays were performed on sections prepared from Swiss rolls (fixation: buffered formalin for 8 -10 h at 24°C). The TUNEL assay protocol of Gavrieli et al. (33) was used with the exception that the sections were incubated with proteinase K (Boehringer Mannheim; 20 g/ml 0.1 M Tris, pH 7.5) for 20 min at room temperature. Digoxigenin-labeled dUTP was used for the in situ terminal transferase reaction. Incorporation was detected with peroxidase-conjugated sheep anti-digoxigenin Fab fragments (Boehringer Mannheim, diluted 1:500 in blocking buffer) and the Vector VIP kit (Vector Laboratories). Sections were counterstained with methyl green (Zymed).
Apoptosis was quantitated by counting TUNEL-positive epithelial cells with apoptotic morphology (6) in all intact jejunal crypt-villus units on at least two nonadjacent sections of a given Swiss roll (n ϭ 3-5 animals, 1600 -3000 crypt-villus units/genotype). All quantitation of apoptosis was done in a blinded fashion by two observers. Mean values Ϯ S.E. were calculated. Statistically significant differences between mice with different genotypes were identified by Student's t test (SigmaPlot).

Expression of Cell Cycle Regulators along the Crypt-Villus
Axis of Normal Adult FVB/N Mice-The crypt-villus units of 6 -10-week-old FVB/N mice were used to initially examine the relationship between execution of a terminal differentiation program and modulation of steady state levels of regulators of the G 1 /S transition.
Jejunal sections were probed with peptide-specific antibodies that do not discriminate between hyper-and un/hypophosphorylated forms of pRB. Intense nuclear staining was observed in crypt as well as villus epithelial cells (Fig. 1, A-C). The intensity of nuclear staining did not vary appreciably from the crypt to the villus tip (Fig. 1A). However, there was a change in intracellular distribution. pRB is present in the cytoplasm and nucleus of epithelial cells located in the crypt and lower villus. Cytoplasmic staining is lost as differentiation is completed (Fig. 1C).
We were not able to detect the pRB-related proteins p107 and p130 (35,36) in jejunal (or duodenal or ileal) epithelium using commercially available antibodies, three different tissue fixation protocols, and two different methods for antigen unmasking.
Studies in cultured cells have shown that cyclin D 1 ⅐cdk4/6 complexes promote the initial hypophosphorylation of pRB in G 1 (37,38). Cyclin D 1 falls below detectable levels just as post-mitotic epithelial cells exit the crypt and begin their terminal differentiation (Fig. 1D). Although levels of cdk4 are higher in the crypt compared with the villus, cdk4 concentrations do not change appreciably as cells move from the cryptvillus junction to the villus tip (Fig. 1, E and F).
Cyclin E⅐cdk2 complexes catalyze further phosphorylations of pRB in late G 1 and are thought to drive entry into S phase (39). Steady state concentrations of cdk2 are highest in crypt epithelial cells. Levels fall rapidly as cells emerge from the crypt and become undetectable by the time that they pass through the lower quarter of the villus (Fig. 1G). Unlike cdk2, cyclin E concentrations do not change as cells journey from the crypt to the villus tip (Fig. 1, H-J). Cyclin E was not detectable in adult FVB/N hepatocytes using the same immunohistochemical methods and reagents (data not shown). In cultured cells, the absence of cyclin E has been used as part of the "signature" of being in G o (40). The presence of cyclin E in terminally differentiating villus enterocytes suggests that they may reside in G 1 rather than in G o during their short lifespan. Un/hypophosphorylated pRB is able to associate with members of the E2F family of transcription factors (41-47). Hyperphosphorylation of pRB in late G 1 leads to the release of E2F, which forms a heterodimeric complex with members of the DP family of proteins (48 -51). These complexes are able to activate genes that participate in entry into and passage through S phase (48,50,52). E2F-1 is detectable in the small intestine of normal adult FVB/N mice by both reverse transcriptase-PCR assays and by Western blotting (Fig. 2). We were unable to detect E2F-1 in sections of duodenum, jejunum, or ileum using the anti-E2F-1 serum employed for Western blotting. However, surveys of adult human jejunum demonstrated comparable levels of E2F-1 in the crypt and villus epithelium (data not shown).
In summary, the immunohistochemical analyses indicate that terminal differentiation of the enterocytic lineage is accompanied by a marked and rapid fall in levels of cyclin D 1 and cdk2. The data also reveal that differentiated cells retain the "partners" of these regulators of the G 1 /S transition, i.e. cdk4 and cyclin E, respectively. The proliferative potential of differentiated enterocytes is maintained in the sense that they have a reservoir of cdk4 and cyclin E. An alternative view is that retention of cdk4 and cyclin E (or loss of cyclin D 1 and cdk2) may contribute to the execution of a terminal differentiation program.
Evidence That pRB Is a Critical Regulator of Proliferative Note the punctate nuclear staining for this cyclin. A similar punctate staining pattern has been noted in the nuclei of HeLa cells (34). J, blocking control performed by preincubating the rabbit anti-cyclin E with its peptide antigen prior to application to a section of a jejunal villus. Note that analogous blocking controls established that preincubation of the cyclin D 1 , cdk4, and cdk2 antibodies with the antigens used for their generation produced no detectable staining of sections (data not shown). Bars, 25 m.
Status along the Crypt-Villus Axis-Although differentiation does not produce an appreciable change in cellular pRB levels, the shift in its intracellular localization from cytoplasmic and nuclear to exclusively nuclear was a consistent finding that was verified by confocal light microscopy. Studies in cultured cell lines have shown that G 1 /S phosphorylation of pRB reduces its binding affinity for nuclear structures, whereas the un/ hypophosphorylated form is an avid binder (53,54). These results led us to speculate that the shift to an exclusively nuclear localization may indicate the presence of an increasing fraction of un/hypophosphorylated pRB in terminally differentiating villus enterocytes.
Wild type and a mutant SV40 T antigen was used to assess the role of pRB in maintaining the growth-arrested state of terminally differentiating enterocytes. SV40 TAg Wt binds pRB, p107, and p130 (55,56). Substitution of Lys for Glu at residues 107 and 108 abolishes the ability of SV40 TAg to bind these three proteins (55,57). Nucleotides Ϫ1178 to ϩ28 of rat Fabpi were used to direct expression of SV40 TAg Wt or SV40 TAg K107/8 in post-mitotic villus enterocytes. Previous studies of FVB/N transgenic mice containing Fabpi Ϫ1178 to 28 /reporter fusion genes have shown that reporter production is first activated on embryonic day 15, is sustained for the next 2 years of life, and extends from the duodenum to the ileum, with highest levels of reporter mRNA and protein accumulation occurring in the jejunum (e.g. Refs. 27 and 58). Expression occurs only in the enterocytic lineage, commences just as these cells emerge from the crypt, and continues throughout their migration-associated differentiation.
Fabpi Ϫ1178 to 28 -directed expression of SV40 TAg Wt results in a return of villus enterocytes to the cell cycle. This re-entry can be defined by the labeling of S phase cells with BrdU (Fig. 3, A and B) or by the appearance of M phase cells. The ratio of BrdU ϩ :M phase cells in jejunal villi was 40, whereas the ratio in crypts, where the transgene is not expressed, was 12. This Cy5-donkey anti-rabbit Ig: the color was arbitrarily assigned). Enterocytes with green nuclear staining for wild type SV40 TAg are distributed along the length of the villus. Expression of the viral protein is associated with re-entry of enterocytes into the cell cycle (defined by incorporation of BrdU; e.g. arrows). (The sections shown in A and B were prepared from Bouin's solution-fixed tissue.) C, a jejunal villus from an Fabpi-SV40 TAg Wt transgenic mouse stained with rabbit anticyclin D 1 serum (detected with Cy3-donkey anti-rabbit Ig). Expression of SV40 TAg Wt results in persistent accumulation of this cyclin, which is normally lost during terminal differentiation of the enterocytic lineage (see Fig. 1D). D, jejunal villus from the same transgenic mouse shown in C but incubated with rabbit anti-cdk2 (visualized with Cy3donkey anti-rabbit Ig). The normal disappearance of cdk2 associated with enterocytic differentiation no longer takes place when SV40 TAg Wt is present (arrowheads point to a crypt-villus junction; compare with Fig. 1G). (The sections shown in C and D were generated from buffered formalin-fixed tissue.) E, confocal image of a Bouin's fixed section from a transgenic mouse expressing the mutant SV40 TAg K107/8 in their villus enterocytes. The lower half of a crypt-villus unit is shown. The section was incubated with goat anti-BrdU (detected as red with Cy3donkey anti-goat Ig) and rabbit anti-SV40 TAg (visualized with Cy5donkey anti-rabbit Ig). S phase cells are confined to the crypt. Immunoreactive SV40 TAg K107/8 (green) is only expressed after enterocytes have exited the crypt. The absence of BrdU-positive villus enterocytes indicates that production of the mutant viral protein is not associated with re-entry into the cell cycle. F, formalin-fixed section from an Fabpi-SV40 TAg K107/8 mouse stained with rabbit anti-cyclin D 1 and Cy3-donkey anti-rabbit Ig. Unlike the transgenic mouse expressing SV40 TAg Wt shown in C, cyclin D 1 is not detectable in terminally differentiating villus enterocytes that produce SV40 TAg K107/8 . Bars, 25 m. statistically significant difference (p Ͻ 0.05) suggests that cell cycle kinetics may differ between undifferentiated SV40 TAg-negative crypt and differentiated SV40 TAg-positive villus populations (59).
Western blots of total jejunal proteins prepared from 6 -10week-old FVB/N Fabpi-SV40 TAg Wt mice and their normal littermates revealed that SV40 TAg Wt -mediated cell cycle reentry is associated with a 2-3-fold increase in steady state levels of cyclin D 1 and cdk2 (Fig. 2). Immunohistochemical studies indicated that both proteins are induced in SV40 TAg Wt -positive villus enterocytes (compare Fig. 3, C and D, with Fig. 1, D and G). The steady state level of cyclin D 1 's partner, cdk4, increases by less than 50% (Fig. 2), and there is no change in its normal distribution along the crypt-villus axis (data not shown). In addition, cdk2's partner, cyclin E, shows no appreciable change in its level or cellular distribution ( Fig.  2 and data not shown).
Western blots of normal and transgenic jejunal extracts were also probed with an antibody specific for the un/hypophosphorylated form of pRB. SV40 TAg Wt expression in villus enterocytes is associated with a 4-fold reduction in the steady state concentration of this pRB species (e.g. Fig. 2; n ϭ 4 animals). Unfortunately, this antibody preparation was not useful for detecting un/hypophosphorylated pRB by immunohistochemistry: negative results were obtained when sections of jejunum, prepared from normal or transgenic littermates, were subjected to each of the three different tissue fixation or two antigen unmasking protocols and when several different sensitive fluorescence techniques were employed to detect antigenantibody complexes (e.g. Cy3-labeled secondary antibodies). Western blots of jejunal extracts disclosed that the depletion of un/hypophosphorylated pRB in SV40 TAg Wt enterocytes is not accompanied by a change in the steady state level of E2F-1 (Fig. 2).
Adult members of two pedigrees of FVB/N Fabpi-SV40 TAg K107/8 mice were analyzed in an analogous fashion. Steady state levels of the mutant SV40 TAg mRNA and protein were equivalent to the steady state levels of SV40 TAg Wt mRNA and protein at comparable positions along the duodenal-ileal axis (Fig. 4). As with SV40 TAg Wt , expression of SV40 TAg K107/8 was confined to villus enterocytes (Fig. 3E). However, surveys of the entire duodenal-ileal axis indicated that the mutant T antigen failed to produce re-entry into the cell cycle (Fig. 3E) or induction of cyclin D 1 or cdk2 ( Fig. 2; plus compare Fig. 3F with Fig.  1D). SV40 TAg K107/8 had no effect on cyclin E or cdk4 expression, nor did it change the steady state levels of un/hypophosphorylated pRB or E2F-1 ( Fig. 2 and data not shown).
These results suggest that SV40 TAg-mediated cell cycle re-entry is dependent upon its ability to interact with pRB. Studies with cultured cells have shown that the shift in pRB phosphorylation that occurs during terminal differentiation involves a down-regulation of cyclin and cyclin-dependent kinase activity and that this down-regulation plays a role in modulating cellular differentiation (60). Our in vivo data support this notion. They suggest that terminally differentiated (G 1 -arrested) enterocytes may maintain pRB in an active un/ hypophosphorylated state by down-regulating expression of cdk2 and cyclin D 1 rather than by changing the levels of their cyclin E and cdk4 partners. The persistence of cdk4 in normal FVB/N villus epithelial cells in the absence of detectable cyclin D 1 raises a question about whether cdk4 exists partnered to other members of the cyclin D family. We were unable to detect cyclins D2 and D3 in normal (or transgenic) small intestine by Western blot analyses or by immunohistochemistry. This does not rule out the possibility that other as yet unidentified cyclins may be expressed in this epithelium.
Forced Expression of E2F1 Is Unable to Induce Villus Enterocytes to Re-enter the Cell Cycle-Field et al. (20) noted that E2F-1 Ϫ/Ϫ mice have no abnormalities in crypt proliferation as judged by BrdU labeling. They were unable to detect any histologic changes in crypt-villus units, suggesting that withdrawal of E2F-1 from villus enterocytes does not affect their differentiation program. Expression of human E2F-1 in cultured, serum-starved, quiescent rat embryo fibroblasts not only induces their proliferation (61)(62)(63) but also results in their dedifferentiation and transformation (62,64). We performed an equivalent in vivo experiment in growth-arrested terminally differentiated enterocytes to determine whether these cells are at all responsive to E2F-1.
Four pedigrees of Fabpi-E2F-1 transgenic mice were established. Reverse transcriptase-PCR analyses established that transgene expression was confined to the small intestine in all pedigrees. Western blot analyses confirmed that human E2F-1 was being produced (Fig. 5A). Forced expression of this E2F family member did not affect the steady state levels of endogenous mouse E2F-1 (Fig. 5A), nor did it induce proliferation of villus enterocytes as judged by the absence of BrdU incorporation and M phase cells (Fig. 5B). E2F-1 activates both the cyclin D 1 and E promoters in cultured cells (65). Immunohistochemical surveys of Fabpi-E2F-1 mice showed no induction of cyclin D 1 (or cdk2) in villus enterocytes (Fig. 5C).
Sections of intestine prepared from these transgenic mice were probed with a large panel of lectins that provide a comprehensive definition of the state of differentiation of each of the gut's four principal epithelial cell lineages (31). Single-and multi-label immunohistochemical studies revealed that forced expression of E2F-1 produced no detectable changes in the state of enterocytic differentiation (data not shown). DP family members must heterodimerize with E2F molecules to generate functional transcription factor complexes. We were unable to detect DP-1 in the small intestine of normal, Fabpi-E2F-1 (or Fabpi-SV40 TAg Wt ) mice by Western blotting or by immunohistochemistry (data not shown). This finding is consistent with a recent report that DP-1 mRNA is not detectable in normal adult mouse intestine (67). Together, our data indicate that the villus enterocyte is not dependent upon or responsive to E2F-1. This insensitivity may reflect a lack of DP-1 or other DP partners.
Cell Cycle Re-entry Caused by Inactivation of pRB Is Associated with a p53-independent Increase in Apoptosis-Loss of pRB function in differentiated lens fiber cells or choroid plexus epithelial populations, whether produced by gene knockout or by oncoprotein-mediated inactivation, is accompanied by an increase in a p53-dependent apoptosis. The apoptosis has been viewed as a compensatory response to the unscheduled proliferation (68 -70).
The inter-relationships between pRB function and the control of proliferative status and apoptosis in terminally differentiated enterocytes was assessed by quantitating apoptosis in normal, Fabpi-SV40 TAg Wt , and Fabpi-SV40 TAg K107/8 jejunal villi. Apoptosis was scored using the TUNEL assay (33) as well FIG. 4. Comparison of steady state levels of SV40 TAg Wt and SV40 TAg K107/8 in the jejunum of adult transgenic mice. Western blot of jejunal protein extracts (80 g protein/lane) were probed with a biotinylated mouse monoclonal antibody to SV40 TAg. Two 10-week-old mice from an Fabpi-SV40 TAg Wt pedigree and two 10-week-old mice from an Fabpi-SV40 TAg K107/8 pedigree were analyzed. Antigen-antibody complexes were detected with alkaline phosphatase-conjugated streptavidin and a chemiluminescent enzyme substrate. as morphologic criteria (6). In nontransgenic littermates, TUNEL-positive cells with apoptotic morphology were seen at the tips of villi, confirming previous observations (6,33). The number and distribution of apoptotic cells in jejunal villi were not significantly different between 6 -10-week-old p53 ϩ/ϩ Fabpi-SV40 TAg K107/8 transgenic mice and their nontransgenic littermates (Fig. 6A). The number of apoptotic cells was increased 4-fold in Fabpi-SV40TAg Wt villi (p Ͻ 0.05 compared with either nontransgenic or Fabpi-SV40 TAg K107/8 animals). Apoptotic cells were typically located in the lower and middle thirds of these villi (Fig. 6, B and C). No statistically significant differences in apoptosis were noted between the jejunal crypts of age-matched Fabpi-SV40 TAg Wt , Fabpi-SV40 TAg K107/8 and nontransgenic mice (data not shown).
Quantitative histologic surveys of BrdU-labeled p53 Ϫ/Ϫ and p53 ϩ/ϩ mice containing wild type and mutant SV40 TAg transgenes revealed that removal of p53 did not affect the proliferative phenotype of their villi (data not shown). TUNEL assays indicated that removal of p53 did not produce a significant change in the magnitude of the apoptotic response to SV40 TAg Wt or the levels of apoptosis in SV40 TAg K107/8 or nontransgenic jejunal villi (Fig. 6A).
These observations establish that SV40 TAg Wt induces a pRB-dependent, p53-independent proliferation of villus enterocytes. The triggering mechanism and the effectors of the apoptosis associated with this proliferation have yet to be determined, but they do not involve p53.
Prospectus-A major conclusion from our studies is that pRB is an important determinant of proliferative status during terminal differentiation of the enterocytic lineage and that this terminal differentiation is associated with loss of cyclin D 1 and cdk2 but not their partners, cdk4 and cyclin E. The latter observations elicit a number of questions about the functional significance of the differentiation-associated loss of cyclin D 1 and cdk2. The loss of cyclin D 1 observed in normal villus enterocytes is consistent with the normal intestinal phenotype observed in mice homozygous for a cyclin D 1 null allele (18,70). The presence of cyclin D 1 's and cdk2's partners, cyclin E and cdk4, in terminally differentiating villus epithelial cells emphasizes that these cells retain part of their "proliferative" heritage, which was expressed 24 -72 h earlier in the crypt. Understanding how expression of these regulators is controlled along the crypt-villus axis should provide more profound insights about the mechanisms that permit the continuum between proliferation and differentiation (and death) to be established and maintained in this dynamic, self-renewing system.  5. Effects of forced expression of human E2F-1. A, Western blot of protein extracts prepared from the proximal, middle, and distal thirds of the small intestine (duodenum, jejunum, and ileum) of FVB/N Fabpi-human E2F-1 transgenic mice and their normal FVB/N littermates. The blots (80 g protein/lane) were probed with rabbit anti-E2F-1 antibodies that recognize the human and mouse proteins. The marked difference in mobilities of the orthologous proteins has been noted previously (66). When replicate blots were probed with E2F-1 antibodies that had been preincubated with the peptide used for their generation, neither the mouse nor the human species indicated by the arrows were detected (data not shown). B, jejunal crypt-villus units from an Fabpi-E2F-1 transgenic mouse that had been pulse labeled with BrdU 1.5 h prior to sacrifice. BrdU-positive cells (visualized with goat anti-BrdU and FITC-donkey anti-goat Ig) are confined to the crypt (e.g. arrows). C, jejunal crypt-villus unit from the same mouse as in B stained for cyclin D 1 using the reagents described in Fig. 1D. This cyclin remains confined to crypt epithelial cells (e.g. arrows). Bars, 25 m.
FIG. 6. Effects of SV40 TAg Wt and SV40 TAg K107/8 on apoptosis in villus enterocytes homozygous for p53 wild type or null alleles. A, apoptosis was defined in sections of Swiss rolls prepared from the jejunum. Sections were stained using the TUNEL assay, and apoptotic cells (6) were counted on villi. Statistically significant differences (p Ͻ 0.05 when compared with nontransgenic p53 ϩ/ϩ mice) are indicated by asterisks. B, hematoxylin and eosin-stained section of a jejunal villus from a Fabpi-SV40 TAg Wt (p53 ϩ/ϩ ) mouse. Apoptotic cells (e.g. arrows) are evident in the lower half of the villus. C, a TUNEL-positive cell (closed arrow) and an M phase cell (open arrow) are shown in the mid-portion of a jejunal villus from the same mouse as in B. Bars, 25 m. for many helpful suggestions throughout the course of this study, and Mike Rosenberg for continued support.