Simian Virus 40 T-Antigen-Mediated Gene Regulation in Enterocytes Is Controlled Primarily by the Rb-E2F Pathway

Abstract

Simian virus 40 large T antigen (TAg) contributes to cell transformation, in part, by targeting two well-characterized tumor suppressors, pRb and p53. TAg expression affects the transcriptional circuits controlled by Rb and by p53. We have performed a microarray analysis to examine the global change in gene expression induced by wild-type TAg (TAg^wt) and TAg mutants, in an effort to link changes in gene expression to specific transforming functions. For this analysis we have used enterocytes from the mouse small intestine expressing TAg. Expression of TAg^wt in the mouse intestine results in hyperplasia and dysplasia. Our analysis indicates that practically all gene expression regulated by TAg in enterocytes is dependent upon its binding and inactivation of the Rb family proteins. To further dissect the role of the Rb family in the induction of intestinal hyperplasia, we have screened several lines of transgenic mice expressing a truncated TAg (TAg^N136), which is able to interfere with the Rb pathway but lacks the functions associated with the carboxy terminus of the protein. This analysis confirmed the pivotal association between the Rb pathway and the induction of intestinal hyperplasia and revealed that upregulation of p53 target genes is not associated with the tumorigenic phenotype. Furthermore, we found that TAg^N136 was sufficient to induce intestinal hyperplasia, although the appearance of dysplasia was significantly delayed.

The large tumor antigen (TAg) encoded by simian virus 40 (SV40) induces transformation of multiple cell lines as well as tumors in experimental animals, and therefore it serves as an important tool to understand the mechanisms of cellular transformation (1). TAg induces transformation, in part, by binding to and disabling the functions of tumor suppressors such as pRb and p53 (6, 7, 14, 22, 23). The Rb protein family (pRb, p107, and p130) regulates cell cycle entry and progression by repressing the E2F family of transcription factors. This, in turn, blocks the expression of a large collection of cellular genes that are E2F dependent. In the absence of active Rb proteins, E2F-dependent genes are expressed, resulting in S-phase entry and progression through the cell cycle. TAg is thought to stimulate cell proliferation by blocking the ability of pRb, p107, and p130 to repress E2F-dependent transcription. The loss of Rb-mediated growth suppression often results in the stabilization of p53 and consequently in a large increase in p53 steady-state levels. Under normal circumstances, this would lead to the expression of p53-dependent genes that arrest cell cycle progression and induce apoptosis. However, TAg binds to p53 and blocks its ability to stimulate gene expression. Thus, TAg-expressing cells proliferate and survive under conditions that would result in growth arrest and/or apoptosis of nontransformed cells.

While the roles of TAg in blocking pRb and p53 functions in SV40-mediated transformation are well established, it is also clear that action on additional cellular factors is sometimes required. For example, TAg interactions with Bub1, Cul7, and CBP/p300 have been implicated in SV40-mediated transformation (2, 5, 24). Furthermore, the SV40-encoded small TAg, which targets the cellular protein phosphatase pp2A, is required for transformation under some conditions (27).

Mouse intestinal epithelial cells offer a useful model for understanding how TAg induces neoplastic transformation. The mouse intestinal epithelium is organized into numerous finger-like projections, the villi, and the structures responsible for their renewal, the crypts of Lieberkühn. Intestinal villi and crypts are localized in different regions of the tissue and therefore can be readily isolated from the underlying submucosa and muscularis. This allows us to prepare proteins or nucleic acids from cell populations greatly enriched for nonproliferating, terminally differentiated cells located in the villi or from their proliferating, multipotent progenitors located in the crypts (20, 33).

We and others have previously described the generation of a series of transgenic mice expressing wild-type TAg (TAg^wt) or TAg mutants (TAg¹¹³⁷, TAg³²¹³, and TAg^D44N) in terminally differentiated enterocytes, using the intestinal fatty acid binding protein promoter (Fig. 1) (12, 15, 16, 26). Under these conditions, expression of TAg extends from the base of the villi to the apical extrusion zone of the villi (12). These transgenic lines do not express detectable levels of small TAg (4). Expression of TAg^wt results in intestinal hyperplasia that progresses to dysplasia by 4 to 6 months of age, while TAg³²¹³, an LXCXE motif mutant unable to bind to the Rb family of proteins, and TAg^D44N, a J domain mutant unable to inactivate Rb family members, do not result in any observable phenotype (15, 26). Thus, TAg-mediated transformation of enterocytes requires pRb binding and a functional J domain. Furthermore, the ability of TAg^wt to induce ectopic enterocyte proliferation depends on the presence of E2F2 (28). Surprisingly, p53 is not stabilized by TAg in enterocytes, nor do TAg-p53 interactions play any role in enterocyte transformation (20).

FIG. 1.

Domain maps of SV40 TAg^wt and mutants, showing the J domain (J), Rb protein (pRb, p130, and p107) binding motif (LXCXE), nuclear localization signal (NLS), origin binding domain (OBD), Zn domain (Zn), ATPase domain (ATPase), and host range domain (HR). Gray circles represent cellular proteins interacting with TAg or mutants. The phenotypic effects of expression of TAg or the mutants in intestinal enterocytes are indicated.

We have also characterized a single line of transgenic mice expressing TAg¹¹³⁷, a truncated TAg that is capable of inactivating Rb proteins and stimulating cell proliferation. Like TAg^wt, TAg¹¹³⁷ induces enterocyte proliferation and intestinal hyperplasia (15). However, these mice did not progress to dysplasia, suggesting a role for the carboxy-terminal portion of TAg in tumorigenic progression. However, only one line of TAg¹¹³⁷ mice was analyzed, and it is therefore possible that the failure to progress to dysplasia was specific to this line. In this study we report the characterization of independent lines of mice expressing an amino-terminally truncated TAg mutant, and we compare the gene expression patterns from transgenic mice expressing TAg^wt or different mutants of TAg.

MATERIALS AND METHODS

Production and maintenance of transgenic mice.

Transgenic mice expressing SV40 TAg^wt under the control of the rat intestinal fatty acid binding protein promoter (pedigree 103) were obtained from Jeff Gordon (16). The construction of transgenic mice expressing SV40 TAg^N136, TAg³²¹³, and TAg^D44N under the control of the rat intestinal fatty acid binding protein promoter is described elsewhere (26, 28). For simplification, we refer to the mice as TAg^wt (Fabpi-SV40 TAg^wt), TAg^N136 (Fabpi-SV40 TAg^N136), TAg³²¹³ (Fabpi-SV40 TAg³²¹³), and TAg^D44N (Fabpi-SV40 TAg^D44N). All lines were maintained by crosses to nontransgenic FVB mice. The genotype of each mouse was determined by PCR amplification as described previously (28). All mice were between the ages of 2.4 and 5 months.

Preparation of intestinal villi samples by LCM.

The intestine was removed in its entirety from the abdominal cavities of mice, from the exit of the stomach to the entry into the large intestine. The intestines were then opened along their cephalocaudal axis, washed with phosphate-buffered saline, and rolled from the duodenum to the ileum. Rolls from TAg^wt and TAg mutant transgenic mice were then processed to isolate villus cell types by laser capture microscopy (LCM). Briefly, the rolls were quickly frozen and stored at −80°C. Frozen sections were cut and immediately dehydrated in graded alcohol solutions. The dried sections were examined microscopically, and the top two-thirds of 50 villous segments from each section were collected using a laser capture microscope (Pix Cell II; Arcturus, Mountain View, CA). The captured tissue was extracted in Pico Pure extraction buffer (Arcturus) according to the manufacturer's directions and stored at −80°C.

RNA extraction.

For the microarray experiments, total RNA was isolated from LCM-prepared villi samples with the PicoPure kit (Arcturus) and amplified twice with the RiboAmp kit (Arcturus). Independent RNA preparations were made from three nontransgenic mice and from three TAg^D44N, three TAg³²¹³, four TAg^wt, and five TAg^N136 transgenic mice.

Microarrays.

Amplified total RNA was sent to the Genomics and Proteomics Core Laboratories (University of Pittsburgh) for hybridization to the Mouse 430 2.0 whole genome array (Affymetrix), which contains 45,101 probe sets representing 21,635 unique genes. CEL files for each array were converted into RMA expression values using BRB-Array Tools (Rich Simon, National Cancer Institute; http://linus.nci.nih.gov/BRB-ArrayTools.html). An average fold change ratio (experimental/control) and a one-sample t test were calculated for each probe set.

Microarray analysis.

The Affymetrix mouse whole-genome chip consisting of 21,635 unique genes was used for microarray analysis. Genes which are threefold up- or downregulated in an individual experimental class (TAg^wt/TAg^N136/TAg³²¹³/TAg^D44N) were selected for further analysis. We applied additional criteria of present and absent calls provided by Affymetrix data files. To be included for consideration, an upregulated gene needed to be present in all the replicates of an experimental class and a downregulated gene needed to be present in all the replicates of the control class (nontransgenic mice).

To analyze the effect of individual domains/regions of TAg on gene expression, we did independent comparisons between TAg^wt and each mutant. Genes with a TAg^wt/control ratio of greater than or equal to 3 and a TAg mutant/control ratio of less than 1.2 were considered upregulated by TAg but not by the mutant. Genes upregulated by the mutant were identified similarly. Furthermore, genes with a TAg^wt/control ratio of less than or equal to 0.33 and a mutant/control ratio of greater than 0.8 were considered downregulated by TAg^wt but not by the mutant. Genes downregulated by the mutant were identified similarly. This strict selection criterion was used in order to identify genes which are up- or downregulated by one class but not by the other. In addition, we have considered present and absent calls as described above.

Reverse transcription-PCR (RT-PCR) analysis.

cDNA synthesis from 1 μg of total RNA was performed using Superscript II reverse transcriptase (Invitrogen). PCR was performed with equal amounts of cDNA using GoTaq polymerase (Promega) for 25 cycles with specific primers for the different transcripts. Amplification with primers for the alcohol dehydrogenase 5 (Adh5) transcript was used as a normalizing control. PCR products were resolved through a 2% agarose gel in 1× Tris-acetate-EDTA and stained with GelStar (Cambrex Bio Science). The cDNAs were amplified with PCR using primers specific for Adh5 (5′-TGCACCACCAACTGCTTAG and 5′-GATGCAGGGATGATGTTC), Fas (5′-GCCGAATGTCGCAGAACCTTAG and 5′-CAGGAGTTGCCAATGTCAATACAG), Lrdd (5′-GAGAGCAGGTTGCCCATACTTG and 5′-GAGCCATCAGGTAGGTAGAGGAAAC), Noxa (5′-TTGTTTGGAGACAAGGGTCCC and 5′-GACCATAAATACCCATTGGGCAAG), and Pten (5′-ATTTGGTCACCCTTTGAGTCCTC and 5′-AGTTTGTCTTGTTGTTAGCCCACC).

To ensure that these reactions were within the linear range of the assay, we optimized the number of cycles required to obtain nonsaturated signals. Exponential amplifications of PCR products were obtained as follows: 2 min at 94°C; a series of 25 cycles of 94°C for 30 s, variable annealing temperatures for 30 s, and 72°C for 30 s; and a final extension step of 5 min at 72°C. The annealing temperatures and product sizes were as follows: 58°C and 152 bp for Adh5, 54.4°C and 204 bp for Fas, 55.7°C and 151 bp for Lrdd, 52.8°C and 189 bp for Noxa, and 52.9°C and 231 bp for Pten. The products were resolved on 2% agarose gels and stained with GelStar (BioWhittaker Molecular Applications).

Immunoblot analysis.

Material enriched in intestinal villi and crypts was prepared and analyzed by Western blotting as described previously (20). An appropriate dilution of TAg-mouse monoclonal PAb416 (11) was used as a primary antibody. Goat anti-mouse antibody linked to peroxidase (Sigma A2554) was used as a secondary antibody and developed with ECL (GE Healthcare) according to the manufacturer's instructions.

Immunohistochemistry.

Small intestines were processed for immunohistochemistry as described elsewhere (20). Assessment of the proliferative status was obtained after mice were given an intraperitoneal injection of 5-bromo-2′-deoxyuridine (BrdU) (Sigma) (120 mg/kg of body weight) 2 hours prior to their sacrifice. Sections were stained with monoclonal rat BU1/75 (dilution, 1:2,000) (Accurate Scientific, Westbury, NY). Antigen-antibody complexes were detected with an ABC Elite Kit Rat (Vector Labs, Burlingame, CA), and development of the peroxidase reaction was performed with diaminobenzidine substrate (DAKO). Stained sections of murine intestines were photographed under a Nikon FXA microscope (original magnification, ×200).

Detection of TAg proteins was performed in 5-μm-thick histological sections using appropriate dilutions of the anti-TAg antibody PAb419 (for TAg^wt) (11) and PAb416 (for TAg^N136) as primary antibodies. Biotinylated anti-mouse-Fab was used as the secondary antibody, followed by use of the streptavidin-peroxidase ARK kit plus development of the peroxidase reaction with diaminobenzidine substrate (DAKO), according to the manufacturer's instructions.

Microarray data accession number.

The microarray data have been submitted to GEO under accession number GSE16389.

RESULTS

Global patterns of cellular gene expression in villus enterocytes of transgenic mice expressing TAg^wt or mutant TAgs.

The intestinal villus enterocytes are composed mainly of nonproliferating, terminally differentiated cells. Expression of full-length TAg in murine enterocytes forces a nonproliferating cell population to restart the cell cycle and results in intestinal hyperplasia that progresses to dysplasia as the animal ages. We have analyzed the RNA profile of villus enterocytes of transgenic mice expressing TAg^wt or mutants using microarray technology. The Affymetrix GeneChip Mouse Genome 430 2.0 Array platform consisting of 21,635 unique genes was used for the microarray analysis. Genes whose expression levels changed at least threefold up or down relative to control nontransgenic enterocytes were considered to be regulated by TAg. Enterocytes isolated from TAg^wt mice showed upregulation of 291 genes and downregulation of 140 genes in comparison to those from their nontransgenic littermates, while those isolated from TAg^N136 mice showed upregulation of 274 genes and downregulation of 197 genes. In contrast, TAg³²¹³ mice showed upregulation of only 19 genes and downregulation of 28 genes, while TAg^D44N mice showed upregulation of 20 genes and downregulation of 61 genes. Thus, mutation of either the LXCXE motif or the J domain results in a large reduction of the number of genes upregulated and downregulated by TAg. These data indicate that nearly all TAg-mediated gene regulation in enterocytes requires a functional J domain and LXCXE motif.

TAg^wt and TAg^N136 regulate similar sets of cellular genes.

The J domain and LXCXE motif of TAg act in cis to disrupt Rb-E2F complexes and to activate E2F-dependent transcription. Amino-terminal fragments of TAg possessing both of these elements are sufficient for this action both in vitro and in vivo (28, 31). We compared the global patterns of cellular gene expression in enterocytes expressing TAg^wt and TAg^N136 with those in enterocytes expressing TAg^D44N and TAg³²¹³ by hierarchical clustering. The average log ratios for the gene probe sets (n = 796) that were threefold up- or downregulated in at least one class of mice (TAg^wt, TAg^N136, TAg³²¹³, and TAg^D44N) were clustered using the Euclidean distance metric and complete linkage (Fig. 2A).

FIG. 2.

Hierarchical clustering of regulated genes reveals two distinct patterns of gene expression. (A) The average log ratios for the probe sets (n = 796) that were threefold upregulated or downregulated by at least one class were clustered using the Euclidean distance metric and complete linkage. The classes are listed at the top of the cluster diagram (left). Probe sets colored red are upregulated (green, downregulated) relative to normal villi. (B) Three groups of probe sets (347 total) showed a pattern of being upregulated in mice expressing TAg^wt and TAg^N136 but not TAg^D44N and TAg³²¹³ (red bar). (C) Conversely, there are two groups of probe sets (44 total) that are downregulated by TAg^wt and TAg^N136 but are not changed in TAg^D44N and TAg³²¹³ (green bar). The line plots for these groups are shown on the right, where the average log ratio is plotted on the y axis.

Three groups of probe sets (347 total) show a pattern of upregulation in mice expressing TAg^wt and TAg^N136 but not in those expressing TAg^D44N and TAg³²¹³ (Fig. 2B). The majority of the genes in these upregulated clusters are involved in replication, mitosis/cytokinesis, transcriptional control, and DNA repair. A list of some of the genes in the major categories are shown in Table 1. Two groups of probe sets (44 total) were downregulated by TAg^wt and TAg^N136 but were not changed by TAg^D44N and TAg³²¹³ (Fig. 2C). The downregulated cluster includes genes belonging to the transmembrane, transcriptional control, immune response, and signaling classes.

TABLE 1.

Genes regulated by TAg^wt and TAg^N136 grouped by ontology^a

Category and gene description	AFFY symbol
Genes upregulated by Tagwt and TAgN136
Replication
Minichromosome maintenance deficient 5, cell division cycle	Mcm5
46 (Saccharomyces cerevisiae)
Minichromosome maintenance deficient 7 (S. cerevisiae)	Mcm7
Minichromosome maintenance deficient 2 mitotin	Mcm2
(S. cerevisiae)
Flap structure specific endonuclease 1	Fen1
Ribonucleotide reductase M1	Rrm1
Ribonucleotide reductase M2	Rrm2
Minichromosome maintenance deficient 4 homolog	Mcm4
(S. cerevisiae)
Minichromosome maintenance deficient 6 (MIS5 homolog,	Mcm6
Schizosaccharomyces pombe) (S. cerevisiae)
Thymidine kinase 1	Tk1
Replication protein A2	Rpa2
Ligase I, DNA, ATP dependent	Lig1
Proliferating cell nuclear antigen	Pcna
Polymerase (DNA directed), alpha 1	Pola1
Minichromosome maintenance deficient 3 (S. cerevisiae)	Mcm3
Replication factor C (activator 1) 4	Rfc4
Topoisomerase (DNA) II alpha	Top2a
DNA primase, p49 subunit	Prim1
Cell division cycle 45 homolog (S. cerevisiae)-like	Cdc45l
Geminin	Gmnn
GINS complex subunit 4 (Sld5 homolog)	Gins4
DNA primase, p58 subunit	Prim2
Deoxyuridine triphosphatase	Dut
Replication protein A1	Rpa1
Polymerase (DNA directed), epsilon 4 (p12 subunit)	Pole4
Polymerase (DNA directed), epsilon 2 (p59 subunit)	Pole2
Lin-9 homolog (Caenorhabditis elegans)	Lin9
GINS complex subunit 2 (Psf2 homolog)	Gins2
Minichromosome maintenance deficient 8 (S. cerevisiae)	Mcm8
Minichromosome maintenance deficient 10 (S. cerevisiae)	Mcm10
Deoxycytidine kinase	Dck
Replication protein A3	Rpa3
Bloom syndrome homolog (human)	Blm
Deoxycytidine kinase	Dck
DNA2 DNA replication helicase 2-like (yeast)	Dna2l
GINS complex subunit 1 (Psf1 homolog)	Gins1
Deoxythymidylate kinase	Dtymk
Replication factor C (activator 1) 5	Rfc5
Replication factor C (activator 1) 2	Rfc2
Origin recognition complex, subunit 6-like (S. cerevisiae)	Orc6l
Topoisomerase (DNA) II beta binding protein	Topbp1
Topoisomerase (DNA) II alpha	Top2a
Mitosis/cytokinesis
Stathmin 1	Stmn1
Cell division cycle 20 homolog (S. cerevisiae)	Cdc20
Cyclin-dependent kinase inhibitor 2C (p18, inhibits CDK4)	Cdkn2c
Budding uninhibited by benzimidazoles 1 homolog	Bub1
(S. cerevisiae)
Cyclin A2	Ccna2
Cyclin B1	Ccnb1
Antigen identified by monoclonal antibody Ki 67	Mki67
Kinesin family member 2C	Kif2c
Cyclin B2	Ccnb2
Cell division cycle 6 homolog (S. cerevisiae)	Cdc6
Cyclin E1	Ccne1
Cell division cycle 2 homolog A (S. pombe)	Cdc2a
Kinesin family member 22	Kif22
Cytoskeleton associated protein 2	Ckap2
Thymopoietin	Tmpo
CDC28 protein kinase regulatory subunit 2	Cks2
Kinesin family member 23	Kif23
Inner centromere protein	Incenp
Kinesin family member 11	Kif11
Polo-like kinase 4 (Drosophila)	Plk4
Centromere autoantigen A	Cenpa
Cell division cycle associated 2	Cdca2
Cell division cycle associated 3	Cdca3
Cell division cycle associated 5	Cdca5
Cell division cycle associated 8	Cdca8
Protein regulator of cytokinesis 1	Prc1
Anillin, actin binding protein (scraps homolog, Drosophila)	Anln
Nucleolar and spindle-associated protein 1	Nusap1
NIMA (never in mitosis gene a)-related expressed kinase 2	Nek2
MAD2 (mitotic arrest deficient, homolog)-like 1 (yeast)	Mad2l1
Thymopoietin	Tmpo
Transcriptional control/chromatin
H2A histone family, member X	H2afx
Retinoblastoma-like 1 (p107)	Rbl1
SMC2 structural maintenance of chromosomes 2-like 1	Smc2l1
(yeast)
E2F transcription factor 2	E2f2
Enhancer of zeste homolog 2 (Drosophila)	Ezh2
High-mobility-group box 2	Hmgb2
DNA methyltransferase (cytosine-5) 1	Dnmt1
Transcription factor 19	Tcf19
ASF1 anti-silencing function 1 homolog B (S. cerevisiae)	Asf1b
Histone aminotransferase 1	Hat1
Chromobox homolog 5 (Drosophila HP1a)	Cbx5
Thyroid hormone receptor interactor 13	Trip13
SMC4 structural maintenance of chromosomes 4-like 1	Smc4l1
(yeast)
DEK oncogene (DNA binding)	Dek
Lamin B receptor	Lbr
Retinoblastoma-like 1 (p107)	Rbl1
DNA repair
Meiotic recombination 11 homolog A (S. cerevisiae)	Mre11a
MutS homolog 6 (Escherichia coli)	Msh6
Exonuclease 1	Exo1
Breast cancer 2	Brca2
RAD54-like (S. cerevisiae)	Rad54l
Timeless homolog (Drosophila)	Timeless
RAD51-associated protein 1	Rad51ap1
Denticleless homolog (Drosophila)	Dtl
Checkpoint kinase 1 homolog (S. pombe)	Chek1
Topoisomerase (DNA) II beta binding protein	Topbp1
Claspin homolog (Xenopus laevis)	Clspn
Ubiquitin-like, containing PHD and RING finger domains, 1	Uhrf1
RAD51 homolog (S. cerevisiae)	Rad51
Breast cancer 1	Brca1
Timeless-interacting protein	Tipin
Apoptosis
Caspase 8-associated protein 2	Casp8ap2
Genes downregulated by Tagwt and TAgN136
Transmembrane
Secreted and transmembrane 1	Sectm1
Aquaporin 1	Aqp1
Protein tyrosine phosphatase, receptor type, E	Ptpre
Solute carrier family 5 (neutral amino acid transporters,	Slc5a4b
system A), member 4b
Inositol 1,4,5-triphosphate receptor 2	Itpr2
Adenosine A2b receptor	Adora2b
Grina
Solute carrier family 5 (sodium/glucose cotransporter),	Slc5a12
member 12
Integrin, alpha E, epithelium associated	Itgae
F-box only protein 32	Fbxo32
Transcriptional control
Laminin, alpha 3	Lama3
Avian musculoaponeurotic fibrosarcoma (v-maf) AS42	Maf
oncogene homolog
High-mobility group box transcription factor 1	Hbp1
Signaling
5′-Nucleotidase, ecto	Nt5e
Tensin 4	Tns4
Insulin receptor substrate 2	Irs2
Oxidoreductase
Aldo-keto reductase family 1, member B7	Akr1b7
Leukotriene B4 12-hydroxydehydrogenase	Ltb4dh
Immune response
Mannose binding lectin (C)	Mbl2
Tumor necrosis factor (ligand) superfamily, member 13b	Tnfsf13b
CD36 antigen	Cd36

^aGenes in the upregulated cluster from Fig. 2 include genes involved in the control of cell growth and division, such as DNA replication, mitosis, chromatin remodeling, and DNA repair. Genes in the downregulated cluster from Fig. 2 include genes involved in transmembrane signaling, transcriptional control, and immune response.

To further determine the effect of individual domains/regions of TAg on gene expression, we did independent comparisons between each mutant and TAg^wt (Fig. 3). Genes with a TAg^wt/control ratio of greater than or equal to 3 and a TAg mutant/control ratio of less than 1.2 were considered upregulated by TAg but not by the mutant. Genes upregulated by the mutant were identified similarly. Furthermore, genes with a TAg^wt/control ratio of less than or equal to 0.33 and a mutant/control ratio of greater than 0.8 were considered downregulated by TAg^wt but not by the mutant. Genes downregulated by the mutant were identified similarly. We used the DAVID 2.0 bioinformatics resource (http://david.abcc.ncifcrf.gov/home.jsp) to group the identified genes into different biological classes.

FIG. 3.

The majority of gene regulation by TAg^wt requires the LXCXE motif and J domain. In each Venn diagram, the intersection represents the number of genes that are commonly upregulated (A, B, and C) or downregulated (D, E, and F) threefold or more by TAg^wt, TAg^N136, TAg³²¹³, or TAg^D44N. The left circle represents genes which are uniquely upregulated or downregulated by TAg^wt, while the right circle represents genes which are uniquely upregulated or downregulated by TAg^N136, TAg³²¹³, or TAg^D44N (see Materials and Methods for details). Genes were assigned to different biological classes using the DAVID 2.0 bioinformatics resource. Biological classes are sorted to show genes with low P values (high enrichment score) in the first position of the list.

Comparison of the gene expression patterns between TAg^wt and TAg^N136 revealed that all the genes upregulated by TAg^wt (199) are also upregulated by TAg^N136 (Fig. 3A). These genes are associated with E2F-dependent transcription, DNA replication and repair, cell cycle regulation, chromatin assembly/modification, and the microtubule cytoskeleton (Table 1). A small number of genes (11) were upregulated by TAg^N136 but not by TAg^wt. These belonged to biological classes such as carbohydrate metabolism and angiogenesis. Similarly, there was a large overlap (93 genes) between genes downregulated by TAg^wt and TAg^N136 (Fig. 3D). These genes belong to different biological classes, such as transmembrane proteins, chromosomal proteins, immune response/antigen processing, and protein kinases (Table 1).

Comparison of the gene expression patterns between TAg^wt and TAg³²¹³ revealed that TAg³²¹³ expression does not affect the genes (251) upregulated by TAg^wt (Fig. 3B), and therefore there is not a significant overlap of genes regulated by both TAg^wt and TAg³²¹³. In fact, no specific class of genes was uniquely upregulated by TAg³²¹³, indicating that gene upregulation by TAg^wt in enterocytes requires an intact LXCXE motif. A total of 76 genes are uniquely downregulated by TAg^wt (Fig. 3E). These genes belong to the class of immunoglobulins, immune response/antigen processing, chromatin/histones, ion transport, and fatty acid metabolism. Only a few genes (6) were uniquely downregulated by TAg³²¹³, but they do not belong to any specific class. As observed with the upregulated genes, there was no significant overlap between the genes downregulated by TAg^wt and TAg³²¹³.

The comparison between gene upregulation in TAg^wt and TAg^D44N shows a pattern similar to that observed in the comparison between TAg^wt and TAg³²¹³. TAg^D44N mice do not show regulation of the genes (238) upregulated by TAg^wt (Fig. 3C), nor did we find any specific class of genes uniquely upregulated by TAg^D44N. Therefore, we did not find any significant overlap of genes between TAg^wt and TAg^D44N, suggesting that all gene upregulation by TAg^wt in enterocytes requires an intact J domain. We found a group of genes commonly downregulated by both TAg^wt and TAg^D44N belonging to the class of major histocompatibility complex and chromatin/histones, suggesting J domain-independent regulation of these genes. A total of 57 genes are uniquely downregulated by TAg^wt (Fig. 3F). These genes belong to the class of peroxisomal proteins, transmembrane proteins, immunoglobulins, oxidoreductase, carboxylic acid metabolism, and protein kinases. Additionally, 13 genes are uniquely downregulated by TAg^D44N, but they do not belong to any specific class. If we relax the cutoff for selecting genes from threefold to twofold, we find additional potential genes downregulated by TAg^D44N. These genes correspond to nucleic acid metabolism, mitosis, cell cycle, and cell division.

Regulation of E2F-dependent gene expression by TAg in villus enterocytes requires inactivation of the Rb family.

The LXCXE motif and J domain are required for the complete inactivation of Rb family members and thus for the induction of E2F target genes (29, 30, 34). Therefore we hypothesized that TAg^N136 should be able to regulate the same set of E2F target genes as TAg^wt, while TAg³²¹³ and TAg^D44N should be defective for this regulation. Consistent with this hypothesis, our microarray data showed upregulation of E2F target genes that was equal to or greater than threefold in mice expressing either TAg^wt or TAg^N136 but not in those expressing TAg³²¹³ and TAg^D44N (Fig. 4). The array results were confirmed by RT-PCR analysis of a series of selected E2F target genes using RNA isolated from enterocytes expressing TAg^wt or mutant TAg proteins (data not shown).

FIG. 4.

Regulation of E2F target genes by TAg and mutants. Mice expressing TAg^wt and TAg^N136 show significant upregulation of E2F target genes, while mice expressing TAg³²¹³ and TAg^D44N do not. The average fold change of the genes was determined by microarray analysis of each category (n = 3): TAg^wt, TAg^N136, TAg³²¹³, and TAg^D44N. “Normalization” represents a fold change of 1 (not regulated).

Induction of intestinal hyperplasia by TAg does not involve regulation of p53-dependent genes.

In many cell culture and transgenic systems, TAg^wt binds the tumor suppressor p53 and blocks p53-dependent transcription (3, 13, 17, 18). In contrast, the expression of amino-terminal truncations of TAg (such as TAg^N136) usually results in p53 stabilization and the activation of p53-dependent transcription and growth suppression (21). However, we have previously shown that enterocytes do not express detectable levels of p53, even in the presence of TAg, and that the interaction between TAg and p53 does not play a role in TAg-mediated enterocyte transformation (20). Consistent with these observations, the array experiments did not show any evidence to indicate that p53-target genes are regulated by TAg^wt or any of the mutants (data not shown). We confirmed the expression levels of selected p53 target genes (Fas, Lrdd, Noxa, and Pten) by semiquantitative RT-PCR analysis (Fig. 5). Nontransformed mouse embryo fibroblasts (MEFs) and MEFs expressing TAg^wt or TAg^N136 were used as controls. As expected, all the p53 target genes were significantly upregulated in MEFs expressing TAg^N136 (lane 3) in comparison to control MEFs (lane 1) and MEFs expressing TAg^wt (lane 2). In contrast, none of the p53 target genes tested was regulated by TAg^wt (lane 5) or mutant TAg (lanes 6 to 8) proteins in enterocytes.

FIG. 5.

p53 target genes are not affected in villus enterocytes expressing TAg and mutants. Transcript levels of p53 target genes in nontransgenic (NT), TAg^wt, TAg^N136, TAg³²¹³, and TAg^D44N villus samples were analyzed. Normal MEFs (control) and MEFs expressing TAg^wt and TAg^N136 were used to show the presence of active p53 in MEFs. cDNAs were reverse transcribed from equal amounts of total RNA extracts and subjected to PCR using specific primers. The transcript level of Adh5 was used as a loading control.

Transgenic mice expressing an amino-terminal truncated TAg develop hyperplasia and show signs of dysplasia.

Comparison of the corresponding gene expression patterns suggests that in villus enterocytes, TAg^N136 regulates almost the same set of genes as TAg^wt. Based on this observation, we predicted that mice expressing TAg^N136 will show the same intestinal phenotype as TAg^wt. To test this, we screened TAg^wt and TAg^N136 mice for their capability to induce neoplasia.

Mice expressing TAg^wt in the intestinal enterocytes develop hyperplasia by 6 weeks after birth, a phenotype that becomes dysplastic by 4 to 6 months (12, 15, 20). Additionally, these mice show a significant increase in the length of their intestines and in the size of intestinal crypts (20). We have previously reported that a line of transgenic mice expressing one amino-terminal truncation of TAg in enterocytes, TAg¹¹³⁷, also develops hyperplasia. However, these mice did not show progression to intestinal dysplasia or increased intestinal length or crypt size (15). We considered two possible explanations for the different phenotypes in TAg^wt and TAg¹¹³⁷ mice. First, one or more TAg functions mapping to the carboxy-terminal end of the molecule could be required for progression to dysplasia and/or increased intestinal length and crypt size. The second possibility is that the failure to display dysplasia and increased intestinal length was specific to the single line of TAg¹¹³⁷ mice used in the study.

To distinguish these possibilities we generated additional lines of transgenic mice expressing a truncated TAg, TAg^N136 (Fig. 1). TAg^N136 is similar to TAg¹¹³⁷, but it contains a nuclear localization signal. Like TAg¹¹³⁷, TAg^N136 is capable of disrupting Rb-E2F complexes and of activating E2F-dependent transcription (28). We confirmed the expression of the N136 protein in transgenic mice by immunostaining and Western blot analysis. Paraffin sections stained with anti-TAg antibody revealed that TAg^N136 expression is confined to the top three-fourths of the villi and is not observed in the villus/crypt junction or the top of the crypts (Fig. 6B). TAg^wt mice also exhibited TAg expression in the villi, but in this case the staining extended to the villus/crypt junction and the top of the crypts (Fig. 6A and C). These results were confirmed by Western blot analysis of fractionated cell extracts (Fig. 6D). The TAg^N136 protein levels remained constant as the animals aged (Fig. 6E).

FIG. 6.

Expression patterns of TAg^wt and TAg^N136 in villus enterocytes. Intestinal tissues were embedded in paraffin and 5-μm sections were stained with anti-TAg antibodies. A brown color reflects the presence of TAg. Mice from 3 to 6 months of age were analyzed and a representative case is shown. The TAg^wt mice present a clear dysplastic phenotype, while TAg^N136 mice are hyperplastic. (A) Nuclear expression of TAg in TAg^wt mice extends from the bottom three-fourths of villus enterocytes into the top of the crypts. (B) TAg^N136 mice express N136 in the nucleus and cytoplasm of villus enterocytes (top three-fourths). (C) Higher magnification showing expression of TAg on the top of the crypts in TAg^wt mice. (D) Expression levels of TAg in TAg^wt and TAg^N136 villus (V) and crypt (C) samples. Protein extracts from villi and crypts of nontransgenic (NT) and TAg^wt and TAg^N136 transgenic mice were subjected to immunoblotting for TAg. Protein levels for β-tubulin were used as loading controls. (E) Expression levels of N136 protein through development. Protein extracts from the middle intestines of nontransgenic (NT) and TAg^wt and TAg^N136 transgenic mice were subjected to immunoblotting for TAg. TAg^wt mice (4 and 7 months) and TAg^N136 mice (4, 9, 12, and 18 months) were used. β-Tubulin was used as a loading control.

To assess the proliferative status of transgenic intestines, mice were injected with BrdU and histological sections were stained with anti-BrdU (Fig. 7). We found numerous BrdU-positive cells in enterocytes of both TAg^wt and TAg^N136 mice, irrespective of the age tested (Fig. 7B and C). Thus, like TAg^wt, TAg^N136 induces enterocytes to reenter the cell cycle. TAg^N136 mice also develop hyperplasia and progress to dysplasia, although these histological changes progress at a slower pace than in TAg^wt mice. We also assessed the intestinal lengths of mice in different age groups (Table 2). As expected, the average intestinal length of TAg^wt mice was much greater than that of control mice (51 cm versus 40 cm). Similarly, TAg^N136 mice show a significant increase in intestinal length (48 cm) in comparison to nontransgenic littermates (40 cm). Unlike TAg^wt mice, TAg^N136 mice do not exhibit enlarged crypts. We conclude that like TAg^wt, TAg^N136 is capable of driving enterocyte proliferation, of inducing hyperplasia, and of increasing intestinal length.

FIG. 7.

TAg^N136 induces ectopic proliferation in villus enterocytes. TAg^wt and TAg^N136 mice (B and C) show ectopic proliferation of enterocytes, while proliferation is restricted to the crypts in nontransgenic mice (A). Intestinal sections were stained with anti-BrdU and show numerous BrdU-positive cells (indicated by arrows) in villus enterocytes of mice expressing TAg^wt and TAg^N136. Mice from 3 to 6 months of age were analyzed, and a representative case is shown. The TAg^wt mice present a clear dysplastic phenotype, while TAg^N136 mice are hyperplastic.

TABLE 2.

Intestinal length of mice according to genetic background and age

Genetic background	Avg total length of small intestine (cm) ± SE (total no. of mice analyzed) for mice at age (mo):
	0-3	3-6	6-9	≥9
Nontransgenic	39.6 ± 0.45 (70)	40.3 ± 0.41 (75)	40.3 ± 0.52 (46)	40.7 ± 0.61 (35)
TAgwt	49.4 ± 1.15 (61)	51.5 ± 0.97 (90)	52.6 ± 1.45 (42)	52.1 ± 1.75 (29)
TAgN136	43.6 ± 0.57 (11)	47.3 ± 1.04 (14)	50.5 ± 1.21 (15)	51.4 ± 0.76 (28)

DISCUSSION

The action of TAg on the Rb family proteins is central to our current understanding of SV40-mediated transformation. Our studies with transgenic mice expressing TAg^wt and mutant TAg in enterocytes is consistent with this view. While TAg^wt induces enterocyte proliferation, TAg mutants that are defective either for Rb binding (TAg³²¹³) or for chaperone action on Rb-E2F complexes (TAg^D44N) fail to do so. Like TAg^wt, truncation mutants such as TAg¹¹³⁷and TAg^N136, which contain a functional J domain and Rb binding motif, induce enterocyte proliferation. The expression of TAg^wt or TAg^N136 results in the degradation of p130 and the loss of p130-E2F4 complexes, the upregulation of E2F2 and E2F3a, and increased expression of E2F-dependent genes (28). Finally, the ectopic enterocyte proliferation induced by TAg is dependent on the presence of E2F2 and perhaps E2F3a (28). Small TAg is not expressed in this system (4), and thus all of the observed effects are due to large TAg.

In this study we have used gene microarray experiments to assess the effects of TAg^wt and mutant TAg on enterocyte gene expression. We have previously reported that TAg^wt alters the levels of a large collection of cellular genes in enterocytes, including an upregulation of E2F-dependent genes (4). In the current study we have used whole-mouse genome arrays to compare mRNA levels in enterocytes expressing TAg^wt, TAg^N136, TAg^D44N, or TAg³²¹³. We found that TAg^wt regulates over 400 genes and that essentially all TAg-mediated gene regulation in enterocytes requires a functional J domain and LXCXE motif. In the present study we have used a strict criterion of a threefold cutoff, which likely results in an underestimation of regulated genes. Nevertheless, TAg^N136 upregulated the same genes as TAg^wt, including E2F-dependent genes. Similarly, almost all of the cellular genes downregulated by TAg^wt were also downregulated by TAg^N136. Taken together, these results indicate that TAg^N136 is fully competent to antagonize Rb proteins and to activate E2F-dependent transcription in intestinal enterocytes.

The fact that all of the DNA tumor viruses alter some aspect of the Rb-E2F axis indicates the central importance of this pathway to cell cycle regulation and tumorigenesis. Thus, each of these viruses would be expected to alter cellular gene expression in a manner similar to that of SV40. However, two factors complicate this prediction. First, each virus exhibits its transforming properties in different host species and/or different cell lineages. For example, we have previously reported that TAg^wt regulates unique sets of cellular genes in transformed MEFs compared to mouse enterocytes (4). Second, the transforming proteins encoded by each of the DNA tumor viruses act by unique mechanisms, and each can have specific effects on cellular gene expression. Nevertheless, in an attempt to discern common effects among these different systems, we have compared our results with three other microarray data sets: (i) the so-called TAg signature, obtained by expression of TAg in three different mouse tissues (8); (ii) human papillomavirus (HPV)-positive tumors (25); and (iii) MEFs conditionally depleted of pRb (19). This comparison is illustrated in Table 3.

TABLE 3.

Comparison of genes regulated in three tumorigenic systems and TAg^wt-expressing enterocytes

System	Upregulated genes		Downregulated genes
	No. of genesa	% Overlap with TAgwt enterocytesb	No. of genesa	% Overlap with TAgwt enterocytesb
TAg signature	83	78	33	9
HPV+	79	35	10	0
Rb−/−	126	67	46	7

^aGenes up- or downregulated in the TAg signature (Table 1) (8), HPV-positive tumors (25), and MEFs conditionally depleted of pRb (Fig. 7) (19) were matched to genes on the TAg^wt enterocyte microarray. A majority of the genes were found.

^bThe percentage represents the overlapping genes between TAg^wt-expressing enterocytes and the other three systems at a twofold cutoff.

We found that a relatively large number of genes upregulated by transformation in the three systems were also upregulated by TAg^wt in enterocytes. For example, of 83 genes identified as being upregulated in the TAg signature, 78% were also upregulated by TAg^wt in enterocytes. This indicates that TAg utilizes some of the same molecular pathways to induce transformation in different systems. Similarly, 35% of the genes upregulated in HPV tumors and 67% of the genes upregulated in MEFs following pRb ablation were also upregulated in TAg^wt-expressing enterocytes. Nearly all of these genes are related to cell cycle regulation and E2F-dependent gene transcription. Thus, transformation by TAg^wt in different tissues and cell types, transformation by HPV, and the loss of pRb activity all appear to lead to similar changes in gene expression. Nevertheless, each system also harbored a collection of genes that were uniquely upregulated. In contrast, there was little or no overlap of cellular genes downregulated in any of the systems. Therefore, the mechanism of gene repression and, consequently, the specific genes downregulated by transformation appear to depend both on the transforming system and on the cell type.

While TAg action on the Rb-E2F axis is necessary for enterocyte transformation, it is unclear if this action is sufficient. If this were the case, then transformed properties of TAg^N136-expressing mice should phenocopy those of mice expressing TAg^wt. To test this, we generated several new lines of TAg^N136-expressing mice and assessed phenotypic characteristics, including enterocyte proliferation, progression to hyperplasia and/or dysplasia, intestinal length, and crypt size. In addition to hyperproliferation, dysplasia is characterized by the release of the cells from the basement membrane, resulting in multiple cell layers of epithelium as well as branching of the villi, increased crypt size, and overall general disorganization of the tissue. The induction of enterocyte proliferation and E2F-dependent gene expression was identical in TAg^wt and TAg^N136 mice. Like TAg^wt mice, TAg^N136 mice exhibited dysplasia and increased intestinal length. However, TAg^N136 induced these phenotypes at a much slower pace. Unlike TAg^wt mice, TAg^N136 mice did not exhibit an increase in crypt size. We have not observed a significant difference in the expression levels of TAg^wt or TAg^N136 with the age of the animal (Fig. 6E), and thus we do not believe that the delayed phenotype observed in TAg^N136 mice is related to a change in protein expression.

Instead, we suggest two possible explanations for this observation. First, we found some significant differences in the patterns of TAg expression in villus enterocytes between TAg^wt and TAg^N136 mice. TAg^N136 mice showed staining only in the top three-fourths of villi, and the staining never went down to the villus/crypt differentiation zone (Fig. 6B). On the other hand, TAg^wt mice showed a dense nuclear staining throughout the villi and villus/crypt differentiation zone which occasionally extended to the top of the crypts (Fig. 6A and C). The expression of TAg in cells that have not differentiated completely could result in a higher degree of dysplastic phenotype and/or the increase in crypt size observed in TAg^wt mice. Experiments directing the expression of TAg^wt and TAg^N136 to the intestinal crypts would test this possibility.

Alternatively, the carboxy terminus of TAg could contribute to the differences observed between TAg^wt and TAg^N136 mice. One obvious candidate that could interact with TAg^wt but not with TAg^N136 is p53, but TAg action on p53 does not appear to play a role in the transformation of enterocytes. TAg expression in enterocytes does not result in p53 stabilization, nor can TAg-p53 complexes be detected in these cells (20). Furthermore, the genetic ablation of p53 has no effect on enterocyte transformation induced by either TAg^wt or truncated TAgs (20). In this study we show that neither TAg^wt nor truncated TAgs alter the transcriptional levels of p53-dependent genes in enterocytes. In this regard, enterocytes differ from some other epithelial cell types, such as the choroid plexus, which exhibit robust p53-dependent apoptosis in response to truncated TAg (32). It is possible that as-yet-unidentified cellular targets are affected by TAg through its carboxy terminus and that their action plays a role in determining dysplasia and/or crypt size.

This is the first study which compares the global changes in the gene expression by TAg^wt and its mutants side by side. Our results confirm and complete previous studies indicating the central and perhaps unique role played by Rb proteins (pRb, p130, and p107) in the control of intestinal tumorigenesis (9, 10). Not only are the first 136 amino acids of TAg sufficient to induce hyperplasia, but the inactivation of Rb proteins appears to be the primary requirement for this induction. The requirement of the LXCXE motif and the J domain for TAg-mediated gene regulation is evident by the fact that TAg^wt and TAg^N136 regulate the same set of genes and the regulation is lost by TAg³²¹³ and TAg^D44N. Furthermore, p53 does not play any role in TAg-mediated gene regulation in the intestinal epithelium, as reflected by the inability of TAg^N136 to induce p53 target genes.