Abstract
To examine the role of Rb1 in gastrointestinal (GI) tumors, we generated mice with an Apc1638N allele, Rbtm2brn floxed alleles, and a villin-cre transgene (RBVCA). These animals had exon 19 deleted from Rb1 throughout the GI tract. We have shown previously that Rb1 deficiency is insufficient for GI tumor initiation, with inactivation of an Apc allele capable of overcoming the insufficiency. In this study we demonstrate that RBVCA mice have reduced median survival because of an increase in tumor incidence and multiplicity in the cecum and the proximal colon. Large intestinal tumors are predominantly adenomas, whereas the tumors of the small intestine are a mixture of adenomas and adenocarcinomas. We find truncation mutations to the second Apc allele in tumors of both the large and small intestine. Expression profiles of duodenal and cecal tumors relative to each other show unique gene subsets up and down regulated. Substantial expression patterns compare to human colorectal cancer, including recapitulation of embryonic genes. Our results indicate that Rb1 has significant influence over tumor location in the GI tract, and that both cecal and duodenal tumors initiate through inactivation of Apc. Expression profile analysis indicates the two tumor types differentially regulate distinct sets of genes that are over-expressed in a majority of human colorectal carcinomas.
Keywords: cancer, mouse, conditional, colorectal, retinoblastoma
Colorectal cancer (CRC) is an important cause of morbidity and mortality. Each year 150,000 new cases are reported in the United States, as well as 56,000 deaths. Studies of two predisposition syndromes have led to the identification of genes involved in the initiation of the disease. Familial adenomatous polyposis (1), a dominant autosomal hereditary syndrome, is caused by germ-line mutations in the adenomatous polyposis coli gene (APC). Many sporadic CRC cases are the result of somatic APC mutation (2). The second familial syndrome, hereditary non polyposis colorectal cancer (HNPCC), is caused by mutations in DNA mismatch repair (MMR) genes that result in an increase in mutation rate (3) responsible for the HNPCC phenotype. Mutations to both copies of APC, resulting from MMR deficiency, are required for HNPCC tumor initiation. Genetic changes in RAS, AKT, TGF-β signaling, and p53 have also been described at different stages of tumor progression. One of the most proximal and frequent changes is the activation of either K-RAS or N-RAS, found to occur in ≈50% of all CRCs. Mutations are also found in the PI3-kinase pathway. The genes or pathways involved in the rest of the tumors are not well understood. The retinoblastoma (RB1) pathway, known to be involved in many types of cancers, is not extensively documented in colon cancer.
In humans, RB1 plays a tumor suppressor role in cancer of the retina (4) and several other tissues (5–9). Abnormalities of the RB1 family (pRB, pRB2/p130, p107) have been identified in a large proportion of cancers and implicated in tumor pathogenesis and progression (10, 11). Some colonic adenocarcinomas undergo allelic loss at the RB1 locus (12, 13), whereas almost half of colorectal carcinomas show nonrandom chromosomal 13 gains (14, 15). The loss of pRB and CDK inhibitor p16(INK4a) correlates with the poor outcome in CRC and cecal tumor-site specificity (16, 17). Approximately two thirds of intestinal tumors paradoxically have increased levels of RB1 expression and protein production when compared with paired normal colonic tissue (18, 19).
Mice with a null mutation of Rb1 do not develop intestinal cancer (20, 21), suggesting that loss of RB1 is not involved in intestinal tumor initiation. Because APC is known to be involved in initiation of colon cancer, we decided to examine the outcome of simultaneous absence of Apc and Rb1 in murine intestinal epithelium. We generated mice with an Apc1638N allele, Rbtm2brn floxed alleles, and a villin-cre transgene (RBVCA). The villin promoter is highly expressed in adult intestine and kidney, and results in removal of exon 19 with corresponding loss of Rb1 function throughout the gastrointestinal (GI) tract. We have shown previously that without Apc mutation, loss of Rb1 function in intestinal epithelium does not result in intestinal tumors. Such mice develop pituitary tumors and highly metastatic C cell carcinoma of the thyroid at 1 year of age (21). In this study we show that the combination of an Apc mutation and Rb1 mutation in the GI tract (RBVCA) significantly decreases median survival when compared with age-matched Apc1638N+/− siblings. Mutations in Apc1638N alone result in a few tumors in the small intestine. Loss of Apc in the Rb1-deficient intestinal epithelium leads to tumors in the small intestine, cecum, and large intestine. Mutational analysis of the WT Apc allele in tumors from RBVCA mice shows similar truncation mutations in all tumor types, indicating large and small intestinal tumors both initiate through somatic mutation of the second Apc allele.
Examination of gene-expression profiles revealed that the two classes of tumors are readily distinguishable from each other, and that expression profiles from the cecum are similar to human colorectal tumors. We observed recapitulation of embryonic gene expression. Our results indicate the Rb pathway has a role to play in intestinal tumorigenesis, and that the presence of Rb mutation results in alteration of signaling pathways and subsequent expansion of the intestinal compartment in which the tumors develop.
Results
RBVCA Mice Have Reduced Median Survival.
A Kaplan-Meier survival plot showed that RBVCA mice had a median survival of 9 months, and died significantly earlier than animals without Apc1638N+/− mutation (RBVC) (Rbtm2Brn −/− Tg Vil-cre, median survival 12.5 months) and Apc1638N+/− siblings that were WT or heterozygous for Rb1 (Fig. 1A).
Fig. 1.
(A) Kaplan-Meier survival plot of RBVCA mice. Time of death or when mice became moribund was recorded. The colors for the different genotypes and the number of mice are as follows: black lines, Rb+/+, Apc+/+ (34 mice); green lines, Rb+/−, Apc+/+ (65 mice); blue lines, Rb−/−, Apc+/+ (120 mice); mauve lines, Rb+/+, Apc+/− (39 mice); red lines, Rb+/−, Apc+/− (41 mice); magenta lines, Rb−/−, Apc+/− (58 mice). All mice carried the Villin-cre transgene. (B) Cecal tumors from an Rb−/−, Apc+/− mouse. (Upper) A picture of multiple gross tumors of the cecum from an Apc1638N mouse with a retinoblastoma deficient intestine. (Lower) A villous adenoma of the cecum stained with hemotoxylin eosin ×100. (C) Papillae of Vater with polyp from an Rb−/−, Apc+/− mouse (Lower) and a WT mouse (Upper).
Rb1 Deficiency in the GI Tract of Apc1638N+/− Mice Promotes Tumor Initiation in the Cecum and Proximal Colon.
Eighty-seven mice were examined for GI tumors. This group included 64 animals of combinatorial genotypes from the RBVCA cohort and 24 mice from other cohorts of interest for comparison. The mean age of each group and the GI tumor incidence is shown in Table 1. All animals with the Apc1638N allele developed small intestinal tumors, with and without Rb1 deficiency. Tumors of the large intestine were found in 87% of the Rbtm2Brn −/− Tg (Vil-cre) Apc1638N+/− animals, compared with 15% seen in the Rbtm2Brn +/+ Tg (Vil-cre) Apc1638N+/− animals (P < 0.001). All Rbtm2Brn −/− Tg (Vil-cre) Apc1638N+/− animals with tumors of the large intestine had tumors of the cecum (Fig. 1B); four of these animals also had tumors in the proximal colon. The tumor incidence in the proximal colon varied significantly from the tumor incidence of the proximal colons of Rbtm2Brn +/+ Tg (Vil-cre) Apc1638N+/− mice (P < 0.05).
Table 1.
Tumor incidence in the GI tract
| Genotype | n | Age (months) mean ± SD | Gender (M:F) | Number of mice with GI tumors (%) |
||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Overall | Stomach | Small intestine | Large intestine |
|||||||
| Total | Cecum | Proximal colon | Distal colon | |||||||
| Rb W+, Apc W+ | 6 | 11.1 ± 0.8 | 1:1.0 | 0 (0)§ | 0 (0) | 0 (0)§ | 0 (0) | 0 (0) | 0 (0) | 0 (0) |
| Rb−/−, ApcW+ | 13 | 8.2 ± 2.3 | 1:0.9 | 0 (0)§** | 0 (0)¶ | 0 (0)§** | 0 (0)** | 0 (0)** | 0 (0) | 0 (0) |
| Rb W+, Apc+/−* | 20 | 7.1 ± 2.1 | 1:1.3 | 19 (95) | 8 (40) | 19 (95) | 3 (15) | 3 (15) | 0 (0) | 2 (10) |
| Rb −/−, Apc+/− | 15 | 8.1 ± 1.7 | 1:0.5 | 15 (100) | 4 (27) | 15 (100) | 13 (87)§** | 13 (87)§** | 4 (27)¶ | 1 (7) |
| Rb+/−, Apc+/− | 15 | 8.5 ± 2.4 | 1:0.7 | 15 (100) | 13 (87)‖†† | 14 (93) | 11 (73)‖ | 11 (73)‖ | 1 (7) | 1 (7) |
| Msh2−/−† | 18 | 9.0 ± 1.1 | 1:0.5 | 16 (89)‡‡ | 0 (0) | 16 (89)‡‡ | 1 (6) | 1 (6) | 0 (0) | 0 (0) |
| Rb W+, Apc+/−‡ | 5 | 5.6 ± 0.6 | 1:0.7 | 5 (100) | 2 (40) | 5 (100) | 1 (20) | 1 (20) | 0 (0) | 0 (0) |
| Rb−/−, VCnegative, Apc +/−‡ | 9 | 7.4 ± 2.9 | 1:0.8 | 8 (89) | 4 (44) | 8 (89) | 2 (22) | 2 (22) | 0 (0) | 2 (22) |
| C57Bl/6, Apc+/−‡ | 6 | 7.6 ± 1.0 | 1:5.0 | 6 (100) | 2 (33) | 6 (100) | 0 (0) | 0 (0) | 0 (0) | 0 (0) |
Genotype positive for Villin-cre transgene unless indicated. n, sample size. SD, standard deviation. M, male. F, female. GI, gastrointestinal.
*Pooled results for groups denoted with “‡”.
†Conditional removal of Msh2 exon 12 by Villin-cre transgene.
§Two-group comparison by Fisher exact probability test: Compared with combined group Rb W +, Apc+/−*: P < 0.001.
¶Compared with combined group Rb W+, Apc+/−*, P < 0.05.
‖Compared with combined group Rb W+, Apc+/−*, P < 0.01.
**Compared with Rb−/−, Apc+/−: P < 0.001.
††Compared with Rb−/−, Apc+/−: P < 0.01.
‡‡Compared with Rb−/−, Apc W+: P < 0.001.
Different control groups were individually and collectively compared with the Rb1 deficient Apc1638N+/− group. In addition to Rbtm2Brn +/+ Tg (Vil-cre positive) Apc1638N+/− siblings, we examined a group of Rbtm2Brn −/− Tg (Vil-cre negative) Apc1638N+/− siblings, Apc1638N+/− on a C57Bl6 background, and the groups combined. All control groups varied significantly from the Rb1 deficient Apc1638N+/− genotype for tumor incidence in the large intestine.
Tumor multiplicity is presented in Table 2. Rbtm2Brn −/− Tg (Vil-cre) Apc1638N+/− mice had a threefold increase in GI tumors overall when compared with Rbtm2Brn +/+ Tg (Vil-cre) Apc1638N+/−; tumors of the large intestine were significantly increased 8.5-fold. Tumors of the cecum were increased 10-fold in mutant animals compared with Apc1638N+/− siblings; tumors of the proximal colon were not detected in the Apc1638N+/− animals.
Table 2.
Tumor multiplicity
| Genotype | n | Number of tumors per mouse (mean ± SEM) |
||||||
|---|---|---|---|---|---|---|---|---|
| Overall | Stomach | Small intestine | Large intestine |
|||||
| Total | Cecum | Proximal colon | Distal colon | |||||
| Rb W+, Apc W+ | 6 | 0.00 ± 0.00g | 0.00 ± 0.00h | 0.00 ± 0.00g | 0.00 ± 0.00h | 0.00 ± 0.00i | 0.00 ± 0.00 | 0.00 ± 0.00 |
| Rb−/−, ApcW+ | 13 | 0.00 ± 0.00gj | 0.00 ± 0.00kg | 0.00 ± 0.00gj | 0.00 ± 0.00fj | 0.00 ± 0.00fj | 0.00 ± 0.00l | 0.00 ± 0.00 |
| Rb W+, Apc +/−a | 20 | 4.15 ± 0.68c | 0.85 ± 0.28c | 2.65 ± 0.34c | 0.65 ± 0.40c | 0.55 ± 0.34c | 0.00 ± 0.00e | 0.10 ± 0.07 |
| Rb−/−, Apc+/− | 15 | 12.33 ± 2.59f | 0.67 ± 0.36 | 5.67 ± 1.77f | 5.53 ± 1.77g | 5.53 ± 1.69g | 0.33 ± 0.16h | 0.13 ± 0.13 |
| Rb+/−, Apc+/− | 15 | 8.60 ± 2.09h | 1.54 ± 0.29hk | 4.93 ± 1.38 | 2.13 ± 0.76fl | 1.93 ± 0.63f | 0.13 ± 0.13 | 0.07 ± 0.07 |
| Msh2−/−b | 18 | 1.61 ± 0.30m | 0.00 ± 0.00 | 1.56 ± 0.28m | 0.05 ± 0.05 | 0.05 ± 0.05 | 0.00 ± 0.00 | 0.00 ± 0.00 |
| Rb W+, Apc+/−d | 5 | 3.4 ± 0.51 | 0.60 ± 0.40 | 2.40 ± 0.40 | 0.40 ± 0.40 | 0.40 ± 0.40 | 0.00 ± 0.00 | 0.00 ± 0.00 |
| Rb−/−, VC negative, Apc+/−d | 9 | 5.11 ± 1.43 | 1.22 ± 0.55 | 2.67 ± 0.60 | 1.22 ± 0.85 | 1.00 ± 0.71 | 0.00 ± 0.00 | 0.22 ± 0.15 |
| C57Bl/6, Apc+/−d | 6 | 3.33 ± 0.72 | 0.50 ± 0.34 | 2.83 ± 0.70 | 0.00 ± 0.00 | 0.00 ± 0.00 | 0.00 ± 0.00 | 0.00 ± 0.00 |
n, sample size. Genotype positive for Villin-cre transgene unless indicated.
aPooled results for groups denoted with “d”.
bConditional removal of Msh2 exon 12 by Villin-cre transgene.
cMultiple comparisons by Kruscal-Wallis test shown on Rb W+, Apc+/−a: P < 0.001.
eMultiple comparisons by Kruscal-Wallis test shown on Rb W+, Apc+/−a: P < 0.01. Two group by Mann-Whitney or binomial test:
fCompared with Rb W+, Apc+/−a: P < 0.01.
gCompared with Rb W+, Apc+/−a: P < 0.001.
hCompared with Rb W+, Apc+/−a: P < 0.05.
iCompared with Rb W+, Apc+/−a: P = 0.055.
jCompared with Rb−/−, Apc+/−: P < 0.001.
kCompared with Rb−/−, Apc+/−: P < 0.01.
lCompared with Rb−/−, Apc+/−: P < 0.05.
mCompared with Rb−/−, Apc W+: P < 0.001.
A significant increase in stomach tumors and expansion of tumors to the large intestine is also found in Rbtm2Brn+/− Tg (Vil-cre) Apc1638N+/− mice, demonstrating that haplo-insufficiency of Rb1 in the context of Apc mutation is sufficient to impact tumor location (Tables 1–3).
Table 3.
Occurrence of adenocarcinomas
| Genotype | n | Number of mice with CA (%) |
Number of CA per mouse (mean ± SEM) |
||||
|---|---|---|---|---|---|---|---|
| Overall | Small intestine | Large intestine | Overall | Small intestine | Large intestine | ||
| Rb W+, Apc W+ | 6 | 0 (0) | 0 (0) | 0 (0) | 0.00 ± 0.00 | 0.00 ± 0.00 | 0.00 ± 0.00 |
| Rb −/−, Apc W+ | 13 | 0 (0)¶ | 0 (0)¶ | 0 (0) | 0.00 ± 0.00‖ | 0.00 ± 0.00‖ | 0.00 ± 0.00 |
| Rb W+, Apc+/−* | 20 | 4 (20) | 4 (20) | 0 (0) | 0.20 ± 0.09§ | 0.20 ± 0.09§ | 0.00 ± 0.00 |
| Rb −/−, Apc+/− | 15 | 5 (33) | 5 (33) | 0 (0) | 0.93 ± 0.42 | 0.93 ± 0.42 | 0.00 ± 0.00 |
| Rb +/−, Apc+/− | 15 | 7 (47) | 5 (33) | 2 (13) | 0.73 ± 0.28 | 0.60 ± 0.29 | 0.13 ± 0.09 |
| Msh2−/−† | 18 | 9 (50) | 9 (50) | 0 (0) | 0.67 ± 0.20 | 0.67 ± 0.20 | 0.00 ± 0.00 |
| Rb W+, Apc+/−‡ | 5 | 0 (0) | 0 (0) | 0 (0) | 0.00 ± 0.00 | 0.00 ± 0.00 | 0.00 ± 0.00 |
| Rb −/−, VC negative, Apc+/−‡ | 9 | 3 (33) | 3 (33) | 0 (0) | 0.33 ± 0.17 | 0.33 ± 0.17 | 0.00 ± 0.00 |
| C57Bl/6, Apc+/−‡ | 6 | 1 (17) | 1 (17) | 0 (0) | 0.17 ± 0.17 | 0.17 ± 0.17 | 0.00 ± 0.00 |
n, sample size. CA, adenocarcinoma. Genotype positive for Villin-cre transgene unless indicated.
*Pooled results for groups denoted with “‡”.
†Removal of Msh2 exon 12 by Villin-cre transgene. Multiple-group comparisons by Kruskal-Wallis test for tumor multiplicity: §, P < 0.05. Two-group by Fisher exact, tumor incidence. Mann-Whitney, tumor multiplicity, compared with Rb−/−, Apc+/−: ¶, P < 0.05; ‖, P < 0.001.
While examining the small intestines of a small cohort of mice (n = 4), it was noticed that tumors of the Papillae of Vater occurred in Apc mutant mice with and without Rb1 mutation but not in WT animals (Fig. 1C). The frequency and progression of these tumors has not been examined.
Histopathological Examination of Tumors.
Approximately 400 tumors were excised from the RBVCA cohort and 70% of them were examined histologically. The results are summarized in Table 3. Adenocarcinomas were predominantly found in the small intestine in all genotypes. Thirty-three percent of the Rbtm2Brn −/− Tg (Vil-cre) Apc1638N+/− mice had adenocarcinomas with a tumor multiplicity of 0.93 carcinomas per mouse, compared with 23% in Rbtm2Brn +/+ Tg (Vil-cre) Apc1638N+/− mice with 0.2 carcinomas per mouse; however, the only significant difference for carcinomas was found between Rbtm2Brn −/− Tg (Vil-cre) Apc1638N+/− and the Rbtm2Brn +/+ Tg (Vil-cre) ApcW+ (P < 0.05). Although the combined Apc control group did not differ significantly, one of the four individual control groups making up the combined control group, Rbtm2Brn +/+ Tg (Vil-cre positive) Apc1638N siblings, had no GI tumors and therefore was significantly different from the mutants.
Villous adenomas are considered to have a high potential to become malignant. Mice without Rb1 were found to have a higher overall incidence and multiplicity of such villous and tubulovillous adenomas date [supporting information (SI) Tables S1 and S2].
Small and Large Intestinal Tumors Initiate by Apc Mutation.
Sixty-nine RBVCA tumors were examined with in vitro transcription and translation for nonsense mutations in the WT Apc allele (Table 4). Our results suggest that the initial events leading to tumor formation in the different tumor types of Apc1638N Rb1−/− mice are similar. We have also compared Apc1638N Rb1−/− results with accumulated data for Apc1638N Rb1+/+ (22). The predominant mutation found in Apc1638N Rb1−/− tumors is base substitution (94.1%), and is also the predominant mutation found in Apc1638N Rb1+/+ tumors (66.7%). By Two-tailed Fisher's exact test this difference is not quite significant (P = 0.0877). The frequency of insertion/deletion mutations is also not quite significantly different (Apc1638N Rb1−/−, 5.9%; Apc1638N Rb1+/+, 33.3%; P = 0.00877).
Table 4.
Apc mutations in Apc1638N tumors with Rb-1 deficient GI tract
| Codon | Mutation | Consequence | WT sequence | Rb−/−Apc1638N | Apc1638N¶ |
|---|---|---|---|---|---|
| 808 | Δ G | Frameshift | GAT AGT AGG TCA GAC | 1 | |
| 854 | C → T | Arg → Stop | AGA GAG CGA GGT | 1* | |
| 867 | G → T | Glu → Stop | ACA GAA AAT | 1‡ | |
| 874 | C → T | Arg → Stop | TCA AAA CGA GGT | 1† | 1 |
| 921 | C → T | Arg → Stop | GCG GCA CGA AGA AGC | 1* | 3 |
| 933 | C → A | Tyr → Stop | AAC ACA TAC AAC TTC | 1‡ | |
| 933 | C → G | Tyr → Stop | AAC ACA TAC AAC TTC | 2*† | 1 |
| 934 | +TACA | Frameshift | AAC ACA TAC AAC TTC | 1 | |
| 939 | G → T | Glu → Stop | AAG TCG GAA AAT | 1§ | 2 |
| 944 | Δ38bp | Frameshift | TCA AAT AGG ACA TGC TCT ATG CCT TAT GCC AAA GTG GAA TAT AAA CGA TCT TCA AAT | 1 | |
| 956 | C → T | Arg → Stop | TAT AAA CGA TCT TCA | 4*‡ | 1 |
| 992 | Δ 8bp + A | Frameshift | AAA TTT TGC AGT TAT | 1 | |
| 993 | C → A | Cys → Stop | AAA TTT TGC AGT | 1‡ | |
| 1018 | G → T | Glu → Stop | GAT GGA GAA CTG | 2† | 1 |
| 1025 | ΔAC | Frameshift | ATA AAT TAC AGT CTT | 1 | |
| 1047 | G → A | Trp → Stop | GAA AGG TGG GCA AGA | 1 | |
| 1078 | G → T | Glu → Stop | TCT GAG AAT | 1‡ | |
| 1115 | +GAATGGGTT | Frameshift | GGT TGA ATG GGT TCT AGT | 1‡ | |
| 1127 | +T | Frameshift | CAG TCT CTG TGT CAG | 1 | |
| 1141 | C → A | Trp → Stop | ACC AAC TAC AGT GAA | 1 | |
| 1143 | G → T | Glu → Stop | TAC AGT GAA CGT TAT | 1 | |
| Total | 17 | 18 |
Apc1638N tumors: *, Cecal; †, Duodenal; ‡, Jejunal; §, Colonic Rb−/−.
¶Data from ref. 22.
Duodenal and Cecal Tumors Have Different Patterns of Gene Expression.
We compared patterns of gene expression in tumors from the small and large intestine. Tumor RNA was examined by using Affymetrix 430_2A gene chips. Three independent tumors were profiled from each location and the top ranked fold-different probe sets were identified from those that had raw robust multichip average signals greater than 6.0 with false discovery rate (FDR) P < 0.1 for cecum versus duodenum (Student's t test and Benjamini Hochberg FDR were calculated) and > 25% fold-difference between cecal and duodenal tumors. This identified 832 probesets overall, with 312 elevated in duodenum and 520 elevated in cecal tumors. Mapping these to human gene orthologs identified 203 nonredundant human genes from duodenal tumors and 367 from cecal tumors. Each of these gene lists were mapped to HG-U133v2.0 Affymetrix GeneChip probesets and examined for their expression in human colorectal cancers relative to normal adult colon (23). Results from this analysis are presented in Fig. 2. The patterns of gene expression in tumors from the two GI compartments are clearly distinguishable from each other and involve a significant number of genes. Focusing on genes over-expressed in both mouse cecal tumors and human colorectal cancers revealed a set of genes associated with early tissue development, growth, and extracellular matrix disruption (Hoxa11, Hoxa10, Hoxd3, Hoxd10, Hoxd9, Hoxd8, Hoxb5, HNF4A, BMP4, IGF1, FGFR1, FGFR2, IL15, Ppp1r14a, and MMP9). In contrast, genes over-expressed in duodenal tumors and human CRC were only marginally enriched with respect to known gene ontologies and pathways. The few genes found, however, were associated with the maintenance of mucosal integrity, lysosome proteolysis, and protein secretion.
Fig. 2.
(A) Cluster analysis of differentially expressed genes from cecal adenomas and duodenal adenocarcinomas. Three duodenal and three cecal tumors from six individual mice were compared, and subsets of genes identified that were relatively up and down regulated for the two tumor types. Red, represents over-expressed; blue, under-expressed. (B) Comparison of gene expression patterns in mouse tumors with human colorectal tumors. Each set of mouse genes that are over- and under-expressed in the cecal or duodenal tumors were compared with the levels of expression in human colorectal cancers. Red, over-expressed; blue, under-expressed.
Discussion
We have generated mice that are deficient for Rb1 throughout the GI tract and are also heterozygous for a mutation in the Apc gene. These mice are viable and have a median survival of 9 months, and die significantly earlier than their Apc1638N+/− and WT littermates. Without the Apc mutation, Rb1-deficient mice survive ≈1 year and then succumb to a latent phenotype that includes tumors of the pituitary and the thyroid. A proportion of RBVCA mice also have early C cell and pituitary tumors (data not presented); all have GI tumors in the small and large intestine that appear to temporally precede the latent phenotype. We have examined the intestines of the RBVCA mice and find a significant increase in tumors of the large intestine, specifically the cecum and the proximal colon, indicating involvement of the Rb1 tumor suppressor function at those sites. We have examined relative expression profiling of tumors of the cecum and duodenum from the Rb−/−, Apc1638N+/− mice. We have found truncation mutations to Apc indicating that tumors of the cecum initiate by the same mechanism as tumors of the small intestine, and that lack of Apc is critical for development of both tumor types.
Mice carrying different mutations in the Apc gene have been previously described (24–27). Some of these mice develop large numbers of adenomas that are restricted to the small intestine and die relatively early in their lives. Others live longer because they develop fewer tumors, also mostly in the small intestine. Beyond the inactivation of the Apc gene, no consistent genetic alterations have been described. Our results suggest that Rb1 mutations in these cells leads to an expansion of the compartment in which tumors develop, and the inactivation of the Rb1 gene also results in more aggressive tumors with a greater potential to become invasive.
Because human intestinal tumors are mostly found in the colon, it has been perplexing that Apc mutant mice predominantly do not. The small percentages of human GI tumors that have little or no pRb1, like the RBVCA mouse model, do significantly correlate with tumors of the cecum (17). Our data show that haplo-insufficiency of Rb1 can also cause expansion of tumors into the large intestine and stomach. Given the altered expression of Rb1 in human CRC, this could have important implications. It has been reported that many CRCs paradoxically have increased levels of Rb1 expression. Relative under- or over-expression of Rb may therefore facilitate the deregulation of its biochemical pathway(s), critical for tumor development in this compartment. Speculating further, it may be that increased Rb1 expression levels are the result of cell cycle adjustment because of progressive Apc mutation. Cross talk between the pRb/E2F and Wnt/beta-catenin signaling pathways is thought to occur (28). Apc can regulate CDKN1A (p21/WAF1) (29), a potent cyclin-dependent kinase inhibitor that binds to CDK2 and CDK4/6, which in turn regulate Rb. In such a model pRb would contribute to cell cycle inhibition in the face of progression toward tumorigenesis by sequential Apc mutation, until even at increased levels it could no longer control the cell cycle, giving way to tumorigenesis. Rb1 deficiency would lower the threshold for tumorigenesis to occur upon Apc mutation.
In our initial expression profile experiments from Rb1-deficient cecum:duodenum, we have identified subsets of genes differentially expressed in the two tumor types. The cecum:duodenum experimental design was established without “normal” murine tissue controls to avoid the problem of having to identify the actual progenitor cells from which the tumors arise. Therefore, the subsets identified likely include both normal cecal and duodenal gene expression as well as deregulated gene expression because of oncogenic transformation. The data have some limitations, particularly when considering tumor versus nontumor oncogenic programs. However, by focusing on differential expression, we identify a unique set of genes that confer differential tumor behavior in the two tissue types. These genes can be considered additive to over- and under-expressed genes that the two locations have in common, and that are by definition tissue-specific independent oncogenes. Interestingly, many of the differentially regulated genes do have prior implication in CRC. For example, we find recapitulation of embryonic transcription patterns, substantiating earlier reports (23). Specifically, duodenal tumors from these animals exhibited higher expression of developmental regulators Gkn1, GATA 4, GATA 5, Lhx6, Foxq1, Barx1, Nfe211, and also terminally differentiated-associated genes Aqp1, Aqp3, and Aqp5. Cecal tumors exhibited relative over-expression of Hif1a, and early GI developmental regulators Hoxa11, Hoxa10, Hoxd3, Hoxd10, Hoxd9, Hoxd8, Hoxb5, Fgfr2, and Ppp1r14a. The over-expression of Hox genes has been substantiated by a recent report that has indirectly linked increased Hox gene expression with Rb1deficiency. Suzuki and Hemmati-Brivanlou have shown that xE2F over-expression in Xenopus embryos induces ectopic expression of Hox genes and results in suppression of dorsoanterior structures of the embryo (30).
We think it will be of interest to compare expression profiles of Apc1638N tumors with and without Rb1 deficiency in the future. Because microarray data for Apc1638N duodenal adenomas with WT Rb1 have been previously deposited in the National Center for Biotechnology Information (NCBI) (31), we compared our data from Rb−/−, Apc1638N+/− with theirs. We first determined, by using their normal mucosa dataset that 30–40% of the genes deregulated in the cross species signature reported by Gaspar et al. (31) were also deregulated in our dataset. Next, we found that direct comparison of duodenal tumor data from Rb−/−, Apc1638N+/− mice to those from Rb+/+, Apc1638N+/− mice showed many differences in regulation of genes known to be involved in cancer, cell cycle, cell growth, cell death, and DNA replication. These differences in canonical pathways included protein ubiquitination, hypoxia signaling, PI3/AKT signaling, oxidative phosphorylation, and ERK/MAPK signaling pathways. Notably, AKT expression was up-regulated in Rb−/−, Apc1638N+/− duodenal tumors, correlating with increased tumor size and progression. Because the comparison of arrays generated in different laboratories has the potential for a higher rate of false-positive and negative values, additional experiments will be required to confirm these data and to make a true comparison.
Materials and Methods
Generation of Apc1638N Mice with Rb1-Deficient GI Tracts.
Mice homozygous for the FVB;129-Rb-1tm2Brn conditional mutation and positive for the B6;D2-Tg(Vil-Cre) 20Syr transgene were developed as previously described (21). These animals were mated to animals bearing the Apc1638N allele on a C57/Bl6 background (26). All mice examined for GI tumorigenesis and general pathology were siblings, except where noted for comparison, and therefore on the same genetic background. Mice were maintained in a barrier facility approved by the Association for Assessment and Accreditation of Laboratory Animal Care and fed Purina's Picolab Mouse Diet 20 Food Type 5058.
Genotyping of Offspring.
Offspring were genotyped by PCR amplification as described previously for Rb1 (32), for Villin-cre (21), and for Apc1638N (26). DNA was made from tail biopsies at 10 days of age by using DNeasy Kits from Qiagen.
Generation of Survival Curves.
Prism software by GraphPad was used to generate survival curves.
Histopathological Analysis of GI Tumors.
Tumors of the GI tract were examined systematically. After killing the mice, the entire GI tract was removed, opened longitudinally, and fixed in 10% neutral-buffered formalin. The gross specimens were examined under a dissecting microscope and the number and location of the tumors was recorded. Representative sections were stained with H&E for histological analysis. By definition, invasive adenocarcinomas were those that had invaded into the muscularis and beyond. The Fisher's exact probability and X2 tests were used for the analysis of tumor incidence, and the Mann-Whitney and binomial exact calculation tests for tumor multiplicity.
Microarray Procedures.
We conducted gene expression analysis by using GeneChip Mouse Genome 430 2.0A arrays (Affymetrix). Duodenal mouse tumors (including tumor tissue arising from the pyloris) and cecal tumors were excised with the aid of a Zeiss Stemi SV6 dissecting microscope, and immediately frozen in liquid nitrogen. RNA was extracted from three duodenal and three cecal tumors from six different individual mice, and was purified by using Qiagen RNeasy Kits, and analyzed at the Harvard Partners Center for Genetics and Genomics Microarray Facility. RNA quality was assessed by RNA Nano LabChip analysis on an Agilent Bioanalyzer 2100. Concentrations were determined by using a NanoDrop 1000 spectrophotometer. Processing of RNA for GeneChip analysis was in accordance with methods described in the Affymetrix GeneChip Expression Analysis Technical Manual revision four. Scanning was carried out on a GeneChip Model 3000 7G scanner with autoloader, controlled by the Affymetrix GCOS v1.3 operating system.
Data Analysis.
Data analyses for cecum and duodenum ratios compared with human CRC were carried out with GeneSpring version 7.3.1 (Silicon Genetics) software, including normalization, T tests, and hierarchical clustering. Relevant files have been submitted to the NCBI gene expression and hybridization array data repository (GEO, www.ncbi.nlm.nih.gov/geo). Genes whose expression was higher in cecal tumors or higher in duodenal tumors were related to normal colon and human colorectal cancers by a meta-analysis procedure that mapped the mouse genes onto the human data set as published by Kaiser et al. (23). This procedure permits recognition that the cecal-elevated set of genes, including the cluster of Homeobox genes, are among those characteristic of human colorectal cancer-reactivated normal embryonic colon genes.
Supplementary Material
Acknowledgments.
This work was supported by National Institutes of Health Grant CA-084301 (to R.K.).
Footnotes
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
This article contains supporting information online at www.pnas.org/cgi/content/full/0802933105/DCSupplemental.
References
- 1.Groden J, et al. Identification and characterization of the familial adenomatous polyposis coli gene. Cell. 1991;66:589–600. doi: 10.1016/0092-8674(81)90021-0. [DOI] [PubMed] [Google Scholar]
- 2.Bodmer WF, et al. Genetic analysis of colorectal cancer. Princess Takamatsu Symp. 1989;20:49–59. [PubMed] [Google Scholar]
- 3.Fishel R, et al. The human mutator gene homolog MSH2 and its association with hereditary nonpolyposis colon cancer. Cell. 1993;75:1027–1038. doi: 10.1016/0092-8674(93)90546-3. [DOI] [PubMed] [Google Scholar]
- 4.Knudson AG., Jr Mutation and cancer: Statistical study of retinoblastoma. Proc Natl Acad Sci USA. 1971;68:820–823. doi: 10.1073/pnas.68.4.820. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Nakamura M, Yonekawa Y, Kleihues P, Ohgaki H. Promoter hypermethylation of the RB1 gene in glioblastomas. Lab Invest. 2001;81:77–82. doi: 10.1038/labinvest.3780213. [DOI] [PubMed] [Google Scholar]
- 6.Simpson DJ, Hibberts NA, McNicol AM, Clayton RN, Farrell WE. Loss of pRb expression in pituitary adenomas is associated with methylation of the RB1 CpG island. Cancer Res. 2000;60:1211–1216. [PubMed] [Google Scholar]
- 7.Gonzalez-Gomez P, et al. CpG island methylation status and mutation analysis of the RB1 gene essential promoter region and protein-binding pocket domain in nervous system tumours. Br J Cancer. 2003;88:109–114. doi: 10.1038/sj.bjc.6600737. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Edamoto Y, et al. Alterations of RB1, p53 and Wnt pathways in hepatocellular carcinomas associated with hepatitis C, hepatitis B and alcoholic liver cirrhosis. Int J Cancer. 2003;106:334–341. doi: 10.1002/ijc.11254. [DOI] [PubMed] [Google Scholar]
- 9.Harbour JW, et al. Abnormalities in structure and expression of the human retinoblastoma gene in SCLC. Science. 1988;241:353–357. doi: 10.1126/science.2838909. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Haigis K, Sage J, Glickman J, Shafer S, Jacks T. The related retinoblastoma (pRb) and p130 proteins cooperate to regulate homeostasis in the intestinal epithelium. J Biol Chem. 2006;281:638–647. doi: 10.1074/jbc.M509053200. [DOI] [PubMed] [Google Scholar]
- 11.Scambia G, Lovergine S, Masciullo V. RB family members as predictive and prognostic factors in human cancer. Oncogene. 2006;25:5302–5308. doi: 10.1038/sj.onc.1209620. [DOI] [PubMed] [Google Scholar]
- 12.Ali AA, et al. RB1 protein in normal and malignant human colorectal tissue and colon cancer cell lines. Faseb J. 1993;7:931–937. doi: 10.1096/fasebj.7.10.8344490. [DOI] [PubMed] [Google Scholar]
- 13.Vogelstein B, et al. Allelotype of colorectal carcinomas. Science. 1989;244:207–211. doi: 10.1126/science.2565047. [DOI] [PubMed] [Google Scholar]
- 14.Reichmann A, Martin P, Levin B. Chromosomal banding patterns in human large bowel cancer. Int J Cancer. 1981;28:431–440. doi: 10.1002/ijc.2910280407. [DOI] [PubMed] [Google Scholar]
- 15.Muleris M, et al. Characteristic chromosomal imbalances in 18 near-diploid colorectal tumors. Cancer Genet Cytogenet. 1987;29:289–301. doi: 10.1016/0165-4608(87)90239-1. [DOI] [PubMed] [Google Scholar]
- 16.Palmqvist R, Stenling R, Oberg A, Landberg G. Expression of cyclin D1 and retinoblastoma protein in colorectal cancer. Eur J Cancer. 1998;34:1575–1581. doi: 10.1016/s0959-8049(98)00162-2. [DOI] [PubMed] [Google Scholar]
- 17.Cui X, et al. Aberrant expression of pRb and p16(INK4), alone or in combination, indicates poor outcome after resection in patients with colorectal carcinoma. Hum Pathol. 2004;35:1189–1195. doi: 10.1016/j.humpath.2004.06.010. [DOI] [PubMed] [Google Scholar]
- 18.Yamamoto H, et al. Paradoxical increase in retinoblastoma protein in colorectal carcinomas may protect cells from apoptosis. Clin Cancer Res. 1999;5:1805–1815. [PubMed] [Google Scholar]
- 19.Lai PS, et al. Overexpression of RB1 transcript is significantly correlated with 13q14 allelic imbalance in colorectal carcinomas. Int J Cancer. 2006;119:1061–1066. doi: 10.1002/ijc.21945. [DOI] [PubMed] [Google Scholar]
- 20.Jacks T, et al. Effects of an Rb mutation in the mouse. Nature. 1992;359:295–300. doi: 10.1038/359295a0. [DOI] [PubMed] [Google Scholar]
- 21.Kucherlapati MH, Nguyen AA, Bronson RT, Kucherlapati RS. Inactivation of conditional Rb by Villin-Cre leads to aggressive tumors outside the gastrointestinal tract. Cancer Res. 2006;66:3576–3583. doi: 10.1158/0008-5472.CAN-05-2699. [DOI] [PubMed] [Google Scholar]
- 22.Chen P, et al. Novel roles for MLH3 deficiency and TLE6-like amplification in DNA mismatch repair-defiecient gastrointestinal tumorigenesis and progression. PLoS Genet. 2008;4(6):e1000092. doi: 10.1371/journal.pgen.1000092. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Kaiser S, et al. Transcriptional recapitulation and subversion of embryonic colon development by mouse colon tumor models and human colon cancer. Genome Biol. 2007;8:R131. doi: 10.1186/gb-2007-8-7-r131. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Oshima M, et al. Loss of Apc heterozygosity and abnormal tissue building in nascent intestinal polyps in mice carrying a truncated Apc gene. Proc Natl Acad Sci USA. 1995;92:4482–4486. doi: 10.1073/pnas.92.10.4482. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Moser AR, et al. ApcMin: A mouse model for intestinal and mammary tumorigenesis. Eur J Cancer. 1995;31A:1061–1064. doi: 10.1016/0959-8049(95)00181-h. [DOI] [PubMed] [Google Scholar]
- 26.Fodde R, et al. A targeted chain-termination mutation in the mouse Apc gene results in multiple intestinal tumors. Proc Natl Acad Sci USA. 1994;91:8969–8973. doi: 10.1073/pnas.91.19.8969. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Kuraguchi M, et al. Adenomatous polyposis coli (APC) is required for normal development of skin and thymus. PLoS Genet. 2006;2(9):e146. doi: 10.1371/journal.pgen.0020146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Heinen CD, et al. The APC tumor suppressor controls entry into S-phase through its ability to regulate the cyclin D/RB pathway. Gastroenterology. 2002;123:751–763. doi: 10.1053/gast.2002.35382. [DOI] [PubMed] [Google Scholar]
- 29.Yang W, et al. p21(WAF1/cip1) is an important determinant of intestinal cell response to sulindac in vitro and in vivo. Cancer Res. 2001;61:6297–6302. [PubMed] [Google Scholar]
- 30.Suzuki A, Hemmati-Brivanlou A. Xenopus embryonic E2F is required for the formation of ventral and posterior cell fates during early embryogenesis. Mol Cell. 2000;5:217–229. doi: 10.1016/s1097-2765(00)80418-9. [DOI] [PubMed] [Google Scholar]
- 31.Gaspar C, et al. Cross-species comparison of human and mouse intestinal polyps reveals conserved mechanisms in adenomatous polyposis coli (APC)-driven tumorigenesis. Am J Pathol. 2008;172:1363–1380. doi: 10.2353/ajpath.2008.070851. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Vooijs M, van der Valk M, de Riele H, Berns A. Flp-mediated tissue-specific inactivation of the retinoblastoma tumor suppressor gene in the mouse. Oncogene. 1998;17:1–12. doi: 10.1038/sj.onc.1202169. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.