Oncogenic K-ras promotes early carcinogenesis in the mouse proximal colon

Authors


Abstract

Oncogenic K-ras mutations are frequently observed in colon cancers and contribute to transformed growth. Oncogenic K-ras is detected in aberrant crypt foci (ACF), precancerous colonic lesions, demonstrating that acquisition of a K-ras mutation is an early event in colon carcinogenesis. Here, we investigate the role of oncogenic K-ras in neoplastic initiation and progression. Transgenic mice in which an oncogenic K-rasG12D allele is activated in the colonic epithelium by sporadic recombination (K-rasLA2 mice) develop spontaneous ACF that are morphologically indistinguishable from those induced by the colon carcinogen azoxymethane (AOM). Similar neoplastic changes involving the entire colon are induced in transgenic mice constitutively expressing K-rasG12D throughout the colon (LSL-K-rasG12D/Villin-Cre mice). However, the biochemistry and fate of K-ras-induced lesions differ depending upon their location within the colon in these mice. In the proximal colon, K-rasG12D induces increased expression of procarcinogenic protein kinase CβII (PKCβII), activation of the MEK/ERK signaling axis and increased epithelial cell proliferation. In contrast, in the distal colon, K-rasG12D inhibits expression of procarcinogenic PKCβII and induces apoptosis. Treatment of K-rasLA2 mice with AOM leads to neoplastic progression of small ACF to large, dysplastic microadenomas in the proximal, but not the distal colon. Thus, oncogenic K-ras functions differently in the proximal and distal colon of mice, inducing ACF capable of neoplastic progression in the proximal colon, and ACF with little or no potential for progression in the distal colon. Our data indicate that acquisition of a K-ras mutation is an initiating neoplastic event in proximal colon cancer development in mice. © 2008 Wiley-Liss, Inc.

Colon cancer is thought to develop in a stepwise fashion in which normal colon epithelium develops precancerous aberrant crypt foci (ACF), some of which progress to adenoma and then to carcinoma. This progression is thought to be mediated by accumulation of specific genetic mutations in proto-oncogenes and tumor suppressor genes. One of the most commonly mutated genes in this process is K-ras, which is mutated in ∼30% of colon cancers.1, 2 Several studies have analyzed K-ras mutations in ACF. Interestingly, although the absolute frequency of K-ras mutation varied considerably between reports, no significant difference in the frequency of K-ras mutations in ACF isolated from colon cancer patients (non-familial adenomatous polyposis) and the carcinoma was reported,3–6 suggesting that K-ras mutations are acquired early in the carcinogenic process and persist throughout neoplastic progression. However, the specific role of K-ras mutation in initiation and neoplastic progression has not been systematically studied.

Several transgenic mouse models allow expression of endogenous oncogenic K-ras in the untransformed colonic epithelium.7, 8 In the K-rasLA2 model, spontaneous intrachromosomal recombination results in expression of oncogenic K-ras under the control of the endogenous K-ras promoter in a small percent of cells throughout the body.7K-rasLA2 mice develop multiple, spontaneous ACF within the colon. These mice also develop lung adenocarinomas, and ultimately die from respiratory failure as a result of lung tumor burden.7 In a second model, the LSL-K-rasG12D mouse, conditional expression of an endogenous latent K-rasG12D allele is activated in a tissue-selective manner.8, 9 Tissue-specific expression of Cre-recombinase and activation of the K-rasG12D allele in the colonic epithelium results in diffuse hyperplasia and an ACF-like morphology throughout the colon.8 Despite the fact that ACF have been shown to be early neoplastic lesions,10 neither K-rasLA2 nor LSL-K-rasG12D mice develop spontaneous colon adenomas and carcinomas. It is possible that K-ras-mediated ACF require additional genetic mutations to promote neoplastic progression, and that these mutations are not acquired during the limited lifetime of these mice.7 Alternatively, it is possible that oncogenic K-ras-mediated ACF are unable to progress along the carcinogenic pathway. Since sporadic ACF are heterogeneous and vary in their ability to progress beyond the early, precancerous stage (reviewed by Alrawi et al.),10 it remains an open question whether oncogenic K-ras-mediated ACF are precursors to colon cancer.

In our study, we evaluated the role of oncogenic K-ras in initiation and neoplastic progression in the colon. We find that expression of oncogenic K-ras differentially activates proliferative and apoptotic signaling pathways in the proximal and distal colon of mice, resulting in increased proliferation in the proximal colon but increased apoptosis in the distal colon. In addition, we show that oncogenic K-ras-initiated ACF progress to dysplastic microadenomas in the proximal, but not distal colon, upon exposure to the colon carcinogen, azoxymethane (AOM). Taken together, our data demonstrate that oncogenic K-ras differentially regulates procarcinogenic signaling and susceptibility to colon carcinogenesis in the proximal and distal colon of mice.

Abbreviations:

ACF, aberrant crypt foci; AOM, azoxymethane; BrdU, 5′-bromodeoxyuridine; ERK, extracellular signal-regulated kinase; H&E, hematoxylin and eosin; K-ras, v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog; MEK, MAP/ERK kinase; RT-PCR, reverse-transcription polymerase chain reaction; TUNEL, TdT-mediated dUTP-biotin nick end labeling.

Material and methods

Mouse breeding and maintenance

Transgenic K-rasLA2 mice7 on a C57Bl/6 background were obtained from Dr. Tyler Jacks and maintained as heterozygotes by breeding with nontransgenic (NTG) C57Bl/6J mice (originally obtained from The Jackson Laboratory, Bar Harbor, ME). LSL-K-rasG12D mice8, 9 were obtained from the Mouse Models of Human Cancers Consortium Repository Frederick, MD, and were maintained as heterozygotes. Mice expressing Cre recombinase under control of the mouse villin 1 promoter (Vil-cre)11 were obtained from The Jackson Laboratory. All animals were housed in microisolator cages in a pathogen-free barrier facility and maintained at a constant temperature and humidity on a 12-hr light/12-hr dark cycle. Mice were provided with a standard irradiated rodent chow and filtered water ad libitum throughout the experiments.

Carcinogen exposure and ACF identification

NTG littermates of K-rasLA2 mice were enrolled in our previously described carcinogen protocol12 at 6 weeks of age. Mice were injected intraperitoneally with 10 mg/kg AOM once a week for 4 weeks. AOM-treated NTG mice, as well as age-matched K-rasLA2 mice, were harvested 10 weeks after the last AOM injection, the colons isolated, flushed with cold saline, slit open and fixed flat in 10% buffered formalin. After 4 hr, the colons were washed in cold PBS and stored in 70% ethanol at 4°C. Fixed colons were stained briefly with 0.5% methylene blue and evaluated for the presence of ACF using previously defined criteria.13–15 The location and multiplicity of each ACF was recorded. ACF were not found in the most proximal, ridged region of the colon, regardless of genotype or treatment, so this region (∼15% of the total colon) was eliminated from the analysis. The remaining colon was divided into 2 equal halves to define distal and proximal regions for analysis of regional differences in distribution of ACF in the colon. For histological analysis, ACF were identified in fixed colon, marked with permanent (red) ink on the luminal surface and isolated with surrounding normal tissue using a 4-mm biopsy punch. Isolated tissues were paraffin-embedded, sectioned and stained with hematoxylin to visualize the crypt structure. ACF were identified histologically as colonic crypts with aberrant morphological features.13–15 In a second experiment, K-rasLA2 mice were injected with AOM (or an equal volume of saline) and evaluated for ACF formation as described earlier.

Genetic analysis of transgenic mice

Genotypic analysis was performed using the protocols and primer sets described at the following web sites: for detection of the mutant K-ras allele in LSL-K-rasG12D mice containing a transcriptional termination Stop codon: the Mouse Repository of Mouse Models of Human Cancers Consortium web site: http://mouse.ncifcrf.gov/protocols.asp?ID=01XJ6&p_allele=Kras2<tm4Tyj>&prot_no=; for the detection of the Vil-Cre transgene construct: The Jackson Laboratory web site: http://jaxmice.jax.org/pub-cgi/protocols/protocols.sh?objtype=protocol&protocol_id=661; and for genotype analysis of K-rasLA2 mice: The Jacks Lab web site: http://web.mit.edu/ccr/labs/jacks/protocols/KrasLA_tables.html. PCR analysis to detect Cre-mediated recombination and deletion of the Stop element in the colonic epithelium of bitransgenic mice was performed using the protocols and primer set described at The Jacks Lab web site: http://web.mit.edu/ccr/labs/jacks/protocols/KrasCond_tables.html.

Colonic epithelial cell cytokinetic analysis

One hour before harvest, mice were injected intraperitoneally with 50 μg/kg 5′-bromodeoxyuridine (BrdU, Sigma-Aldrich, St. Louis, MO). Formalin-fixed proximal and distal colon tissues were analyzed for proliferation as determined by BrdU labeling.16, 17 Ten to twenty longitudinally cut colonic crypts were analyzed for BrdU-labeled cells and the proliferative index (average number of BrdU-labeled cells/crypt height) was calculated for each colon region. Apoptosis was detected by TdT-mediated dUTP-biotin nick end labeling of fragmented DNA (TUNEL) staining of formalin-fixed colon tissues using the DeadEnd™ Colorimetric TUNEL System (Promega, Madison, WI). The number of cells undergoing apoptosis in 100 longitudinally cut colonic crypts was quantified and expressed as apoptotic index as previously described.14, 18

K-ras expression and activity assay

Colonic epithelial cells were isolated from proximal and distal colon as previously described.19 K-ras activity was assessed by affinity isolation of GTP-bound Ras using a Raf-1 binding domain coupled to agarose beads (Upstate Biotechnology, Lake Placid, NY) as previously described.20 Active (GTP-bound) and total K-ras were detected by immunoblot analysis using a K-ras specific antibody (Santa Cruz Biotechnology, Santa Cruz, CA).

Immunohistochemical analysis

After deparaffinization and rehydration, tissue sections were processed for antigen retrieval as described by the manufacturer (DAKO, Carpinteria, CA) and treated with 3% hydrogen peroxide in methanol to inhibit endogenous peroxidases. Phospho-T202/Y204 ERK1/2 (p42/44) was detected using rabbit mAb (No. 4376, Cell Signaling, Danvers, MA) and Envision Dual + Labeled Polymer Detection System (DAKO, Carpinteria, CA). Tissues were counterstained with hematoxylin.

RT-PCR and immunoblot analysis

Colonic epithelium was isolated as previously described14 and subjected to immunoblot analysis for phospho-ERK1/2 and total ERK1/2 (Nos. 9101 and 9102, respectively, Cell Signaling). Densitometric analysis of immunoblots was performed using the Kodak Molecular Imaging System. Purified colonic epithelial cells were subjected to immunoblot analysis for PKCβII expression as previously described.19 Total RNA was isolated and subjected to quantitative RT-PCR for PKCβII mRNA as previously described,19 and normalized to glyeraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA abundance to control for RNA concentration.

ACF characterization

Colon lesions were categorized based on their size as follows: small ACF (1–5 crypts/focus), large ACF (6–10 crypts) or microadenoma (>10 crypts). Hematoxylin-stained slides of microadenoma from K-rasLA2 mice were evaluated histologically in a blinded fashion for the presence of the following characteristics of dysplasia21: (i) increased crypt height, (ii) crypt branching, (iii) serration of the lumen, (iv) thickening of the crypt wall, (v) nuclear stratification, (vi) loss of mucins and (vii) nuclear crowding. For each characteristic of dysplasia, the lesion was given a score of 0 (no change) to 4 (extensive alterations). The sum total of the scores (0–28) was used to establish a dysplasia score that represents the level of dysplasia in each lesion. At least 2 different H&E-stained slides were evaluated for each lesion and the highest dysplasia score was used for that lesion.

Statistical analysis

One-way ANOVA was used to compare the means of ACF number, size and distribution between experimental groups. ANOVA analysis was used to compare the means of the BrdU labeling index, mRNA abundance and level of dysplasia between experimental groups. Linear regression analysis was used to evaluate the interaction between dysplasia score of each lesion and its location within the colon (distance from rectum).

Results

Expression of oncogenic K-ras results in spontaneous ACF formation

Oncogenic K-ras mutations are detected in human colon carcinoma and ACF, suggesting that they play a role in early colon carcinogenesis.1 Consistent with this hypothesis, expression of the K-rasG12D allele in the colonic epithelium of K-rasLA2 mice results in spontaneous formation of ACF that are morphologically indistinguishable from ACF induced in NTG mice by the well-characterized, colon-selective carcinogen, AOM (Figs. 1a1e).13, 22 Both K-ras- and AOM-induced ACF are characterized by aberrant, disorganized crypt structure, increased crypt height and crypt branching (Figs. 1d and 1e). K-rasG12D and AOM induces similar numbers of ACF per colon (13.5 ± 5.7 and 8.8 ± 2.9, respectively, difference not significant) (Fig. 1f). However, the distribution of these precancerous lesions is strikingly different in these 2 models (Fig. 1g). Whereas AOM-induced ACF are largely confined to the distal colon (5.8 ± 2.1 ACF in the distal colon, compared to 3.0 ± 0.9 in proximal colon), K-ras-induced ACF are evenly distributed within both the distal and proximal colon (6.7 ± 3.5 and 6.8 ± 2.6, respectively).

Figure 1.

Oncogenic K-ras induces formation of ACF indistinguishable from AOM-induced ACF. (a and b) ACF were identified in methylene blue stained, formalin-fixed (a) proximal and (b) distal colons. ACF ranged in size from a single aberrant crypt (white arrow) to 10 or more aberrant crypts/foci (black arrow). (c–e) Hematoxylin-stained (c) normal colonic epithelium from a K-rasLA2 mouse, (d) ACF isolated from a K-rasLA2 mouse, and (e) ACF isolated from an AOM-treated mouse. The red color on the luminal surface of the ACF is the dye used to mark the ACF upon microscopic identification in the intact colon. (f and g) NTG littermate controls of K-rasLA2 mice were treated with AOM as described in Material and Methods. AOM-treated mice were harvested at 10 weeks after the last AOM injection. Colons from AOM-treated NTG mice and age-matched K-rasLA2 mice were analyzed for ACF number and distribution (see Material and Methods for details). Number of ACF in the (f) total and (g) distal and proximal colon is plotted as a box plot (first quartile, median and third quartile are illustrated). n = 6 mice/group. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

To confirm our results regarding the ability of oncogenic K-ras to induce early neoplastic changes in the colon, we turned to a second model of K-ras-mediated oncogenesis, the LSL-K-rasG12D mouse.9 These mice harbor a latent oncogenic K-ras allele flanked by a transcriptional termination Stop element that can be removed by Cre-recombinase.8 We crossed LSL-K-rasG12D strain to a transgenic Vil-cre mouse strain expressing Cre-recombinase throughout the colonic epithelium under control of the villin 1 promoter11 to generate bitransgenic LSL-K-rasG12D/Vil-cre mice. PCR analysis of genomic DNA from Vil-Cre, LSL-K-rasG12D and LSL-K-rasG12D/Vil-cre mice confirmed tissue-specific, Cre-mediated recombination and activation of the oncogenic K-ras allele in the proximal and distal colon of LSL-K-rasG12D/Vil-Cre mice (Fig. 2a). PCR using primers specific to the unrecombined, conditional LSL-K-ras allele (Stop) confirmed the presence of this allele in genomic DNA isolated from the tail (Tl), proximal (P) and distal (D) colon of LSL-K-rasG12D mice and the absence of this allele in Vil-Cre mice (Fig. 2a). As expected, bitransgenic LSL-K-rasG12D/Vil-cre mice exhibit loss of the unrecombined K-ras allele in the proximal and distal colon as a result of Cre recombinase-mediated recombination that excises the Stop element and activates oncogenic K-ras expression. Recombination was tissue-selective since the unrecombined K-ras allele was still detected in genomic tail DNA from LSL-K-rasG12D/Vil-cre mice (Fig. 2a). PCR reactions using primers specific for the wild type K-ras allele in the same genomic DNA samples were used as a loading control. Analysis of multiple LSL-K-rasG12D/Vil-cre mice indicates >90% recombination efficiency in both the proximal and distal colonic epithelium (data not shown). Histological analysis demonstrates that whereas LSL-K-rasG12D mice exhibit histologically normal colonic epithelium (Fig. 2b), LSL-K-rasG12D/Vil-cre mice exhibit widespread, disorganized crypt structure (Fig. 2c). Similar morphological changes have been observed in the colonic epithelium of bitransgenic LSL-K-rasG12D/FABP-cre mice.8 The morphological alterations in the colonic crypts of LSL-K-rasG12D/Vil-cre mice are remarkably similar to the ACF seen in K-rasLA2 and AOM-treated mice (Figs. 1d and 1e).7 Thus, oncogenic K-ras expression in the colonic epithelium mediates formation of ACF-like lesions in 2 separate mouse models of K-rasG12D expression.

Figure 2.

Constitutive expression of oncogenic K-ras results in ACF-like morphology throughout the colon. (a) Genomic DNA was isolated from tail (Tl), proximal colon (P) and distal colon (D) of Vil-Cre mice (Vil-cre/+), LSL-K-rasG12D mice (LSL) and LSL-K-rasG12D/Vil-Cre/+ mice (LSL/cre). The unrecombined mutant K-ras allele was detected by PCR analysis using primers specific to the Stop element (Stop). Amplification of the wild type K-ras allele (WT Kras) is shown as a control for DNA content. (N) negative control containing no genomic DNA. See Material and Methods for links to PCR primer sequences. (b and c) H&E stained colonic epithelium from the distal colon of (b) LSL mice and (c) LSL/Vil-Cre mice. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Expression of oncogenic K-rasG12D induces region-selective procarcinogenic alterations in the colonic epithelium

Oncogenic K-ras promotes oncogenesis by stimulating mitogenic signaling to induce cellular proliferation.8 To evaluate the effect of oncogenic K-ras expression on colonic epithelial cell proliferation in vivo, we analyzed 5′-bromodeoxyuridine (BrdU) incorporation into colonic epithelial cells from LSL-K-rasG12D/Vil-cre and LSL-K-rasG12D mice (Fig. 3a). LSL-K-rasG12D/Vil-cre mice exhibited a significant increase in proliferative index (defined as the number of BrdU-labeled cell/crypt) in the proximal colon when compared to LSL-K-rasG12D mice (Fig. 3a). Interestingly, however, no increase in proliferative index was observed in the distal colon of LSL-K-rasG12D/Vil-cre mice, demonstrating that oncogenic K-ras selectively promotes proliferation in the proximal colon but not the distal colon of these mice. The proliferative index of LSL-K-rasG12D mice did not differ from that observed in NTG or Vil-cre/+ mice (data not shown).

Figure 3.

Oncogenic K-ras induces differential effects on proliferation and apoptosis in the distal and proximal colon. Distal and proximal colonic epithelium from LSL-K-rasG12D mice (LSL) and LSL-K-rasG12D/Vil-Cre/+ mice (LSL/cre) were evaluated for BrdU incorporation and TUNEL-staining cells by immunohistochemistry. (a) Proliferative cells were detected by incorporation of BrdU. The proliferative index (number of BrdU-labeled cells/colonic crypt) is plotted for distal and proximal colon. (b) Cells undergoing apoptosis were detected by TUNEL staining. The apoptotic index (number of apoptotic cells/100 colonic crypts) is plotted for distal and proximal colon. Values represent mean ± SEM; n = 4–5 animals/group. (c) Distal and proximal colonic epithelium was isolated from LSL and LSL/cre mice and assayed for active (GTP-bound) Ras using the Ras-binding domain of Raf-1 as an affinity ligand. Active and total K-ras were detected by immunoblot analysis with an antibody specific for K-ras. Blot is representative of 3 independent assays.

Oncogenic K-ras signaling has been implicated in both procarcinogenic and proapoptotic signaling.23 We therefore analyzed the effect of activation of the oncogenic K-ras allele on apoptosis in the colonic epithelium. TUNEL staining revealed a significant increase in apoptotic index in the distal colon of LSL-K-rasG12D/Vil-cre mice compared to LSL-K-rasG12D mice (Fig. 3b). This effect is specific for the distal colon, since no difference in apoptotic index was observed in the proximal colons of LSL-K-rasG12D/Vil-cre compared to LSL-K-rasG12D mice (Fig. 3b). Thus, expression of K-rasG12D selectively increases proliferation in the proximal colon and apoptosis in the distal colon of these mice.

It is possible that the region-specific effects of oncogenic K-ras observed in this model are due to differences in the level of expression or activity of the oncogenic K-rasG12D allele within the proximal and distal colon. To address this issue, we determined the level of expression and activity of K-ras in the distal and proximal colon of LSL-K-rasG12D/Vil-cre mice (Fig. 3c). A significant >5-fold increase in K-ras activity (as measured by GTP-bound, active K-ras) was observed in both the proximal and distal colon of LSL-K-rasG12D/Vil-cre mice when compared to LSL-K-rasG12D mice (Fig. 3c, top panel). As expected, a concomitant ∼2-fold increase in expression of K-ras protein consistent with activation of the oncogenic K-rasG12D allele was observed in both the distal and proximal colon of LSL-K-rasG12D/Vil-cre mice when compared to LSL-K-rasG12D mice (Fig. 3c, bottom panel). Taken together, these data demonstrate that the region-specific effects of the oncogenic LSL-K-rasG12D allele on proliferation and apoptosis are not due to differences in the expression or activity of K-rasG12D in the proximal and distal colon of LSL-K-rasG12D/Vil-cre mice.

The Raf→MEK1→ERK1/2 signaling axis is a Ras effector pathway thought to play a critical, requisite role in oncogenic K-ras-mediated proliferation and cellular transformation.24 K-ras-mediated activation of the mitogenic Raf-MEK-ERK pathway, as characterized by elevated, active (phosphorylated) ERK1/2 levels, has been observed in colon tumors from humans and carcinogen-treated rats.25, 26 Thus, we assessed the level of phospho-ERK1/2 in the colonic epithelium of LSL-K-rasG12D and LSL-K-rasG12D/Vil-cre mice by immunohistochemical staining using a phospho-specific ERK antibody validated for use in the mouse intestinal epithelium (Fig. 4a).27LSL-K-rasG12D/Vil-cre mice exhibited a significant increase in phospho-ERK1/2 staining in the proximal colon but little or no change in phospho-ERK1/2 staining in the distal colon when compared to LSL-K-rasG12D mice (Fig. 4a). Immunoblot analysis of isolated colonic epithelial cells for phospho-ERK1/2 and total ERK1/2 confirmed the selective activation of ERK phosphorylation in the proximal colon of LSL-K-rasG12D/Vil-cre mice (Fig. 4b). When normalized to total ERK1/2 expression, we observed an increase of ∼2-fold in phospho-ERK1/2 in the proximal colon but no change in ERK1/2 phosphorylation in the distal colon (Fig. 4c). These results are consistent with both the region-specific increase in proliferation in the proximal colon induced by K-rasG12D and the well-documented role of MEK-ERK signaling in K-ras-mediated oncogenesis.

Figure 4.

Oncogenic K-ras induces ERK activation selectively in the proximal colon. (a) Immunohistochemical detection of active (phosphorylated) ERK1/2 (p42/44) in distal and proximal colon from LSL-K-rasG12D mice (LSL) and LSL-K-rasG12D/Vil-Cre/+ mice (LSL/cre). Preincubation of primary antibody with immunizing (blocking) peptide completely ablates immunohistochemical staining in the epithelial cells. (b) ERK1/2 phosphorylation and total ERK1/2 expression was determined by immunoblot analysis of distal (D) and proximal (P) colon extracts from LSL and LSL/cre mice. (c) The ratio of phospho-ERK1/2 to total ERK1/2 detected by immunoblot analysis is plotted (average of 2 experiments).

We previously demonstrated that the MEK-ERK pathway regulates expression of the procarcinogenic protein kinase C isozyme PKCβII in intestinal epithelial cells.19 PKCβII expression is induced early in colon carcinogenesis and overexpression of PKCβII in the colonic epithelium of transgenic mice promotes colonic epithelial cell proliferation and enhanced sensitivity to AOM-induced colon carcinogenesis.12, 14, 19 Furthermore, PKCβ null mice are highly resistant to AOM-mediated colon carcinogenesis.19 Thus, PKCβII is a critical downstream target of MEK/ERK signaling that is required for colon carcinogenesis.19, 20 We therefore assessed the effects of K-rasG12D activation on PKCβII expression in the proximal and distal colon by immunoblot analysis. PKCβII is expressed at a low but detectable level in the proximal colon of LSL-K-rasG12D mice and is significantly induced in the proximal colon of LSL-K-rasG12D/Vil-cre mice (Fig. 5a). A corresponding increase in PKCβII mRNA was observed in the proximal colon of LSL-K-rasG12D/Vil-cre mice when compared to LSL-K-rasG12D mice (Fig. 5b). In contrast, expression of PKCβII was significantly reduced in the distal colon of LSL-K-rasG12D/Vil-cre when compared to LSL-K-rasG12D mice (Fig. 5a). This decrease in PKCβII expression was accompanied by a significant decrease in PKCβII mRNA abundance in the distal colon of LSL-K-rasG12D/Vil-cre mice compared to LSL-K-rasG12D mice (Fig. 5b). Thus, oncogenic K-ras differentially regulates PKCβII mRNA abundance and protein expression in the murine colon in a manner consistent with the differential effects of K-ras on MEK/ERK signaling, proliferation and apoptosis in the proximal and distal colon.

Figure 5.

Oncogenic K-ras differentially regulates PKCβII expression in the proximal and distal colon. (a) Immunoblot analysis of PKCβII protein expression in colonic epithelium isolated from the distal and proximal colon of 2 LSL-K-rasG12D mice (LSL) and 2 LSL-K-rasG12D/Vil-Cre/+ mice (LSL/cre). Actin protein levels are shown as a loading control. Immunoblot results are representative of 3 independent experiments. (b) PKCβII mRNA abundance was determined by qRT-PCR analysis of mRNA isolated from the colonic epithelium of LSL and LSL/cre mice. RNA abundance was normalized to GAPDH. Values represent mean ± SEM; n = 3–6.

K-ras-mediated ACF in the proximal colon are differentially capable of progressing to microadenoma

Given the differential effects of oncogenic K-ras expression on colonic epithelial cell proliferation and apoptosis in the proximal and distal colon, we wished to assess whether K-ras-mediated ACF in the proximal and distal colon differ in their ability to progress to more malignant lesions. For this purpose, K-rasLA2 mice harboring K-ras-mediated ACF were treated with AOM and assessed for neoplastic progression. Neoplastic progression in the colon is characterized by an increase in the multiplicity of ACF (with increased crypt multiplicity correlating with progression)28 and the development of highly dysplastic microadenomas.29 Therefore, we assessed colonic lesions from AOM-treated K-rasLA2 mice using these parameters. Interestingly, AOM treatment of K-rasLA2 mice caused an increase in ACF number in the proximal colon but had no effect on ACF number in the distal colon (Fig. 6a). Evaluation of the multiplicity of ACF in the proximal colon of AOM-treated K-rasLA2 revealed that AOM treatment did not significantly affect the number of small and large ACF (Figs. 6b and 6c). However, AOM induced a significant increase in the number of large, high multiplicity microadenoma selectively in the proximal colon (Fig. 6d).

Figure 6.

Oncogenic K-ras- and AOM-induced ACF distribute differently Throughout the Colon. K-rasLA2 mice treated with saline (sal) or AOM (AOM) were examined for ACF number and distribution (see Material and Methods for details). (a) Number of ACF in the total colon, distal colon and proximal colon is plotted as a box plot (first quartile, median and third quartile are illustrated). (b–d) The number of (b) small ACF (1–5 aberrant crypts/foci), (c) large ACF (6–10 aberrant crypts/foci) and (d) microadenoma (>10 aberrant crypts/foci) in the proximal colon of saline and AOM-treated K-rasLA2 mice is plotted. n = 6 mice/group.

Colonic epithelial lesions are commonly categorized as aberrant crypt structures without dysplasia (ACF), or as dysplastic microadenomas (Fig. 7a). ACF without dysplasia exhibit elongated and branched crypts that are otherwise similar to normal colonic crypts. Dysplastic microadenomas have all of the characteristics of ACF without dysplasia, but in addition exhibit loss of mucins, nuclear crowding, stratification of the crypt epithelial lining, serration of the crypt lumen and thickening of the crypt walls. Lesions with these features of increased dysplasia are associated with an increased risk for colon cancer development.28–30 We therefore examined microadenomas from AOM-treated K-rasLA2 mice for evidence of dysplasia. Lesions were examined microscopically and graded on a scale of 0–4 for crypt height, branching, serration of the lumen, thickening of the wall, epithelial stratification, loss of mucins and nuclear crowding as well-established indices of dysplasia30 (described in Material and Methods) to generate an aggregate dysplasia score (Fig. 7b). As can be seen, the dysplasia score was significantly higher in proximal colon lesions than distal colon lesions. Furthermore, a significant positive correlation was observed between dysplasia score and distance of the lesion from the rectum (Fig. 7c). We conclude that oncogenic K-ras-induced ACF in the proximal colon of mice selectively undergo oncogenic progression in response to AOM, where as ACF in the distal colon do not.

Figure 7.

K-ras-initiated ACF in the proximal colon are capable of carcinogenic progression. (a) Hematoxylin-stained colon tissue from K-rasLA2 mice demonstrating normal, aberrant (nondysplastic) and dysplastic morphology. The red color on the luminal surface of the ACF is the dye used to mark the ACF upon microscopic identification in the intact colon. (b) Microadenomas isolated from K-rasLA2 mouse colon were evaluated for the level of dysplasia using the scoring system described in Material and Methods. Dysplasia scores for microadenoma in the distal and proximal colon are plotted as a box plot (first quartile, median and third quartile are illustrated). Microadenomas in the proximal colon are significantly more dysplastic than those in the distal colon, p = 0.03. (c) The level of dysplasia in K-rasLA2 microadenomas correlates with location in the colon. The dysplasia score for each microadenoma is plotted versus its distance from the rectum (n = 4 distal colon microadenoma and n = 6 proximal colon microadenoma). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Discussion

Mutation activation of the K-ras proto-oncogene occurs in ∼30% of sporadic colon cancers, making it one of the most frequent mutational events in colon cancer. Despite the prevalence of K-ras mutations, their role in initiation and progression of sporadic colon cancer is unclear. In humans, K-ras mutations are observed in precancerous ACF, suggesting a role for oncogenic K-ras in the early stages of transformation.1, 4, 5 However, since ACF are heterogeneous and not all ACF undergo neoplastic progression to adenoma and carcinoma10, 28 it remains an open question as to whether K-ras-mediated ACF are capable of neoplastic progression to adenoma. The K-rasLA2 mouse model, in which oncogenic K-ras can be expressed in the colonic epithelium, confirms that oncogenic K-ras drives formation of precancerous ACF, providing direct genetic proof that K-ras can initiate colon carcinogenesis in mice.7 However, since these mice do not generate spontaneous adenomas or carcinomas, it was unclear whether K-ras-mediated ACF are capable of progressing to adenoma. In this report, we assessed whether oncogenic K-ras-induced ACF have the capacity to undergo neoplastic progression. Treatment of K-rasLA2 mice with the colon-specific carcinogen AOM resulted in an increase in large, highly dysplastic microadenoma, demonstrating that oncogenic K-ras-induced ACF in the mouse colon can undergo neoplastic progression.

Despite the fact that K-ras induces ACF formation evenly throughout the distal and proximal mouse colon, the ability of these lesions to progress to microadenomas differs depending upon their location within the colon. Specifically, proximal colon ACF progress to highly dysplastic microadenoma after exposure to AOM whereas ACF in the distal colon do not. These results are particularly interesting in light of numerous reports on K-ras mutation frequency in human colon cancers. Whereas oncogenic K-ras mutations are observed in both proximal and distal colon cancers,2, 31–33 they occur at a significantly higher frequency in proximal versus distal colon cancers.2, 32, 34 Furthermore, K-ras mutation does not correlate with tumor progression or metastasis in proximal colon tumors, suggesting that proximal colon tumors acquire K-ras mutations in earlier stages of neoplastic transformation.32 In contrast, K-ras mutations in distal colon tumors correlate with tumor stage suggesting that these mutations may be acquired later in the carcinogenic process in the distal colon.32 Our data are consistent with these observations and demonstrate that K-ras-induced ACF in the mouse proximal colon are more likely to progress to colon cancer than those found in the distal colon.

The difference in the fate of K-ras-induced ACF in the proximal and distal mouse colon suggests that oncogenic K-ras induces different biochemical and cellular effects in these 2 regions. Indeed, biochemical characterization of the effects of oncogenic K-ras expression revealed striking region-specific differences. Oncogenic K-ras expression selectively induces procarcinogenic PKCβII expression and ERK1/2 activation, as well as hyper-proliferation, in the proximal colon of mice. In contrast, oncogenic K-ras expression induces an increase in apoptosis and marked suppression of PKCβII expression in the distal colon. The observation that oncogenic K-ras regulates PKCβII expression is particularly intriguing, since we have previously demonstrated that PKCβII plays a critical, indispensible role in colon carcinogenesis in mice.14, 19 PKCβII is required for AOM-induced colon carcinogenesis, and its expression is induced early in the carcinogenic process.12, 14 PKCβII expression is induced in a MEK/ERK-dependent manner, suggesting that oncogenic K-ras may regulate PKCβII expression by activation of the MEK1→ERK1/2 pathway.20 Overexpression of PKCβII in the colonic epithelium induces hyper-proliferation, and an increase in AOM-induced ACF and colon tumors.14 Our current results indicate that oncogenic K-ras-mediated regulation of PKCβII may be a critical determinant in whether oncogenic K-ras expression results in hyper-proliferation and neoplastic progression or apoptosis.

Expression of oncogenic K-ras in the colonic epithelium promotes proliferative signaling in the mouse proximal colon and apoptotic signaling in the distal colon. The proximal and distal colonic epithelia differ in their embryonic origins, cellular function, vascular blood supply and gene expression profiles.35 These cellular differences are thought to mediate the differential susceptibility to human colorectal carcinoma with distinct molecular characteristics (such as microsatellite instability or CpG island methylator phenotypes)36 and are also likely to determine the cellular response to expression of oncogenic K-ras. Differences in exposure to environmental factors, such as secondary bile acids which promote apopotosis at high concentrations37 (and should be more concentrated in the fecal water of the distal colon), may also influence the region-specific effects of oncogenic K-ras in mouse colonic epithelium. Therefore, both cellular and environmental differences are likely to contribute to the cellular response to oncogenic K-ras expression. The underlying mechanism(s) for the observed regional differences are likely complex and will be the subject of further investigation.

In summary, we report several novel and important observations regarding the role of oncogenic K-ras in colon carcinogenesis. First, we demonstrate that oncogenic K-ras induces formation of precancerous ACF throughout the colon in mice. Second, we demonstrate that these K-ras-mediated ACF differ in their potential to progress to microadenoma depending upon their location within the colon. In the mouse proximal colon, K-ras-induced ACF are capable of neoplastic progression to adenoma, whereas those in the distal colon are not. Third, we demonstrate that oncogenic K-ras induces markedly different biochemical and cellular responses in the proximal and distal colon in these mice, consistent with the ability of oncogenic K-ras to produce ACF with the potential for neoplastic progression. Our data may have important implications for the use of K-ras mutations as a diagnostic for colon cancer. Specifically, our data suggest that the presence of a K-ras mutation in ACF in the proximal colon may be more predictive of colon cancer formation than those found in distal colon ACF. Evaluation of the distribution of K-ras mutations in proximal and distal colon ACF in humans will be needed to test this hypothesis. Our data suggest that by combining high resolution chromoendoscopic imaging of the proximal colon to identify and biopsy proximal colon ACF with genetic analysis for K-ras mutation in these lesions, we may increase our ability to identify patients predisposed to proximal colon cancer.

Acknowledgements

The authors acknowledge Ms. Sofija Rak and Ms. Deborah Huveldt for providing excellent technical support. They thank Dr. Tyler Jacks, Dr. David Tuveson and the Massachusetts Institute of Technology for supplying the LSL-K-rasG12D mice. AOM was provided by the National Cancer Institute's Chemical Carcinogen Reference Standard Repository. These studies were supported by NIH grants CA94122 (N.R.M), CA081436 (A.P.F) and The Mayo Clinic Foundation.

Ancillary