Numerous experimental, epidemiologic, and clinical studies indicate that nonsteroidal anti-inflammatory drugs (NSAID) act as chemopreventive agents for colorectal cancer.1, 2, 3, 4, 5 Despite the clear involvement of COX-2 inhibition,6, 7 other molecular mechanisms seem also to be involved.8, 9, 10 Studies demonstrate that certain NSAID possess an ability to induce apoptosis and inhibit cell proliferation.11, 12, 13 COX-2 inhibition plays a role in these effects.14, 15 But as oncogenesis is a complex process with multiple molecular pathways and biological disorders of genomic instability,16 it is uncertain whether the biological actions of NSAID can protect against oncogenesis across this diversity.
Sulindac is representative of nonselective NSAID that have significant activity in inhibiting colon tumor formation in several settings. Sulindac inhibits azoxymethane (AOM)-induced tumor incidence and increases apoptosis in rat tumors17, 18 where the molecular pathway frequently involves K-ras defects19, 20, 21 but not other human-relevant genes such as APC and p53.22, 23 It also protects in Min mice,24 where APC function is defective. Additionally, induction of apoptosis by sulindac was reported in cases of familial adenomatous polyposis.11, 25, 26 Ability of sulindac or other NSAID to protect in the face of p53 dysfunction is unclear.
A proapoptotic effect of NSAID might regulate genomic defects at various stages throughout oncogenesis. Most of the data suggest that sulindac-induced apoptosis in the late promotion or progression stage is clinically relevant in the chemoprevention of colon neoplasms.17 Some studies also show that sulindac suppresses tumor incidence by enhancing apoptosis in carcinogen-initiated cells27, 28 and induces apoptosis in early appearing premalignant lesions.29 This raises the possibility that sulindac might also play an important role in apoptotic regulation of initiating events.
Carcinogen-based animal models provide a unique opportunity to study this as they are driven by carcinogen-induced mutational load. We have found previously that the acute apoptotic response to carcinogen (AARGC) in the mouse colon is p53-dependent with a gene-dosage effect and that decreased AARGC in p53+/− and p53−/− mice is associated with increased susceptibility to carcinogen-induced colorectal oncogenesis.30 This model presents the opportunity to determine whether chemopreventive agents can overcome the defects in homeostatic control mechanisms subsequent upon p53 dysfunction, including mechanisms controlling mutational load such as AARGC.
We test the ability of sulindac to reverse these p53-dependent defects. We tested whether it could restore the abnormal AARGC seen in p53+/−and p53−/− mice and determined if this was associated with protection against the consequent enhanced colorectal oncogenesis due to defective p53 function.
Material and methods
AOM and sulindac was purchased from Sigma (St. Louis, MO). Sulindac was dissolved in 1 M Tris-HCL, pH = 7.5; the final pH was adjusted to 7.2 with 10 N NaOH.
p53+/− mice on a background of C57BL/6J were a gift from Dr. A. Harris (Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia). The mice carry a target neo insert disrupting exon 2–6 of the p53 gene and were originally purchased from Jackson Laboratories (Bar Harbor, ME). Mice were housed in cages (4/cage) and maintained in a temperature and humidity-controlled animal facility with a 12-hr light-dark cycle. Mice were fed ad lib on rodent chow (Milling Industries Stockfeeds, Adelaide, Australia) and autoclaved water. All the mice were observed daily for clinical signs of ill health and weight was measured weekly. Mice appearing sick were euthanized. p53+/− mice were interbred at the Flinders Medical Centre Animal Facility to generate the range of genotypes studied, i.e., wild-type, p53+/− and p53−/−.
DNA from mice tail was extracted using DNeasy Tissue Kit (QIAGEN Inc., Hilden, Germany). Mice were genotyped by PCR analysis of tail DNA with primers (neo) oIMRO13 (5′ CTT GGG TGG AGA GGC TAT TC 3′) and oIMRO14 (5′ AGG TGA GAT GGC AGG AGA TC 3′) to amplify the disrupted allele, and primers oIMRO 336 (5′ ATA GGT CGG CGG TTC AT 3′) and oIMRO 337 (5′ CCC GAG TAT CTG GAA GGC AG 3′), to amplify the wild-type allele. The PCR product was then size-fractionated on 2% Tris-acetate-EDTA agarose gels and stained with ethidium bromide; the disrupted allele generated a 280-bp fragment and the wild-type allele generated a 500-bp fragment.
For apoptosis and cell proliferation assays, mice allocated to sulindac treatment received sulindac at 160 ppm in drinking water for a period of either 1 week or 4 weeks. For the tumor study, mice allocated to drug treatment received the same dosage of sulindac from 4 weeks of age until the termination of experiment. The average consumption of sulindac was 0.5 ± 0.1 mg/day/mouse, an amount equivalent to the dose that inhibits tumor growth reported in APCMin/+ mice and AOM-induced colon carcinogenesis rat model, without showing toxicity.18, 31
Experiment 1: induction of apoptosis by AOM.
Ten-week-old C57BL/6J mice were randomized into 9 groups (10–12/group) according to the p53 genotypes and treatments. Three groups were studied from each genotype. One did not receive sulindac treatment and served as control, whereas the other 2 received sulindac for either 1 or 4 weeks. All groups in each genotype received a single i.p. injection of AOM (10 mg/kg) to induce the acute apoptotic response to DNA damage at the end of the drug period as described previously.32 Eight hours after giving AOM, animals from all of the 9 groups were killed by CO2-induced narcosis. Immediately after death, the colon was rapidly removed, flushed clean with ice-cold saline and immediately fixed in 10% paraformaldehyde overnight at room temperature, followed by 70% ethanol treatment. A standardized segment of distal colon just proximal to the colorectal junction was embedded in paraffin. Trans-axial sections of 5-μm thickness were taken for histological and immunohistological examination.
Experiment 2: induction of colon tumor by AOM.
At 4 weeks of age, C57BL/6J mice were randomized into 6 groups (8–29/group) according to the p53 genotypes and treatments. Two groups were studied from each genotype. One did not receive sulindac treatment and served as controls, whereas the other received sulindac from age 4 weeks for the duration of the experiment. All groups received 3 weekly i.p. injection of AOM (20 mg/kg) commencing 2 weeks after starting sulindac treatment. All mice were maintained on rodent chow and were observed daily for clinical signs of ill health. The experiment was terminated 20 weeks after last AOM injection. Mice were euthanized by CO2-induced narcosis and colons were immediately removed and flushed with iced saline to remove debris. Colons were opened longitudinally, flattened on Hibond C paper, and fixed in 10% paraformaldehyde overnight at room temperature and 70% ethanol for future inspection.
Detection and measurement of apoptosis
The frequency of epithelial cells undergoing apoptosis was determined on paraffin-embedded sections stained with H&E as described previously.32 Apoptotic cells were identified by characteristic morphology, i.e., by cell shrinkage and nuclear condensation and formation of apoptotic bodies. A halo was often seen around apoptotic cells. Twenty separate crypts from distal colonic segments were chosen and counted by an independent observer who was unaware of the experimental condition, at a magnification of ×100. The apoptotic index was calculated as the number of apoptotic cells per crypt column divided by the total number of cells in the column and multiplied by 100.
Apoptotic cells in tumors were quantified in a section taken from the middle of each tumor and stained with H&E, as described previously.30 A grid added to the ocular lens was used to count apoptotic epithelial cell nuclei. Ten random grid fields were chosen for each tumor sample and counted by an independent observer who was unaware of the experimental condition at a magnification of ×100. The total number of apoptotic epithelial cell nuclei from 10 grid fields was summed, than the total number of apoptotic nuclei was divided by the total number of epithelial cell nuclei to give the apoptotic index.
Measurement of epithelial proliferation
Proliferative activity was measured in epithelial cells using immunohistochemical staining with anti-proliferating cell nuclear antigen (PCNA) monoclonal antibody (PC-10 clone, Santa Cruz, CA) as described previously.32 In brief, sections were dewaxed by xylene, endogenous peroxidase activity quenched by 0.3% hydrogen peroxide for 20 min in 50% ethanol and 10% normal horse serum applied for 30 min to block non-specific staining. The PCNA antibody was applied at 1:500 dilutions overnight at room temperature. A Level 2 Ultra Streptavidin detection system (Signet Laboratories, Inc., Dedham, MA) was used utilizing biotinylated goat anti-mouse as the secondary antibody, followed by 3,3′-diaminobenzidine (Sigma, St. Louis. MO) for 5–10 min and hematoxylin for 3 min. In all cases, an independent observer who was unaware of the experimental condition counted the PCNA positive cells. The scoring for cell proliferation was the same as the method used to score apoptosis. PCNA-positive staining was determined by counting 20 separate crypts per section. Proliferation was expressed as cell turnover, i.e., proportion of cells stained by PCNA. The proliferation index was calculated as the number of DAB-positive cells stained by PCNA divided by the total number of cells in each crypt column multiplied by 100.
Colon tumor analysis
Using a dissecting microscope, colon was scored for tumor number and location by an independent observer who was unaware of mouse genotype and sulindac status as described previously.30 Tumors were removed and embedded in paraffin (5 μm) for histopathological analysis. All tumors were examined histologically and evaluated by an independent observer based on the criteria described previously.33 Adenoma was characterized by expansion of the mucosa layer, reduction in goblet cell number, cellular dysplasia, moderate loss of mucosal architecture by glandular growth and lack of invasion through the basement membrane. Adenocarcinoma was identified when there was typical cytological change, prominent cellular atypia, loss of cell polarity, marked distortion of glandular architecture and invasion.
Statistical analyses were carried out using both SPSS for Windows, version 10.0 (SPSS Inc., Chicago, Illinois) and State version 8 for Windows. Data are expressed as means with standard errors of mean. Comparisons of apoptosis, PCNA positive cells and body weight between control and sulindac treatment groups were analyzed using one-way ANOVA with correction for multiple comparisons by Tukey's post-hoc test. All incidence data (the proportion of mice with tumors) were analyzed using generalized linear model to compare the control to sulindac treatment groups. All nonparametric data (tumor multiplicity) were analyzed using Poisson regression. A probability value of p = 0.05 was used as the critical level of significance.
The body weights of mice who received sulindac were slightly but not significantly different from those not given sulindac (Fig. 1). Body weights were much lower in p53−/− mice whether given sulindac or not compared to wild-type and p53+/− mice, toward the end of experiment (p < 0.05). Three p53−/− mice on control diet died before the end of Experiment 2, the major cause of death was tumor burden. Tumor end-points were available in these mice.
Effects of sulindac on AOM-induced apoptosis and epithelial cell proliferation
The effect of sulindac (administered for 1 or 4 weeks) on apoptosis was examined and compared to controls not given sulindac (Fig. 2). Sulindac significantly increased apoptosis in wild-type mice and p53+/− mice (p < 0.05) compared to control groups. In p53+/− mice, sulindac restored the apoptotic response to the level seen in wild-type mice not given sulindac. Sulindac did not affect rates of apoptosis in p53−/− mice; p53−/− mice showed a near zero response to the carcinogen and sulindac. Sulindac treatment for either 1 week or 4 weeks had no effect on PCNA labeling index or crypt column height (Fig. 3).
Chemopreventive effect of sulindac in wild-type and p53 knockout mice
The effects of sulindac on AOM-induced colon tumor incidence and multiplicity in wild-type and p53 knockout mice are summarized in Table I. The incidence and multiplicity of colon tumors in AOM-treated wild-type, p53+/−mice and p53−/− mice was 36% (with an average of 0.8 tumors/mouse), 64% (with an average of 1.63 tumors/mouse) and 90% (with an average of 1.74 tumors/mouse) respectively. Administration of sulindac significantly reduced the incidence and multiplicity of AOM-induced colon tumors in all genotypes, i.e., wild-type, p53+/− (p < 0.01) and p53−/− mice (p < 0.05). With sulindac, incidence was 17% (with an average of 0.3 tumors/mouse) in wild-type mice, 38% (with an average of 0.8 tumors/mouse) in p53+/−mice and 63% (with an average of 1.0 tumors/mouse) in p53−/−mice. This represents significant reductions of tumor incidence and multiplicity versus controls by 53% and 63% in p53 wild-type mice, 41% and 51% in p53+/−mice, and 37% and 43% in p53−/−mice respectively. There was no apparent difference in the ratio of adenomas to adenocarcinomas with or without sulindac treatment (Table I).
Table I. Chemopreventive Effects of Sulindac on AOM-Induced Colon Oncogenesis in Wild-Type, p53+/− and p53−/− Mice
Rates of apoptosis in tumors and impact of sulindac are shown in Figure 4. There was a trend to an increase of apoptosis in wild-type and p53 knockout mice with sulindac treatment compared to no sulindac treatment although this did not reach statistical significance.
Our study shows that the reduced colonic apoptotic response to genotoxic carcinogen (AARGC) characteristic of p53+/−mice30 is corrected by sulindac. Sulindac also increases AARGC in wild-type mice. Previous studies have shown that certain other chemopreventive agents can increase AARGC in wild-type animals from normal baseline levels; such as dietary fiber,32, 34 nonstarch polysaccharides and resistant starch,35 and fish oil,36 but this is the first study to demonstrate that a defective AARGC can be reversed.
An apoptotic response to DNA damage is proposed to be an innate biological mechanism for protection against oncogenesis.37, 38 In the case of carcinogen-induced damage, apoptosis is one mechanism for reducing mutational load in the colon by eliminating DNA-damaged cells that might represent mutations capable of leading to neoplastic transformation.36, 38, 39 Our previous study showed that in the presence of p53 haploinsufficiency, the apoptotic response to carcinogen is defective and is associated with an increased risk of progression to cancer after carcinogen administration.30 Sulindac reduces this risk as we found that progression to colonic tumour formation was reduced in both wild-type and p53+/− mice by 53% and 41% respectively. In effect, p53+/− mice were not at increased risk for AOM-induced oncogenesis when given sulindac. As sulindac did not change cell proliferation rates, its protective effect might be mediated by the enhanced apoptotic deletion of cells initiated by the carcinogen.
Ability to reverse p53-related biological defects is of relevance to the control of human colorectal oncogenesis. Although evidence now points to several different molecular pathways of colorectal cancer development,16 the most common pathway often involves p53 abnormalities.40, 41 Whereas some attention has been paid to how exogenous factors interact with genetic factors in cancer development in terms of understanding the subtle and complex mechanisms that maintain epithelial homeostasis and genomic stability,42 there has been relatively little work addressing whether chemopreventive agents are effective in specific genetic contexts.
Sulindac has been shown to be able to reverse defective apoptosis characteristic of APCMin/+ mice due to overexpression of COX-2.18 Because p53 dysfunction has been shown to play a role in susceptibility to several types of tumors, including our previous study,30, 43, 44, 45 we tested whether sulindac would be protective in this human-relevant genetic context. Our results show that not only does sulindac overcome the colonic apoptotic defect found in p53 haploinsufficiency, it also reverses the enhanced risk of progression to cancer. This suggests that sulindac-enhanced AARGC reduces the colonic mutational load resulting from the carcinogen by eliminating DNA-damaged cells.
Several rodent models of colorectal cancer, especially the rat azoxymethane/dimethylhydrazine model and the APCMin/+ mice model, have proved to be valuable models for studying cancer prevention strategies.46 A recent article has compared the effect of dietary and chemopreventive agents on tumor incidence in rodent studies with those of clinical intervention studies of adenoma recurrence. Results for most agents tested were consistent across the animal and clinical models.47 The rodent models have the advantage of allowing more detailed analysis of the mechanisms (molecular and biological) and genetic context.48 Although our present study gives no insight into the molecular pathways by which sulindac is protective, it confirms that sulindac does protect in the context of reduced levels of active p53 and suggests that this is biologically mediated through reversal of an apoptotic defect.
Our results show, however, that the effect of sulindac in relation to p53 status is a complex one. Reversal of AARGC seems to depend on the presence of some functional p53 as AARGC is not restored in p53−/−mice (where it is absent). Sulindac does reduce the risk of AOM-induced tumors in the p53−/−mice. This further supports the notion that protection by sulindac is p53-independent but it demonstrates that protective actions in addition to enhanced AARGC must be operative, perhaps in the post-initiation stages. As we better understand the importance of genomic instability to colonic oncogenesis and the critically important role of p53,49 our results suggest that sulindac acts in some way, perhaps by activating apoptosis, to reduce genomic instability whether or not p53 is available.
Most studies suggest that sulindac-induced apoptosis in the later stages of oncogenesis is the important protective biological action17 with regard to the growth inhibition by apoptosis that has been reported in many animals18, 50, 51 or human studies.11 Sulindac does not always require COX inhibition or p53 induction52, 53 for apoptosis induction and is able to induce or restore apoptosis through activating a proapoptotic gene54 or inactivating antiapoptotic gene expression.13, 55 For example, apoptosis is completely abolished in response to NSAID in colonic carcinoma cells with a bax deficiency.56, 57, 58 In contrast to the requirement for p53 to mount a normal acute apoptotic response to DNA damage, we found previously that apoptosis in tumors is not dependent on p53 status.30 In our present study, sulindac treatment tended to increase apoptosis in tumors, even when one allele or both alleles of p53 were lost, although the effect was not significant. Others have shown that certain NSAID increase levels of apoptosis in tumors.17, 18, 59
This raises the question as to what stage of oncogenesis is the proapoptotic action of sulindac important? The recent review has concluded that NSAID suppression of tumors is most effective when begun before exposure to the carcinogen.4 Another report has shown that sulindac treatment suppressed AOM-induced oncogenesis at both early and late stages, however, it was only proapoptotic in the early stages.60 Yet a study from Reddy et al.28 has shown that early administration of sulindac is necessary for a chemopreventive effect colon cancer, which was related to increased apoptosis. As sulindac enhances apoptotic response to DNA damage, and given the apparent importance of this effect in the control of mutational load, sulindac seems to exert a significant proapoptotic effect early in oncogenesis. It is likely however, that it has a continuing effect throughout the process of oncogenesis as it is also preventive in p53−/− mice where AARGC is absent. Both might act to correct genomic instability.
In conclusion, although reduced levels of p53 impairs the acute apoptotic response to AOM-induced DNA damage in colonic epithelial cells, sulindac is capable of restoring this defect. Consequently, sulindac corrects the increased risk for AOM-induced oncogenesis found in p53 haploinsufficiency. Sulindac is also an effective chemopreventive agent in the absence of any p53 as it also protects against AOM-induced tumors in p53−/− mice. Clearly, sulindac is an affective chemopreventive agent in the colon in the context of p53 dysfunction. Some of the protective action of sulindac can be ascribed to a proapoptotic action early in oncogenesis, perhaps through reduction of the mutational load resulting from carcinogen administration. It would also seem to effectively suppress oncogenesis at later stages, perhaps by reducing genomic instability, as it remains protective when AARGC is absent. Further understanding of how chemopreventive agents interact with and modulate the apoptosis pathway may enhance the context for optimal use of chemopreventive agents.
The authors wish to thank O. Pennino for technical assistance and A. Harris (Walter and Eliza Hall Institute) for provision of the initial p53+/− mice.