Colon cancer chemoprevention by a novel NO chimera that shows anti-inflammatory and antiproliferative activity in vitro and in vivo

The initiation of colon cancer is thought to begin with a single mutational event within an isolated colon crypt. The aberrant crypt foci (ACF), or a specific dysplastic subset of these lesions, are seen as an early precursor stage to adenomas and cancer. The ACF itself is a monoclonal structure that arises from mutations within a single crypt stem cell. That these ACF structures directly develop into adenomas is still a matter of debate, but ACF are strongly linked to colon cancer risk, and in animal models, such as the murine azoxymethane carcinogen model, a good correlation between ACF number and tumorigenesis has been reported for a wide variety of chemopreventive agents (1). The ACF number is a reliable biomarker of colon cancer in preclinical models, and furthermore, in humans, ACF show increased expression of markers of proliferation [proliferating cell nuclear antigen (PCNA)], and inflammation [inducible NO synthase (iNOS); ref. 2].

Colorectal cancer is a leading cause of death, and chemoprevention of colorectal cancer represents an important therapeutic target and an unmet need. Chemopreventive strategies for colorectal cancer have targeted antiproliferative and anti-inflammatory actions on colonocytes containing populations subject to carcinogen-induced DNA damage. In many cases, these chemopreventive agents act to modulate the levels of abnormally expressed proteins and to inhibit proliferation. In most cases, the exact mechanism of action is uncertain, although again, there are clear biomarkers that correlate with drug intervention in the development of ACF and tumorigenesis, in particular biomarkers of inflammation and of proliferation.

Nonsteroidal anti-inflammatory drugs (NSAID), such as acetylsalicylic acid (ASA), have shown promise in colorectal cancer clinical trials (3), but carry the burden of severe gastrointestinal side effects and loss of efficacy at low doses. Selective cyclooxygenase-2 (COX-2) inhibitors, which possess attenuated gastrointestinal side effects, have recently been confirmed by the U.S. Food and Drug Administration to manifest serious cardiovascular side effects. A further class of anti-inflammatory agents, nitric oxide donor NSAIDs (NO-NSAIDs), was originally designed to use the biological activity of NO to mollify the gastrointestinal damage caused by NSAIDs. NCX 4016 is an NO-ASA that is currently in National Cancer Institute (NCI)–sponsored clinical trials for colorectal cancer prevention, having shown promise in preclinical studies (Fig. 1; ref. 4). NO-NSAIDs are aliphatic nitrates conjugated via a labile ester linkage to an NSAID (5) and represent a new direction for nitrate therapeutic agents, which have been in clinical use for more than 130 years in cardiovascular therapy.

Rigas et al. (6) have positively assessed the potential for NO-ASA in cancer therapy and chemoprevention, but pointed out key gaps in our knowledge: the mechanism of action is not clearly defined; it is not known how classic nitrates compare with NO-NSAIDs; and drug optimization of improved NO-NSAIDs has been superficially explored. The novel nitrate, GT-094, is a prototype NO chimera, containing pharmacophores known to be effective in chemoprevention, including an NSAID (Fig. 1). To validate GT-094 as a lead compound for optimization of a nitrate specifically for colorectal cancer chemoprevention, GT-094 was studied in the traditional rat/azoxymethane colorectal cancer model. An extended rat/azoxymethane study provided tissues to examine drug effects on protein regulation and proliferation. Studies in cell culture showed that anti-inflammatory, antiproliferative, and cytoprotective effects seen in vivo were recapitulated in vitro. The NO chimera, GT-094, is reported for the first time as a lead for drug discovery and a probe compound for exploration of mechanisms of colon cancer chemoprevention.

Azoxymethane was obtained from the NCI Chemical Carcinogen Reference Standard Repository, MidWest Research Institute. Buffered formalin, xylene, and alcohols (histology grade) were from Fisher Scientific. Vincristine and all other biochemical reagents were from Sigma. The iNOS and poly(ADP-ribose)polymerase (PARP) antibodies were from Santa Cruz Biotechnology. Male Fisher 344 rats were obtained from Harlan Sprague-Dawley Labs and studied from 8 weeks of age (200–250 g).

Rats were allowed free access to water and Harlan Teklad 8640 mash diet. The rats were housed in micro-isolator cages with GreenTru laboratory bedding (Green Products) and exposed to a controlled light cycle of 14:10 h (light/dark) at 25°C. This study was approved by the University of Illinois at Chicago IACUC.

Rats received 15 mg/kg of azoxymethane s.c. twice, 7 days apart (total dose of 30 mg/kg). Drug administration was initiated 2 days after the last dose of azoxymethane. In the 8-week treatment protocol, the drug was given in feed for 8 weeks after initiation of drug treatment to animals in four arms (n = 8). Treatment diets were supplemented with arm 1: ASA (117 ppm); arm 2: GT-094 (233 ppm); or arm 3: ASA (117 ppm) + nitrates [ISDN 93 ppm; mannitol dinitrate (MDN) 80 ppm]. The control group (arm 4) received standard diet. The amount of ASA in feed was calculated to deliver ∼10 mg/kg/day, the same dosage used in the study of Bak et al. (4). The amounts of GT-094 and of ASA/nitrates in feed were calculated to be roughly equimolar with the ASA dosage. In arm 2 only, GT-094 was supplemented with twice-weekly administration of GT-094 by gavage [3.5-mg dose, i.e., 1 mL of a 5% DMSO solution (70 mg/mL) made up in 1% carboxymethylcellulose]. In the 28-week treatment protocol, a separate longer term study (n = 11) was conducted in which drugs were given 2 days after the last dose of azoxymethane for 8 weeks, using identical dosing to arms 2 and 4 in the 8-week treatment protocol. Immediately after the 8-week drug administration (i.e., in week 11 of the study), the administration of GT-094 by gavage was terminated, and the GT-094 treatment arm received GT-094 in feed alone for a further 20 weeks. Immediately after this 20-week period, animals were sacrificed. For reasons not associated with drug treatment, one animal was lost in the control group. In both protocols, rats were weighed weekly to examine drug perturbations of feeding and body weight (which were not observed) and were euthanized by CO₂ asphyxiation either 8 or 28 weeks after drug administration was initiated. Two hours before sacrifice, rats received vincristine 1 mg/kg by i.p. injection, thereby allowing the proliferative effects of azoxymethane on the colon to be assessed. In colon tissue from the 28-week study, proliferation was measured by vincristine metaphase arrest; inflammation was quantitated by measuring iNOS protein in Western blots and by immunohistochemistry; DNA damage was estimated by immunoassay of PARP protein; and p27 levels were estimated by Western blots.

Immediately postsacrifice, colons were excised, rinsed with ice-cold PBS, and opened longitudinally under direct vision using a Nikon SMZ-U stereomicroscope. Overall colon length as well as polyp and tumor location was determined by mounting the specimen on a glass slide marked in 1-cm increments. The colon was then stained in ice-cold 0.3% (v/v) methylene blue in PBS for 40 to 60 min. After staining, the surface of the colon was again examined but now at 40× to identify ACF. ACF were identified using the criteria of Bird and McLellan (7), and their position was recorded. To be considered an ACF, each structure had to have four of the five following criteria: crypts that were two to three times larger than normal; a thickened layer of epithelial cells; an increased pericryptal area; slit-shaped lumina; and be microscopically elevated above the plane of normal crypts in the preparation.

Gross examination of the liver, lungs, and peritoneum for secondary tumors was done in all animals. In the dissected rat colon, tumor number, size, weight, and location were recorded. Most colonic tumors were adenocarcinomas, and thus, we used only those for the analysis of variables. Each tumor was fixed individually, with the remainder of the colon processed as single specimen. All tissues were formalin fixed and paraffin embedded according to standard Armed Forces Institute of Pathology (AFIP) protocol. Blocks were sectioned (5 μm) using a Spencer Model 820 microtome (American Optical), heat fixed at 70°C for 20 min, and stained with H&E according to standard AFIP protocol.

The number of proliferating colonic crypt cells was determined by counting vincristine-arrested cellular metaphases from freshly dissected crypts according to the method of Goodlad (8). Briefly, after ACF analysis (described above), the proximal and distal colon were separately fixed in Carnoy's solution for 3 h and then stored in 70% alcohol. Individual crypts were prepared by hydrating in graded alcohols followed by immersion in 1 mol/L HCl for 10 min at 60°C. Colon tissue was counterstained with Schiff's reagent for 45 min until a deep magenta color developed. Crypts were isolated under direct stereomicroscopic vision after adding two drops of 45% (v/v) acetic acid and spreading under a large (24 × 40 mm) coverslip. Metaphase nuclei from 20 separate crypts were identified and counted at 400× from both the distal and proximal colon of all animals.

Protein lysates were prepared from excised rat colon in cold NaCl (150 mmol/L), EDTA (1 mmol/L), EGTA (1 mmol/L), sodium PPi (2. 5 mmol/L), β-glycerophosphate (1 mmol/L), Na₃VO₄ (1 mmol/L), 1% (v/v) Triton X-100, 1 μg/mL leupeptin, phenylmethylsulfonyl fluoride (1 mmol/L), Tris-HCl (20 mmol/L; pH, 7), before electrophoretic resolution via SDS-PAGE and transfer to nitrocellulose for immunoblotting. The blots were developed on Kodak Biomax film using a Luminol enhanced chemiluminescence kit. Quantitation of band intensity at 130 kDa is done using Adobe Photoshop image analysis software (Adobe Systems Inc.).

Immunohistochemistry was done using a two-stage modification of an immunoperoxidase technique that we have previously described (9). Briefly, after quenching peroxidase activity with 3% hydrogen peroxide and blocking nonspecific binding with goat serum, a 1:50 dilution of primary iNOS antibody was applied to 5-μm tissue sections and incubated at 4°C overnight. The sections were washed in PBS, followed by incubation with a horseradish peroxidase–labeled anti-rabbit immunoglobulin G for 30 min using the DAKO EnVision+ System. After a final wash in PBS, the polymer-bound antibody is detected with liquid 3,3′-diaminobenzidine substrate chromogen system for 3 to 5 min. The chromogen-stained tissue is counterstained for 30 s in Gill's modified hematoxylin and iNOS expression measured by Q-IHC as we have previously described (9).

Caco-2 human colonic adenocarcinoma cells obtained from the American Type Culture Collection were grown in DMEM/F-12 supplemented with 1% penicillin-streptomycin, 20% fetal bovine serum, nonessential amino acids, sodium pyruvate, and 1.5 g/L sodium bicarbonate and incubated in 5% CO₂ at 37°C. At confluency, cells were seeded in 96-well plates at a density of 2 × 10⁴ cells/mL in 190 μL media. After 24 h incubation, test samples were added to each well, and the cells were incubated for an additional 48 or 96 h. The cells were fixed with ice-cold 20% trichloroacetic acid, and cell count was done using sulforhodamine B (SRB) staining. The absorption was measured at 515 nm using a plate reader. Assays were done in three separate cell cultures. Cell cycle fluorescence-activated cell sorting (FACS) analysis was done by the propidium iodide–trypsin method on two separate cell cultures. After 24 h incubation with or without drug, cells were lysed with trypsin/EDTA and centrifuged (600 × g), and the pellet was resuspended in RPMI/PBS/10% FCS. Aliquots of 1 × 10⁶ cells were transferred to FACS tubes and centrifuged (600 × g), and the pellet was resuspended in citrate buffer (100 μL) before tryptic digest, RNase treatment, and staining with ice-cold propidium iodide (416 μL/mL) and spermine tetrahydrochloride solution (1.16 mg/mL) before FACS analysis.

The annexin V apoptosis assay was carried out on an alternative human colon adenocarcinoma cell line, HT-29, supplied by Dr. Murillo (Illinois Institute of Technology, Chicago, IL). Cells were maintained in RPMI 1640, supplemented with 1% antibiotic-antimycotic, 1% l-glutamine (200 mmol/L), and 10% fetal bovine serum (Atlanta Biologicals), and incubated in 5% CO₂ at 37°C. Incubations of HT-29 cells with GT-094 (50 or 100 μmol/L) were done according to standard methods (annexin V–FITC apoptosis detection kit; Sigma). After the indicated time points, the media were harvested. After workup, both floating and trypsinized cells were treated with annexin V–FITC and propidium iodide and analyzed by flow cytometry (Beckman Coulter Elite ESP) to quantitatively assess cells undergoing apoptosis.

The mouse RAW 264.7 cells, provided by Dr. J. Cook (University of Illinois at Chicago, Chicago, IL), were grown in DMEM supplemented with 10% heat-inactivated fetal bovine serum, 50 μg/mL penicillin, and 50 μg/mL streptomycin in bacteriologic plates at 37°C in humid 5% CO₂. Cells were cultured for no more than four passages. Cells were plated in DMEM and incubated at 37°C for 12 h. The medium was changed, allowing cell induction by addition of lipopolysaccharide (LPS) in the medium. Cell lysate supernatant from drug-treated and control incubations was taken for assay at 5, 12, and 24 h. Supernatant was assayed for NO₂⁻ using the Griess assay. Cell lysates were assayed for cell number using the SRB method or for iNOS content using Western blot immunoassay.

In the 8-week study, mean ACF number was significantly reduced in the GT-094–treated group (24 ± 4) of rats compared with azoxymethane-treated control rats (44 ± 9; P < 0.05; Fig. 2) by one-way ANOVA using Dunnett's post-test. Furthermore, the effects of ASA (44 ± 4) or ASA plus nitrates (42 ± 7) failed to reach significance for the reduction of ACF formation. In the 28-week study, the incidence of tumors in rats treated with azoxymethane alone was 50% (5:10), whereas tumors developed in 36% (4:11) of GT-094–supplemented animals; similarly, tumor multiplicity was reduced 1.8 versus 1.5 tumors per animal, as well as mean tumor weight 169 ± 81 mg versus 96 ± 36 mg in the GT-094–supplemented animals, but these differences did not reach statistical significance by unpaired t test.

In response to azoxymethane, normal crypts residing in the proximal and distal colon of the rat undergo an increase both in the number of proliferating crypt cells and an apical migration of metaphase-arrested cells up the native crypt (10). We evaluated proliferation 28 weeks after azoxymethane administration by counting metaphase-arrested cells in whole crypt preparations from five treated and untreated animals. In the proximal colon, GT-094 reduced proliferation by 69% [13.9 ± 7.6 versus 4.4 ± 2.6 metaphases per crypt (mean ± SD); P < 0.05]. Crypt proliferation rates in the distal colon were more modestly reduced [13.9 ± 2.9 versus 9.1 ± 2.9 metaphases per crypt (mean ± SD); P < 0.05], but this 30% reduction remained significant (Fig. 3).

Expression of the cell cycle inhibitor p27 was not significantly altered within the proximal and distal colon of untreated animals after 28 weeks of azoxymethane administration. However, GT-094–supplemented animals displayed marginally reduced p27 expression in the proximal colon (despite decreased levels of proliferation) compared with untreated animals, but also showed a marked (∼3-fold) increase in expression in the distal colon where DNA damage and ACF formation occur (P < 0.05; Fig. 3).

The iNOS expression identified in the 10-week study by immunohistochemistry was predominantly within and adjacent to colon crypt as previously shown in other azoxymethane models (11). ASA alone reduced iNOS expression by 27%, whereas GT-094 further reduced iNOS production by 67%, a ∼2.5-fold reduction compared with azoxymethane alone. The extent of this iNOS inhibition in GT-094–treated animals did not persist through 30 weeks of carcinogen exposure, declining to 33% of the levels expressed in untreated animals, although this reduction remained significant (Fig. 3; P < 0.05).

PARP-1 expression and cleavage were evaluated as a secondary marker of inflammation and DNA damage (12). Again, GT-094–treated animals showed significantly reduced amounts of cleaved and uncleaved PARP-1 expression in the distal colon compared with that of azoxymethane-treated animals. Uncleaved PARP was ∼4-fold higher in the azoxymethane-alone group, whereas the cleaved protein, representing cellular damage, was 2-fold higher in the animals not protected by GT-094 (Fig. 3).

Data collected in vivo showed antiproliferative and anti-inflammatory actions of GT-094 in rat colon; therefore, studies were conducted to determine if these effects were recapitulated in vitro. A concentration-response curve was obtained for GT-094 in the colon cancer, Caco-2, cell line, using SRB dye staining to obtain cell number at 48 h incubation, giving IC₅₀ ∼40 μmol/L (Fig. 4). At 100 μmol/L, GT-094 was cytostatic. The antiproliferative effects of GT-094 were further studied in Caco-2 cells by cell cycle FACS analysis, demonstrating a transient G₂-M phase block that was apparent at 6 h, but which was diminished at 24 h (Fig. 5). The Caco-2 cell cycle analysis gave no qualitative evidence for apoptosis induced by GT-094, but this aspect was studied in more detail by FACS analysis using annexin and propidium iodide staining in the HT-29 colon cancer cell line. The data analysis showed that GT-094 does not induce apoptosis in HT-29 colon cancer cell culture (Fig. 5).

The murine alveolar macrophage-like RAW 264.7 cell line has frequently been used to examine interference of agents with cellular inflammatory response. LPS treatment induces iNOS, which can be quantified by Western blot or by measurement of NO₂⁻ production because NO₂⁻ is the ultimate product of NO oxidative metabolism. Using the Griess assay, ∼30 μmol/L NO₂⁻ was measured in LPS-induced RAW 264.7 cells after 24 h incubation. The majority of nitrates, including ISDN, undergo denitration to NO₂⁻ as the major product (13). The supernatant from RAW 264.7 cells was assayed for NO₂⁻ in the absence of LPS induction: no significant amount of NO₂⁻ was detected in untreated cells; whereas ISDN-treated cells accumulated 2.7 ± 0.5 μmol/L NO₂⁻. Cells treated with GT-094 (10, 50, and 100 μmol/L) contained 3.4 ± 0.2, 16.2 ± 0.3, and 38.7 ± 0.7 μmol/L NO₂⁻, respectively, corresponding to ∼33% yield from GT-094. No significant differences were seen in [NO₂⁻] produced from nitrates at 9, 12, and 20 h. Simple subtraction of NO₂⁻ in the absence and presence of LPS induction yielded values for endogenous NOS-mediated NO₂⁻ production (Fig. 6). GT-094 substantially inhibited NO₂⁻ production in a concentration-dependent manner, which was not seen for ISDN at high concentration (100 μmol/L) with or without added ASA (100 μmol/L). Western blots of lysates from treated and untreated RAW 264.7 cells, with and without LPS induction, confirmed that GT-094 inhibited iNOS protein expression in response to LPS induction.

NO-NSAIDs are nitrates conjugated to an NSAID drug moiety via a labile linker. Collectively, NO-NSAIDs containing an ASA moiety are termed NO-ASA; one isomer, NCX 4016, has entered clinical trials for colorectal cancer chemoprevention. NCX 4016, together with its isomer, NCX 4040, have been the subject of elegant studies by Rigas et al. (6). NO-ASA molecules were originally designed to exploit the NO bioactivity of the nitrate group to attenuate the serious gastrointestinal toxicity caused by the action of NSAIDs; for example, NCX 4016 first entered clinical trials in 1996 for use in pain and inflammation. Despite NO-ASA not having been chemically optimized for chemoprevention, the published results are impressive (14). Given this promise, a prototype nitrate was selected, GT-094, to provide a lead compound for further structure-activity studies toward optimization for colorectal cancer chemoprevention (Fig. 1). Herein, we report in vivo and in vitro studies on GT-094, showing attributes considered beneficial for a chemopreventive agent.

The classic nitrate nitrovasodilator, nitroglycerin (GTN), has been in clinical use in therapy of angina for more than 130 years. The classic nitrate, ISDN, is important because it is now being used as a cardioprotective agent, expanding the clinical paradigm of nitrates to chronic, prophylactic therapy (15). Organic nitrates contain the nitrooxy group (–ONO₂) that provides bioactivity that mimics that of NO (13). Nitrates are also readily able to act as oxidizing agents toward thiols. Bioactivation of nitrates provides NO bioactivity, with no evidence for production of the higher levels of NO that can be observed from NO donors such as NONOates (13). Novel nitrates containing a disulfide linkage (e.g., GT-094) have been shown to differ from classic nitrates, manifesting NO bioactivity including antioxidant capacity and cytoprotection (16–19). GT-094 is coined an NO chimera because it incorporates ancillary pharmacophores, a disulfide and an NSAID (thiosalicylate), both shown to be effective in colorectal cancer chemoprevention (Fig. 1). Thiosalicylates have been reported as NSAIDs (20, 21), and a thiosalicylate derivative has been explored in colon cancer therapy (22). Disulfides, notably diallyl disulfide (DADS), have been studied as garlic-derived chemopreventive agents that reduce ACF formation in the rat/azoxymethane model (23, 24). Reports suggest that the disulfide group of DADS is responsible for the antiproliferative activity (16). DADS was reported to induce a G₂-M phase cell cycle block in HCT-15 cells (25) and to be anti-antiproliferative in Caco-2 cells (24, 26). The antiproliferative colorectal cancer chemopreventive agent, butyrate, was also reported to induce a G₂-M phase block in Caco-2 cells (at 5 mmol/L; refs. 24, 27).

Epidemiologic data suggest an inverse relationship between colorectal cancer risk and regular use of NSAIDs. Clinical trials with NSAIDs also showed that NSAID treatment caused regression of pre-existing colon adenomas in patients with familial adenomatous polyposis (FAP). However, even the use of low-dose ASA can damage the gastric mucosa. NSAID gastrotoxicity seems to be closely related to the inhibition of COX and PG biosynthesis. NO has cytoprotective properties in the stomach and in other organs, exhibiting actions in the gastrointestinal tract similar to those of PGs, such as the stimulation of mucus secretion and maintenance of mucosal blood flow. NO solutions and GTN were reported to be able to substitute for PG action in the gastrointestinal tract to counterbalance pharmacologic COX inhibition (28), providing the impetus for development of NO-NSAIDs; hybrid nitrates that conjugate a nitrate group to an NSAID via a labile acyl linker (4–6, 29, 30).

There is an extensive literature on the in vitro activity of NO-NSAIDs and NO-ASA. Importantly, in cell culture, NO-NSAIDs are reported to be potent COX-independent, antiproliferative agents in contrast to the parent NSAIDs (6, 31, 32). Moreover, the combination of ASA with very high concentrations of true NO donors (0.7 mmol/L) was only weakly antiproliferative (33), supporting the argument that antiproliferative activity is a property of the benzyl nitrate pharmacophore. Interestingly, two recent studies reported that the activity of NCX 4016 resulted at least in part from the oxidant properties of the compound (34, 35).

The role of inflammation in colorectal cancer is well accepted: iNOS is invoked as a component of inflammatory carcinogenesis. The use in colorectal cancer of drugs that are traditionally regarded as NO donors seems counterintuitive, but the correlation between NO, NOS, and colorectal cancer is not so straightforward. Levels of iNOS and eNOS are reported to be elevated in carcinoma tissue in response to azoxymethane, and similar observations on iNOS in human colorectal cancer have been made (36), but increased eNOS has also been correlated with improved survival in colorectal cancer (37). Inhibitors of iNOS have been observed to reduce ACF levels in the rat/azoxymethane model (38), but in one report in this model, a nonspecific NOS inhibitor increased ACF levels (39). In the Apc(min/+) FAP model, iNOS^−/− knock-outs had more intestinal adenomas (40). Ornithine decarboxylase activity is raised in colorectal cancer and is reduced by NO, providing a possible rationale for some of these observations (39–41). The evidence for a causative role for NO from NOS in colorectal cancer is problematic, but the evidence supports iNOS as a biomarker for colorectal cancer. Furthermore, agents that lower iNOS in RAW 264.7 cells have been shown to reduce ACFs in the rat/azoxymethane model (1), supporting the cellular iNOS level as a biomarker in drug discovery.

The progression of NCX 4016 into clinical trials for colorectal cancer is supported by rodent models. The first study on NCX 4016 in a modified rat/azoxymethane model used as an intrarectal dose of trinitrobenzene-sulfonic acid normally used in models of inflammatory colitis, and the drug was given concurrently with carcinogen (4). Drug treatment (p.o. ASA, 10 mg/kg/day, or NCX 4016, 18.5 mg/kg/day) yielded a 65% reduction of ACFs by ASA alone and an 85% reduction by NCX 4016. In a second animal study, NCX 4016 and NCX 4040, a regioisomer of NCX 4016 with more potent cytotoxic activity were studied in the Apc(min/+) mouse model of FAP, in which tumors develop spontaneously in the small intestine (42): delivered intrarectally at high dose (100 mg/kg/day; 21 days), a reduction of 60% in tumor incidence was observed for NCX 4040, but the effect of NCX 4016 was not significant. Tissues from these NO-ASA animal studies showed no evidence for inhibition of COX or PG synthesis and no evidence for an antiproliferative effect. A very recent rat/azoxymethane study reported significant reductions in tumor incidence and multiplicity after 46 weeks of treatment with high-dose NCX 4016 (14).

The actions of GT-094 in vitro and in vivo are not identical to those reported for NO-ASA. The G₂-M phase cell cycle block observed for GT-094 has not been reported for NO-ASA and may result from the disulfide pharmacophore of GT-094 because DADS was reported to show a similar antiproliferative profile. The antiproliferative activity of DADS in Caco-2 and HCT-29 cells has been ascribed to the modulation of histone acetylation and p21^waf1/cip1 expression (26). NO-ASAs are antiproliferative but, in addition and in contrast to GT-094, have been reported to induce apoptosis and PARP cleavage in colon cancer cells (6, 31, 33, 34). However, the differences in the properties of the NO-ASA isomers should be noted: NCX 4040 and NCX 4060, in contrast to NCX 4016, were significantly cytotoxic: in HT-29 cells, NCX 4016 gave an antiproliferative IC₅₀ of 200 μmol/L and only 5% apoptosis, whereas under similar conditions, the isomers were 100- to 200-fold more potent and induced 100% apoptotic and atypical cell morphology (33, 34). GT-094 was seen to be antiproliferative in rat and in human colon cancer cells, but showed evidence neither for inducing apoptosis in HT-29 cells nor for PARP cleavage in vivo. The reported differences between GT-094, NCX 4016, and NCX 4040 emphasize that the nitrate structure influences activity, one rationale being the well-known capacity of NO to exert both pro- and antiapoptotic actions dependent on NO flux, cell type, and other factors (43). Induction of apoptosis in neoplastic cells in vivo is seen as a contributor to chemoprevention, although not a definitive requirement. Further studies are needed to compare the activity of chemopreventive nitrates toward apoptosis and differentiation in vivo.

In the present study, GT-094 is shown to act as an antiproliferative agent that reduces ACF in the rat/azoxymethane model and reduces iNOS expression in vitro and in vivo. The strongest evidence for NO-ASA actions on iNOS was reported in the recent rat/azoxymethane study delivering high-dose NCX 4016 in feed, where both antiproliferation and iNOS inhibition were observed (14), in contrast to two previous NO-ASA animal studies that did not show such effects. There is a known sensitivity to dosage in rat/azoxymethane studies, and in this respect, the observation in the present study that a single-dose combination of ASA with ISDN did not significantly reduce ACF cannot be viewed as definitive; however, it is noted that in vitro, the combination therapy was also not efficacious.

In the newly reported rat/azoxymethane work on NCX 4016, drug was delivered in feed at 3,000 ppm (14). The initial azoxymethane/ACF study on NCX 4016 used a lower dose, and in our work, an approximately equimolar dosage of GT-094 in feed was chosen (233 ppm). GT-094 gave a significant reduction in ACF, but at this dose after 28 weeks, data on tumor reduction did not reach significance relative to azoxymethane-treated control (n = 10). At the 10-fold higher dose in the NCX 4016 study, a significant reduction in both the incidence and multiplicity of noninvasive colon adenocarcinomas was reported after 46 weeks (n = 36). NCX 4016 was also antiproliferative, reducing colonocyte PCNA expression by 14%, and was anti-inflammatory, reducing iNOS activity by 47%, but not having any significant effect on iNOS expression. In comparison and in contrast, GT-094 reduced azoxymethane-induced colon crypt proliferation by 30% to 69% and iNOS expression in the rat colon by 33% to 67%. NCX 4016 reduced azoxymethane-induced COX activity in the colon, which was not measured for GT-094; and GT-094 reduced azoxymethane-induced PARP cleavage and increased p27 expression in the distal colon, which was not reported for NCX 4016. The nitrates, NO-ASA and GT-094, thus do not show identical activity, but do manifest a variety of beneficial actions, both in vitro and more importantly in vivo, that support the further exploration and optimization of nitrates for colorectal cancer chemoprevention.

Daniel D. Lantvit is thanked for technical assistance. This project was done in part using compound(s) provided by the National Cancer Institute's Chemical Carcinogen Reference Standards Repository operated under contract by Midwest Research Institute, no. N02-CB-07008.