Non-Steroidal Anti-Inflammatory Drugs, DNA Repair and Cancer

Colorectal cancer is the third most common cancer in women and fourth in men, with respect to incidence, and 529,000 deaths occurred worldwide in 2002 (Parkin et al., 2005). Approximately 2 ‐ 5% of cases of colorectal cancer are due to a genetic predisposition of which the most common is hereditary non‐polyposis colorectal cancer (HNPCC). HNPCC is an autosomal dominant disorder with high penetrance and exhibits allelic and locus heterogeneity (Aarnio et al., 1995; de la Chapelle, 2004; Dunlop et al., 1997). In HNPCC there are heterozygous germline mutations in the DNA mismatch repair (MMR) genes MutS homologue 2 (MSH2), MutL homologue 1 (MLH1), MutS homologue 6 (MSH6), post‐meiotic segregation increased 2 (PMS2) and post‐meiotic segregation increased 1 (PMS1) (Bocker et al., 1999; Buermeyer et al., 1999; Jiricny, 1998; Jiricny & Marra, 2003; Lucci-Cordisco et al., 2003; Mitchell et al., 2002; Narayan & Roy, 2003; Nicolaides et al., 1998; Plaschke et al., 2004; Zabkiewicz & Clarke, 2004). Germline mutations in hMLH1 and hMSH2 are the most common with abnormalities in these genes found in more than 90% of HNPCC mutation carriers (Abdel-Rahman et al., 2006; de la Chapelle, 2004; Hampel et al., 2005; Lagerstedt Robinson et al., 2007). The phenomenon of transmission of an epimutation in hMLH1 has also been reported (Hitchins et al., 2007). The DNA MMR system plays an essential role in identifying and correcting any replication errors and any additional errors which arise through physical or chemical damage. These errors may be base‐base mismatches, short insertions/deletions and heteroduplexes, which can occur during DNA replication and recombination (Jiricny, 1998; Jiricny & Marra, 2003). The DNA MMR system therefore maintains genomic integrity and stability and in essence provides a tumour suppressor function. Deficiencies in DNA MMR lead to the accumulation of mutations in repetitive nucleotide regions, a phenomenon termed microsatellite instability (MSI) (Parsons et al., 1993; Parsons et al., 1995; Thibodeau et al., 1993; Thibodeau et al., 1998). Microsatellites are classically defined as simple tandem nucleotide sequence repeats of 1 – 6 base pairs in the genome (Hancock, 1999). Changes in the number of the repeat units due to defective DNA MMR are potentially cancer causing (Riccio et al., 1999; Yamamoto et al., 1998). The MSI phenotype or replication error positive (RER+) phenotype can be considered as an almost canonical feature of DNA MMR deficiency (Kinzler & Vogelstein, 1996; Parsons et al., 1993). This MSI phenotype is observed in approximately 15% of all human colorectal cancer, gastric and endometrial carcinomas (Lothe et al., 1993; Seruca et al., 1995; Shibata, 1999; Umar et al., 1994). Somatic inactivation of DNA MMR largely


Introduction
Colorectal cancer is the third most common cancer in women and fourth in men, with respect to incidence, and 529,000 deaths occurred worldwide in 2002 (Parkin et al., 2005).Approximately 2 -5% of cases of colorectal cancer are due to a genetic predisposition of which the most common is hereditary non-polyposis colorectal cancer (HNPCC).HNPCC is an autosomal dominant disorder with high penetrance and exhibits allelic and locus heterogeneity (Aarnio et al., 1995;de la Chapelle, 2004;Dunlop et al., 1997).In HNPCC there are heterozygous germline mutations in the DNA mismatch repair (MMR) genes MutS homologue 2 (MSH2), MutL homologue 1 (MLH1), MutS homologue 6 (MSH6), post-meiotic segregation increased 2 (PMS2) and post-meiotic segregation increased 1 (PMS1) (Bocker et al., 1999;Buermeyer et al., 1999;Jiricny, 1998;Jiricny & Marra, 2003;Lucci-Cordisco et al., 2003;Mitchell et al., 2002;Narayan & Roy, 2003;Nicolaides et al., 1998;Plaschke et al., 2004;Zabkiewicz & Clarke, 2004).Germline mutations in hMLH1 and hMSH2 are the most common with abnormalities in these genes found in more than 90% of HNPCC mutation carriers (Abdel-Rahman et al., 2006;de la Chapelle, 2004;Hampel et al., 2005;Lagerstedt Robinson et al., 2007).The phenomenon of transmission of an epimutation in hMLH1 has also been reported (Hitchins et al., 2007).The DNA MMR system plays an essential role in identifying and correcting any replication errors and any additional errors which arise through physical or chemical damage.These errors may be base-base mismatches, short insertions/deletions and heteroduplexes, which can occur during DNA replication and recombination (Jiricny, 1998;Jiricny & Marra, 2003).The DNA MMR system therefore maintains genomic integrity and stability and in essence provides a tumour suppressor function.Deficiencies in DNA MMR lead to the accumulation of mutations in repetitive nucleotide regions, a phenomenon termed microsatellite instability (MSI) (Parsons et al., 1993;Parsons et al., 1995;Thibodeau et al., 1993;Thibodeau et al., 1998).Microsatellites are classically defined as simple tandem nucleotide sequence repeats of 1 -6 base pairs in the genome (Hancock, 1999).Changes in the number of the repeat units due to defective DNA MMR are potentially cancer causing (Riccio et al., 1999;Yamamoto et al., 1998).The MSI phenotype or replication error positive (RER+) phenotype can be considered as an almost canonical feature of DNA MMR deficiency (Kinzler & Vogelstein, 1996;Parsons et al., 1993).This MSI phenotype is observed in approximately 15% of all human colorectal cancer, gastric and endometrial carcinomas (Lothe et al., 1993;Seruca et al., 1995;Shibata, 1999;Umar et al., 1994).Somatic inactivation of DNA MMR largely arises as a consequence of epigenetic silencing of hMLH1 (through hypermethylation of promoter CpG islands) rather than via classic mutational inactivation (Herman et al., 1998;Jacinto & Esteller, 2007;Jones & Laird, 1999;Peltomaki, 2001;H. Yamamoto et al., 1998).Genes particularly prone to MSI include Bax, TGF-β receptor II, hMSH3 and hMSH6 (Yamamoto et al., 1998); other susceptible genes include the DNA glycosylase MBD4 (Bader et al., 2000;Bader et al., 1999;Riccio et al., 1999) and the epidermal growth factor receptor (EGFR) (Woerner et al., 2010).Additionally, as a consequence of MSI, inactivation of proteins in the Wnt signalling pathway (eg TCF-4) has been reported (Shimizu et al., 2002).Clinically defined MSI is where at least two of the loci tested in a panel out of five exhibit instability (Boland et al., 1998).Mutation frequencies in cells defective in DNA MMR can be increased 100-1000 fold (Parsons et al., 1993;Shibata, 1999).The protective role of the DNA MMR system in suppressing the mutator phenotype in a range of common cancers is thus well established.The hypothesis that aspirin may potentially modulate this pathway to prevent carcinogenesis (Goel et al., 2003) is of central interest, and has so far received relatively little attention and merits further investigation.The aim of this chapter will be to review the evidence that non-steroidal anti-inflammatory drugs (NSAIDS), including aspirin, celecoxib, sulindac and so on, affect DNA repair mechanisms and pathways and to further examine the consequences of this in relation to cancer development and progression.As NSAIDs are considered to be one of the most widely used over-the-counter drugs, we will also discuss the potential effect of NSAID use on cancer treatment.Aspirin and other NSAIDs may have the capacity to perturb DNA repair pathways and this may have important implications for the patient response to chemotherapeutic agents.It is also worth noting that inflammation -the 'seventh hallmark' of cancer (Colotta et al., 2009) -can possibly repress (by epigenetic mechanisms) DNA mismatch repair.

Cancer and NSAIDs
From evidence adduced from epidemiological studies and clinical trials, it has been proposed that regular ingestion of aspirin and other non-steroidal anti-inflammatory drugs (NSAIDs) can promote colorectal tumour regression and reduce the relative risk of developing colorectal cancer (CRC) in the general population and in genetically susceptible individuals (for example see; (Baron et al., 2003;Chan et al., 2009;Cuzick et al., 2009;Giovannucci, 1999;Imperiale, 2003;Logan et al., 1993;Paganini-Hill, 1993;Sandler et al., 2003;Thun et al., 2002).NSAID use is also associated with a reduced risk of oesophageal adenocarcinoma particularly in patients with high risk molecular abnormalities, for example, with 17p LOH, 9p LOH, and DNA content abnormalities (Galipeau et al., 2007).Recent meta-analyses of randomised clinical trials have strengthened the contention that aspirin has protective effects against CRC (Din et al., 2010) and non-CRC related adenocarcinomas, including oesophageal and lung cancer (Rothwell et al., 2011).Although such studies have now provided substantial evidence that regular use of aspirin based medication can reduce the risk of colorectal cancer (Bosetti et al., 2002;Muscat et al., 1994;Thun et al., 1991) the molecular basis for the protective effect of aspirin vis-à-vis CRC and other cancers is rather controversial.A substantial number of theories are now in circulation.Whilst much of the focus in the recent past has understandably been on the intrinsic anti-inflammatory nature of the compounds in use, see for example, (Giovannucci, 1999;Keller & Giardiello, 2003), there are a number of intriguing findings which suggest that NSAIDs can impact upon genetic stability and it is these aspects which we wish to highlight in this chapter.However, the optimism that arises from the findings that NSAIDs may offer protection against cancer should be tempered: NSAID use can result in serious side effects (gastrointestinal disturbances and cardiovascular events) in susceptible individuals (Cuzick et al., 2009).This has lead to considerable debate amongst clinicians over recent years as to whether NSAIDs should be prescribed as chemopreventative agents, in particular, to those individuals at high risk of developing colorectal cancer, such as HNPCC or Familial Adenomatous Polyposis (FAP) patients.Animal studies confirm that NSAIDs can protect against the development of colorectal neoplasia (Corpet & Pierre, 2003, 2005).For example, aspirin has been shown to suppress spontaneous intestinal adenoma formation and reduce incidence and volume of colon tumour induced by the carcinogen 1,2-dimethylhydrazine in rat models (Barnes & Lee, 1999).Continuous administration of a clinically relevant aspirin dosage was crucial in these studies in comparison to other studies where aspirin was administered at the start of carcinogenesis or one week after carcinogen exposure (Craven & DeRubertis, 1992).Taken together with findings from epidemiological studies in humans, there is thus the suggestion that long term, continuous usage of aspirin is required to gain any beneficial chemopreventative effects.Several studies carried out in the 1990s demonstrated inhibition of carcinogen induced tumour development by NSAIDs including aspirin and sulindac in rats (Rao et al., 1995;Reddy et al., 1993) and also in a murine model of FAP (Barnes & Lee, 1998;Chiu et al., 1997;Jacoby et al., 1996;Mahmoud et al., 1998;Oshima et al., 1996).Familial adenomatous polyposis (FAP) is a colorectal cancer syndrome inherited in humans in an autosomal dominant manner caused by an absence of a functional caretaker APC protein (Narayan & Roy, 2003).Lifetime administration of aspirin to a mouse model with germline defects in both the APC and Msh2 gens (APC Min/+ , Msh2 -/-) suppresses intestinal and mammary neoplasia formation (Sansom et al., 2001).Epidemiological studies have identified environmental and dietary factors which alter the risk of developing colorectal cancer.Protective dietary factors include NSAIDs, fruit, vegetables, and folic acid and possibly calcium, whilst red and processed meat ingestion, alcohol use and obesity are perceived to increase risk (Forte et al., 2008;Key, 2011;La Vecchia et al., 2001;Ryan-Harshman & Aldoori, 2007;Scheier, 2001;Serrano et al., 2004).An assessment of chemopreventative measures, such as NSAID and micronutrient intake, for the general population and individuals with an increased risk for colorectal cancer based on family history, has recently been published (Cooper et al., 2010).

Inflammation and cancer
Debate has arisen with regards to the molecular mechanism of action of aspirin and other NSAIDs in reducing the incidence of certain cancers.Based on the anti-inflammatory effects of these agents, one obvious explanation is that inflammation can drive cancer development.Indeed, inflammation is becoming increasingly recognised as being critical to cancer formation, and building on the framework proposed by Hannahan and Weinberg (Hanahan & Weinberg, 2000), it has been proposed that inflammation should be considered the seventh hallmark of genetic instability (Colotta et al., 2009).It has been estimated that one in four cancers are linked to infection and chronic inflammation (Hussain & Harris, 2007).In an inflammatory microenvironment mutation frequency is increased (Bielas et al., 2006).There is strong evidence for an increased risk of cancer in individuals with chronic inflammatory states (inflammatory bowel disease, gastroesophageal reflux disease, asthma) with or without attendant bacterial or viral infection (eg Helicobacter pylori, hepatitis) (Grivennikov & Karin, 2010;Schetter et al., 2010;Xie & Itzkowitz, 2008).The molecular drivers of cancer formation resulting from the interaction of pre-cancerous cells with activated immune cells and the surrounding stroma are complex.The host response to infection, injury and wound repair through production of reactive oxygen (ROS) and nitrogen oxide (RNOS) species, and pro-inflammatory cytokines (eg TNF-α, IL-1, IL-6) and chemokines (eg IL-8) (Wang et al., 2009) is of central importance.Such 'micro-cytokine' storms can stimulate activation of transcriptions factors (eg Nuclear Factor-κB, AP-1, STAT3) fundamentally altering the expression profile of a cell and promoting cell survival, proliferation, angiogenesis, motility and invasion.It is clear that for individual transcription factors their activation can be context (eg tissue) dependent and they may not always promote tumour cell formation, and may indeed exhibit tumour suppressor activity (eg NF-κB in skin cancer) (Chaturvedi et al., 2011;Ditsworth & Zong, 2004).Microsatellite instability has been observed in non-neoplastic tissue in patients with chronic inflammatory conditions prior to the presence of dysplastic tissue, indicating that defects in DNA MMR can be an early event in inflammation-associated cancers (Brentnall et al., 1995;Park et al., 1998).There are tantalising findings which hint at the molecular basis for this MSI.Oxidative stress, in the guise of H 2 O 2 , increases frameshift mutations (Gasche et al., 2001), and can inactivate the DNA MMR system (Chang et al., 2002).Moreover, in a p53 and p21 dependent fashion, activated neutrophils induced replication errors and a G2/M arrest in colonic epithelial cells (Campregher et al., 2008).In inflammatory bowel disease neoplasia, hypermethylation of the hMLH1 gene and reduced hMLH1 protein expression occurs frequently (Fleisher et al., 2000).In an animal model of colorectal cancer, inflammation and hypoxia were found to epigenetically silence hMLH1 expression: down-regulated expression of this DNA mismatch repair gene occurred as a consequence of decreased acetylation, which was reversible when the animals were treated with the HDAC inhibitor suberoylanilide hydroxamic acid (SAHA) (Edwards et al., 2009).Decreased expression of hMLH1 can alter hPMS2 stability and consequently, genetic integrity.Moreover, hMLH1 and hPMS2 can have a role in activating cell-cycle checkpoints and promoting apoptosis (Cejka et al., 2003;Davis et al., 1998;Ding et al., 2009;McDaid et al., 2009;Sansom et al., 2003;Yanamadala & Ljungman, 2003;Zhang et al., 1999).Thus, a functional DNA mismatch repair system prohibits expansion of cells containing DNA damage (Carethers et al., 1996;Papouli et al., 2004).Dysregulated hMLH1/hPMS2 expression could conceivably impact on the survival of cells containing damaged DNA and promote cancer progression.hMLH1 expression has been reported to be altered by tobacco usage and inflammatory state in the epithelium of the oral mucosa (Fernandes et al., 2007).Increased tissue specific hMLH1 hypermethylation has been observed in the progression of oesophageal cancer (Vasavi et al., 2006); the authors also reported that patients with gastroesophageal reflux disease (GERD) exhibited a very significant degree of hMLH1 hypermethylation prompting the suggestion that reflux can promote hypermethylation.The relationship between acid reflux and inflammation in GERD has been recently reviewed (Orlando, 2010).The phenomenon of elevated microsatellite instability at selected nucleotide repeats (EMAST) is also seen in 60% of sporadic colon cancers, is more common in individuals of African-American origin and has been linked to reduced expression of the hMSH3 protein.Moreover, EMAST is also more prevalent in rectal cancer with immune cell infiltration (Devaraj et al., 2010;Lee et al., 2010).As MSI in CRC can result in products with potentially increased immunogenicity, it is possible that this further stimulates inflammation (Banerjea et al., 2004).Tumour infiltrating lymphocytes in CRC with MSI are activated and cytotoxic (Phillips et al., 2004).In a model of experimental colon carcinogenesis in rats, long-term, low-dose administration of aspirin significantly reduced cytokine and matrix metalloproteinase release (Bousserouel et al., 2010).Based on the above findings, it would appear to be reasonable to suppose that research into the epigenetic modifying effects of NSAIDs, and inflammation itself (Maekawa & Watanabe, 2007), particularly with respect to alterations in DNA repair protein expression and MSI, should be a focus in the future.Oxidative stress occurs as a consequence of the activation of the immune system, where oxidative bursts have a role in protecting the host against microbial invaders.Reactive oxide and nitrogen oxide species (ROS and RNOS) thus produced can ultimately cause DNA damage such as abasic sites, oxidised bases, DNA-intrastrand adducts, strand breaks, as well as RNA alkylation, and protein damage (Hussain et al., 2003).Reaction with lipids produces extremely reactive peroxidation products, including malondialdehyde (MDA) and trans-4-hydroxynonenal (4-HNE).4-HNE can form etheno adducts in DNA and has been found to preferentially form adducts in codon 249 of the human p53 gene (Federico et al., 2007;Hu et al., 2002).Mutations in the tumour suppressor p53 are found in inflamed tissue in ulcerative colitis (UC) patients (Hussain et al., 2000).Accelerated telomere shortening, and DNA damage -as assessed by analysis of phosphorylated histone H2AX (γH2AX; a measure of DSBs) occurs in the bowel of UC patients (Risques et al., 2008).ROS and RNOS -whilst generally perceived as having a negative impact on cell function -do have important roles as secondary messengers (Valko et al., 2006).However, if the cellular defence mechanisms -anti-oxidant enzymes such as manganese superoxide dismutase and glutathione peroxidise, detoxification, and DNA repair systems -are overwhelmed, these reactive species are potentially mutagenic (Ferguson, 2010).The bystander effects from released cytokines and reactive signalling species may also occur some distance from the initial trauma.Redon et al. have shown that increased levels of DNA damage (measuring γH2AX levels) can occur systemically in mice implanted with a non-metastatic tumour (Redon et al., 2010).Animal model studies confirm the role of RNOS and the inflammatory process in contributing to cancer development (Hussain et al., 2003;Itzkowitz & Yio, 2004).The base excision repair (BER) system is critically important for dealing with oxidative damage to DNA, such as removal of 8-oxo-G lesions (David et al., 2007;Lindahl & Wood, 1999;McCullough et al., 1999;Wood et al., 2001).Increased expression of the BER proteins AAG (a 3-methyladenine DNA glycosylase) and APE1 (apurinic endonuclease1; Ref-1) is seen in inflamed tissue from UC patients.Paradoxically a positive correlation of MSI was noted with overexpression of AAG, and for MSI-high tissues with increased APE1 expression.In model systems this adaptive response was confirmed to positively correlate with an in increase in MSI in human cells and frameshift mutations in S. cerevisiae (Hofseth et al., 2003).Dysregulated expression of BER proteins can generate a mutator phenotype (Glassner et al., 1998) (as cited in Hofseth, 2003).In a mouse model of UC, expression of the Mutyh BER protein, which can recognise 8-oxoG:A mispairs and oxidised adenines was actually found to influence the inflammatory response to dextran sulphate sodium induced oxidative stress (Casorelli et al. 2010).The biomediator signalling molecule nitric oxide (NO .) formed during inflammation has the capacity to inhibit in vitro and in vivo the formamidopyrimidine-DNA glycosylase, which can recognise abasic sites and 8-oxoguanine lesions (Wink & Laval, 1994).Thus, oxidative stress can compromise genetic stability.RNOS, and inflammatory cytokines and chemokines can activate the transcription factors activator protein-1 (AP-1), HIF-1 and NF-κB, the latter resulting in transcription of target genes including COX-2 and TNF-α (for example, see Olson & van der Vliet, 2011;Valko et al., 2006).DNA double strand breaks can also activate NF-κB (Rakoff-Nahoum, 2006), thus potentially creating a feedback loop where the inflammatory response is perpetuated.Persistent DNA damage can also trigger secretion of cytokines such as IL-6 (Rodier et al., 2009).There is evidence that NF-κB and cyclooxygenase activation and HIF-1 signalling are interlinked (Jung et al., 2003;Qiao et al., 2010).Hypoxia can induce MSI (Kondo et al., 2001): in human sporadic colon cancers HIF-1α over-expression is associated with loss of hMSH2 expression, and in vitro experiments confirm that in a p53 dependent manner HIF-1α can repress hMSH2 and hMSH6 (Koshiji et al., 2005;To et al., 2005).Recently HIF-1α has been suggested to regulate expression of an inhibitor of apoptosis protein, Survivin (Wu et al., 2010), which is notably present in CRC (Chen et al., 2004).There is increasing evidence that microRNAs (short, non coding RNAs that regulate translation) are mediators of the inflammatory process (Schetter et al., 2010).Two micro-RNAs, miR-210 and miR-373 are up regulated in an HIF-1α dependent manner in hypoxic cells: forced expression of mir-210 reduced expression of homologous recombination factor RAD52, whilst reduced mir-373 expression was found to suppress RAD52 and RAD23B (Crosby et al., 2009).Overexpression of another micro-RNA, miR-155 has been reported in colorectal cancer, and can regulate DNA MMR protein expression (Valeri et al., 2010).Inflammation can up-regulate miR-155 expression and increase mutation frequency two to threefold in spontaneous hypoxanthine phosphoribosyltransferase gene mutation assays (Tili et al., 2011).MiR-155 expression is associated with a poor prognosis in lung cancer (Yanaihara et al., 2006).The cyclooxygenase family of enzymes (COX-1: constitutively expressed; COX-2: induced in inflammation; COX-3: splice variant of COX-1) catalyse the conversion of arachidonic acid into prostanoids (prostaglandins and thromboxanes).Cyclooxygenase expression is increased in CRC (Kutchera et al., 1996).Prostaglandins promote epidermal growth factor receptor (EGFR) transactivation, increased cell proliferation, motility, invasion and angiogenesis (eg by altering vascular endothelial growth factor expression) and inhibit apoptosis, for example, by increasing Bcl-2 expression (Ghosh et al., 2010;Pai et al., 2003;Sheng et al., 1998;Wang & DuBois, 2008).To summarise: inflammation and hypoxia can result in genetic instability and suppression of apoptosis.

Mechanistic aspects of NSAID cytotoxicity
The mechanism by which NSAIDs protect the host from colorectal cancer development has been under investigation for decades, and as a consequence a plethora of hypotheses (some of which may be competing) have been proposed to explain this phenomenon.In addition to the evidence alluded to above of epidemiological studies and animal models examining the protective effect of NSAIDs, there also exists a substantive literature reporting that NSAIDs exhibit a degree of specific toxicity in vitro to colorectal cancer cell lines.Because of the intrinsic anti-inflammatory activity of NSAIDs a significant number of researchers have focused on this aspect as a protective mechanism.For example, aspirin can acetylate the cyclo-oxygenases (COX) significantly reducing arachidonic acid metabolism and prostaglandin production, thereby reducing inflammation (Elwood et al., 2009).Expression of the inducible COX, cyclooxygenase-2 is notably elevated in colorectal malignancies and in other cancers (Ferrandez et al., 2003;Kutchera et al., 1996;Soslow et al., 2000), and this over-expression has been actively implicated in the metastasic potential of tumours (Jang et al., 2009;Tsujii et al., 1997).These effects, however, may not be restricted only to colorectal cancer: a case control study has found that use of selective and non-selective COX-2 inhibitors (celecoxib, refecoxib and aspirin and ibuprofen) has utility in the chemoprevention of lung cancer (Harris et al., 2007).NSAIDs do not only affect cyclooxygenase activity: exposure to these drug can significantly alter gene expression and thus NSAIDs can be reasonably described as having pleiotropic effects, some of which may be relatively compound dependent.Chronic NSAID use can suppress CpG island hypermethylation of tumour suppressor genes [p14(Arf), p16(INK4a), E-cadherin] in the human gastric mucosa (Tahara et al., 2009).Furthermore, NSAID toxicity may not be absolutely dependent on inhibition of COX activity: the proliferation of COX-2 negative cell lines can also be inhibited by NSAIDs (Lai et al., 2008;Richter et al., 2001) and the chemical precursor of aspirin -salicylate -which has weak anti-COX activity, can itself be anti-inflammatory (Amann & Peskar, 2002) and pro-apoptotic to CRC cells (Elder et al., 1996).Additionally, whilst the NSAID sulindac sulfide and its sulfone derivative can both inhibit the HT-29 CRC cell line growth, sulindac sulfone is "devoid of prostaglandin inhibitory activity" (Piazza et al., 1995).Additionally, aberrant crypt foci formation in a chemically induced rat model of CRC were suppressed by treatment with sulindac sulfone (Charalambous & O'Brien, 1996).Supporting these findings, a mechanism for NSAID toxicity (anti-proliferative and inducing apoptosis) -testing sulindac sulfide and piroxicamtoward a CRC line lacking cyclooxygenase activity (HCT-15) and thus independent of prostaglandin inhibition, has also been reported (Hanif et al., 1996).Sulindac metabolites (sulphide and sulfone; see Fig. 1) can inhibit the activation and expression of the EGF receptor (Pangburn et al., 2005), with the down-regulated activity mediated by lysosomal and proteasomal degradation (Pangburn et al., 2010).Rigas and others have cogently proposed that the anti-neoplastic activities of NSAIDs can be categorised as either being COX-dependent or COX-independent (Keller & Giardiello, 2003;Shiff & Rigas, 1999).This concept has been reviewed in some detail in (Ferrandez et al., 2003).Smith et al examined the effects of NS-398 (a selective COX inhibitor), indomethacin (a non-selective COX inhibitor) and aspirin on the HT29Fu, HCA-7, SW480 and HCT116 CRC cell lines with respect to effects on cell proliferation, cell cycle arrest, apoptosis induction, β-catenin and COX-1 and COX-2 protein production.They concluded that NSAIDs act via COX-dependent and COXindependent mechanisms (Smith et al., 2000).Indeed, acetylsalicylic acid regulates MMP-2 activity and inhibits colorectal invasion of B16F0 melanoma cells (Tsai et al., 2009).Intriguingly, although medium conditioned by cultured colorectal cancer cell lines is capable of inducing endothelial cell (EC) tube formation (as a model for angiogenesis); aspirin and salicylate (both at 1mM) can reduce the ability of the conditioned medium from treated DLD-1, HT-29 and HCT116 CRC cell lines to promote EC tube formation (Shtivelband et al., 2003).Aspirin is principally metabolised to salicylate in vivo (Law et al., 2000;Paterson & Lawrence, 2001), with plasma salicylate concentrations of 0.95-1.9mM achievable in patients receiving aspirin as an anti-inflammatory agent (Amann & Peskar, 2002;Urios et al., 2007;Yin et al., 1998).The effect of significantly higher concentrations than that achievable physiologically has been used by a number of investigators in a number of in vitro studies and one must interpret cautiously data produced from these analyses.To facilitate interpretation, we have incorporated the concentrations utilised in the relevant publications.We should also point out that whilst vegetables and fruits have been considered to be a natural source of salicylate, where it functions as a plant signalling molecule (Paterson & Lawrence, 2001;Schenk et al., 2000, and refs therein), it has been reported that SA may be an endogenous compound, with SA found in the blood of carnivorous animals, for example in the burrowing owl (Paterson et al., 1998;Paterson et al 2008).

COX independent effects of NSAIDs
A substantial number of hypotheses regarding NSAID cytotoxicity not directly involving inhibition of COX activity exist (reviewed in: Goke, 2002;Watson, 2006).Much of the data is obtained from examining the effect of NSAIDs on cultured cell lines.For example, salicylate can cause cell cycle arrest: in a range of adenoma and carcinoma derived cell lines, incubation with 5 mM resulted in a decrease of cells in the S-phase with cells accumulating in the G o -G 1 phase (Elder et al., 1996).Activating mutations stabilizing the Wnt pathway and β-catenin are common in CRC (Behrens, 2005;Bienz & Clevers, 2000).In four CRC cell lines (SW948, SW480, HCT116, LoVo), aspirin (5 mM) and indomethacin (400 μM) downregulated cyclin D1 expression, and the authors proposed this was a consequence of reduced transcriptional activity of the β-catenin/TCF complex (Dihlmann et al., 2001).Bos et al using reporter assays also showed that aspirin (up to 5 mM) can down-regulate APC-β-catenin-TCF4 signalling in the SW480 CRC cell line and increase the phosphorylation of protein phosphatase 2A (PP2A), resulting in decreased PP2A enzymatic activity.This alteration is stated to be essential for the observed effect on the Wnt/β-catenin pathway activity (Bos et al., 2006).Mutations in the APC or β-catenin genes may also result in increased peroxisome proliferator-activated receptor δ (PPAR δ) activity; indeed, sulindac sulfide and indomethacin have been reported to suppress (by direct inhibition) in a dose-dependent fashion, PPARδ activity (He et al., 1999).However, it should be noted that regulation of PPARδ by the APC/ β -catenin/TCF4 pathway and the effect of NSAIDs on PPAR function is a field not without some controversy; for example, Foreman et al recently reported not finding decreased PPARβ/δ expression following NSAID treatment in a number of human CRC cell lines (Foreman et al., 2009).

NSAIDS, signalling pathways and apoptosis
Other transcriptional pathways affected by NSAID use include NF-κB signalling, which is critically involved in regulating immunity, inflammation and apoptosis and in cancer development (Nakano, 2004;Olivier et al., 2006;Staudt, 2010).Aspirin and sodium salicylate can antagonise the NF-κB pathway (Kopp & Ghosh, 1994), via inhibition -by direct, but not covalent binding-of IκB kinase-B (IKK β kinase) (Yin et al., 1998).Yamamoto et al extended this hypothesis by reporting that sulindac, sulindac sulfide and sulindac sulfone (but not indomethacin) also inhibits the NF-κB pathway by decreasing IKK β kinase activity (Yamamoto et al., 1999).In HT-29 CRC cells, nitric oxide (NO)-donating aspirin inhibited NF-κB transcription more potently than aspirin as assessed by electrophoretic mobility shift assays (Williams et al., 2003).Zerbini et al reported that NF-κB can mediate the repression (assessed by RT-PCR analysis) of the growth arrest and DNA damage inducible (GADD) 45α and GADD45γ genes and that this repression is both necessary and sufficient for cell survival, and that GADD45α and γ activity can contribute towards apoptosis of prostate cancer cells (assessed by transfection studies in DU145 and PC-3) (Zerbini et al., 2004).The melanoma differentiation associated gene 8 (mda-7/IL-24) was induced by NSAIDs and affected growth arrest and apoptosis in vitro, in a manner dependent on GADD45α and γ expression affecting the p38/JNK pathway (Sarkar et al., 2002;Zerbini et al., 2006).However, it has been suggested that recombinant IL-24 lacks apoptosis inducing properties to melanoma cells (Kreis et al., 2007).In a recent review, Siafakas and Richardson have suggested that the GADD family of proteins can be considered to be molecular targets for anti-tumour agents; indeed, commonly used NSAIDs can up-regulate GADD45α expression (Rosemary Siafakas & Richardson, 2009).GADD45α overexpression in NIH 3T3 cells has been shown to promote global DNA demethylation and loss of expression has been shown to induce DNA hypermethylation (including in hMLH1).Such findings led Barreto et al to conclude that GADD45α can relieve epigenetic silencing by promoting DNA repair (Barreto et al., 2007).It is noteworthy that GADD45γ can be inactivated by epigenetic mechanisms in multiple tumours (Ying et al., 2005).In a review of the role of the NF-κB pathway in inflammation and cancer, the authors suggested that, '...constitutive NF-κB activation is likely involved in the enhanced growth properties seen in a variety of cancers' (Yamamoto & Gaynor, 2001).Specifically targeting the NF-κB pathway may thus have utility in cancer treatment (Olivier et al., 2006): for example, attempts are ongoing to identify clinically useful and novel IKK-β inhibtors (for example see (Lauria et al., 2010)), and inhibition of NF-κB (by siRNA) can enhance the chemosensitivity of HCT116 CRC cells to the DNA topoisomerase inhibitor, irinotecan (Guo et al., 2004).In marked contrast however, Stark et al, have suggested that aspirin can activate NF-κB signalling in vitro (SW480, HRT-18, HCT116, CT26 cell lines) and has the capacity to induce apoptosis in in vivo (xenografts and APC Min+/-) models of CRC (Stark et al., 2001;Stark et al., 2007).Moreover, these effects are cell type specific to aspirin (for CRC cells): the induction of apoptosis was independent of COX-2 expression, APC and β-catenin mutation status, and DNA mismatch repair proficiency (Din et al., 2004).A number of other hypotheses have been invoked for the observed cytotoxicity of NSAIDs to CRC cell lines.Cytosolic phospholipase A 2 (cPLA2) expression is decreased in NIH 3T3 cells treated with either aspirin or sulindac (Yuan et al., 2000) and in CRC cell lines (SW480, Colo320 and HT-29) when treated with low mM (2.5-10) concentrations of aspirin (Yu et al., 2003).This observation is intriguing given that cPLA 2 expression has been proposed to participate in intestinal tumorigenesis (eg (Lim et al., 2010) and refs therein).Law et al have indicated that (high mM concentrations of) salicylate (in Balb/MK cells) can inhibit the activity and phosphorylation of the mitogen activated protein kinase, p70 s6k independent of p38 MAPK, with a concomitant reduction in DNA synthesis, cell proliferation and in expression of proliferation associated proteins such as c-MYC, cyclin D1, and PCNA and led the authors to conclude that salicylate may act via the mTOR pathway (Law et al., 2000).In contrast, it has been suggested that salicylate-induced apoptosis in HCT116 CRC cells occurs through activation of p38 MAPK and p53; however, the concentration tested was -in our opinion -high (10mM), and thus casts doubt on the utility of the findings (Lee et al., 2003).NO-donating aspirin has also been reported to activate p38 and JNK MAP kinase pathways in HT-29 colorectal cancer cells (Hundley & Rigas, 2006).Using subtractive hybridization, Baek et al identified an increase in the expression of the NSAID activated gene (NAG-1) -a member of the TGF-β superfamily, with pro-apoptotic properties -in indomethacin treated HCT116 cells.A range of NSAIDs including sulindac sulfide and aspirin also increased NAG-1 expression (Baek et al., 2001).NCX-4040 (para-NO-aspirin), an NO donating aspirin derivative can also induce NAG-1 expression and it was confirmed that NAG-1 has a proapoptotic role (Tesei et al., 2008, and refs therein).Another nitro-derivative of aspirin, NCX-4016 inhibited EGFR and STAT3 signalling in cisplatin-resistant human ovarian cancer cells (Selvendiran et al., 2008).NSAIDs can also up-regulate 15-lipoxygenase-1 (15-LOX-1) in CRC cells: NS-398 and sulindac sulfone both induced expression in the DLD-1 (COX-1 and COX-2 negative) cell line; significantly, inhibiting 15-LOX-1 blocked the induction of apoptosis (Shureiqi et al., 2000).With respect to characterisation of the pathway by which NSAID toxicity occurs, reports have been rather contradictory: in vitro studies have suggested that whilst aspirin and other NSAIDs can cause cell cycle arrest and inhibit CRC proliferation, this may, (Din et al., 2004;Elder et al., 1996;Piazza et al., 1995;Yu et al., 2002;Yu et al., 2003) or may not (Shiff et al., 1996;Smith et al., 2000) occur with the induction of apoptosis, or may occur as consequence a combination of activation of both apoptotic and necrotic 'pathways' (Lai et al., 2008).Notwithstanding the apparently contradictory reports, an absence of Bax expression in CRC cells has been found to abolish the apoptotic response to NSAIDs (Zhang et al., 2000).These authors reported finding Bax mutations in indomethacin resistant cells.Bax and BCl-2 expression can be up and down-regulated respectively, in a dose-dependent fashion, (up to 10 mM) in SW480 cells incubated with aspirin (Lai et al., 2008).BCL-2 expression can be reduced by aspirin in SW480 cells (Yu et al., 2002) and can also suppress apoptosis in CRC cells (Jiang & Milner, 2003).Pretreating LNCap (human prostate) and CX-01 (colorectal carcinoma) cell lines with aspirin was found to enhance the capacity of tumour necrosis factor-related apoptosis-inducing ligand (TRAIL) to intiate apoptosis, an effect -according to the authors -related to down regulation of BCL-2 gene expression resulting from the inhibition of NF-κB (of which BCL-2 is a known target) (K.M. Kim et al., 2005).HeLa cells incubated with aspirin undergo apoptosis as assessed by cleavage of procaspase 3, PARP and PKC-δ, and annexin V staining.Moreover, the caspase inhibitor zVAD-fmk suppressed cell death, Bax was noted to translocate from the cytosol to mitochondria, and cytochrome c release from mitochondria was seen using time-lapse confocal microscopy (Zimmermann et al., 2000).Sulindac sulfide (up to 500 μM tested) can inhibit growth of NIH3T3 and SAOS cells: an inhibtion of ras mediated proliferation and transformation was noted (Herrmann et al., 1998).Hughes et al., (2003) showed that NSAID treatment of colorectal cancer cell lines caused a decrease in intracellular polyamine content and was cytotoxic.Polyamines are growth factors, and involved in protein synthesis.An increase in intracellular polyamine concentration is observed in the early stages of carcinogenesis.When polyamines were reintroduced to NSAID treated cells, apoptosis was inhibited suggesting that the polyamine pathway is affected by NSAID treatment in CRC cell, and that modulation of this pathway may explain the chemoprotective effects of NSAIDs (Hughes et al., 2003).Dikshit et al., reported that aspirin disrupted proteasome and mitochondrial function in mouse Neuro 2a cells (Dikshit et al., 2006).Hardwick et al., (2004) analysed changes to gene expression employing DNA microarray in the colorectal cancer cell line HT-29 upon aspirin treatment, and found a significant increase in Rac1 gene and protein expression in a time and concentration dependant manner.Rac1 is involved in intestinal epithelial cell differentiation (Stappenbeck & Gordon, 2000) proliferation, motility and resistance to apoptosis (Parri & Chiarugi, 2010).Spitz et al report that aspirin and salicylate can inhibit purified phosphofructokinase (PFK) in a dose-dependent manner: as cancer cells utilise glycolysis for energy production in which PFK is the rate limiting step, the authors suggest that these NSAIDs may be pro-apoptotic through disturbing the glycolytic pathway (Spitz et al., 2009).We have recently identified novel derivatives of di-aspirin (bis-carboxyphenol succinate) to have potential as anti-colorectal cancer agents (Deb et al., 2011).

NSAIDS and DNA repair
A relatively unexplored area of research is that the anti-tumour mechanism of action of aspirin may be via 'interaction' with the DNA repair systems.This is of key interest particularly in the context of hereditary and sporadic colorectal cancer where DNA repair defects can be a causative factor.It has been shown that the MSI mutator phenotype is suppressed by aspirin.For example, treatment of the colorectal cancer cell line HCT116which is DNA MMR deficient with a defect in hMLH1 -reduced the MSI phenotype and induced apoptosis (Ruschoff et al., 1998).It is speculated that aspirin exposure resulted in genetic selection against cells with MSI and that MSI unstable cells were weeded out by apoptosis (Ruschoff et al., 1998).This observation is important as MSI resulting from a genetic mutation in hMLH1, accounts for 5 -10% of colonic tumours.Nitric oxide releasing NSAIDS (NO-NSAIDS) suppressed MSI in DNA mismatch repair deficient colorectal cancer cells (McIlhatton et al., 2007), and the authors suggested that since these NO containing derivatives of aspirin are more effective than aspirin itself, they should be considered for use in chemopreventitive trials in patients at high risk of developing CRC.Recently, in a mouse model of HNPCC, McIlhatton et al., reported that aspirin and low dose NO-aspirin increased life span.However, though the intestinal tumours in aspirin treated animals showed less microsatellite instability, low dose NO-aspirin had 'minimal effect on MSI status' and high dose NO-aspirin decreased life span and increased MSI (McIlhatton et al., 2011).In addition to the phenomenon of aspirin selecting for microsatellite stability in colorectal cancer cells (Ruschoff et al., 1998), aspirin in vitro can inhibit CRC growth and increase the level of the DNA MMR proteins hMLH1, hMSH2, hMSH6 and hPMS2 in DNA MMR proficient cells (Goel et al., 2003).Increased DNA MMR levels could conceivably facilitate programmed cell death.This is a highly significant observation as defects in DNA MMR proteins are ultimately responsible for HNPCC and this study speculates in particular on the involvement of MLH1 function in the chemoprotective effect of aspirin.In addition to this, recent evidence has shown that although MLH1 expression is decreased in cases of sporadic colorectal adenoma, MLH1 expression was found to be increased in cases of sporadic colorectal adenomas with regular aspirin use (Sidelnikov et al., 2009).Aspirin is not alone in having the capacity to select for microsatellite stability: the NSAID mesalazine (5-aminosalicylic acid), used to treat patients with inflammatory bowel disease (IBD), experimentally reduced frameshift mutations in cultured CRC cells (HCT116 and HCT116+chr 3) independent of DNA mismatch repair proficiency (Gasche et al., 2005).In vitro, mesalazine inhibits the growth of HCT116 and HT-29 cells, a result reported a consequence of the compound inhibiting CDC25A, a cell cycle protein (Stolfi et al., 2008).Studies have also implicated cell-cycle arrest as a cellular response to aspirin exposure (Lai et al., 2008;Ricchi et al., 1997).Decreased transcription of CCNB1 (Cyclin B1) which regulates cell-cycle progression at the G 2 /M phase by sulindac treatment has been observed (Iizaka et al., 2002).To summarise: NSAIDs can select for microsatellite stability and cause cell-cycle arrest.

NSAIDs and DNA repair protein expression
Microarray analysis has been extensively utilised to examine alterations in gene expression in response to NSAID exposure in a number of colon cancer cell lines (Germann et al., 2003;Hardwick et al., 2004;Huang et al., 2006;Iizaka et al., 2002;Yin et al., 2006).Hardwick et al., (2004) reported that cell-cycle related genes and NF-κB were repressed upon aspirin treatment.Iizaka et al., (2002) reported an increase in GADD45α upon treatment with the NSAID sulindac.However, although, microarray analysis has been previously carried out to determine the effects of aspirin on colorectal cancer (Hardwick et al., 2004;Iizaka et al., 2002) these studies have not looked specifically at DNA damage signalling pathways.In addition, Hardwick et al., (2004) and also Yin et al., (2006) have shown that notably different gene expression patterns are seen when different concentrations of NSAID are used.Both studies showed that at low concentrations (1 mM) aspirin elicited activation of a different set of genes in the colorectal cancer cell line HT-29, compared to cells incubated with 5 mM aspirin.There is a lack of consistency in the literature regarding dosages utilised for both clinical and experimental studies.Intriguingly, clinical trials have shown that low dose aspirin (81mg) is more protective against colorectal cancer than 'high' dose aspirin (325mg) (Baron et al., 2003).The variation in experimental design, especially with regards to what dose is clinically relevant makes it problematic when deciding on appropriate drug concentrations for experimental work.
In addition to the report by Iizaka et al., (2002), a number of other studies have also shown NSAID regulation of growth arrest and DNA damage (GADD) gene expression.Germann et al (2003) analysed gene expression of CC531 colorectal cancer cells treated with 4.5mM Butyrate or 3mM aspirin and showed that GADD153 is up-regulated upon aspirin treatment.GADD153 expression has also been shown to be induced by celecoxib treatment in cervical cancer cells (Kim et al., 2007).Microarray analysis of the colorectal cancer cell line HT-29 treated with 1mM and 5mM aspirin showed an up-regulation of GADD45α (Yin et al., 2006).Further to this, microarray analysis also shows GADD45 gene expression to be upregulated by NS-398 (Huang et al., 2006) and celecoxib (Fatima et al., 2008) in colorectal cancer cells and by ibuprofen in human gastric adenocarcinoma cell line (Bonelli et al., 2010).Suppression of GADD45α expression confers resistance to sulindac and indomethacin induced gastric mucosal injury and apoptosis (Chiou et al., 2010) .Other genes involved in DNA repair have also been shown to be regulated by NSAIDs.High concentrations of NS-398, ibuprofen, and RNAi mediated inhibition of COX-2 in human prostate carcinoma cells affected genes involved in DNA replication, recombination and repair (John-Aryankalayil et al., 2009).PCNA gene expression has been shown to be up-regulated by ibuprofen treatment of a human gastric adenocarcinoma cell line (Bonelli et al., 2010), and also in the human hepatocellular carcinoma cell line HepG2 treated with 25mM vanillin (Cheng et al., 2007).In contrast, down-regulation of PCNA gene expression has been reported in colorectal cancer cell lines treated with celecoxib suggesting that these compounds may modulate cell cycle regulation in these model (Fatima et al., 2008).Interestingly, FADD gene expression is down-regulated by NS-398 but upregulated by indomethacin in the colorectal cancer cell line Caco-2 (Huang et al., 2006).Sulindac has been reported to down-regulate XRCC5, ERCC5 and UNG gene expression in colorectal cancer cell lines (Iizaka et al., 2002).Celecoxib up-regulated ATM, MAP3K2, CDKN1A and Bax gene expression in the colorectal cancer cell line HCA-7; in contrast, Bax gene expression was down-regulated in the HCT116 cell line upon celecoxib treatment (Fatima et al., 2008) suggesting that there may be variation in response from cell line to cell line to NSAID exposure.We have recently reported finding that XRCC3 protein expression in SW480 cells was increased upon 1 mM aspirin treatment for 48 hours (Dibra et al., 2010).The altered expression of XRCC3 upon aspirin exposure may have implications to the sensitivity of cells to chemotherapeutic agents.Indeed, previous research in the breast cancer cell line MCF7 demonstrated that over-expression of XRCC3 induced cisplatin resistance (Xu et al., 2005).Studies have shown that in contrast to XRCC3 over-expression and cisplatin resistance, XRCC3 deficient HCT116 cells have increased sensitivity to cisplatin and also mitomycin C (Yoshihara et al., 2004).Depletion of XRCC3 by siRNA in MCF7 cells inhibited cell proliferation, leading to accumulation of DNA breaks and triggering activation of p53dependant cell death (Loignon et al., 2007).Although some studies have shown no association between polymorphisms in XRCC3 and colorectal cancer risk (Mort et al., 2003;Yeh et al., 2005;Tranah et al., 2004;Skjelbred et al., 2006) some studies have (Improta et al., 2008), and in XRCC3 polymorphisms have also been associated with breast and lung cancer susceptibility (Smith et al., 2003;Jacobsen et al., 2004).One essential point to note is that there is a lack of information in current literature about the effects of NSAIDs on normal human colonocytes.The information that we have relates to cell lines/in vitro studies and murine models.There is a need to understand the effects of these compounds on normal, healthy colon cells and how these effects prevent carcinogenesis in these cells.With the majority of the models used in studies already established as cancerous, it is difficult to separate the effects of these compounds on normal cells rather than cells which are already cancerous and subject to pathway dysregulation.In addition to NSAIDs, flavonoids have also been associated with anti-carcinogenic properties.It is interesting to note that a recent study has shown that these compounds, which occur naturally in fruits and vegetables, not only have a protective effect against oxidative DNA damage but also increase repair activity in vitro (Ramos et al., 2010).To summarise: NSAIDs can alter the expression of DNA repair, and pro-and anti-apoptotic proteins.

Curcumin and DNA repair
Curcumin is a naturally occurring turmeric derivative with anti-inflammatory properties and apoptotic, anti-proliferative, anti-oxidant and anti-angiogenic effects (Shehzad et al., 2010).In vivo studies have shown that curcumin decreased intestinal polyp formation in the APC min/+ mouse model (Murphy et al., 2010).Curcumin is considered an attractive compound for chemopreventative use and there are at present clinical trials ongoing testing the effectiveness of the compound against different types of cancer (Shehzad et al., 2010).However, as with aspirin, a known mechanism of action of curcumin is yet to be elucidated with a range of molecular targets proposed (as reviewed in Shehzad et al., 2010 andRavindran et al., 2009).As discussed in a recent review (Burgos-Moron et al., 2010), although curcumin is widely regarded as a potential chemopreventative drug its safety and efficiency is yet to be fully elucidated and results from studies, clinical or otherwise, should be interpreted with this in mind.Curcumin has been shown to affect DNA damage and repair.Curcumin has been found to induce DNA damage in mouse-rat hybrid retina ganglion cells (Lu et al., 2009).The study reported that curcumin decreased expression of the DNA repair genes ATR, ATM, BRCA1, DNA-PK and MGMT.Curcumin has also been seen to cause DNA damage, as tested in the Comet Assay, in gastric mucosa cells and human peripheral blood lymphocytes (Blasiak et al., 1999).Curcumin has been shown to induce DNA single strand breaks (Scott & Loo, 2004) and induce the expression of the pro-apoptotic gene GADD153 in HCT116 colonocytes.It is suggested that the up-regulation of GADD153 is a direct response to DNA damage caused by curcumin, ultimately resulting in the induction of apoptosis (Scott & Loo, 2004).Microarray analysis of human lung cancer cells after curcumin treatment saw an upregulation of GADD45 and GADD153 gene expression (Saha et al., 2010) and microarray analysis has also shown an up-regulation of GADD45 by curcumin treatment in a human breast cancer cell line (Ramachandran et al., 2005).It has recently been proposed that DNA MMR may play a role in the cellular response to curcumin (Jiang et al., 2010): DNA MMR proficient cells showed a greater accumulation of double strand breaks (DSB) upon curcumin treatment compared to MMR deficient cells suggesting that DSB formation induced by curcumin is primarily a DNA MMR-dependent process; further to this, curcumin was reported to activate ATM/Chk1 and cause cell-cycle arrest and apoptosis in human pancreatic cancer cells.DNA MMR proficient cells showed activation of Chk1 and induction of the G(2)-M cell cycle checkpoint (Jiang et al., 2010) suggesting that the curcumin induced checkpoint response may be a DNA MMR dependent mechanism.Interestingly, microarray analysis of gene expression of invasive lung adenocarcinoma (CL1-5) cells exposed to curcumin found an induction in MLH1 gene expression and reduction in MMP expression (Chen et al., 2004).Curcumin is also reported to affect the Fanconi anemia (FA)/BRCA pathway.Curcumin sensitises ovarian and breast cancer cells with a functional FA/BRCA pathway to cisplatin (Chirnomas et al., 2006).This is a clinically significant finding as resistance to chemotherapeutic drugs such as cisplatin is common and monoketone analogs of curcumin are now being developed as a new class of FA pathway inhibitors (Landais et al., 2009).Recently, curcumin inhibition of the FA/BRCA pathway has also been suggested to be a mechanism for the reversal of multiple resistance in a multiple myeloma cell line (Xiao et al., 2010).In a recent phase IIa clinical trial report, curcumin reduced aberrant crypt foci formation (Carroll et al., 2011).

Interaction of NSAIDs with anti-cancer treatments
The capacity for specific cyclooxygenase inhibitors and NSAIDs to enhance the cellular response to chemotherapeutic and radiotherapeutic agents has been examined.Additive, synergistic and antagonistic effects have been reported.For example, the effect of a selective COX-2 inhibitor SC-236 (4-[5-(4-chlorophenyl)-3-trifluoromethyl)-1H-pyrazol-1-yl) benzene sulphonamide) was tested on cells derived from a murine sarcoma in the absence and presence of γ-ray irradiation, and a clonogenic cell survival assay confirmed that the compound significantly enhanced cell radiosensitivity (Raju et al., 2002).In the human hepatocellular carcinoma cell line HepG2, indomethacin and SC-236 enhanced doxorubicin toxicity reportedly via inhibiting P-glycoprotein and the multidrug resistance-associated protein 1 (MRP1) (Ye et al., 2011).Exogenous prostaglandin E2 addition failed to reverse the cellular accumulation and retention of doxorubicin and the authors concluded that the action of the drugs was via a COX-independent mechanism.Synergistic cell death characterised as being apoptotic based on Bax expression, DNA fragmentation and TUNEL assay, was observed in HT-29 cells co-treated with aspirin and 5-fluorouracil (Ashktorab et al., 2005).In marked contrast to the above findings, an antagonistic activity of celecoxib and SC-236 to cytotoxicity mediated by cisplatin to human esophageal squamous cell carcinoma cells has been observed; mechanistic analysis indicated that the compounds decrease cisplatin accumulation and DNA platination, in a COX-2 independent manner (Yu et al., 2011).As previously intimated, there is evidence that NSAIDs can affect double strand break repair pathways.It is known that the homologous recominbation protein Rad51 is overexpressed in chemo-radioresistent carcinomas.A study by Ko et al., (2009) showed that celecoxib enhanced gefitinib induced cytotoxicity in NSCLC cells: combined celecoxib/gefitinib treatment resulted in the reduction of Rad51 protein levels.Degradation of Rad51 occurred via a 26S proteasome-dependent pathway.Celecoxib has been shown to inhibit growth of head and neck carcinoma cells and enhance radiosensitivity in a dosedependent manner: celecoxib downregulated Ku70 protein expression and inhibited DNA-PKcs kinase activity which is known to be involved in DSB repair (Raju et al., 2005).Further to this, sodium salicylate has been shown to inhibit the kinase activity of ATM and DNA-PK which suppresses their DNA damage response (Fan et al., 2010).A recent study in prostate cancer demonstrated that treatment of the prostate cancer cell line PC-3 with NO-sulindac increased the rate of single strand DNA breaks and that there was slower repair of these lesions (Stewart et al., 2011).We thus suggest that caution is exercised in situations where patients are prescribed NSAIDs when undergoing chemotherapy or radiotherapy as unexpected additive or antagonistic reactions may arise, and thus potentially compromise treatment effectiveness.However, there is also the promise that NSAIDs (either known or novel) may significantly synergise and enhance responses to chemotherapeutic, biologic, or radiotherapeutic modalities.

Conclusion
The accumulated findings from a significant body of research indicates that NSAIDsincluding aspirin-act via multiple mechanisms in reducing the morbidity and mortality of cancer, and that aspirin and other NSAIDs reduce inflammation via COX-dependent and COX-independent pathways.NSAIDs can select for DNA mismatch repair competency and microsatellite stable cells thus inhibiting at least one well recognised pathway of colorectal cancer progression; furthermore, in vitro data suggest that NSAIDs exhibit 'direct' and relatively specific and rapid toxicity to colorectal cancer cells, with evidence suggesting that this may (but almost certainly not exclusively) involve DNA repair pathways.There is a paucity of information regarding the effect of NSAIDS on gene expression in nontransformed cell lines, including colonocytes and stromal cells and stem cells.Epidemiological evidence strongly indicates that NSAIDs are particularly protective against adenocarcinoma formation.Given the potential as chemopreventative agents, significant effort should be directed into producing novel NSAID derivatives that do not produce the adverse gastrointestinal and cardiovascular effects but retain the multiple and potent protective actions that are involved in suppressing adenocarcinoma formation.We feel that the following questions need to be addressed: a.What is the mechanism by which aspirin selects for microsatellite stability?b.What is the mechanism by which aspirin and other NSAIDs show their relatively specific toxicity to colorectal cancer cells?c.Does long term usage of aspirin and other NSAIDs preferentially reprogram the colonic epithelium via epigenetic mechanisms eg altering histone deacetylase and DNA methyltransferase activities?d.How do NSAIDS affect gene expression in non-transformed cells, including colonic stem cells?Do NSAIDs affect epithelial miRNA profiles?

Acknowledgement
We wish to thank Dr Paul Hooley for helpful discussions and comments during the preparation of this chapter.