Applicability domain of the GADD45a reporter assays: non-steroidal anti-inflammatory drugs do not produce misleading genotoxicity results

Abstract

Increased expression of the GADD45a gene is a very specific biomarker of genotoxin exposure in TK6 cells, and has been exploited in green fluorescent protein and luciferase reporter genotoxicity assays. A recent European Food Safety Authority suggested that the GADD45a-reporter genotoxicity assays might produce misleading positive results for non-steroidal anti-inflammatory drugs. This study was conceived to test the hypothesis that these drugs should be excluded from the applicability domain of the TK6 cell GADD45a GreenScreen HC and BlueScreen HC reporter assays. Data from published screening and validation studies have been reviewed, and new test data have been generated from 20 NSAIDS, from both reporter assays, both in the presence and absence of metabolic activation. The data fail to support this hypothesis: the high specificity of the GADD45a reporter assays is maintained amongst NSAIDs.

---

Introduction

The GADD45a gene has a central role in the maintenance of genome stability, and is a key component of the cellular response to DNA damage. It is a downstream target of the p53 transcription factor¹ and early studies implicated the gene in genotoxin induced cell cycle arrest,² DNA repair³ and apoptosis.⁴ The GADD45a gene product is a small globular protein, which modifies DNA accessibility in damaged chromatin, and associates with nuclear factors involved in the mitotic checkpoint. These findings are consistent with its function. Interestingly however, the GADD45a p53 response element (RE) is not found in the promoter region but in intron 3. Mutational analysis of the GADD45a p53 RE has shown that this site, along with an intronic WT1 binding site, is indeed critical in the proper regulation of the gene.⁵ Despite the key role of GADD45a in apoptosis, it has been demonstrated that its induction in TK6 cells is specific to compounds that trigger apoptosis as a consequence of DNA damage: non-genotoxic apoptogens do not induce the gene.⁶ In this respect the GADD45a gene has been shown to have a critical role in the successful application of DNA damaging anti-cancer therapeutics, which kill cancer cells through induction of apoptosis. Reduced GADD45a expression can produce resistance to cytotoxic anti-cancer drugs.^7,8

A GADD45a expression plasmid has been developed in which a reporter gene (encoding Green Fluorescent Protein ‘GFP’, or Gaussia luciferase ‘GLuc’) replaces a region spanning the translational start codon and intron 2 of the gene. It conserves the promoter, the p53 RE and the 5′ and 3′ un-translated regions of the gene. The transfection of the plasmid into the p53 competent TK6 human lymphoblastoid cell line has produced the GFP GreenScreen human cell (GSHC)⁹ genotoxicity assay. Systematic validation studies, both in the absence¹⁰ and presence^11,12 of S9 metabolic activation, have demonstrated that the assay is highly sensitive to agents that cause chromosome mis-segregation (aneugens) and breakage (clastogens), as well as diverse classes of DNA damage that can produce mutations (mutagens), whilst maintaining a high specificity. Together these properties have been demonstrated to make it an effective tool in the early screening for genotoxic liability in collections of novel chemicals, such as those developed as pharmaceuticals,¹³ pesticides¹⁴etc. This has been recognised by international committees, such as IVGT (ILSI HESI)¹⁵ as well as the UK Committee on Mutagenicity of Chemicals in Food, Consumer Products and the Environment (COM),¹⁶ and the European Food Safety Authority (EFSA).¹⁷ More recently, a modified version of the assay has been reported in which the GFP gene is replaced by a gene encoding the luciferase (Luc) protein from the marine copepod Gaussia princeps (BlueScreen HC, “BSHC”). It maintains the high sensitivity and specificity of the GFP assay.^18,19

A recent EFSA report regarding the scientific opinion on genotoxicity testing strategies applicable to food and feed safety assessment¹⁷ drew attention to the finding that GADD45a could be induced by various non-steroidal anti-inflammatory drugs (NSAIDs), such as sodium salicylate,²⁰ and warned that the mechanistic basis of such transcriptional assays “does not guarantee that DNA damage response (DDR)/stress pathways gene activation will necessarily involve DNA damage”. The GADD45a reporter assays are increasingly being adopted into early genotoxicity hazard assessment screening, so it is important to investigate any potential limitations to their applicability. This paper was conceived to assess whether the NSAID therapeutic class could be included or should be excluded from the pharmaceuticals applicability domain.

NSAIDs, such as aspirin, ibuprofen, diclofenac, indomethacin and sulindac are widely available and safely used in the suppression of inflammation, pain and fever. Pharmacologically they inhibit the cyclooxygenase (COX)-mediated production of prostaglandins, which affect muscle and blood vessel functions, pain sensing and body temperature. Two isoforms of the COX enzyme exist, which are known to catalyse the biosynthesis of prostaglandins COX-1 and COX-2. COX-1 is constitutively expressed at high levels in cells and tissues such as endothelium, monocytes and platelets, indicating that it is developmentally regulated. COX-2 however is induced by inflammatory mediators in a wide variety of cells and tissues such as vascular endothelium, osteoclasts and macrophages.²¹ NSAIDs can exhibit different inhibitory activities towards the two COX isoforms. For example, celecoxib, etodolac and meloxicam selectively inhibit COX-2, whereas, aspirin, ibuprofen and diclofenac are able to inhibit both COX-1 and COX-2 isoforms with little selectivity.²¹ A third distinct COX isoenzyme, COX-3, has not been shown to exhibit any cyclooxygenase activity, suggesting a lack of involvement in prostaglandin mediated pain and inflammation.²²

In early studies (reviewed in ref. 23), prostaglandins were found at higher levels in human colon, lung and breast tumours, compared to the surrounding tissues. There is a lower risk of colon cancer amongst regular users of NSAIDs and in rodent studies high doses of NSAIDs have been shown to suppress chemically induced carcinoma.²³ These studies suggest that rather than being carcinogens, NSAIDs can be protective against carcinogens in the colon. Previous studies have demonstrated that multiple NSAIDs can up-regulate GADD45a expression in various cancer cell lines.^20,24–28 Zerbini and co-workers demonstrated that NSAIDs, such as sulindac sulphide, diclofenac, flufenamic acid, flurbiprofen, sulindac sulphone, and NS-398 induce apoptosis in the DU145 prostate cancer cell line.²⁷ Furthermore, sulindac sulphide, the active metabolite of sulindac, induces apoptosis in numerous other prostate, renal, breast and stomach cancer cell lines. However, when tested in untransformed cells, such as F12 foreskin fibroblasts, no induction of apoptosis by sulindac sulphide was observed.

Since GADD45a has a key role in apoptosis, the same study²⁷ compared GADD45a, GADD45b and GADD45g mRNA expression in response to sulindac sulphide treatment, in a variety of cell lines using real-time PCR. GADD45a was found to be significantly induced in breast, renal, and stomach cancer cell lines. It was further demonstrated that MDA-7/interleukin-24 protein mediates the NSAID induction of apoptosis and GADD45a in cancer cell lines. This occurs via c-Jun NH₂-terminal kinase (JNK) activation and growth arrest induction through inhibition of Cdc2-cyclin B checkpoint kinase.²⁹ However, mRNA expression does not always translate directly into protein levels, and for GADD45a in particular it has been shown in mouse embryonic fibroblasts that whilst transcription is increased by many different stresses, the protein accumulates only for a subset of these.²⁹ The study by Zerbini et al.²⁷ reported an investigation of GADD45a protein expression in DU145 and PC-3 prostate cancer cells for only one compound, sulindac sulphide, and was found to increase.

More recently Chiou and Hoa reported that indomethacin and sulindac sulphide induce GADD45a mRNA and protein synthesis in three human colon cancer cell lines.²⁸ GADD45a up-regulation was accompanied by apoptotic and necrotic cell death, which is consistent with the anti-neoplastic effect of NSAIDs in colon tumourigenesis and cancer growth. The GADD45a reporter assays use a lymphoblastoid (TK6) cell line with a wild type p53 gene. Many of the studies reviewed above use cancer cell lines with a dysfunctional p53 gene: DU145 is p53 mutant²⁷ and PC-3 is p53 null.³⁰

The GADD45a gene is a p53 target, and whilst the results from the p53 mutant lines discussed above might not be relevant to the GSHC assay, they stimulated the two studies reported here that were conceived to test the hypothesis that NSAIDs are liable to produce misleading positive results in the GADD45a reporter assays. One is a retrospective analysis of NSAID data from previously published GSHC studies, and the second presents new data from 20 NSAIDs with both GSHC and BSHC, with and without S9.

Materials and methods

Compound selection

An historical review of published GSHC assay data revealed that 21 NSAID and related compounds had been tested: 4 of these were included in a study of 75 marketed pharmaceuticals using the standard assay format,¹³ without S9. A further 17 were included in a study of the 1266-compound Sigma-Aldrich Library of Pharmacologically Active Compounds (LOPAC),³¹ also without S9. The top test concentration in the LOPAC study was limited to 0.1 mM because compounds were supplied at 10 mM in 100% DMSO (the assay requires dilution to 1% DMSO), and the study was limited to 3 dilutions (cf. 9 in the standard GSHC protocol, and 8 in the BSHC assay).

For this new study, 20 NSAIDs and related compounds were selected for assessment in both the GSHC and BSHC assays, both in the presence and absence of S9 metabolic activation, and up to a top test concentration of 10 mM, where not limited by solubility or toxicity (i.e. the limit prior to ICH S2(R1)). A subset of NSAIDs listed in the Merck Index³² was selected on the basis of availability, compound purity, coverage of sub-classes, and the availability of comparative genotoxicity test data (including GSHC) in order to create overlap with previous studies. Table 1 lists the compounds selected for testing and summarises information on source, purity and published in vitro and in vivo genotoxicity data, as well as carcinogenicity study data. These published data were drawn from the extensive review by Brambilla and Martelli³³ unless otherwise indicated in Table 1. The selected compounds included: 11 previously tested in the LOPAC study;³¹ 4 from the marketed pharmaceuticals study;¹³ 2 NSAIDs (sulindac and indomethacin) previously shown to increase GADD45a mRNA levels;²⁸ 2 sulindac metabolites, sulindac sulphide and sulindac sulphone; the active/inactive analogues, celecoxib and 2,5-dimethyl-celecoxib. All compounds were tested in triplicate in each reporter assay (±S9). Published results from GADD45a mRNA expression and protein expression studies are included in the final columns of Table 2.

Table 1 Compounds selected for testing

Compound name	CAS Registry no.	Chemical supplier and grade		Alternative test results
				Ames		MCM		In vitro CAb/MNT		In vivo CAb/MNT	Carc. studies
				Result		Result		Result		Result	Result
		Supplier	Grade	−S9	+S9	−S9	+S9	−S9	+S9
Key: Ames, Salmonella reversion test; MCM, mammalian cell mutation (mouse lymphoma tk assay or hypoxanthine-guanine phosphoribosyltransferase assay); CAb, chromosome aberration; MNT, micronucleus test; Carc, carcinogenicity; S9, metabolic activation, ‘−’, negative; ‘+’, positive; ‘±’, positive and negative studies; I, inconclusive results; superscripted numbers refer to reference source at end of manuscript.
Aspirin	50-78-2	Sigma	Pharmaceutical secondary standard	−	−	−	−	±	±	±	−
Celecoxib	169590-42-5	Sigma	≥98% (HPLC)	−	−	−	−	−	−	−	−
Diclofenac sodium	15307-79-6	Sigma	Pharmaceutical secondary standard	−	−	−	−	−	−	−	−
Diflunisal	22494-42-4	Fluka	≥98% (HPLC)	−	−	−	−			+	−
2,5-Dimethyl-celecoxib	457639-26-8	Fluka	Analytical standard
Etodolac	41340-25-4	Sigma	Meets USP testing specifications	−	−	−	−	+		−	−
Ibuprofen sodium	31121-93-4	Fluka	≥98% (GC)	−	−					+	−
Indomethacin	53-86-1	Sigma	Meets USP testing specifications	−	−			I		±	±
Ketoprofen	22071-15-4	Fluka	Analytical standard	−	±					±	−
Mefenamic acid	61-68-7	Sigma	≥98% (TLC)	−	−			−	−
Meloxicam sodium	71125-39-8	Sigma	99% (HPLC)	−	−			−	−	−	−
Nabumetone	42924-53-8	Fluka	≥98% (TLC)	−	−			+	+	−	−
Naproxen	22204-53-1	Sigma	Meets USP testing specifications	−	−					+	−
Nimesulide	51803-78-2	Sigma	≥98% (TLC)							+
Oxaprozin	21256-18-8	Sigma	≥98% (TLC)	−	−	−	−	−	−	−	±
Piroxicam	36322-90-4	Sigma	Meets USP testing specifications							I	−
Sulindac	38194-50-2	Sigma	≥98%								+^34
Sulindac sulphide	32004-67-4	Sigma	≥98% (HPLC)
Sulindac sulphone	59864-04-9	Sigma	≥97% (TLC)								−^35
Valdecoxib	181695-72-7	Sigma	≥98% (HPLC)	−	−	−	−	−	−	−	−

---

Table 2 Published GreenScreen HC and GADD45a mRNA/protein expression results for 21 NSAIDs and related compounds

Compound name	Highest test concentration (mM)	Top dose limited by	GreenScreen HC results				Induction of GADD45a expression
			Growth inhibition		Genotoxicity, GFP induction		mRNA	Protein
			Result	LEC (mM)	Result	LEC (mM)	Result	Result
Abbreviations: Top dose was limited by either ‘prep’, compound preparation (supplied at 10 mM in 100% DMSO); ‘Cyto’, cytotoxicity; ‘Sol’, solubility. LEC, lowest effective concentration; ‘−’, negative; ‘+’, positive; ‘±’, positive and negative studies, #, compounds also re-tested in the current study, superscripted numbers refer to reference source. All GreenScreen HC results refer to data from tests performed in the absence of S9 with a 48 hour test compound exposure period.
Aspirin^#^31	0.10	Prep	−		−		+^20
Diclofenac^#^13	1.71	Cyto	+		−		+^24,36
Diflunisal^#^13	0.16	Sol	±		−
Etodolac^#^13	1.00	Sol	±		−
Fenspiride hydrochloride^31	0.10	Prep	−		−
Ibuprofen^#^31	0.10	Prep	−		−		+^26
Indomethacin^#^31	0.10	Prep	+		−		+^28	+^28
Ketoprofen^#^31	0.10	Prep	−		−
Ketorolac^31	0.10	Prep	−		−
Loxoprofen^31	0.10	Prep	−		−
Meclofenamate sodium^31	0.10	Prep	+	0.03	+	0.03
Meloxicam^#^31	0.10	Prep	−		−
Nabumetone^#^13	0.13	Sol	±		−
Naproxen^#^31	0.10	Prep	−		−		+^36
Niflumic acid^31	0.10	Prep	+	0.05	−
Nimesulide^#^31	0.10	Prep	+	0.01	−
Oxaprozin^#^31	0.10	Prep	−		−
Piroxicam^#^31	0.10	Prep	−		−
SC-560^31	0.10	Prep	+	0.03	−
Sulindac^#^31	0.10	Prep	−		−		+^25	+^25
Sulindac sulphone^#^31	0.10	Prep	−		−		+^36

---

GSHC and BSHC reporter assays

The GSHC assay methods, both without S9⁹ and with S9,¹² as well as the BSHC assay with and without S9¹⁸ have been described in detail elsewhere. Briefly, both are 96-well microplate assays in which 4 compounds are each tested at multiple dilutions (9 in GSHC, 8 in BSHC). Appropriate cell-free control wells containing solvent alone, assay medium alone, and test compounds alone in assay medium are provided on each microplate. Cell-containing control wells are also provided on each microplate, which include vehicle (1% DMSO) and high and low doses of control genotoxins or pro-genotoxins, as appropriate to the S9 condition of the assay. The GSHC assay employs 2 strains of TK6: a fluorescent ‘test’ strain expressing GFP, and a ‘control’ strain, in which an additional base pair after translational start prevents GFP protein expression. One set of compound dilutions has test cells added, a second has control cells added. The control cell wells identify fluorescent compounds which might confound data analysis. Data are collected by flow cytometry for the S9 assay protocol. The BSHC assay utilises a single luminescent strain of TK6. An additional strain to control for luminescent compounds is not needed because this property is extremely rare (compound-only control wells would identify compound luminescence). In a typical experiment test compounds are tested in a 2-fold dilution series (microplate layouts have been published previously for GSHC⁹ and BSHC¹⁸). Compounds tested in the absence of S9 were incubated with cells for 48 hours. Compounds tested in the presence of S9 were incubated with cells for 3 hours, washed to remove the S9, then incubated for a further 45 hours in fresh medium without compound, prior to data collection.

At the end of the assay, reporter outputs (GFP fluorescence, or GLuc luminescence) were recorded together with a measure of culture density: light absorbance (620 nm) for GSHC; thiazole orange (TO) fluorescence for BSHC. TO is a nucleic acid binding dye, detectable using standard fluorescein (FITC) filters (Ex. 485 nm, Em. 535 nm). The TO dose response is more sensitive and remains linear to lower cell densities than the light absorbance method, and this is critical in the 384-well version of the assay¹⁹ where lower cell numbers are available for assessment. The fluorescent/luminescent reporter output is divided by a measure of cell density to produce a ‘brightness’ value. In this way the assay discriminates between wells containing low numbers of strongly light-emitting cells and wells containing high numbers of weakly light-emitting cells. A positive result is concluded where exposure within the acceptable toxicity range produces a significant increase in brightness. This is defined as greater than 3 times the standard deviation derived from studies of toxic and non-toxic non-genotoxins. From the triplicate testing, compounds producing 3/3 or 2/3 positive results were recorded as positive for each assay.

Results

Historical GADD45a-GFP assay data review

There are published data for 21 NSAIDs or NSAID derivatives tested using the GSHC assay in the absence of S9. A study of 75 marketed pharmaceuticals, each tested to a top dose of 10 mM unless limited by toxicity or solubility, included 4 NSAIDs: diclofenac, diflunisal, etodolac and nabumetone. All produced negative results (Table 2), allowing the initial conclusion that NSAIDs as a class are not generally liable to produce misleading positive results in the GADD45a-GFP assay. A much larger study of the 1266 compounds from the LOPAC collection included 17 NSAIDs. In that study, testing was limited to three doses (100 μM, 50 μM and 25 μM), but where any dose exceeded the toxicity limit, the compound was retested in a lower dose range. One compound out of 17, meclofenamate sodium, produced a positive GADD45a result in the GSHC assay at an LEC of 0.03 mM.

2 of the 21 compounds had also been tested by Chiou and Hoa²⁸ (see Introduction). Sulindac sulphone, the inactive sulindac metabolite did not stimulate GADD45a expression in the Chiou and Hoa study, and produced a negative result in the GADD45a-GFP LOPAC screening study.³¹ Indomethacin, which was found to induce both GADD45a mRNA and protein synthesis in the Chiou and Hoa study, was also negative in GSHC. Overall 20 of the 21 compounds (95%) in this data review produced negative GSHC results, supporting the contention that NSAIDs do not generally produce misleading positive data in the assay.

Study of 20 NSAIDs and derivatives in the GSHC and BSHC assays

Tables 3 and 4 summarise the new experimental data generated from 20 NSAID or NSAID-derived compounds, using the GSHC assay (Table 3) and the BSHC assay (Table 4), respectively. All were tested with and without S9. All compounds produced negative results in the GSHC assay, both with and without S9. All compounds produced negative results in the BSHC assay with S9, and all except mefenamic acid produced negative results in the BSHC without S9.

Table 3 Results from 20 NSAID/related compounds tested in the GreenScreen HC assay in both the presence and absence of S9

Compound name	Test conc. limitation^a	Highest test conc.		GreenScreen HC results
				Growth inhibition		Genotoxicity, GFP induction		Growth inhibition		Genotoxicity, GFP induction
				Without S9, 48 h exposure				With S9 – 3 h exposure^b
		mM	μg ml^−1	Result^c	LEC (mM)	Result^d	LEC (mM)	Result	LEC (mM)	Result	LEC (mM)
a See the footnote in Table 2.b Treatment of cells with test article was for a 3 h period in the presence of S9 followed by 45 h recovery without S9.c All cytotoxicity and genotoxicity results are for three independent experiments.d A positive result for genotoxicity is observed when the GFP induction is greater than 1.5-fold (without S9) st or greater than 1.3-fold (with S9). ‘+’, positive, ‘−’, negative.
Aspirin	mM	9.83	1771	+++	9.83	−−−		−−−		−−−
Celecoxib	Sol	0.05	19	−++	0.02	−−−		−−−		−−−
Diclofenac sodium	Cyto	1.04	331	+++	0.03	−−−		−−−		−−−
Diflunisal	Sol	0.17	43	+++	0.04	−−−		−−−		−−−
2,5-Dimethyl-celecoxib	Sol	0.02	7.91	+++	0.01	−−−		−−−		−−−
Etodolac	Sol	0.26	75	+++	0.13	−−−		−−−		−−−
Ibuprofen sodium	Sol	0.31	71	−−−		−−−		−−−		−−−
Indomethacin	Sol	0.29	104	+++	0.04	−−−		−−−		−−−
Ketoprofen	Sol	1.00	254	+++	0.50	−−−		−−−		−−−
Mefenamic acid	Sol	0.18	43	+++	0.09	−−−		−−−		−−−
Meloxicam sodium	Sol	1.48	553	+++	0.19	−−−		−−−		−−−
Nabumetone	Sol	0.22	50	+++	0.03	−−−		−−−		−−−
Naproxen	Sol	0.25	58	−−−		−−−		−−−		−−−
Nimesulide	Sol	0.28	86	+++	0.14	−−−		−−−		−−−
Oxaprozin	Sol	0.21	62	+++	0.21	−−−		−−−		−−−
Piroxicam	Sol	0.65	215	+++	0.65	−−−		−−−		−−−
Sulindac	Sol	0.19	68	−−+	0.19	−−−		−−−		−−−
Sulindac sulphide	Sol	0.18	61	+++	0.02	−−−		−−−		−−−
Sulindac sulphone	Sol	0.06	22	−−−		−−−		−−−		−−−
Valdecoxib	Sol	0.11	35	+++	0.03	−−−		−−−		−−−

---

Table 4 Results from 20 compounds tested in the BlueScreen HC assay in both the presence and absence of S9

Compound name	Test conc. limitation^a	Highest test conc.		BlueScreen HC results
				Growth inhibition		Genotoxicity, GLuc induction		Growth inhibition		Genotoxicity, GLuc induction
				Without S9 – 48 h exposure				With S9 – 3 h exposure^b
		mM	μg ml^−1	Result^c	LEC (mM)	Result^d	LEC (mM)	Result	LEC (mM)	Result	LEC (mM)
See the footnotes in Table 3.
Aspirin	mM	9.83	1771	+++	4.92	−−−		−−−		− − −
Celecoxib	Sol	0.05	19	+++	0.02	−−−		−−−		− − −
Diclofenac sodium	Cyto	1.04	331	+++	0.13	−−−		+−−	1.04	− − −
Diflunisal	Sol	0.17	43	+++	0.08	−−−		+−−	0.17	− − −
2,5-Dimethyl-celecoxib	Sol	0.02	7.91	+++	0.01	−−−		−−−		− − −
Etodolac	Sol	0.26	75	+++	0.13	−−−		−−−		− − −
Ibuprofen sodium	Sol	0.31	71	−−−		−−−		−−−		− − −
Indomethacin	Sol	0.29	104	+++	0.15	−−−		−−−		− − −
Ketoprofen	Sol	1.00	254	+−+	1.00	−−−		−−−		− − −
Mefenamic acid	Sol	0.18	43	+++	0.05	+++	0.09	−−−		− − −
Meloxicam sodium	Sol	1.48	553	+++	0.37	−−−		−−−		− − −
Nabumetone	Sol	0.22	50	+++	0.22	−−−		−−−		− − −
Naproxen	Sol	0.25	58	−−−		−−−		−−−		− − −
Nimesulide	Sol	0.28	86	+++	0.07	−−−		−+−	0.28	− − −
Oxaprozin	Sol	0.21	62	+−+	0.21	−−−		−−−		− − −
Piroxicam	Sol	0.65	215	+++	0.65	−−−		−−−		− − −
Sulindac	Sol	0.19	68	−−−		−−−		−−−		− − −
Sulindac sulphide	Sol	0.18	61	+++	0.05	−+−	0.05	−−−		− − −
Sulindac sulphone	Sol	0.06	22	−−−		−−−		−−−		− − −
Valdecoxib	Sol	0.11	35	+++	0.11	−−−		−−−		− − −

---

Discussion

Both the historical data and the results from the experimental study reported here refute the hypothesis that NSAIDs should be excluded from the applicability domain of the GSHC or BSHC reporter assays. In the broader chemical space encompassed by the ECVAM recommended lists, 96% of non-carcinogens have produced negative data in the GSHC assay.³⁷ This high specificity for the detection of non-carcinogens is also maintained in the BSHC assay.¹⁸

Amongst the 21 NSAIDs with previously reported GSHC assay data, 20 (95%) produced negative results. Those studies were carried out in the absence of S9 metabolic activation, and whilst some compounds were tested to a maximum of 10 mM, others were tested in a screening mode, up to 0.1 mM. The single positive result was produced by meclofenamate sodium without S9 (from the LOPAC study³¹): it has produced negative results in all but one of the published genotoxicity studies identified – a chromosome aberration study produced a positive result with CHO cells, only with metabolic activation.³⁸ It produces a plausible DEREK alert for mammalian carcinogenicity, but no published positive carcinogenicity studies were identified.

Twenty additional compounds were tested using GSHC and BSHC, with and without S9, and to 10 mM (unless limited by solubility or toxicity). Five of these had not been tested before, and 11 had only been tested to 0.1 mM in previous studies.³¹ GSHC produced negative results for all 20 (100%) and BSHC produced negative results for 19 of 20 (95%). Thus high specificities of the GSHC and BSHC reporter assays were maintained in this targeted NSAID study. Mefenamic acid, without S9, was the lone BSHC positive, and produced negative results in other published genotoxicity studies. Since this was the only compound to produce different results between GSHC and BSHC, the data were re-analysed, using light absorbance rather than TO fluorescence to normalise data. This produced an overall negative result (−−+), with and without S9, in agreement with the GSHC assay results. The presence of a single outlier using the standard protocols demonstrates that the assays are substantially equivalent in the testing of NSAIDs. There was also just a single outlier in previously published results from the ECVAM recommended lists with both assays, with and without S9. In that study EDTA produced a positive result for GSHC only (EDTA, 5 mM).³⁷

The new data reported here were generated from assays in which compounds were tested to 10 mM, unless limited by solubility or toxicity. Only 3 compounds, aspirin, diclofenac and meloxicam, were soluble and in the acceptable toxicity range above 1 mM and all produced negative results. The single compound producing a reproducible positive result had an LEC of 0.09 mM. Hence, if the new 1 mM top dose ICH S2(R1) guidance³⁹ had been applied, the overall findings of this study would not have been changed.

The qualitative and quantitative results from cell based assays are expected to vary according to the cell type, not least because they are derived from different individuals. In addition, different tissues from the same individual have intrinsically different patterns of gene expression. Finally, cell lines derived from tissues or tumours are all mutants that have escaped the normal restrictions of proliferation. Of particular relevance to genetic toxicology, many cell lines carry p53 mutations leading to an impaired DNA damage response. The TK6 cells utilised in the GSHC and BSHC assays are p53 wild type. However, whilst many of the reports of NSAID-induced GADD45a expression discussed in the introduction were from p53-deficient cell lines, the RKO human colon cancer cell lines are p53 wild type,²⁸ so this alone does not account for the differences in GADD45a expression between the different cell types. Thus until the basis for the difference between cell lines in their GADD45a response is fully understood, any new cell line to be used to screen for genotoxicity using this GADD45a reporter assay expression would need to be fully evaluated for its specificity and sensitivity, as has been done for the TK6 cell line. This argument should be extended to any assay transferred into a new cell line.

In the current study, although 16/20 NSAIDs and related compounds produced cytotoxic results in the GADD45a reporter assays at the concentrations tested, only mefenamic acid gave rise to an increase in GADD45a expression in all 3 repeat tests (BSHC assay without S9). This suggests that NSAID induced cytotoxicity of TK6 cells does not stimulate GADD45a expression. This finding is consistent with a previous study from our laboratory which investigated the ability of non-genotoxic apoptosis inducers to produce misleading positive results in the GSHC assay.⁶ Since GADD45a expression can be induced by both apoptotic pathways and DNA damage responsive (DDR) pathways, it is important to assess the liability of the GSHC assay to produce misleading positive results due to compounds that induce GADD45a via apoptotic pathways and not DDR pathways. In that previous study a panel of 17 compounds potentially capable of stimulating apoptosis in the GADD45a-GFP assay or of inducing GADD45a in the absence of genotoxic stress, such as p53 activators, NF-κB and Bcl-2 inhibitors were tested. Most did not have carcinogenicity or in vivo genotoxicity data. Only 4/17 induced the GADD45-GFP reporter, demonstrating that positive results are not prevalent amongst apoptogens. The 4 exceptions produced positive results in concurrent comet assay and/or micronucleus tests. This is significant because several studies have shown that some NSAIDs and related compounds, such as sulindac sulphide and diclofenac, induce apoptosis in cell lines derived from human melanoma and prostate cancers.^20,27,40 Apoptosis occurs via an mda-7 (IL-24) mediated mechanism involving the coordinated overexpression of the GADD family of genes by means of p38 MAPK1.⁴⁰ However, this has not been found to be the case in non-cancerous immortalised cell lines.

The TK6 cell line was derived from a spontaneously immortalized lymphoblast culture, WIL237. WIL2 is a cell line derived from a patient who at the time of culture initiation did not show evidence of malignant diseases.⁴¹ TK6 was isolated from WIL2 through a series of mutagenic and selection procedures to produce a cell line heterozygous for thymidine kinase.⁴¹ Although the properties of TK6 cells are similar to those derived from Burkitt's Lymphoma or leukaemic patients,⁴² the cell lines used for NSAID testing in previous studies by Zerbini et al.²⁷ and Chiou and Hoa²⁸ were cancer cell lines isolated from patients actually presenting these forms of cancer. Therefore differences between cancer cell lines and lymphoblastoid cell lines may account for differences in GADD45a induction observed in these cell lines. TK6 is a cell line recommended for genotoxicity testing due to its p53 status. Fowler and co-workers⁴³ performed micronucleus testing of 19 compounds that are accepted as producing misleading or “false” positive results in in vitro mammalian cell assays on a range of both rodent cell lines such as V79, CHL and CHO and p53 competent cell lines such as TK6.⁴¹ This was done in order to determine if choice of cell type could reduce the frequency of misleading positive results in genotoxicity testing. They concluded that the use of a p53-competent, non-rodent, cell type such as TK6, HuLy or HepG2 would, according to data generated in that study, reduce the incidence of misleading positives and the authors strongly recommended using p53-competent cell types.

Conclusions

This study reports an investigation designed to test the hypothesis that NSAIDs as a therapeutic class should be excluded from the applicability domain of the TK6 cell GADD45a GSHC/BSHC reporter assays. The results presented in this study do not support this hypothesis. NSAIDs are not prone to the generation misleading positive results in the GSHC and BSHC assays and since NSAIDs are not genotoxic carcinogens, these results confirm the high specificity of the GSHC and BSHC assays. NSAIDs need not therefore be excluded from the applicability domain of the GSHC/BSHC reporter assays.

Conflict of interest

RMW is founder and CSO of Gentronix Ltd which developed and sells the GSHC and BSHC genotoxicity assays.

Abbreviations

GADD	Growth arrest and DNA damage
Luc	Luciferase
GFP	Green fluorescent protein
NSAIDs	Non-steroidal anti-inflammatory drugs
GSHC	GreenScreen HC
BSHC	BlueScreen HC
HESI	Health and Environmental Sciences Institute
ILSI	International Life Sciences Institute
IVGT	Project Committee on Relevance and Follow-Up of Positive Results in In Vitro Genetic Toxicity Testing
TO	Thiazole orange

Acknowledgements

Christopher Jagger is thanked for critically reading the manuscript.

References

1. M. B. Kastan, Q. Zhan, W. S. El-Deiry, F. Carrier, T. Jacks, W. V. Walsh, B. S. Plunkett, B. Vogelstein and A. J. Fornace Jr, A mammalian cell cycle checkpoint utilizing p53 and GADD45 is defective in ataxia telangiectasia, Cell, 1992, 71, 587–597.
2. Q. Zhan, M. J. Antinore, X. W. Wang, F. Carrier, M. L. Smith, C. C. Harris and A. J. Fornace Jr, Association with Cdc2 and inhibition of Cdc2/cyclin B1 kinase activity by the p53-regulated protein Gadd45, Oncogene, 1999, 18, 2892–2900.
3. F. Carrier, I. Bae, M. L. Smith, D. M. Ayers and A. J. Fornace Jr, Characterization of the GADD45 response to ionizing radiation in WI-L2-NS cells, a p53 mutant cell line, Mutat. Res., Fundam. Mol. Mech. Mutagen., 1996, 352, 79–86.
4. M. Takekawa and H. Saito, A family of stress-inducible GADD45-like proteins mediate activation of the stress-responsive MTK1/MEKK4 MAPKKK, Cell, 1998, 95, 521–530.
5. D. Johnson, P. W. Hastwell and R. M. Walmsley, The involvement of WT1 in the regulation of GADD45a in response to genotoxic stress, Mutagenesis, 2013, 28(4), 393–399.
6. C. H. Topham, N. Billinton and R. M. Walmsley, Nongenotoxic apoptosis inducers do not produce misleading positive results in the TK6 cell-line, Toxicol. Sci., 2012, 128, 79–91.
7. A. Bahar, J. E. Bicknell, D. J. Simpson, R. N. Clayton and W. E. Farrell, Loss of expression of the growth inhibitory gene GADD45γ, in human pituitary adenomas, is associated with CpG island methylation, Oncogene, 2003, 23, 936–944.
8. K. Al-Romaih, B. Sadikovic, M. Yoshimoto, Y. Wang, M. Zielenska and J. A. Squire, Decitabine induced demethylation of 5 CpG-island in GADD45A leads to apoptosis in osteosarcoma cells, Neoplasia, 2008, 10, 471–480.
9. P. W. Hastwell, L. L. Chai, K. J. Roberts, T. W. Webster, J. S. Harvey, R. W. Rees and R. M. Walmsley, High-specificity and high-sensitivity genotoxicity assessment in a human cell line: validation of the GreenScreen HC GADD45a-GFP genotoxicity assay, Mutat. Res., Genet. Toxicol. Environ. Mutagen., 2006, 607, 160–175.
10. N. Billinton, P. W. Hastwell, D. Beerens, L. Birrell, P. Ellis, S. Maskell, T. W. Webster, S. Windebankd, F. Woestenborghs, A. M. Lynch, A. D. Scott, D. J. Tweats, J. van Gompel, R. W. Rees and R. M. Walmsley, Inter-laboratory assessment of the GreenScreen HC GADD45a-GFP genotoxicity screening assay: an enabling study for independent validation as an alternative method, Mutat. Res., Genet. Toxicol. Environ. Mutagen., 2008, 653, 23–33.
11. C. Jagger, M. Tate, P. A. Cahill, C. Hughes, A. W. Knight, N. Billinton and R. M. Walmsley, Assessment of the genotoxicity of S9-generated metabolites using the GreenScreen HC GADD45a-GFP assay, Mutagenesis, 2009, 24(1), 35–50.
12. N. Billinton, S. Bruce, J. R. Hansen, P. W. Hastwell, C. Jagger, C. McComb, M. L. Klug, K. Pant, A. Rabinowitz, R. Rees, M. Tate, A. M. Vinggaard and R. M. Walmsley, A pre-validation transferability study of the GreenScreen HC GADD45a-GFP assay with a metabolic activation system (S9), Mutat. Res., Genet. Toxicol. Environ. Mutagen., 2010, 700, 44–50.
13. P. W. Hastwell, T. W. Webster, M. Tate, N. Billinton, A. M. Lynch, J. S. Harvey, R. W. Rees and R. M. Walmsley, Analysis of 75 marketed pharmaceuticals using the GADD45a-GFP ‘GreenScreen HC’ genotoxicity assay, Mutagenesis, 2009, 24, 455–463.
14. A. Knight, S. Little, K. Houck, D. Dix, R. Judson, A. Richard, N. McCarroll, G. Akerman, C. Yang, L. Birrell and R. Walmsley, Evaluation of high-throughput genotoxicity assays used in profiling the US EPA ToxCastTM chemicals, Regul. Toxicol. Pharmacol., 2009, 55, 188–199.
15. A. M. Lynch, J. C. Sasaki, R. Elespuru, D. Jacobson-Kram, V. Thybaud, M. De Boeck, M. J. Aardema, J. Aubrecht, R. Daniel Benz, S. D. Dertinger, G. R. Douglas, P. A. White, P. A. Escobar, A. Fornace Jr, M. Honma, R. T. Naven, J. F. Rusling, R. H. Schiestl, R. M. Walmsley, E. Yamamura, J. van Benthem and J. H. Kim, New and emerging technologies for genetic toxicity testing, Environ. Mol. Mutagen., 2011, 52, 205–223.
16. Committee on Mutagenicity of Chemicals in Food, Consumer Products and the Environment, http://www.iacom.org.uk/index.htm: March 2010.
17. EFSA, Scientific opinion on genotoxicity testing strategies applicable to food and feed safety assessment, EFSA J., 2011, 9, 2379–2438.
18. C. Hughes, A. Rabinowitz, M. Tate, L. Birrell, J. Allsup, N. Billinton and R. M. Walmsley, Development of a high-throughput Gaussia luciferase reporter assay for the activation of the GADD45a gene by mutagens, promutagens, clastogens, and aneugens, J. Biomol. Screening, 2012, 17, 1302–1315.
19. K. Simpson, N. Bevan, P. W. Hastwell, P. Eidam, P. Shah, E. Gogo, S. Rees and A. Brown, The BlueScreen-384 assay as an indicator of genotoxic hazard potential in early-stage drug discovery, J. Biomol. Screening, 2013, 18(4), 441–452.
20. D. Sarkar, Z.-Z. Su, I. V. Lebedeva, M. Sauane, R. V. Gopalkrishnan, K. Valerie, P. Dent and P. B. Fisher, mda-7 (IL-24) mediates selective apoptosis in human melanoma cells by inducing the coordinated overexpression of the GADD family of genes by means of p38 MAPK, Proc. Natl. Acad. Sci. U. S. A., 2002, 99, 10054–10059.
21. P. N. Praveen Rao and E. E. Knaus, Evolution of nonsteroidal anti-inflammatory drugs (NSAIDs): COX inhibition and beyond, J. Pharm. Pharm. Sci., 2008, 11, 81–110.
22. N. V. Chandrasekharan, H. Dai, K. L. T. Roos, N. K. Evanson, J. Tomsik, T. S. Elton and D. L. Simmons, COX-3, a cyclooxygenase-1 variant inhibited by acetaminophen and other analgesic/antipyretic drugs: Cloning, structure, and expression, Proc. Natl. Acad. Sci. U. S. A., 2002, 99, 13926–13931.
23. M. J. Thun, Aspirin, NSAIDs, and digestive tract cancers, Cancer Metastasis Rev., 1994, 13, 269–277.
24. R. Singh, R.-P. Cadeddu, J. Fröbel, C. Wilk, I. Bruns, L. Zerbini, T. Prenzel, S. Hartwig, D. Brünnert, T. Schroeder, S. Lehr, R. Haas and A. Czibere, The non-steroidal anti-inflammatory drugs Sulindac sulfide and Diclofenac induce apoptosis and differentiation in human acute myeloid leukemia cells through an AP-1 dependent pathway, Apoptosis, 2011, 16, 889–901.
25. S.-K. Chiou, A. Hodges and N. Hoa, Suppression of growth arrest and DNA damage-inducible 45α expression confers resistance to sulindac and indomethacin-induced gastric mucosal injury, J. Pharmacol. Exp. Ther., 2010, 334, 693–702.
26. M. Aryankalayil, S. T. Palayoor, D. Cerna, M. T. Falduto, S. R. Magnuson and C. N. Coleman, NS-398, ibuprofen, and cyclooxygenase-2 RNA interference produce significantly different gene expression profiles in prostate cancer cells, Mol. Cancer Ther., 2009, 8, 261–273.
27. L. F. Zerbini, A. Czibere, Y. Wang, R. G. Correa, H. Otu, M. Joseph, Y. Takayasu, M. Silver, X. Gu, K. Ruchusatsawat, L. Li, D. Sarkar, J.-R. Zhou, P. B. Fisher and T. A. Libermann, A novel pathway involving melanoma differentiation associated gene-7/interleukin-24 mediates nonsteroidal anti-inflammatory drug-induced apoptosis and growth arrest of cancer cells, Cancer Res., 2006, 66, 11922–11931.
28. S.-K. Chiou and N. Hoa, Up-regulation of GADD45a expression by NSAIDs leads to apoptotic and necrotic colon cancer cell deaths, Apoptosis, 2009, 14, 1341–1351.
29. H.-Y. Jiang, L. Jiang and R. C. Wek, The eukaryotic initiation factor-2 kinase pathway facilitates differential GADD45a expression in response to environmental stress, J. Biol. Chem., 2007, 282, 3755–3765.
30. S. Supiot, R. P. Hill and R. G. Bristow, Nutlin-3 radiosensitizes hypoxic prostate cancer cells independent of p53, Mol. Cancer Ther., 2008, 7, 993–999.
31. A. W. Knight, L. Birrell and R. M. Walmsley, Development and validation of a higher throughput screening approach to genotoxicity testing using the GADD45a-GFP GreenScreen HC assay, J. Biomol. Screening, 2009, 14, 16–30.
32. M. J. O'Neil, The Merck Index: An Encyclopedia of Chemicals, Drugs, and Biologicals, Merck Research Laboratories, Merck & Co., Inc., Whitehouse Station, NJ, 13th edn, 2001.
33. G. Brambilla and A. Martelli, Genotoxicity and carcinogenicity studies of analgesics, anti-inflammatory drugs and antipyretics, Pharmacol. Res., 2009, 60, 1–17.
34. Toxicology Data Network (TOXNET), http://www.toxnet.nlm.nih.gov.
35. The Carcinogenic Potency Project, http://www.potency.berkeley.edu.
36. L. F. Zerbini, R. E. Tamura, R. G. Correa, A. Czibere, J. Cordeiro, M. Bhasin, F. M. Simabuco, Y. Wang, X. Gu, L. Li, D. Sarkar, J.-R. Zhou, P. B. Fisher and T. A. Libermann, Combinatorial effect of non-steroidal anti-inflammatory drugs and NF-κB inhibitors in ovarian cancer therapy, PLoS One, 2011, 6(9), e24285.
37. L. Birrell, P. Cahill, C. Hughes, M. Tate and R. M. Walmsley, GADD45a-GFP GreenScreen HC assay results for the ECVAM recommended lists of genotoxic and non-genotoxic chemicals for assessment of new genotoxicity tests, Mutat. Res., Genet. Toxicol. Environ. Mutagen., 2010, 695, 87–95.
38. M. L. Kropko, J. E. Fitzgerald and F. A. De la Iglesia, Lack of in vivo and in vitro genotoxicity with the nonsteroid, antiinflammatory agent sodium meclofenamate, Teratog., Carcinog. Mutagen., 1984, 4, 329–340.
39. ICH, Guidance on genotoxicity testing and data interpretation for pharmaceuticals intended for human use S2(R1), 2011, 1–2934.
40. P. Schwenger, P. Bellosta, I. Vietor, C. Basilico, E. Y. Skolnik and J. Vilček, Sodium salicylate induces apoptosis via p38 mitogen-activated protein kinase but inhibits tumor necrosis factor-induced c-Jun N-terminal kinase/stress-activated protein kinase activation, Proc. Natl. Acad. Sci. U. S. A., 1997, 94, 2869–2873.
41. J. A. Levy, M. Virolainen and V. Defendi, Human lymphoblastoid lines from lymph node and spleen, Cancer, 1968, 22, 517–524.
42. H. L. Liber and W. G. Thilly, Mutation assay at the thymidine kinase locus in diploid human lymphoblasts, Mutat. Res., Fundam. Mol. Mech. Mutagen., 1982, 94, 467–485.
43. P. Fowler, K. Smith, J. Young, L. Jeffrey, D. Kirkland, S. Pfuhler and P. Carmichael, Reduction of misleading (“false”) positive results in mammalian cell genotoxicity assays. I. Choice of cell type, Mutat. Res., Genet. Toxicol. Environ. Mutagen., 2012, 742, 11–25.

---