Resolution of historically discordant Ames test negative/rodent carcinogenicity positive N-nitrosamines using a sensitive, OECD-aligned design

ABSTRACT

The in vitro bacterial reverse mutation (Ames) test is crucial for evaluating the mutagenicity of pharmaceutical impurities. For N-nitrosamines (NAs) historical data indicated that for certain members of this chemical class, the outcomes of the Ames test did not correlate with their associated rodent carcinogenicity outcomes. This has resulted in negative outcomes in an OECD (Organization for Economic Cooperation and Development)-aligned Ames test alone (standard or enhanced) no longer being considered sufficient by regulatory authorities to assess potential carcinogenic risk of NAs if present as impurities in drug products. Consequently, extensive follow-up in vivo testing can be required to characterize the potential mutagenicity and genotoxic carcinogenicity of NA impurities (i.e. beyond that defined in the ICH M7 guideline for non-NA impurities). We previously demonstrated that the mutagenicity of alkyl-nitrosamines can be detected by the appropriately designed, OECD-aligned Ames test and identified those conditions that contributed most to assay sensitivity. This OECD-aligned Ames test design was used to assess seven NAs, i.e. (methyl(neopentyl)nitrosamine, N-methyl-N-nitroso-2-propanamine, N-nitrosodiisopropylamine, bis(2-methoxyethyl)nitrosoamine, N-nitroso-N-methyl-4-fluoroaniline, dinitrosoethambutol, (R,R)- and mononitrosocaffeidine) that were reported to be negative in historical Ames tests but positive in rodent carcinogenicity studies. All seven of the NAs were demonstrated to be mutagenic in the OECD-aligned Ames test and therefore these compounds should no longer be considered as discordant (false negatives) with respect to the correlation of the Ames test and rodent carcinogenicity. These results confirm the sensitivity of the OECD-aligned Ames test for the detection of NA mutagenicity and provides further support of its pivotal placement within the ICH M7 framework for the assessment of mutagenic impurities in pharmaceuticals to limit potential carcinogenic risk. In addition, we present data for 1-cyclopentyl-4-nitrosopiperazine, that indicates it could serve as a suitable positive control to provide further confidence in the sensitivity of the Ames test for the NA chemical class.

Introduction

N-Nitrosamines (NAs) are a class of chemical compound that have been demonstrated to be potentially mutagenic and carcinogenic in nonclinical safety studies, and certain exemplars have been demonstrated to be potent rodent carcinogens as compared to non-NAs [1,2]. Some NAs are formed endogenously, within the body, while others are present in the environment e.g. in water, air, and soil, or in the diet (food and beverages) and from other sources, e.g. tobacco smoke, and certain consumers and/or industrial products [3]. More recently, certain NAs were observed as trace impurities in some pharmaceutical products [4,5]. Existing regulatory guidelines indicated that additional considerations were required regarding the assessment and control of NAs if present as impurities in pharmaceuticals given the nonclinical data for the NA chemical class [6]. More recently additional regulatory guidance has also been published to clarify the details of these additional considerations to ensure appropriate control of NA impurities in pharmaceutical products [7–10].

As outlined in ICH M7, the Ames test is the pivotal assay for the assessment of DNA reactive (mutagenic) impurities in pharmaceuticals and the outcome of the assay is used to determine whether an impurity is, or is not, a potential mutagenic carcinogen (i.e. an ICH M7 class 2 or class 5 impurity, respectively). However, historically there were some concerns regarding the capacity of the Ames test to predict the carcinogenic potential of NAs [11,12], due to reports of NAs that were non-mutagenic in the Ames test but carcinogenic in rodent studies [13]. These historical studies raised questions as to whether assessment of an NA impurity in the Ames test alone was sufficient to inform on the appropriate control strategy for the said NA impurity in a drug product. As a consequence, the demonstration that a NA impurity is negative (non-mutagenic) in the Ames test alone is no longer considered sufficient by regulatory authorities to classify it as an ICH M7 class 5 impurity. Therefore, in the absence of additional data (e.g. read-across to existing rodent bioassay data on a similar structure), extensive additional follow-up in vivo testing can be required to assess NA impurities when they cannot be removed during manufacture and/or adequately controlled to set limits in drug product defined by the Carcinogenicity Potency Categorisation Approach (CPCA) [14]. Recently, there have been examples of NA impurity assessments requiring extensive ICH S2 R1 characterization, combining Ames test data and mammalian cell mutation data (e.g. from rodent transgenic gene mutation assays—typically Muta™mouse or BigBlue®), to address concerns regarding the potential for mutagenicity mediated rodent carcinogenicity of NA impurities [15]. Whereas drug safety is of foremost importance, this extensive characterization requires significant resource and animal use, which can potentially impact patient access to medicines. This would not be required if the Ames test were considered sufficiently sensitive for the assessment of potential NA mediated mutagenicity and associated rodent carcinogenicity. Indeed, as a ‘cornerstone’ assay in nonclinical testing, it is important for all stakeholders (regulatory authorities, industry, academia, and the public) to have confidence that the Ames test can support robust decision-making.

In our preceding paper, Thomas et al. [16] the historical inconsistencies in the Ames tests ability to predict the carcinogenic potential of NAs in rodents were addressed. Various assay parameters were explored that contribute to NA mutagenic potency in the Ames test, such as the choice of vehicle solvent, the species of liver-S9 mix used for metabolic activation, the bacterial strains employed, the concentrations of the compound tested, and the method of the test (i.e. pre-incubation vs. plate incorporation). Examination of two alkyl-nitrosamines, N-nitrosodimethylamine (NDMA) and N-nitrosodiethylamine (NDEA) identified the key parameters that enhance Ames test sensitivity. The most sensitive conditions for detecting the mutagenicity of NDMA and NDEA were use of the pre-incubation method (with a 30-min incubation period), the choice of water or methanol as vehicle solvent, inclusion of 10% rat or hamster induced liver-S9 mix and inclusion of base-substitution sensitive bacterial strains (Salmonella typhimurium strains TA100 and TA1535, and Escherichia coli strain WP2uvrA(pKM101)).

To further evaluate the performance of the Ames test under these ‘optimal conditions’ [16], the potential mutagenicity of seven NAs reported to be ‘false-negatives’ [13] (i.e. those that were reported to be non-mutagenic in the Ames test but carcinogenic in rodents) was re-evaluated. In many cases the legacy Ames test data were published before the introduction of the OECD (Organization for Economic Cooperation and Development) 471 guideline [17]) and in retrospect, the experimental assay design for these NAs would be considered incomplete in some way (e.g. an insufficient number of bacterial strains were employed, the top concentration used did not meet current guidelines, or a single species liver-S9 was used); as such the results for these legacy studies would actually be considered inconclusive (as opposed to negative) by current standards. Upon re-evaluation, all seven of these NAs were demonstrated to be mutagenic in the OECD-aligned Ames test, data that indicates the test outcome is concordant with reported rodent carcinogenicity outcomes. These results support the position that an appropriately designed Ames test can detect the potential mutagenicity of NAs, re-affirming its position as a pivotal test in the ICH M7 framework for the assessment and control of DNA reactive (mutagenic) impurities in pharmaceuticals to limit potential carcinogenic risk.

Lastly, Ames test data for 1-cyclopentyl-4-nitrosopiperazine is presented, which indicates it is well-suited for use as a second NA positive control in alignment with recently published guidance when testing nitrosamine drug substance-related impurities (NDSRIs) [9,10].

Materials and methods

Compounds and solvents

Methyl(neopentyl)nitrosamine (CAS 31820-22-1), N-methyl-N-nitroso-2-propanamine (CAS 30533-08-5), N-nitrosodiisopropylamine, (CAS 601-77-4), bis(2-methoxyethyl)nitrosoamine (CAS 67856-65-9), N-nitroso-N-methyl-4-fluoroaniline (CAS 937-25-7), dinitrosoethambutol, (R,R)- (CAS 52322-22-2), mononitrosocaffeidine (CAS 145438-96-6) and 1-cyclopentyl-4-nitrosopiperazine (CAS 61379-66-6) were purchased from Enamine (Kiev, Ukraine) or ArZa Bioscience Ltd (Bolton, UK). Ultrapure water or methanol were used as solvent vehicles (Sigma–Aldrich, Haver Hill, UK). Supplementary Table S1 lists the test chemicals, their structure, purity, and sources/batches used. Compound stability in vehicle was also confirmed (data not presented).

Metabolic activation system

Phenobarbital and 5,6-benzoflavone induced Sprague–Dawley rat liver post-mitochondrial fraction (S9) or Aroclor-1254 induced Golden Syrian hamster liver post-mitochondrial fraction (S9) purchased from Molecular Toxicology Incorporated, NC, USA (MolTox™) was used as an exogenous metabolizing system. Sprague–Dawley rats were prepared using the treatment schedule described in Matsushima et al. [18], while Golden Syrian hamsters were prepared as described in ASTM E1687 [19]. The batches of S9 fraction were thawed immediately prior to use and prepared as detailed in Thomas et al. [16] (which includes concentrations of components), with an NADPH generating system (which included NADP and glucose-6-phosphate). The several lots of S9 used for this manuscript contained approximately 36.1 to 40.5 mg/ml of rat S9 protein or 33.6 to 35.3 mg/ml of hamster S9 protein. The final S9 mix contained 10% (v/v) of induced S9-liver fraction only, as a review of the NTP database did not reveal any uniquely mutagenic NAs in the Ames test using 30% (v/v) S9 (data not reported). However, when testing 1-cyclopentyl-4-nitrosopiperazine, 30% (v/v) of S9-liver fraction was also used, as this compound is currently considered to be a suitable second NA positive control in the enhanced Ames test (EAT) [10].

Bacterial strains

Five bacterial strains (S. typhimurium strains TA98, TA100, TA1535, TA1537, and E. coli strain WP2uvrA (pKM101)) were used to identify the mutagenic capacity of the selected NA compounds. Additional details for each bacterial strain, including origin and strain genotype are as described in Thomas et al. [16]. Strains were maintained as frozen stocks with their genotype characteristics having been confirmed as described by Mortelmans and Zeiger [20]. Cultures used in the present investigation were prepared from these bacterial strain stock samples.

Ames test experimental conditions

The selective agar plates were purchased from E&O Laboratories, product number PP2037 (agar volume 24–26 ml). Bacteria were propagated from frozen stocks. The Ames test was conducted following the recommendations of the OECD 471 test guideline [17]. Bacteria were grown in nutrient broth with shaking (150 pm at 37°C) for 10 h until the exponential growth phase was completed and treatments were completed within 3 h from the end of incubation to ensure bacteria were in a stationary phase of growth. Both ‘Ames’ plate incorporation and/or ‘Yahagi’ pre-incubation methods, in the absence and presence of induced rat or hamster liver S9-mix, respectively, were used for mutagenicity assessment.

In the plate incorporation method, 100 µl bacterial suspension, 100 µl test compound or vehicle solvent or positive control, and 500 µl phosphate buffer were mixed directly with 2 ml of histidine, biotin, and tryptophane-containing molten top agar (50°C) and the mixture poured onto the selective agar plates (i.e. without essential amino acids). In the ‘Yahagi’ pre-incubation test, 100 µl bacterial suspension, 50 µl test compound or vehicle solvent or positive control and 500 µl induced rat or hamster liver S9-mix were placed into a shaking incubator for 30 min at 37°C (the pre-incubation phase), before 2 ml of histidine, biotin, and tryptophane-containing molten top agar (50°C) were added and the mixture poured onto selective agar plates. All plates were incubated for 3 days at 37°C prior to colony counting. Five to seven concentration levels were evaluated per study arm and test concentrations ranged from 5 to 5000 μg/plate. The number of replicate plates used, positive controls assessed (in the presence and absence of S9-mix—including concentration and positive control vehicle), data acquisition and criteria for toxicity were as described in Thomas et al. [16]. For evaluation of treatment-related effects, a fold increase was defined as positive for mutagenicity, if a biologically relevant increase in the mean number of revertant colonies above a threshold of 2-fold (TA98, TA100, WP2 uvrA (pKM101)) or 3-fold (TA1535, TA1537) as compared to the concurrent negative controls was observed. Table 1 lists the basic elements of study composition per test article.

Results

Details for each Ames test, including vehicle and positive controls, concentrations, mean revertant colony counts per concentration, etc., are presented in the common technical document tables within the Supplementary Tables S2–S12. The negative ‘solvent-vehicle only’ treatments resulted in strain-specific revertant colony frequencies consistent with the laboratory historical control values and the standard ‘non-nitrosamine’ positive controls induced clear unequivocal increases in numbers of revertant colonies and thus confirmed the validity of assays conducted (Supplementary Tables S13 and S14).

Nitrosamine Ames test results using methanol as solvent

A summary of outcomes (positive or negative) per strain and per test condition (pre-incubation with rat or hamster-induced liver S9 activation or plate incorporation without S9 activation) for each NA, using methanol as a solvent is presented in Table 2.

Methyl(neopentyl)nitrosamine (CAS 31820-22-1)

Methyl(neopentyl)nitrosamine (NMNA) was mutagenic in the Ames test when using methanol as a solvent (data presented in Supplementary Table S2 and Supplementary Figs. S1 and S2). In the pre-incubation arm, NMNA was mutagenic in Salmonella strains TA100 and TA1535 at concentrations of 500 and 1500 µg/plate in the presence of 10% hamster S9-mix but was negative at higher concentrations where bacterial toxicity was observed (NMNA was tested up to the limit concentration of 5000 μg/plate). NMNA was negative in the other bacterial strains and in all strains when using 10% induced liver rat S9-mix. Interestingly, a positive NMNA response was first observed in E. coli WP2uvrA (pKM101) in the plate incorporation assay in the absence of S9-mix, but only at the limit concentration tested (5000 μg/plate). As NAs generally require metabolic activation to exert their mutagenic effect [1,21,22] this result was considered atypical and additional tests were conducted for verification. Since the positive response could not be reproduced in two additional independent tests in E. coli WP2uvrA (pKM101) NMNA was considered negative overall in this particular arm of the study. As such it was concluded NMNA is not a direct acting mutagen and requires metabolic activation (using hamster S9) to elicit mutagenicity.

Bis(2-methoxyethyl)nitrosoamine (CAS 67856-65-9)

Bis(2-methoxyethyl)nitrosoamine was mutagenic when using methanol as a solvent (data presented in Supplementary Table S3 and Supplementary Fig. S3). Bis(2-methoxyethyl)nitrosoamine was mutagenic in the pre-incubation method with 10% hamster S9-mix in Salmonella strain TA100 at concentrations 2500 µg/plate and above, and in TA1535 at all concentrations tested (50–5000 µg/plate). In contrast, bis(2-methoxyethyl)nitrosoamine was negative in all bacterial strains when 10% rat S9-mix was used and in a plate incorporation test in the absence of S9-mix.

N-Nitroso-N-methyl-4-fluoroaniline (CAS 937-25-7)

N-Nitroso-N-methyl-4-fluoroaniline was mutagenic when using methanol as a solvent (data presented in Supplementary Table S4 and Supplementary Fig. S4). A positive response was observed in the pre-incubation arm with 10% rat S9-mix in Salmonella TA98 at concentrations of 50 to 500 µg/plate and in E. coli WP2uvrA (pKM101) at concentrations of 150 and 500 µg/plate, but not at higher concentrations where bacterial toxicity was observed. In the presence of 10% hamster S9-mix, N-Nitroso-N-methyl-4-fluoroaniline was mutagenic in TA98 at a single concentration only (500 µg/plate) and was toxic at concentrations ≥ 1500 µg/plate in all bacterial strains. N-Nitroso-N-methyl-4-fluoroaniline was negative in the plate incorporation arm without S9-mix.

Dinitrosoethambutol, (R,R)-(CAS 52322-22-2)

Dinitrosoethambutol was mutagenic when using methanol as a solvent (data presented in Supplementary Table S5 and Supplementary Fig. S5). Increased mutagenicity was observed only in Salmonella TA1535 in a pre-incubation assay with 10% hamster S9-mix and in the plate incorporation assay in the absence of S9-mix, both at the limit concentration (5000 µg/plate). Dinitrosoethambutol was negative in the pre-incubation arm with 10% rat S9-mix.

N-Methyl-N-nitroso-2-propanamine (CAS 30533-08-5)

N-Methyl-N-nitroso-2-propanamine was mutagenic when using methanol as a solvent (data presented in Supplementary Table S6 and Supplementary Fig. S6). In the pre-incubation arm in the presence of 10% hamster S9-mix N-Methyl-N-nitroso-2-propanamine induced increased revertant colonies above the threshold for a positive response in Salmonella TA100 (≥ 1500 µg/plate) and in TA1535 (≥ 500 µg/plate). N-Methyl-N-nitroso-2-propanamine was negative in the pre-incubation arm with 10% rat S9-mix and plate incorporation arm in the absence of S9-mix.

N-Nitrosodiisopropylamine (CAS 601-77-4)

N-Nitrosodiisopropylamine (NDIPA) was not mutagenic when using methanol as a solvent and tested up to 5000 μg/plate, the maximum concentration in accordance with current guidelines (data presented in Supplementary Table S7 and Supplementary Fig. S7).

Mononitrosocaffeidine (CAS 145438-96-6)

Mononitrosocaffeidine was not mutagenic when using methanol as a solvent and tested up to 5000 μg/plate, the maximum concentration in accordance with current guidelines (data presented in Supplementary Table S8 and Supplementary Fig. S8).

1-Cyclopentyl-4-nitrosopiperazine (CAS 61379-66-6)

1-Cyclopentyl-4-nitrosopiperazine (CPNP) was mutagenic when using methanol as a solvent (data presented in Supplementary Table S9 and Supplementary Figs. S9 and S10). Increased mutagenicity was observed in Salmonella TA100 and TA1535 in the pre-incubation assay with rat and hamster liver S9-mix (10 or 30%) and E. coli strain WP2uvrA (pKM101) with hamster liver S9-mix (10 or 30%) or rat 30% liver S9-mix (but not rat 10% S9-mix). Indeed, highly potent responses were observed in Salmonella TA1535 (up to ≥ 200-fold increases compared to solvent control). No mutagenic response was seen in the absence of liver S9-metabolic activation, or in strains TA98 and TA1537 under any metabolic condition.

Nitrosamine Ames test results when water was used as a solvent

As NDIPA and mononitrosocaffeidine were negative when tested in methanol additional studies were undertaken (in a five strain Ames test) using water as a solvent. Additional studies were also undertaken with dinitrosoethambutol, (R,R)- (but only in Salmonella TA1535) using water as a solvent. A summary of results is presented in Table 3.

N-Nitrosodiisopropylamine (CAS 601-77-4)

N-Nitrosodiisopropylamine (NDIPA) was mutagenic when water was used as a solvent (data presented in Supplementary Table S10 and Supplementary Fig. S11). Positive responses were observed in Salmonella TA1535 (at 500–2500 µg/plate) in the pre-incubation assay with 10% hamster S9-mix. NDIPA also showed increased mutagenicity (> 2× fold) in Salmonella TA100 (≥ 500 µg/plate), however, when the test was repeated the revertant frequency (maximum 1.9-fold) did not exceed the protocol threshold for a positive response. NDIPA was negative when tested in water with 10% rat induced liver S9-mix.

Dinitrosoethambutol, (R,R)- (CAS 52322-22-2)

Dinitrosoethambutol, (R,R)- was mutagenic when water was used as a solvent (data presented in Supplementary Table S11 and Supplementary Fig. S12) and induced reproducible increases in revertant colonies in Salmonella TA1535 using the pre-incubation assay with 10% rat S9-mix (@ 5000 µg/plate in Test 1 and ≥ 3750 µg/plate in Test 2) and 10% hamster S9-mix (≥ 2500 µg/plate in Tests 1 and 2). An equivocal result was observed in the plate incorporation assay (-S9-mix), i.e. there was a mutagenic response at the limit dose of 5000 µg/plate (Test 1) but this was not reproducible (Test 2).

Mononitrosocaffeidine (CAS 145438-96-6)

Mononitrosocaffeidine was mutagenic when water was used as a solvent (data presented in Supplementary Table S12 and Supplementary Fig. S13) with increases in revertant colonies in Salmonella TA98 using the pre-incubation assay with 10% rat or hamster S9-mix and in the absence of S9-mix when tested up to 5000 μg/plate, the maximum concentration in accordance with current guidelines. However, a different batch of mononitrosocaffeidine was used with methanol as a vehicle (see above), further work to explore batch differences is planned in collaboration with another laboratory (and will be the subject of a separate manuscript).

Discussion

The principal aim of this study was to assess the mutagenicity of NAs that have historically shown discordant results between the Ames test and rodent carcinogenicity studies, using an enhanced study design described by Thomas et al. [16],. The Ames test is pivotal for evaluating the mutagenicity of pharmaceutical impurities, and its reliability is essential for predicting potential mutagenicity mediated carcinogenesis and enabling decision-making by key stakeholders. However, for NAs, the sensitivity of the Ames test has been questioned due to historical data suggesting it did not accurately predict the carcinogenic potential of various NAs (see Rao et al. [11], Andrews et al. [12]).

In our preceding paper [16], we qualitatively and quantitatively showed the mutagenicity of alkyl-nitrosamines can be readily detected in the Ames test and established, via benchmark dose modelling, the key assay parameters that contributed most to mutagenic potency. We identified the pre-incubation method (compared with plate incorporation), inclusion of 10% rat and hamster induced-liver S9-mix and S. typhimurium TA1535 and TA100 and E. coli WP2uvrA (pKM101) bacterial strains as key parameters. Solvent choice was not deemed to be a significant discriminatory factor, with water and methanol equally appropriate.

In the current study, several NAs with reported discordant bacterial mutagenicity and rodent carcinogenicity data were investigated (Table 1). The results indicated that these NAs were mutagenic in the OECD-aligned Ames test and should not be considered ‘false negatives’ with respect to the correlation with rodent carcinogenicity. Moreover, our study confirms the OECD-aligned Ames test is a highly sensitive test for the assessment of NA-mutagenicity.

Other notable outcomes were that one of the NAs tested, dinitrosoethambutol (R,R)- showed mutagenic capacity in the absence of S9-mix (when methanol or water was used as a solvent vehicle). In addition, mononitrosocaffeidine was mutagenic in the presence of induced rat or hamster liver S9-mix and in the absence of S9-mix. For the remaining studies, the use of induced hamster liver S9-mix was found to be a superior exogenous metabolic activation system compared to induced rat liver S9-mix, identifying 7/7 NAs versus 3/7, respectively. It is noted that mononitrosocaffeidine and N-nitrosodiisopropylamine also required water as a solvent to obtain a positive response, whilst dinitrosoethambutol, (R,R)- required water as a solvent to obtain a positive response specifically with induced rat liver S9-mix. There were no rat S9 unique mutagenic NAs, however, five out of seven NAs (with methanol as solvent) and three out of three NAs (with water as solvent) were mutagenic with hamster S9. Four of these compounds—(methyl(neopentyl)nitrosamine, bis(2-methoxyethyl)nitrosoamine, N-methyl-N-nitroso-2-propanamine and N-nitrosodiisopropylamine), were considered uniquely positive with hamster liver S9, and only NDIPA and mononitrosocaffeidine were uniquely positive in water.

N-Nitroso-N-methyl-4-fluoroaniline was previously reported negative in Salmonella TA1535 when tested up to 1000 μg/plate in both plate incorporation and pre-incubation assays, using induced rat or hamster liver S9-mix [23]. In the current studies, a negative response in TA1535 was also observed, however, N-nitroso-N-methyl-4-fluoroaniline was positive in E. coli WP2uvrA (pKM101) in the presence of rat-induced liver S9-mix, and more interestingly in TA98 with rat and hamster liver S9-mix. Salmonella TA98 is generally considered less sensitive towards NA-induced mutagenicity [16,24], because NAs typically cause base-substitution mutations due to alkylation and the TA98 strain requires frameshift mutations for reversion from auxotrophy to prototrophy [25,26]. As such, the mutagenic profile of N-nitroso-N-methyl-4-fluoroaniline seems somewhat atypical relative to other NAs. Moreover, N-nitroso-N-methyl-4-fluoroaniline is a potent rodent carcinogen (Gold TD₅₀ of 0.255 mg/kg/day) in male F344 rats (relative to non-NAs) and specifically induces oesophageal tumours, whereas the liver is reported to remain tumour free [27]. This may suggest that in vivo the rat oesophagus, but not liver has the required metabolic capacity to activate N-nitroso-N-methyl-4-fluoroaniline to mutagenic intermediates that are potentially carcinogenic in rodents. This may be the result of differential CYP (cytochrome p450) expression or specific ADME (Absorption, distribution, metabolism, and excretion) properties, e.g. site of contact exposure via drinking water, or a combination of the two. Nevertheless, the positive outcome observed in the current OECD-aligned Ames test represents an outcome i.e. predictive of N-methyl-4-fluoroaniline rodent carcinogenicity, whereas the positive result in TA98, a frameshift sensitive bacterial strain, may suggest an alternate mutagenic mechanism could be at play in the rat oesophagus.

Legacy methyl(neopentyl)nitrosamine (NMNA) Ames data, mostly in Salmonella TA1535 with induced rat S9-mix, was negative for mutagenicity [11,12,23,28,29]. One study, however, did report a positive response in TA100 using Aroclor-induced rat liver S9 [28]. In male F344 rats, NMNA induced 100% tumour incidence in the oesophagus and mortality by 45 weeks of treatment, whereas the liver was reported to be tumour free [30]. In the current Ames study, NMNA was positive in TA100 and TA1535 at concentrations of 500 and 1500 µg/plate in the presence of 10% hamster liver S9-mix, but not with 10% rat S9-mix. As such, the historically negative data may simply reflect the limited assessment performed with rat liver S9.

Likewise, the legacy N-methyl-N-nitroso-2-propanamine Ames test data is inconclusive since a single strain (TA1535) was tested and only up to a limit concentration of 2000 µg/plate (albeit in pre-incubation and plate incorporation assays with and without induced rat liver S9-mix) [11]. N-methyl-N-nitroso-2-propanamine has been described as a weak rodent carcinogen (no indication of potency was provided) [11,31] and our current Ames test study confirms N-methyl-N-nitroso-2-propanamine as negative when tested with rat induced liver S9-mix in all bacterial strains, including TA1535. In contrast, N-methyl-N-nitroso-2-propanamine was positive in Salmonella strains TA100 and TA1535 when hamster induced liver S9 was used, and again, this result leads to the conclusion that the negative historical data may reflect the limited (i.e. rat S9 only) bacterial mutagenicity test design in place at the time of the assessment.

In the current study, bis(2-methoxyethyl)nitrosoamine was uniquely positive in the Ames Test with induced hamster liver S9-mix, inducing mutagenicity in the base-pair substitution sensitive strains, Salmonella TA1535 and TA100. Legacy bis(2-methoxyethyl)nitrosoamine data was negative in Salmonella TA1535 when tested up to 2000 µg/plate, in both pre-incubation and plate incorporation assays, with and without rat-induced liver S9-mix [11], whereas Lee and Guttenplan [28] observed a positive response in Salmonella TA100 under similar testing conditions. Bis(2-methoxyethyl)nitrosoamine was reported to induce 100% tumours in Sprague–Dawley rats, with animals developing both hepatocellular carcinomas and nasal tumours [32]. The current Ames test result (positive) therefore correlates with the reported rodent carcinogenicity outcomes for bis(2-methoxyethyl)nitrosoamine.

NDIPA is another weak rodent carcinogen, associated with the induction of nasal tumours in Sprague–Dawley rats [33]. At higher doses in Sprague–Dawley and Berlin Drucksey IX Rats NDIPA has also been shown to cause hepatocellular carcinomas [31,33]. Legacy Ames data is conflicting: NDIPA was reported negative in Salmonella TA1535 when tested up to 2000 µg/plate (pre-incubation and plate incorporation assays with and without induced rat liver S9-mix) [11] but positive in TA100 [28] when tested in a pre-incubation assay with rat S9. In the current study, NDIPA was negative when tested in methanol but positive in water, but only in bacterial strain Salmonella TA1535 when using hamster induced liver S9-mix (although the result in TA100 was equivocal). NDIPA appears to be a weak mutagen in the Ames test and this may be the result of steric hindrance associated with its chemical structure; indeed, NDIPA is classified category 5 by the CPCA framework as is not predicted to be metabolically activated via α-hydroxylation [34]. Nevertheless, albeit weak, the positive Ames test result in the current study preserves the correlation between mutagenicity and rodent carcinogenicity for NDIPA.

Investigations into the mutagenicity of mononitrosocaffeidine yielded interesting results, this NA was only mutagenic when water was used as a solvent. The mutagenicity of mononitrosocaffeidine is also atypical to other NAs, being positive in TA98 only, in the presence of induced 10% rat or hamster S9-mix and, surprisingly, in the absence of S9-mix. In addition, there were potential batch differences, and this is the subject of ongoing work. Mononitrosocaffeidine-treated rats show localized tumours in the nasal cavity (neuroepitheliomas of the olfactory epithelium and squamous cell carcinoma of the nasal cavity) [35] but reported as negative in the Ames test using Salmonella strains TA100, TA1535, TA102, TA98, and TA1537, both in the presence and absence of S9-mix, when tested up to 1690 nmol (333 µg) per plate [36]. The present studies were tested at higher concentrations and included hamster S9, which may go some way to explain the inconsistency. Moreover, given its solvent specific profile in the Ames test, additional studies using the in vitro mammalian cell mutation assay are warranted and this will be the subject of a future manuscript.

Similarly, dinitrosoethambutol, (R,R)- was reported negative in Salmonella TA100 and TA98 in a pre-incubation assay in the absence or presence of rat S9-mix [37] but is a rodent carcinogen in BALB/C mice inducing tumours in the lung in male and female animals [38]. In the current study, dinitrosoethambutol, (R,R)- was mutagenic in Salmonella TA1535, in the absence of S9-mix or when pre-incubated with hamster S9-mix with methanol as the solvent vehicle. Interestingly, when water was used as a solvent, dinitrosoethambutol, (R,R)- was positive in TA1535 with rat-induced liver S9-mix, but in the absence of S9, the response was equivocal (i.e. non-reproducible). Nevertheless, here again, the dinitrosoethambutol, (R,R)- Ames test result correlates with the reported rodent carcinogenicity outcome.

Lastly, we investigated the rodent carcinogen 1-Cyclopentyl-4-nitrosopiperazine (CPNP) [39], a NA which induces olfactory and liver tumours in rats. CPNP was identified as a potent mutagenic NA in the Ames test and was positive in strains TA100, TA1535, and WP2uvrA (pKM101) with both rat and hamster 30% S9-mix and in TA100 and TA1535 with 10% rat S9. Given the fold increases observed in mutagenic response (up to ~80-fold in rat and ~200-fold in hamster) in the two base-substitution strains, TA100 and TA1535 (with both rat and hamster liver S9) we conclude CPNP would make a robust second positive control in the Ames test when assessing NAs. This would align with the requirements of recently published EMA (European Medicines Agency) regulatory guidance on the EAT [10].

To conclude, our results demonstrate the sensitivity of the Ames test to robustly predict the potential rodent carcinogenicity of NAs. Additionally, this confirms the key study design parameters associated with NA mutagenic potency as established in our previous work [16] and aligns with published Ames data on nitrosamines generated by the FDA (Food and Drug Administration) [40]. However, we note that for two of these difficult-to-test nitrosamines, NDIPA and mononitrosocaffeidine, positive results were only observed when water was used as a solvent. Whether these results represent the exception to the rule, they clearly demonstrate water to be the more appropriate solvent in these two cases. Water, however, is not always the most appropriate solvent, e.g. with more polar NDSRIs, and discretion is warranted when trying to prove a negative. As such, it is interesting to note the latest FDA guidance [9] recommends water, methanol or DMSO (dimethyl sulfoxide) as solvent (the latter using minimal practical volumes). Any concerns associated with primary solvent choice would also be addressed by supporting in vitro metabolism data, which will highlight the metabolic competency of the Ames test and is now recommended in the latest FDA guidance [9]. We show CPNP represents an ideal positive-control candidate when testing NAs using the Ames test. if required in addition to the standard positive controls that are used routinely. Strong mutagenic responses were observed using Salmonella strains TA1535 (77 or 212-fold) and TA100 (9 or 46-fold) in the presence of induced liver S9-mix (10% rat or hamster, respectively) and this should be sufficient to confirm the performance of the Ames test to detect NA-induced mutagenicity. CPNP was also positive in E. coli strain WP2uvrA (pKM101), but potency was significantly reduced, and therefore we would not recommend including this strain nor the two negative strains (TA98 and TA1537) when using CPNP as a positive control. Moreover, the need for a nitrosamine specific positive control is, in of itself, questionable, particularly given recent experience with the performance of the Ames test with nitrosamines reported by our laboratory and by others. It is likely that such an adjustment was initially requested, at least in part, because of the historically discordant nitrosamines, and in our view a specific nitrosamine positive control is not warranted.

Supplementary data

Supplementary data are available at Mutagenesis Online.

Acknowledgements

None.

Conflict of interest statement

None declared.

Data availability

The data underlying this article are available in the article and in its online supplementary material.

References