Adjusting the Preincubation Conditions to Enhance the Ames Test for Detecting the Mutagenicity of N‐Nitrosamines

ABSTRACT

Conducting S9‐mediated preincubations at slightly acid pHs (e.g., pH 5.0 to pH 6.5) has been reported to enhance the mutagenicity of small‐molecule nitrosamines in the Ames test. In this study, we have evaluated the effect of adjusting the preincubation mix used in the Enhanced Ames Test (EAT) from pH 7.4 to pH 6.0 and supplying the activation mix with preformed reducing equivalents (NADPH and NADH). Abbreviated EAT assays were conducted on five small‐molecule N‐nitrosamines and 10 nitrosamine drug substance‐related impurities (NDSRIs) with tester strains TA1535 and WP2 uvrA (pKM101) and using S9 mixes containing 30% hamster liver S9. Testing on small‐molecule nitrosamines found that the pH 6.0 preincubation mix enhanced the mutagenicity of N‐nitroso‐dimethylamine and N‐nitroso‐diethylamine but not N‐nitroso‐diphenylamine, N‐nitroso‐methyl‐4‐aminobutyric acid or N‐(2,2‐diethoxyethyl)‐2,2‐diethoxy‐N‐nitrosoethanamine. Of the 10 NDSRIs that were tested, N‐nitroso‐phenylephrine was consistently positive with the pH 6.0 preincubation mix, while it was generally negative with pH 7.4 preincubation mixes. The mutagenicity of the nine other NDSRIs that were tested was not changed by the slightly acid preincubation mix. The results indicate that performing preincubation reactions under slightly acid conditions increases the mutagenicity of some small‐molecule nitrosamines, and in one case, produced a positive mutagenic response with an NDSRI that was otherwise negative in the EAT.

1. Introduction

The carcinogenicity of N‐nitrosamines (or nitrosamines), especially the most potent carcinogenic nitrosamines, is widely viewed as due to their mutagenicity. Therefore, an accurate assessment of their mutagenicity is important for evaluating the human health risks posed by nitrosamine drug impurities. The Enhanced Ames Test (EAT) was introduced by health authorities for the purpose of detecting the mutagenicity of N‐nitrosamine drug impurities with the greatest possible sensitivity while retaining reasonable specificity (EMA ²⁰²⁴; FDA ^2024a; Health Canada ²⁰²⁴). The EAT employs several modifications of the standard preincubation version of the Ames test (OECD ²⁰²⁰) that have been reported in previous studies to increase the mutagenicity of nitrosamines (see discussion in Heflich et al. ²⁰²⁴). Besides using the preincubation version of the test, these modifications include S9 mixes containing 30% liver S9 from both hamsters and rats pretreated with enzyme inducers, and the use of five tester strains, Salmonella typhimurium strains TA1535, TA100, TA98, and TA1537, and E. coli tester strain WP2 urvA (pKM101). A recent ring trial conducted as a public‐private partnership organized by the Health and Environmental Sciences Institute Genetic Toxicology Technical Committee (HESI/GTTC), indicates that the EAT is highly successful at identifying carcinogenic nitrosamines (Bercu et al. 2025).

Regardless of the predictive performance reported by Bercu et al. (2025), uncertainty remains regarding the ability of the EAT to classify nitrosamines as to their carcinogenic potential. Because the HESI/GTTC trial used nitrosamines with rodent cancer data, only small‐molecule nitrosamines (and a few drug‐like N‐nitroso compounds) were tested. An important subcategory of nitrosamine drug impurities, nitrosamine drug substance‐related impurities (NDSRIs), was not included in the trial since little or no cancer data are available for these compounds. While having the N‐nitroso functional group, NDSRIs are derived from the drug substance they are associated with and generally are larger and more complex than the small‐molecule nitrosamines for which carcinogenicity data are available (FDA ²⁰²³; FDA ^2024b; FDA ^2024a).

Heflich et al. (2024) showed that modifications to the Ames test that increase the sensitivity of the assay for detecting the mutagenicity of small‐molecule nitrosamines also tend to increase the sensitivity of the assay for detecting the mutagenicity of NDSRIs. This observation suggests that conclusions about the sensitivity of the EAT for detecting the mutagenicity of carcinogenic small‐molecule nitrosamines likely are true for NDSRIs. However, given the concerns that remain regarding the ability of the Ames test for predicting the carcinogenicity of nitrosamines, the U.S. Food and Drug Administration does not rely solely on a negative EAT to conclude that NDSRIs pose a negligible cancer risk (FDA ^2024a).

As described in Heflich et al. (2024), there are additional modifications to the Ames test that were not included in the development of the conditions used in the EAT and that potentially can increase the sensitivity of the Ames test for detecting the mutagenicity of nitrosamines. One such modification is to employ a slightly acidic pH in the preincubation mix (Guttenplan 1979, 1980; Negishi and Hayatsu ¹⁹⁸⁰). At least two variants of this approach have been described that involve adjusting the preincubation to pHs between 5.0 and 7.0. In the method described by Guttenplan (1980), the preincubation was conducted in two phases: first NADPH was generated from NADP by incubation with mouse liver S9 at a neutral pH (pH of 6.9–7.4); then the pH was adjusted to pH 6.5, the test compound added, and the incubation continued for up to 60 min. The other approach involves conducting the entire preincubation at a pH of 6.0 or 5.0, but supplying the activation mix with preformed reducing equivalents (NADPH and NADH) (Negishi and Hayatsu 1980).

In the present study, we have combined methods used for the EAT (30 min preincubations using 30% hamster liver S9 and tester strains TA1535 and WP2 uvrA [pKM101]) with those of the technically simpler method of Negishi and Hayatsu (a preincubation mix pH of 6.0 and NADPH and NADH cofactors). We used this assay to retest five of the small‐molecule nitrosamines and 10 NDSRIs we previously reported on, including two small‐molecule nitrosamines and seven NDSRIs that had previously tested negative using the EAT (Heflich et al. 2024).

2. Materials and Methods

2.1. Materials

2.1.1. N‐Nitrosamine Test Substances

The compounds evaluated for mutagenicity in this study (and their CAS numbers) are five small‐molecule N‐nitrosamines, N‐nitroso‐dimethylamine (NDMA) (CAS 62‐75‐9) from Chem Service Inc. (West Chester, PA) N‐nitroso‐diethylamine (NDEA) (CAS 55–18‐5) from TCI (Portland, OR), N‐nitroso‐methyl‐4‐aminobutyric acid (CAS 61445‐55‐4) from Chem Space (Monmouth Junction, NJ), N‐nitroso‐diphenylamine (CAS 86‐30‐6) from Toronto Research Chemical (Vaughan, Ontario, Canada), and N‐(2,2‐diethoxyethyl)‐2,2‐diethoxy‐N‐nitrosoethanamine (or N,N‐bis(2,2‐diethoxyethyl) nitrous amide) (CAS 67856‐67‐1) from Clearsynth Canada Inc. (Brampton, Ontario, Canada), and ten NDSRIs, N‐nitroso‐phenylephrine (CAS 78658‐64‐7), N‐nitroso‐varenicline (CAS 2755871‐02‐2), N‐nitroso‐diclofenac (CAS 66505‐80‐4), N‐nitroso‐sertraline (CAS 3006789‐98‐3), N‐nitroso fluoxetine (CAS 150494‐06‐7), N‐nitroso‐paroxetine (CAS 2361294‐43‐9) all from Clearsynth Canada Inc., N‐nitroso‐desvaleryl‐valsartan (CAS 2254485‐68‐0) from LKT Lab (St. Paul, MN), N‐nitroso‐bumetanide (CAS 2490432‐02‐3) from Clearsynth, N‐nitroso‐desvaleryl‐valsartan methyl ester (CAS 137862‐54‐4) from LKT Lab, and N‐nitroso‐dabigatran etexilate (CAS 2892260‐29‐4) from Clearsynth Canada Inc. Additionally, information on the sources of these test substances and their properties can be found in Heflich et al. (2024).

2.1.2. Tester Strains and S9

The tester strains used were the Salmonella typhimurium histidine auxotroph TA1535 and the E. coli tryptophan auxotroph, WP2 uvrA (pKM101). The tester strains are reverted by base pair substitution at G:C (TA1535) and A:T [WP2 uvrA (pKM101)]. The tester strains were obtained from Molecular Toxicology Inc. (MOLTOX, Boone, NC) and stored either on paper disks at 4°C or as frozen stocks at −80°C. S9 prepared from the liver of male Syrian hamsters pretreated with a mixture of phenobarbital and β‐naphthoflavone (PB/BNF) was obtained from Molecular Toxicology, stored at −80°C and thawed on ice just before use.

2.1.3. Reagents and Medium

D‐Glucose 6‐phosphate sodium salt (Sigma, St. Louis, MO), NADP monosodium salt (Millipore, Burlington, MA), NADPH tetrasodium salt (Millipore), NADH (Roche, Basal, Switzerland), and Adenosine 5′‐triphosphate disodium salt hydrate (ATP) were purchased from Sigma‐Aldrich (St. Louis, MO).

Acetone and molecular grade reagent water (used for both test agent solutions and S9 mix preparation) were supplied by Thermo Fisher (Waltham, MA); dimethyl sulfoxide (DMSO) and agar were obtained from Sigma‐Aldrich.

Nutrient Broth No. 2 was from Oxoid (Basingstoke, UK).

2.2. Ames Testing Methods

Three different assay protocols, involving three different preincubation mixes, were used in this study. One, the Abbreviated Enhanced Ames Test or AEAT, was derived directly from the methods recommended for the EAT (EMA ²⁰²⁴; FDA ^2024a; Health Canada ²⁰²⁴), except that only 30% hamster liver S9 was used for the S9 mix and only tester strains TA1535 and WP2 uvrA (pKM101) were used for the assays. These conditions were sufficient to produce positive responses with all the small‐molecule nitrosamines and NDSRIs that were tested in our previous studies (Li et al. 2023; Heflich et al. ²⁰²⁴). The conditions of the AEAT also represent a practical compromise for testing the modifications to the assay described by Negishi and Hayatsu (1980).

The second testing protocol, referred to as NH7, kept the pH of the preincubation mix at the same pH as the EAT and AEAT (pH 7.4) but substituted NADPH and NADH as cofactors as described by Negishi and Hayatsu (1980). The third protocol, referred to as NH6, used the same preincubation mix as in NH7, but adjusted its pH to 6.0. By using these three preincubation mixes, we were able to test the effect of changing the cofactors and the cofactors plus the pH of the preincubation mix on the mutation responses produced by the EAT.

2.2.1. Solvents Used for Test Substances and Controls

Organic solvents used to dissolve test substances were delivered at 25 μL per 700 μL preincubation reaction and test substances dissolved in water were delivered at 100 μL per 700 μL preincubation reaction. NDMA and NDEA were dissolved in water, while DMSO was used as the solvent for N‐nitroso‐varenicline. Acetone served as the solvent for the remaining test chemicals.

2.2.2. Control Assays

Vehicle controls were included in each assay. In most assays, negative controls without the organic solvent and a positive control that also demonstrated the effect of the acid pH preincubation mix on nitrosamine mutagenicity (termed a ‘pH control/positive control’ or pHC/PC) were included. The pHC/PCs generally contained 300 μg/assay of NDMA for TA1535 and 15 μg/assay for WP2 uvrA (pKM101). These NDMA concentrations were chosen to reliably demonstrate both an S9‐mediated increase in mutagenicity and the effect of the NH6 preincubation conditions on the mutagenicity response. In some cases, other concentrations of NDMA were used (see Supporting Information).

2.2.3. Test Substance Concentrations

The concentrations of the small‐molecule nitrosamines and NDSRIs assayed in this study were based on the mutagenicity data generated in Heflich et al. (2024). Compounds testing negative were assayed up to either 5 mg/plate or the lowest concentration producing a precipitate or toxicity in the assay. For compounds that tested positive in our previous study, concentrations were chosen producing negative responses up to concentrations that yielded moderate mutagenic responses. Repeat assays were sometimes necessary to achieve this range of responses or to confirm small effects of the preincubation conditions on the mutational responses.

2.2.4. Preincubation Reactions

The compositions of the three S9 mixes used for the AEAT, NH7, and NH6 assays are described in Table 1. When an organic solvent was used to dissolve the test substance, 0.5 mL of the S9 mix was combined in a 2.0‐mL polypropylene blunted‐bottom microcentrifuge tube with 75 μL of water (typically included with the S9 mix to reduce pipetting steps), 0.1 mL of the tester stain in late logarithmic growth (~14 h of growth in Oxoid nutrient medium) and 25 μL of the solvent or the solvent containing the test chemical. When water was used to dissolve the test substance, 0.5 mL S9 mix was combined in a 2.0 mL tube with 0.1 mL of the tester strain culture and 0.1 mL of water or water containing the test chemical. The combination of S9 mix with the additional components resulted in what we termed preincubations mixes that all had a final volume of 0.7 mL. The preincubation mixes were then incubated at 300 to 350 rpm at 37°C for 30 min using a FINEPCR shaker‐incubator (Gyeonggi‐do, Korea; model confide‐S202H). All assays were conducted in triplicate. The pHs of the different preincubation mixes were checked using pH‐paper following the 30‐min preincubation, just before plating as described in Section 2.2.5.

TABLE 1.

Concentrations of components in the S9 mixes used to evaluate bacterial mutagenicity.

Abbreviated enhanced Ames test (AEAT) S9 Mix					Negeshi and Hayatsu (NH) S9 Mixes
Component	Stock solution concentration	Volume stock solution per mL of S9 mix	Final concentration	Component	Stock solution concentration	Volume stock solution per mL of S9 mix	Final concentration
Phosphate buffer (pH 7.4)	0.2 M	500 μL	100 mM	Phosphate buffer (pH 7.4 for NH7 or 6.0 for NH6)	0.2 M	500 μL	100 mM
β‐Nicotinamide‐adenine dinucleotide phosphate (NADP)	0.1 M	40 μL	4 mM	β‐Nicotinamide‐adenine dinucleotide phosphate tetrasodium salt (NADPH)	0.1 M	40 μL	4 mM
				β‐Nicotinamide adenine dinucleotide, reduced disodium salt hydrate, (NADH)	0.1 M	40 μL	4 mM
				Adenosine 5′‐triphosphate disodium salt hydrate (ATP)	0.09 M	5 μL	5 mM
Glucose‐6‐phosphate	1 M	5 μL	5 mM	Glucose‐6‐phosphate	1 M	5 μL	5 mM
Potassium chloride a	1.65 M	20 μL	33 mM	Potassium chloride a	1.65 M	20 μL	33 mM
Magnesium chloride a	0.4 M		8 mM	Magnesium chloride a	0.4 M		8 mM
Hamster S9 homogenate	33–39 mg protein/mL	300 μL	30%	Hamster S9 homogenate	33‐39 mg protein/mL	300 μL	30%

Note: S9 mix components were added as concentrated stock solutions (stored at room temperature, 4°C, −30°C or −80°C, as appropriate). Water was added to the S9 mix to make up any needed volume before the addition of freshly thawed S9. The preincubation mixes contain 500 μL S9 mix; an additional 75 μL water (if needed), 100 μL bacterial tester strain and 25 μL (if an organic solvent is used to dissolve the test compound) or 100 μL (if water is used to dissolve the test compound) of freshly prepared test compound or vehicle, with the components added in the order stated. NH7, for preincubation mixes using Negishi and Hayatsu (1980) cofactors at pH 7.4; NH6, for preincubation mixes using Negishi and Hayatsu (1980) cofactors at pH 6.0.

^a Added as a single solution containing both Magnesium chloride and Potassium chloride.

2.2.5. Plating, Incubation, and Colony‐Counting

After the 30‐min preincubation, the mixes were transferred into glass tubes containing 2 mL of 45°C molten top agar (0.6% agar/0.5% NaCl containing either 0.05 mM tryptophan for assays conducted with WP2 uvrA (pKM101) or 0.05 mM histidine/biotin for assays conducted with TA1535). The tube contents were mixed briefly, immediately spread onto 100‐mm minimal glucose plates (MOLTOX) and allowed to solidify. The plates were then inverted and placed in a 37°C incubator for 2 days. Assays conducted with WP2 uvrA (pKM101) were routinely incubated for one additional day at room temperature to encourage colony growth.

Signs of toxicity were evaluated by visually inspecting the background bacterial lawn of the plates. Visual signs of cytotoxicity included: a decrease in colonies at higher positive doses, a thin lawn with small pinpoint colonies, no colonies present, or test plates showing a decrease in revertant colonies of 50% or more relative to the vehicle control or highest revertant frequency induced by the test chemical. Any precipitation that occurred during the preparation of the preincubation mixes, plating, or after incubation was documented.

Colonies were counted using an automated plate counter (ProtoCOL3 version 1.0.27.0). Plates were also counted manually when automated counting was compromised by precipitation or signs of toxicity. Colony counts distorted by contamination or toxicity were documented and not used for evaluating mutagenicity.

2.2.6. Data Evaluation

The mean and standard deviations of each experimental group were placed in summary data tables. These tables can be found in the Supporting Information section. Test substances tested in WP2 uvrA (pKM101) were considered positive when the mean colony counts were 2‐fold or greater than the concurrent vehicle control; when tests were conducted with TA1535, the test substance was positive when colony counts in the treatment assays were a minimum of 3‐fold greater than the concurrent vehicle control value.

2.2.7. Determining the Effect of Assay Conditions on Mutagenicity Responses

For compounds that tested negative under the AEAT conditions, adjusting the pH of the preincubation mix was considered to have an effect when the test substance produced a consistently positive response using the criteria indicated in Section 2.2.6.

For test substances that were positive using the AEAT conditions, quantitative comparisons between revertant frequencies induced using the AEAT conditions (see below) and modified preincubation conditions (NH6 or NH7) were made using Benchmark Concentration (BMC) mutagenicity‐response ranking (Guo et al. 2016; Wills et al. ²⁰¹⁶; Mittelstaedt et al. ²⁰²¹). The analysis was conducted using web‐based PROAST software, version 70.1 (available at https://proastweb.rivm.nl/). BMC₁₀₀ values (concentrations producing a doubling of the vehicle control revertant frequency) and their 90% Confidence Intervals (CIs) were calculated using the type of preincubation mix as a covariate and model averaging with 200 iterations. Concentration‐related responses with non‐overlapping BMC₁₀₀ CIs were considered significantly different. The Lowest Observed Genotoxicity Level (LOGEL) and the magnitude of the mutagenic response also were considered when evaluating the relative responses produced by AEAT‐positive nitrosamines using the different preincubation conditions.

Assays, whose results were clear, whether they were positive or negative, were not repeated. Assays whose results were suggestive of an effect (e.g., an increasing response with concentration, with a maximum approximately twofold [for WP2 uvrA (pKM101)] or threefold [for TA1535] of the solvent control frequency), were replicated, occasionally multiple times.

3. Results

Table 2 summarizes results obtained with tests conducted to evaluate the effect of the cofactors used in the preincubation mix and the pH of the preincubation mix on the mutagenicity of five small molecule nitrosamines and 10 NDSRIs. Responses that we conclude were increased significantly by the pH 6.0 preincubation mix (i.e., the NH6 mix) are highlighted in green. Detailed mutagenicity data can be found in the Supporting Information. The overriding purpose of the study was to determine if reducing the pH of the preincubation mix would produce a positive response for a nitrosamine that was otherwise negative in the EAT; thus, of the 15 compounds tested, two of the small‐molecule nitrosamines and 7 NDSRIs tested in this study were reported previously to be negative in the EAT (Heflich et al. 2024; Table 2).

TABLE 2.

Effect of preincubation conditions on the mutagenicity of small molecule nitrosamines and NDSRIs in the Ames test.

Test chemical (CAS number)	TA1535		WP2 uvrA (pKM101)
	EAT response a	Relative extent of mutation, NH6 vs. AEAT b	EAT response a	Relative extent of mutation, NH6 vs. AEAT
Small molecule nitrosamines
N‐nitroso‐dimethylamine	Positive	NH6>AEAT	Positive	NH6>AEAT
N‐nitroso‐diethylamine	Positive	NH6>AEAT	Positive	NH6>AEAT
N‐nitroso‐methyl‐4‐aminobutyric acid	Negative	NH6, NH7, AEAT all negative	Negative	NH6, NH7, AEAT all negative
N‐nitroso‐diphenylamine	Negative	NH6, NH7, AEAT all negative	Negative	NH6, NH7, AEAT all negative
N‐(2,2‐diethoxyethyl)‐2,2‐diethoxy‐N‐nitrosoethanamine	Positive	n.s. differences	Negative	NH6, NH7, AEAT all negative
NDSRIs
N‐nitroso‐phenylephrine	Negative	NH6>AEAT c	Negative	NH6, NH7, AEAT all negative
N‐nitroso‐varenicline	Positive	n.s. differences	Positive	n.s. differences
N‐nitroso‐diclofenac	Negative	NH6, NH7, AEAT all negative	Negative	NH6, NH7, AEAT all negative
N‐nitroso‐sertraline	Negative	NH6, NH7, AEAT all negative	Positive	n.s. differences
N‐nitroso fluoxetine	Positive	n.s. differences c , d	Positive	n.s. differences
N‐nitroso‐paroxetine	Negative	NH6, NH7, AEAT all negative	Negative	NH6, NH7, AEAT all negative
N‐nitroso‐desvaleryl‐valsartan	Negative	NH6, NH7, AEAT all negative	Negative	NH6, NH7, AEAT all negative
N‐nitroso‐bumetanide	Negative	NH6, NH7, AEAT all negative	Negative	NH6, NH7, AEAT all negative
N‐nitroso‐desvaleryl‐valsartan methyl ester	Negative	NH6, NH7, AEAT all negative	Negative	NH6, NH7, AEAT all negative
N‐nitroso‐dabigatran etexilate	Negative	NH6, NH7, AEAT all negative	Negative	NH6, NH7, AEAT all negative

Note: All preincubations conducted for 30 min with 30% liver S9 from hamsters pretreated with a mixture of phenobarbital and beta‐naphthoflavone. AEAT, preincubation conditions based on the abbreviated version of the Enhanced Ames Test (EAT); NH7, preincubations using Negishi and Hayatsu (1980) cofactors at pH 7.4; NH6, preincubations using Negishi and Hayatsu (1980) cofactors at pH 6.0. See Section 2 and Table 1 for full description of preincubation conditions. Significant differences for the extent of mutagenesis determined by non‐overlapping 90% confidence intervals at BMC₁₀₀. Significant increases in mutagenicity with NH6 are highlighted in green.

^a EAT response as reported in Heflich et al. (2024).

^b Determined by PROAST dose‐response analysis using model averaging and a Critical Effect Size (CES) of 1.0 (i.e., twice the vehicle control); n.s., responses not significantly different based on non‐overlapping 90% confidence intervals for the Benchmark Concentration at a CES of 1.0.

^c Analysis performed on data combined from three trials.

^d One out of three trials (Trial 1) produced a NH6>NH7 result of doubtful relevance (see text).

The three preincubation mixes that were used included the 30% hamster S9 preincubation mix used in our previous study (Heflich et al. 2024). This will be referred to here as ‘abbreviated EAT’ or (AEAT) conditions, because the EAT also recommends testing be conducted using 30% rat liver S9 (FDA ^2024a, Health Canada ²⁰²⁴; EMA, 2024). Also, the EAT uses five bacterial tester strains, whereas the AEAT used only two, TA1535 and WP2 uvrA (pKM101).

The mutagenicity responses in the current study for the AEAT preincubation mix were qualitatively similar to those reported previously. Positive responses were detected for three small molecule nitrosamines, NDMA, NDEA, and N‐(2,2‐diethoxyethyl)‐2,2‐diethoxy‐N‐nitrosoethanamine (TA1535 only), and for three NDSRIs, N‐nitroso‐varenicline, N‐nitroso‐fluoxetine, and N‐nitroso‐sertraline [WP2 uvrA (pKM101) only]; all the remaining test substances were negative under the AEAT conditions as they were previously under EAT conditions (Heflich et al. 2024).

The mutagenicity of both NDMA and NDEA was increased using the NH6 preincubation mix as judged by their relative BMCs and CIs (Table 2; PROAST graphical output for NDEA shown in Figure 1A). Increases in the mutagenicity of NDMA and NDEA can also be seen in the relative Lowest Observable Genotoxicity Levels (LOGELs) and large differences in the magnitude of the mutagenicity responses using the AEAT and NH6 preincubation mixes. For instance, the data shown in Supporting Information, Table A for NDMA mutagenicity in TA1535 indicate that the LOGEL for the AEAT mix was 400 μg/plate, while the LOGEL for the NH6 preincubation mix was 25 μg/plate and that the magnitudes of the differences in mutant frequency for preincubations conducted using the two conditions were in the 100s to 1000s of revertants/plate. The mutagenicity of NDMA was also increased in TA1535 using the Negishi and Hayatsu preincubation mix at a pH of 7.4 (NH7 responses), but not to the degree that it was with the NH6 preincubation mix (Supporting Information, Table A (NDMA/TA1535)).

FIGURE 1.

BMC dose‐response analysis showing relative mutagenicity of NDEA (A), N‐nitroso‐varenicline (B), and N‐nitroso‐phenylephrine (C) in TA1535 under different preincubation conditions. Tabular data used for the PROAST analysis can be found in Supporting Information, that is, Table C for (A) with red, black, and green curves from standard (S data), pH 6 Negishi and Hayatsu (NH6 data), and pH 7.4 Negishi and Hayatsu (NH7 data) S9 mixes, respectively; Supporting Information, Table M‐4 for (B) with all S9 preincubation mixes producing similar responses; and Supporting Information, Tables K, K‐1, and K‐2 for (C) with data pooled from three trials (red, S data; black, NH6 data). PROAST output used by permission from RIVM.

Effects on the mutagenicity of the other EAT‐positive small‐molecule nitrosamines and NDSRIs were not as easily determined. The magnitude of the effects, if any, was much less than with NDMA and NDEA, and often several repeat assays were conducted to study these small effects. For example, BMC analyses conducted on N‐nitroso‐varenicline (Table N‐1) (N‐nitroso‐varenicline/WP2 uvrA (pKM101) in Supporting Information) and N‐nitroso‐fluoxetine (Table S (N‐nitroso‐fluoxetine/TA1535)) occasionally showed a significant pH‐related effect on mutagenicity. Also, in some assays, the Negishi and Hayatsu (1980) cofactors by themselves (tests run with preincubations at pH 7.4) affected NDSRI mutagenicity (Tables M‐1 (N‐nitroso‐varenicline/TA1535), R (N‐nitroso‐sertraline/TA1535), S‐2 (N‐nitroso‐fluoxetine/TA1535)), both positively and negatively. These apparent effects for EAT‐positive NDSRIs, however, were not reproduced in repeat assays, were generally small in magnitude, and in some instances may have been influenced by relatively low vehicle control revertant frequencies (e.g., Table M‐1 (N‐nitroso‐varenicline/TA1535), and S (N‐nitroso‐fluoxetine/TA1535)). The BMC analysis relies upon comparisons between treated samples and the solvent control, and when tests are run with TA1535, the control frequencies are numerically small and small differences in revertant counts can have large effects on the BMC estimation. Thus, these apparent effects, especially with TA1535, that were not repeated in subsequent tests, may represent responses at the limit of the power of the assay conditions to detect an effect. We have not included these responses in Table 2 where responses that we consider altered by the modified preincubation conditions are indicated.

The effects of the different preincubation mixes were much easier to determine for nitrosamines that were negative in the EAT. One of the 9 EAT‐negative nitrosamines examined in our study, N‐nitroso‐phenylephrine, tested positive, albeit weakly positive, using the NH6 preincubation mix. In each of three independent trials, N‐nitroso‐phenylephrine produced a positive response with tester strain TA1535 when the NH6 preincubation mix was employed. Only one of these three trials showed a pH 6.0‐related positive increase as judged by BMC analysis (Supporting Information Tables K, K‐1, and K‐2; Trial 3 results from Table K‐2 had BMC₁₀₀ values whose CIs did not overlap). By pooling data from all three trials, however, BMC dose response analysis indicated that the mutagenic response produced by N‐nitroso‐phenylephrine using the modified pH 6.0 preincubation mix was significantly greater than when using the AEAT conditions (Figure 1C; upper vs. lower Cl for the AEAT, 2080 vs. 840 μg/plate; for the NH6 assay, 527 vs. 220 μg/plate).

4. Discussion

The alternative preincubation mixes employed in this study were based on the mix used by Negishi and Hayatsu (1980). Thus, we employed an S9 preincubation mix adjusted to a pH of 6.0 (the NH6 mix) that contained preformed reducing equivalents (NADPH and NADH) as did Negishi and Hayatsu. Unlike Negishi and Hayatsu, however, our mixes used 30% hamster liver S9 for metabolic activation and tester strains TA1535 and WP2 uvrA (pKM101) to detect mutagenesis. In our previous study, we observed that all positive mutagenicity responses for the 29 small‐molecule nitrosamines and NDSRIs that were tested could be detected with a combination of TA1535 and WP2 uvrA (pKM101) and that preincubation mixes containing 30% hamster S9 generally produced the strongest mutagenic responses (Heflich et al. 2024). Negishi and Hayatsu used TA100 in their study, which also was a sensitive tester strain in our previous study, but not as sensitive for detecting nitrosamine mutagenicity as TA1535 or the combination of TA1535 and WP2 uvrA (pKM101).

Compared with responses in the AEAT, the NH6 preincubation mix increased the mutagenicity of NDMA in TA1535 by nearly 20‐fold and NDEA by 12‐fold; smaller, but still significant increases in mutagenicity were found in assays employing WP2 uvrA (pKM101) (3‐ and 5‐fold for NDMA and NDEA). These fold changes were based on comparing BMC₁₀₀ values generated by BMC dose–response analysis using the model that best fit the concentration‐response data (data in Supporting Information Tables A (NDMA/TA1535), B (NDMA/WP2 uvrA (pKM101)), C (NDEA/TA1535), D (NDEA/WP2 uvrA (pKM101))). These increased responses with slightly acidic preincubation mixes were consistent with the findings of Guttenplan (1979, 1980) and Negishi and Hayatsu (1980); NDMA and NDEA were the only two nitrosamines used in common between this present study and the previous studies. The mutagenicity of two EAT‐positive NDSRIs (N‐nitroso‐varenicline and N‐nitroso‐fluoxetine) also was increased by the NH6 preincubation conditions in some trials conducted with TA1535 (see Supporting Information). We consider these enhancements to be questionable because they were not replicated consistently (see Section 3).

NDMA and NDEA were the two smallest (lowest molecular weight) of the small‐molecule nitrosamines tested. Guttenplan (1980), who tested a series of small‐molecule nitrosamines using pH 6.5 and 7.4 preincubation mixes, found that the lower pH generally increased the mutagenicity of the test compounds, and that the magnitude of the increase was inversely related to the molecular weight of the nitrosamine. Excluding the questionable enhancements mentioned above, this relationship between molecular weight and enhancement generally was seen in our study: the smallest nitrosamines tested, NDMA, followed by NDEA, had the greatest increases in mutagenicity using the NH6 preincubation mix.

In addition, N‐nitroso‐phenylephrine was the only EAT‐negative nitrosamine that produced a positive, albeit a very weakly positive, mutagenicity response with the NH6 preincubation mix. It was also, at 196.21 Da, the smallest of the NDSRIs tested in our study. However, N‐nitroso‐phenylephrine has structural features that classify it by the Carcinogenic Potency Categorization Approach (CPCA) in Potency Category 2, indicating that it has a structure suggesting that it can be metabolized by α‐carbon hydroxylation to a mutagenic diazonium ion (FDA ^{2023, 2024b}; Kruhlak et al. ²⁰²⁴). It is possible that a diazonium ion produced from N‐nitroso‐phenylephrine is sufficiently labile that it fails to interact with tester strain DNA under EAT conditions. Note that there is a shallow dose‐responsive increase in mutagenicity induced by N‐nitroso‐phenylephrine under the AEAT preincubation conditions that usually does not quite reach the three‐fold increase relative to the concurrent control frequency necessary for a positive response in TA1535 (Supporting Information Tables K, K‐1, K‐2). Note also, that of the five trials conducted with N‐nitroso‐phenylephrine in TA1535 under EAT and AEAT conditions (including the two trials reported in Heflich et al. ²⁰²⁴), one (Supporting Information Table K‐1) resulted in a response for the high dose (5 mg/plate) that exceeded the threefold concurrent negative control frequency. Thus, we conclude that N‐nitroso‐phenylephrine is a compound whose mutagenicity is generally below the mutagenicity detection limit of the EAT (and AEAT) assay, but that the mutagenicity enhancement provided by the NH6 preincubation mix is sufficient to result in a consistently positive response, at least when mutagenicity is evaluated using the fold‐rule.

It is not clear why acidic conditions increase the mutagenicity of at least some small‐molecule nitrosamines and NDSRIs. Guttenplan (1980) found that slightly acid pHs reduce the generation of NADPH from NADP. Thus, the preincubations used in his studies were split into two phases, one at neutral pH to generate NADPH, followed by one at pH 6.5 to metabolize the test nitrosamine in the presence of a bacterial tester strain. Negishi and Hayatsu (1980) avoided the problem of generating NADPH in the preincubation mix by supplying their slightly acidified preincubation mix with preformed NADPH and NADH. Guttenplan (1980) also found that pH 6.5 inhibited NDMA demethylation at high, but not low concentrations of NDMA; Negishi and Hayatsu (1980) found that preincubation mixes at pH 6.0 and 5.0 inhibited NDMA demethylation at all the concentrations of NDMA that they tested. This would tend to decrease, not increase NDMA mutagenicity. One theory is that acidified preincubation mixes serve to stabilize reactive metabolites of nitrosamines (e.g., as happens with the direct‐acting mutagen, N‐ethyl‐N‐nitrosourea [de Kok et al. ¹⁹⁸³]), making them more likely to interact with the DNA of the tester strains and cause mutations. Thus, it may be that the acid pH used for the assay results in a balance between increasing mutagenesis by stabilizing metabolites and decreasing mutagenesis by inhibiting the metabolic activation pathways for nitrosamines. The balance that is struck at pH 6.0 resulted in an enhancement of mutagenesis for the smallest of small‐molecule nitrosamines (NDMA and NDEA) and at least one NDSRI (N‐nitroso‐phenylephrine). It is conceivable that other pHs in the acid range may be more suitable for mediating the mutagenicity of nitrosamines having specific structural features. This speculation remains to be tested.

In this study, we evaluated a further modification of the Ames test that may improve the detection of the mutagenicity of nitrosamines. OECD TG 471 does not comment on the pH of the preincubation mix used for the preincubation version of the assay; therefore, the modified NH6 preincubation mix used in the current study appears to be consistent with the recommendations for conducting the assay given by TG 471. We conclude that adjusting the pH of the preincubation mix to pH 6.0 and using NADPH and NADH as cofactors, as described in this study, could produce data that add to the weight of evidence supporting a negative response in the EAT. The assay also could serve to identify mutagenic NDSRIs, most likely NDSRIs with lower molecular weights, whose mutagenicity is below the level of detection for the standard EAT.

Although we found that detecting small effects on the mutagenicity of EAT‐positive nitrosamines could be challenging, often requiring repeat testing, the effect of the NH6 preincubation mix on the mutagenicity of nitrosamines that otherwise tested negative in the EAT was relatively straightforward. Based on the findings in this report, it may be prudent to test any EAT‐negative small‐molecule nitrosamine or NDSRI, especially those in CPCA Potency Categories 1 or 2, using the method described in this report before conducting tests using more resource‐intensive mammalian cell or in vivo assays.

Author Contributions

Drs Aisar H. Atrakchi, Timothy J. McGovern, Robert H. Heflich, Naomi L. Kruhlak, Robert T. Dorsam, Sruthi T. King, Nan Mei, and Ms. Michelle E. Bishop designed the study. Dr. Robert H. Heflich and Ms. Michelle E. Bishop, Audrey M. Sims, Sharon K. Guerrero, and Kamela Mitchell conducted the study and analyzed the data. Dr. Nan Mei prepared the draft figure and tables. Dr. Robert H. Heflich prepared the manuscript draft with important intellectual input from Drs. Sruthi T. King, Aisar H. Atrakchi, Timothy J. McGovern. All authors approved the final manuscript. Dr. Robert H. Heflich and Ms. Michelle E. Bishop have complete access to the study data and primary data are available upon request.

Disclosure

This article reflects the views of its authors and does not necessarily reflect those of the U.S. Food and Drug Administration. Any mention of commercial products is for clarification only and is not intended as approval, endorsement, or recommendation.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: Supporting Information.

Bishop, M. E. , Sims A. M., Guerrero S. K., et al. 2026. “Adjusting the Preincubation Conditions to Enhance the Ames Test for Detecting the Mutagenicity of N‐Nitrosamines.” Environmental and Molecular Mutagenesis 67, no. 3: e70041. 10.1002/em.70041.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.