Ames Testing Reveals Structural Nitrogen Alerts Governing Heterocycle Mutagenicity

A comprehensive toxicological investigation evaluating simple nitrogen-containing aromatic heterocycles has identified specific ring configurations that govern bacterial mutagenicity in the Ames test. Writing in Genes and Environment, an investigative team evaluated twenty-four unsubstituted heterocyclic compounds, comprising twelve quinoline analogues and twelve indole analogues, across five standard bacterial tester strains under preincubation conditions with and without mammalian metabolic activation. The experimental findings revealed that five quinoline analogues and two indole analogues induced significant mutagenic activity. By mapping positive responses directly to the atomic architecture of each core structure, the investigators demonstrated that the exact placement and quantity of nitrogen atoms dictate whether these heterocyclic compounds trigger base-pair substitutions or frameshift mutations. These empirical results address persistent predictive blind spots in computational toxicology, yielding refined structural alerts capable of correctly discriminating mutagens from non-mutagens across these fundamental chemical scaffolds.

Heterocyclic ring systems containing nitrogen atoms are often used as basic structural moieties in pharmaceuticals. Historically, the bacterial reverse mutation assay, commonly known as the Ames test, has served as the worldwide gold standard for identifying chemical mutagens and evaluating potential genotoxic hazards. Over recent decades, computational in silico models have gained regulatory acceptance, most notably under the International Council for Harmonisation M7 guideline for assessing mutagenic impurities in pharmaceuticals. Despite substantial progress, the predictive reliability of both knowledge-based and statistical in silico screening platforms remains compromised for specific chemical categories due to an acute shortage of robust empirical Ames data. A significant portion of legacy testing historical records relied on non-standard experimental protocols, failed to utilize the full battery of guideline-recommended bacterial strains, or omitted tests conducted in both the presence and absence of metabolic activation systems.

To overcome historical inconsistencies and systematically establish baseline structure-activity relationships, the research team selected twelve quinoline analogues and twelve indole analogues that completely lacked functional side moieties or peripheral substituents. The quinoline series comprised naphthalene ring systems substituted with one or two nitrogen atoms, encompassing quinoline, isoquinoline, cinnoline, quinazoline, quinoxaline, phthalazine, and six distinct naphthyridine isomers: 1,5-naphthyridine, 1,6-naphthyridine, 1,7-naphthyridine, 1,8-naphthyridine, 2,6-naphthyridine, and 2,7-naphthyridine. The indole series comprised indene cores containing between one and three ring nitrogen atoms, encompassing indole, indazole, benzimidazole, 4-azaindole, and polyaza derivatives including 5H-pyrrolo[3,2-d]pyrimidine and pyrazolo[3,4-b]pyridine. Crucially, seventeen of these twenty-four compounds had never been evaluated in documented Ames test literature, while six of the remaining seven compounds possessed only partial historical evaluations conducted with incomplete strain batteries, leaving parent quinoline as the sole compound previously tested across all guideline-stipulated bacterial strains.

The experimental protocol strictly complied with international standards, following testing recommendations outlined in Organisation for Economic Co-operation and Development guideline 471 and the ICH S2 guidance. The researchers utilized four Salmonella typhimurium strains—TA100, TA1535, TA98, and TA1537—alongside the Escherichia coli strain WP2uvrA. Liquid preincubation assays were executed at thirty-seven degrees Celsius for twenty minutes prior to plate incorporation in minimal-glucose agar. Exogenous metabolic activation was supplied using rat liver post-mitochondrial S9 supernatant prepared from male Sprague-Dawley rats pretreated with enzyme inducers phenobarbital and 5,6-benzoflavone. The standard S9 mix incorporated a standardized cofactor solution containing magnesium chloride, potassium chloride, glucose-6-phosphate, nicotinamide adenine dinucleotide phosphate, reduced nicotinamide adenine dinucleotide, and sodium phosphate buffer at physiological pH. For assays executed in the absence of exogenous activation, sodium phosphate buffer replaced the S9 preparation, ensuring strictly uniform reaction volumes across all negative controls, dose-range iterations, and positive control validations.

Mutagenic evaluation followed standardized threshold benchmarks to ensure reproducible interpretations across testing tiers. The researchers classified a chemical as mutagenic only if the maximum revertant colony yield satisfied the two-fold rule for strains TA100, TA98, and WP2uvrA, or the three-fold rule for strains TA1535 and TA1537, while simultaneously exhibiting a reproducible, dose-dependent escalation above concurrent negative vehicle controls. Background lawn integrity was microscopically verified to differentiate true mutation induction from growth inhibition and general cytotoxicity. To quantitatively express the magnitude of mutagenic potency among confirmed positive agents, the authors derived maximum mutagenic activity, measured in revertants per milligram of test compound, calculated by subtracting mean negative control revertant counts from the peak revertant yield at a specific non-toxic dose level and dividing the difference by the corresponding administered dose.

Within the quinoline cohort, five of the twelve evaluated molecules demonstrated clear mutagenic potency. Parent quinoline exhibited exceptionally potent activity, generating 30,758 revertants per milligram in strain TA100 solely in the presence of the S9 metabolic activation mixture. Among the naphthyridine isomers, 1,5-naphthyridine produced 419 revertants per milligram, 1,7-naphthyridine produced 255 revertants per milligram, and 1,8-naphthyridine produced 170 revertants per milligram, each demonstrating peak activity in strain TA100 under metabolic activation. Furthermore, 1,7-naphthyridine exhibited dual reactivity, inducing mutagenicity both with and without S9 mix in strains TA100 and TA98. In marked contrast to the other quinoline derivatives, cinnoline induced mutagenic activity exclusively in the absence of S9 mix, generating 13 revertants per milligram within strain TA1537. Because TA100 primarily detects base-pair substitution mutations at guanine-cytosine sites while TA98 and TA1537 identify distinct frameshift mutations at guanine-cytosine pairs, the observed response profiles demonstrate that quinoline analogues predominantly generate DNA adducts targeting guanine-cytosine genomic loci.

The structural divergence among quinoline analogues provides clear mechanistic insight into metabolic activation pathways. The presence of a nitrogen atom at position 1 appeared indispensable for metabolic activation; analogues lacking a nitrogen at position 1, including isoquinoline, phthalazine, 2,6-naphthyridine, and 2,7-naphthyridine, failed to exhibit mutagenicity under any condition. Conversely, the introduction of a second ring nitrogen at position 3 (quinazoline) or position 4 (quinoxaline) completely suppressed mutagenic capacity, indicating that nitrogens at these adjacent locations structurally impede enzymatic epoxidation. The non-mutagenicity of 1,6-naphthyridine further suggests that a nitrogen at position 6 disrupts activation pathways. Meanwhile, the unprompted mutagenicity of cinnoline and 1,7-naphthyridine without mammalian S9 implies alternative bioactivation pathways, likely driven by endogenous bacterial oxidases generating reactive intermediates directly within the bacterial cytoplasm.

Among the twelve indole analogues, mutagenic responses were remarkably sparse, with only two compounds producing positive revertant induction. Indazole displayed weak mutagenic potency in strain WP2uvrA, yielding 38 revertants per milligram in the presence of S9 mix and 40 revertants per milligram in its absence. Because the dose-response trajectories for indazole were virtually identical regardless of whether exogenous liver enzymes were present, the compound functioned as a direct-acting mutagen, inducing base-pair substitutions at adenine-thymine base pairs rather than guanine-cytosine sites. Meanwhile, 4-azaindole exhibited weak mutagenicity restricted to the absence of S9 activation, yielding 4 revertants per milligram in strain TA1537 via frameshift mutation induction. Parent indole, benzimidazole, and polyaza derivatives such as pyrazolo[3,4-b]pyridine and 5H-pyrrolo[3,2-d]pyrimidine were non-mutagenic, demonstrating that introducing nitrogen atoms outside positions 2 or 4 fails to induce mutagenicity.

The researchers subsequently tested all twenty-four heterocycles across established computational platforms to assess in silico predictive performance against the newly gathered laboratory data. Knowledge-based expert systems—specifically DEREK Nexus version 6.1.1 and CASE Ultra GT_EXPERT version 1.8.0.5—demonstrated poor sensitivity. Both platforms achieved merely 20 percent sensitivity for quinoline analogues by detecting only parent quinoline while misclassifying or yielding inconclusive verdicts for the remaining four mutagens. For the indole analogues, both expert systems demonstrated 0 percent sensitivity, completely failing to detect indazole and 4-azaindole as mutagens. In contrast, the statistical machine-learning model CASE Ultra GT1_BMUT exhibited 100 percent sensitivity across both chemical families by identifying positive alerts in all seven active mutagens. However, this high sensitivity came at the cost of poor specificity, dropping to 43 percent for quinoline analogues and a meager 10 percent for indole analogues, generating widespread false-positive alerts across safe compounds.

To reconcile the performance shortcomings of commercial software, the authors formulated dedicated structural alert rules grounded directly upon their empirical findings. Under the proposed framework, a quinoline analogue without peripheral substituents is classified as mutagenic when it possesses ring nitrogen atoms at positions 2, 5, 7, or 8 in addition to position 1, or retains the unsubstituted quinoline scaffold itself. For unsubstituted indole analogues, the structural alert designates compounds as mutagenic only when a nitrogen atom is positioned at ring position 2 or position 4 in addition to position 1, explicitly excluding unsubstituted parent indole. When applied retrospectively to the experimental data, these targeted alerts achieved 100 percent predictive accuracy across the twenty-four scaffolds. The authors emphasized that their empirical dataset remains focused on unsubstituted heterocyclic cores; incorporating reactive chemical substituents such as nitro groups, amino groups, or alkyl halides, or accounting for in vivo heterocyclic N-oxide metabolic transformations, will require systematic experimental verification in subsequent testing campaigns.

Subject of Research: Structure-mutagenicity relationships of quinoline and indole analogues in bacterial reverse mutation assays

Article Title: Structure-mutagenicity relationships on quinoline and indole analogues in the Ames test

Article References: Kurakami, M., Hakura, A., Sato, R., Kawade, A., Yamagata, T., Koyama, N., Kakiuchi, D., & Asakura, S. (2024). Structure-mutagenicity relationships on quinoline and indole analogues in the Ames test. Genes and Environment, 46(1), Article 23. https://doi.org/10.1186/s41021-024-00316-6

Image Credits: AI Generated

DOI: 10.1186/s41021-024-00316-6

Keywords: Ames test, bacterial mutagenicity, quinoline analogues, indole analogues, structural alerts, Salmonella typhimurium, Escherichia coli WP2uvrA, in silico toxicology, DEREK Nexus, CASE Ultra, enamine epoxide theory, S9 metabolic activation, naphthyridines, cinnoline, indazole, 4-azaindole, genetic toxicology, ICH M7 guideline