Article pubs.acs.org/crt
Investigating a Relationship between the Mutagenicity of Arylboronic Acids and 11B NMR Chemical Shifts Maria L. Pellizzaro,† Elizabeth M. Covey-Crump,† Julie Fisher,‡ Anne-Laure D. Werner,† and Richard V. Williams*,† †
Lhasa Limited, Granary Wharf House, 2 Canal Wharf, Leeds, LS11 5PS, U.K. School of Chemistry, University of Leeds, Woodhouse Lane, Leeds, LS2 9JT, U.K.
‡
S Supporting Information *
ABSTRACT: The mutagenicity of arylboronic acids has recently become an important area of research because of their potential to be genotoxic impurities in active pharmaceutical ingredients. There is no known mechanism, so currently all structure−activity relationships have been derived using Ames test data. We present preliminary data supporting a hypothesis that the mutagenicity of arylboronic acids is related to the 11B NMR chemical shift. This could indicate that the mutagenic activity of the arylboronic acids is related to the reactivity of the boron center.
■
INTRODUCTION Understanding the mutagenicity of arylboronic acids has become increasingly important because of potential exposure to such molecules in pharmaceutical products. Although arylboronic acids are not currently present in an active pharmaceutical ingredient (API), they are common intermediates in their synthesis. Arylboronic acids are reagents in C−C bond forming reactions,1 for example, the Suzuki-Miyaura reaction,2 which is commonly used in the synthesis of APIs.3 Therefore, arylboronic acids could be present in the final API at low levels as an impurity. The mutagenicity of arylboronic acids in the Ames test was first discussed in the public domain in 2010.4,5 Since then, O’Donovan et al. have published data describing the mutagenicity of 10 arylboronic acids,6 all of which gave mutagenic responses in the Ames test in the absence of metabolic activation. Salmonella typhimurium TA100 and Escherichia coli WP2 uvrA were found to be the sensitive strains. Although many arylboronic acids are mutagenic in the Ames test, most are nongenotoxic in mammalian cells.7 The reason for this difference in activity is unclear but does suggest that the mechanism of mutagenicity in the Ames test is not applicable to mammalian cells. Understanding the mechanism of mutagenicity in the Ames test and/or obtaining a more robust structure−activity relationship (SAR) could indicate why © 2015 American Chemical Society
differences in genotoxicity are observed between bacterial and mammalian cell lines. Currently, the mechanism by which the arylboronic acids exert their mutagenic activity is unclear. Therefore, any SARs must be derived from Ames test data alone. In order to more accurately characterize the hazard posed and reduce the potential for duplicate testing, Lhasa Limited established a data sharing group consisting of eight pharmaceutical companies. This collaboration rapidly expanded the chemical space to cover 64 arylboronic acids, 43 mutagenic, 19 nonmutagenic, and 2 equivocal (unpublished data). The increased data set allowed the identification of SARs which improved, but did not perfect, the prediction of arylboronic acid mutagenicity in Derek Nexus.8 Although the mechanism of action is unknown, it can be hypothesized that mutagenicity is related to the intrinsic reactivity of the boron atom. Indeed Hammett constants,9 a measure of the electronic effects of phenyl substituents, have been shown to correlate to the mutagenicity of the arylboronic acids.10 However, Hammett constants cannot be used to predict mutagenicity for the wider data set because they cannot be calculated for ortho substituents, they are not available for Received: February 18, 2015 Published: May 27, 2015 1422
DOI: 10.1021/acs.chemrestox.5b00078 Chem. Res. Toxicol. 2015, 28, 1422−1426
Article
Chemical Research in Toxicology Table 1. 11B NMR Chemical Shift for 14 Arylboronic Acids Which Have Ames Test Results Publicly Available compound name
11
B NMR chemical shift (CDCl3/CD3OD 4:1)
4-fluoro-2-methoxyphenylboronic acid 4-carboxy-3-fluorophenylboronic acid 3-carboxyphenylboronic acid 5-fluoro-2-hydroxyphenylboronic acid 4-methoxycarbonylphenylboronic acid 2-methoxyphenylboronic acid 2,5-dimethoxyphenylboronic acid 3-hydroxyphenylboronic acid 4-carboxyphenylboronic acid p-tolylboronic acid 2-isopropylboronic acid 2-biphenylboronic acid 2,6-dimethylphenylboronic acid 2,4,6-triisopropylphenylboronic acid
27.658 27.707 27.976 28.031 28.053 28.062 28.243 28.502 28.504 28.808 30.229 30.346 30.908 31.634
nonmutagenic4 nonmutagenic21 nonmutagenic21 nonmutagenic21 mutagenic21 nonmutagenic4 mutagenic6 mutagenic21 nonmutagenic21 mutagenic6 nonmutagenic21 nonmutagenic21 nonmutagenic21 nonmutagenic21
28.217 ppm; CDCl3/CD3OD (4:1), 23.661 ppm). This is presumably a result of interactions between 3,5-difluorophenyl boronic acid and methanol, possibly resulting in a boronate species forming in the presence of methanol, which would lower the NMR chemical shift.20 Because of this behavior, 3,5difluorophenylboronic acid was not included in subsequent analyses. 11 B NMR spectra were obtained in CDCl3/CD3OD (4:1) for 14 phenylboronic acids (Table 1), all of which have publicly available Ames test data.4,6,21 The chemical shift of the boron atom will vary according to its chemical environment; for example, how sterically hindered it is and how much shielding from extranuclear electrons occurs. Compounds with a sterically hindered boron atom will have a low field chemical shift, and those that contain more electron donating groups will have a high field chemical shift. The chemical shifts of the phenylboronic acids were found to be between 27 and 32 ppm. The majority of the molecules tested produced clean spectra (see Supporting Information), i.e., there was no indication of additional boron containing compounds. However, additional signals were detected in the spectra for 2,6-dimethylphenylboronic acid (∼1% at 18.509 ppm), 5-fluoro-2-hydroxyphenylboronic acid (∼1% at 18.509 ppm), and 2,4,6-triisopropylphenylboronic acid (∼1% at 18.528 ppm). A slight shoulder is also noted to low field of the signal at 28.243 ppm for 2,5dimethoxyphenylboronic acid, which could not be resolved. Resonance lines in general were broad, particularly for 2,4,6triisopropylphenylboronic acid and 4-carboxyphenylboronic acid. The latter molecule was only partially soluble in the CDCl3/CD3OD mix, and the nonhomogenous nature of the solution undoubtedly contributed to the increased line width. A decision tree classifier in KNIME22 was used to classify mutagenic and nonmutagenic compounds based on their chemical shift (Figure 1). It was found that all of the mutagenic phenylboronic acids in the data set exhibited a chemical shift between 28.042 and 29.519. All phenylboronic acids with a chemical shift >29.519 ppm or 29.519 and
Investigating a Relationship between the Mutagenicity of Arylboronic Acids and 11B NMR Chemical Shifts Maria L. Pellizzaro,† Elizabeth M. Covey-Crump,† Julie Fisher,‡ Anne-Laure D. Werner,† and Richard V. Williams*,† †
Lhasa Limited, Granary Wharf House, 2 Canal Wharf, Leeds, LS11 5PS, U.K. School of Chemistry, University of Leeds, Woodhouse Lane, Leeds, LS2 9JT, U.K.
‡
S Supporting Information *
ABSTRACT: The mutagenicity of arylboronic acids has recently become an important area of research because of their potential to be genotoxic impurities in active pharmaceutical ingredients. There is no known mechanism, so currently all structure−activity relationships have been derived using Ames test data. We present preliminary data supporting a hypothesis that the mutagenicity of arylboronic acids is related to the 11B NMR chemical shift. This could indicate that the mutagenic activity of the arylboronic acids is related to the reactivity of the boron center.
■
INTRODUCTION Understanding the mutagenicity of arylboronic acids has become increasingly important because of potential exposure to such molecules in pharmaceutical products. Although arylboronic acids are not currently present in an active pharmaceutical ingredient (API), they are common intermediates in their synthesis. Arylboronic acids are reagents in C−C bond forming reactions,1 for example, the Suzuki-Miyaura reaction,2 which is commonly used in the synthesis of APIs.3 Therefore, arylboronic acids could be present in the final API at low levels as an impurity. The mutagenicity of arylboronic acids in the Ames test was first discussed in the public domain in 2010.4,5 Since then, O’Donovan et al. have published data describing the mutagenicity of 10 arylboronic acids,6 all of which gave mutagenic responses in the Ames test in the absence of metabolic activation. Salmonella typhimurium TA100 and Escherichia coli WP2 uvrA were found to be the sensitive strains. Although many arylboronic acids are mutagenic in the Ames test, most are nongenotoxic in mammalian cells.7 The reason for this difference in activity is unclear but does suggest that the mechanism of mutagenicity in the Ames test is not applicable to mammalian cells. Understanding the mechanism of mutagenicity in the Ames test and/or obtaining a more robust structure−activity relationship (SAR) could indicate why © 2015 American Chemical Society
differences in genotoxicity are observed between bacterial and mammalian cell lines. Currently, the mechanism by which the arylboronic acids exert their mutagenic activity is unclear. Therefore, any SARs must be derived from Ames test data alone. In order to more accurately characterize the hazard posed and reduce the potential for duplicate testing, Lhasa Limited established a data sharing group consisting of eight pharmaceutical companies. This collaboration rapidly expanded the chemical space to cover 64 arylboronic acids, 43 mutagenic, 19 nonmutagenic, and 2 equivocal (unpublished data). The increased data set allowed the identification of SARs which improved, but did not perfect, the prediction of arylboronic acid mutagenicity in Derek Nexus.8 Although the mechanism of action is unknown, it can be hypothesized that mutagenicity is related to the intrinsic reactivity of the boron atom. Indeed Hammett constants,9 a measure of the electronic effects of phenyl substituents, have been shown to correlate to the mutagenicity of the arylboronic acids.10 However, Hammett constants cannot be used to predict mutagenicity for the wider data set because they cannot be calculated for ortho substituents, they are not available for Received: February 18, 2015 Published: May 27, 2015 1422
DOI: 10.1021/acs.chemrestox.5b00078 Chem. Res. Toxicol. 2015, 28, 1422−1426
Article
Chemical Research in Toxicology Table 1. 11B NMR Chemical Shift for 14 Arylboronic Acids Which Have Ames Test Results Publicly Available compound name
11
B NMR chemical shift (CDCl3/CD3OD 4:1)
4-fluoro-2-methoxyphenylboronic acid 4-carboxy-3-fluorophenylboronic acid 3-carboxyphenylboronic acid 5-fluoro-2-hydroxyphenylboronic acid 4-methoxycarbonylphenylboronic acid 2-methoxyphenylboronic acid 2,5-dimethoxyphenylboronic acid 3-hydroxyphenylboronic acid 4-carboxyphenylboronic acid p-tolylboronic acid 2-isopropylboronic acid 2-biphenylboronic acid 2,6-dimethylphenylboronic acid 2,4,6-triisopropylphenylboronic acid
27.658 27.707 27.976 28.031 28.053 28.062 28.243 28.502 28.504 28.808 30.229 30.346 30.908 31.634
nonmutagenic4 nonmutagenic21 nonmutagenic21 nonmutagenic21 mutagenic21 nonmutagenic4 mutagenic6 mutagenic21 nonmutagenic21 mutagenic6 nonmutagenic21 nonmutagenic21 nonmutagenic21 nonmutagenic21
28.217 ppm; CDCl3/CD3OD (4:1), 23.661 ppm). This is presumably a result of interactions between 3,5-difluorophenyl boronic acid and methanol, possibly resulting in a boronate species forming in the presence of methanol, which would lower the NMR chemical shift.20 Because of this behavior, 3,5difluorophenylboronic acid was not included in subsequent analyses. 11 B NMR spectra were obtained in CDCl3/CD3OD (4:1) for 14 phenylboronic acids (Table 1), all of which have publicly available Ames test data.4,6,21 The chemical shift of the boron atom will vary according to its chemical environment; for example, how sterically hindered it is and how much shielding from extranuclear electrons occurs. Compounds with a sterically hindered boron atom will have a low field chemical shift, and those that contain more electron donating groups will have a high field chemical shift. The chemical shifts of the phenylboronic acids were found to be between 27 and 32 ppm. The majority of the molecules tested produced clean spectra (see Supporting Information), i.e., there was no indication of additional boron containing compounds. However, additional signals were detected in the spectra for 2,6-dimethylphenylboronic acid (∼1% at 18.509 ppm), 5-fluoro-2-hydroxyphenylboronic acid (∼1% at 18.509 ppm), and 2,4,6-triisopropylphenylboronic acid (∼1% at 18.528 ppm). A slight shoulder is also noted to low field of the signal at 28.243 ppm for 2,5dimethoxyphenylboronic acid, which could not be resolved. Resonance lines in general were broad, particularly for 2,4,6triisopropylphenylboronic acid and 4-carboxyphenylboronic acid. The latter molecule was only partially soluble in the CDCl3/CD3OD mix, and the nonhomogenous nature of the solution undoubtedly contributed to the increased line width. A decision tree classifier in KNIME22 was used to classify mutagenic and nonmutagenic compounds based on their chemical shift (Figure 1). It was found that all of the mutagenic phenylboronic acids in the data set exhibited a chemical shift between 28.042 and 29.519. All phenylboronic acids with a chemical shift >29.519 ppm or 29.519 and
