NMR Assignments of Six Asymmetrical N-Nitrosamine Isomers Determined in an Active Pharmaceutical Ingredient by DFT Calculations

N-nitrosamines, which are well-known pro-mutagens, are found in drugs, pickled food and tobacco. Therefore, controlling their concentrations is very important. When an HPLC, GC or NMR analysis is conducted to investigate certain asymmetrical N-nitrosamines, two sets of signals attributed to the asymmetric N-nitrosamine isomers are usually observed. However, few reports on the NMR assignment of asymmetrical N-nitrosamine isomers have been published. In this study, we investigated the NMR assignments of the Z/E isomers of six asymmetrical N-nitrosamines by means of density functional theory (DFT) calculations. The configuration of the major isomer of asymmetrical N-nitrosamine 3 was the Z-configuration. The configuration of the major isomers of asymmetrical N-nitrosamines 4–7 was the E-configuration. Then, we determined the Z/E ratios of these asymmetrical N-nitrosamines by means of variable temperature (VT) and room temperature (RT) 1H-NMR experiments. The ratios of the Z/E isomer 3 quickly increased beyond 100% in the VT 1H NMR experiments. The ratios of Z/E isomers 4–7 were increased in the range of 10–60% in the VT 1H NMR experiments. The results of this study indicate that identifying the isomers of asymmetrical N-nitrosamine is necessary to control the quality of N-nitrosamines for active pharmaceutical ingredients (APIs).


Introduction
N-Nitrosamines are well-known pro-mutagens that can react with DNA following metabolism to produce DNA adducts, such as O 6 -alkyl-guanine. These adducts can result in DNA replication miscoding errors, leading to GC > AT mutations and an increased risk of genomic instability and carcinogenesis [1]. In 2018, N-nitrosodimethylamine (NDMA, 1), a genotoxic carcinogen, was detected as a synthesis impurity in some valsartan drugs, while other N-nitrosamines, such as N-nitrosodiethylamine (NDEA, 2), were later detected in other sartan products. In September 2019, the FDA stated that a low amount of NDMA had been detected in ranitidine. The FDA also announced that it had found excessive levels of NDMA in metformin in February 2022. Some N-nitrosamines, such as N-nitrososarcosine (NSAR, 3), N-nitrosomethylvinylamine (4), 3-(methylnitrosamino)propionitrile (MNPN, 5), 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK, 6), N-nitrosornicotine (NNN, 7) and N-methyl-N-nitrosourea (MNU, 8), occur not only in drugs but also in pickled foods and tobacco ( Figure 1). Therefore, controlling their concentrations in drugs, foods and tobacco is very important. When we performed HPLC or GC analyses of certain asymmetrical N-nitrosamines, we often observed two peaks for one nitrosamine. This finding was attributed to the fact that asymmetrical N-nitrosamine may have configurational isomers due to the hindered rotation of a single bond (N-N), resulting in strong variations in the anisotropic effects. The two conformers have features similar to those of the E/Z isomers relative to a double bond ( Figure 2). A similar phenomenon has been reported, in which the stereospecific response of the E/Z isomers of NSAR (3) was determined by LC-ESI-MS/MS [2]. NSAR (3) and MNPN (5) have also been shown to produce two isomer peaks in the UPLC-MS/MS assay [3]. In this paper, we report a series of asymmetrical N-nitroamines (3)(4)(5)(6)(7) displaying two groups of NMR signals. However, few studies have reported the NMR assignment of asymmetrical N-nitroamines isomers. Inspired by the above phenomenon, variable-temperature (VT) 1 H-NMR experiments were carried out to determine the percentage changes of the two configurational isomers, which revealed the configurational isomerism phenomenon. As density functional theory (DFT) calculations are widely used to determine NMR assignments for the characterization of complex structures [4,5], we performed DFT calculations to assign the NMR signals of these conformers. To our knowledge, this is the first report of the NMR assignment of configurational isomers of N-nitrosamines. When we performed HPLC or GC analyses of certain asymmetrical N-nitrosamines, we often observed two peaks for one nitrosamine. This finding was attributed to the fact that asymmetrical N-nitrosamine may have configurational isomers due to the hindered rotation of a single bond (N-N), resulting in strong variations in the anisotropic effects. The two conformers have features similar to those of the E/Z isomers relative to a double bond ( Figure 2). A similar phenomenon has been reported, in which the stereospecific response of the E/Z isomers of NSAR (3) was determined by LC-ESI-MS/MS [2]. NSAR (3) and MNPN (5) have also been shown to produce two isomer peaks in the UPLC-MS/MS assay [3]. In this paper, we report a series of asymmetrical N-nitroamines (3-7) displaying two groups of NMR signals. However, few studies have reported the NMR assignment of asymmetrical N-nitroamines isomers. Inspired by the above phenomenon, variabletemperature (VT) 1 H-NMR experiments were carried out to determine the percentage changes of the two configurational isomers, which revealed the configurational isomerism phenomenon. As density functional theory (DFT) calculations are widely used to determine NMR assignments for the characterization of complex structures [4,5], we performed DFT calculations to assign the NMR signals of these conformers. To our knowledge, this is the first report of the NMR assignment of configurational isomers of N-nitrosamines.

Results and Discussion
As shown in Figures S1-S12, the 1 H-NMR spectrum of N-nitrososarcosine 3 showed one group of major signals (δH 3.79 and 4.28) and a set of minor signals (δH 3.01 and 5.01). In addition, the major carbon signals of 3 were observed at δC 40.0, 47.3 and 167.6, and the minor signals were observed at δH 33.0, 54.6 and 170.3. Similarly, the 13 C-NMR spectrum

Results and Discussion
As shown in Figures S1-S12, the 1 H-NMR spectrum of N-nitrososarcosine 3 showed one group of major signals (δ H 3.79 and 4.28) and a set of minor signals (δ H 3.01 and 5.01). In addition, the major carbon signals of 3 were observed at δ C 40.0, 47.3 and 167.6, and the minor signals were observed at δ H 33.0, 54.6 and 170.3. Similarly, the 13 C-NMR spectrum of N-nitrososarcosine 4 showed two sets of different carbon signals. However, some of the 1 H-NMR signals had differences, such as signals of -NCH 3 (δ H 3.15 vs. δ H 3.89) and H-1 (δ H 7.89 vs. δ H 7.58). Two conformers of 5 showed two groups of distinct 1D NMR signals, of which the maximum difference in the 1 H-NMR and 13 C-NMR spectra between the two isomers was 0.77 ppm for -NCH 3 (3.03 vs. 3.80 ppm) and 8.6 ppm for C-3 (49.1 vs. 40.5 ppm), respectively. Differences in the 1 H-NMR spectra between the two isomers of 6 were present in the alkyl chain, including H-2, H-3 and H-4. Furthermore, the differences in their 13 C-NMR spectra were associated with -NCH 3 and the chain from C-2 to C-4. Two groups of NMR signals in the spectrum of 7 can be easily distinguished. Above all, the major and minor signals were also assigned based on the peak integration (Tables 1 and 2). Furthermore, the configurational exchange and conformer ratios of 3-7 were investigated via a VT 1 H-NMR experiment. To further investigate the configurational behavior of asymmetrical N-nitrosamines 3-7, DFT quantum chemical calculations were conducted [6]. Because the hindered rotation of the nitryl formed E and Z configurations, resembling the Z/E isomers relative to the double bond, two configurational isomers (a/b) were converted to Z/E for further calculations ( Figure 2).
Compound 3 may contain 4 isomers 3a-3d ( Figure 3). The DFT calculations showed that the Gibbs free energies of isomers 3c and 3d are higher than those of 3a and 3b (Figure 3), suggesting that they are more unstable than 3a and 3b; thus, we mainly considered the contributions of 3a and 3b to the NMR data.

6′
153.8 153.8 To further investigate the configurational behavior of asymmetrical N-nitro 3-7, DFT quantum chemical calculations were conducted [6]. Because the hinder tion of the nitryl formed E and Z configurations, resembling the Z/E isomers re the double bond, two configurational isomers (a/b) were converted to Z/E for fur culations ( Figure 2).
Compound 3 may contain 4 isomers 3a-3d ( Figure 3). The DFT calculations that the Gibbs free energies of isomers 3c and 3d are higher than those of 3a and 3b 3), suggesting that they are more unstable than 3a and 3b; thus, we mainly consid contributions of 3a and 3b to the NMR data. Compounds 4 and 7 have sp 2 CH or CH2, which could affect the stability of mers. For compound 4, we considered four possible stable conformers, and their were calculated. As shown in Figure 4, the interaction of N=O and sp 2 CH can b sented through the energy difference between E1 and E2. Similarly, the interaction b N=O and sp 2 CH2 can be shown through the energy difference of E4-E1. In addi interaction between the nitrogen atoms of N=O and sp 2 CH2 can be interpreted by ergy calculation of E3 and E1. On the basis of their energy differences, the closer sp of CH or CH2 and N=O are, the more unstable they are. Thus, for compound 7, t dine ring is rich in electrons, similar to the double bond in compound 4, which re N=O-containing electrons. Meanwhile, considering the steric hindrance of pyrid E-configuration of 7 is more stable than the Z-configuration, which is consistent energy calculation.  Compounds 4 and 7 have sp 2 CH or CH 2 , which could affect the stability of the isomers. For compound 4, we considered four possible stable conformers, and their energies were calculated. As shown in Figure 4, the interaction of N=O and sp 2 CH can be represented through the energy difference between E 1 and E 2 . Similarly, the interaction between N=O and sp 2 CH 2 can be shown through the energy difference of E 4 -E 1 . In addition, the interaction between the nitrogen atoms of N=O and sp 2 CH 2 can be interpreted by the energy calculation of E 3 and E 1 . On the basis of their energy differences, the closer sp 2 values of CH or CH 2 and N=O are, the more unstable they are. Thus, for compound 7, the pyridine ring is rich in electrons, similar to the double bond in compound 4, which repels the N=O-containing electrons. Meanwhile, considering the steric hindrance of pyridine, the E-configuration of 7 is more stable than the Z-configuration, which is consistent with the energy calculation.  Intriguingly, N-nitrososarcosine 8 showed only one set of NMR signals, sug that only one optimized conformer was present in 8, which was caused by the key gen bond between the oxygen or nitrogen in the nitryl moiety and the hydrogen Intriguingly, N-nitrososarcosine 8 showed only one set of NMR signals, suggesting that only one optimized conformer was present in 8, which was caused by the key hydrogen bond between the oxygen or nitrogen in the nitryl moiety and the hydrogen in urea, restricting its configurational exchange. The presence of hydrogen bonds was established by energy calculations at the M062X/Def2TZVP level of theory. Both the E configuration (8a) and the Z configuration (8b) might form a hydrogen bond. The E configuration (8a) was predicted to be 4.97 Kcal/mol lower in energy than the Z configuration (8b), indicating that the E configuration (8a) may be the stable configuration, with an intermolecular hydrogen bond of approximately 2.225 Å ( Figure 5A). The DFT quantum chemical calculations showed that the calculated 13 C NMR data for the E configuration (8a) were less different from the experimental data. Based on the above evidence, one set of NMR signals was concluded to be from the E configuration (8a).
Intriguingly, N-nitrososarcosine 8 showed only one set of NMR that only one optimized conformer was present in 8, which was caus gen bond between the oxygen or nitrogen in the nitryl moiety and th restricting its configurational exchange. The presence of hydrogen bo by energy calculations at the M062X/Def2TZVP level of theory. Both (8a) and the Z configuration (8b) might form a hydrogen bond. The was predicted to be 4.97 Kcal/mol lower in energy than the Z config ing that the E configuration (8a) may be the stable configuration, wi hydrogen bond of approximately 2.225 Å ( Figure 5A). The DFT quan lations showed that the calculated 13 C NMR data for the E configu different from the experimental data. Based on the above evidence, on was concluded to be from the E configuration (8a). A summary of these calculated NMR data and their comparison values are presented in Tables 3 and 4, and the correlation coefficie Table 5; these data were used to assign the NMR signals for the Z an N-nitrososarcosine 3-7. Table 6 shows the Gibbs free energy values isomers for compounds 3-7 at the M062X/Def2TZVP level of theor correction. The major calculated molecular models of 3-7 are shown A summary of these calculated NMR data and their comparisons with experimental values are presented in Tables 3 and 4, and the correlation coefficients are presented in Table 5; these data were used to assign the NMR signals for the Z and E configurations of N-nitrososarcosine 3-7. Table 6 shows the Gibbs free energy values (G, Kcal/mol) of Z/E isomers for compounds 3-7 at the M062X/Def2TZVP level of theory with Grimme's D3 correction. The major calculated molecular models of 3-7 are shown in Figure 6.  6. Optimized conformers derived from DFT calculations for asymmetrical N-nitrosamines 3-7. To quantify the ratios of isomers and the changes in the ratio at different temperatures, we carried out VT NMR spectroscopic studies (Figure 7). All ratios of the 3-7 isomers were changed in the VT-NMR experiments. To determine whether these changes were affected by temperature or time, control NMR experiments were performed at room temperature (RT). Figure 7A shows that the Z/E ratio of 3 quickly increased in the VT-NMR experiment. This ratio was maintained at 120~130% even though the NMR probe temperature changed from 90 °C to 30 °C. In the control RT-NMR experiment of 3, the Z/E ratio increased slowly from 2 to 12% within seven hours. A similar phenomenon was also observed in the VT/RT-NMR experiments of 6 ( Figure 7D). The Z/E ratios of 4 and 5 exhibited small changes of approximately 12 and 24%, respectively ( Figure 7B, C). In addition, 7 was shown to exhibit different changes in the Z/E ratios in the VT/RT-NMR experiment, but they ultimately showed a similar Z/E ratio of approximately 50%. Based on these VT/RT-NMR experiments, the rapid changes in the Z/E ratios of isomers 3-7 were temperature-dependent. To our surprise, when the NMR probe temperature returned to 30 °C from higher temperatures, the Z/E ratios did not show a significant decrease. This means there might be a balance between the two isomers in solvents. To quantify the ratios of isomers and the changes in the ratio at different temperatures, we carried out VT NMR spectroscopic studies ( Figure 7). All ratios of the 3-7 isomers were changed in the VT-NMR experiments. To determine whether these changes were affected by temperature or time, control NMR experiments were performed at room temperature (RT). Figure 7A shows that the Z/E ratio of 3 quickly increased in the VT-NMR experiment. This ratio was maintained at 120~130% even though the NMR probe temperature changed from 90 • C to 30 • C. In the control RT-NMR experiment of 3, the Z/E ratio increased slowly from 2 to 12% within seven hours. A similar phenomenon was also observed in the VT/RT-NMR experiments of 6 ( Figure 7D). The Z/E ratios of 4 and 5 exhibited small changes of approximately 12 and 24%, respectively ( Figure 7B, C). In addition, 7 was shown to exhibit different changes in the Z/E ratios in the VT/RT-NMR experiment, but they ultimately showed a similar Z/E ratio of approximately 50%. Based on these VT/RT-NMR experiments, the rapid changes in the Z/E ratios of isomers 3-7 were temperature-dependent. To our surprise, when the NMR probe temperature returned to 30 • C from higher temperatures, the Z/E ratios did not show a significant decrease. This means there might be a balance between the two isomers in solvents. Molecules 2022, 27, x FOR PEER REVIEW 8 of 10 Figure 7. The Z/E ratios of asymmetrical N-nitrosamines 3-7 (A-E, respectively) at different temperatures. Figure 7. The Z/E ratios of asymmetrical N-nitrosamines 3-7 (A-E, respectively) at different temperatures.

NMR Experiments
NMR samples were prepared in DMSO-d6 with 0.03% tetramethylsilane (TMS). The chemical shifts are quoted in ppm relative to TMS. 1 H-NMR and 13 C-NMR spectra were recorded on a 400 MHz Bruker NMR spectrometer (Bruker BioSpin GmbH, Ettlingen, Germany). VT 1 H-NMR spectra were recorded at 30 • C, 40 • C, 50 • C, 60 • C, 70 • C, 80 • C and 90 • C. Before each sample was subjected to the VT 1 H-NMR experiment, it was heated in an NMR probe at the experimental temperature (from 30 • C to 90 • C, then back to 30 • C) for at least 10 min. TopSpin 2.1 software (Bruker BioSpin, Billerica, MA, USA)was used for the acquisition and processing of the NMR data.

Computational Details
A conformational search was performed using Crest software (Loughborough University, Loughborough, The United Kingdom), and the conformers within an energy window of 5 kcal·mol −1 were optimized with DFT calculations at the M062X/Def2TZVP level of theory with Grimme's D3 correction [7] using the Gaussian 09 program (Gaussian, Inc.: Wallingford, CT, USA) [8]. A frequency analysis was performed at the same level of theory to ensure that no imaginary frequencies existed and to determine the Gibbs free energies for the subsequent population analysis. Room-temperature (298.15 K) equilibrium populations were calculated according to the Boltzmann distribution law. Those conformers, accounting for over 99% of the population, were subjected to subsequent calculations.
The GIAO method [9][10][11][12][13] at the mPW1PW91/B3LYP/6-31+G(d, p) level of theory (in DMSO) in corresponding solvents with the IEFPCM solvent model [14] was used for the NMR calculation. The chemical shifts were calculated from shielding constants by referencing TMS at 0 ppm (δ calcd = σ TMS − σ calcd ), where σ TMS is the shielding constant of TMS calculated at the same level of theory. For each possible candidate, the parameters of the linear regression δ cal = a × δ exp + b and the correlation coefficient, R 2 , were determined.
The hydrogen bond energy (E H ) was calculated using the equation E H = E 8a − E 8 , where E 8a is the energy of the conformer without hydrogen bonding by twisting the N-N to break the hydrogen bond, and E 8 is the energy of the optimized conformer [15].

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are contained within the article and supplementary materials.