Selective Oxidation of Clopidogrel by Peroxymonosulfate (PMS) and Sodium Halide (NaX) System: An NMR Study

A selective transformation of clopidogrel hydrogen sulfate (CLP) by reactive halogen species (HOX) generated from peroxymonosulfate (PMS) and sodium halide (NaX) is described. Other sustainable oxidants as well as different solvents have also been investigated. As result of this study, for each sodium salt the reaction conditions were optimized, and four different degradation products were formed. Three products were halogenated at C-2 on the thiophene ring and have concomitant functional transformation, such as N-oxide in the piperidine group. A halogenated endo-iminium product was also observed. With this condition, a fast preparation of known endo-iminium clopidogrel impurity (new counterion) was reported as well. The progress of the reaction was monitored using nuclear magnetic resonance spectroscopy as an analytical tool and all the products were characterized by 1D-, 2D-NMR and HRMS.


Introduction
The direct insertion of halogens in (hetero)aromatic drugs, in a selective way, has been the object of much interest by the synthetic community [1]. The inclusion of a new C−X bond in these bioactive heterocyclic compounds can improve their physical and biological properties, increase potency, and be used as a handle in the further design and construction of pharmaceuticals [2]. Thiophene rings are five-membered heterocycles bearing sulfur atoms in their structure. Connected to a halogen, halothiophenes represent a class of bioactive molecules with extraordinary pharmacological properties [3,4], including the FDA-approved drugs Avatrombopag, Tioconazole, Lornoxicam, Rivaroxaban, and Brotizolam ( Figure 1A).
In recent years, several methods of direct activation of halogens in organic compounds have been reported using safe halogen sources such as HX, NH 4 X and NaX (X = Cl, Br and I). To sustainably transform these halides into more reactive species, the use of oxidizing agents that conform to the principles of Green Chemistry is essential [1].
In our previous work, we documented the reactive sequence of the oxidationchlorination of Ticlopidine hydrochloride using PMS. We observed the formation of the reactive intermediate species (DP-1) containing a chlorine group at the C-2 carbon of the thiophene and the oxidation of cyclic amine to N-oxide in the piperidinic structure ( Figure 1B) [31]. 2 of 12 tion methods for degradation of active pharmaceutical ingredients, we wanted to understand the oxidative halogenation reaction of heterocycles containing non-hydrohalic acids with safe halogen sources. Here, we show the oxidation of CLP using PMS/sodium halides (NaX, X = Cl, Br and I; Figure 1D). We observed the formation of four interesting classes of products that can predict the degradation of thienopyridines under high salinity media. To the best of our knowledge, detection, characterization, and selective preparation of the products in this oxidative method are first reported herein.

Influence of Chloride
Our experimental work started with the optimization of conditions for the oxidative chlorination of CLP with sodium chloride (NaCl) as a chlorine source, and different oxidant agents (Table 1). We monitored the reaction by NMR and HPLC. In the initial experiments, the treatment of CLP with H2O2 or tert-butyl hydroperoxide (TBHP) in the presence of NaCl did not lead to the formation of products after 24 h at room temperature ( Table 1, Entries 1 and 2). Using PMS in D2O, we observed the complete consumption of CLP and the formation of various degradation products after one minute. Inspired by our Clopidogrel hydrogen sulfate (CLP, Plavix TM ) is another thienopyridine drug that has powerful antiplatelet properties, and it plays an important role in the treatment of coronary, peripheral vascular and cerebrovascular diseases [32][33][34]. Several researchers have reported functional transformations of CLP using halogenated succinimide reagents ( Figure 1C). Padi and coworkers have used N-bromosuccinimide (NBS) for the preparation of endo-iminium impurity DP-2 on a large scale [35]. Jiao and coworkers have developed an efficient method using N-chlorosuccinimide (NCS) and dimethyl sulfoxide (DMSO) as a catalyst for the preparation of 2-Cl-clopidogrel DP-3a [36].
Continuing our efforts in the development and optimization of sustainable prediction methods for degradation of active pharmaceutical ingredients, we wanted to understand the oxidative halogenation reaction of heterocycles containing non-hydrohalic acids with safe halogen sources. Here, we show the oxidation of CLP using PMS/sodium halides (NaX, X = Cl, Br and I; Figure 1D). We observed the formation of four interesting classes of products that can predict the degradation of thienopyridines under high salinity media. To the best of our knowledge, detection, characterization, and selective preparation of the products in this oxidative method are first reported herein. Our experimental work started with the optimization of conditions for the oxidative chlorination of CLP with sodium chloride (NaCl) as a chlorine source, and different oxidant agents (Table 1). We monitored the reaction by NMR and HPLC. In the initial experiments, the treatment of CLP with H 2 O 2 or tert-butyl hydroperoxide (TBHP) in the presence of NaCl did not lead to the formation of products after 24 h at room temperature (Table 1, Entries 1 and 2). Using PMS in D 2 O, we observed the complete consumption of CLP and the formation of various degradation products after one minute. Inspired by our previous work [31], we decided to investigate the co-solvent effect for this reaction. We performed some experiments with dichloromethane-d 2 (CD 2 Cl 2 ), chloroform-d (CDCl 3 ), dimethyl sulfoxide-d 6 ((CD 3 ) 2 SO), benzene-d 6 (C 6 D 6 ), and toluene-d 8 (C 7 D 8 ), but these deuterated co-solvents did not lead to the formation of any products. However, with acetone-d 6 , we observed the formation of an N-oxide product, DP-6 (48%), as well as the chlorinated products DP-3a (4%) and DP-4a (25%) after 5 h (Entry 4). performed some experiments with dichloromethane-d2 (CD2Cl2), chloroform-d (CDCl3), dimethyl sulfoxide-d6 ((CD3)2SO), benzene-d6 (C6D6), and toluene-d8 (C7D8), but these deuterated co-solvents did not lead to the formation of any products. However, with acetone-d6, we observed the formation of an N-oxide product, DP-6 (48%), as well as the chlorinated products DP-3a (4%) and DP-4a (25%) after 5 h (Entry 4).

Influence of Bromide
In parallel to the chlorination reaction, we submitted CLP to the oxidation process using NaBr as a bromine source ( Table 2). We observed a quick conversion of the brominated products DP-3b and DP-5b with all solvents studied. Then, we extended our study varying the amount of PMS and NaBr. The best condition for the formation of DP-3b was with D2O:CD3CN (2:1), PMS (0.5 equiv), and NaBr (1.0 equiv; Table 2, Entry 6). The combination of PMS (1.5 equiv) and NaBr (1.0 equiv;  To investigate the selectivity between chlorinated products DP-3a and DP-4a, the same reaction was performed with other polar solvents (Entries 5-7): methanol-d 4 (32%), THF-d 8 (51%) and acetonitrile-d 3 (51%). We observed increased formation of DP-4a, as well as small amounts of DP-3a. With the optimized solvent in hand, we extended the method to optimize DP-3a or DP-4a using varying amounts of PMS and NaCl (Entries 8-13). We did not obtain DP-3a with a conversion higher than 20%. Thus, our best conditions for DP-4a formation used a mixture of D 2 O:CD 3 CN (2:1), PMS (2.0 equiv), and NaCl (2.0 equiv; Entry 8).

Influence of Bromide
In parallel to the chlorination reaction, we submitted CLP to the oxidation process using NaBr as a bromine source ( Table 2). We observed a quick conversion of the brominated products DP-3b and DP-5b with all solvents studied. Then, we extended our study varying the amount of PMS and NaBr. The best condition for the formation of DP-3b was with D 2 O:CD 3 CN (2:1), PMS (0.5 equiv), and NaBr (1.0 equiv; Table 2, Entry 6). The combination of PMS (1.5 equiv) and NaBr (1.0 equiv; Table 2, Entry 8) provided DP-5b in excellent yield.

Influence of Iodide
When we used the iodide reagent NaI and other co-solvents (Entries 1-3), solubility problems were observed after the in situ oxidative conversion of the iodide to the reactive iodine species was generated. With NaCl, NaBr and NaI, acetonitrile-d3 was the most efficient deuterated co-solvent. Unlike with the halides described above, no substitution at the thiophene ring occurred (no formation of DP-3c, entries 4 and 5). Instead, we discovered a fast method (as compared to Padi's) [36] to prepare DP-2 in excellent yield by using the PMS/NaI system ( Table 2, Entry 4). Table 3. Screening of conditions for oxidative iodination reaction. (a).

With NaCl
In Figure 2, a compilation of the 1 H NMR spectra on the main functional transformations that occurred in the oxidative process, which resulted in the formation of the products in high yields, is shown. In addition, the data of the starting materials (as

Influence of Iodide
When we used the iodide reagent NaI and other co-solvents (Entries 1-3), solubility problems were observed after the in situ oxidative conversion of the iodide to the reactive iodine species was generated. With NaCl, NaBr and NaI, acetonitrile-d 3 was the most efficient deuterated co-solvent. Unlike with the halides described above, no substitution at the thiophene ring occurred (no formation of DP-3c, entries 4 and 5). Instead, we discovered a fast method (as compared to Padi's) [35] to prepare DP-2 in excellent yield by using the PMS/NaI system (Table 3, Entry 4).

Influence of Iodide
When we used the iodide reagent NaI and other co-solvents (Entries 1-3), solubility problems were observed after the in situ oxidative conversion of the iodide to the reactive iodine species was generated. With NaCl, NaBr and NaI, acetonitrile-d3 was the most efficient deuterated co-solvent. Unlike with the halides described above, no substitution at the thiophene ring occurred (no formation of DP-3c, entries 4 and 5). Instead, we discovered a fast method (as compared to Padi's) [36] to prepare DP-2 in excellent yield by using the PMS/NaI system ( Table 2, Entry 4). Table 3. Screening of conditions for oxidative iodination reaction. (a).

With NaCl
In Figure 2, a compilation of the 1 H NMR spectra on the main functional transformations that occurred in the oxidative process, which resulted in the formation of the products in high yields, is shown. In addition, the data of the starting materials (as

With NaCl
In Figure 2, a compilation of the 1 H NMR spectra on the main functional transformations that occurred in the oxidative process, which resulted in the formation of the products in high yields, is shown. In addition, the data of the starting materials (as reference) and non-chlorinated intermediates DP-6 (mixture of diastereomers) are also included in this compilation of spectra, as described in Figure 2A,B [37]. The main observation among the NMR spectra of this work is the disappearance of the doublet (δ H~6 .7 ppm, CD 3 CN) of proton H-3 of the thiophene ring and a conversion into a singlet in that same region which results from the insertion of a heteroatom at carbon-2 of this heterocycle ( Figure 2C). reference) and non-chlorinated intermediates DP-6 (mixture of diastereomers) are also included in this compilation of spectra, as described in Figures 2A and 2B [38]. The main observation among the NMR spectra of this work is the disappearance of the doublet (δH ~6.7 ppm, CD3CN) of proton H-3 of the thiophene ring and a conversion into a singlet in that same region which results from the insertion of a heteroatom at carbon-2 of this heterocycle ( Figures 2C). Figure 2C represents the progress of oxidative chlorination of CLP and NaCl after two hours (Table 1, Entry 12). In the first two hours of reaction, we observed the appearance of non-halogenated diastereomers of intermediate DP-6 with a new N + -OH bond, as well as the mono-chlorinated compound DP-3a. As the reaction progressed towards the production of DP-4a, the signals of DP-6 and DP-3a decreased in the spectrum. This led us to conclude that the reaction pathway to obtain DP-4a occurred via two simultaneous processes: oxidation-chlorination via DP-6 and chlorination-oxidation via DP-3a ( Figure  3). The final product, DP-4a, is represented in Figure 4D, and bears a chlorine on the thiophene ring and an N-oxide in the piperidine structure. The mechanism of this reaction was shown in our previous work [31].  Table 1, Entry 12); (D) Crude 1 H NMR spectrum of diastereomers DP-4a after 12 h (see Table 1, Entry 8).  Table 1, Entry 12); (D) Crude 1 H NMR spectrum of diastereomers DP-4a after 12 h (see Table 1, Entry 8). Figure 2C represents the progress of oxidative chlorination of CLP and NaCl after two hours (Table 1, Entry 12). In the first two hours of reaction, we observed the appearance of non-halogenated diastereomers of intermediate DP-6 with a new N + -OH bond, as well as the mono-chlorinated compound DP-3a. As the reaction progressed towards the production of DP-4a, the signals of DP-6 and DP-3a decreased in the spectrum. This led us to conclude that the reaction pathway to obtain DP-4a occurred via two simultaneous processes: oxidation-chlorination via DP-6 and chlorination-oxidation via DP-3a (Figure 3). The final product, DP-4a, is represented in Figure 4D, and bears a chlorine on the thiophene ring and an N-oxide in the piperidine structure. The mechanism of this reaction was shown in our previous work [31].

With NaBr
Contrary to the PMS/NaCl process, the reaction with NaBr did not present N-oxi products, but a C-2 halogen/endo-iminium DP-5b when PMS (1.0 equiv) was used ( Figu  4B and 4D). To understand the development of this reaction, we started with 0.5 equiv lent of PMS (Table 2, Entry 6; Figure 4A) and observed a mixture between CLP and mon brominated product DP-3b. In addition, we increased the amount of PMS to provide D 5b in a quantitative yield ( Table 2, Entry 8; Figure 4D).    Table 1, Entry 8).

Figure 3.
Reaction progress for oxidative chlorination reaction of CLP with PMS/NaCl system measured by 1 H NMR (see Table 1, Entry 8).

With NaBr
Contrary to the PMS/NaCl process, the reaction with NaBr did not present N-oxide products, but a C-2 halogen/endo-iminium DP-5b when PMS (1.0 equiv) was used ( Figure 4B,D). To understand the development of this reaction, we started with 0.5 equivalent of PMS (Table 2, Entry 6; Figure 4A) and observed a mixture between CLP and mono-brominated product DP-3b. In addition, we increased the amount of PMS to provide DP-5b in a quantitative yield ( Table 2, Entry 8; Figure 4D).

With NaI
With NaI, we tried to obtain an iodinated product, DP-3c, under our experimental conditions. Instead, we observed the formation of endo-iminium DP-2 in high yield ( Table 3, Entry 4; Figure 5B). Reducing the molar amounts of PMS, we detected a mixture between CLP and endo-iminium DP-2 (Table 3, Entry 5, Figure 5A).

With NaI
With NaI, we tried to obtain an iodinated product, DP-3c, under our experimental conditions. Instead, we observed the formation of endo-iminium DP-2 in high yield ( Table  3, Entry 4; Figure 5B). Reducing the molar amounts of PMS, we detected a mixture between CLP and endo-iminium DP-2 (Table 3, Entry 5, Figure 5A).  Table 3, Entry 4.

Characterization of the Products.
Degradation products DP-2, DP-3a, DP-4a, DP-3b and DP-5b were characterized directly from the reaction mixtures by HPLC-MS, HRMS, as well as by 1D-and 2D-NMR spectroscopy. Two-dimensional correlations 1 H-1 H COSY, 1 H-13 C HSQC and 1 H-13 C HMBC were used for the elucidation of the structure.

Characterization of DP-2
Using NMR spectrometry (in CD3CN, see Table 4 and SI), fifteen protons were detected, a value consistent with the molecular formula of DP-2. In addition, the bidimensional 1 H-1 H COSY spectrum showed important correlations for structural elucidation of this molecule, including a cross peak between endo-iminium H-4 (δH 8.81, δC 162.40) and the singlet for H-10, which is part of a CH group (δH 6.39, δC 72.08). Shift data of the latter are similar to that observed for this position in the other molecules (Table 4). This feature proves that an endo-iminium compound, and not an exo-iminium one, is indeed formed. The edited HSQC exhibited the CH2 groups H-6 (δH 4.30-4.23 and 3.95-3.86, δC 49.94) and H-7 (δH 3.52-3.36, δC 24.04). With HRMS/ESI-TOF analysis, the molecular formula of DP-2 was determined as C16H15ClNO2S + , with m/z calculated at 320.0517, and m/z observed at 320.0516 (−0.3 ppm error), indicating loss of one hydrogen atom in the molecule and the formation of the endo-iminium product. These analyses do not show a halogenation bond in the molecule and confirm the structure of this degradation product.  Table 3, Entry 4.

Characterization of the Products
Degradation products DP-2, DP-3a, DP-4a, DP-3b and DP-5b were characterized directly from the reaction mixtures by HPLC-MS, HRMS, as well as by 1D-and 2D-NMR spectroscopy. Two-dimensional correlations 1 H-1 H COSY, 1 H-13 C HSQC and 1 H-13 C HMBC were used for the elucidation of the structure.

Characterization of DP-2
Using NMR spectrometry (in CD 3 CN, see Table 4 and SI), fifteen protons were detected, a value consistent with the molecular formula of DP-2. In addition, the bidimensional 1 H-1 H COSY spectrum showed important correlations for structural elucidation of this molecule, including a cross peak between endo-iminium H-4 (δ H 8.81, δ C 162.40) and the singlet for H-10, which is part of a CH group (δ H 6.39, δ C 72.08). Shift data of the latter are similar to that observed for this position in the other molecules (Table 4). This feature proves that an endo-iminium compound, and not an exo-iminium one, is indeed formed. The edited HSQC exhibited the CH 2 groups H-6 (δ H 4.30-4.23 and 3.95-3.86, δ C 49.94) and H-7 (δ H 3.52-3.36, δ C 24.04). With HRMS/ESI-TOF analysis, the molecular formula of DP-2 was determined as C 16 H 15 ClNO 2 S + , with m/z calculated at 320.0517, and m/z observed at 320.0516 (−0.3 ppm error), indicating loss of one hydrogen atom in the molecule and the formation of the endo-iminium product. These analyses do not show a halogenation bond in the molecule and confirm the structure of this degradation product.

Characterization of DP-3a
The molecular formula of product DP-3a was established as C 16 H 11 O 2 NSCl 2 by HRMS/ESI-TOF data in which m/z of 356.0279 was observed and was calculated for [M + H] + 356.0288 (2.5 ppm error), indicating the insertion of a new chlorine atom into the molecule. In the NMR spectrum (in CD 3 OD, see Table 4 and SI), the addition of the chlorine atom at carbon 2 resulted in an absence of the H-2 signal and the appearance of the singlet associated with H-3 (δ H 6.50, δ C 125.54) consistent with the molecular formula of DP-3a. 1

Characterization of DP-4a
The oxidative chlorinated product DP-4a was found to have the molecular formula C 16  . This mass can also be seen in two retention times in the HPLC-MS chromatogram of Figures S1 and S2 (see SI) indicating the formation of two diastereomers with N + -OH bonds. In comparison to the data of 1 H NMR with CLP, and based in our previous report about DP-6 ( Figure 2B,D in CD 3 CN; also see ref. [37]), it is possible to notice a significant change in the chemical shift in H-4 (δ H 6.62 and 6.52 ppm; δ C 62.50, 62.27 ppm) due to the asymmetric electric field (AMEF) generated in the molecule. This electrical field is the result of the polarization of the molecule through the dipole N + -OH generated in this oxidative process.

Characterization of DP-3b
Similar to the results seen for the compound DP-3a, the product DP-3b showed the molecular formula C 16 Table 4 and SI), the inclusion of a bromine atom at carbon 2 resulted in an absence of the H-2 signal and the appearance of a signal associated with H-3 (δ H 6.70, δ C 129.36), consistent with the molecular formula of DP-3b. In addition, other changes in chemical shift in H-4 (δ H 3.63 and 3.53, δ C 51.10), H-6 (δ H 2.95-2.80, δ C 49.21) and H-7 (δ H 2.77-2.74, δ C 26.02) were noted. HMBC correlations from proton signals for H-3, H-4, H-6 and H-10 were observed.

Characterization of DP-5b
Finally, in the NMR spectrum (in CD 3 OD, see Table 4 and SI), fourteen protons were observed, a value consistent with the molecular formula of DP-5b. In addition, HSQC showed the existence of an endo-iminium group in the piperidine moiety as a singlet (δ H 9.05, δ C 162.47), two methylene groups in H-6 (δ H 4.44-4.37 and 4.00 -3.91, δ C 49.65), H-7 (δ H 3.59-3.40, δ C 24.38) and H-10 protons (δ H 6.62, δ C 72.72). HMBC correlations from proton signals for H-3, H-4, H-6 and H-10 were observed. Also, the molecular formula of DP-5b was determined to be C 16 H 14 BrClNO 2 S + by HRMS/ESI-TOF, with an observed m/z of 397.9622 and a calculated m/z of 397.9623 (0.3 ppm error) indicating the insertion of a bromine atom into the molecule, as well as the formation of the endo-iminium. This analysis corroborated with the data obtained from the NMR spectra.

Nuclear Magnetic Resonance Spectroscopy
All NMR spectra were recorded on a Bruker AVANCE III HD 400 MHz spectrometer at 297 K in D 2 O/CD 3 OD or D 2 O/CD 3 CN (2:1, v/v) solvents. 19 mg of CLP were dissolved in 0.6 mL of deuterated solvent mixture and used for 1 H, 13 C NMR, 1 H-1 H COSY, HSQC and HMBC analysis. Chemical shifts are reported in ppm (δ) and residual CD 3 OD (δ H = 3.31 ppm, δ C = 49.0 ppm) or CD 3 CN (δ H = 1.94 ppm, δ C = 1.32 ppm). Processing of the raw data was performed using Bruker TOPSPIN software.

Mass Spectrometry
The compounds were dissolved (about 0.05 mg·mL −1 ) in acetonitrile, using a solvent system of acetonitrile: formic acid, 0.1% in water [90:10, v/v] at a flow rate of 0.5 mL·min −1 . The mass spectrum of the isolated products was acquired on a Xevo G2-XS Tof Mass Spectrometry instrument from Waters (Wilmslow, UK) in positive spray ionization (ES+) mode. The column used was an ACQUITY UPLC BEH C18 1.7 µm and with the following dimensions: 2.1 mm × 50 mm. The ES+ capillary was set at 3.0 kV, the source temperature at 120 • C and the desolvation temperature at 500 • C. Mass range was scanned between 50 and 750 amu.

General Description of the Experiment
To a glass vial with a solution of CLP (19 mg; 45.49 µmol; 1.0 equiv) in D 2 O/CD 3 CN (0.6 mL, 2:1 v/v) was added the corresponding halide salt (1.0 equiv) and PMS (2KHSO 5 ·KHSO 4 ·K 2 SO 4 , MW = 614.74 g·mol −1 )-see Tables 1-3 for specific data. The solution was stirred before being transferred to an NMR tube and continuously monitored by 1 H NMR at room temperature. After complete conversion, the reaction mixture was quenched with sodium thiosulfate and extracted with dichloromethane (3×). The organic layers were combined, dried over anhydrous MgSO 4 , filtered, and concentrated in vacuo. The desired halogenated products were then characterized without further purification. Characterization of the products by LC-MS is also possible directly from the reaction mixture.

Conclusions
In summary, we showed a selective transformation of clopidogrel hydrogen sulfate (CLP) by a PMS/halide system in aqueous acetonitrile media without employing a metal catalyst. With this method, we have prepared three major halogenated products using different halide salts. With this condition, a fast preparation of known endo-iminium clopidogrel impurity (DP-2, new counterion) was described as well. The new degradation products DP-3a-b, DP-4a and DP-5b were characterized using spectroscopic techniques (namely 1D-NMR, 2D-NMR and HRMS). We believe that this procedure is not only useful for generating clopidogrel derivatives but also very important to the study of drug degradation under hypersaline conditions. We are currently extending this study with other active pharmaceutical ingredients.