Isolation and structural characterization of novel thermal degradation impurity of nafcillin sodium using spectroscopic and computational techniques

Thermal degradation of nafcillin sodium ( NS ) gave rise to an unknown impurity. This unknown thermal degradation impurity ( TDI ) was evaluated using a reverse-phase high performance liquid chromatography, where it was eluted at 1.31 relative retention time to NS peak. TDI was isolated using preparative HPLC from degradation mixture and fully characterized using the various spectroscopic techniques like high resolution MS, multidimensional NMR and FTIR. Based on the available spectroscopic data, the probable structure for the impurity is proposed and is named as (2R,4S)-2-((R)-carboxy(2-ethoxy-1-naphthamido)methyl)-3-((2-ethoxy-1-naphthoyl)glycyl)-5,5-dimethylthiazolidine-4-carboxylic acid. The proposed structure is further supported by the density functional theory calculations. In addition, a two-step mechanism of the degradation is proposed. To the best of our knowledge, it is a novel impurity and not reported elsewhere.


Introduction
Nafcillin sodium (NS) is a parenteral, second generation penicillinase-resistant penicillin antibiotic used largely to treat moderate to severe staphylococcal infections.It is a semi-synthetic naphthalene, beta-lactam antibiotic and the chemical structure is shown in Figure 1.These semi-synthetic agents are engineered to be resistant to hydrolysis by most staphylococcal β-lactamases by virtue of a substituted side chain that acts by steric hindrance at the site of enzyme attachment. 1 Its bactericidal activity is triggered by inhibition of bacterial cell wall synthesis by forming covalent bond to one or more of the penicillin binding proteins that play a critical role in the final transpeptidation process.Binding to penicillin-binding proteins inhibits the transpeptidase and carboxypeptidase activities conferred by these proteins and prevents the formation of the crosslinks. 2 This has been linked to rare occurrences of clinically apparent, idiosyncratic liver injury. 3It is administrated orally and by intramuscular and intravenous injections.However, it is poorly absorbed after oral administration, and absorption is further depressed if the drug is given with food.The levels are low even if administered by intramuscular injection.Therefore, most dosing is now intravenous. 4To have a benchmark quality of the drug substance and to meet various guidelines such as International Conference on Harmonization (ICH) and regulatory requirements, it is important to identify the impurities present in the drug substance. 5,6 he requirement of stability testing data is to understand how the quality of a drug substance and drug product changes with time under the influence of various environmental factors. 5,6 is indeed helps in selecting proper formulation and package as well as providing proper storage conditions and shelf life. 7To this end, Prasada Rao et.al. reported the formation of various degradation impurities of NS during the stress and formal stability storage conditions as per ICH requirements. 8Ashline et.al. has reported a novel thietan-2-one degradation impurity in aqueous solution 9 while Jagadeesh et.al. has reported a new degradation impurity under thermal and humidity condition. 10n the present study, NS subjected to degradation under thermal conditions.The HPLC chromatogram of NS showed the formation of an unknown impurity along with four known impurities at the relative retention time (RRT) of around 1.31 with respect to NS peak.The four known impurities obtained at RRT of ~0.49 (14.10 %), ~0.67 (12.05%), and ~0.84 (8.23 %) are found to be Penicilloic acid of Nafcillin, Penilloic acids of Nafcillin-1 and 2 and 2-ethoxy-1-napthoic acid respectively as reported by Prasada Rao et.al. 8Here we present in detail the identification, isolation, structure elucidation and the mechanism of formation of this unknown thermal degradation impurity (TDI).1).The significant addition of 273.1011Da in TDI against NS might be due to the addition of C15H15NO4 (exact mass 273.1001Da).Molecular formula of protonated TDI worked out as C36H38N3O9S + , with mass error, 2.0 ppm (exact mass 688.2323Da).Double bond equivalence (DBE) 19.5 and odd number (three) nitrogen atoms were considered to determine the structure.The chemical structure of TDI was investigated thoroughly using HRMS fragmentation and multi-stage mass fragmentation (MS n ) techniques.The obtained data is presented in Table 1 provided the evidence to the proposed chemical structure of TDI.The possible structures were proposed for fragments of TDI and shown in Figure 2

NMR structural characterization studies
The values of 1 H and 13 C NMR chemical shifts (δ) and proton-proton coupling constants (J) of TDI are presented in Table 2. 1 H and 13 C NMR displayed 37 protons and 36 carbons respectively.DEPT90, DEPT135 and 13 C NMR experiments verified that TDI contains 4-CH3, 3-CH2, 15-CH and 14-C carbons.The experimental number of carbons precisely matched with the theoretically calculated number of carbons at proposed chemical structure of TDI as shown in Figure 2. Based on the MS n fragmentation and the NMR data, the proposed chemical structure is shown in Figure 3.Although this impurity is formed at 1.31 RRT which is very close to the impurity reported by Prasad et.al at 1.26 RRT having the chemical structure TDI-1 10 as shown in Figure 3, the structure proposed here is entirely different.The major difference between these two structures is that TDI has a thiazolidine ring whereas TDI-1 has an open chain structure.This is evidenced by the NMR data that in TDI a doublet signal for proton is observed at 5.85 ppm and 66.02 ppm which correspond to -CH, (sp3 carbon) while in TDI-1, a multiplet signal for proton is observed at 7.33 ppm and 161.4 ppm which corresponds to ethylene =CH, (sp2 carbon) at C25.The proposed structure is further supported by notable HMBC correlations between H2/C4, H9/C7, H23&H23'/C19 and H25/C27, Figure 4. HSQC, COSY and TOCSY data was consistent and further supports the proposed chemical structure of TDI (refer supplementary material for spectra).

Structure confirmation by Computational NMR
To support further the proposed structure of TDI and to show the differences between the TDI and TDI-1 as shown in Figure 3, both the 1 H and 13 C NMR chemical shift values were predicted using the computational techniques.The geometries of both the structures were fully energy minimized at the same level as discussed in the computational details and found to be the true minima on the potential energy surface characterized by the real values for frequencies.The energetic analysis suggests that TDI is relatively more stable than TDI-1 by 6.2 kcal/mol (the calculated Boltzmann weighted population is 100 %).
The calculated 1 H (excluding labile protons) and 13 C NMR chemical shifts were correlated against the experimental values and the statistical parameters are presented in Table 3 and the correlation plots are shown in Figure 5.The complete chemical shift values are presented in Table S1 of supplementary material.From the correlations, it is observed that the correlation coefficient R 2 with respect to protons of TDI is 0.9641 which is close to unity compared to TDI-1 (0.9204).The MAE, MaxE, CMAE and CMaxE of TDI (0.48, 1.48, 0.34 and 1.15) are relatively small compared to TDI-1 (0.54, 3.17, 0.41 and 2.65).Similarly, the R 2 value with respect to carbons of TDI is 0.9956 which is close to unity compared to TDI-1 (0.8589).Also, the statistical parameters of errors are small for TDI compared to TDI-1 (Table 3).In addition, the formation of TDI is evidenced by the 13 C chemical shift value of C25 atom observed at 66.02 ppm and computationally at 70.96 ppm (Table S1) which is in line with the standard range of ~50-75 ppm for >CH−N (sp 3 carbon).This contrasts with the standard chemical shift range of ~145-160 ppm for CH=N sp 2 carbon (165.18calculated, Table S1). 10Taken collectively all the statistical correlations, the given experimental 1 H and 13 C NMR chemical shifts support to the proposed chemical structure of TDI.

IR structural characterization studies
The observed IR spectra of TDI showed a broad peak at 3600 to 2500 cm -1 , 3405.48 cm -1 , 1733.07 to 1515.11 cm -1 corresponding to O−H, N−H and C=O stretching's respectively.The presence of these functional groups confirms and validates the proposed structure of TDI (refer supplementary material for spectra).To support the observed values for the presence of these functional groups, computational simulations were performed at the same level.During the simulations, the vibrational wavenumbers are usually calculated using the simple harmonic oscillator model.Therefore, they are typically larger than the fundamentals observed experimentally and therefore the values are scaled with a scaling factor of 0.9648 as proposed by Merrick et.al. for this computational model. 11The obtained values are presented in Table 4 and observed that the obtained values are within the range of the respective characteristic wavenumbers.This indeed supports the structure of TDI.

Mechanism of the Degradation Reaction
The proposed reaction mechanism for the formation of TDI is shown in Figure 6.The mechanism initiates by the hydrolytic ring opening of -lactam in NS to obtain intermediate NS-1 followed by the self-condensation to produce dimer, NS2 which then undergoes elimination of 5,5-dimethyl-4,5-dihydrothiazole-4-carboxylic acid (DHTC) producing TDI.

Conclusions
The assessment of the novel degradation impurity of nafcillin sodium drug substance under thermal condition is presented.The impurity was isolated and fully characterized by various spectroscopic techniques like LC-MS, NMR and FTIR.Based on the spectroscopic data, probable structure for the degradation impurity is proposed.The proposed structure is further confirmed by the density functional calculations by predicting the NMR chemical shift values and performed statistical correlations against the available experimental data.In addition, a two-step mechanism of the degradation is proposed and involves the opening of the -lactam followed by self-condensation.The present work will help the scientists and industry engaged in the method development and stability studies of Nafcillin.

Experimental Section
Chemicals and Reagents NS sample was procured from RIA International (Hangzhou Dawn Ray Pharmaceutical make).HPLC grade solvents such as acetonitrile (ACN), n-hexane and isopropyl alcohol (IPA) and other chemicals such as ammonium acetate, glacial acetic acid (AcOH), hydrochloric acid (HCl), sodium hydroxide (NaOH), dichloromethane (DCM) and sodium chloride (NaCl) were purchased from Merck (Mumbai, India).HPLC grade purified water was obtained using a Milli-Q water purification system (Millipore Corporation, Billerica, MA, USA).The proton and carbon chemical shift assignments were carried out with the help of two-dimensional (2D) correlation spectroscopy (COSY), total correlation spectroscopy (TOCSY), hetero nuclear single quantum correlation spectroscopy (HSQC), hetero nuclear multiple bond correlation spectroscopy (HMBC).All the experiments were carried out in the phase sensitive mode. 12The spectra were acquired with 2×256 or 2×192 free induction decays (FID) containing 8-16 transients with relaxation delays of 1.0-1.5 sec.The 2D data were processed with Gaussian apodization in both the dimensions.Fourier Transformed Infrared Spectroscopy (FTIR).The FT-IR spectra were recorded in the range of 4000 -400 cm -1 using Perkin-Elmer Spectrum 100 spectrophotometer (Waltham, MA, USA).The sample pellet was prepared using dried KBr powder into a mortar with 1% of the sample.Computational Details.All the calculations were performed using the Gaussian 09 program 13 with hybrid DFT-B3LYP functional 14 in conjunction with 6-31+G(d,p) basis set.This combination of the functional and basis set proved to provide accurate geometries and molecular properties for organic molecules at a reasonably low computational cost. 15The gas phase molecular geometries of both TDI and TDI-1 were energy minimized without any symmetry constraints and with the same stereo-configuration as in NS.The obtained geometries were then subjected to the vibrational frequency analysis at the same Hessian to ensure that the obtained geometries represent minima on the potential energy surface.

Instruments and instrumental parameters
The isotropic 1 H and 13 C NMR shielding constants were calculated by using the Gauge Independent Atomic Orbital (GIAO) 16 approximation.The equilibrium solvation effects were considered through an implicit solvation model for dimethylslufoxide (DMSO) by employing the self-consistent reaction field (SCRF) method in conjunction with the conductor-like polarizable continuum model (C-PCM). 17,18 he chemical shifts were calculated as  = ref−, where ref is the shielding constant of the reference TMS calculated at the same level of theory (ref = 31.6764and 193.5149 respectively for 1 H and 13 C atoms).
To compare the calculated data with the experimental results, a linear fit of calculated versus experimental shifts ( calc =a+b exp ) is performed to obtain the intercept a, slope b and correlation coefficient R 2 .The results were then evaluated in terms of mean absolute error, MAE = Σn|calc -exp|/n, the corrected mean absolute error CMAE = Σn|scaled -exp|/n, where scaled = (calc -a)/b measures the distance between the experimental value and the value predicted by the linear fitting, the maximum error MaxE = max(|calc-exp|) and the corrected maximum error CMaxE = max(|scaled-exp|).It is noted that during the statistical analysis, the labile protons (NH and OH) are excluded from the correlation which would flatten all statistical parameters focusing on a narrower range allowed us to highlight the differences in the region of interest for the comparison.It is remarkable that the sources of variance in the calculated chemical shifts like conformational degrees of freedom in the ethoxy groups and flexibility of hydroxyl groups were neglected during the calculations.

Sample preparation
Degradation reaction and isolation of TDI.To a 250 ml single-neck round bottom flask equipped with magnetic stirrer was added 10 g (0.22 mol) of NS at room temperature (24-25 °C).The solid was heated to 105 °C and allowed to stir under nitrogen atmosphere.Degradation was monitored by HPLC and 13.21% product formation was observed after 88 hrs. Figure 7, top.The reaction was continued till 150 hrs, and no change in the percentage of product formation by HPLC is observed (Figure S1 of supplementary material).After completion of the reaction, 5 g crude solid was collected and dissolved in 60 mL of diluent (50% acetonitrile in water) for purification by preparative HPLC.A suitable preparative HPLC method was developed to separate and isolate TDI from residual drug substance and other impurities present in the mixture.The eluent (aq.acetonitrile) containing TDI was collected and solvent was evaporated.Solid compound was precipitated from the aqueous medium, filtered and dried at room temperature under vacuum.Weight of the resultant compound was about 170 mg which has 97.03% purity by HPLC, Figure 2, bottom.Reverse phase achiral HPLC sample preparation.Weighed and transferred about 5 mg of TDI into a clean 10 mL volumetric flask.To this, 7 mL of diluent Acetonitrile was added and sonicated to dissolve.The volume of the solution was made up to the mark with diluent and mixed well.

Figure 3 .
Figure 3. Proposed structure for TDI along with the reported structure of TDI-1 (1.26 RRT).

Figure 6 .
Figure 6.Proposed reaction mechanism for the formation of TDI.

Figure 7 .
Figure 7. Crude HPLC chromatogram of NS obtained at 88 hrs. of thermal degradation (top) and chromatogram after purification (bottom).

Mass structural characterization studies
LC-MS analysis indicated the presence of protonated ions [M+H] + of TDI and NS drug substance at m/z 688.42 Da and 415.26Da respectively.HRMS data indicated that the molecular ions [M+H] + of TDI and [M+Na] + of NS at m/z 688.2343Da and 437.1152Da (Table . Molecular ion 688.2343 fragmented to 670.2236 due to the water loss in TDI.Molecular ion 688.2343 fragmented to 433.1483 due to the amide bond cleavage at five membered ring.The fragmented ion 433.1483 further fragmented in to 389.1547 with the loss of CO2.The fragment ions 670.2236 and 389.1547 further fragmented and produced a common fragment ion 199.0770 due to the amide bond cleavage.

Table 1 .
HRMS and multi-stage mass fragmentation (MS n ) data of TDI

Table 3 .
13lative energies (kcal/mole) and statistical parameters for the correlations of 1 H and13C NMR chemical shifts of TDI and TDI-1 obtained at the B3LYP/6-31+G(d,p)//B3LYP/6-31+G(d,p) in DMSO solvent R 2 is the correlation coefficient, MAE is the mean absolute error, CMAE is the corrected mean absolute error, MaxE is the maximum error and CMaxE is the corrected maximum error with respect to the linear fit.

Table 4 .
Calculated .2 mL/min and the detection was carried out at 228 nm on a photodiode array (PDA) detector.The compounds were eluted with mobile phases having components A and B where A was 0.1% (v/v) formic acid in water while B was 100% ACN.A timed program of T/%B: 0/25, 40/65, 50/65, 51/25, 60/25 was followed for linear gradient.LC-MS analysis was performed with Thermo Ion Trap Mass spectrometer (Thermo scientific, Waltham, MA, USA) coupled to an UHPLC quaternary gradient to study thermal degradation sample.Multi-stage fragmentation studies performed to evaluate the structures of NS and TDI.MS analysis was performed by applying Heated Electrospray Ionization (HESI) source in positive and negative ion modes.HESI source parameters setting involved source voltage 4.0 kV, capillary temperature 350°C, source heater temperature 350 °C, sheath gas flow 45.0, auxiliary gas flow 20.0, S-Lens RF level 60.0%.The data was acquired using Xcalibur software.MS analysis was performed by applying electrospray ionization (ESI) source in positive ion mode.ESI source parameters setting involved capillary voltage 3.0 kV, source temperature 110 °C, desolvation temperature 250 °C, sampling cone 35 V, extraction cone 4.0 V, and desolvation gas flow 600 L/hour.Collision energy 18 eV was used for TDI to generate fragments.The data was acquired using Masslynx 4.1 software.
Analytical high-performance liquid chromatography.A Waters Alliance HPLC (Waters Corporation, Milford, MA, USA) equipped with PDA detector was employed for the analysis of TDI and NS.The separation of TDI from NS was carried out using YMC ODS Aqua, 150 mm × 4.6 mm column packed with 5 μm particles.The flow rate was 1High resolution Mass Spectrometry (HRMS).HRMS analysis was carried out on SYNAPT G2 Q-TOF Mass spectrometer (Waters Corporation) coupled to an ACQUITY H-Class UPLC having quaternary gradient pump to identify the NS and TDI with accurate mass measurements.Q-TOF-MS/MS analysis was carried out to study the fragmentation patterns for TDI and NS.