Practical and Economical Implementation of Online ... - ACS Publications

Oct 11, 2013 - In order to improve the efficiency, high-throughput identification based on LC-MS strategy is being widely adopted.(1-3) However, the s...
1 downloads 7 Views 3MB Size
Download PDF
Article pubs.acs.org/ac

Practical and Economical Implementation of Online H/D Exchange in LC-MS Ravi P. Shah,*,† Amit Garg,‡,■ Siva Prasad Putlur,‡ Santosh Wagh,‡ Vineet Kumar,‡,▲ Venugopala Rao,†,■ Saranjit Singh,§ Sandhya Mandlekar,⊥ and Sridhar Desikan∥ †

Analytical Research and Development, Pharmaceutical Development, Biocon Bristol-Myers Squibb R&D Center, Syngene International Ltd., Bangalore, India ‡ Pharmaceutical Candidate Optimization, Biocon Bristol-Myers Squibb R&D Center, Syngene International Ltd., Bangalore, India § Department of Pharmaceutical Analysis, National Institute of Pharmaceutical Education and Research (NIPER), S.A.S. Nagar, India ⊥ Pharmaceutical Candidate Optimization, Biocon Bristol-Myers Squibb R&D Center, Bristol-Myers Squibb India Pvt. Ltd., Bangalore, India ∥ Analytical Research and Development, Pharmaceutical Development, Biocon Bristol-Myers Squibb R&D Center, Bristol-Myers Squibb India Pvt. Ltd., Bangalore, India S Supporting Information *

ABSTRACT: Structural elucidation is an integral part of drug discovery and development. In recent years, due to acceleration of the drug discovery and development process, there is a significant need for highly efficient methodologies for structural elucidation. In this work, we devised and standardized a simple and economical online hydrogen−deuterium exchange methodology, which can be used for structure elucidation purposes. Deuterium oxide (D2O) was infused as a postcolumn addition using the syringe pump at the time of elution of the analyte. The obtained hydrogen/deuterium (H/D) exchange spectrum of the unknown analyte was compared with the nonexchanged spectrum, and the extent of deuterium incorporation was delineated by using an algorithm to deconvolute partial H/D exchange, which confirmed the number of labile hydrogen(s) in the analyte. The procedure was standardized by optimizing flow rates of LC output, D2O infusion, sheath gas, and auxiliary gas using the model compound sulfasalazine. The robustness of the methodology was demonstrated by performing sensitivity analysis of various parameters such as concentrations of analyte, effect of matrices, concentrations of aqueous mobile phase, and types of LC modifiers. The optimized technique was also applied to chemically diverse analytes and tested on various mass spectrometers. Moreover, utility of the technique was demonstrated in the areas of impurity profiling and metabolite identification, taking pravastatin-lactone and N-oxide desloratidine, as examples.

S

early stages of drug discovery. Unambiguous structure elucidation of impurities, degradants, and metabolites is becoming increasingly important to understand potential toxicity liabilities earlier in the development to avoid latephase development surprises. Hydrogen−deuterium (H/D) exchange in LC-MS is a powerful technique capable of

tructure elucidation of impurities, degradants, and metabolites is critical in the development of new drug candidates. The conventional approach for structure elucidation, which involves separation of the desired analyte from a mixture followed by isolation, enrichment, and synthesis for spectral analysis, is tedious. In order to improve the efficiency, highthroughput identification based on LC-MS strategy is being widely adopted.1−3 However, the structures proposed by these techniques with limited information such as molecular ion peak or a few fragments are tentative and fit-for-purpose suitable for © 2013 American Chemical Society

Received: July 27, 2013 Accepted: October 11, 2013 Published: October 11, 2013 10904

dx.doi.org/10.1021/ac402339s | Anal. Chem. 2013, 85, 10904−10912

Analytical Chemistry

Article

Table 1. Intensity Correction Using the Equation for Sulfasalazine

relative intensity m/z

a

before online H/D after online H/D

corrected

calculation as per the equationa

possible ions after H/D exchange

5.33 − 0 = 5.33 8.04 − [(5.33 × 22.0)/100] = 6.86

[M] + H+ [M] + D+; [M − Ha+D] + H+; [M − Hb + D] + H+; [M − Hc + D] + H+ [M − Ha + D] + D+; [M − Hb + D] + D+; [M − Hc + D] + D+; [M − Ha − Hb + 2D] + H+; [M − Ha − Hc + 2D] + H+; [M − Hb − Hc + 2D] + H+ [M − Ha − Hb + 2D] + D+; [M − Ha − Hc + 2D] + D+; [M − Hb − Hc + 2D] + D+; [M − Ha − Hb − Hc + 3D] + H+

399 400

100.00 (M) 22.00 (M + 1)

5.33 8.04

5.33 6.86

401

7.43 (M + 2)

23.07

21.16

23.07 − [{(6.86 × 22.0)/100} + {(5.33 × 7.43)/100}] = 21.16

402

1.77 (M + 3)

55.75

50.49

403

0.44 (M + 4)

100.00

87.20

55.75 − [{(21.16 × 22.0)/100} + {(6.86 × 7.43)/100} + {(5.33 × 1.77)/100}] = 50.49 100 − [{(50.49 × 22.0)/100} + {(21.16 × 7.43)/100} + {(6.86 × 1.77)/100}] = 87.20

[M − Ha − Hb − Hc + 3D] + D+

For calculation, n = 3 was considered, as peak intensity (0.44) of m/z + 4 was ignored.

enon, where previously exchanged deuterium reverts back to hydrogen may occur, necessitating the need to maintain the LC system at extremely low temperatures and developing a short LC method. Keppel et al. reported a modified/refrigerated LC system to get efficient H/D exchange data,13 whereas Wu et al. reported the advantage of UPLC over HPLC in back exchange by decreasing run time.14 In other reported techniques, ND3 or CH3OD were used as the nebulizer gas or part of the curtain gas to achieve effective H/D exchange.15−17 However, additional setup and expensive deuterated gas are limiting factors for wide acceptance of this technique. Lam and Ramanathan reported the use of a sheath liquid inlet of LCMS LCQ for infusion of D2O using a separate LC pump.18 The D2O was infused at a constant rate of 0.2 or 0.4 mL/min throughout the run. A similar technique was also reported using a dual-spray API source and an additional LC pump to infuse D2O at a 0.6 mL/min rate.19 These setups can only be achieved with an additional LC pump in Thermo LCQ or instruments equipped with a sheath liquid inlet or dual spray. In this work, we describe experimental efforts in achieving an easy-to-adopt online H/D exchange methodology that can be utilized for structure elucidation of unknowns. Instrument method conditions and an algorithm to deconvolute partial H/ D exchange are explained in detail with sulfasalazine as a model compound possessing diverse functional groups (−NH, −OH, and −COOH).

determining presence, number, and position of exchangeable hydrogen(s) in unknowns, enabling their confirmation.4 The principle behind the technique is based on the mass difference of hydrogen (1 amu) and deuterium (2 amu) as labile hydrogen(s) [e.g., −OH, −SH, −N(R)H, and −COOH] present in the molecule is replaced by deuterium in solution or gas phase, yielding a higher m/z, corresponding to the number of labile Hs.5 In the literature, a few reports are available to carry out H/D exchange LC-MS studies using the deuterated mobile phase.6−8 In a conventional H/D exchange experiment reported by Siegel, 2−15 μL of sample was injected, preceded and followed by 50 and 100 μL of D2O, respectively, which was further infused to a nondeuterated solvent carrier stream for mass analyses in a TSP analyzer. The system was named as “sandwiched slug injection technique”.9 Similar types of sample injection techniques such as direct inlet, flow injection, and syringe pump were reported by Ohashi et al.10 However, both of these techniques are not applicable for unknown trace level analytes present in samples, where LC separation is essential. The most widely used technique involves deuterated mobile phase, where the analyte elutes in a deuterated solvent environment.6 The requirement for equilibrating and running the chromatography systems with deuterated solvents makes the technique difficult to adopt and expensive. Additionally, differences in polarity of the deuterated and aqueous mobile phases and related retention time shifts may impact structure elucidation of the unknowns.5 In order to overcome high usage of expensive D2O, Liu et al. reported online H/D exchange using a narrow-bore column.11 However, the development of all analytical methods on a narrow-bore column is difficult to adopt. Another reported H/D exchange method includes incubation of sample in a deuterated solvent, followed by injection into a chromatographic system utilizing conventional nondeuterated solvents.12 In this case, back-exchange phenom-

■

HYPOTHESIS AND EQUATION FOR PARTIAL EXCHANGE In order to minimize usage of a deuterated solvent, its postcolumn addition is proposed in this study. The instrument setup was modified in a simple manner by incorporating a tee union before the MS source. The mobile phase eluting from HPLC and the infused D2O from the syringe pump are mixed at the tee union and advanced to the MS source. Due to the 10905

dx.doi.org/10.1021/ac402339s | Anal. Chem. 2013, 85, 10904−10912

Analytical Chemistry

■

presence of deuterated and nondeuterated solvents in the MS source, partial H/D exchange would be achieved for the analyte. For instance, if acetonitrile (ACN) and water (H2O) are used as LC mobile phases and D2O is added postcolumn, small droplets of the solvent mixture contain ACN, H2O, HOD, D2O, and the analyte. However, in the mass source, solvents evaporate based on their boiling points. Thus, a major portion of ACN [bp 81.6 °C] would evaporate first, followed by H2O (bp 100 °C), HOD and D2O (bp 101.22 °C). As a result, the solvent mixture would be enriched with HOD and D2O, and more analyte molecules would come in contact with D2O, which would enable further exchange of labile H with D. This is explained in the pictorial presentation in the graphical abstract. That said, the hypothesis ignores the formation of azeotropic mixtures during the evaporation of solvents. Thus, a partial H/D exchange is a likely outcome, considering the short residence time of D2O with the analyte and the presence of H2O molecules in the MS source. In order to understand the partially exchanged analyte spectra, the isotopic contribution of previous peak(s) in each peak needs to be subtracted. The following equation is proposed to accurately estimate the extent of H/D exchange, to determine the total number of exchangeable hydrogens, irrespective of the functional groups [HD( mz )]

=H

O

D( m z)

Article

EXPERIMENTAL SECTION

Chemicals. Sulfasalazine, verapamil, ketoconazole, pimozide, haloperidol, ranitidine, metoprolol, isoniazid, labetolol, ornithine, terbutaline, pravastatin, desloratadine, and D2O were purchased from Sigma Aldrich (St. Louis, MO). Ritonavir was obtained as USP reference standard. HPLC grade acetonitrile (ACN), methanol, formic acid, acetic acid, and ammonia solution were purchased from Merck Specialties Private Limited (Mumbai, India). Ammonium acetate and ammonium formate were obtained from S D Fine-Chem Limited (Mumbai, India). The excipients used in the study were methocel E4M (Colorcon, Goa, India), lutrol E300 (BASF, Germany), mannitol (BASF, Germany), solutol HS-15 (BASF, Germany), carboxymethyl cellulose sodium salt (St. Louis, MO), povidone (BASF, Germany), simple syrup (Humco, TX), and sodium lauryl sulfate (BASF, Germany). Instruments. LC-MS systems used in the study were as follows: (i) linear ion trap LC-MS LTQ XL (Thermo Scientific, San Jose, CA) equipped with Shimadzu HPLC as the front end; (ii) LC-MS Orbitrap (Thermo Scientific, Bremen, Germany) equipped with Agilent 1200 HPLC; (iii) QTRAP 5500 (AB Sciex, Ontario, Canada) equipped with Waters ACQUITY UPLC; (iv) triple-quadrupole LC-MS (API 4000, AB Sciex, Toronto, Canada) equipped again with ACQUITY UPLC, and (v) single-quadrupole LC-MS (ACQUITY SQD, Waters, Manchester, U.K.) also equipped with ACQUITY UPLC. Optimization Studies. Optimization studies were performed using a 0.01 mg/mL sulfasalazine solution in ACN:H2O (50:50 v/v) with three specific objectives: (1) D2O infusion inlet as postcolumn addition either through sheath liquid inlet or through a tee union; (2) ratio of LC flow rate to D2O infusion rate; and (3) sheath and auxiliary gas flow. In order to investigate postcolumn addition mode, 4 μL of sulfasalazine solution was injected to LC-MS LTQ-XL using union instead of column and ACN:H2O (50:50 v/v) as the mobile phase. The LC flow was set at 100 μL/min, and D2O was pumped at a rate of 126 μL/min through the in-built syringe pump. LC flow and D2O infusion rate were optimized by varying LC flow from 50 to 250 μL/min and D2O infusion from 50 to 126 μL/min through a tee union. Effect of sheath and auxiliary gas flow was studied by maintaining LC flow rate at 50 and 100 μL/min and constant D2O flow rate at 126 μL/min. Robustness Studies. For robustness studies, five parameters, viz., analyte concentrations, aqueous mobile phase concentrations, aqueous LC modifiers, organic LC modifiers, and different matrices were selected. For all the parameters, except different matrices, 4 μL of sulfasalazine solution was passed through union at a 100 μL/min LC flow rate. The concentration of sulfasalazine was varied from 0.0005 to 0.1 mg/mL for understanding the effect of analyte concentration, whereas it was selected as 0.01 mg/mL for all other parameters. The flow rate of D2O through tee union was kept constant at 126 μL/min. In order to understand the effect of aqueous mobile phase concentrations, the composition of ACN:H2O was varied from 5:95 v/v to 90:10 v/v. For aqueous LC modifiers and organic LC modifiers, aqueous phase was replaced by 10 mM ammonium formate, 10 mM ammonium acetate, 0.1% formic acid, and 0.1% NH3, whereas MeOH was selected in place of ACN. In order to investigate the effect of various matrices, spiking studies were done. Pharmaceutical excipients were weighed in equal quantity (1 mg each) and mixed through a mortar and pestle for preparation of the

⎡ n [H m ][I ] ⎤ D( z − i) i ⎥ − ⎢∑ ⎢⎣ i = 1 ⎥⎦ 100

where [HD(m/z)] is the corrected relative intensity of m/z after H/D exchange, [HOD(m/z)] is the observed relative intensity of m/z after H/D exchange, [HD((m/z)−i)] is the corrected relative intensity of m/z − i peak after H/D exchange, Ii is the relative intensity observed during nondeuterated analysis for ith isotopic abundance, and n is the total number of isotopic peaks considered from nondeuterated analysis. As per the equation, isotopic abundance of all previous mass peaks (m/z − 1; m/z − 2 .... m/z − n), contributing in m/z peak intensity, could be subtracted effectively. As a result, corrected relative intensity of the m/z peak would only reflect incorporation of D atom into the ion structure, as clarified in Table 1 for sulfasalazine (m/z: 399). The mass spectrum of sulfasalazine contains five isotopic peaks of m/z 399, 400, 401, 402, and 403 with relative intensities of 100, 22.00, 7.43, 1.77, and 0.44 corresponding to M, M+1, M+2, M+3, and M+4 isotopes, respectively. During online H/D exchange, all five isotopes exchange their labile hydrogen(s) with deuterium(s), irrespective of their isotopic mass and relative intensities. For understanding the equation for corrected relative intensity, the relative intensity of m/z 400 before and after H/D exchange is illustatrated here: Before H/D exchange, the relative intensities of the isotope m/z 400 (M+1) and m/z 399 (M) are 22.00 and 100%, respectively. After online H/D exchange, the relative intensity of m/z 400 is 8.04%. This includes contribution from the unexchanged M+1 isotope (m/z 400) and singly exchanged M isotope (m/z 399). The contribution from the M+1 isotope would be 22% of 5.33 (1.17), as relative intensity of m/z 399 is 5.33 after the H/D exchange. Thus, the actual/corrected relative intensity of singly exchanged M isotope would be 6.86 (1.17 subtracted from 8.04). Similarly, the calculation of corrected relative intensities of the other m/z are tabulated in Table 1. 10906

dx.doi.org/10.1021/ac402339s | Anal. Chem. 2013, 85, 10904−10912

Analytical Chemistry

Article

Figure 1. Effect of ratio of LC output vs D2O infusion. (a) LC flow 50 μL/min. (b) LC flow 100 μL/min. (c) LC flow 150 μL/min. (d) LC flow 200 μL/min. (e) LC flow 250 μL/min. (f) LC flow, D2O infusion flow, and corrected relative intensity of m/z 403.

impurity was analyzed on LC-MS using a Zorbax C-8 column (250 mm × 4.6 mm, 5 μ). For chromatographic elution, ACN (A) and H2O (B) were varied in a linear gradient program (Tmin/A:B; T0−5/10:90; T30/60:40; T35/70:30; T36/10:90). The LC flow was set at 0.4 mL/min, which was split to 1/4th using Dionex fixed volume splitter after elution from the column. Hence, the flow moving toward the mass spectrometer was 100 μL/min. The D2O was infused just before the retention time of the impurity through a tee union at 126 μL/ min flow. The D2O was infused through a tee union at a flow rate of 126 μL/min, just before the retention time of the impurity. Metabolite Identification. For metabolite identification, the mixture of 1 mL in vitro rat liver microsomal (RLM) suspension (1 mg/mL), desloratadine (30 μM), and 100 mM phosphate buffer (pH 7.4) was preincubated at 37 °C for 5 min. Subsequently, 1 mM NADPH was added, and the combination was incubated at 37 °C for 1 h in a shaking water bath. The incubation was terminated by addition of 1:1 (v/v) of ACN. After precipitation of proteins, the sample was centrifuged at 14 000 rpm for 5 min, and the supernatant was analyzed using LC-MS LTQ-XL, as described in the Impurity Profiling section above. The analysis revealed formation of the N-oxide metabolite of desloratadine. Instruments. Sulfasalazine solution (0.01 mg/mL) was injected in different LC-MS systems having different mass analyzers, viz., single quadrupole, triple quadrupole, ion trap, quadrupole trap, and Orbitrap. The composition of the mobile phase was ACN:H2O (50:50 v/v), and flow rates were 100 μL/ min for LC and 126 μL/min for D2O infusion. The D2O was

formulation matrix, and 0.2 mL of rat plasma, microsomes, bile, and urine were considered as endogenous matrices. Ten microliters of a 100 μM sulfasalazine solution in ACN was added to each matrix and further diluted to 0.4 mL with ACN. The supernatant samples were centrifuged at 14 000 rpm for 5 min and injected on a Zorbax C-8 column (250 mm × 4.6 mm, 5 μ; Agilent Technologies, U.S.A.). For chromatographic elution, ACN (A) and H2O (B) were varied in a gradient program (Tmin/A:B; T0−4/30:70; T8/90:10; T14/90:10; T20/ 30:70. Application. Analytes. Samples were prepared by weighing and dissolving 1 mg of analyte (verapamil, ketoconazole, pimozide, haloperidol, ranitidine, metoprolol, isoniazid, sulfasalazine, ritonavir, labetolol, and ornithine) in different volumetric flasks with 100 mL of ACN:H2O (50:50 v/v). Samples were analyzed on LC-MS LTQ-XL through a union employing ACN:H2O (50:50 v/v) as the mobile phase. The LC and D2O flow rates were maintained at 100 μL/min and 126 μL/min, respectively. Other than sulfasalazine, which was used as a model drug, the utility of the methodology was evaluated through its application to several drugs with different structures, viz., verapamil, ketoconazole, pimozide, haloperidol, ranitidine, metoprolol, isoniazid, ritonavir, labetolol, and ornithine. In each case, 1 mg of drug was dissolved in 100 mL ACN:H2O (50:50 v/v) and an aliquot of each solution was analyzed by LC-MS LTQ-XL using dissolving solvent as the mobile phase. The LC and D2O flow rates were 100 and 126 μL/min, respectively. Impurity Profiling. A known impurity of pravastatin was generated by exposing 30 μM of pravastatin to 0.1 N HCl for 1 h at room temperature. The generated sample having a lactone 10907

dx.doi.org/10.1021/ac402339s | Anal. Chem. 2013, 85, 10904−10912

Analytical Chemistry

Article

Figure 2. (a) H/D exchange efficiency at constant SG 20 and varied AG. (b) H/D exchange efficiency at constant AG 10 and varied SG. (c) H/D exchange efficiency at various analyte concentrations (mg/mL). (d) H/D exchange efficiency at various matrices (M-microsomes; P-plasma; B-bile; U-urine; Ex-excipients). (e) H/D exchange efficiency at various aqueous mobile phase concentrations of acetonitrile (H2O). (f) H/D exchange efficiency at various aqueous LC modifiers.

plot of LC output, D2O infusion rate, and corrected H/D exchange intensity. LC flow rate and D2O infusion rate of ≤100 μL/min and ≥107 μL/min, respectively, were required to achieve more than 80% corrected H/D exchange efficiency. Sheath and Auxiliary Gas Flow. Postcolumn addition of D2O increases the total volume of solvent to be nebulized. Therefore, sheath gas (SG) and auxiliary gas (AG) could play a potential role in the ionization step. Also, based on mobile phase composition, these gases need to be optimized during mass tuning, a general practice in mass spectrometry method development. Based on the results of section Ratio of LC Flow Rate to D2O Infusion Rate, LC flow was selected as 50 and 100 μL/min. For the first set of analyses, SG was kept constant at 20, and AG was varied from 10 to 40. For the second set of analyses, AG was kept constant at 10, and SG was varied from 20 to 60. The results of both sets are shown in Figure 2a,b, respectively. For all the experiments, the corrected intensity of m/z 403 varied from 82 to 87%. As the difference in results was insignificant, the effects of SG and AG were considered insignificant for H/D exchange efficiency. Based on optimization experiments, LC flow rate and D2O infusion rate were finalized as 50 and 126 μL/min, respectively. However, dependent on the specific elucidation requirement, LC flow could be increased to 100 μL/min with minimal loss of H/D exchange efficiency. As far as the mass method was concerned, there was no restriction on D2O infusion addition mode, including SG and AG flow. Robustness of the Technique. Considering the nature of the sample and the mobile phase as two critical factors, the robustness of the proposed H/D exchange technique was

infused through the tee union using an external pump (Harvard Apparatus, Holliston MA).

■

RESULTS AND DISCUSSION H/D Exchange Technique Optimization. Based on the current approach, three potential factors were envisaged to affect the partial H/D exchange phenomenon: (1) D2O infusion inlet as postcolumn addition mode; (2) ratio of LC flow rate to D2O infusion rate; and (3) sheath and auxiliary gas flow. D2O Infusion Inlet as Postcolumn Addition Mode. There are two possibilities for postcolumn infusion in the existing Thermo LC-MS LTQ XL setup for carrying out the H/D exchange: (i) through the sheath liquid inlet of ESI source and (ii) through a tee union. In order to understand the effect of D2O addition, ESI source was selected for both the possible inlets. The corrected intensity of m/z 403 (maximum deuterium exchange) for sheath liquid inlet and tee union was 83 and 80%, respectively. As these results were comparable, the effect of D2O addition was considered insignificant. Ratio of LC Flow Rate to D2O Infusion Rate. The ratio of LC flow rate to D2O infusion rate needs to be optimized to get significant H/D exchange of the analyte, as nondeuterated LC mobile phase can also consume D2O. A statistical approach (bracketing and matrixing) was designed to optimize LC flow rate and D2O infusion rate by varying their ratios from 1:0.25 to 1:2.5. The results are shown in Figure 1a−e. H/D exchange efficiency increased with an increase in D2O infusion rate. The best results were obtained with an LC output of 50 μL/min and D2O infusion rate of 126 μL/min. Figure 1f shows the contour 10908

dx.doi.org/10.1021/ac402339s | Anal. Chem. 2013, 85, 10904−10912

Analytical Chemistry

Article

Table 2. Chemical Diversity of Labile H and Their H/D Exchange Efficiency

evaluated on the basis of the following parameters: (i) concentration of analyte; (ii) sample matrices; (iii) relative ratio of aqueous and organic components in the mobile phase; (iv) aqueous LC modifiers; and (v) organic LC modifiers. Effect of Analyte Concentrations. From the previous discussion (Ratio of LC Flow Rate to D2O Infusion Rate), H/D exchange efficiency was identified to raise with an increase in D2O infusion rate. This could be attributed to the increased ratio of D2O to analyte along with a decreased ratio of LC mobile phase to D2O in nebulized droplets. Thus, the concentration of the analyte was one of the factors that could

potentially affect the ratio of D2O to analyte. A comparison of corrected intensity of m/z 403 for different concentrations (0.0005 to 0.1 mg/mL) is shown in Figure 2c. The intensities were in the range of 59 to 84%, with a trend of lowering of H/ D efficiency with an increase in concentration. This is explained by the change in the ratio of D2O to analyte concentration. However, this change is not significant (84 to 59%; ∼30%), in comparison with the concentration change (0.0005 to 0.1 mg/ mL; ∼20 000 fold). This can be attributed to the fact that molar concentration of D2O would be much higher (∼10 000 fold) than the working concentration of the analyte in nebulized 10909

dx.doi.org/10.1021/ac402339s | Anal. Chem. 2013, 85, 10904−10912

Analytical Chemistry

Article

Figure 3. (a) MS spectrum of pravastatin lactone after H/D exchange. (b) MS spectrum of desloratadine N-oxide after H/D exchange.

droplets. In light of this analysis, this technique is capable of handling a wide range of working concentrations of analytes. Effect of Sample Matrices. As H/D exchange is most widely used for elucidating unknowns present as impurities, degradants, and metabolites, it is important to understand the effect of sample matrices on the sensitivity of the technique. The matrices could be a complex mixture of excipients, denatured proteins, oligopeptides, lipids, sugars, and so forth. While sample preparation (extraction) and LC separation would remove these nonanalyte substances, a few of them could coelute with the analyte. In such instances, these extraneous substances would compete with the analyte for H/D exchange, as they may also have significant number of labile hydrogens in the form of −NH, −OH, −COOH, and −SH. A comparison of corrected intensity of m/z 403 for different matrices is shown in Figure 2d. The corrected intensity remained consistent between 71 and 75%, except in the case of bile (60%). This decrease in efficiency in bile could be due to coelution of bile salts, but this did not significantly impact H/D exchange of sulfasalazine. On the basis of this

analysis, it can be inferred that number of labile hydrogens present in an analyte may be easily distinguished in all matrices. Effect of Relative Ratio of Aqueous and Organic Components in the Mobile Phase. Almost all LC-MS analyses are carried out with RP HPLC using an aqueous mobile phase and an organic modifier. As H2O in the aqueous phase could potentially compete with the analyte for D2O, the relative ratio of the aqueous mobile phase can affect H/D exchange efficiency. Moreover, the elution of the unknown compounds could vary significantly because of their inherent physicochemical properties such as pKa and log P/D. Thus, the effect of the aqueous mobile phase component on H/D exchange efficiency was studied at five different relative ratios. The results are shown in Figure 2e. It was observed that the higher the water content in the mobile phase, the lower the H/D exchange efficiency. However, even at 95% aqueous content, corrected intensity of m/z 403 was 78%, which is adequate to identify the number of labile hydrogens in the structure of an analyte. Effect of Aqueous LC Modifiers. The aqueous modifiers in LC are usually required in most analyses, owing to the benefits 10910

dx.doi.org/10.1021/ac402339s | Anal. Chem. 2013, 85, 10904−10912

Analytical Chemistry

Article

(−OH), whereas lactone formation would involve the loss of two labile hydrogens (one from −OH and other from −COOH). Post H/D exchange LC-MS showed an m/z value of 431 (MD + Na), indicating the presence of only two labile Hs, thus confirming the lactone structure. Metabolite Identification. The RLM samples of desloratadine showed multiple mono-oxidative metabolites of m/z 327 in ESI + ve mode. It is reported that hydroxyl metabolites (labile Hs: 2) elute before desloratadine (RRT < 1), whereas the N-oxide metabolite (labile Hs: 1) elutes after desloratadine (RRT > 1).21 In order to extend the proposed online H/D exchange technique in this case, the N-oxide metabolite having an RRT of 1.1 was analyzed. The possible structure of the mono-oxidative metabolite and the line-spectrum after H/D exchange is shown in Figure 3b. The intensity after correction (74.8%) indicated the presence of one labile H, thus confirming the N-oxide metabolite. Instruments. Initial studies were carried out on LTQ XL instrument, so the proposed technique was evaluated on other in-house available LC-MS systems of different types, viz., single quadrupole, triple quadrupole, quadrupole trap and Orbitrap. Results were comparable on all the instruments [S4], highlighting that the proposed H/D exchange technique can be employed using any type of LC-MS.

they have with respect to the peak shape. They provide appropriate pH to the mobile phase, based on pKa of the analyte. However, most commonly used modifiers in LC-MS analysis themselves have labile hydrogen(s) and can compete with the analyte for D2O. In order to investigate this, four commonly used modifiers (ammonia, ammonium formate, ammonium acetate, and formic acid) were screened along with water. Even though the H/D exchange efficiency in all the four modifiers decreased slightly as compared to water alone (Figure 2f), the number of labile Hs could be easily distinguished in all cases. Effect of Organic LC Modifiers. The most commonly used organic LC modifiers are methanol and ACN. Because methanol is an alcohol with one labile hydrogen, it has the potential to decrease the effect of H/D exchange. The influence of both organic modifiers was screened for H/D exchange efficiency, and it was found that methanol had a considerable negative effect on corrected intensity of m/z 403 (64 vs 84%). Even with the reduced H/D exchange efficiency with methanol, the number of labile Hs present in analyte could easily be distinguished in both the organic modifiers. Hence, it is desirable but not necessary to have ACN as the modifier in H/ D exchange experiments. Applications of the Technique. In order to establish a wide applicability, the optimized online H/D exchange technique was tested on different analytes and instruments. Analytes having a different number of labile Hs were tested. The applicability of the technique in impurity profiling and metabolite identification was also demonstrated. Analytes. As optimization and robustness of the technique was established using a single compound (sulfasalazine) having three labile Hs, it was necessary to check the potential of the technique with a different number of labile Hs. Moreover, it was considered important to check compounds with varying chemical diversity as the exchange rate at the molecular level would be dependent on the pKa of the chemical group. A range of analytes based on the number of labile Hs and the diversity of chemical groups were selected. In all these cases, corrected relative intensity of the completely exchanged moiety was sufficient to distinguish the number of labile Hs (Table 2). The observed difference in H/D exchange efficiency having the same number of labile Hs was due to a different exchange rate at the molecular level. In order to confirm this, experiments were carried out using acidic or basic D2O using an acidic modifier (nondeuterated formic acid, acetic acid, and trifluoroacetic acid) or a basic modifier (nondeuterated ammonia, diethyl amine, and triethyl amine). The result [S3] supports the effect of pKa as a contributor to the H/D exchange efficiency. Impurity Profiling. As per different regulatory requirements, impurity identification is one of the most important areas to comply with as it regards the quality specification of the drug candidate. Pravastatin impurity was selected to show the applicability of this technique. It was reported that pravastatin on acid degradation forms pravastatin lactone as an impurity, and so samples for the present study were generated by adding pravastatin in acidic medium.20 Acid degradation samples of pravastatin were subjected to online H/D exchange study. Before online H/D exchange, one of the degradation impurities showed an m/z of 429 (M + Na), 18 mass units less than the parent drug. The loss was attributed to removal of H2O from any of the three −OH groups, as shown in Figure 3a. It was apparent that alkyl dehydration would result in the formation of one π-bond with the loss of one exchangeable hydrogen

■

CONCLUSIONS A simple and cost-effective technique for H/D exchange has been developed on existing LC-MS instrumentation. The technique does not call for extended system setup and equilibration and a large volume of deuterated water consumption. Further, the algorithm for delineating partially exchanged moieties was demonstrated with a simple equation. The technique was also used for distinguishing the number of labile hydrogens in multiple structures using calculated relative intensities. In an ideal situation, any positive value of corrected relative intensity for m/z is indicative of deuterium exchange. However, an acceptable limit is required to avoid any falsepositive interpretation. From the current study, the lowest corrected relative intensity was 30.3%, which was observed for ritonavir. Thus, considering method capability and potential risk of false positives, a conservative acceptable limit of >15% has been proposed for the interpretation of the data. As such, the technique was applicable under a variety of analyte concentrations, mobile phases, and matrices. As we have shown in this work, this technique could easily be extended to degradation impurity and metabolite identification. We see the potential applications of this technique to different fields such as counterfeit identification, mass fragmentation pattern assignment, intermediate identification in chemical/degradation reactions, and rank ordering of molecules based on reactivity of functional groups.

■

ASSOCIATED CONTENT

S Supporting Information *

LC-MS spectra, including observed relative intensities and corrected intensities. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. 10911

dx.doi.org/10.1021/ac402339s | Anal. Chem. 2013, 85, 10904−10912

Analytical Chemistry

Article

Present Addresses ■

Novartis India Limited, Dr Annie Besant Road, Worli, Mumbai 400 018, Hyderabad, Andhra Pradesh, India. ▲ University of Washington, Seattle, Washington 98195, U.S.A. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank Dr. Ramaswamy Iyer, Dr. Jonathan Josephs, Dr. Mark Bolgar, and Dr. Ragu Ramanathan from the BristolMyers Squibb Company, U.S.A., for their helpful comments and discussion.

■

REFERENCES

(1) Singh, S.; Handa, T.; Narayanam, M.; Sahu, A.; Junwal, M.; Shah, R. P. J. Pharm. Biomed. Anal. 2012, 69, 148−173. (2) Pan, C.; Liu, F.; Ji, Q.; Wang, W.; Drinkwater, D.; Vivilecchia, R. J. Pharm. Biomed. Anal. 2006, 40, 581−590. (3) Prasad, B.; Garg, A.; Takwani, H.; Singh, S. TrAC, Trends Anal. Chem. 2011, 30, 2360−2387. (4) Shah, R. P.; Sahu, A.; Singh, S. Anal. Bioanal. Chem. 2013, 405, 3215−3231. (5) Novak, T. J.; Helmy, R.; Santos, I. J. Chromatogr., B 2005, 825, 161−168. (6) Oslen, M. A.; Cummings, P. G.; Kennedy-Gabb, S.; Wagner, B. M.; Nicol, G. R.; Munson, B. Anal. Chem. 2000, 72, 5070−5078. (7) Pan, C.; Jiang, Y.; Wen, H.; Motto, M. J. Pharm. Biomed. Anal. 2012, 57, 99−103. (8) Liu, T.; Du, F.; Zhu, F.; Xing, J. Rapid Commun. Mass Spectrom. 2011, 25, 3303−3313. (9) Siegel, M. M. Anal. Chem. 1988, 60, 2090−2095. (10) Ohashi, N.; Furuuchi, S.; Yoshikawa, M. J. Pharm. Biomed. Anal. 1998, 18, 325−334. (11) Liu, D. Q.; Hop, C.; Beconi, M. G.; Mao, A.; Chiu, S. Rapid Commun. Mass Spectrom. 2001, 15, 1832−1839. (12) Zhang, Z.; Smith, D. L. Protein Sci. 1993, 2, 522−531. (13) Keppel, T. R.; Jacques, M. E.; Young, R. W.; Ratzlaff, K. L.; Weis, D. D. J. Am. Soc. Mass Spectrom. 2011, 22, 1472−1476. (14) Wu, Y.; Engen, J. R. J. Am. Soc. Mass Spectrom. 2006, 17, 163− 167. (15) Gucinski, A. C.; Somogy, A.; Chamot-Rooke, J.; Wysocki, V. H. J. Am. Soc. Mass Spectrom. 2010, 21, 1329−1338. (16) Felix, T.; Reyzer, M.; Brodbelt, J. Int. J. Mass. Spectrom. 1999, 190/191, 161−170. (17) Evans, S. E.; Lueck, N.; Marzluff, E. M. Int. J. Mass. Spectrom. 2003, 222, 175−187. (18) Lam, W.; Ramanathan, R. J. Am. Soc. Mass Spectrom. 2002, 13, 345−353. (19) Wolff, J.; Laures, A. J. Rapid Commun. Mass Spectrom. 2006, 20, 3769−3779. (20) Nigovi, B.; Vegar, I. Croat. Chem. Acta 2008, 81, 615−622. (21) Chen, G.; Daaro, I.; Pramanik, B. N.; Piwinski, J. J. J. Mass. Spectrom. 2009, 44, 203−213.

■

NOTE ADDED AFTER ASAP PUBLICATION After this paper was published ASAP on November 8, 2013, a correction was made to Table 1. The corrected version was reposted November 19, 2013.

10912

dx.doi.org/10.1021/ac402339s | Anal. Chem. 2013, 85, 10904−10912